Essays in Development Economics in Honor of Harry T. Oshima by Ronald Yacat

Essays in Development in Honor of Harry

Economics

T. Oshima

Essays in Development

Economics

in Honor of Harry

T. Oshima

PHILIPPINE INSTITUTE FOR DEVELOPMENT STUDIES

The Philippine Institute for Development Studies is a non-stock, non-profit government research institution engaged in long-term policy-oriented research. It was established on September 26, 1977 by virtue of Presidential Decree No. 1201. Through the Institute's activities, it is hoped that policy-oriented research on Philippine social and economic development can be expanded in such a manner as to more directly and systematically assist the government in planning and policymaking. PIDS publishes the output of its research program as part of its effort to promote the utilization of research findings and recommendations. The views expressed in published reports, however, do not necessarily reflect those of the Institute. No part of this book may be reproduced in any form or by any means without permission in writing from the Publisher. Brief quotations in reviews; however, may be made if properly acknowledged.

Printed in the Philippines

NOTE

This volume was planned by J. Encarnaci6n and H. Choo and carded out by R. M. Bautista and E. M. Pemia under the auspices of the Philippine Institute for Development Studies and the University of the Philippines School of Economics. The Research Information and Dissemination Department og the institute provided editorial assistance.

CONTENTS

Tables ...........................................

Figures ..........................................

xvi

Florian A. Alburo, Agricultural Modernization and Non-Agricultural Employment." An Asian and Pacific Perspective ..............................

Romeo M. Bautista, The Inflationary Effect of LDC Exchange Rate Changes Under Generalized Currency Floating ..............................

Bryan L. Boulier, An Evaluation of Time Budget Studies as Complements to Conventional Labor Force Surveys .......................................

Burnham O. Campbell, Age Structure and Development in A SEAN and Japan: 1950-2015, A Preliminary Report .......................................

Amado A. Castro, Foreign Trade and Economic Welfare in the Last Half-Century of Spanish Rule .............

Hakchung Choo, Widening Urban-Rural Income Differentials in Korea: A Reexamination .............

115

Rolando A. Danao, On the Use of Linear Programming for Family Planning Resource Allocation .............

149

Jos6 Encarnaci6n Jr., An Auxiliary Model for Quantifying the Socio-Economic Impact of a Development Project ...........................

159

Alejandro N. Herrin, Philippine Demographic Development: Problems and Prospects ...............

175

G. L. Hicks and S. G. Redding, Culture and Corporate Performance in the Philippines: The Chinese Puzzle ........................................

199

vii

viii

CONTENTS

Helen Hughes, Long Run Trends in the Growth of Developing Countries ............................

217

Shigeru Ishikawa, Relevance of the Experiences of Japan to Contemporary Economic Development ............

255

H. N. S. Karunatilake, The Impact of Sri Lanka's Economic Reforms in 19 77 on Employment and lncome Distribution .........................

281

Shirley W. Y. Kuo, Technical Change and Factor Utilization: The Case of Taiwan, 1952-1980 ..........

297

Toshiyuki Mizoguchi, Laboring Poor of Japan in the Early Stages of Industrialization .................

315

Konosuke Odaka, An Analysis of the Personal Consumption Expenditures in Japan, 1892-1967 ....................................

335

Mitsuo Ono, Growth and Equity in a Sub-Regional Economy: Hawaii, 1900-1976 .....................

357

Ernesto M. Pernia, Asian Urbanization and Development: A Comparative View .............................

383

Miyohei Shinohara, Growth and Stability in Selected Newly-Industrializing and Socialist Economies (With Particular Reference to Investment Cycles)

405

Edita A. Tan and Wannasiri Naiyavitit, Distribution of Education in Thailand .........................

......

Flow 421

Kunio Yoshihara, Sogo Shosha and Japanese Industrialization ................................

459

Research Publications

483

of Harry T. Oshima ...............

TABLES

Album Paper 1 2 3

Regressions of Non.Agricultural Employment: Summary Asia and Pacific Countries ................... Regressions of Non.Agricultural Employment: Summary South and Southeast Asian Countries ............ Regressions of Non-Agricultural Employment: Summary Asia and Pacific Countries (Log Transformation) ............................. Extended Regressiofis of Non-Agricultural

Employment: Summary Asia and Pacific Countries A.1 Definitions of Variables ............................. A.2 List of Asia and Pacific Countries Included in the Analysis ................................. A.3 Data Tabulation Asia and Pacific Countries

.........

.................

15 17 18 19

Bautista Paper 1 2

Regression Results A (Dependent variable: _b; 20 observations) ................................ Regression Results B (Dependent variable: P; 16 observations) ................................

3 Regression Results C (Dependent variable: P) ............... 4 Regression Results D (Dependent variable: jb) ............... Appendix A Definition of Variables and Data Sources ............. Appendix B Values of E, V and Pf ..........................

31 32 34 35 38 40

Boulier Paper 1

A Comparison of Time Allocation Estimates Obtained from Retrospective Questionnaires and from Observation ............................... ix

TABLES

Campbell Paper 1

Net Relative Income and Composition Effects of Changes in Age Structure: Selected Variables .....................................

Castro Paper 1 2 3 4 5

Growth of Philippine Foreign Trade in the Nineteenth Century .............................. Quantity Index of Major Exports (1876 - 100) .............. Philippine Exports ................................. Philippine Imports ................................. Compound Rates of Growth (Percent) ....................

99 102 104 106 110

Choo Paper 1

2 3

Agricultural Household Income as Percentage of Non-Agricultural Household Income, 1965-1979 .................................... Farm Income as Percentage of City Wage Earners' Income, 1965-1979 ........................ Adjusted Farm Household Income as Percentage of City Wage Earners' Household Income, 1965-1977 .................................... Adjusted Farm Per Capita Income as Percentage of City Wage Earners' Per Capita Income, 1965-1977 .................................... Ratio of Direct Taxes and Levies to Farm and City Wage Earners' Household Income, 1965-1979 (In Percent) ............................

International Comparison of Urban and Rural Cost of Living Differentials, 1973 (Rural Cost of Living -- 100.0) ............................ Average Family Size of City Wage Earners' Household and Farm Households in Korea, 1965-1979 ....................................

118 120

122

123

125

127

128

Standardized Cost of Living Scales of the U.S. Poverty Line, 1974 ..............................

129

Sample of Procedure for Determining Farm Output in Current Prices ...........................

131

TABLES

11 12

Imputed Income from Housing Ownership and Rentals of City Wage Earners as Percentage of Household Income, 1963-1979 ..................... Rates of Increase of Korean Farm Prices Paid by Major Categories, 1962-1979 (In Percent) ............. Assignment of Relative Weights of Major Expenditure Items of City and Farm Consumer Price Indices ................................... Comparison of City and Farm Consumer Price Increases by Major Expenditure Categories, 1970-1973 and 1974-1978 (In Percent) ................. Farm Household Effective Level of Living income as Percentage of City Wage Earners' Household Income, 1963-1979 (Estimation I) ............. Farm Household Effective Level of Living Income as Percentage of City Wage Earners' Household Income, 1963-1979 (Estimation II) ............ Farm Household Effective Level of Living Income as Percentage of City Wage Earners' Income by Income Classes, 1976 .....................

133 135

136

137

140

141

143

Encarnaci6n Paper 1

Means of Variables .................................

161

Regression Equations for CEB

164

Coefficients of YHiCand

.........................

YH2C

.......................

169

Hicks and Redding Paper 1

Sales Volume and Industry Sector by Culture of Ownership. The Top 259 Philippine Companies. All Figures in Million Pesos ................. Manufacturing Industries (out of top 259 companies), No. of Companies, Chinese and Non-Chinese, by Industrial Sector ..................... Commercial Industries (out of 259 companies), No. of Companies, Chinese and Non-Chinese, by Industrial Sector .............................. Manufacturing Sector, Comparison of Chinese and Non.Chinese Financial Performance (In Million Pesos) ................................

205

206

207

208

xii

5 6

TABLES

Manufacturing Sector ............................... Commercial Sector .................................

208 209

Hughes Paper 1

2 3

4 5 6

7 8

9 10

11 12

Growth Trends in GNP and GNP Per Capita (Average Annual Percentage Growth Rates in 1980 US$ at Official Exchange Rates) .................. Developing Countries' Population and Per Capita income by Region, 1980 ...........................

220 222

Developing Countries' GNP and Per Capita Growth by Region (Average Annual Percentage Growth Rates in 1980 US$ at Official Exchange Rate) ................................. Magnitude of Remittance Inflows into Major Labor-Exporting Countries, 1978-1979 ................. Country Composition of World Exports, 1960, 1970 and 1978 (Percent) ...................... Composition and Growth of World Merchandise Exports by Commodity Groups, 1965, 1970 and 1978 (Percent) .............................. Regional Structure of Merchandise Trade Flows, 1970 and 1979 ............................

223 236 237

238 240

Share of Imports in the Apparent Consumption of Manufactured Goods in Industrial Countries, 1970-1979 ............................. Non-Factor Services: Shares in Total

243

Exports, 1970, 1978 (Percent) ....................... Composition of Net Capital Flows to Developing Countries, 1960-1962 and 1976-1978 (Percent) ............................. Developing Country Debt and Debt Service Indicators, 1970-1979 (Percent) ....................... Shares of World Population and GNP (percent) ..............

244

246 250 252

Karunatilake Paper 1

Average Size of Household, Number of Spending Units, Income Receivers and Dependents Per Household, 1963, 1973 and 1978/79 ...................

291

TABLES

4 5

Consumer Finance Survey Estimates of Average One Month Income (Rs) per Income Receiver, 1953, 1963, 1973 and 1978/79 ...................... Consumer Finance Survey Estimates of the Gini Coefficient for Income Receivers, 1953-1978/79 .................................. Percentage of Income Received by Decfles, All Island ...................................... Mean Income Per Income Receiver by Decfles, All Island, 1973 and 1978/79 ........................

xiii

293

293 294 295

Kuo Paper 1

3 4

5 6

7 8

Rates of Technical Change and Their Contributions to the Growth Rates in the Non-Agricultural, Manufacturing and Service Sectors, 1952-1980 (Percent) ..................................... Growth Rates of Capital Stock, Labor and Value.Added in Manufacturing, 1952-1980 (In Percent) .................................... Rate of Technical Change in the Manufacturing Industries (In Percent) ............................ Rate of Technical Change in the Non-Agricultural Sectors of Taiwan, ROC (Based on I-0 measurement) (In Percent) ......................... Rates of Technical Change, Japan and the U.S. (Based on I-0 measurement) (In Percent) ................ Growth Rates of Gross Output, Value Added and Exports of the Four Leading Industries (in Percent) ............................. Changes in the Shares of Leading Industries in Manufacturing (In Percent) ....................... Sources of Output Growth (Percentage Distribution, 1961-1976) ..........................

300

302 305

308 310

310 311 312

Mizoguchi Paper 1 2

Occupational Distribution of Family Members in Tokyo Slum in 1911-1912 ........................ Average Income and Labor Days Per Month in 1912 Survey ...................................

322 325

xiv

TABLES

3-A Income Distribution Data of Tokyo Informal Sector in 1921 Data, by Monthly Receipts ............... 3-B Data Per Adult Equivalent Disposable Income ............... 4 Ratios to Income of Household Heads by Kinds of Receipts (Percent) ............................. 5 Composition of Consumption Expenditures (Percent) .....................................

327 328 329 330

Odaka Paper 1 2 3 4

Personal Consumption Function, Japan (1892-1938 and 1954-1967) ........................ Personal Consumption Function in Japan: Prewar (1892-1938) and Postwar (1954-1967) Personal Consumption Function, UX., U.S. and Canada ....................................

342 ............

Effects of Income Growth on Consumption Propensities ...................................

347 349 352

Ono Paper 1

Hawaii's Ranking on Selected Socio-Economic Indicators .....................................

358

Percent of Food and Housing Expenditures to Total Expenditures .............................

375

Standard Deviation of Quintfle Shares, Hawaii, 1949, 1959, 1975 ..........................

377

Pernia Paper 1 2 3 4

5 6

Urbanization Indicators for Selected Asian Regions/Countries, 1950-1980 ....................... Urban and Rural Populations, and Growth Rates: Asian Regions/Countries, 1950-1980 ................... Urban Concentration Indicators: Asian

396

Countries, 1960-1980 ............................. Average Annual Percent Growth Rates of Population, GDP and Sectoral Production: Asian Countries, 1960-1970, 1970-1978 ................ Percentage Distribution of GDP: Asian Countries, 1960-1978 ............................. Export and Imports of GDP (In Percent) ..................

399

398

400 401 402

TABLES

Shinohara Paper 1

Personal Savings Ratio in Korea and Taiwan ................

408

External Deficit/Gross Domestic Investment Ratio in Korea and Taiwan .........................

409

Tan and Naiyavitit Paper 1 2

4 4a 5

Attendance Rate by Age Range 1960, 1975 Rate of School Attendance of Children and

................

Youth by Age, by Region, by Municipal/ Village, 1975 .................................. Percentage Distribution of Youth Not Attending School by Ago, Highest Level Attained, Municipal/Non.Municipal .......................... Attendance Rate by Age, by Family Income, 1975 ........... Percentage Distribution of Population by Age ............... Distribution of Students by High School Test Scores, by Region and Father's Occupation, 1975 ........................................

5a Mean Test Scores of Grade III Pupils by Type of School, Region, Father's Occupation and Urbanization of Location, 1973 ...................... 6

Percentage Distribution of Children and Youth Attending School by Type of School and Distance to School, 1975 ........................

Average Annual Expenditure on Education per Person by Level of Education Attended and

8 9 10 A.1 A.2 A.3 A.4

423

424

426 427 427

429

430

Type of Expenditure ............................. Results of Linear Regressions of Attending or Not Attending School ........................... Range of Values of Probability of Attending School by Age 1975 .............................. Value of Beta in Legit .............................. Regression Results on Alternative Selection

431

of Independent Regression Results of Independent Regression Results

7.10 .................... Selection 11-13 ................... Selection

447

of Independent Variables, Age 14-18 ................... Regression Results on Alternative Selection of Independent Variables, Age 19-24 .....................

451

Variables, Age on Alternative Variables, Age on Alternative

443 445 446

449

453

FIGURES Alburo Paper 1 2

Percent Non-Agricultural Employment and Paddy Rice Yield, 1975 ................................... Percent Non-Agricultural Employment and Percent of Agriculture GDP to Total GDP ......................

8 9

Campb ell Paper 1 2 3 4a 4b 5

The Key Ratio: ASEAN Members and Japan, 1950 to 2015 ..... Fertility Rates and the Key Ratio, ASEAN Members and Japan, 1950 to 2015 ............................. Growth Rates of the 15-19 and 30-64 Age Classes in ASEAN Countries and Japan, 1950-1955 to 2010-2015 ........... Total Educational Burdens in ASEAN Members and Japan: Population 5-24/Population 30-64 .................... Higher Educational Burdens in ASEAN Members and Japan: Population 15-24/Population 30-64 .............. Dependency Ratios in ASEAN Members and Japan, 1950-2015 ...................................

74 78 83 87 88 91

Encarnaci6n Paper 1

..............................................

163

Ocla_ Paper 1 2

Average Propensity to Consume in Three Countries ........... Average Propensity to Consume in Japan: Actual and Estimated (Equation (8)) ............... ........

337 346

Shinohara Paper 1 2 3 4

Full Employment Fixed Investment Fixed Investment Fixed Investment

Ceiling Vs. Resource Ceiling .............. Cycle in Korea ....................... Cycle in Taiwan ...................... Cycle in Some Socialist Economies .........

407 411 413 416

Tan Paper 1

.............................................

439 xvi

AGRICULTURAL MODERNIZATION AND NON-AGRICULTURAL EMPLOYMENT: AN ASIAN AND PACIFIC PERSPECTIVE Florian A. Alburo*

1. Introduction A small number of studies in the literature on agricultural modernization and development argue that non-farm economic activities grow from within the agricultural sector itself rather than being externally propelled. The more meaningful or comprehensive models posit the existence of non-manufactured non-farm goods as part of the range of choices determining economic as well as social and demographic behavior in an agrarian environment. 1 When viewed against prevailing policy thrusts for rural and agricultural development, this area in the field offers an interesting ground for empirical analysis. This paper aims to provide a cross-country aggregate perspective of the interactions between agricultural modernization and non-agricultural employment in Asia and the Pacific. The second section further elaborates the context of this study relying on some micro analytic foundation as well as on the practical aspects of the problem. The'third section defines and *University of the Philippines School of Economics. This paper was written while visiting with the Resource Systems Institute, East-West Center Hawaii in November and December 1981. The support of the Center during the visit is gratefully acknowledged. I. See for example the work of S. Hymer and S. Resnick,"A Model of An Agrarian Economy with Non-Agricultural Activities," Amerk_n Economic Review (September 1967), 493-506. Although basically neo-classical in approach, this is probably the seminal theoretic work on this area. The new household economics, while somewhat peripheral partly alludes to optimal behavior. In this paper, non-farm and non-agricultural are considered similar in concept. 1

FLORIANA. ALBURO

tests the various measures used with the interactions empirically specified. The fourth section draws out some behavioral comparisons. The last section concludes. 2. Context One departure point for looking at agricultural modernization and employment is the common policy thrust in developing country plans of increasing rural employment opportunities through rural industrialization. 2 This kind of strategy is construed to provide linkages with larger urban based industries, stimulate auxiliary industries, prevent agglomeration, and expand the base for manufactured exports. Rural industrialization would accelerate non-agricultural employment. Although the degree of employment generation may vary with the scale and kind of industries promoted, it is generally accepted that the employment needs are met. Moreover, the seeming acceptance of redundant employment in agriculture justifies the alternative or complementary emphasis on rural industries. Depending on the extent to which this strategy is pursued, one would normally expect structural changes in agriculture in terms of products as well as magnitudes of non-agricultural employment. On the analytical aspects of agricultural modernizationnon-agricultural activities, allowing for Z-goods implies an optimal basket of goods which include them. The theoretical foundations, results and implications are familiar in the literature and need no further review here. At low levels of development, Z-goods are predominant but sensitive commodities. At more sophisticated levels of modernization, factory type non-inferior manufactured products prevail in an optimal consumption basket. 3 The new household economics likewise suggests that household decisions follow a behavior which can be explained along this framework. 2. The magnitudes and scale of rural industrialization in the current Philippine Development Plan are not really dear-cut. HoWever,it implies some amount of deliberate policy to directly promote non.farm activities. In the plan, "...rural-based non-farm activities will be promoted to increase employment and income opportunities of farm families. The regional industrial dispersal program of the government will support this strategy.... "Five Year Philippine Deve!opmentPlan 1978-1982(Mardla: September 1977), p. 107. 3. S. Hymer and S. Resnick, op.cit.

AGRICULTURE AND NONAGRICULTURAL EMPLOYMENT

The associated

empirical

evidence

is quite limited

and fragment-

ary. 4 Where there are studies, one is basically left with a gap especially on the dynamic expectations regarding non-agricultural employment. For example, Anderson and Leiserson, s while suggesting a framework for tracing the evolution of non-farm activities, fail to provide orders of magnitude on the effects of development on non-agricultural employment. In Gibb's work, different categories of non-agricultural economic activities are postulated based on responsiveness to agricultural productivity changes. The direct effect industries respond to immediate income increases and are mainly of "inferior goods" variety (the Z-goods industry in Hymer and Resnick). Two other industry categories are postulated to grow as on-farm development spreads - indirect effect industries which are the backward linkage types of production activities and public service effect industries which presumably cater to tertiary demands. The micro data from Nueva Ecija, Philippines support the range of hypotheses implied in the framework. However, although some comparisons are made especially in terms of employment among agricultural towns and with agriculturally-based cities, one is left with the impression that it is the static distribution in non-agriculture that is adequately substantiated. 6

4. One survey is E. Chuta and C. Liedholm, "Rural Non.Farm Employment: A Review of the State of the Art," MSU Rural Development Paper No. 4, Department of Agricultural Economics, Michigan State University (1979). A pioneering effort in the Philippines are the research works of A. Gibb, among which are "A Note: Def'ming the Non-Farm Employment Question," L_'scussion Paper No. 71-14 UP Institute of Economic Development and Research (August 1971); "Report on On-Going Research: Some Evidence on the Impact of Agricultural Modernization on Non-Agricultural incomes in Agricultural Market Center," D_cussion Paper No. 72-4 UPIEDR (April 1972) and "Preliminary Data on Non-farm Employment Changes in an Agricultural Sub-Resion" Discussion Paper No. 72-19 UPIEDR (June 1972). A current large scale project is on-going in Thailand on off-farm employment reported in T. Onchan, P. Charsombut, R. L. Meyer and D.C. Mead, "Description of the Rural Off-Farm Employment Assessment Project in Tbsilsnd,'" Research Paper No. 1 Center for Applied Economics Research, Kasetsart University (Bangkok, Thailand, October 1979). 5. D. Anderson and M.W. Leiserson, Rural Enterprise and Nonfarm Employment (World Bank, January 1978). Also Ident Economic Development and Cultural Change (January 1980). 6. A. Gibb, "Some Evidence on Agriculturally-based Cities," _ion PaperNo. 72-17 UPIEDR (June 1972).

FLORIANA. ALBURO

Following an essentially similar framework, a study of two agricultural towns, at two different levels of modernization, does not appear to refute most of what is implied in the modernization non-agricultural employment nexus. 7 Comparisons in terms of employment growth show increasing employment at initial stages of agricultural development, followed by a decline in employment growth. Analysis of the kinds of industries belonging to the specified categories indicated previously, validates theoretical expectations. What is relevant to note is that the employment structure is associated with capital resource needs that go with modernity. In particular, as development proceeds, employment opportunities may fall while capital needs accelerate even though the scale of establishments as measured by the number of workers increases. Regardless of the manner in which employment may be measured, it is argued that one can expect its behavior to be sytematically related to the agricultural sector's overall modernization and growth. Whether over a span of time or looking at comparative agricultural development, it would seem that employment increases at first before failing out. s Several policy implications are discernible in the empirical studies. First is simply the main finding of a large scope of non-agricultural activities and employment associated with agricultural productivity changes even without explicit drive for such. Second, modernization creates a phase of low labor absorption. This partly depends on the degree and sophistication of development taking place. In concrete terms, a higher rate of employment is expected with increases in agricultural productivity than in, say, rural agro-based industries. 9 Third, rural industrial development at early growth stages appears to be internally self-generated and self-financed. 10 This would mean 7. F.A. Alburo, "Comparative Agricultural Modernization and Non.Farm Economic Activities," Journal of Philippine Development (Second Semester 1979), 102-116 and ldem, "Comparative Agricultural Modernization and Non-Farm Employment," Discussion Paper No. 8_12 UPIEDR (December 1980). 8. Ibid. 9. Since the income spread of agricultural productivity changesis achieved more throughdirect installation of industries, the income effect mechanism is more responsive to the usual Engers law and thus stimulates non-agricultural economic activities faster. 10. The evidence is in the two papers cited in footnote 7.

AGRICULTUREAND NONAGRICULTURAL EMPLOYMENT

that assuming that stage is identifiable, scarce development resources can be alternatively channeled and still obtain non-agricultural rural employment opportunities. Finally, and in relation to the previous point, even narrow types of agricultural development policies have a significant contribution to broadening the base of modernization. In short, employment creation of the type deliberately sought through rural development can be satisfied by concentration on narrow agricultural strategies. While the framework is intuitively logical, the fragmented analyses seem to rest on crude quantitative measures and simple statistical manipulations on a small data base. The limit and paucity of data as basis for important policy implications tend to raise questions of validity. Indeed some of the conclusions are not strictly warranted by data and simple analysis. 11 Given this basic micro analytic framework, however, it is possible to test it out at an aggregate level by undertaking a cross-country analysis looking into agricultural modernization and non-agricultural employment relationships. By examining a number of countries which are in a wide spectrum of agricultural modernization, it is possible to see the patterns of non-agriculturai employment. The next section provides a perspective using data from 26 countries in Asia and the Pacific. 12 The primary interest is to understand the interactions between modernization and non-agricultural employment without necessarily imputing a strict behavioral pattern. With appropriate assumptions it is of course possible to advance behavioral notions on the data (see Section 4). The initial task at hand is to determine the measures of agricultural modernization that would reflect various stages of development as hypothesized or pursued by policy. An associated task is to postulate what might be the behavior of the relationships within the .framework.

11. See especially section II in F.A. Alburo, "ComparativeAgricultural Modernization and Non-Farm Employment," Discussion Paper No. 8042 UPIEDR(December1980). 12. See Appendix 2 for a list of the Asia and Pacific countriesincludedin this classification.Not all countries were included in the analysis for lack of availablecompletedata.

FLORIANA. ALBURO

3. Data and Analysis It is argued here that a narrow measure of agricultural modernization is farm productivity while a broader index is gross domestic product (GDP) from agriculture. As a measure alone, farm productivity is really a specific area crop yield while agricultural GDP is value added in agriculture that comprises crop and non-crop economic activities. It is true that crop value added is, by definition, a component of agricultural GDP. But one is not necessarily similar or positively related to the other. Over time, the values of one measure need not consistently vary with the other since agricultural GDP is arrived at from a variety of agricultural outputs which have differences in patterns. More specifically, as used here, productivity is paddy rice yield while the broader measure is the proportion of GDP coming from agriculture. In summary, these two indices reflect the stages of agricultural modernization implied in the framework in the previous section. The data come from various issues of the Food and Agriculture Organization (FAO) Yearbooks and the World Bank country tables. 13 As defined in these sources, paddy rice yield is measured in kilograms per hectare. The broader index of agricultural development is the ratio of agricultural GDP to total GDP at current market prices of the country currency. Non-agricultural employment is defined in the absolute and as a ratio. In the former this is the residual from subtracting agricultural employment from total employment. Economically active employment means "'all persons engaged in economic activity, whether as employees, own-account workers, salaried employees or unpaid workers assisting in the operation of a family farm or business. 'q 4 Agriculture includes forestry, hunting and fishing. It is assumed that while there are possibly country nuances that prevent strict comparability, the definitions are commonly applied and fairly comparable across countries. The ratio measure is simply the ratio of non-agricultural employment to total employment. Two other measures are applied - index of agricultural concentration and per capita food production index. The index of 13, For a description of the data set see C.C. Schlegel and E. Ng "The User's Guide to the PSI World Data,set," Technical Memorandum TM 81-1, Resource Systems Institute, East-WestCenter Hawaii (March 1981). 14. /b/d, 56. See Appendix 1 for complete definitions.

AGRICULTUREAND NONAGRICULTURALEMPLOYMENT

agricultural concentration measures the dominance, in terms of area, of a single crop in the agricultural sector. 15 1975 is the common year for which data are available. Some of the analysis results do not include all countries in the list of Appendix 2. It is also not possible to obtain yearly data for some specific variables (e.g., non-agricultural employment) since the original country sources do not report yearly data. The analysis will follow two tracks. The first compares the responsiveness of non-agricultural employment to agricultural modernization that reflects an early, and a later stage. The second examines the probable path of employment as development is sustained using the same measure. Some description will be utilized and linear regression results will be reported. Figures 1 and 2 plot the 1975 percentage of non-agricultural employment to total employment on paddy rice and the percentage of agriculture in the GDP for the Asia and Pacific countries, respectively. In Figure 1, the pattern seems to be of a positive relationship between _Â˘ield and non-agricultural employment. There are extremes that do not fall neatly such as those of Japan, Korea and Hongkong. The latter essentially exhibits a high percentage of non-agricultural employment and a low paddy rice productivity which is partly explained by its nature as a city-state, i.e., without a substantial agricultural sector. Singapore is not found in Figure 1 because of data unavailability. Both Korea and Japan have been pursuing industrial sub-contracting that is consistent with agricultural development. The same three countries however do not exhibit similar extremities in Figure 2. To ascertain the strength of the responsiveness, an OLS regression equation was fitted into the data. The purpose of the fitting is essentially to ascertain some notion of the elasticities involved in the relationship. The equations are not purported to underlie a complete specification. The regression results in the estimation of elasticities, at the means, of non-agricultural employment to the two measures of agricultural modernization. The estimating form from which the response behavior is derived is

15. The food production index is broader than cereals and thus encompass other agricultural products that are basic parts of food. See Appendix 1.

PERCENT NON-AGRICULInJRAL

i .O

EMPLOYMEI_T

_longkong

'*USA

.90'

_t_lla

.80'

_apan

.70"

.60" llran

*rip ._o-

*K_o

"_ o

PN_ .Sri_M_Y

sio

_r_ .4 0 -

째Poklm_

.3o-

_. m

_째

> t_

L_ _m_oa_, _o. d "_,m_ .20"

_honlltln Papua., NewC,ulrma "B_glad_h

.10'

;

2 PADDY

RICE YIELD

( in '000

KILOGRAMS

CHART PERCENT

NON-AGRICULTURAL

PER HECTARE)

EMPLOYMENT

AND

PADDY

RICE

YIELD,

1975

d_O ")V£O.L0£ cl(]_ 3_Ifl.L'IfY31U$¥-40 J.h]'qO._l._Id

Z m

d(ID "IV,LO.LO,.LdGD _IfLL'I ftDI_IDV _IOzN'_D'd_d QNV ,Llq_I_IAOldI4I_ IV_If'LL'IflDI_IDV-NON INaD_Iad Z ,.L'_I VHD

O',

I o_"

_"|

I o_"

_"I

| o_

I _z"

I oz"

I _r

I or

I co"

Pmlml..L

,0 Z"

- oi" _¢

DmUOPUI

_, .< Z O

o_," _uD_!Js"

D!,_to.n slU_lllqd

Og"

o_" oz

[.-,

uodDe

olto,£111'IV

• os" "T_I'I,L"_rIO1b_F * NON

,UC_l_d

(1)

FLORIANA. ALBURO

PNAGREit

=a+b 1 AGRit

+ b 2 [AGRit] 2 + Ut

where PNAGREit is the percentage of non-agricultural employment of country i at time t, AGRtt is the measure of agricultural development of country i at the same time t and Ut is the error term assumed to 4aave standard normal properties. Non-agricultural employment is defined by substitution as a ratio or in absolute terms (NAGRE). 16 The values and signs of b 1 and b2 would indicate the contribution of development and of its increasing value, respectively. One would expect that elasticities will presumably be lower with a greater spread of modernization, other things being equal. With reference to Figures 1 and 2, it can be noticed that the South and Southeast Asian countries cluster closer than all countries in the Pacific region taken together. This would mean a lesser fit with data points covering only this subset. However, across various specifications it should be expected that (a) the elasticity at the mean of paddy rice yield will be greater than that of the broader measure, agricultural gross domestic product, (b) the incremental change in non-agricultural employment from changes in paddy rice yield will be positive, and negative for the late stage of modernization, and (c) there is a lower fit of the specification for the South and Southeast Asian countries. The estimating equations using four indicators of agricultural development for all countries of Asia and the Pacific and for the sub-group of South and Southeast Asian countries reveal some interesting results. Agriculture's share in GDP and the index of agricultural concentration have strong influences on the variation of non-agricultural employment. Although both have negative coefficients, only the former has been given sufficient explanation with regard to expectations. The negative effect of concentration may be explained in two ways. On the one hand, high concentration of land on a single crop could imply extensive mechanization under a basic enclave-type of agriculture and would not significantly affect non-agricultural employment. On the other hand, a similar high concentration results in skewed income distribution that does not tend to favor the emergence of non-farm economic activities which would normally 16. See Aptmndix 1.

AGRICULTURE AND NONAGRICULTURAL EMPLOYMENT

come

under

a broader

yields. 17 The use modernization fails in the

development

path,

such as increase

in rice

of per capita food production index to reflect is really a misspecification of the problem. For one, it standard significance tests. For another, given its

definition, it is difficult to conceptualize the mechanisms by which it affects or is affected by non-agricultural employment in part because of its encompassing nature. Table 1 summarizes the relevant results for the Asia and Pacific countries. Table 1 - Regressions of Non-Agricultural Employment: Summary Asia and Pacific Countries

VARIABLE

Elasticity a

PDYRY PAGGDP INAGRC PCFDPR

:>0 <0 <0 <0

<0 :>0 :>0 2>0

6.62 1.45 28.58 0.96

R2 0.11 0.81 0.52 0.06

aof b 1 at the meanvaluesof the variables PDYRY - Paddyrice yield (kilogram/hectare) PAGGDP - Ratio of agricultuxeGDPto total GDPat marketprices INAGRC - Index of AgricultuxalConcentration(See Appendix1) PCFDPR - Percapitafood productionindex (See Appendix1)

The table

indicates

GDP, the former coefficient discernible

that

between

has a higher elasticity

palay

yield

and agricultural

than the latter. Moreover

the

of b 1 is positive as expected. The only drawback from the table is the low value of/_- as compared with

the specification using PAGGDP as the modernization index. Within the context outlined in the previous section, the coefficients of the development indices are not as critical as the signs and the implied elasticities. But even if the coefficient values ate taken into account, the estimates show that PDYRY yields the largest positive (non-

17. It is probably reasonable to suppose that rice production is more or less governed by a large number of farms independently distributed compared with other agricultural crops (e.g. export).

FLORIAN A. ALBURO

agricultural) employment cance is low. The

results

essentially reveal although the R-

for

contribution

the

South

the same perceptibly

although

and

the level of signifi-

Southeast

Asian

countries

relative magnitudes of the elasticities suffer in significance.l 8 The directions

of the relationships between modernization and non-agricultural employment are the same with the larger group of Asia and Pacific countries. Table 2 is a summary of the results for the South and Southeast Asian countries.19 Table 2 - Regres_tionsof Non-Agricultural Employment: Summarya South and Southeast Asian Countries

VARIABLE

Elasticity

PDYRY

PAGGDP

>t3

0.84

0.26

INAGRC

28.13

0.29

PCFDPR

0.52

--0.08

8.36

_2 -0.13

a See Table 1 for legend.

In the context of the framework presented and the specifications followed in the data analysis, it is not really necessary to compare the two indices of agricultural development in order to trace effects on non-agricultural

employment.

It is sufficient

and magnitudes of the modernization second coefficient simply measures

to examine

the signs

coefficients - b 1 and b 2. The the incremental change in

18. In a separate specification using PNAGRE as dependent variable with the PDYRY indicator, the R2is 0.21 but the coefficients are opposite to that of Table 2. Either one however does not yield significant coefficients. 19. The number of cases for the South and Southeast Asian countries is 13 for PAGGDP, 15 for INAGRC and 17 for PDYRY. The classification in Appendix 3 however indicates only 16 countries for the South and Southeast Asia.

AGRICULTUREAND NONAGRICULTURALEMPLOYMENT

non-agricultural employment relative to a continually increasing change in the variable. In other words, one need not look for different measures of the independent Variable but assume that development takes place as the index increases at increasing rates. It is apparent from Table 1 that at early development stages nonagricultural employment is positive but declines as this kind of growth proceeds. On the other hand as b 1 and b 2 for PAGGDP show, as a broader form of agricultural modernization spreads later, non-agricultural employment declines before picking up again. One may argue however that as PDYRY increases, this contributes to PAGGDP and therefore transmits effects on non-agricultural employment. This point is only partly true since other agricultural growth takes place and thus the impact on employment must net out the various responses. Note that the experience in agricultural development is the predominance of large export crops along with or prior to food crop expansion. Equations (2) and (3) report the specifications for Asia and the Pacific in 1975 (2) PNAGREit = 0.98 - 2.95 PAGGDPit + 2.69 [PAGGDPit]2 + Ut (- 0.59) (1.11) k2

-- 0.81

-- 39.36

(3) NAGREIt = 58015 + 43.72 PDYRY - 0.005 [PDYRYit] 2 _2 = 0.11 (34.27)

(-F 0.005)= 2.40

The numbers underneath the coefficients are the standard errors of the coefficients. The coefficients, while as expected, do not appear to be significant at usual levels and the _2 is quite low. This could partly be corrected by assuming a non-linearrelation. It can be seen that PAGGDP initially reduces non-agricultural employment before increasing it by equivalent magnitudes whereas PDYRY increases it by an amount relatively greater than what it reduces with continuous improvements of palay yield. All this seems consistent with the relationships argued in the framework. The preceding relationships, when investigated in logarithmic transformation, lend an even str92nger support to the hypothesis advanced. In particular, the fit ( R ) improves substantially for the independent variable paddy rice yield and weakens for all other indices of agricultural modernization. The strength of the elasticities

FLORIANA. ALBURO

remain consistent with the results in the simple linear form (Table 1). Some differences are observable however in terms of the signs of the coefficients. Table 3 summarizes the results for the log-linear regressions as basis for deriving non-agricultural employment responses to the same set of modernization indicators. Table3 - Regressionsof Non-AgriculturalEmployment: SummaryaAsia andPacific Countries (LogTransformation) VARIABLE

Elasticity

PDYRY PAGGDP INAGRC PCFDPR

>0 <0 (0 <0

<0 <0 >0 <0

2.34 1.20 46.65 0.75

_2 0.26 0.63 0.35 0.01

aSeeTableI forlegend.

4. Some Behavioral Patterns It is possible to extend the previous analysis towards some behavioral notion of non-agricultural employment, partly to examine the consistency of the earlier results. By postulating, for example, homogeneity of non-agricultural employee characteristics, one can specify a model explaining non-agricultural employment. An important variable that can obviously determine non-agricultural employment behavior is some indicator of wage rates. However, there is no index available from the data set reflecting wages for a comtLon year or across the Asia and Pacific countries. Hence it is difficult to build this into a specification. Another determining variable would be the past non-agricultural employment behavior in the sense of some lag structure. The danger here, as expected, will be the statistical problem of autoeorrelation andnon-independence. For example a specification of the type (4) PNAGREIt = a

+ bI

AGRIt 4. b2 [AGRu]2 4. b3 PNAGREI,t.I 4. Ut

will likely encounter problems that would call to question its proper

AGRICULTUREAND NONAGRICULTURAL EMPLOYMENT

interpretation. Nevertheless Comparisons can perhaps be made with the parameters derived previously. Table 4 summarizes the results using (4) but allowing past non-agricultural employment behavior (1970) to influence current behavior. It has not been possible to use the past year's non-agricultural employment as an independent variable since the earliest available data for the variable is 1970. Note that the data for nonagricultural employment is 1975, which is being explained. Thus the specifications are at best exploratory. Nevertheless, as pointed out, the aim is simply to discern some consistency with the previous results and not attribute particular behavioral patterns. Table4 - ExtendedRegressionsof Non-AgriculturalEmployment: SummaryAsia andPacificCountries VARIABLE

Elasticitya

PDYRY PAGGDP INAGRC PCFDPR

>0 >0 <0 .....

<0 <0 >0

>0 >0 >0

0.18 0.09 0.004

_2 0.99 0.99 0.99

aOfblat themeanvaluesof thevariables.

Elasticity values for PDYRY and PAGGDP are absolutely lower than the values in Table 1. The responsiveness of non-agricultural employment to PDYRY remains relatively greater than PAGGDP even though the signs are the same. The2weakness of this specification is evident in the high values of the R . 5. Conclusions The rationale set out for this study was to look at the interactions between agricultural modernization and non-farm employment in Asia and the Pacific. Based on some micro analytic foundations, the analysis sought to provide amore quantitative meai),s to the perspectives drawn in small studies.

FLORIANA. ALBURO _

It turns out that the results tend to be supportive of micro studies. In particular it is found that early stages of modernization, reflected in increases of palay yield, lead to increases in non-agricultural employment. The mechanisms by_which this takes place work through demand considerations or income effects and through supply possibilities, or non-farm economic activities. Higher stages of agricultural modernization reflected in increases in the fraction of agricultural GDP to total GDP lower non-agricultural employment. Part of the reason for this phenomenon is the capital intensification of industries which result in lower labor absorption rates. To be able to pinpoint the development level at which higher forms of non-agricultural employment and thus, lower labor absorption, take place, more micro studies would be needed.. The strong conclusion that can be drawn here is that narrower forms of agricultural modernization, such as productivity improvement programs do stimulate non-agricultural economic activities which also tend to increase the overall agricultural GDP. A policy thrust therefore that combines rural industrialization with yield acceleration will require a closer scrutiny in order to specify potential trade-offs as implied in the results. Indeed it may even be feasible that rural industrialization and manufacturing development will have stronger linkages through small-scale industrial employment of less than 10 workers that spins off from agricultural modernization. This is not inconsistent with the experiences of some developed countries. 20 In short, non-agricultural employment may also be served equz'_y well with on-farm, on-agricultural development. The specific policy conclusions attendant to the findings here are simply a re-echo of previous works. For example, in order to smoothen adjustments for non-agricultural economic activities, capacity building programs would make sense - skills training for Occupation in rural industries_(of inferior goods), technical support in management and financial administration of rural non-agricultural establishments, and at some later stage financial resources to support capitalization and rural industrial development. Although these points do not follow from the analysis given in this paper, the f'mdings, coupled with associated evidences from other studies support them and make the directions more concrete. 20. H.T. Oshirna, "Labor Force Explosion and the Labor-Intensive Sector in Asian Growth," Economic Development and CulturalChange(January 1971).

AGRICULTUREAND NONAGRICULTURAL EMPLOYMENT

Appendix 1 Definitions of Variables PNAGRE-

NAGRE

PAGGDP-

PDYRY INAGRC -

PCFDPR

percent of non-agricuRural employment to total employment. Employment is total economically active population (000) which means all persons engaged in economic activity, whether as employers, own-account workers, salaried employees, or unpaid workers assisting in the operation of a family farm or business. Economically active population in agriculture includes forestry, hunting and fishing. economically active non-agricultural population (000) derived by subtracting agricultural employment from total employment. percent of agriculture GDP (at current prices, in million units) to total GDP (at current prices, million units). paddy rice yield in kilograms per hectare percentage computed by dividing the area planted in the largest single crop by the total area in the five largest crops. per capita food production index (1952-56 = 100) calculated on a calendar-year basis; The number represents price-weighted production volume for production flowing to all sectors other than agriculture itself. This figure is compared with average volume during the base period. The food production index includes cereals, starchy roots, sugar, edible oil crops, wine, livestock, livestock products, etc.

FLORIANA. AI._URO Appendix 2 List of Asia and Pacific Countries Included in the Analysis 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26.

Afghanistan Australia Bangladesh Burma Cambodia China Fiji Hongkong India Indonesia Iran Japan Korea Laos Malaysia Nepal New Zealand Pakistan Papua New Guinea Philippines Singapore Sri Lanka Taiwan Thailand U.S.A. Vietnam

AGRICULTURE

AND

NONAGRICULTURAL

EMPLOYMENT

Appendix 3 Data Tabulation Asia and Pacific Countries The data used for the analysis in the text axe presented in this Appendix with Appendix 1 providing the complete definitions and measures of the variables. There are some arbitrary points in the regional classification. For example, in the East Asia group of countries, Fiji and PapuaNew Guinea are included although they are not conventionally done so. The Developed Pacific countries do not include Japan. All are for the year 1975. SOUTHEAST

ASIA

Indonesia Malaysia Ph_ppiaes Singapore Thailand Burma Cambodia Laos Vietnam

NAGRE

PNAGRE

PDYRY

PAGGDP

INAGRC

PCFDPR

17,574 1,979 7,489 830 4,280 5,726 764 381 5,367

.374 .483 .504 .973 .223 .445 .239 .236 .264

2,629 2,625 1,721 1,825 1,816 1,429 1,338 2,260

.348 .277 .287 .019 .311 .360 ,368

63 39 79 75 90 97 88

113 153 108 221 113 94 39 121 111

1,330 3,780 80,348 5,072 410 8,608 2,167

.202 .149 .334 .578 .068 .438 .457

2,071 1,853 1,858 3,102 2,074 2,296 2,271

.570 .478 .091 .308 .361

70 91 39 73 59 54 50

95 94 101 117 94 114 88

142,175 1,810 48,972 7,015 101 218

,361 .967 .852 .553 .561 .158

3,507 1,792 6,187 5,324 2,255 1,740

.013 .054 .251 .140 .187 .259

34 48 68 -

113 54 114 112 81 108

5,369 1,093 91,872

.932 .895 .972

5,135 5,109

.064 .044

70 -

116 104 121

SOUTH ASIA

Afghanistan Bangladesh India Iran Nepal Pakistan Sri Lanka EAST ASIA China Hongkong Japan Korea Talwan Fiji Papua New Guinea DEVELOPED PACIFIC Australia New Zealand U.S.A. Source:Seetext.

FLORIAN A. ALBURO

REFERENCES Alburo, F. A. (1979), "Comparative Agricultural Modernization and Non,Farm Economic Activities," Journal of Philippine Development 6:I02-116. .... . 0980), "Comparative Agricultural Modernization and Non-Farm Employment," Discussion Paper No. 80-12, U.P. Institute of Economic Development and Research (UPIEDR). Anderson, D. and Leiserson, M. W. (1978), Rural Enterpr_es and Non-Farm Employment, World Bank. Chuta, E. and Liedholm, C. (1979), "Rural Non-Farm Employment: A Review of the State of the Art," MSU Rural Development Paper No. 4, Department of Agricultural Economics, Michigan State University. Gibb, A. (1971), "A Note: Defining the Non-Farm Employment Question," Discussion Paper No. 71-41, U.P. Institute of Economic Development and Research (UPIEDR). .... . (1972), "Report on On-Going Research: Some Evidence on the Impact of Agricultural Modernization on Non-Ag_'icultural Incomes in Agricultural Market Center," Discussion Paper No. 72-4, UPIEDR. .... . (1972), "Some Evidence on Agriculturally-based Cities," Discussion Paper No. 72-17, UPIEDR. .... . (1972). "Preliminary Data on Non-Farm Employment Changes in Agricultural Sub-Region," Discussion Paper No. 72-19, UPIEDR. Hymer, S. and Resnick S. (1967), "A Model of an Agraria_ Economy with Non-Agricultural Activities," American Economic Review: 493-506. National Economic and Development Authority (1977), F_ve-Year Development Plan 1978-1982, Manila. Onchan, T., Chatsombut, P., Meyer, R. L., and Mead, D. C. (1979), "Description of the Rural Off-Farm Employment Assessment Project in Thailand, Research Paper No. 1, Bangkok: Center of Applied Economics Research, Kasetsart University. (1971). Oshirna, H. T. "Labor Force: Explosion and the Labor-intensive Sector in Asian Growth," Economic Development and C'_ltural Change. 19 (2): 161-183. Schlegel C. C. and Ng, E. (1981), "The User's Guide to the RSI World Dataset," Technical Memorandum TM 81-1, Hawaii: Resource Systems Institute, East-West Center.

THE INFLATIONARY EFFECT OF LDC EXCHANGE RATE CHANGES UNDER GENERALIZED CURRENCY FLOATING Romeo M. Bautista*

1. Introduction A major reason for the widespread reluctance of policymakers in less developed countries (LDCs) to undertake exchange rate adjustments, despite the recognition of domestic currency overvaluation, is the anxiety about induced general price increases which may "rapidly undercut the improved competitiveness that the devaluation is designed to achieve" (Cooper, 1971, p. 21). There is a pervasive belief that devaluation is inflationary, not only in the sense of a once-and-for-all price increase but also through some cost-push mechanism (e.g., a wage price spiral) which make a higher rate of inflation inevitable. While the nature and empirical significance of such mechanisms have been debated in the literature for some time, the evidence from cross-country studies based on LDC experience under the Bretton Woods system seems to indicate that the worst fears concerning the inflationary effect of devaluation are unfounded. In a study of 24 devaluation episodes involving 19 countries, most of them developing, during the period 1959-1966, Cooper *The University of the Philippines School of Economics and Philippine Institute for Development Studies, This paper is part of a largerstudy on the comparative experience of developing countries under generalized currency floating done by the author as consultant to the Economic Analysis and Projections Department of the World Bank during 1979-80. The author is indebted to JamesRiedel for helpful discussionsand to WilliamShawfor data collection and computational as_tance. Views expressed in the paper ate personaland should not be attributed to the WorldBank. 21

ROMEOM. BAUTISTA

(1971) t'mds that increases in wholesale and consumer prices have been far less than devaluation - on average, by 32 percent and 42 percent, respectively, of the devaluation. Connolly and Taylor (1976) have likewise observed that eight devaluations in five LDCs during 1962-1970 resulted in faster increases in wholesale and consumer prices in the year following devaluation but at much lower rates than export and import prices; over two years "the devaluation would have added 13 and 14 1/2 percent to the export and import price indices" (p. 857). Finally, in a regression analysis of the inflationary impact of devaluation in ten developing countries (included in the recently completed NBER project on foreign exchange regimes and economic development) using data thro'tigh 1972, Krueger (1978) concludes that, after taking account of monetary behavior, the net effect of other influences on the rate of price increase in each of the four quarters following devaluation has not been statistically significant. The devaluations examined in the abovementioned studies took place under the Bretton Woods system of adjustable par values, involving large once-and-for-all exchange rate changes. Similar crosscountry studies on LDC experience with small, gradual exchange rate adjustments under generalized currency floating, which has been in effect since March 1973, are not yet available. 1 The "crawling peg" system (frequently involving relatively large sustained currency depreciation in nominal terms), which has gained legitimacy in the post-Bretton Woods period, represents still another form of exchange rate adjustment whose inflationary effect has yet to be compared with the other exchange rate regimes on an intercountry basis. For several reasons the response of domestic prices to exchange rate changes may be quantitatively different under different exchange rate regimes. One possibility is that producers, in their pricing decisions, may react less significantly to smaller changes over a given period in the exchange rate, in view of the transactions cost of changing prices in imperfectly competitive markets. Moreover, the publicity attending policy decisions involving large, discrete exchange rate changes may lead to greater domestic price repercussions. Some producers may even take the opportunity to raise prices of certain 1. Published research on selected developed countries (ef. Goldstein, 1977) appears to indicate mixed results concerning differences in estimated price equations using time series data including and excluding the years 1971 to 1973 which are transition periods of exchange rate regimes.

INFLATION ANDEXCHANGE RATES

products which was difficult to do before the exchange rate adjustment "for reasons of law, custom, fear of public opprobrium or simply inertia" (Cooper, 1971, p. 27);2 the large, once-and-for-all devaluation serves as an excuse for producers to raiseprices even for unrelated reasons. On the other hand, continuous exchange rate adjustments, whether large or small, may generate expectations of sustained movements in the same direction, which add to inflationary pressure. The so-called "ratchet effect", due to a possible asymmetry in the effects of positive versus negative cost changes induced by fluctuating exchange rates on domestic prices, also needs to be considered. With increased variability in an LDC's effective exchange rate under generalized floating, downward rigidity of domestic prices could result in a cumulative rise in the general price level. While each of the above hypotheses is intuitively plausible, their practical significance remains an open question. In this paper, we attempt to relate observed movements in the effective exchange rate, measured as a trade-weighted averageof the domestic currency's exchange rates with foreign currencies, to changes in the general price level in a sample of 22 developing countries during 1973-1979. This is done with due consideration of other possible influences on the inflation rate based on previous contributions to the empirical literature. Section 2 describes various inflation models which have appeared prominently in the literature, specifying the variables that may contribute to the explanation of observed differences in intercountry inflation rates. Certain aspects of exchange rate changes are also discussed, laying the groundwork for the subsequent estimation of their influence on the domestic price level In Section 3 the collective experience of 22 sample LDCs is examined, utilizing cross-country data expressed in average annual levels or rates of change over the period 1973-1979. One advantage of using this type of cross-section data over time series analysis is that dynamic considerations (e.g., adjustment lags) and short-term disturbances can be ignored; averaging of values over a span of six years serves to "smooth out' fluctuations from these sources, so that comparison across countries provides a meaningful basis for estimating at least the intermediate-term response of domestic prices to 2. Primeexamplesare publicutility companies.andotherfirmswhose productsaresubjectto government pricecontrol.

ROMEO M. BAUTISTA

exchange rate changes and other influences. It also affords an opportunity to discern differences in the estimated effects under different exchange rate regimes. The final section discusses certain implications of the fmdings for LDC policymaking and gives some concluding remarks. 2. Alternative Specifications of the Price Equation While there exists a wide variety of empirical models explaining domestic price behavior, we are interested here only in those that have immediate relevance to LDCs and include the exchange rate (explicitly or implicitly) as an explanatory variable. Moreover, data limitations require that our model specification be kept simple; one would be ill-advised to insist on analytical sophistication given the data constraints in developing countries. It is necessary however to consider various models of inflation in open economies as basis for the empirical tests of the effect of exchange rate changes on the general price level The objective is to reduce the possibility that the test results are peculiar to only one set of assumptions concerning the determinants of inflation in developing countries. We start with the following variant of a fairly standard cost-push type specification of the price equation: (1) P = ao + al (t' + a2 Prn + a3 _r, al ,a2 > O,a3 < 0 where a6 I_

---

/_.m " Y --

general price level money wage rate import price, in domestic currency real output

and the dot (.)above each variable denoting a proport_ionat_e - change. It is hypothesized in eq. (1) that the change in domestic prices bears a linear relationship to the changes in unit labor cost and import price, the latter embodying exchange rate changes. Because information on man-hours worked is generally not available, output level is used instead of labor productivity which is the u_sual

INFLATIONAND EXCHANGERATES

explanatory variable in this type of inflation model (cf. Ball and Duffy, 1972). The variable Y is expected to have a negative coefficient. This would follow from a constant mark-up assumption, since an increase in real output, other things constant, will lower unit costs. Alternatively, where the mark-up is positively related to the level of excess demand in the product market (or any other measure of market tightness), a rise in output supply for the same amount of effective demand will reduce excess demand, which is disinflationary. Imports are viewed in the above formulation "more as a factor of production than as a finished good that competes with domestically produced finished goods" (Goldstein, 1977, p. 579), which seems a realistic assumption in the context of developing countries. It is possible that the money wage rate depends in part on the general price level, in which case, there will be a simultaneity problem in the estimation of eq. (1). Indeed, the wage-price spiral is widely assumed to provide the mechanism, or in the "structuralist" terminology (cf. Argy, 1970, p. 80), the propagating element that sustains a higher rate of inflation once a price increase, from any source, has been initiated. We hypothesize the following relationship: (2)

IÂ˘ = be + bt P + b2 _r + bsL,

bl, b2, b 3 > 0

where L is a shift variable representing the level of economic development (measutved by per Capita GDP). Other things being equal, an increase in domestic prices or real outputâ&#x20AC;˘ is assumed to result in a rise in nominal wages. This might be justified even in labor surplus LDCs on the grounds that employers will find it easier to comply with minimum wage requirements, or accede to union demands for pay increases, when product prices and volumes are rising. The coefficient b3 is expected to be positive, on the assumption that the pressure t_ raise wages (or the bargaining strength of labor unions) is greater in higher income countries..If higher income LDCs have higher levels of labor productivity (which seems reasonable), then b_ greater than zero would be consistent with Glytsos' (1977) hypothesis, apparently borne out by his empirical analysis using data for nine LDCs in the 1960s, that the "rate of change of wages depends (positively) on the level of productivity and not on the rate of change of productivity" (p. 318). Substituting (2) into (1) yields: (3)

i _ = ao' + al'Pm

+ a2' t" + a_ L,

al' a_ > O, a2' _

ROMEO M.BAUTISTA

where a° + al b° ao - 1 - a1 b 1 '

a2 , a_ = 1 - a x b 1 al b3

al b2 + a3 -" 1 - aI b I '

i|l

1 - a1 b1 The above discussion suggests two ways of estimating the price equation which could be the basis for an investigation of the inflationary effect of exchange rate .changes (which are embodied in l_n). First, assuming exogeneity of W (i.e., it is independent of P and _), one can use eq. (1); and second, assuming that eq. (2)holds, one can use the reduced form eq. (3). Another approach to the empirical analysis of inflation in developing countries is via the relationship between money supply and the general price level. Justification for the use of money supply as an explanatory variable in the price equation rests on the quantity theory of money. One advantage of using this approach is that it avoids the structural representation of various influences on the general price level which, it has been argued, no simple model can adequately provide in a cross-country study (cf. Krueger, 1978, p. 180). Furthermore, it recognizes that for any cost-push type disturbance to be a source of continuing inflation, there has to be an accompanying stimulus to aggregate demand, as would be provided by an increasing money supply. While studies of cost-push •inflation tend to ignore the role of money supply (e.g., Goldstein, 1977 and Kwack, 1977), it is clear that an accommodating monetary policy serves to validate the higher prices and sustain wage-price spirals. Finally, monetary theories of inflation also emphasize the tendency for government expenditures in developing countries to be financed notby higher taxes or bond sales to the public but by money creation. The increased spending thus tends to stimulate faster money supply growth which results in accelerated inflation. Here we follow Cooper (1971) and, to some extent, Argy (1970) and Glytsos (1977), in expressing the change in the general price level as: 3 3. Instead of changes in real output (4/) and in importprices(i_m),Cooper made use of variations in food production and devaluationrate, respectively,as explanatory variablesin eq. (4). Argy, and Glytsosin his demandpull submodel, did not include Pm in their regressionspecifications.

INFLATIONAND EXCHANGERATES

(4)

/_=co

+ cl._l

+ c2 ir + c3 i'm,

c,,

c 3 > 0,

c2 <0

which allows for the fact that increases in money supply are not inflationary to the extent that they are absorbed by changes in real output. As shown by Aghevli and Rodriguez (1979), a price equation similar to (4) can be derived using a two-sector open economy model in which home goods prices adjust in part to the excess supply of money while the price in the import-competing sector is determined directly by the world price and current exchange rate. As argued some time ago by Harberger (1963) in his study of Chilean inflation, autonomous wage increases could lead in the short run to price rises even in the absence of money supply expansion, adding to the explanation given by monetary movements in the rate of inflation. Of course, "if the monetary authorities always increased the money supply by enough (or more than enough) to assure full employment at the new wage level, there would be no strong reason to expect that wages would add significantly to a monetary explanation of inflation" (p. 228). It seems worthwhile in the present study to run regressions on a specification with the wage variable included in the R.H.S. of (4), i.e.,

Similar formulations have also been used by Diaz-Alejandro (1965) and Diz (1970) for Argentina, and Colaco (1969) for Brazil and India. A final specification of the price equation deserving some attention concerns the use of l_I- Y as a measure of excess domestic demand, which "captures the excess liquidity created by the government fiscal deficit as well as changes in the demand for money caused by income growth" (Blejer and Halevi, 1980, p. 119). Equation (5) might then be simply modified as:

(6) l; = Co+ c1

+ c3I'm +

Focusing now on the import price variable, influence on the domestic price level need to the effect of Changes in Pm may vary from depending on the extent of reliance on imported allowance for this by multiplying the observed

several aspects of its be recognized. First, country to country, goods. We.can make values of Pm by the

ROMEOMIBAUTISTA

factor (1 + m) where m is the share of imports to GNP representing a fairly standard measure of the "degree of openness". More highly open LDCs might be expected to exhibit a greater response in domestic prices to a change in import price, based on a cost-push view of inflation. 4 Second, iJ,n represents the combined changes in foreign currency import prices, in tariff, indirect tax and other policies affecting the domestic price of imported goods, and under generalized CUrrency •floating in bilateral exchange rates with trade partner currencies. We do not take into account here changes in domestic policies rePated to the so-called "non-formal" component of the exchange rate which, given the complexity of the trade and payments regime in most developing countries, would be exceedingly difficult to make comparable across countries and over time. There is a presumption however that "formal" exchange rate changes, being more widely publicized and having a greater impact on expectations, have a much stronger and faster effect on the rate of domestic inflation (Blejer and Halevi, 1980). We need then an index of import prices in currencies at which they are set and an effective exchange rate (EER) index• reflecting "what is in some sense the •average change in the country's exchange rate" (Rhomberg, 1976, p. 88). Import price and EER indices which we have calculated in an earlier work (Bautista, 1980), are used here based on the "price-currency import shares" of twelve major currencies, assuming that world prices of primary agricultural products and fuels are set in U.S. dollars, metals and minerals in U.K. pounds, and other traded goods in the currencies of the developed country origin of imports. 5 4. There is a competing hypothesis that views openness as a reflectionof an LDC's capacity to draw from foreign sources in times of reduced domestic availability for whatever reason (e.g., poor crop harvest), which implies a negative relationship between the degree of openness and inflation rate. Additionally, more open economies are faced with a higher degree of foreign competition, and hence are less subject to inflationary •pressures arising from monopolistic behavior of domestic producers. It would be appropriate therefore to include m also as an intercept dummy whose coefficient is expected to be negative. 5. See Bautista and Riedel (forthcoming) for a detailed discussion. This weighting scheme in EER ealculatinn differs from the traditional use of bilateral trade shares which does not allow for pricing of traded goods in third-country currencies.

INFLATIONAND EXCHANGERATES

A third consideration, particularly warranted in the assessment of the inflationary effect of exchange rate changes, is the possible nonlinearity of the domestic price response. As indicated above, large EER changes may or may not have a more significant impact on domestic prices than small-step exchange rate adjustments, in part depending on what expectations on future exchange rates and costs are generated as well as on the relative cost of frequent price adjustments. This can be dealt with by including the square of the effective exchange rate change (EER 2) as an additional explanatory variable in the regressions. Bruno and Sussman (1978), for example, have reported a positive coefficient for this variable, "although of not very high degree of significance" (p. 23), in an examination of inflation and devaluation in Israel based on quarterly data for the period 1967-75. Fourth, one would also want to test the ratchet effect associated with exchange rate fluctuations under generalized floating which, as pointed out earlier, generates an inflationary bias duo to domestic price inflexibility in the downward direction. 6 An extension of this hypothesis would bc that upward and downward deviations of the effective exchange rate from expected or trend values have asymmetrical effects on the domestic price level, so that countries faced with greater EER variability around the trend would tend to be more inflationary, other things the same. The explanatory variable used here to capture the ratchet effect of EER variability consists of the standard error of estimate in the logarithmic trend regression of monthly import-weighted EER values for each sample LDC over the period March 1973 to December 1979. A final consideration concerns the differences among sample LDCs in the type of exchange rate adjustment adopted during the observation period. Some of the characteristics of exchange rate regimes would ha4e been captured already in the other explanatory variables mentioned earlier. For instance, if the crisis atmosphere in which large, discrete devaluations take place makes it easier for the national authorities to adopt contractionary monetary policy and to rally labor groups behind a more restrained wage increase than would be the case with small, gradual exchange rate changes, the 6. Inconclusive results have been obtained in a recent systematic in_sl_Igation of the practical significance of ratchet effects, which however pertainonly to some selectedmajorindustrialcountries(Goldstein, 1977).

ROMEOM. BAUTISTA

estimated coefficients for the money supply and wage rate variables will already reflect the corresponding effects on the domestic price level. In introducing dummy variables for particular exchange rate regimes in the regression equation, one is representing other aspects of exchange rate adjustment and attendant policy measures having a separate influence on the inflation rate. Thus, if the adoption of a crawling peg meant, in an ex post sense, the relaxation of import restrictions to a greater extent than what accompanied other exchange rate regimes, the dummy coefficients allowing the intercept of the price equation to shift might turn out significant. It is of course hazardous to speculate on the various relevant aspects of economic policy, which may have offsetting influences on the domestic price level, accompanying exchange rate adjustments and to assess their separate inflationary effects in the sample countries. However, we can attempt to generalize from the collective experience of the sample LDCs by examining the significance of the dummy variables that would reflect the net effect on the domestic price level of other policy measures attributable, at least indirectly, to the type of exchange rate regime being adopted. Additionally, it is instructive to run regressions excluding the crawling peg countries (Argentina, Brazil, Colombia and Uruguay) and to compare the estimated coefficients with the corresponding results of the regressions based on the entire sample.

3. Empirical Results The results of cross-country regressions testing for the foreign inflation rate (Pf), wage rate changes and level of economic development as determinants of domestic inflation are presented in Tables 1 and 2 for sample LDCs including and excluding the crawling peg countries, respectively. 7 Other explanatory variables appearing in the various specifications are changes in money supply, real GDP and effective exchange rate (l_.).s

7. Indonesia and Thailand,which are part of our sample of 22 countries, do not have availablewage series and hence have not been included in these regressions. 8. See AppendixA for a furtherdescriptionof the variables(and sources of data) appearingin these and other regressionresultsreported in this paper. AppendixB givesdataobservationsnot drawnfrompublishedsources.

Table I - Regression (Dependent

Equation

I II llI

Const M-V

17.94 .281 (1.32) • (1.55) 15.90 .310 (1.27) (1.79) 5.282** .360* (2.98) (2.39)

.013 (.10) --.033 (-.28) --

--.067 (-.66) -

.9944

3.09

.9946

3.03

n_ ;_

.9950

2.92

-.728 (1.90)

.928** (5.26)

-1.963 (-1.40)

.003 (.02)

-.132 (-1.19)

.9946

3.01

-.589

.931"*

- 1.524

.9945

3.06

(1.59) -.507

(5.20) .836**

.9947

2.98

(-1.49) -.528 (-1.42) -.274 (-.78) -.082 (-.25)

(5.56) 1.137"* (10.93) 1.204"* (12.14) 1.190"* (46.88)

.9942

3.12

.9937

3.27

_e _" ,.-! t_

.9933

3.36

21.91

.311

(1.52) 6.088*

-- 1.160 (-.90) -1.091 (-.86) --

VIII

.912"* (5.05) .918"* (5.19) .848** (5.83)

.256 (1.44)

(2.47) 37.30 * (2.61) 31.29 * (2.11) 9.668** (4.38)

VII

28.15 (1.86)

(1.78) .367* (2.38)

variable: P; 20 observations)

Results A

(-1.11)

-2.664 (-1.96) -2.259 (-1.61) -

-.066 (-.53)

.047 (.34) -.037 (-.28) -

-.173 (-1.56)

• Significant at 5 percent level • *Significant at 1 peroent level Numbers in parentheses underneath the coefficient estimates ate t-values. ho

Table2 (Dependent

Equation Numb I

Const. M.320 (1.24) .367 (1.59)

IlI

4.529• (1.93)

.395" (2.18)

27.I0 (1.24) 20.40 (1.02) 4.965 (1.58 ) 31.78 (1.42) 22.29 (1.03) 9.016"* (3.23)

V VI VII VII! IX

Variable:

i'S .919"* (3.98) .926"* (4.16) .898** (4.67)

.327 (1.26) .398 (1.65) .401

-.818 (-1.38) -.695 (-1.23) -.475

.859"* (3.56) .893 * * (3.81) .879**

(2.11) -

(-1.18) -.474 (-.88) -.145 (-.30) .022 (.06)

(4.04) .983"* (4.33) 1.084"* (4.95) 1.195"* (6.73)

Results B

P; 16 observations)

'/'

11.80 (.82) 11.39 (.82)

-- Regression

•Sit,niflcant at 5 percent leveL • *Significant at I percent level Numbers in parentheses underneath the coefficient estimates are t-values.

-.707 (-.48) -.708 (-.50) -1.912 (-.98) - 1.444 (-.782) .... -2.132 (-i.06) -1.433 (-.72) -

L (

.035 .18 ) --.020 (-.13)

-.056 (-.48) -

--.017 (-.08) -.083 (-.46)

.112 (.63) .047 (.27) -

--.110 (-.840) -

-.164 (-1.29) -

_2 .7986

3.28

.8126

3.16

.8377

2.94

.7962

3.30

.8022

3.25

.8251

3.05

.7843

3.39

.7714

3.49

.7787

3.43

INFLATIONAND EXCHANGERATES

Given the consistently low t-values and sometimes perverse signs of the estimated coefficients for PL Wand L in the relevant specifications, these variables were not included in subsequent regressions. The lack of significance of the foreign price variable was not entirely surprising in view of the observed similarity in Pf trends among the sample LDCs having the same principal trade par[nets. In the Case of the wage variable, the statistical insignificance of the estimated coefficient might have been due in part to the poor quality and low degree of comparability of wage data among developing countries. Note however that dropping W is consistent with equation (2) above, which postulates simultaneity in the determination of 1) and W. Moreover, to the extent that wage increases had been accommodated by monetary expansion, there would be no additional explanation of domestic inflation that is provided by _V. The regression results in Tables 1 and 2 also indicate that _" appearing in the various specifications without M yields insignificant coefficient estimates. Furthermore, use of I(/I- _" is clearly superior, based on the comparative values of the t-statistics, to having I_1and as separate explanatory variables. Another striking observation from Tables 1 and 2 is the consistently significant coefficient estimates for the exchange rate variable in either set of regressions. Based on the specification having only 1_I- _" and I_ as explanatory variables, the estimated elasticity of the domestic price level with respect to the exchange rate is .848 for the sample of 20 LDCs and .898 excluding the crawling peg countries which, using the t-test, are not significantly different at the 5 percent level. Table 3 contains the results of regressions including the degree of openness as an additional explanatory variable, based on the entire sample of 22 LI)Cs and on a smaller sample excluding the four crawling peg countries. The coefficient estimates for mE, which show contrasting signs in the two sets of regressions, are found to be no significantly different from zero. The coefficient of the intercept dummy, while having the expected sign, is also seen to be insignificant in every specification. The estimated elasticity of the domestic price level with respect to the exchange rate is .776 for the entire sample and .846 for the subsample without the crawling peg countries. Again, the difference between these coefficient estimates is not statistically significant.

ROMEO

Table

3 -

M. BAUTISTA

Reweuion

(Dependent

Regalia

variable:

i _)

Equation

No.

Con,t. M._?

s.,.o.

22 Ob#ervations I

5.670* (2.89)

,432'* (2.96)

.804** (5.72)

-.390 (-.55)

-.019 (-.77)

.9939

3.08

4.911"* (2.89)

,439"* (2.99)

.786"* (5.74)

-,275 (-.40)

,9940

3.04

IlI

5.441"* (2.89)

,417"* (2,99)

.787** (5.85)

-.016 (-.68)

.9941

3.02

4.828** (2.96)

.432** (3.18)

.776** (5.89)

.9943

2.97

18 observations (excluding crawling peg countries) V

4.541 (1.71)

,54" (2.66)

.818" (2.76)

.132 (1.38)

-.010 (-.38)

.8147

3.18

3.941 (1.90)

.472** (2,98)

.815" (2.84)

.95 (.15)

.8260

3.08

VII

4.578 (I.80)

.453" (2.77)

.839"* (4.37)

.8278

3.06

VIII

3.969 (1.99)

.471"* (3.08)

.846** (4.55)

.8373

2.98

-.011 (-.41) -

*Siglxificantat 5 percent leveL **8i/Â˘tiflcant at 1 percentlevel Numbers in parenthesesunderneath the coefficitmtÂ˘stlmM_l_e t-values.

In Table 4 regression results, testing for the significance of exchange rate variability and nonlinear price effects, are presented. Consistently low values of the t-statistic are shown for the estimated coefficients of V in the various specifications, indicating that exchange rate variability does not add to the explanation of intercountry differences in inflation rates. In the case of I__, insignificant coefficient estimates are also observed in the regression equations having ]_ as another explanatory variable; however, where E is not included in the specification (equations III and V for the entire sample, VIII and X for the subsample excluding the crawling peg countries), there is apparently a significant influence of 1_2. Even so, having 1_ instead of E2 seems superior, judging by the usual statistical criteria (higher t-statistic and _2 and lower s.e.e.). No conclusive evidence is therefore provided with respect to the differential effects of large versus small exchange rate changes on the domestic price level.

INFLATION

]bible

AND

4 -- Rest_eeaion

(Dependent

Equat/on

No.

EXCHANGE

Results

variable:

Contr. M-,_

RATES

.OlE2

,.e.o.

22 obser_tions I

5.046* (2.60)

.434"* (3.05)

.728** (4.39)

.030 (.57)

.814 (.05)

.9937

3.12

5.085** (2.96)

.434"* (3,14)

,730"* (4.69)

.030 (.58)

.9941

3.03

IIl

-.310 (-.15)

,954** (8.55)

.146" (2.21)

20.33 (.88)

.9874

4.42

4.797* (2,59)

.432** (3.10)

.774** (5.47)

.651 (.04)

.9940

3.06

.9875

4.39

.8217

3.12

.8288

3.05

2,797 (.12)

.7871

3.40

V 18 observations

.306 ,993"* (.15) (9,74) (excluding crawling peg countries).

.154" (2.37)

4.355 (2.03)

,438* (2,43)

1.826 (1.93)

-2.301 (-.63)

VII

4.060 (1.97)

,431" (2,45)

1.142 (1.85)

-1.775 (--.51)

VIII

3.288 (1.45)

,624"* (3,76)

4.330** (3.08)

4.182 (2.01)

,486** (3,06)

.922"* (3.91)

12.45 (-.55)

.8294

3.05

3,320 (1.53)

,631"* (4,22)

4.414"* (3.76)

.8011

3.29

*$isaJfi_tat at 5 percentlevel **Sl_dficant at 1 pero_ntleveL Numher_impat_nthemsunderneaththe coefficient est/mates avet-vatues.

-15.66 (-.66)

â&#x20AC;˘ROMEOM. BAUTISTA

The coefficient estimates for lVl - _' and 1_are significant at the 5 percent level, but not those for the exchange rate regime dummy variables. Thus, one may accept the hypothesis that the method of exchange rate adjustment adopted in the sample countriesdoes not shift the price equation in the sense indicated above. 9 It is also worth noting that the estimated coefficient for:/_ (.795) is within the range of the corresponding estimates presented earlier. Given the limited sample size and data imperfections, the relationship between the exchange rate and domestic price level as represented in the above regression results appears quite robust. Moreover, the high values of the coefficient of determination are suggestive of the general validity of the hypotheses tested. The implication from the above results, that on average, the general price level rises by about 80 percent of the rate of currency depreciation, is at variance with the findings in earlier studies cited above. While such magnitude of the inflationary effect does not completely nullify the improvement in international competitiveness that a devaluation is designed to confer on the country's products, it represents a much more severe repercussion on domestic prices than indicated in past empirical investigations. Part of the reason would be that longer term price effects are captured in the present stu.dy by using average levels and rates of change over a six-year period. The results from other empirical studies using cross-country data pertaining to a shorter time span would reflect to some extent dynamic adjustments and short-run disturbances which render more difficult the taking into account of the cumulative effect of exchange rate changes on the domestic price level. Hence, the long-run inflationary effect might tend to be underestimated if one relied simply on estimates based on cross-country data over one, two or three years. Another explanation for the higher price response coefficient estimated in this study would be that the exposure of developing countries, or at least those included in our sample, to the international economy had presumably become stronger over the years. The actual extent of such exposure is not measurable simply by the weight of imports in the final expenditure basket - our "degree of openness" variable, whose coefficient did not prove 9. Blejer and Halevi (1980) have found that the inflationary effects of exchangerate changesin Israelduringthe period 1968-1976 are not sensitiveto the change in exchange rate regime, using also intercept and slope dummy variablesto distinguishthe crawlingpegsubperiodbeginningmid-1975.

INFLATIONAND EXCHANGERATES

significant in the above regressions - but "of the relevant group of tradables, that is, including close domestically produced import substitutes" (Bruno, 1978, p. 379). A final consideration is that the dramatic developments in the world economy during the 1970s might have altered the economic structure underlying the price equation that prevailed under the Bretton Woods regime. Price expectations, whose influence on domestic price behavior has been virtually left out of account in the present study as well as in most other empirical investigations of the determinants of LDC inflation rates, would likely have reinforced the inflationary effect of exchange rate changes in the recent period compared to the calmer decades of the 1950s and 1960s. The adoption of generalized currency floating among developed countries would have contributed to the widespread breakdown of money illusion, particularly "exchange rate illusion," in developing countries in view of the greater attention now being given to the real international value of an LDC's currency. Interpreting the E coefficient in the above regressions as a long-run elasticity measure, the resulting estimates of around .80 would not seem unreasonable. They are evidently much higher than the findings of empirical studies pertaining to postwar LDC experience through the early 1970s, which however are not comparable in many ways. It is worth noting that, in a recent investigation of the influence of exchange rate changes on prices in eighteen developed countries, "the main conclusion.., is that parity changes have a larger and a quicker effect on the rate of inflation than they used to have" (Robinson et. al., 1979, p. 48). A related finding is obtained in an analysis of inflation rates during 1972-76 in sixteen developed countries, "pointing to a very marked direct role of import costs in the inflationary process in these countries"; moreover, "even for a country as 'closed' as the United States empirical studies point to a large and significant effect of import prices in recent years" (Bruno, 1978, p. 389). 4. Conclusion The f'mdings of this empirical study indicate that changes in the effective exchange rate and excess domestic demand can explain to a very large extent inter-LDC differences in inflation rates for the period 1973-1979. The sensitivity of the domestic price level to exchange

ROMEOM. BAUTISTA

rate movements is seen not to differ significantly among countries adopting different methods of exchange rate adjustment. Tests of the significance of the inflationary effects of the "d'egree of openness", level of economic development and exchange rate variability, and of nonlinear price effects, have also yielded inconclusive results. The observed robustness of the coefficient estimates for the effective exchange rate suggests the stability of the exchange rate-domestic price level relationship. In contrast to the findings of earlier studies on the price effects of devaluations in fhe 1950s and 1960s, it would appear that the price response to exchange rate adjustment in the post-Bretton Woods period is of a very high order of practical importance to LDC policymakers, since about 80 per cent of exchange rate changes is estimated to have been added to the domestic price level. This would seem to provide justification for the caution with which governments in developing countries tend to view currency depreciation, at least under the conditions prevailing in the 1970s. The implication for LDC policymaking seems clear. The cost of achieving a given "real" devaluation, in terms of a higher domestic inflation rate, is now greater than what it used to be, reducing the attractiveness of exchange rate adjustment as a means of improving international competitiveness (and hence allocative efficiency, external balance and real income, given the high degree of market distortion in most developing countries). What other policy options are available to LDC policymakers and how would they affect the various objectives of development policy? This and other related policy issues deserve careful study in the context of individual LDCs. Appendix A Definition of Variables and Data Sources P

M Y m W E Pf

= -= = =

wholesale price index or, if complete series is not available â&#x20AC;˘ (Malaysia, Morocco and Singapore), consumer price index money supply gross domestic product (at 1975 prices) ratio of imports to gross national product non-agricultural (industrial, manufacturing) wage rate import-weighted effective exchange rate index import-weighted index of foreign wholesale prices

INFLATIONAND EXCHANGERATES

V ffi standard error of estimate in the logarithmic trend regression of monthly E values from March 1973 to December 1979 DC ffi dummy variable: 1 for crawling peg countries, 0 for other sample countries DS = dummy variable: 1 for countries with small, gradual exchange rato changes, 0 for other sample countries L = per capita gross domestic product, (in hundred U.S. dollars, 1975 prices) Data for P, M,Y and m were drawn entirely from the IFS computer file of the IMF. The ILO Yearbook of Labor Statistics provided most of the wage data, which were supplemented by published materials from individual countries. The average annual _ercentage change in E for each sample LDC was calculated as (1+I_*) 12 - 11 t x 100, where 1_* is the coefficient estimate in the logarithmic trend regression of monthly values of E drawn from an earlier work of the author (cf. Bautista, 1980). A similar transformation was done on the monthly rates of change in Pf, defined as

Pr=

I_WxiR_l/P II

iR/P

where wxt is the "price currency export share" of developed country t, R t is bilateral exchange rate and Pi is the wholesale price index in country i. As indicated in the text, the measure of EER variability V is represented by the standard error of estimate in the logarithmic trend regression of monthly E values from March 1973 to December 1979.

ROMEO M. BAUTISTA

Appendix B Vahtes of 1_,V and Pf Country

l_/-

Argentina Brazil Colombia Costa Rica Egypt Greece India indonesia Malaysia Mexico Morocco Pakistan Philippines Portugal Singapore South Korea

153.25 29.31 10.86 4.14 4.93 5.59 1.43 6.56 -0.22 14.72 0.36 0.70 3.04 14.8 -0.44 5.16

.4251 .1022 .0269 .0647 .1620 .0297 .0239 .1390 .0267 .1447 .0206 .0330 .0268 .0891 .0235 " .0472

8.405 9,413 8.649 8.462 8.091 8.534 10.086 8.445 10.069 9.332 9.278 8.888 9.544 8.723 8.618 7.711

.0644 .0335 .0492 .0255 .1500 .1131

8.578 8.198 8.722 8.131 8.280 8.291

Spain Taiwan Thailand Tunisia Turkey Uruguay

4.42 1.06 1.69 -0.90 15.12 45.87

Sottrce: Basicdata drawnfrom Bautista (1980).

REFERENCES Aghevli, B. B. and Rodriguez, C. A., (1979), "Trade, Prices and Outpu t in Japan: A Simple Monetary Model," IMF Staff Papers, 26: 38-54. Argy, V., (1970), "Structural Inflation in Developing Countries," Oxford EconomicPapers, 22: 73-85. Ball, R. J. and Duffy, M., (1972), "Price Formation in European Countries,'; in The Econometrics of Price Determination, ed. O Eckstein, Washington D. C., pp. 347-368. Bautista, R. M. (1980), "Exchange Rate Adjustment Under Generalized Currency Floating: Comparative Analysis Among Developing Countries," World Bank Staff Working Paper No. 436, Washington, D.C. Bautista, R. M. and Riedel, J. (forthcoming), "M_or Currency Realignments and the Terms of Trade in Developing Countries," Journal of Developmen t Econ om ics.

INFLATION

AND EXCHANGE

RATES

Blejer, M. I. and Halevi, N. (1980), "Effective Devaluation and Domestic Rate of Inflation," Journal of Development Economics, 7:117-122. Bruno, M., (1978), 'Exchange Rates, Import Costs and Wage-Price Dynamics," Journal of Political Economy, 86: 379-403. Bruno, M. and Sussman, Z. (1978) "Exchange Rate Flexibility, Inflation and Structural Change: Israel Under Alternative RegimeS," paper presented at the Ford Foundation/Central Bank of Barbados seminar on Developing Countries and the International FInancial System, January 12-14. Colaco, F. X., (1969), "Harberger's Inflation Model: A Critique and Test using Data for Brazil and India," Indian Economic Journal, 16: 434-444. â&#x20AC;˘ Connoily, M. and Taylor, D. (1976), "Testing the Monetary Approach to Devaluation in Developing Countries," Journal of Political Economy, 84: 849-859. Cooper, R. N., (1971), "An Assessment of Currency Devaluation in Developing Countries," in Government and Economic Development, ed. G. Ranis, New-Haven: Yale University Press, pp. 472-515. Diaz-Alejandro, C. F. (1965), Exchange-Rate Devaluation in a Semi-Industrial Country: The Experience of Argentina 1955-1961, Cambridge: The MIT Press. Diz, A. C. (1970), "Money and Prices in Argentina, 1935-1962," in Varieffes of Monetary Experience, ed. D. Meiselman, Chicago: University of Chicago Press, pp. 111-122. Glytsos, N. P. (1977), "Determinants of Wage and Price Changes in Less DÂ˘_ veloped Countries," Journal of Development Economics, 4: 315-342. Goldstei_ M. (1977), "Downward Price Inflexibility, Ratchet Effects, and the Inflationary Impact of Import Price Changes: Some Empirical Tests," IMF Staff Papers, 24: 569-612. Harbetger, A. (1963), "The Dynamics of Inflation in Chile," in Measurement in Economics, ed. C. F. Christ, Stanford: Stanford University Press, pp. 219-254. Kxueger, A. O. (1978), Foreign Trade Regimes and Economic Development: Liberalization Attempts and Consequences, New York: Columbia University Press. Kwack, S. Y. (1977), "The Effect of Foreign Inflation on Domestic Prices and the Relative Price Advantage of Exchange Rate Changes," in The Effec_ of Exchange Rate Adjustments, ed. P. B. Clark, D. E. Logue and R. J. Sweeney, Washington, D.C.: U.S. Department of the Treasury, pp. 387-398. Rhomberg, R. R., (1976), "Indices of Effective Exchange Rates," IMF Staff Papers, 23: 88-I12. Robinson, W., Webb, T. R., and Townsend, M. A., (1978), "The Influence of Exchange Rate Changes on Prices: A Study of 18 Industrial Countries," Economica, 26: 27-50.

AN EVALUATION OF TIME BUDGET STUDIES AS COMPLEMENTS TO CONVENTIONAL LABOR FORCE SURVEYS Bryan L. Boulier*

1. Introduction Labor force, employment, and unemployment data are gathered for many purposes. These data are used as indicators of aggregate economic activity, as measures of labor supply or potential labor supply, and as variables explaining various kinds of economic and non-economic behavior. The current approach to measuring employment and unemployment in developing countries derives from labor force concepts used in developed countries. The applicability of these concepts to developing countries has come under sharp attack (Myrdal, 1967 and Hauser, 1974): The purpose of this paper is to examine the role of time budget studies as complements to, or substitutes for, the usual questions asked in labor force surveys. Section 2 of this paper evaluates conventional labor force questions from the perspective of economic models of household tinge allocation. The value of time budget studies for the analysis of questions for which labor force data are gathered is outlined. Section 3 discusses alternative ways of gathering time use data, and Section 4 *George WashingtonUniversity.This study uses data from the Lagunacurve households study. Initial funding for the survey came from the Agricultural Development Council and additional funding came from the Population Center Foundation of the Philippines,the InterdisciplinaryCommunicationsProgramof the Smithsonian Institution, the Rockefeller Foundation, and the School of Economicsof the Universityof the Philippines.The project workinggroupwas composed of Robert E. Evenson, Bai-ryPoplcin,CeciliaFloreneio, TeresaJayme Ho, Emiqueta Tortes, and the author. I am indebted to BarryPopkinfor much instructionin gatheringtime data, althoughhe bearsno responsibilityfor errors. 43

BRYANL. BOULIER

states the major problems in interpreting these data. Section 5 presents time allocation data gathered from households in Laguna, Philippines. Section 6 concludes. 2. Labor Force Questions and Economic Models of Time Use Conventional labor force measures are classifications of how people use their time or would like to use their time. People who supply labor "for the production of economic goods and services" (United Nations, 1967) (i.e., engage in a specified set of activities) are counted as employed. The number of hours worked is ordinarily tabulated. People who do not work are classified as either unemployed or not in the labor force. Persons are counted as unemployed if they would like to work but have been unable to find employment. Active search for a job is sometimes a requirement for a person to be considered unemployed. Individuals who work, and would like to work more hours, are sometimes classified as underemployed. Since labor force measures are classifications of time use in economic activities, it is important to consider economic models of individual and household time allocation before evaluating their usefulness in describing or analyzing behavior. The traditional model of individual labor supply behavior classifies the time of an individual into two categories-work and leisure. Work is a source of income which, when combined with non-labor income, enables an individual to purchase market produced goods. These goods and leisure yield satisfaction or utility to the individual. The amount of work an individual performs depends on his preferences for goods and leisure, non-labor income, and the wage rate he faces. An increase in the wage rate raises the value of each hour worked, leading the individual to work longer hours (a substitution of work for leisure). Simultaneously, the individual enjoys a higher income for any given number of hours worked, which may lead him to work fewer hours since leisure yields satisfaction. The net effect is a priori indeterminate. An immediate implication of this model is that classification of individuals by hours worked is not simple. For example, an increase in hours worked may indicate improved employment opportunities, or, alternatively, an increase in hours worked may be the supply response to reduced wage rates. Besides variations in hours worked data, variation in unemploy-

TIMEBUDGETSTUDIES

ment rates are used as indicators of economic activity. The unemployment category is designed to measure persons who are involuntarily out of work. Unfortunately, the unemployment rate is also difficult to evaluate without further information. For instance, the number of measured unemployed in developing countries is sometimes low because sources of non-labor income are limited, so that virtually any work, no matter how low the wage, is better than no income at all. In addition, measured unemployment may rise during periods of economic expansion, because the presence of employment opportunities paying high wages encourages persons previously not in the labor force to seek employment. They contribute to unemployment during their job search. Another important problem in the application of conventional labor force definitions is determining the set of activities to be defined as involving "the production of economic goods and services". Following the work of Becker (1965) and Lancaster (1966), economists have begun to view households as little factories, in which the time of family members, purchased goods, and household capital are used to produce "consumables" (e.g., the health status of family members) yielding utility to the family. Family time, capital, and purchased inputs can also be used to produce goods which are marketed and for which the family receives income. In the simplest case, the family exports only time to the market and receives wage income. In turn, the income earned from the sale of goods and time and non-labor income (plus borrowing) is used to purchase market goods and time of nonhousehold members or for saving. Given prices of market goods and wages, the family maximizes its utility function subject to a set of production functions, the budget constraint, time constraints for each household member, and constraints that the services of each capital good used for production be less than or equal to available capital services. In this model, all time of family members is considered to be productive-even sleep is a productive activity which uses time and capital goods and yields utility. Only a small and rather arbitrary subset of these activities is included in the list of activities considered to be supplied for the production of economic goods and services in labor force surveys. Thus, all wage earning time is ordinarily included and time spent in producing agricultural commodities is typically included regardless of whether the output is sold in the marketplace or consumed at home, while the treatment of, say, the time spent

BRYANL. BOULIER

washing clothes is included only if not for home consumption. Substantial amounts of household time spent in productive activities may be missedinconventional surveys. Inrural Nepal,forexample,a majorityof women are classified as economicallyinactive, yet a study of eightvillages by Meena Acharya (1981,Table6,p. 22) found thatwomen overage 15 worked an averageof 4.6hoursper day in conventional economicactivities (animalhusbandry,agriculture,manufacture,and income earningactivities outsidethehousehold)compared to 5.8 hours for men in the same age group.In addition, women spent2.1hoursper day on huntingandgathering, fueland watercollection, householdconstruction and foodprocessing(excluding cookingand servicing)compared to .9hoursperday by men. Clearly, conventional surveysunderstate considerably women's economicactivities, and omissionof thelatter setofactivities formen yieldsan underestimate of theirhoursofwork by about13 percent, i As developmentproceeds,changesin prices, wages,production processes, andpreferences leadto shifts inthelocation ofproduction of many goods and services from the home to th_ market place, thcrebyalteringthe proportionof time or number of persons includedin the employed categoryby laborforcesurveys. Consequently,the concentration ofthesesurveyson onlya smallsubsetof productiveactivities yieldsan inaccurate pictureof contemporary economic activityand an inadequate,and possiblymisleading, description oftrendsinactivity. One approachto measuringtimeusewhich meetstheobjectives raisedaboveistoanalyzethehousehold's useoftimeinallactivities. Variations in time use from year to year can shedlighton labor supply questionsand on shifts between market and non-market production. If,inaddition, thevalueoftimeinvariousactivities can bc measured,one canthenanalyzethegainsand losses tohouseholds withvariations inaggregatc economicactivity, aswellasthenetgains from participating in marketactivities. The remainderof thispaper concentrates on methods for gatheringtime data.Methods of measuringhouseholdoutput and valuingthe time of household members arc discussedin Boulicr(1977),Butz and Greenburg (1975),Gronau (1976),and Havrylyshyn(1976).Boulier(1977) 1. Boserup (1976) provides a good description of the problems encountered in measuringthe labor force participation of women.

TIMEBUDGETSTUDIES provides estimates for the sample of in Section 5.

Philippinehouseholds described

3. Methods of Gathering Time Data There are three major methods of gathering time data: 1) retrospective questionnaires, 2) observation, and 3)self-kept records.2 These methods are described below. In retrospective questionnaires, individuals are asked about the time devoted to various activites in a recent period of time-the last hour, the last day, etc. Combinations of time units can also be used. For instance, one might ask how many hours the individual cared for poultry today and how many days in the last week, if weekly averages are sought. The choice of time interval is not a simple manner. Presumably, the shorter the interval, the more accurate the data. For a sample of West Java village households interviewed twelve times in one year, Wigna (1980, p. 10) reports that average annual hours spent in income-producing work by men and by women estimated from interviews based on thirty-day recall periods averaged oniy about 60 to 70 percent of hours estimated from interviews based on one-day recaUs. If there is considerable variation in time use from interval to interval, the researcher faces the choice of gathering time data for longer intervals or asking the question more often to the same set of households. Tradeoffs of cost and data accuracy will undoubtedly vary from country to country. There are several other choices to be made in gathering the data. First, there is the matter of the choice of respondent. Each individual may be asked separately about his own activities, or some individuals may be asked about the time allocation of others. The latter alternative is ordinarily less reliable, except in the case of young children. That estimates of time spent in various activities depends on the respondent was shown in our Philippine survey, since it was found that husbands' reports of wives' time spent in activities involving home production (e.g., weaving) were often inconsistent with wives' reports of their own time. There was no persistent pattern of inconsistency. That is, some husbands reported more time 2. lttrdsall(1981) and S,nlni (1972) includeb_llographies of time budget _udies; Bird_ll (1981) and Stone (1972) discuss the analysis of time budget data;and Eagle .andButz(1981) discussmethodologicalissuesin collecting time budget data.

BRYANL. BOULIER

spent by their wives and others less. Inaccurate responses to questions about the time use of others is not necessarily due to lack of knowledge. White (1975) reports that Indonesian parents commonly depreciate the time contributions of children to household activities. A second important matter is the organization of time questions.â&#x20AC;˘ There are several possible procedures for gathering time data. One method is to ask an individual to recount his or her day beginning with the moment the person awakens (White, 1975). Thismethod yields information about the sequencing of activities and gives estimates which exhaust totally the twenty-four hours in a day. However, this procedure may yield inaccurate information if respondents omit activities which require little time or seem unimportant to them. Consequently, the method should probably be supplemented by probing for such activities. A second method is to organize the questionnaire by individual and then ask about a set of activities for each individual. A third method is to list activities and then ask who performed the activities and the amount of time each person spent performing the activity. We found this last method to work successfully in the Philippines. Considerable experimentation is needed to select the list of activities, so that important activities are not omitted. Our questionnaire asked women about the time they spent in performing activities occurring in and around the home. Men were asked about agricultural activities, wage earning jobs, and various kinds of market activities. If more than one respondent answers questions about time use, care must be taken to ensure that duplication of reports does not occur (i.e., women reporting aboutâ&#x20AC;˘ income producing home production and men reporting the same time under business activities). The second major approach to gathering time data is to observe activities directly. One or more interviewers are stationed in or by the house to re_ord activities on time sheets. The method still requires some retrospective reports of time use since not all â&#x20AC;˘_individuals spend a full day near the home. A possible serious bias is that the presence of the interviewer will influence the activities of family members. A check on observer bias can be attempted by stationing an interviewer for two or more days at some households, If average behavior on the second day or subsequent days differs markedly from behavior on the first day and if differences show consistent patterns, there is some indication of observation bias.

TIMEBUDGETSTUDIES

A third method is to have respondents keep time charts of their own activities, noting activities performed at (say) hourly intervals. Aside from the obvious bias that keeping the time chart requires, like being time-consuming and interfering with the usual flow of activities, the method demands relatively well-educated and highly motivated respondents It is possible to combine diaries of time use with retrospective reports. In a study involving twelve countries, respondents were contacted on the day prior to the one for which time use data were sought and were instructed to keep a time record (Szalai, 1972). On the day following the day of observation, respondents were interviewed to supplement recorded events. We found this approach unsuccessful in the Philippines. In practice, the method yielded retrospective reports of the previous days' activities. In comparing the methods, one can say that self-kept time charts are probably not usable in developing countries, and that observation is likely to yield the most accurate estimates of time allocations, although some care must be taken to determine whether there is observer bias. In our Philippine survey, we observed households for two days and used only the reports from the second day, because interviewers stated that households appeared to be modifying their behavior because of the presence of the interviewer. When households do not have watches or clocks or other methods for telling time, retrospective reporting or time charts cannot be used as substitutes for observation. Observation is, of course, the most expensive method of gathering time data. 4. Problems of Interpreting Time Data There are two major problems in measuring and interpreting time data. The fn,st has to do with "jointness" of activities and the second with the intensity of effort. Jointness arises when individuals perform two or more activities simultaneously. For instance, _a woman can breastfeed and stir the supper soup at the same time or mind the family business and care for children simultaneously. If each activity is listed separately, it is even possible for retrospective reports to yield more than 24 hours of time allocated to a set of activities. A related problem occurs in observational studies. If intervals over which time observations are noted are long, one can perform more than one activity in the interval. Thus, if a woman

BRYANL. BOULIER

spends one-half hour caring for children and one-half hour cooking supper and the minimum unit over which time is recorded is one hour, the woman has performed two activities. This difficulty is reduced by shortening the time interval. Since interviewers cannot be expected to record activities second by second, a compromise time, determined by experimentation, is needed. A partial solution is to have the observer code two activities for the time interval. While this method does not solve the problem, it at least gathers information on the ability of individuals to combine activities, an important question in its own right (McCabe and Rosenzweig, 1976). In the Philippines, we used open ended intervals and permitted the recording of at most two simultaneously performed activities. The second difficulty in the measurement and interpretation of time data is the problem of intensity of effort. Suppose a respondent says that he spent eight hours in crop cultivation. The investigator cannotdetermine without considerable expense (e.g, measurement of calories expended)whether the respondent worked hard or enjoyed considerable on-the-job leisure. Variation in effort intensity may be large in agricultural areas with pronounced seasonal labor requirements. And, if effort intensity is variable, it is quite possible that a reduction in the demand for labor will not lead to a reduction in reported hours worked, only effort. 3 5. Time Allocation in Laguna Households Table 1 presents estimates of time allocation in a sample of households in Laguna, Philippines. 4 Part A of Table 1 gives the 3. Stafford and Duncan (1977, Table 2, p. 15), who have addressed these issues with time diary data for the United States, fred that about 10percent of work time is spent in coffee and other scheduled work breaks, informal breaks, personal business, and lunch breaks more than 60 minutes, and suggest that failure to adjust conventional estimates of work time for such breaks yields erroneous patterns of age-work profdes, givesmisleadingestimates of union and non-unionwage differentials,and may account for a portion of the declinedin estimated labor productivity if the extent of on-the.job leisure has increased throughtime. In addition,by comparingworkers'evaluationsof workintensity relative to workers' evaluations of the energy and effort involvedin watching television,StaffordandDuncanconcludethat there are substantialdifferencesin workeffort by occupation,union status,andgender. 4. Households in the sample were chosen by simple random sampling from household lists prepared for 34 ruralbarriosselected to give a mixtureof

Table I - A Comparison of TmaeAllocation Estimates Obtained from Re_mpeetive Questiennakes and from Obaervatioa. t

A. RETROSPECTIVI_ QUESTIONNAIREa (Hours per Week)

B, OBSERVATIOI_ (Hours per Day) Father

Father

(a) Crops (b) Fishing

(c) Pout_and Liveaock (d) Wage (e) Business and Professions (f) Income Producing Home Production (g) Repairs (h) Markettn8 Total Income Earning Time (i) Non-income Home Production (j) Child Care Total Non-Income Time _Dalafrom Boalle_(1977). DatafromKin8(1970.

22.03 .43

9.84

Mother

1,31

Mother

Children

1.54

Cycle 1

Cycle 2

Cycle 1

Cycle 2

Cycle 1

Cycle 2

2.84 .24

3.22 .31

.77 .01

.81 .01

2.29 .18

2.68 .29

4.47

1.38

.36

.21

.51

15.72

4.29

6.67

1.34

1.96

.50

.79

1.22

.96

2.06

3.76

.22

.87

.72

.45

,56

.70

.58

2.31

.05

.13

.39

.12

.14

.05

,08 .23

.01 .05

.06 .05

.07

.14 .01

.01

2.04

2.80

52.13

.61

.82

.36

16.33

12.43

6,26

7.22

2,59

2.57

5.04

5.08

2.85 .30

40.94 9.38

17.00 3.16

.68 .31

.95 .53

5.55 2.33

5.50 1.88

2.66 .65

3.13 .34

3.15

50.22

20.16

.99

1.48

7.88

7.38

3.31

3.47

-52

BRYAN L, BOULIER

average number of hours worked per week by family members for 366 farm households interviewed in May through July of 1975. The Cycle 1 and Cycle 2 data in Part B of Table 1 are based on the second day of a two-day observation of a sub-sample of 98 households, where the observations took place approximately three months (Cycle l) and six months (Cycle 2) after the date of the original survey. The estimates of child time refer to an average of about four children per household. Because the households are not the same for the time estimates and because the data are gathered in differentseasons of the year, the retrospective report estimates of time use are not directly comparable to the observation estimates. The retrospective report data are estimates of the average number of hours worked per week in the year preceding the survey. For crop cultivation and fishing, the average number of hours per week is calculated by multiplying the average number of hours per day spent in the activity in the past month times the number of days worked and dividing by 4.35 weeks per month. The last month is assumed to be a typical month for the purpose of converting these data to average hours worked in the past year. May, June, and July are months of relatively intense activity of crop cultivation in Laguna, so that estimates of time spent in crop cultivation are probably overstated. On the other hand, the time of children devoted to economic activity may be somewhat understated since school begins in June. A similar procedure is used for time caring for livestock and poultry for which questions asked minutes per day and days worked in the last month. For wage earning time and time spent engaged in a profession, the number of hours worked per week in the last month is multiplied by the number of months worked in the last year divided by twelve for both primary and secondary jobs. For business activities, the number of days worked per week in the last month is multiplied by eight hours per day times the number of months worked in the past year divided by twelve. Time spent in income-producing home production is summed from responses to questions about hours worked in the past week devoted to home gardening, washing and ironing for sale, .furniture making and handicrafts, food preservation for sale, woven crafts, and other householdeconomic activities (e.g., fishing barriosand rice cultivationbarrios). Boulier (1976) and King (1976) analyze the determinants of time allocation u_ng the Lagunadata.

TIMEBUDGETSTUDIES

activities. Together these activities are-denoted "work time," even though some of the time in crop cultivation, poultry and livestock care, fishing, and gardening is really time devoted to production for home consumption. Childcare time is time devoted to feeding, bathing and dressing and cuddling and watching infants (0-2 year olds) and other pro-school children. Non-income home production time is time spent in the last week in the following activities: (a) (b) (c) (d) (e) (t3 (g) (h) (i)

Marketing of food, Washing dishes, Cleaning baekyard, Cleaning house, Cooking and preparing food, Other feeding time, Washing and ironing clothes, Getting water and f_rewood, and Mending, sewing, or repairing children's clothes.

All home production time, whether income or non-income, and child care time data are taken from questionnaires given to mothers. Other income-earning time data are taken from the fathers' questionnaires. Questionnaires w_re checked to eliminate duplication in the reporting of time (e.g., weaving reported as home production by mothers and as business time by fathers). When such duplication occurred, mothers' reports of their own time and children's time were used. In comparing the time estimates by the two methods, let us consider household members separately. For fathers, total income earning time per week derived from the retrospective questionnaire exceeds the estimates from the observation method (the daily average multiplied by seven). The excess arises from more time spent in the care of poultry and livestock, wage earning activity, and income-producing home production, with somewhat less time spent in business and professional activity and in t'_hing. These differences seem to be largely differences in sampled households. The retrospective reports include many non-rice producing farmers, while the observations include only rice-producing households, although they also engage in other kinds of farming (e.g., coconut raising). Compared to the observation estimates, retrospective reports show only about one-half the amount of time devoted to non-income

BRYAN L. BOULIER

home production and about one-seventh the amount of time devoted to child care. In part, the differences may stem from the fact that retrospective reports on fathers' time spent on these activities are obtained from their spouses. The two estimates of time spent by mothers in income-producing activities are similar, although the detailed composition of time is somewhat different. Time devoted to non-income home production is about 15 percent higher in the retrospective reports, and child care time is about 30 percent lower. The retrospective reports of children's time spent in incomeearning activities is only about one-third the amount obtained by the observation method, with large differences occurring in crop cultivation, care of poultry and livestock, and business and professions. Child care time estimates are similar and non-income home production time is about four hours per week lower. All things considered, the retrospective reports of child activities are far smaller than the time estimates obtained from the observation method. 6. Conclusion In this paper, conventional methods of gathering labor force data have been described and criticized, and methods of gathering time budget data have been described and problems of data interpretation outlined. A comparison of two methods of gathering time data for a small sample of households was attempted, although the comparison was hindered by differences in the sampled households and differences in the dates at which data were gathered. For adults, retrospective reports of time allocation were broadly similar to estimates of time allocation obtained by observation, although estimated nonincome home production time of fathers is markedly lower in the retrospective reports. Retrospective reports of child activities are far smaller than those obtained by direct observation. Time budget data when combined with other economic and demographic information can provide answers to many of the questions for which labor force data are gathered. However, because of our limited experience in gathering and analyzing time budget data and because of the cost of gathering such information, time budget surveys are probably not a substitute for conventional labor force surveys. Rather, the principal benefits of time budget surveys would appear to be guiding the formulation of labor force questions

TIME BUDGET STUDIES

to insure that important productive categories of time use are not omitted from labor force surveys and studies on the iuicroeconomics of household

behavior.

REFERENCES Acharya, M. (1981), "Time Use Data from Nepalese Villages: Policy Implications," Unpublished manuscript. Becket, G. (1965), "A Theory of the Allocation of Time," Economic Journal, 75:493-517. Birdsall, N. (1981), "Measuring Time and Nonmarket Exchange," in Third World Poverty. ed. W. Mcgreevey, Lexington, Massachusetts: Lexington Books. Rosemp, E. (1976), "Employment of Women in Developing Countries," in Population Growth and Economic Development in the Third World, ed. L. Tabah, Liege, Belgium: International Union for the Scientific Study of Population, 1: 79-108. BouUer, B. (1976), "Children and Household Economic Activity in Laguna, Philippines," Discussion Paper No. 76-19, Quezon city: Institute of Economic Development and Research (UPIEDR), School of Economics, University of the Philippines. Butz, W. and Gteenbetg, D. (1975), An Economic Methodology for Measuring the Benefits from Children, The Rand Corporation R-1792-RF. Cain, M. (1977), "Household Time Budgets, \Village Fertility Study Methodology Report No. 1," Unpublished manuscript. Durand, J. (1975), The Labor Force in Economic Development: A ComparLson of International Data, 1946-1966, Princeton: Princeton University Press. Engie, P. and Butz, W. (1981), "Methodological Issues in Collecting Time Use Data in Developing Countries," Unpublished manuscript. Gronan, R. (1976), "Leisure, Home Production and Work: The Theory of the Allocation of Time Revisited," NBER WorMng Paper No. 13 7. I-Iauser,P. (1974) 'Whe Measurement of Labor Utilization," Malayan Economic Review: 1-15. Havrylyshyn, O. (1976), "The Value of Household Services: A Survey of EmpiricatlEstimates," Review of Income and Wealth, 22: 101-131. King, E. (1976), "Time Allocation in Philippine Rural Households," Discussion Paper No. 76-20, Quezon City: Institute of Economic Development and Research, School of Economics, University of the Philippines. Lancaster, K. (1976). "'A New Approach to Consumer Theory," Journal of Political Economy, 74: 132-157. McCabe, J. and Rosenweig, M. (1976), "Female Labor Force Participation Occupational Choice, and Fertility in Developing Countries," Journal of Developmen t Economics 3: 141-160. Meeller, E. (1977), "Desisn of In-Depth Employment Surveys for Human Resources Analysis in LDCs," Unpublished manuscript.

BRYAN L. BOULIER

Myrdal, G. (1967), Asian Drama, New York: Twentieth Century Fund, 2: 961-1028. Stafford, F. and Duncan, G. (1977), "The Use of Time and Technology by Households in the United States," Unpublished manuscript. Stone, P. (1972), "The Analysis of Time Budget Data," in The Use of Time ed. A. Szalai. The Hague and Paris: Mouton Press. 89-112. United Nations (1967), Principles and Recommendations for the 1970 Population Censuses, Statistical Papers, Series M, No. 44. White, B. (1976) "Economic Value of Children," Unpublished Ph.D. Dissertation, Columbia University. Wigna, W., Surnayata, K. and White, B. (1980), "Comparison of the Results of Time Allocation Research Using Two Different Recall Periods," Unpublished manuscript.

_GESTRUCTURE AND DEVELOPMENT IN ASEAN AND JAPAN: 1950-2015, A PRELIMINARY REPORT Burnham 0. Campbell*

1. Introduction Although population changes have been given increasing attention in both theoretical and empirical explanations of growth, the age composition of population changes remains a relatively neglected factor. Recently, especially in the work of Richard Easterlin and Michael Wachter, this neglect has been questioned and theoretical and preliminary empirical evidence (mostly for the US) advanced to suggest that changes in age structure may have a pervasive effect on economic, social, and political outcomes2 Unlike earlier empirical studies including my own on demographic changes and long-swings in residential construction, 2 Easterlin and Wachter emphasize the effect of changes in age structure on various age specific rates (e.g., fertility rates). Since the effects of age structure changes have been generally neglected, perhaps even more so in studies of growth in developing countries than elsewhere, this preliminary paper is designed to accomplish three tasks. FirSt, the major possible influences of shifts *Univexsityof Hawaii.I would like to acknowledgethe use of the research facifitizs of the Institute of Developing Economies, Tokyo, and of the Population Research Institute, Nihon University, Tokyo, and the research as_stance of MasakiOhkubo and Yasuniko Saito of the Population Research Institutestaff. 1. See Easterlin, R. (1978), Easterlin 17,.and Wachter, M.L. and S.M. (1978), Wachter,M.L.(1975, 1976), andWachter,M.L.andS.M.(1980). 2. Campbell,B. (1966). 57

BURNHAMO. CAMPBELL

in age structure, both given age-specific rates and through feedbacks from induced changes in age-specific rates, will be set out. Then, since if age composition does not change it makes no difference what the effects of changes would be, the extent of past and likely future age structure changes in the ASEAN countries and in their major trading partner, Japan, will be investigated at this stage in relative broad terms. Finally, given the assumed relations between different economic, demographic, and social variables and changes in age structure, some of the past and likely future effects of changes in age structure for the ASEAN countries and for Japan will be asse'ssed. If the possible impact of shifts in age composition turns ottt to be important, this should encourage research to test the hypothesized relations - the major goal of this study. Putting population changes in a growth context can most easily be accomplished through the long-swing growth pattern that characterized economic development in the US and industrial Europe from the early nineteenth to the mid-twentieth century and possibly beyond.3 From the 1830s to the 1930s, until the 1924 Immigration Act ended the process, the expansion phase of the long-swing in the US put upward pressure on real wages in the US which encouraged immigration and hence growth in the labor supply. The responsiveness of the supply of labor to growth in demand kept the expansion going longer than otherwise would have occurred., raising the capital stock, the quality of the capital stock and per capita incomes above levels that otherwise would have been achieved. Slowing growth, as the longswing peak was passed, at a time when children born to the last wave of immigrants were entering the labor force, then created labor market conditions leading to reduced immigration. The resulting long-waves in population growth and in the age structure of the population gave rise to the residential building cycle that led the way into the depressions that terminated each long-swing from the end of the Civil War through the 1930s. Over this period, partly because the ebb and flow of immigration tended to stabilize the relative income of the young, age-specific rates, at least headship 3. See Abramovitz, M. (1961) for the standard description of the long-swingprocessandEasterlin,R. (1968) for an analysisof the contribution of demographicfactorsto this process. Oshima,H. (1980) contains artapplication of the long-swingconcept to the growth experiencein East Asia, one of the few applicationsto developingeconomies.

AGESTRUCTUREAND DEVELOPMENT

rates, remained remarkably constant.4 Then, after the 1930s, population change was no longer dominated by swings in immigration. When the long-swing expansion during and after World War II raised the demand for labor at a time when the entering labor force was growing relatively slowly, the absolute and relative incomes of the young could increase sharply without inducing an offsetting flood of immigrants but bringing about, however, increased ruralurban migration. One outcome was a large increase in household formation rates among the young, partly responsible for the period's housing boom and the continued role of residential construction in the long-swing. Another, according to Easterlin, was a large increase in fertility rates, which more than offset the effects of the contemporary change in age structure and raised crude birth rates. Bythe early 1960s the increase in births began to be reflected in rapid increases in the entering labor force, both absolutely and relative to the established or basic labor force. This change in the age composition of the labor force redt_ced the relative incomes and increased the rate of unemployment of the young and so, among other effects, reduced fertility rates. The resulting waves in population change and in age composition will continue to influence the prospects for economic growth as well as for such diverse outcomes as divorce, suicide, and political alienation rates in the US far into the future, s In most developing countries immigration is ruled out and, with some important exceptions, emigration is not encouraged. 6 Thus, the most recent US experience is more relevant than that based on the pre-World Wax II waves of immigration. Most changes in the labor force base and in age structure in developing countries are the result of past swings in births and of the aging of the indigenous 4. The response of immigrationto labor marketconditions also tended to stabilize the absolute real wages of unskilled and semi-skilledlabor near the trend growth rate, but below it, in productivity. Presentimmigrationin the US from Latin Americamay partlyhave this effect but it seems basedmore on the large existing inter-country differentials in real wages and so may prove less responsiveto long-swingsin labordemand. 5. The analystsin the text is based on my Universtiyof Hawaiilecture notes (availablein mimeographedform) coveringthe 1967 to 1981 period and on Easterlin(1968, 1978). 6. Exceptions in the ASEAN regionin the period covered include the effects of emigration, permanent and otherwise, from the Philippines and immigrationfromMalaysiato Singaporein the late 1960s and early 1970s.

BURNHAMO. CAMPBELL

population. Thus, the size and skill Composition of the labor force and the relative • income and labor •market experience of the young, with all the. feedbacks and effects suggested by Eastcrlin-Wachter, will depend on the current, age Composition of the population and hence on past, not present, economic conditions. •However, for much of the world, until quite recently, the age structure of the population probably did not vary widely. Population • growth was slow, despite high birth rates, because of high infant mortality, pestilence and war. And, because productivity was low and growing very slowly, if these brakes on population expansion temporarily disappeared , the Malthusian"constraint of starvation would take their place. With• all surviving the high initial mortality rates equally affected by.these limiting forces there was little scope for wide swingS, at least as a continual experience, in age. composition. The whole population increased or decreased with the cycles of pestilence .and war, leaving the .age composition relatively unchanged, Then in the twentieth century, mostly for developing countries since World War II, a combination of medical, economic, and political changes removed the major past constraints on •population growth. •With a much higher percentage of Children born surviving the first dangerous years, with increased •incomes allowing improved sanitation, better nutrition for ever larger numbers, mass preventive inoculations, and better medical .care when illness does arise, the result was population explosion. Accompanying this explosion was a sharp twist in the age structure, with-the very young accounting for a much greater proportion.of the total population. One result, since the very young add little to output, was the dilution' of hard won increases in total output •even. among more rapidly increasing populations,, keeping per capita growth rates at very low levels or even reducing them.• Oncethis population wave reached the age of first employment the spectre of social and economic malaise and of still further increases in .births led to .a sense of emergency that gave rise to numerous national efforts to control birth rates. These efforts, through the auspices of the UN and others, became world-wide, the goal being to control birth rates before Malthusian limits again come into plaY, this time involving the earth's environment and lifesupporting resource limits! This effort or, more likely, the effect of the increased entering population on the relative and often absolute incomes and job. market experiences of the young had the desired .effect. in many countries. Thus, fertility rates fell, giving rise to

AGE STRUCTUREAND DEVELOPMENT

swings in the age structure and in population growth rates that will affect economic activity in these countries through the year 2000 and beyond. In the present context, the key point is that the ASEAN countries are among those experiencing these waves. If per capita rates of economic, demographic or social variables differ between age classes, then the swings in age composition just described can affect all these variables directly. For example, to mention only a few important possibilities, household formation or headship rates, savings rates, work experience and up-to-dateness of education, kinds of goods and services demanded, labor force participation rates, school attendance, adaptability to changed conditions, ate all known to vary between age classes. Thus, if age-specific rates are relatively constant and the age composition of the population changes, the above variables will also change and with them the economicand other activities they help determine, including activities that may influence births and thus future population and age structure swings. In addition, swings in age composition can affect the relative incomes of different age groups, if they are not perfect substitutes in production, and their relative labor market experiences (e.g. unemployment or underemployment rates, income expectations). In turn, changes in relative income, in income expectations, and labor market experience can alter the pattern of age-specific rates. In fact, the change in age-specific rates may overwhelm the opposite effect, if present, of changes in age composition, given age-specific rates. This clearly happened in the US just after World WariI when a change in headship rates at early life-cycle stages turned a prospective decrease in household formations, at given headship rates, into a large increase. 2. The Relative Income Effects of Changes in Age Structure With some differences, the following analysis is based on the Easterlin-Wachter approach. Essentially, this approach divides the working population into two age groups. The argument is that the relative income of the first or entering labor force group, defined as the 15-29 age class, varies inversely with. the ratio, hereafter called the key ratio, of this age class to the second or basic labor force group, det'med as the 30-64 age class. This inverse variation is based

BURNHAMO. CAMPBELL

on the assumption that workers in the two age classes are not close substitutes. A simple argument in support of this assumption, from the viewpoint of employers, would be that the greater experience and the greater stability of the services of the 30-64 age class make the expected productivity of the average member of this class considerably higher than for the 15-29 age class. Thus an increase in the relative numbers in the 15-29 class will have the effect of lowering the relative wages of this age class.7 Actually, the relative productivity and degree of substitutability between these broad age classes would be expected to vary with the stage of development. Four stages may be usefully recognized in this context. The earliest stage, involving a predominantly agricultural economy, would exhibit relatively minor differences in relative productivity between age classes and gives little importance to the effect of shifts in age composition on the economy via shifts in relative income. Only the effect on economic, demographic, and social aggregates of changes in age composition given age-specific rates would remain. Then in stage two, when labor intensive industrialization based on relatively unskilled labor is underway, output tends to grow at a pace constrained, with a lag, by the growth of the entering population. The established population partly substitutes for the enteringpopulationinlaborintensive industries and partlycomplements thisageclass inmanagement and service roles. The established populationalsoprovidesan increasing shareofagricultural labor. At thisstage, swingsinthekey ratiobegintoaffect therelative incomes of the two age classes, both becauseexperience beginsto countin productionand becauseshifts inthekey ratiowillgiverisetoexcess supplyor demand forolderworkersintheirrolesthatcomplement theentering laborforce. As tradeand industrial productiongrow inimportanceand in 7. There is some product mix, some growth in the mix of capital specialized to different labor characteristics,some adaptations of the skilled/ unskilledmix in producingdifferentproductsthat together orindividu"llycould keep the relative demands for the two age classesin line with their relative numbersat existingrelativewages.It is assumedthat changesin relativenumbers .lone cannot, i.e., Viachanges in the composition of demand or in expected labor constraints at different skiff levels, bring about the shifts described. Thereforerelativeexcess supplies or demandswill appearand relativewageswill be changed.

AGESTRUCTUREAND DEVELOPMENT

complexity, the learned skills required also become more complex and a premium is placed on the expected stability of the labor supply. These characteristics of stage three indicate that the degree of substitution between younger, less skilled and less stable workers and more experienced, more settled workers declines and relative incomes tend to shift more widely with the same ehange in the key ratio. Education is an important input, but the basic ideas tend to serve for an entire working life. At this stage, the younger population tends to complement the basic population and their jobs tend to be tied to the employment of the basic population, via pipeline and apprentice relations. Substitution takes place through changes in the mix of products requiring different input skills and occurs in response to the relative wage changes brought about by changes in the key ratio. Finally, as up-to-date education becomes more valuable, because of rapid technological change in a "high teeh" economy, the relative productivity of theyounger population will increase, increasing substitutability and reducing again the relative income effect of key ratio changes. At some point, the productivity of the entering population may even surpass that of the basic population, reducing substitutability once again. In this fourth stage, probably not yet attained by any country, the average product per worker will move directly with rather than inversely to the key ratio. With the exception of Indonesia and possibly Thailand, Japan and the ASEAN countries were at or beyond stage two by the beginning of the 1960s. With the exception of Indonesia, all were almost into or into stage three by the beginning of the 1970s. So, in general, the conditions were present for the expected relation between the key ratio and relative incomes to hold in the ASEAN countries and Japan over the postwar period. Although some countries may move into stage four (e.g. Japan and Singapore) by the time the period to be covered ends (2015), the analysis will proceed on the assumption that all will be in stage three through this period. For changes in relative income to have the effects postulated, the impact of changes in the relative income of the 15-29 age class must not be outweighed by the effects of opposite changes in absolute income. For example, if the 15-29 age class grows more rapidly than the 30-64 age class but there is a large excess demand for both age classes, it is doubtftil that the fact income increases 25 percent for the entering age class while increasing 35 percent for the 30-64 age

BURNHAMO. CAMPBELL

class will reduce fertility or headship rates. They may not increase as much as they otherwise would have, but that is all. Thus, arguing that relative income varies inversely with the key ratio is not sufficient; something must be said about the total supply and demand for labor as well. If, as in growth stage three, the demand for labor varies with the 30-64 population, then, of course, the market for this age group would always be in balance at the existing wage and increases or decreases in the key ratio would signal excess supply or demand for the 15-29 age class. 8 Thus, relative real income would fall and unemployment would increase in the entering age class if the key â&#x20AC;˘ratio increased. What would then happen to absolute real income would depend on what is happening to labor productivity. All that can be said is that the 15-29 age groups share of increased real output per unit of labor input would tend to fall. Of course, unless productivity changes are really large, the young will be more impressed by the decline in their relative position than by any increase in the purchasing power of their nominal income, especially if consumption standards are based on the now relatively higher real incomes of the 30-64 age class. Even with higher incomes it may seem more difficult than before for the 15-29 age class to attain the desired consumption standard. A relatively simple model that would give the relation just described, is that over five year periods (the time period used in the empirical work that follows) with the growth in the basic labor force, the 30-64 population, plus the rate of increase in output per unit of labor input, determining the rate of growth in GNP. In other words, over such periods the economy adjusts to the basic supply of labor. There is no deficiency of demand and the capital stock adjusts to keep the 30-64 labor force employed at the same average product, net of increases in labor productivity. If the latter are assumed to proceed at a relatively constant rate over time, then the GNP growth â&#x20AC;˘rate and the growth rate in the basic population will vary together. If it is additionally postulated â&#x20AC;˘that the demand for the entering 8. If, alternatively, labor demand grows with the total population or with the growth in the 15-29population as, with a lag, might happen in stage two, the relative income effects of changes in the key ratio would be the same as described for stage three. However, the unemployment and absolute income effects would tend to differ and so the net effect of relative income changes might also differ, a possibility ignored in the following analysis.

AGESTRUCTUREAND DEVELOPMENT

population is some function of the rate of change in the basic population, given relative wages, then the fortunes of the entering population would clearly vary inversely with the key ratio. 9 Direct reasons for such a relation could involve modern variants of an apprentice or helper system or a fixed proportion of unskilled to skilled intensive outputs all related to the 30-64 age class labor constraint. To sum, in this supply determined model the total demand for labor varies proportionately with the rate of growth of the basic population. Thus, excess supply or demand for labor is restricted to the entering population. The unemployment rate and the relative income of this age group will vary with the excess supply or demand â&#x20AC;˘ and, unless labor productivity changes are large, so will absolute income. The extent of the variation will depend on how the product mix responds to the changes in relative income. However, the absolute income relation will be assymetrical. Since the trend in labor productivity generally has been positive, the real income of the entering population will increase if the labor market is in equilibrium (i.e., if the key ratio is unchanged from one period to the next). This productivity effect must be added to excess demand and subtracted from excess supply situations to get the net effect of these differing circumstances on wages. If, in fact, the growth in the 30-64 population is not in line with the growth in labor â&#x20AC;˘demand, then the outcomes become ambiguous. In this case, if there is an excess supply of basic labor, the relative income of the entering population can be expected to fall (in stage three) even if the key ratio is constant and may, up to a point, d_p even if the key ratio drops. With an aggregate excess supply, more experienced workers will be Substituted for less experienced ones, workers representing a bigger investment to employers will be kept at the expense of new entrants, etc. On the other side, if there is a basic excess demand, entering workers will be upgraded to fill the gap and 9. However, if for example, excess supply of the entering age group appearsandrelativeincomes fall,some substitutionof entering for basicworkers will occur, either directly, by assumption very limited, or by changesin the product mix as the relativeprices of goods and servicesusing relatively more unskilledlabor and, usually,relativelyless capitalper unit of output tend to fall. The more significantthis indirect substitution effect between age classes, the more importantwill be the composition effect comparedto the relativeincome effect of key ratiochanges.

BURNHAMO.CAMPBELL

their relative income may increase even if the key ratio is unchanged. Certainly their absolute income and so, presumably, their behavior will change under these circumstances. However, in the following discussion these ambiguities will be ignored. The economic implications of the model presented will be discussed in the next section. An important qualification must be introduced first. Breaking the labor force into two age groups ignores possible variation within the age groups selected. Optimally, the variation within should be nil and the variation between the age groups selected should be considerable. In fact, if only two broad age classes are used, the best that can be expected is that variation between the age classes be greater than variations within. In that case the age â&#x20AC;˘classes selected should be the ones that maximize the ratio of inter to intra-elass variation. For present purposes it is sufficient to assume that this is the case and to ignore the effect of variation within the age Classes involved, even though this is large for several variables to be discussed, e.g., headship rates. However, in any subsequent study the ideal of selecting age classes to exclude any significant variation within the age classes should be pursued. 3. Compositional and Relative Income Effects of Changes in the Key Ratio Given that various economic, social, and political activities vary relative to the population in each age class, changes in age structure have a role in determining changes in the level and in the rate and direction of change of the activities involved. Since changes in age structure, at least of the labor force and household formation population bases, can be readily forecasted (net of immigration or emigration), they provide an important input in making base-line (i.e., given age-specific rates) forecasts of labor market conditions, housing needs, education demands, GNP growth rates, etc. In the following, the effect of changes in the age structure of the population, as measured with age-specific rates held constant, is called, directly enough, the composition effect or the CE for short. Changes in age structure affect the relative supplies of labor in different age classes and, in the manner described above, the changes â&#x20AC;˘ in relative supplies affect the relative incomes of the different age classes. Changes in relative incomes and, perhaps more importantly, the unemployment rates and lifetime expectations that move with

AGESTRUCTUREAND DEVELOPMENT

relative incomes, will affect the age-specific rates for economic, social and political variables. In the following this effect is labeled the relative income effect or simply RE. This effect, ff the age-specific rate concerned is a ratio to income (e.g., the propensity to consume), will be the net of a "pure" RE or "keeping up with the Joneses" effect and of the fact that changes in relative income alter the aggregate income weights attached to the different age-specific activity rates. Where these two effects pull in opposite directions the net result may not be predictable on a priori grounds (see the discussion of the propensity to save below). If the neoclassical assumptions discussed roughly hold, the key ratio or some variant is probably not too bad a measure of the relative income effect. For this to be so, it is necessary that increases in the ratio do not always match with large aggregate excess demand for labor and decreases with large aggregate excess supply. However, it is much â&#x20AC;˘less satisfactory as a measure of the composition effect. For many activities, e.g., household formations, saving rates, there is no way the population between 15 and 64 years of age can be split into two groups without there remaining much intra-class variation in one or both of the groups. Nor, if the full range of the composition effect on the economy is of interest, is it possible to cover this range if the population outside the 15-64 age class is ignored. A more complete study would have to look carefully at the gains in explanatory and predictive power attainable with the further disaggregation possible on the basis of available data. Given these qualifications, the expected relations between an increase in the key ratio and a long but incomplete list of dependent variables, based on the net impact of the CE and RE, are given in Table 1. For some variables, these two effects complement one another and pull in the same direction when the key ratio changes. For example, an increase in the key ratio will tend to reduce the households to population ratio both because headship rates are lower at early than at late life-cycle stages and because the induced fall in the relative income of the 15 to 29 age class will reduce the age-specific headslfip rates in this age class much more than the increase in relative income of the basic population will raise headship rates at later life-cycle stages. I0 Essentially, the age of leaving home, 10. When relative incomes change, the responseof age-specificratescan be expected to differ between the age classesincluded in the kzy ratio. For

BURNHAM O. CAMPBELL

Table 1 - Net Relative Income and Composition Effects of Changes in Age Structure: Selected Variables I. Variables Reduced By An Increase In The Key Ratio 1 1. 2. 3. 4. 5. 6.

Income Equality a (0, -) Fertility Rate b (0+, -) Immigration Rate (0, -) Average Headship Rate (-, _)c Skill Intensify Of Output (-, _)d Average Product of Labor And Per Capita Income (-,-)

II. Variables Increased By An Increase In The Key Ratio 1. 2. 3. 4. 5. 6. 7.

Unemployment (Underemployment) Rates (+, 0-) Female Labor Force Participation Rates (0+, +) Emigration Rates (+, +) Age of First Marriage, Age of Leaving Home (0, +) Suicide, Divorce, Alienation Rates (Jr,+) Responsiveness To Change (+, +) Demand For Loanable Funds (+, +)

IIL Variables With A Priori Uncertain Responses To An Increase In The Key Ratio i. 2. 3. 4. 5. 6. 7. 8.

Savings Rate (--, ?) Aggregate Demand (?, ?) Goods Demanded By Young/Goods Demanded By Basic (+, -) Demand For Higher Education (+, ?) Velocity Of Money (+,-) Crude Birth Rate (+, -) MaleLabor Force Participation Rates (-, ?) Internal Migration Rate (+, -)

a. Range Of average incomes across quintfles and skill distribution. b. For females 15-34. c. Also reduced ate the ratio of owners to renters and the average quality of housing demanded. d. Capital intensity is often :elated to skill intenaity and an increase in the relative cost of unskilled workers would likely increase capital intensity. 1. The first sign in parentheses after each variable is the expected direction of the composition effect and the second sign in the expected direction of the relative income effect.

AGESTRUCTUREANDDEVELOPMENT

the age of first marriage, and the size of non-normal households all depend on job market experience and the headship rates of the young vary with these factors. 11 In fact, all the per capita expenditure, running from demand for appliances to the services of various utilities, related to the average headship rate will also tend to fall. Other major economic activities in which the composition and relative income effects are complementary or in which one or the other is the only effect involved include the relative attractiveness of skill intensive versus unskilled intensive products, the average productivity of labor, the relative capital intensity of production, the household and skill related income distributions, and the rate of immigration. All these variables move inversely with the key ratio. On the other side, expected to vary directly with the key ratio are, to mention a few, such outcomes as the natural rate of unemployment (so the inflation necessary to achieve a given rate of unemployment), the rate of emigration, the age of marriage and of leaving home, and political alienation, suicide and divorce rates. Many other variables, including such key outcomes as crude birth rates, internal geographic mobility, and the savings ratio, may increase or decrease in response to a key ratio increase since for these variables, the composition and relative income effects can pull in opposite directions. Because age-specific fertility rates are higher for the 15-29 age class, the composition effect of an increase in the key ratio will be to raise crude birth rates. However, the relative income effect of the same change will tend to reduce crude birth rates by raising the age of marriage and the age of leaving home, by increasing example, sinceit is much simplerto choose between livingwith one's familyor forming an independent household when young and unmarried than after marriageand children, the response of headship rates to changes in relative income will also differ between the entering and basic populations. That there will be a differential response for many other age.specific rates, e.g., fertility rates, demand for education, immigration,emigration, migration, suicide and divorcerates,is also appparent. 11. These changesdo not mean householdformations will fall, only that they will be lower than if the key ratio had not increased. They will fall if the combination of the CE andRE more than offsets the currentpopulationgrowth. After World War II what would have been a large decrease in household formationsat beginningperiod headshipratesin the US waspartlyoffset by the currentCE and turned into a largeincreaseby the even largerimpactof the RE on age-specificheadshiprates of the young.

BURNHAM O. CAMPBELL

the female labor force participation rate, and by decreasing the perceived well-being and foreseen accomplishments of the most fertile age groups. No a priori conclusions can be drawn, but Easterlin shows a clear inverse relation for the US between the key ratio and the crude birth rate. A necessary link in the relation, if increases in the key ratio are to lead to reduced crude birth rates, is that the fertility rate must move inversely with the key ratio, a relation investigated in Section 5. Internal migration falls in the "unpredictable" category because, given the greater geopgraphic mobility of young people, the CE effect of an increase in the key ratio is to raise migration, v_hile the decline in relative income, especially the associated increase in unemployment rates or difficulty in locating a job, has the opposite effect. Because the CE and RE have opposing directional pulls, the demand for education and the relative markets for goods specialized to the two age classes involved also belong to the "uncertain" category. Finally, the saving ratio is a priori unpredictable because, while the CE and the "pure" relativeincome or "keeping up with the Joneses" effect work in the same direction, the "aggregate" relative income effect can pull in the opposite direction. The first two will reduce the aggregate propensity to save as the key ratio increases, but the last may increase, decrease, or leave it unchanged, depending on what happens to the share in total income of the 30-64 age class. If the income share of the basic population increases, then the higher age-specific saving ratio for the 30-64 age class wil_ receive a proportionately greater weight on the average, raising the propensity to save and possibly more than offsetting the impact of the other two effects. If the change in relative numbers is exactly offset by the change in relative wages so that the income shares of the two age classes are unchanged, then only the CE and the "pure" relative income effect will be involved and the propensity to save will fall. Predicting the effect of key ratio changes on the propensity to save (or other ratios involving income as a denominator) then involves predicting the probable response of the aggregate income share of the 30-64 or of the 15-29 age group, which requires country-specific knowledge. 12 12. In addition to the effects discussed, changesin age structure are likely to affect the composition of demand and the allocation of resources since tastes differ widely between age classesand the dynamics of taste development depend

AGE STRUCTUREAND DEVELOPMENT

Since the effect of key ratio changes on the propensity to consume is uncertain, the effect on aggregate demand must also be uncertain. This conclusion is supported by the fact that the effect of key ratio changes on investment and on government expenditures and taxes relative to income is also uncertain. Although the neoclassical or supply side argument used in this paper essentially assumes away any significance for aggregate demand variations, the latter may be important in a cyclical context, (business cycle or long-swing) and country-specific research to reduce the uncertainties described would clearly be desirable. However, in the following analysis, the aggregate demand effects of age structure changes will have to be ignored. Before turning to the empirical part of this study, it is important to remind the reader that the relations between relative income changes and the social, demographic, and economic outcomes listed have been asserted, not adequately derived or tested. They are based on simple ecomomic analysis or partial empirical evidence. However, for any but the stated "educational" purpose of this paper, the theoretical underpinning would have to be improved and the theoretical implications derived from the improved underpinning would have to be empirically tested in a model sufficiently large to allow holding the effects of other major determinants constant. This is obviously so for attempts to explain how age structure changes did or will affect the economy or society of a specific country. As is, the conclusions reached below must be interpreted as stating "what would have happened" and "what will happen" if changes in the key ratio are the only changes involved and have the asserted effects via relative income and employment changes on the economic, demographic, and social outcomes listed. However, the CE component of the explanation, being based on information about the actual age-specific rates at different life-cycle stages, can be taken as the actual past and future contribution of the composition effect alone to these outcomes.

on the relativesizes of differentageclasses.Underthe_ circumstances,since age will be one determinantof the marketsize for differentproducts,they will also be one determinantof the relativecosts, via economies of scale, of diff_mztproducts,%_ affec_g the allocatica of t_sources.

BURNHAMO. CAMPBELL 4. Key Ratio Changes in ASEAN and Japan; 1950-2015

Before looking at the evidence, three important qualifications to the basic results presented must be discussed. First, the age classes 15-29 and 30-64, combined in the key ratio may, on closer inspection, not be the best in the sense of maximizing between class and minimizing within class variation. In fact, it is likely that different dichotomies would be "better" in this sense in different countries, depending on the stage of development and on cultural and institutional constraints. For example, in some developing countries substantial numbers may enter the labor force in their early teens (10-14) and, because of this and the fact that schooling stops early for the vast majority, most may reach "'experienced" or "skilled" status and thus have a stable family life by their twenty-fifth year. At the other end of the life cycle, because of earlier physical aging as well as cultural practices, effective retirement may come considerably before the sixty-fifth year. These considerations might suggest a 10-24/25-54 ratio instead of the one used. Although most ASEAN countries (Indonesia is a possible exception) seem to have achieved a stage of educational and industrial development consistent with the age classes included in the key ratio, instead of simply assuming this to be true, the proper procedure would be to def'me the key ratio independently for each country after investigating labor market conditions in each country.13 Also, as noted, although some two-way dichotomy of the labor force may be reasonably appropriate for Capturing the relative income effect, no two-way breakdown of the population can give anything but a very rough approximation of most of the composition effects of changes in age structure. Further, if a two-way classification is used, the best classification almost certainly would not be the same for all the variables exhibiting an age-specific life cycle. Again, the composition effects based on the breakdown used have to be considered as only roughly indicative of the role that age structure changes might play. Finally, the results presented do not take account of possible 13. If this is done,two key ratios, one for agricultureand one for the rest of the economy, with weights changingas agriculture'ssharechanges,may turn out best.

AGESTRUCTUREAND DEVELOPMENT

short-run income effects resulting from the impact of age structure changes on aggregate demand, of long-run feedbacks from changes in saving ratios or investment rates, etc. to growth, or of other possible feedbacks from the changes induced by the composition and relative income effects to future age structure and population changes. In other words, the closed system required for a complete analysis of the effects of an initial change in age structure on the economy and on the society and so back to future changes in age composition is missing. With these qualifications in mind, from 1950 through 1990, the period over which changes in age structure are fact (i.e., do not depend on birth forecasts), the variation in the key ratio follows a _yclical or long-swing path in all the ASEAN countries and in Japan (see Figure 1). 14 This long-swing variation is particularly clear in Indonesia and Thailand, involving a peak-trough-peak contraction sequence. It is also dominant in Malaysia and Singapore, where the key ratio traces an expansion-peak-contraction path. The Philippines follows the same sequence but with a much smaller amplitude. In Japan, on the other hand, the downward trend dominates the key ratio through 1980 after which there is a very minor cyclical movement through 1990. Although moving in the noted long-swing fashion, the key ratios of Indonesia and Singapore also show a downward trend for the entire period, with Singapore's coming entirely after 1975 and Indonesia's entirely before 1975. Again over 14. The data used in Figure 1 andin subsequentfiguresare from or are based on the medium estimates contained in UN (1980). For the countries coveredthese estimatesare mostly derivedfrom Censusand inter-Censussurveys from 1950 to 1975, though some interpolation for inter-censal years was probably necessary. From 1980 to 2000 the 0.4 age class "forecasts" arebased on estimates of births and infant survivalrates and the "forecasts" of the remainingage classeson the agingof the populationby 5 year agecohorts, e.g., from 0-4 in 1985 to 5-9 in 1990, etc., usingvariousfife tables. Migrationby age classis also "forecasted", but is assumedto be largelyinsignificantin the future for the countries covered. From 2005 to 2015 the estimates were derived by u_ng the ratio of the population0-4 in 2000 to the femalepopulation 15-34in 1995 along with the female population 15-34 in 2000 to estimate the 0-4 population in 2005. The ratio of each ageclassin 1995 to the age class fiveyears older in 2000 was used with the populationby ageclass in 2000 to estimatethe population by age class five years old and up in 2005. The processwas then repeated to obtain the population by ageclassin 2010 and 2015. Basedon data derivedas described, the key ratio will be independentof birth forecasts until 1990.

BURNHAM O. CAMPBELL

Based on ¢_8t 0818 (lgS0-1975) and totac_gt egir_ of existing DoDulatJort (t eeo-1 El90)

1 2G I il_

.4_\

11¢

, /

\ \p-'C

_°_

i_.\ _I ." "_ 17 ; p.__ ._,? ,_'=', Jk ,.

It",. %" e.

! Balled on fot_aSt birth rules and I i) tion (1e95-201 fomc.a_ eOin_ sof existing po_ule-

: _ "t I

' 7O -

_.i -..,,'.,.-\ )r'l "" " .'", "_.\. ,,, -" "l ".'_. \ '\

'_L \ \I Philippines

•

"_ ,

_ Mslayeie

"_1.-.'

,o • I 1950

I 1960

I _ 970

J 19B0

_.,; _ 990

[

I 20OO

• $ingstoore _) Japan

I 20_ 0

J 2020

Year

110

Ae,.,.,....I.,,.

_4;$e_ Of1 ailSt data (I 9S0-197 =;) and fOreCast aging of existing population ,... (1 aS0.1SSO)

100 _ njde--- -,._ ...,.. 90"_

' _

Je.--..,._,,_..

tlk .,... t.....,

Baee,'l on foreclist blrlh t&tes _ntt Iorec._stlging Ofexisting pODulation (leS5-2015)

lk"_Q._..._.__.e.....l_..4.

,__._. _

,,_.

4050

I_ .... 1950

, ]

_ 1960

t 1970

Japan

" i l, 1990

t 1980

t _ . 2000

t 2010

I 2020

Year $0um.,e¢ Derived from UN (1fiB01,

FIGURE i THE KEY RATIO: ASEAN MEMBERSAND JAPAN, 1950 TO 2015

AGESTRUCTUREAND DEVELOPMENT

the period, the Philippines shows a slight upward trend while Thailand and Malaysia show no trend at all Reflecting these varying trends, there were many shifts in relative positions over the 1950 to 1990 period. Indonesia, for example, starts with the highest ratio, falls to the lowest (for ASEAN) in the 1970s and 1980, and will increase to third place in 1990. Malaysia's key ratio rank also varies considerably, moving from fourth place in 1950 to first place in 1980 and forecasted to fall to third (with Indonesia) in 1990. Thailand and the Philippines generally have high key ratios, ranking first, second, or third in most periods. Singapore's key ratio moves from last place in 1950 to second in 1975 before dropping, if the population forecasts hold, to a level far below the other ASEAN countries between 1980 and 1990. Japan's key ratio, from an initial level slightly above Singapore, was less than that of the other countries throughout the rest of the period, falling, as noted, sharply from 1950 to 1980 and reaching a level less than half that of the lowest ASEAN country in the latter year. Given the relatively large share of Indonesia in the ASEAN total, it is not surprising that for ASEAN as a whole the key ratio mirrors that of Indonesia in both long-swing and trend aspects and in relation to Japan's experience. Based on the UN birth rate forecasts, discussed (and questioned) below, after 1990 the variation across countries will be greatly reduced, with the Philippines, Thailand, Malaysia and Indonesia all falling sharply and monotonically toward the Singapore level, which will remain relatively unchanged from 1990 on. Japan, also relatively trondless, will trace out a contraction-through-expansion cycle between 1995 and 2015. If the forecasted drop in birth rates after 1975 does not occur, as it will not if relative incomes are effected as expected by key ratio changes between 1980 and 2000 and crude birth rates are effected as expected by the change in relative incomes, then, instead of falling after 1995, the ASEAN key ratios will increase. Clearly, as measured by the key ratio, the age structure has changed and will change significantly in every ASEAN country and in Japan over the 65 year period covered in Figure 1. And, in addition to the variation in age structure within the different countries, there has been and will be considerable inter-country variation. Thus, if the asserted relations hold, changes in age structure could have played and could play a significant role in

BURNHAMO. CAMPBELL

determining what happened and what will happen to key economic, demographic, and social variables in the ASEAN countries and in Japan and in determining the comparative fortunes of these countries over the period considered. The basic relation, it will be recalled, is that, ceteris paribus, relative incomes of the twoage groups involved vary with changes in the key ratio and that numerous economic, demographic, and social outcomes vary with changes in relative income and with the composition effects of changes in the key ratio. In more detail, if, for example, all other determinants of the following variables were the same in all countries, the country with the highest key ratio, e.g., Indonesia in 1950, would have the lowest average product and per capita income, the lowest desired capital-labor and capital-output ratios, the highest rate Of emigration and the lowest rate of immigration, the lowest ratio of households to the total population, the highest unemployment or underemployment rate, the lowest fertility rate and the greatest inequality of household incomes, to name a few of the outcomes for which the direction of the age structure effect is known. Given the many other factors involved, the relative rankings based on age structure differences of the variables just discussed clearly, would not be expected to match the actual country rankings of the same variables. However, Japan's relative position on most of the outcomes a priori predictable conforms to what would be expected on the basis of age structure, is For the other countries involved, it is left to the reader to carry out the exercise. The relative rankings sometimes conform to age structure differences and some times do not. Only a larger model, allowing measurement of the influence of other determinants, would make it possible to meaningfully assess the probable past and likely future contribution of the large swings in age structure that have been and will be experienced on the relative performance of the countries covered. The main point to be made is that age structure, a variable often not included in such models, could be an important determinant of the final outcome. 5. Fertility Rates and Age Structure For most variables influenced

by changing age structure,

quin-

15. Fertility rates are an exception for both Japan and Indonesia, suggestingthe importance of absoluteincome differences.

AGESTRUCTUREAND DEVELOPMENT

quennial data axe not readily available for a sufficiently long period to make possible even roughly relating changes in the variables to changes in age structure for all the ASEAN countries. However, there are comparable data on the population 0-4 years of age in all countries covered from 1950 to 2015. Thus, a crude measure of fertility rates is available, comparable across countries and based on the ratio of the 0-4 age class in a given year to the female population 15-34 years of age five years before. These data allow a rough test of the hypothesis that fertility rates vary inversely with the key ratio (via the labor market experience or relative income effect). The fact that a small number of births taking place will have been to mothers outside the 15-34 age group may bias the results slightly when changes in the fertility ratio are very small, but more important is the potential for error arising from the fact that the measure used records improvements in infant survival rates as well as what is happening to fertility rates. 16 This error is likely to be bothersome, considering the five year periods used, only in the 1950s and 1960s for the countries covered. Finally, fertility rates differ between the age classes included in the 15-34 age group and so the n_easure used will partly reflect compositional changes as well as relative income changes. The more refined studies, hopefully encouraged by our results, will have to take account of these potential biases as well as other influences on fertility. In Figure 2, the key ratio, the actual crude fertility rate as just defined for the 1955 to 1975 period, and two sets of forecasts of this fertility rate are shown. The first set is based on the UN forecasts of future 0-4 and 15-34 populations and the second on the implications of the 1955 to 1975 relation between the key ratio and the crude fertility rate for future fertility rates, and, with a lag, future changes in the key ratio. In general, the evidence presented is consistent with the expectation that fertility rates will vary inversely with the key ratio. Looking at the country relations, the hypothesized inverse relation is found in all the countries covered except the Philippines, with the relation strongest in Indonesia, Thailand, Malaysia, and Singapore, 16. Another source of possible error is that the 0-4 age class may include migrantsor that the 15-34 age class five years before may havebeen increased duringthe intervalby immigrationor decreasedby emigration.A biasin the 0-4 age class data for this reason is unlikely in the countriesinvolved,except that a bias,based on emigration,in the 15-34age classis possiblein the Philippines.

BURNHAM O. CAMPBELL

Philippines 11'C ?-,

140 "_

100

_ ..o--.-e

__-_._

t 90 . rr

130

....

12o_

IO0

- 90 =

50--

I 1950

_. _ . I " 1960

" I. • 1970

I _ 1960 Year

1990

2000

'2 0

110. c=i 2020

2010

Singapore 120 -

--150 /

100

130 _

90 110 -

12o _ KR

_ 140 "

70-

lOO

60 -

50 -

__--_%.

' 30 20

• _ I 1950

I 1960

I 1970

_ " 1980

I 1990

I 2000

50 60

_ ..... 2010 ,2020

Year

i ....

UN mediumpopulatlonfotec_dlts_OCfore¢asts(based on 1955-1980 relation)

Note:Crudofel'tilityrateistheratlootthepo_ulatlon0-4 at one pointin limeto the female population15-34 five yearsbefore,

*Sources:DerivedfromUN(1980).

FIGURE2 FERTILITYRATESANDTHEKEYRATIO,ASEANMEMBERS ANDJAPAN,1950 TO 2015

AGE STRUCTURE

AND

DEVELOPMENT

Figure 2 (Continued)

Indonesia 120--

,- 150

110-

140

1o0 _

_30 _. n-

==90-

_2o_

80_

110¢

7o-

1oo

eo-

6o ,_

50_

80 o

40 --

70 I 1950

J 1960

• 1g70

I 1980

1 1990

I 2000

) 2010

2020

Year

Malaysia

::r, CF I00

130

_ |

• o

8o F

11o _;

"0 `'_`0"

.o F

,oi

-- 70

,of

-- 60 I 1950

1960

1970

1980

1990

2000

2010

-,o<>i 2020

Year

UN medium poilulation Io,'ecasts....

EIOCtc4_ClB_ (h=Sed 0¢11955-1180 mkiilON

Note; Cruile terlllity lille is the mtlo el the eoe_ion 0-4 lilt One !0oinl in lime tO the tenile polxdilior115-34 tlve years belo,re,

-5oumes; Oedved from UN (llieO),

BURNHAM O. CAMPBELL

Figure 2 (Continued)

Thailand 120

150

==, _.._

•110 i 1 O0 0.g0

__*'r"

130 140

....,..l,

CFR

, "" "O._

110

701

100

70 1 1960

1950

_ e O.--

120

801

_ -,_"

I 1970

1 1 g80

1 1990

I • 2000

i 2010

2020

Year

Japan 9(_

• '

120

i :¢

E oo=* U

eo =

_".o

70 ,,=

60 •

2 (3

- ..O- -- "-O,,_

20 i 1950

,..o..,._ .o. J

lg60

1970

1980

I 1990

I 2000

- --o

I 2010.

2020

Year

....

u N medium population forecasts" 90C forecasts.(based on 1955-1980 relation)

NOte; Ctu(:le tenillty rate iSthe ratio of the populatlon 0-4 al one point in time to the female population 1S-34 live years before,

*Sources; Derived from UN (1960),

AGESTRUCTUREAND DEVELOPMENT

and weakest in Japan. For pooled ASEAN data, the relation is as expected in 14 cases and the opposite in 6 cases. Four of the six exceptions came in periods and were of a direction (an increase in the key ratio matched by an increase in fertility rates) that suggests the possible influence of reduced infant mortality. On the other hand, no adjustment was made for the long-run downtrend in fertility rates noted by many observers. If, as this evidence suggests is possible, fertility rates do depend on the relative income-labor market experience effect, then the decline in the key ratio underway in Singapore since 1975, beginning in 1980 in Thailand and Malaysia, and that will take place from 1985 on in Indonesia and less dramatically in the Philippines, will lead to an increase in fertility rates, from the dates indicated, in all of these countries. Although the response may not be as great as in the past, if it is as just discussed, then the UN forecasts of the 0-4 population from 1980 on will be erroneous. As a reflection,of this, so will the forecasts of the 15-29 population and thus the key ratio from the year 1995 on. Alternative estimates of future fertility rates, and with a lag, of future key ratios based on this historical experience, done simply by inspecting the graph, are given to suggest another possible set of outcomes. Although this result is subject to all the qualifications discussed above, the mere fact that there exists an apparent inverse relation between the key ratio and crude fertility rates, and hence births, suggests that this is a topic deserving of careful further study. 6. Age Structure Changes and Country Developments: An Example Just as for cross-country comparisons, a multi-equation or general equilibrium model is required to correctly measure the contribution of age structure changes in one country over time. However, it is unlikely there will be as much variation from one five year period to the next in the other determinants of the variables being explained within a country as there is at any time across countries. Thus, the contribution of age structure changes to changes in these Variables may be more easily traced for individual countries over time. In any event, to illustrate what .the effect of age structure changes would be if all the hypothesized relations noted in Table l hold, the effect of changes in age structure alone on economic,

BURNHAMO. CAMPBELL

demographic,and socialoutcomes in Indonesiaisbriefly discussed below.The analysis isbasedon therelative growthratesofthe15-29 and 30-64 age classes shown in Figure3. Comparing thesegrowth ratesprovidesa visually descriptive way ofpresenting changesinthe key ratio. Ifitisassumed thatIndonesiawas moving from growth stagel to 2 in the 1950s,was beginningto move from stage2 to stage3 from the mid-1960sthroughthe 1970s and isnow about there,then therewould be arelative income effectfrom the late 1950son inIndonesia. Growth would havevariedwiththe15-29age class throughmost of the 1960s,withtransition toa 30-64ageclass constraint underway inthe1970sand dominatingthefuture. From 1950 to 2015, thegrowth in the 15-29age classswings widelyaround thegrowth rateofthe 30-64ageclassinIndonesia, accountingforthe observedlongwave in thekey ratio. From 1950 to 1970 the 30-64ageclass grew much more rapidly thanthe15-29 group,which,ifagestructure was theonlyvariable involved, should have causedan increase in therelative wagesof the 15-29ageclass and servedto discourage laborintensive industrialization intheearly part of this period.Thus, the thrustof age structurechanges, combiningthe CE and RE, would have been to increase average productivity andper-capita income(bythecomposition effect only), stimulateskillintensive and possiblycapitalintensive production relatively speaking, raise fertility rates, discourage emigration, reduce householdincome inequality, increasethe nationalheadshiprate, and decreasesuicide ratesand thepolitical alienation of theyoung, toname two of thepossible social outcomes. With a largejump inthegrowth rateoftheentering population, from zeroto almostI0 percentbetweenthefirst and secondhalves of the 1960s,thechangesbroughtaboutby a falling key ratiowould have sloweddown and then haltedin the first halfof the 1970s. Then, in the secondhalfof the 1970s and in the first halfof the 1980s allthe above outcomes would be reversed, withtherelative and possiblythe absoluteincomes of the young decreasing. The initial impactof theabsolutely and relatively high15-29growthrate would have been sharplyincreasing underemployment(unemployment,althoughlessimportantintheIndonesian context, would also increase). If these age structure changes alone are considered, the result would be decreased average productivity and per-capita incomes, a stimulus to less capital intensive and skill intensive production, reduced average headship rates, lower fertility rates,

AGE STRUCTURE AND DEVELOPMENT

;ndoeesia 20

_ ,--"_"-.

30-6,_

A /

'_t- ,"_" "_'.-/

"L.=_.t_

,.3¢ r- 1950.65

Malaysia

Based on UN birth rats Iocecasts

1960 S5 _0

Years

-15

35-_ased 0¢=UN b)_

_.,° 10 _-

,o.j_ "_ - "°-'/'--JZ-

,._

.'"

_ ._.. \

+ Philippines

20 F .=_ 15_--

A15-29 _

Years

30/2T

ea,ed o. UN b,,,.

f _ou oO

e_ O_

"2g

1970-75

19,EIO--S5 _0"15 Years

/P_

Sa'_'d °n u_ airth ra'e ,0 ....

_°_-_o--,\ / sI

,._, -.-0-- I *_,.

°°k Singapore

_".

/ t'/

J'at_fOre.sis

..o_._o i

• •

Yea:$

000-05

_010-15

SOurce_: Derived from UN (1_80).

FIGURE3

GROWTHRATES OF THE 15-29 AND30.64 AGECLASSESIN A_EANCOUNTRIESANDJAPAN,19S0-1955 TO 2010-2015

BURNHAM

O. CAMPBELL

Figure 3 (Continued)

25-

Tha,and 2C.

;

-,.. : IB_t_edon UN birth rate Ioreca_;tB ,,,_...o_.._.___._ .."\ i "'o-,-o..

/ ,_3o_, _

, L

[

1950-55

1960-65

I..,

1970-75

" I

1980-85

]

1990-95

2000-05

I 2010-15

,YearS

25 A-EAN$

*I B86e rl On UN birth rate fors¢_stl; "

g- s

1 ,9V

0 i

I 1060-65

......

. . . [ i 1990-95

I 1980-85

I 1970-75

i 2000-05

I 2010-15

Years

15 Japan

...f"

=_lO == s L_ C ¢

O 30-64 _''O,_

' "_i._'___

_,_ _ _r_5_29

_ ' ; i Baaed on UN birth rete iorecasts

-,,. _o_

p,,

•

0 -

"°""

-10 -

""o"'_

J t m

I 1950-55

J 1960-65

I 1970-75

I 1980-85

1990-95

2000-05

I 2010-15

Years Sources: Derived trom UN (1980),

AGESTRUCTUREAND DEVELOPMENT

more household income inequality, higher suicide rates and increases in other indices of social instability. In the second half of the 1980s, the job market would once again turn to the advantage of the 15-29 age class, with this advantage increasing especially rapdily from 1990 to 1995. At that time, demographic variables will be favorable to the completion of the transition to growth stage 3. However, the further demographic stimulus to increased productivity, more capital and skill intensive production, income equality, increased average headShip rates, and social stability implied in Figure 3 down to 2015 will not occur if the response in fertility rates to changes in relative income in the next decade is the same as this response was in the past (see Figure 2). The foregoing catalogues the changes in various economic, demographic, and social indicators that would have occurred or will occur in Indonesia if changes in age composition alone determined and will determine the outcome. However, since other things were not equal and will not be equal, it should not surprise anyone to find the actual outcomes opposite to those expected on the basis of age composition changes only. For example, Sukarno's policies and the disruptions surrounding and following his removal had a major impact on what occurred in Indonesia in the 1960s and on how the 1970s compare with the 1960s. Even so, the record of possible effects given, when it is considered that age structure changes affect the entire economy and can be easily forecasted, should be sufficient incentive to carry this research further. And the fact that the changes in one of the variables dependent on the key ratio, fertility rates, occur as hypothesized in Indonesia should provide encouragement for the task. The reader can trace, using Figure 3, the possible impact of experienced and predicted changes in age structure (key ratio) on the other ASEAN countries and Japan in the same fashion as was done for Indonesia. In all, there are interesting conclusions to be drawn. For example, taking the ASEAN countries, they all have gone through a period and, except for Singapore, will for a while continue to go through a period of more rapid growth in the 15-29 than in the 30-64 population. In all, but least in Singapore, evidence of the expected effects of the consequent decrease in the key ratio can be found. And all will, in the 1990's, experience the opposite pressures resulting from a relative shortage of young compared to established workers. This situation will reverse again in the early part of the 21st

BURNHAMO. CAMPBELL

century if the above arguments about the key ratio and fertility rates are correct, setting up a demographic-based repetitive long-swing. Finally, the stimulus to increased capital and skill intensive production, to increased income equality, and to social stability in Japan coming from changes in age structure alone over the last thirty years is noteworthy as are the changes, especially the decline in the growth of the 30-64 age class, just ahead. 7. Other Age Related Outcomes: Education and Dependency Burdens in ASEAN and Japan Two effects of changing age structure not measured by the key ratio are the educational burden and the dependency burden. The education burden can be measured in various ways but to roughly illustrate how countries rank in terms of this burden and to indicate the changes that have been and will be occurring, the 5-24 age class is related to the 30-64 age class in Figure 4.17 The former is taken to measure the school age population and the latter the population providing most of the taxable surplus or private spending used to support a country's educational effort. Measuring the educational effort in terms of this ratio implicitly holds the quality of this effort constant over time. When the ratio increases the quality may be reduced, with long-run implications, to keep the burden relatively constant, and vice-versa. Since there is no way to monitor how each country responds to the pressures involved, this aspect is simply ignored in the following discussion. In addition to the total educational burden, the higher education (high school plus college) burden, measured by the ratio of the 15-24 age class to the 30-64 age class is also given for each country in Figure 4. Through 1985 the educational burden Clearly follows a longswing pattern in all the ASEAN countries. However, the pattern differs with Indonesia's swing moving inversely to and generally at a lower level than Thailand's, Malaysia's, and the Philippines'. Singa17. This ratio "measures" the burden, given quality, of the current or operating costs,includingthe opportunity costs of keepingthe 15-24 age classin school, of education. The share of available investment funds required to maintainplant and equipmentin educationat a givenlevel wouldbe "measured" by the ratio of the change in the 5-24 populationto the basicpopulation. Since education as used here covers pre-school to graduate school, there will be a composition effect within the 5-24 age group. This is also ignoredin the rough index presented.

AGE STRUCTURE

200

AND DEVELOPMENT

IjKase(:lonactuStelataorforeoB6tMe4)i_l of people already born (19._0-1_R_O)

190!

-d"

\¢'"

" .e/ _50 Mo'" 140

(_ SSS-201S)

%.. " .yr"

_-_

{"_s

",o ,,

,; _.-

: :

"\:_=

_ I_.

'-,! -

S e...4v"

"x-x ¶ \

",,,-'\:", ", •

_. : _-

1,o o.-._

,,,',_,.,

',\

\AN

",X,"X

_,,

\ \

_\ %',.

_,:

,,o',,

"...

,. __ 'j _

-k 90-

\ '_

100--

sO-

Bale¢l on UN birth rate toreca_i a.u forecast agi.g of existing pop!dlatiOn

_%.._,

,-,.. _,.-_ 160

-_\ •

_'_

Indonesia

"e Malaysia

-.._,..

'__'"

''O I

50_

', "..

;

60-

'%."$ Thailand "4

70-

'o Philippines

x.-,,.

"t" - .,._.B

_,__e,_

Japan

- "'m SingaPore.

J ,

I 1960

1 1950

I 1970

I ', 1980

I 1990

I 2000

I 2010

I 2020

Year

;::r - .... o=0..o ..... o.o._0oou_ 0._.o o_.o0 of people slready born (lSS0-1980)

160

A _,. ,.,,

_'_

_.., _°_

lot°cast aging of _xlsttng population

. 11985"20151 ._e,_. _--,_''_ ""_"'_, "e_ .,,

! ,_o[-

""x,

120 F

_..

60 80 f

Japan I t 950

i 1960

I 1g 70

A , 1980

I 1990

I 2000

1 2010

ASEAN I 2020

Year 6ource: Oetlved Irom UN ( 1_).

FIGURE 4a TOTAL EDUCATIONAL BURDENS IN ASEAN MEMBERS AND JAPAN: POPULATION 5-24/POPULATION 30-64

BURNHAM

10C

Burdenbased on actualdata oraging of i_ople already born(1985, 1990,1995)

pOpulation

.,D'T_t__ e

8o T--._o % -

--M

._,'._&_..._.._=--_,

----I _

j __e.,._.....4jp. .......

_ 60

SO-_EL_--

Burdenbased on UN forecasts Of births aS well as aging of existing

I 0,-,., _

O. CAMPBELL

/..'_ ", _

. ,s.._,.,t....,...,.;..."

"t...._.__o.-,

", I _'_ Thailand • '.,_ ' "l Indonesia ,, -,_. Malaysia = F'__--...e-.---_I ".e."'"L.._.,O ".-_Philippines. _ _ Japan _,_e_ _i' Singapore

50-30 --

,.._-'-w_.....

20I 1950

t 1960

I 1970

I 1980

I 1 1990

I 2000

I 2010

I 2020

Year

90BurOenbased onactual date Oraging of people already born(1985.1990,1995) •q0 -- A IP _.--S_._

7060-

\,,___r.,,_"

Ii '

Burden based on UN forecastsof births as well asaging ofexisting

population Burdenbased on past

e',_._e_

"_'_ -

4C -

3C I 1950

I 1960

I 1970

-_B-.,-_i__j.lj I I I 1980 1990 Year

'N"'_e ASEAN

I 2000

I 2010

Japan I 2020

Source: Derivedfrom UN (f 980).

FIGURE 4b HIGHER EDUCATIONAL BURDENS IN ASEAN MEMBERSAND JAPAN: POPULATION 15-24/POPULATION 30-64

AGESTRUCTUREAND DEVELOPMENT

pete's burden is much less than and peaks considerably ahead of the others. The experiences of Thailand, Malaysia, and the Philippines are very similar, with the education burden trending upward from the 1950s to a long-swing peak in the second half of the 1970s. Clearly, if they all had the same real income per capita (for the 30-64 age class), the burden of educating the young would be much greater and the resources available then for other uses much less, at least from 1960 to 1980, in Thailand, Malaysia, and the Philippines than in Indonesia and Singapore. By 1980, the last year in which birth forecasts are not involved, Singapore's burden will be far less than for any other ASEAN country. Since Singapore's real income is much higher than the others, on a per capita basis, the relative advantage of Singapore is much greater than indicated as is the actual improvement in Singapore's position from 1975 on. The contrast in educational burden between Japan and ASEAN is equally extreme given Japan's much higher per capita real income and much lower relative burden. Japan's burden has steadily fallen relative to that of the ASEAN countries except Singapore after 1970. This must have made growth easier, especially skill intensive growth, in Japan over the post-war years than in ASEAN and by analogy, in Singapore than in the other ASEAN countries. In fact, even if the overly optimistic UN forecasts of birth reductions take place, it will not be until the year 2000 that the education burden in Malaysia and IndoneSia falls to the 1950 level in Japan. Thailand's burden will not reach this level until 2005 and the Philippines until 2010! If all these countries acted as one nation, government policy might well be to tax the Japanese "province" to share the burden of education in the other "provinces" since this redistribution would seem both equitable and likely to raise the long-run real income of all "provinces" A similar relative picture emerges for the higher education burden. Japan has been and for the rest of the century will continue to be much more favorably placed to afford high quality higher education (or mass higher education) than the ASEAN countries, again except for Singapore after 1990. And, with this exception, none of the ASEAN countries will reach Japan's 1950 burden until the year 2000, a fact that shows how much more difficult it will be for these countries to meet the educational challenge brought about by a rapidly changing high technology world than for Japan. In this context, the fact that the long-swing in the higher education burden for ASEAN as a whole moves very nearly inversely to that of Japan

BURNHAMO. CAMPBELL

suggests that cooperation, at a relatively low cost to Japan, to share this burden would have been possible in the 1970s and will be • feasible in the 1980s. Figure 5 gives the dependency burden, measured by the ratio of the dependent population, defined here as 0-14 and 65 and up, without adjustment for the different costs of maintaining individuals in these two groups, to the 30-64 age group, taken again as primarily shouldering the tax or direct burden of dependency. The time pattern of changes in the dependency burden across the countries included is similar to that of the education burden with the peaks coming earlier since the dependency ratio is dominated (except for Japan after 1990) by changes in the 0-14 age class. To the extent that the savings ratio moves inversely to the dependency ratio, is Japan's much higher savings ratio could be explained by this source alone. On this basis, Japan's savings rate would have been expected to increase dramatically between 1950 and 1975 and to be much higher than for any ASEAN country through 1980. From 1990 on, because of the rapid increase in the population 65 and over, Japan's dependency ratio will increase continuously, surpassing the Singapore and Malaysian ratios in 2015. However, Japan's burden will remain •very low by past standards and, even in 2015 and with some very optimistic ASEAN birth rate forecasts, below• the ASEAN average. 19 The dependency burden for Malaysia, Thailand, and the Philippines increased sharply in the first two postwar decades and has remained high in the latter two countries. Remarkably, this ratio slowly improved in Indonesia over the same period and by 1975 was much lower than for the other ASEAN countries except Singapore. If births follow the UN's expectations, dependency ratios will fall sharply for all ASEAN countries through the remaining years of the 20th century, stimulating increased savings and growth. However, to put these changes in perspective, the dependency ratios for all the ASEAN countries except Singapore and for ASEAN as a whole will not fall to Japan's 1950 level until 1995 or later. Thus it will be some time until •these demographic factors offer as much stimulus to 18. Conclusions/based on the dependency ratio ignore the concurrent effects of changes in the key ratio on savings.Although this effect is not a priori predictable, this does not mean it is necessarily negligible. 19. For an analysis of the economic and demographic effects of the future changesin Japan's age structure see Ogawa, H. (I982).

AGE STRUCTURE

200

/ Mif

A //

150

14o

"'-_._' "_ •_

"_ ,.

,'_, ' ,

_, \.% "it %'\ ' _..,_ " ,\_

_. : ,.,: ,., : q ,I .,., !,,

..." s--

_ g 13o J= _120 \ 110 "_

\ _,x ,_..,_.. ,,-,_.¢.. ,% _ ' i. .. ,\v. _,,:_, ""-,.

,'-,

_o090 --

I :

_ "%. Ill

"_.% e. "-

"_I_ ':,,,, I Thailend "_---Indonesia

"',=,.,.

.1,. 1950

180 -A _

1960

1970

: L

-e"_'"e..

1990 1990 Year

BurdenI_tGedOnpestdett_ -4-- _ _ "-4.-.

160 -

,I 2000

t._..al)

Singapore

2010

2020

Bumeabassoon UNbirthIorecastsaS _II_6|=_Icist llglng ofsxlstlng populallon

"e"_ "O"

".=e...

140--

120

'"_1) Philippines

\_._

9o-

•

Buqltenha.sodonpa_t0ata :_ Burdenbatedon UN birthforecasts_ ..-4k_ I welletlforeclst8glngofexistingpopu_tlon _h_' ._e. -,...e

190

180

AND DEVELOPMENT

"_"_

J_._

100

_ • [ 1950

"%e,.,. ,...e,_ "_ ll" _--t ASEAN

Japan I 1960

I 1970

1980 Year

i 1990

I 2000

t 2010

,l 2020

Source:DOrive¢l fromUN(198o),

FIGURE S DEPENDENCY RATIOS IN ASEAN MEMBERS AND JAPAN. 1950 TO 2013 (Population 0-14 Plus 65 + 1 Population 30-64)

BURNHAMO.CAMPBELL

savings and growth in ASEAN as they did in Japan 40 years ago! And, if births follow less optimistic forecasts, dependency ratios in ASEAN will swing up again in the late 1990s. 8. Summary One important goal of this paper was to show that significant shifts in age composition have occurred and will occur in the ASEAN countries and Japan. Based on the key ratio, the ratio of the entering labor force population, 15-29, to the established labor force population, 30-64, and on the education and dependency burdens, the evidence is clear that such shifts have taken and will take place. A second goal was to trace the potential impact of these age structure changes on the economy and society of the ASEAN countries and Japan, assuming a set of hypothetical relations between age structure changes, as measured by the key ratio, and various economic, demographic, and social outcomes. This was done for Indonesia very briefly and left for the reader to do for the other countries involved. The Indonesian "case study" led to the conclusion that past and future changes in age structure could have led and could lead to wide swings or long waves in per capita incomes or the growth rate in per capita incomes, the skill and possibly capital intensity of production, household formation rates, emigration and immigration rates, fertility rates, and suicide rates and other indicators of social instability. Similarly interesting conclusions emerged from the analysis of education and dependency burdens, especially the inverse relation between Japan's and ASEAN's burdens on both indices. The low educational and dependency burdens Japan enjoyed must have given Japan a distinct advantage in providing resources for infrastructure, plant and equipment, etc.., an advantage that Japan will maintain (joined by Singapore) over the remainder of this century. The ramifications of the conclusions reached are wide and it is entirely possible that the relative economic performance of Japan versus the ASEAN countries or the ASEAN countries versus one another could not be fully explained or adequately forecasted without considering the impact of age structure and changes in age structure. And it seems unlikely that the past or future performance of these countries individually can be adequately understood or forecasted without considering the long-swing apparent in the key ratio and in the education and dependency burdens of each. For one

AGE STRUCTURE AND DEVELOPMENT

example, future fertility rates in several of the ASEAN countries and in Japan will be different from UN forecasts if they Vary with the key ratio in the future as they have in the past. Even if this result is only partially correct, ignoring age structure changes could lead to major forecasting and planning errors. The significance of these possible outcomes highlights the final and major goal of this paper, the encouragement of the research necessary to test the hypothetical relations on which these outcomes are based and to empirically determine the contribution of age structure changes. 2째 A VLTStstep would be to develop a more sophisticated to carry out and cultural

theoretical framework than that used in this study and country by country investigations of the institutional constraints that influence the relations involved. The

implications for the effects of age structure changes thus derived would then have to be empirically tested in a model sufficiently complete to allow holding "other things equal". The model shOuld also make possible the explanation of past and the stimulation of future feedbacks from age structure changes through economic, demographic, and social changes to age Structure changes. Obviously there is much to be done and completing bits and pieces of the mosaic can be very helpful. Perhaps the many conjectures advanced in this paper will encourage the doing.

REFERENCES Abramovitz, M. (1961), "The Nature and Significance of Kuznets Cycles," Economic Development and Cultural Change, : 225-248. Abramovitz, M. (1968), "THe Passing Of The Kuznets Cycle," Economica, 35:349-367. Campbell, B. (1966), Population Change and Building Cycles, Urbana: Bureau of Economic and Business Research, University of Illinois. Easterlin, R. (1968), Population, Labor Force, and Long Swings in Economic Growth: The American Experience, New York: N.B.E.R. Euterlin, R., (1978), "What Will 1984 Be Like? Socioeconomic Implications of Recent Twists In Age Structure," Demography, 4: 397-421. Easterlin, R., Wachter, M. L., and Wachtet, S. M. (1978), "Demographic Influences on Economic Stability: The United States Experience," Population and Development Review, 4: 1-23. 20. One research area not touched upon that could be significant is the contribution of age structure changes to the dynamic comparative advantage and hence to trade relations between ASEAN countries and Japan.

BURNHAM

O. CAMPBELL

Ogawa, H., (1982), "Economic Implications of Japan's Aging Population: A Macro-Economic Demographic Modeling Approach, International Labor Review, 1: 17-33. Oshima, H. (I 980), "Perspectives on the Prospects for Southeast Asian Growth in the 1980s," Journal of Philippine Development, 14:89-116. United

Nations (1980), World Population and its Age-Sex Composition by Country 1950-2000; Demographic Eatimation and Population as Assessed in 1978, New York,: Population Division, UN Secretariat.

FOREIGN TRADE AND ECONOMIC WELFARE IN THE LAST HALF-CENTURY OF SPANISH RULE Amado A. Castro*

When the galleon Magallanes arrived in Manila from Acapulco in 1815, having departed from Manila on its outward voyage in 1811, it not only marked the end of over two centuries of Spanish mercantilist restrictionism through the galleon trade monopoly but also set the stage for an era of more liberal Spanish economic policy in the Philippines. In Spain itself, there had been stirrings of economic liberalism in the last half of the seventeenth century. The Bourbon kings had taken over the Spanish throne from the Hapsburgs and the French antecedents of these new monarchs had exposed them to French ideas of liberalism not only in politics but also in economics. The British occupied Manila in 1762-63 and they discovered commercial possibilities in the archipelago. Thus, in the last quarter of the eighteenth century, there had been a flurry of activities in an attempt to trigger the economic dewlopment of the colony. For example, in 1779, less than a year after he took office, Governor Jose Basco y Vargas presented an economic development plan for the Philippines. This was not the first plan however because there had been previous plans by other Spanish officials such as Viana and Simon de Anda. In 1781, Basco founded the Economic Society of Friends of the Country (Sociedad Economica de Amigos del Pals) and in 1781, he organized the tobacco monopoly which made the colony financially *University of the Philippines School of Economies. I am indebted to Mrs. Celia Reynes Salazar, Mrs. Loreli Cataylo-de Dios and Miss Gilda M. Rojas for assistance in the preparation of th_ paper. 95

AMADOA. CASTRO

stable. In 1 785, the Royal Company of the Philippines (Real Compania de Filipinas), a trading and development company, was chartered under part ownership of the Spanish crown. Early in the nineteenth century, Spain started to lose her colonies in the new world as national wars of liberation were waged. She was left with only her colonies in the Antilles and in the far Western Pacific (Cuba, Puerto Rico, and the Philippine Islands, plus other scattered islands in those areas) and her attention was turned perforce to these hitherto neglected colonies. In the Philippines, writers such as Manuel Azcarraga y Palmero (La Libertad de Comercio en Filipinas) and Tomas de Comyn (Estado de las Islas Filipinas) began espousing freer trade and more vigorous economic development policies. Quotations from the writings of the Frenchman Jean Francois Galoup de la Perouse and the last Spanish Governor in the eighteenth century, Don Rafael Aguilar, doubtless offered consolation. In 1787, after a three-month visit in the archipelago, de la Perouse had said, "I will not be afraid to assert that a very great nation, which should have no other colony than the Philippines and which should establish the best kind of government that could be constituted, might, without the least envy, behold all the European settlements in Africa and America". Writing in 1973, Aguilar predicted that in three or four years, the Philippines would become ,the most valuable colony in the world" for "the King's dominions here are so extensive, so valuable and so productive that there are no limits to their possibilities". Generous words indeed, but how realizable were they? Unfortunately, the ascendancy of liberal ideas in Spain was not assured and Spain in the nineteenth century saw a distressing alternation of liberal and Conservative regimes. The unsettled political situation prevented the sustained pursuit of enlightened economic and political policy not only in Spain but also in the colonies. Nevertheless, the new, more liberal orientation also reached the Philippines. There was, for example, the Royal Company of the Philippines, which, although a failure, had some worthwhile activities during its existence. In trying to enlarge the Company's sources of supply and markets, foreign ships were allowed to call in Manila starting in 1789, provided they carried only Asian cargos. Thereafter, a small but steady stream of foreign ships came to transact business in the port. In 1976, the American ship Astrea did stop at Manila harbor, the first known

FOREIGNTRADE UNDERSPAIN

direct sailing between the United States and the Philippines. On October 25, 1813, the suppression of the galleon trade was decreed by the new King Ferdinand VII and with that, a rival enterprise which had competed with the Royal Company for the •attention of the officials in Manila ended. Despite the disappearance of rivalry as well as the reorganization in 1803, the Royal Company continued to flounder. Its doom was probably sealed by excessive lending to the crown for royal military activities. Without doubt, however, inept management of the Royal Company also contributed greatly to its failure. 1 Thus, on September 6, 1834, the Royal Company was fimaUy dissolved. On that same day, the port of Manila was officially opened to all foreign traders, Previous to this, in the 1810's, English traders and in 1820's, Americans such as Peele, Hubbell and Company and Russell and Sturgis were already actively conducting business in Manila, fomenting native agricultural and industrial v.entures in the process. The ftrst foreign consul, an American, was assigned to Manila in 1816, and he was followed by the Consuls of France (1836), Belgium (1842), and Great Britain (1844). It was not till the second half of the nineteenth century, however, that commercial development and, as a consequence, agricultural expansion took place in earnest. A signal event was a royal decree on September 29, 1855 which opened three additional Philippine ports aside from Manila to foreign traders. Of the three Iloilo, Zamboanga, and Sual in Pangasinan - only the first is significant today for Philippine foreign trade; Sual was a rice transhipment port for interisland traffic which a few years after its opening silted up and was not used again. Subsequently numerous •other ports were opened and at the end of the Spanish regime, some •63 ports were open for foreign trade. The inauguration of the Suez Canal on November 17, 1869 also stimulated Philippine foreign trade as it made Spain and the rest of Europe closer to the Philippines. The advent of the steamship further shortened the Manila-Europe voyage, and the industrial revolution in England and the British rise to commercial and political power spurred not only world trade but also Philippine commerce. Other signs of increasing commercialization were the establishment of the Banco Espafiol Filipino in 1. Benito Legarda y Fernandez (November-December, 1967), "The PhilippineEconomy under SpanishRule," Solidity, pp. 8-11.

AMADO A.CASTRO

1851, the growing activities of foreign merchant bankers (American and British), and the opening of branches by the Chartered Bank of India, Australia, and China (1873) and the Hong Kong and Shanghai Banking Corporation (1876). The growth of Philippine foreign trade in the nineteenth century is shown in estimates as well as official records (Table 1). In 1810, Philippine exports were P4.8 million and imports were P5.3 million. The drop in estimated exports to P1.2 million in 1818 is easy to explain. The galleon trade had ended and the re-exports of Chinese and Indian goods carried on the galleon to Acapulco no longer dominated the statistics, although in that year, re-exports by private traders still came to 40 percent of recorded exports. In 1831, the exports had risen to P1.96 million or $2 million and in the decade of the 1840's, exports were close to $3 million. To avoid overstatement in values on account of the long term fall in the value of the Mexican peso (then the unit of value in the Philippines), the peso values can be converted to U.S. dollar equivalents, using the conversion rates between the Mexican peso and the U.S. dollar prepared by the U.S. Mint and published in the Census of 1903. In 1850, exports were valued at $3.694 million and began to rise slowly, reaching $6.4 million in 1855. In the next year, 1856, which was the year after the opening of the ports of Iloilo, Zamboanga and Sual, exports rose by 50 percent to $9.6 million. From then on, the trend was definitely upward, though somewhat unsteady. In 1870, an apparent peak value of exports of $29 million was hit, although Legarda has pointed out that this is not a reliable figure, z Again in 1889, exports reached $25.6 million) The values of exports were no doubt affected by prices. Thus, an attempt to measure the change in volume of exports is presented in the quantity index in Table 2. The index is based on export volumes for the seven leading Philippine exports in the second half of the nineteenth century and uses 1876 as the base period. A complication in the preparation of the index was that one export, coffee, tapered off in the late 1880s because of a blight which hit the coffee plantations of Batangas. Coffee finally disappeared as an export at the end of the century. At about thesame time, however, a new export, 2. Benito Legarda y Fernandez, Foreign Trade, Economic Changeand Entrepreneurship (unpublished Ph.D. dissertation), p. 185. 3. Ibid., pp. 196-198 and Table 1

Table I - Growth of Philippine Foreign Trade in the Nineteenth Centmry

TOTAL IMYEAR

IMPORTS Pesos

1831 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856

1,249,148 2,060,143 2,710,456 2,153,247 1,844,424 2,252,997 2,856,096 2,191,685 3,309,312 3,934,824 2,639,494 3,429,931 3,149,164 2,443,215 3,178,249 3,301,334 3,951,333 4,004,530 3,756,345 4,243,602 6,959,254

Dollars 1,208,676 2,112,471 2,777,133 2,238,516 1,917,463 2,329,824 2,924,357 2,233,546 3,390,721 4,013,127 2.696,248 3,525,283 3,226,633 2,513,091 3,286,309 3,468,382 4,116,894 4,241,598 3,978,721 4,459,177 7,312,784

EXPORTS Pesos 1,185,009 1,956,754 2,894,068 2,674,220 2,475,942 3,36¢_,734 3,073,580 2,923,795 3,242,392 3,020,717 2,972,967 3,126,141 2,975,807 3,723,921 3,573,067 •4,172,274 5,016,313 5,778,676 6,352,348 6,121,622 .9,133,317

PORTS AND EXPORTS

Dollars

1,146,615 2,00.6,456 2,965,262 2,780,119 2,573,989 3,481,540 3,t47,039 2,979,639 3,322,155 3,030,829 3,036,886 3,21.3,048 3,049,012 3,830,425 3,694,551 4,383,391 5,226,497 6,120,774 6,723,407 6,432,600 9,597,290

2,355,291 4,118,927 5,742,395 5,018,635 4,491,452 5,811,364 6,071,396 5,213,185 6,712,876 7,093,956 5,733,129 6,738,331 6,275,645 6,343,516 6,980,860 7,851,773 9,343,391 10,362,372 10,707,128 I0,891,777 16,910,074

EXCESS OF EXPORTS OVER IMPORTS Dollars 62,06la 106,015 a 188,129 541,603 565,526 1,151,716 222,682 746,093 68,566 a 932,298 a 340,643 312,235 a 177,621" 1,317,334 403,242 915,009 1,109,603 1,879,176 2,749,686 1,973,423 2,284,506

Z _e

_e >. Z

Table 2 (Continued)

YEAR

IMPORTS Pesos

Dollars

EXPORTS Pesos

Dollars

TOTAL IM째 PORTS AND EXPORTS Dollars

EXCESS OF EXPORTS OVER IMPORTS Dollars

1857 1858 1859 1860 1861

9,907,299 5,798,720 6,271,560 8,739,474 " 1O,148,160

10,532,4.50 l1,895,821 6,093,295 9,416,975 6,701,789 9,082,868 8,286,565 9,509,481 10,650,494 8,065,530

12,643,447 9,895,357 9,705,953 10,104,775 8,464,774 .

23,178,897 15,988,652 16,407,742 19,391,340 19,115,268

2,113,997 3,802,062 3,004,164 818,210 2,185,720 a

1862 1863 1864 1865 1866 1867 1870 1872 1873 1874 1875 1876 I877 1878 1879 1880

6,841,735 7,465,063 10,901,584 17,870,523 17,711,791 15,180,853 23,500,000 22,163,142 13,21.7,836 13,704,254 12,215,153 11,987,162 19,535.,864 I7,292,84.7 18,03.1,547 25,493,319

7,341,579 7,889,079 11,520'794 18,789,068 18,634,575 15,841,220 24,522,250 23,023,072 19,478,227 13,770,034 11,924,432 10,964,657 18,449,670 15,674,237 " 15,923,659 22,938,888.

9,625,003 10,628,045 11,262,345 22,008,554 23,338,232 22,964,100 29,218,000 17,068,164 23,985,923 17,386,031 18,470,168 13,572,132 15,452,692 15,835,084 16,614,159 2t,100,566

16,988,582 18,517,124 22,783,139 40,797,622 41,972,807 38,805,320 53,740,250 40,09t,236 37,464,150 31,156,065 30,394,600 24,536,789 33,902,362 31,509,321 32,537,818 44,039,454

2,283,424 2,738,966 258'449_ 3,219,486 4,703,657 7,122,880 4,695,750 5,954,908 a 4,507,696 3,615,997 6,545,736 2,607,475 2,996,978 a 160,847 " 690,500 1,839,322 e

9,100,797 10,056,818 10,657,026 20,932,617 22,182,523 22,006,804 28,000,000 16,430,655 23,522,529 17,302,977 18,920,475 14,837,796 16,36.2,444 17,470,305 18,813,452 23,450,285

._ 0 U'3

.-.]

1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893

20,777,209 21,260,762 21,308j074 21,246,241 19,199,468 20,073,598 17,530,198 21,208,482 24,790,906 19,797,257 21,647,280 23,817,373 25,922,515

18,485,483 18,970,978 18,584,902 18,579,838 16,027,716 15,689,524 13,491,240 15,660,343 18,216,358 16,277,305 I6,804,783 16_310,137 15,893,094

24,579,006 20,673,333 26,380,727 22,672,833 24,553,885 25,721,032 25,257,139 26,293,271 34,926,969 26,213,554 26,905,102 27,976,569 35,275,566

21,867,942 18,446,815 23,009,270 19,827,392 20,49.7,416 20,103,559 19,437,894 19,414,95.1 25,684,337 21,652,784 20,886,431 19,158,354 22,240,550

40,353,425 37,417,793 41,594,172 33,407,230 36,525,132 35,793,083 32,929,134 35,075,294 43,880,695 37,830,089 37,691,214 35,468,491 38,133;644

3,382,459 524,163 a 4,424,368 1,247,554 4,469,700 4,414,035 5,946,654 3,754,608 7,447,979 5,275,479 4,081,648 2,848,217 .6,347,456

1894 1895 1898 b 1899 1900 1901 1902

28,558,552 25,398,798

14,245,006 13,052,442 5,380,963 19,192,985 24,863,779 30,162,471 33,342,166

33,149,984 36,655,727

16,535_212 18,837,378 5,165,356 14,846,582 22,990,373 24,503,353 28,671,904

30,780,218 31,889,820 10,546,319 34,039,568 47,854,152 54,665,824 62,014,070

2,290,206 5,784,936 215,607 a 4,346,404 a 1,873,406 a 5,659,118 a 4,670,262 a

aExcess of imports over exports blrrce months, August to December, inclusive. Source Census of the Philippine Islands, Vol., IV, United States Bureau of the Census, Washington, 1905, pp. 564-565.

t.a

102

AMADO A. CASTRO

Table 2 -- Quantity Index of Major Exportsa (1876 = 100) Ye_

Expo_s

1854 1855 1856

35.9 30.2 40.2

1857 1858 1860 1861 1862 1863 1864 1865 1866 1867 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894

34,4 27.0 42.2 38.8 51.5 44.1 39.9 39.1 36.4 47.4 71.0 b 81.9 96.9 100.0 94,6 96.6 101.6 137.8 159.0 115.9 147.7 99.0 153.8 139.6 140.1 136.2 176.7 113.3 121.5 191.8 205.5 172.9

alncludes sugar, caw abaca, leaf tobacco, manufactured tobacco, coffee, and rice, and copra. b Raw hemp is included with manufactwed hemp, not having been separately zecorded.

FOREIGNTRADE UNDERSPAIN

103

copra, began to take the place of coffee. Another export product which dropped out was rice. It is clear however that exports as well as imports rose in the period under study. The pattern of Philippine exports and imports changed over the period, as is seen in Tables 3 and 4. Three years are chosen to illustrate the changes: 1856, which was the beginning of the period; 1876, the middle; and 1894, two years prior to the 1896 revolution. Table 3 shows the changes in the positions of the five leading exports as well as the directions of the trade. In 1856, sugar was already the leading export with England as the largest market, followed by Australia and the United States. For abaca, the second largest export, the United States was the largest outlet, absorbing 88 percent of total abaca sold. American merchant houses, Peele, Hubbell and Company and Russell and Sturgis, dominated this trade. The next largest exports were raw tobacco, rice, and manufactured tobacco. In 1876, sugar and raw abaca ranked number one and two respectively again, although the ranking of markets had changed. The United States was now the largest buyer of sugar and England of abaca. A new export, coffee, appeared; manufactured tobacco and raw tobacco exchanged ranks as numbers four and five. Rice, the subsistence crop, disappeared from the list of top exports. In other words, the subsistence crop was no longer exported in significant volumes and in its stead, cash crops - sugar, abaca, tobacco became dominant. A shift had taken place in emphasis on agricultural production. The Philippine economy had been transformed from a subsistence to an export economy. In 1894, further intensification in export orientation continued. The leading exports were still the same, although in this particular year, abaca earned more than sugar. The United States again was the chief market for abaca while China was for sugar. As has been mentioned, coffee was no longer exported in large quantities but a new product, copra, which was to become important in the next century, emerged. There were also changes in imports. In 1856, imports were only $7.3 million. There were large imports of gold and silver from China and copper sheets from England for minting. Apart from these, cloth muslin and cotton - was the largest item accounting for 11 percent of total imports. Otherwise, imports were of various sorts, largely items for the upper class of Manila, not one of these items, except for muslin cloth, came to more than 10 percent of total imports, There

104

AMADO A. CASTRO

Table 3 - Philippine Exports

Commodity

Value (pesos)

Rank

Percentage of X째 to total exports

Percentage of Country X to X째

1856 Total Exports

9,136,310.18

Sugar

3,526,297.55

39%

England Australia United States Raw abaca

44% 29% 20% 2,618,863.88

29%

United States England Tobacco, raw

88% 10% 1,081,234.00

12%

Spain Rice China California Manuf. tobacco Australia Singapore China

100% 352,427.73

4% 92% 7%

336,858.00

4% 32% 17% 15%

1876 Total Exports

14,721,746.00

Sugar U.S.A. England P. Inglesas

7,404,807.00

Raw abaca

50% 57% 32% 5%

974,308.00

26%

England U.S.A. P. Inglesas Coffee P. Inglesas

51% 37% 7% 1,113,269.00

7% 45%

FOREIGN TRADE UNDER SPAIN

105

Table 3 (Continued)

Commodity

Value (pesos)

Rank

Percentage of X째 to total exports

England Spain Manuf. tobacco

35% 15% 786,725.00

P. Inglesas England P. Holand Raw tobacco

Percentage of Country X to X째

87% 8% 1% 104,474.00

P. Inglesas England

52% 47% 1894

Total exports

33,176,984.00

Raw abaca U.S.A. England China

14,516,717.00

Sugar China England Australia

10,975,185.00

Coconuts

2,349,080.00

43% 45% 36% 6%

33% 36% 24% 17%

France England Spain Manuf. tobacco

45% 20% 20% 1,750,006.00

China Singapore Spain Raw tobacco Spain Egypt Singapore

40% 24% 9% 1,408,662.00

4% 80% 11% 3%

106

AMADO A. CASTRO

Table 4 -- Philippine Imports

Commodity

Value (pesos)

Rank

Percentage of M째 to total imports

Percentage of Country M to M째

1856 Total Imports Gold & silver bars China Singapore California Muslin cloth

6,949,254.10 2,681,I51.20

38% 77% 5% 3%

779,609.93

11%

England China Singapore Shawls

73% 18% 5% 265,041.17

England China Singapore Cotton cloth

74% 17% 5% 171,358.50

England China Copper sheets

65% 34% 62,367.34

England China Singapore

72% 22% 5% 1876

Total Imports

11,319,925.00

Cotton fabrics P. Inglesas England China

5,196,615.00

Cottonyarns P.Inglesas England

46% 65% 33% 2%

711,820.00

6% 70% 28%

FOREIGN TRADE UNDER SPAIN

107

Table 4 (Continued)

Commodity

Abaca fabrics

Value (pesos)

398,369,00

Rank

Percentage of M째 to total imports

P.lnglesas England China Chemical & pharmaceutical products

56% 24% 16%

383,694.00

P. Inglesas China England Wheat flour p. Inglesas China U.S.A.

Percentage of Country M to M째

44% 37% 10% 332,317.00

2% 56% 27% 21%

1894 Total Imports

28,443,624.00

Closely-woven fabrics

7,208,140.00

25%

England Spain China

42% 39% I1%

Cotton yarn Spain England China

2,506,249.00

Wine Spain France

1,918,093.00

Mineral Oils

1,409,710.00

Russia U.S.A. China

8% 46% 34% 8%

6% 97% 1%

4% 53% 30% 14%

108

AMADOA. CASTRO

Table4 (Continued)

Value (pesos)

Commodity Rice Salgon China Singapore

1,130,018.00

Rank 5

Percentageof M째 to total imports

Percentageof CountryMto Me

3% 75% 14% 9%

were shawls, fans, soap, perfume, pianos and so on. Twenty years later, in 1876, cotton fabrics and yarn, almost all coming from England, either directly or indirectly through English possessions, were the biggest items, and accounted for over half of all imports. New items appeared in the list of the five largest imports: chemical and pharmaceutical products (3 percent) and wheat flour (2 percent) Finally, in 1894, cloth and yarn were againthe largest imports (33 percent). Spain, :though, was coming up as an important source of textiles because of the tariff of 1891 which was intended to favor Spanish products. In 1894, the next three largest imports differed from those in the same positions in 1876, the new imports being wine from Spain, petroleum products from Siberia (Sakhalin) and the United States, and rice from Indochina. An idea of the growth of Philippine trade can be gleaned by a calculation of the rates of growth, using figures from 1850, the beginning of the period, to the last normal year, 1895. The regression equation for this calculation is fnY = fnA + bt where Y represents exports (or imports, as the case may be); A and b, constants, and t, year. The results were! for value of exports (in dollar terms)

fnY

= -38.13750

+ 0.02917t

for export quantities

= -84.68861

+ 0.0475 It

for value of imports (in dollar terms)

= --36.46605

+ 0.02817t

FOREIGNTRADE UNDERSPAIN

109

In other words, for the years from 1850-1895, the growth rate of the value of exports in dollars was 2.9 percent and of the quantity of exports, 4.7 percent. Over the same period, the value of imports rose somewhat less, by 2.8 percent annually. 4 Economic Welfare The main concern of this paper is to determine what trade expansion meant in terms of economic welfare. One eminent historian has said :5 The opening of Manila to foreign trade in 1834, followed by the opening of other ports in succeeding years, led to commercialprogress. This was reflectedin the value of exports which,in 1839, amountedto only _2,500,000 but which,in 1870, roseto PI08,000,000. Whilethis prosperity was obviousin the mannerof living of the new rich amongthe nativesand the mestizos, it was neverthelessnot palpablein the common people who had been groaningunder the heavy burdenof taxation. The prosperitythat the grossnationalproductindicated touched only those who participated in the businessandcommercialventuresbut was an illusionto the great_asses of the people. TO what extent is the above harsh judgement warranted? There is a minor detail which, however, • does not affect the argument. The official figures for export values given in the compilation for the Census of 1903 are P2,674,220 for 1839 and P28,000,000 for 1870 (not P108 million). 6 The real issue here is the economic welfare effects of the expansion of trade, the five-fold increase in the value of exports and the four-fold rise in imports. •Table 5 shows their compound growth rates. There are no measures of national income for the Spanish period, • but perhaps exports and imports, for which there are data, can be 4. The simpleaverageannualratesof growth were: for value of exports (in dollars) : 6.39 percent for export quantifies : 7.66 percent for value of imports (in dollars) : 8.21 percent 5. Teodoro A. Agoncillo (1974), Prelude to 1896, ProfessorialChair Lecture Series, Monograph No. 2, Quezon City: Universityof the Philippines Press,p. 1. 6. See Legarda'spoint on the apparently overstated figurefor 1870.

110

AMADO A. CASTRO

used to indicate at least the trends if not year by year levels of changes in GNP. This is a reasonable assumption. In fact, for the periods for which there are such data (i.e., the last part of the American regime and the entire postwar independence period), there has been a close correspondence over time between growth rates in exports and GNP, although in the Spanish and American periods exports probably grew at somewhat higher rates than GNP. In Table 5, the rates of growth of exports and imports and population have been compared. AS was pointed out earlier, for the forty-five years from 1850-1895, exports grew, in dollar terms, by 2.9 percent a year and in quantity by 4.7 percent while imports rose by 2.8 Table 5 -Compound Rates of Growtha (Percent) 1850-1S 95

1902-I 941

Exports (values)

2.9

4.7

8.2

Exports (quantities)

4.7 b

4.4 c

5.6d

Imports (values)

2.8

3.9

8.1

5.9

1.3

2.0

2.97 d

GNP Population

1949-1979

aTheregressionequationswereof the form: fnY = fnA + bt whereY reFresentsexport (or imports,as the casemaybe); Aand b, constants;and t, year; and the resultswere: 1850-1895 for valueof exports(in dollars) : -38.13750 for export quantities : -84.68861 + 0.04751t for value of imports (in d011ars) ; -36.46605 + 0.02817t 1902-1941 for value of exports(in dollars) : -,71.69708 + 0.04716t for export quantities : -79.79979 + 0.04380t forvalue of import (in dollars) : -57.30193 + 0.03960t 1949-1979 for value of exports (in dollars) : -154.78766 + .08223t for export quantities : - 105.66736 + .05592t for value of imports (in dollars) : -151.96260 + .08088t bFot the years1854 to 1894. eFor the yeats 1902 to 1938. Quantityindexescomputedby GlicerioAbed. dFor the yeats 1950 to 1979.

FOREIGNTRADEUNDERSPAIN

111

percent annually. Philippine population for the year 1858 was 4,296,300 as ftrmly estimated by Englishman John Bowring and in 1903, according to the Census of 1903, it was 7,635,426, giving a growth rate of population of 1.3 percent a year for the forty-five year period from 1858-1903. Comparing export or import growth (as proxies for national income) and population growth, a crude growth rate of per capita income in the order of 1.5 percent annually is indicated. This was not far from the growth rates of per capita incomes in England, Europe, the United States and Japan during the period of expansion of the world economy in the nineteenth century. For reference purposes, the growth rate of exports, imports and population from 1902-1941 and from 1949-1.979 are also given in Table 5. For the latter period, the growth rate for GNP is also shown. For 1902-1941, exports (stated in terms of dollars) rose at the rate of-4.7 percent annually, imports by 3.9 percent, and population by 2 percent. For the postwar period (excluding the years of rapid rehabilitation from 1945-1948), the growth rate for GNP at constant prices was 5.9 percent, for export values 8.2 percent, for export quantifies 5.6 percent, for imports 8.1 percent, and for population 3.0 percent. In all years, trade, and by inference, national income, rose faster than population. Granted that an average figure does not per se show the distribution of the income, still it is hardly plausible that the broad masses of the people had no welfare gains at all The pattern of imports gives another indication of the welfare effects of the trade expansion. Table 4 may again be examined. It can be noted that by 1876, a shift in import composition had taken place. The leading imports were finished cotton textiles and yarn. Cotton cloth is not an item of luxury but of mass consumption. In 1876, 3,320,637 kilograms of cotton and other fabrics and 774,637 kilograms of yam were imported, quantities which not all the weU-dressed aristocratic ladies of Intramuros, the Spanish administrative city, could have consumed in a year. R is clear then that cotton was being sold and worn widely among the populace. For a population of perhaps 5.5 million, this works out to 0.6 kilograms of cloth per person per year or from 3 to 7 meters of cloth, depending on how thick and heavy one wants to assume the cloth to have been. In 1894, consumption was higher still; for a population of 6.7-6.8 million, imports of fabrics of all kinds were 6,039,408 kilograms or about 0.9 kilograms per person. As a reference f'_gure,

112

AMADOA. CASTRO

it will be noted that the cloth ration in China, a country which has gone a long way towards eradicating poverty, was only four• meters of cloth a year in the 1970's. It may be argued that the large imports of cheap English cloth killed the native textile weaving industries, found in such places as • Miagao in Iloflo or around Vigan in the Ilocos. It is claimed that 20,000 looms in Ilocano homes were thus made idle. However, there could be no comparison between the coming productions from handlooms operated in houses during the off-season •or idle periods and the enormously prodnctive power-driven factory looms of Lancashire. 7 In welfare terms, the gains from the cheap imports far outweighed the losses from the handlooms left idle. Similarly it should not be forgotten• that against the production losses of the handweavers in Miagao were the gains of the sugar and rice farmers in other parts of Iloilo. On to the export side, what can be inferred from examining the composition of the Philippine products sold abroad? As has been pointed out, by 1876, the Philippines had made the transition from a subsistence economy to an agricultural export economy, as cash crops were now being cultivated for export rather than the subsistence crop, rice. The shift must have been profitable, for cash crops by their nature are more profitable than a subsistence crop. Otherwise, the shift would not have been made. The early days of the sugar industry saw the cultivation of sugar in Luzon and Iloilo, and when cultivation could no longer be expanded in Iloilo, also saw the colonization and rapid expansion of the population of Negros by migrants and farmer entrepreneurs from Parlay and managers and laborers recruited from Spain. The sugar industry was also witness to the activities and accomplishments of the Englishman Nicholas Loney who sold English sugar machinery, and the rise and fall of the American merchant houses, followed by the takeover of the business by British banks and merchant houses. It may be objected that sugar is precisely an industry which 7. Some idea of the productivity differencesis givenby Henderson:"In the 1840s, a cotton mill employing 750 workers and using a 100 h.p. steam engine could run 50,000 spindles and produce as much yam as 200,000 operativesusing spinningwheels;a calico-printingmachineoperatedby one man could print as much cloth in four colors in an hour as 200 men could printby hand." W.O. Henderson (1969), The Industrialization of Europe, !780.1914, London: Thamesand Hudson,p. 46.

FOREIGN TRADE UNDERSPAIN

113

favors an upper class of large farmers. This can be granted but it should be noted that whatever upper class was created originally was not all upper class but had large middle class or lower origins. These were persons who were able to seize an economic opportunity and prosper through their enterprise. They became upper class as a result of hard work. For the other products however it is clear that then as now, it was not only large landowners but also smallholders who benefited because the production pattern in agriculture in the country, unlike in other colonies such as Indonesia and Malaya, is geared toward smallholder agriculture. This was the case for abaca which was the leading export in 1894, for coffee and coconut, and even for tobacco except for that grown in the Tabacalera haciendas in Cagayan valley. It is hard to believe that the producers of these exports did not benefit and that the gains from the export trade were not diffused widely. Concluding Remarks There is no suggestion in this paper that the Philippines became a prosperous country as a result of foreign trade expansion in the last half of the nineteenth century. That is not true even today. When-the Americans took over, the Philippines was in many respects still a backward economy. Much development work had to be undertaken from the grassroots. Finally, it may be pointed out that in the last half century of Spanish rule, exports were a leading sector in Philippine economic development. This was true in the American period and may also be true in the coming years. However this stages interpretation of Philippine economic development could perhaps be the subject of another study, rather than this one.

REFERENCES Agoncillo, T.A. (1974), Prelude to 1896, Professorial Chair Lecture Series, Monograph No. 2, Quezon City: Universityof the PhilippinesPress. Azcarraga, M.P. (1969), La Libertad de Comercio en Filipinas, Manila: The Filipiniana Book Guild. De Cornyn, T. (1969), Estado de laslslas Filipinas, Manila:The Filipiniana Book Guild. De la Perouse (1969), Travel Accounts of the lslands, Manila: The Filipiniana Book Guild.

114

AMADO A. CASTRO

Henderson, W.O. (1969), The Industrialization Thames and Hudson.

of Europe,

Legarda, B.F. (1955), Foreign Trade, Economic Unpublished Ph.D. dissertation. ....

(1967), "The Philippine November-December.

United

States, Bureau of the Vol. IV, Washington.

Economy

Census

(1905),

1780-1914,

London:

Change and Entrepreneurship,

Under

Spanish

Rule,"

Solidarity,

Census of the Philippine

Islands,

WIDENING URBAN-RURAL INCOME DIFFERENTIALS IN KOREA: A REEXAMINATION Hakchung Choo *

1. Introduction The intersectoral income differentials between the urban and rural households have long been a standing issue in income distribution among the developing countries. The widening urbanrural income differential has been reemphasized in a number of recent studies in Korea. 1 This is essential since acceptance of this fact without critically examining both theoretical and empirical evidence may mislead researchers and policymakers in identifying the critical distributive issues. The purpose of this paper is, by reexamining the issue of widening intersectoral income inequality, to caution those economists who have a tendency to dwell on the past and established concepts despite the rapid pace at which the national and world economies are changing. There are a few sufficient observations to support this caution. First, despite the growing list of literature in income distribution studies, economists have not yet reached a *iKoreaDevelopment Institute,Seoul, Korea. 1. Bong Soon Kang and Pal Yong Moon, An Analysis of Determinants of Farm Income KoreanDevelopmentInstitute (KDI) WorkingPaper77-09, 1977 (in Korean). Sung Hwan Ban, "Determinants of Farm Income and Its Distribution," I-IakchungChoo, ed., Income Distribution and Its Determinants in Korea, Vol. I, Korean Development Institute (I(.DI), 1979, pp. 111-179 (in Korean). Dwight H. Perkins, et al, Rural Development, Studies in the Modernization of the Republic of Korea: 1945-75, Harvard University Press. 1980. 115

116

HAKCHUNGCHOO

consensus on what is meant by 'income' in such studies. 2 Furthermore, the comparison of urban and rural income is based on either gross income or disposable income, which is an inadequate representation of the living standard and the level of welfare. Second, though the conclusions of such a study may be far-reaching, little serious effort is being made to critically evaluate the limitations of the empirical bases used, let alone to improve them, especially for the developing countries. Third, there is a question of priority in persistently attaching such importance to the issue of widening urban-rural income differentials when Korea's rural population has declined significantly in the past twenty years, from 56.3 percent in 1961 to 25.8 percent in 1981, and is expected to decrease further in view of its resource endowment and development strategy. The sin of omission here is that the growing urban poverty and the disparities within have been neglected. This paper critically assesses the validity of both theoretical and empirical evidence of the widening intersectoral inequality and reexamines the evidence in the light of recent attempts at income adjustment. This reexamination begins with a review of both theoretical and empirical problems in comparing the incomes of urban and rural households and in the light of the shortcomings of conventional income concept and existing survey data. Both gross and disposable incomes of the two types of households are then adjusted to present a realistic level of living, and are compared in time series and cross section analyses. Because of data constraints, the adjustments in this study represent a secondbest approximation, maintaining as much consistency as possible.

2. The concept of income used in distributive studies has evolved from income before tax and disposable income to income allowing for the incidences of indirect taxes and government expenditures. Furthermore, such income needs to be standardized with respect to the differences in family size and cost-of-living. If one takes a non-conventional view of income, this conventionally adjusted income needs to be further adjusted to represent welfare GNP, measurement of economic welfare or net national welfare. Final adjustment would require allowing for capital gains and underground incomes. For details, see Hakchung Choo, "The Concepts of Income in Distributive Studies." Hakchung Choo, ed., Op. Cir., pp. 42-62.

URBAN-RURALINCOMESIN KOREA

117

2. Empirical Grounds for Growing Income Differentials A. National Income Accounts

Data

The most often cited empirical sources of growing income differentials in Korea are: first, comparison of the average household and per capita incomes in the agricultural and non-agricultural sectors derived from the national income accounts 3; and second, the comparison of city wage earners' income and farm household income derived from the city and farm household income and expenditure surveys, 4 respectively. These comparisons are made constant prices. Table 1 compares agricultural household and per capita incomes to that of non-agricultural household income in current and constant prices from 1965 to 1979 in Korea. Except for the fact that the income differentials in constant prices appear somewhat higher than current prices from 1965 to 1973 and somewhat lower from 1974 to 1979 because of the implicit bias of the deflator of 1975, 5 the income differentials in both current and constant prices show more or less similar trends from 1965 to 1979. The ratio of agricultural household income to non-agricultural household income in current and constant prices slightly deteriorated from 48.9 and 61.4 percent, respectively, in 1965, to 37.2 and 49.5 percent in 1968. Since 1968, except for some deviations due to harvest conditions, there has been an improving trend. A comparison of per capita income in current and constant prices over these years shows similar patterns as those of the household income comparison. The differences between the two are accounted 3. W.I. Abraham, "Observations on Korea'sIncome Distribution and the Adequacy of the Statistical Base," April 1976 (Mimeographed) Korean Development Institute (Y_DI),pp. 4-7. 4. B.S.Kang and P.Y. Moon, Op. Cir., pp. 11-17. S.H. Ban, Op. C/t., p. 114. 5. The change in the base year from 1970 to 1975 results in the following differences in constant price farm household income as percentage of city wage earners' household income.

1970 Base Year 1975 Base Year (A)- (B)

1965

1970

1975

106.9 111.2 --4.3

67.2 74.6 -7.4

86.7 101.6 -14.9

Table 1 -- Agricultural

Household Income as Percentage Household Income, 1965-1979

of Non-agricultural Non-Agricultural Income= 100.0

Household

Income

Household

Per Capita Income

Percentage of Off-farm Income

Current Price Income

1975 Constant Price Income

Current Price Income

1975 Constant Price Income

1965 1966 1967

48.9 45.4 38.0

61.4 63.7 52.2

59.5 53.1 45.3

73.9 75.4 63.1

20.9 22.1 22.1

1968 1969

37.2 39.2

49.5 50.8

44.1 46.2

57.6 59.8

23.5 23.3

1970 197I 1972 1973 I974 1975 1976 1977 1978 1979

45.5 45.5 45.8 44.3 52.0 55.0 56.3 59.6 61.9 63.4

54.0 49.6 49.2 46.2 53.3 55.0 56.8 55.9 51.3 57.4

49.6 48.1 51.2 49.2 54.5 61.2 60.9 60.9 63.7 64.3

59.5 52.4 54.5 50.4 55.8 61.2 61.7 57.8 52.5 58.3

24.2 t 8.1 17.7 18.8 19.6 18.1 20.3 27.7 28.1 31.3

Source: Derived from the Bank of Korea, Nationallncome

in Korea, for respective years.

,_ O O

URBAN-RURALINCOMESIN KOREA

119

for by the changes of family sizes of the two household categories as shown later in Table 7. This comparison between agricultural and non-agricultural household income derived from the national income accounts usually understates the agricultural household income, because calculation of this income excludes off-farm income, as is shown on the right hand column of Table 1. Furthermore, average non-agricultural household income is strongly influenced by a small number of extremely high household incomes. Thus, it is safe to say that a comparison of household incomes based on the national income accounts gives a rather distorted picture by exaggerating the income differentials. B. Household Survey Data Another empirical source of income comparison of the two types of households is the result of the household income and expenditure surveys, shown in Table 2. This table implicitly assumes the homogeneity of incomes of the households compared. The city wage earners' household income is primarily comprised of employment remunerations while farm income is composed of the imputed wage component of self-employed labor and of the returns on such capital investment as land and intermediate inputS. Thus, there remains a question as to whether or not it is appropriate to compare directly the incomes of these two different types of households. It would be more appropriate to compare the incomes of city proprietors and farm households, considering the nature of their income composition. This type of income comparison has merit in that such a comparison encompasses the two most numerous types of households in the economy. The comparison of current incomes shown in Table 2 indicates a trend toward widening income differentials against the farm households in the late 1960s, rapid improvement in the early 1970s, and another period of deterioration in the late 1970s. The comparison of household incomes in constant prices shows a similar trend as that of current income, except for one notable difference. Because of the index problem, 6 farm household incomes in constant prices, before and after the base year 1975, are evaluated, respectively, somewhat higher and lower than current prices. Per capita income comparison

6. See Footnote 5.

120

HAKCHUNG

CHOO

Table 2 - Farm Income as Percentage of City Wage Earners' Income, 1965-1979 City Wage E_rners' Income=lO0.0 Household Income Current Price Income 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979

99.1 80.2 59.8 62.6 65.3 67.2 78.8 83.0 87.5 104.5 101.6 100.3 102.0 98.3 84.7

Per Capita Income

1975Constant PriceIncome

CurrentPrice Income

111.2 89.6 67.0 67.1 72.0 74.6 89.8 95.3 98.6 105.7 101.6 97.9 96.3 89.7 74.7

1975Constant PriceIncome

87.3 70.6 53.3 56.5 59.1 60.6 71.4 76.6 80.2 96.2 93.0 91.5 87.9 85.5 75.6

98.0 78.8 59.8 60.7 65.2 67.3 81.3 88.0 90.5 97.3 93.0 89.2 83.0 78.0 66,7

Sources: Derivedfrom EconomicPlanningBoard,Annual Report on the Family Income and Expenditure Survey, and Ministryof Agricultureand Fisheries Report on the Results of FarmHousehold Economy Survey, for respective years.

further accentuates difference in family In interpreting One implicit equal weight

the income sizes. these

data,

differentials,

two subtle

points

mainly

due

to the

must be considered.

assumption in per capita is given to all the members

income comparison is that of a household. This means

that all data must be adjusted to the application of which reduces the income

adult equivalent scale, the differential in favor of farm

households. The second point to note is that the income includes a component that accounts for the cost-of-living differentials. increase income differential

wage earners' city and rural

Neglecting two factors has a tendency against farm households.

URBAN-RURALINCOMESIN KOREA C Comparison With Adjusted

121

Farm Income

Upon examination of the Korean farm household income data, W. I. Abraham asserts that the data include, as a significant component, undue capital gains from grain inventories. 7 Although this assertion is erroneous as shown below, a number of agricultural economists made farm household income adjustments, 8 the results of which are compared with city wage earners' household income and shown in Tables 3 and 4. In addition to adjusting for capital gains from inventories of grains and intermediate inputs, these adjustments are also corrected for the over-representation of farmers with relatively large land holding in the farm household survey. The adjustments by three representative studies shown in these, tables vary in degree because of the differences in adjustment procedures. However, the adjustments are consistently downward, in marked contrast to the evidence presented in the earlier tables. As shown in the tables, the adjusted series of Bong Soon Kang and Pal Yong Moon represent the least downward adjustments for the years after 1970, while the adjustments by Sung Hwan Ban were relatively less than those of Kang and Moon for the years before 1969 and greater after that year. The biggest adjustments of the three, after 1971, had been made by D. H. Perkins, et al. The magnitudes of downward adjustments also differ significantly depending on the year under consideration and the authors. Kang and Moon made the smallest downward adjustment, 3 percentage points for 1975, and the highest one, 13.1 percentage points, for 1974, while Ban's adjustments made the lowest 3.1 percentage points for 1966, and the highest, 13.9 percentage points, for 1973. In the case of Perkins, et al, the levels of adjustment were the most profound, from the lowest of 2.7 percentage points for 1965 to the highest of 30.9 percentage points for 1974. The adjusted per capita income comparisons show similar trends to those of adjusted household incomes, except that the differences of the two are explained by changes in average family sizes over time.

7. W.I.Abraham,Op. Cit., p. 4. 8. S_I. Ban, Op. Or., pp. 120-122. B.S. Kangand P.Y. Moon, Op. Cir., pp. 26-27. DJI. Perkins,et. al, Op. Cir.,pp. 429-432.

Table 3 - Adjusted Farm Household Income as Percentage of City Wage Earners' Household Income, 1965-1977

City Wage Farmers' Income= 100.0 Current Price Income

1975 Constant B.S. Kang & P.Y. Moon

Price Income

B.S. Kang & P.Y. Moon

Sung Hwan Ban

D. Perkins et al.

1965 1966 1967 1968 1969 1970

93.8 75.7 56.1 56.9 59.8 61.7

94.7 77.8 56.6 57.3 60.2 59.6

97.4 72.8 55.8 56.3 56.0 60.4

102.8 82.7 62.7 63.1 65.2 68.7

103.6 84.8 63.3 63.6 65.6 66.4

106.6 79.7 62.3 62.4 61.0 67.3

1971 1972 1973 1974 1975 1976 1977

74.0 76.9 79.0 91.0 98.5 -

73.9 74.7 75.3 80.8 88.2 89.8 91.8

70.1 69.4 77.6 72.3 84.1 --

82.7 84.8 82.3 90.0 98.6 -

82.7 82.3 78.6 79.7 88.2 82.9 79.7

78.6 76.7 80.9 71.4 84.2 ---

Sung Hwan Ban

D. Perkins et al.

Souzces: Bong Soon Kang and Pal Yong MoOnAn Ar_lysis of Deterrninants of Farm Income, Korean Development Institute (KDI), 1977, pp. 26-27. Sung Hwaa Ban, "Determinants of Farm Income and Its Distribution )'. Income Distribution and Its Deterrainants in Korea, Hakchung Choo, ed., VoJ. I, Korean Development Institute (KDI), 1979, pp. 120-22. Dwight H. Perkins, et. al, Rural Development, S_dies in the Modernization of the Republic of Korea: 1945-75, HaivaId, 1980, p. 432.

C_ O O

Table 4 - Adjusted Farm Per Capita Income Earners' Per Capita Income,

as Percentage 1965-1977

of City Wage

City Wage Earners' Income= 100.0 Current Price Income

B.S. Kang & P.Y. Moon

Sung Hwan Ban

1975 Constant Price Income

D. Perlins et al.

B.S. Kang & P.Y. Moon

Sung Hwan Ban

D. Perkins et al.

1965 1966 1967 1968 1969

82.4 66.2 50.0 51.4 54.1

83.4 68.6 50.5 51.8 54.5

84.9 64.2 49.8 50.8 0.6

90.2 72.5 55.9 57.0 58.9

91.2 74.6 56.5 57.5 59.4

93.0 70.0 55.6 56.4 55.2

1970 1971 1972 1973 1974 1975 1976 1977

55.7 67.0 71.1 72.3 83.7 90.2 --

53.8 67.0 68.9 69.1 74.4 80.7 81.8 79.1

54.5 63.5 64.2 71.2 66.5 77.0 ---

61.9 75.0 78.3 75.6 82.6 90.2 -

59.9 75.0 76.0 72.2 73.4 80.7 75.6 68.7

60.7 7 i.i 70.8 74.3 65.7 77.0 -

Source: Derived from the same sources, as in Table 3.

_, O ;_

124

HAKCHUNGCHOO

3. Problems Households

Comparing

the Income

of City and Farm

If the problems of the widening urban and rural income differentials constitute a distributive issue, then the concept of income needs to be theoretically and empirically expanded and modified to represent the level of living and welfare. In this section, we will examine three major conceptual modifications necessary in comparing the incomes of the two sector households. Then, we will turn to a number of data problems pertinent to such a comparison in Korea. A. Some Conceptual Problems 1) Burdens of Direct Taxes and Levies Direct and indirect taxes as well as public expenditures are generally recognized as important means of facilitating income redistribution. However, the incidence studies of indirect taxes necessarily make many assumptions that limit the validity of conclusions. 9 Recent studies on the redistributive effects of government expenditures need much more extensive data than is generally available in the developing countries. 1째 Therefore, we will confine our discussion here to the burdens of direct taxes and levies on the two categories of households under different direct tax and levy system. In Korea, farm incomes are still exempted from income tax. However, farm income is subject to a local direct farm land tax. Both types of households pay various other duties and levies. As shown in Table 5, city wage earners' households pay consistently higher portions of their income in direct taxes and levies than do farm households on the average, though the ratios of these burdens vary over time. It is, therefore, clear from the table that the farm households would have more disposable income than their city counterparts for the same level of income. The relatively wide variation in the direct tax and levy burden 9. Luc Henry de Wuff, "Fiscal Incidence Studies in DevelopingCountries: A Survey and Critique," International Monetary Fund (1MF)Staff Papers,Vol. XXII, Mrach, 1975, p. 102. 10. Jocob, Meerman, Public Expenditure in Malaysia: WhoBenefits and Why, Oxford University Press. 1979.

URBAN-RURAL INCOMES IN KOREA

125

Table $ - Ratio of Direct Taxes and Levies to Farm and City Wage Earners' Household Income, 1965-1979. (In Percent) Year 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979

Farm Household

City Wage Earners' Household

2.73 2.59 1.75 1.69 1.98 1.28 1.16 1.00 1.33 1.29 1.45 1.69 1.87 1.68 2.10

3.31 3.38 4.70 7.91 7.24 6.78 7.05 6.60 6.99 1.87 2.85 3.60 3.04 2.38 2.12

Sources: Economic PlanningBoard. Annual Report on the Family Income and Expenditure Survey, fo_ respectiveyear_. Ministryof Agricultuxeand Fisheries,Report on the Resuits of Farm Household Economy Survey, for respective years.

the two types of households stems from the fact that tax exemptions and tax rates on wages change annually, while the land exemption and rates change once every several years. However, in conditions of high inflation, the effective fixed land tax exemptions rates over a few years rise significantly. The tax burden ratios of earners are also affected by the imposed upper income ceiling also changed only intermittently until 1976. The 1974 tax and burden ratio of the city wage earners' households exemplifies distortions caused by the imposition of a 2.4 million won per Upper income ceiling adjusted from 2 million won in 1973, an of 20 percent, when nominal per capita GNP increased by percent, from 153.6 thousand won in 1973 to 211.4 thousand 1974. Despite these variations, the ratios of direct tax and burden of farm households are generally lower than those on wage

earners'

households

which

should

appropriately

126

HAKCHUNGCHO0

accounted for in the income households. 1t

comparison

of the two

types

2) Cost of Living Differentials In comparing the household income of farm and city dwellers, the differences in lifestyle and subsequent cost of living differentials should be properly recognized. These differences may be explained by three major factors: necessary expenditures of city dwellers or so-called "regretables", the difference in urban and rural purchasing power parity, and the difference in the quality of life between urban and rural living. Including necessary expenditures for urban living in the conventional concept of income has been opposed by many who criticize the inadequacies of the national income concepO2. For example, the expenditure on bus fares for commuting to work should be reimbursed as cities become more urbanized, although the time spent for commuting could not be compensated. Therefore, even if the income levels of the urban working population and farmers are the same, the farmers would enjoy a higher level of real income than the urban population to the extent equal to the portion of income expended on these "regretables". However, there remains the question as to whether or not one can equate the value of commuting to work by mass transportation by city workers with that of commuting to fields on foot by farmers, which is,related with the qualitative aspect of living in this comparison. The logic of allowing for the cost-of-living differentials among the countries in an international comparison of income 13 could also be applied to an urban-rural income comparison. Although there remain technical problems of gathering appropriate data and deriving convincing formulas for adjustment, even the prices of necessities differ substantially between the urban and rural areas. For these 11. For details, see: Hakchung Choo, "Redistributive Effects of Agriettltural Taxcum-Subsidios," Korea Development Review, Vol. I, No. 3, September, 1979 (in Korean) pp. 43-68. 12. A.W. Sametz, "Production of Goods and Services: The Measurement of Economic Growth,"E.B. Sheldon and W.E. Moore, eds., Indicators of Social Change, Russel Sage Foundation, 1968. Economic Council of Japan, Measuring Net National Welfareof Japan, 1973. 13. Irving B. Kravis et. al, A System of International Comparisons of GrossProduct and PurchasingPower, Johns Hopkins Press, 1975, Oh. II.

URBAN-RURAL INCOMES IN KOREA

difference foodstuff

in the cost of living, expenditures and others, should be considered.

127

such

as dwelling,

Table 6 provides a quantitative base allowing for the cost of living differentials between cities and rural areas for a few selected countries in 1973. According to this table, the cost of living is about 10 percent higher in the medium-size cities than in rural areas, and about 20 percent higher in large cities. In Korea, the cost-of-living in 1973 is estimated to have been 19.4 percent higher for urban dwellers. However, a portion of this is attributable to qualitative differences between urban and rural lifestyles. Since there is no established method of accounting for such qualitative differences in lifestyle, no attempt is made here to modify further the cost-ofliving differentials, except to use most relevant existing data. Table 6 - International Comparison of Urban and Rural Cost of Living Differentials, 1973. (Rural Cost of Living ffi100.0)

Rural Area

Small and Medium Cities

Large City

Canada

100.0

115.0

128.8

Japan

100.0

110.9

Korea

100.0

119.4

U.S.

100.0

117.1

121.9

Source: SandMokSuh,et. al,Patternsof Poverty and Anti-Poverty Programsin Korea. KoreanDevelopmentInstitute (KDI),1981, p. 98.

3) Adjusting

for Family Size

The difference in average family size between city wage earners' and farm households (Table 7), could be a cause of a downward bias in the per capita income comparison of the two types of households. This bias, however, could theoretically be adjusted by using an adult equivalent scale.14 An attempt was made in Korea to derive an adult equivalent scale from the city household income and expenditure survey, but it did 14. Belvit Singh and A. L. Nagar, "Determination Scales," Economctrica, Vol XXXXI, No. 2, March 1973.

of Consumer Unit

128

HAKCHUNGCHO0

Table 7 - Average Family Size of City Wage Earners' Household and Farm Househo_s in Korea, 1965-79 (Unit: person) Year 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979

Farm Household 6.29 6.22 6.12 6.02 5.99 5.92 5.83 5.71 5.72 5.66 5.63 5.54 5.52 5.38 5.20

CRy Wage Eaners' Households 5.54 5.47 5.46 5.44 5.42 5.34 5.28 5.27 5.25 5.21 5.15 5.05 4.76 4.68 4.64

Sours: E_nomic Planning Bo_d, Annual Report on _e Family lncomeand Expendit_eSu_ey.

not yield a meaningful result. 15 However, in the case of the United States, for example, as shown in Table 8, the difference in adult equivalent scale between a two-member family and a four-member family is about 50 percent. Although this adult equivalent scale is derived from the poverty line cost-of-living data, and thus, may not accurately represent the true picture of a developing country, it is sufficient to support an argument that the per capita income concept in comparing income differentials of the two sectors gives an undue bias against large families. Besides, the income comparison of the two sectors is made in terms of average of convenience. Although arithmetic mean is a widely used representative measure for mass observations, the question of how significant a mean is should not be overlooked, 15. Kwang Suk Kim and Dai Young Kim, The Effects of Households Size, Structure and Income on Expenditure Patterns, Korean Development Institute (KDI) Working Paper 7510, May, 1975.

URBAN-RURALINCOMESIN KOREA

129

Table8 -- StandardizedCostof LivingScalesof the U.S. PovertyLine, 1974 (Non-Farm2-PersonHouseholdffil00.0) Sizeof Household

Non-Farm Household

Farm Household

1Person 2 Persons 3 Persons 4 Persons 5 Persons 6 Persons Over7 Persons

77.7 100.0 122.6 156.9 185.3 208.6 257,0

65.2 84.3 103.7 134.0 157.5 177.5 218.6

Average 77.5 99.4 121.8 156.0 184.1 207.1 254.3

Source:DerivedfromU.S. Bureauof Census,"Characteristics of thePopulation Belowthe PovertyLevel:1974"CurrentPopulationReports,SeriesP-60,No. 102,Table A-2.

especially in distributive studies. For Korean farm households, the average income is more significant than for city households because of a low measure of inequality for farm households. 16 B. Problems of Survey Data Since the household income and other related data are derived from different survey sources, it is important to establish comparability of the data used for an income comparison of the two types of households. Of the many differences, the following three deserve careful notice: 1)inventory evaluation in the farm household survey, 2) imputation of dwelling ownership in the city household survey, and 3) price indices used for household income deflators in Korea. 1) Farm Inventory Evaluation The most controversial issue in the income comparison of city and farm households in Korea has been the alleged inclusion in farm income of capital gains from inventory evaluation. W. I. Abraham argues that "the farm survey figures include capital gains in inventory 16. Hakchung Choo, "Economic Growth and Income Distribution," Chong Kee Park, ed., Human Resources and Social Development in Korea, Korean Development Institute, 1980, pp. 291-2.

130

HAKCHUNGCHOO

accumulation and therefore do not measure current income (or saving out of current income) - the steeper the rise in g_ain prices, the bigger the capital gain portion which is included. ''z7 This observation by Abraham has been taken literally and many agricultural economists have attempted to adjust farm income accordingly, as shown in Tables 3 and 4. This plausible point is in fact a misundertanding by a consultant to a developing country, the implications of which were never critically examined by the followers of this contention. Because of the importance of this adjustment in the attempt to conduct an income comparison in this paper, some clarifications on the accounting procedures for determining the farm income in Korea is in order. Accoring to Abraham, changes in inventories, Air, result even when no actual change in the quantity of inventory occurs, since the value of inventory at the beginning of a given year, It- l, is value in the prices of the previous year, Pt- l, and that value at the end of that year is in the prices of that year, Pt. This change in nominal value due to price changes under rather high inflationary pressure and agricultural price support policy is considered capital gains by the proponent of this adjustment. Table 9 compares the current official estimation survey procedures and Abraham's adjustment procedures. For simplicity, this table gives an illustration for estimating the total output as a basis for determining income by using hypothetical figures. It seems that Abraham conceived output to be the sum of self-consumption, sales, and change in inventories, without paying due attention to its identified relationship to production. The current survey procedures would yield the production total of 960 while Abraham's adjusted procedures would result in the allocative sum of 920, which is diffe* rent from the production for that year. Therefore, Abraham's undue exclusion of capitalgains lacks a logical basis. Genuine capital gains could occur only when beginning inventories exceed annual dispositions and the inventories continue to accumulate to the following year. Considering the demand for farm products in Korea, however, continued increases in farm inventories are not conceivable. Even if such unlikely increases in farm inventories did occur, it would not present a problem to account for them as a part of current income from gains due to price changes as with a 17. W.I. Abraham,Op. Ot., p. 4.

Table 9 - Sample of Procedure for Determining

Farm Output in

Current Prices. Survey Procedures Quantity

Unit Price

Abraham's Adjusted Procedures Value

Quantity

Unit Price

;_ Z

Value _o

Beginning Inventory (Ib) Current.Output (Q)

20 80

10 12

200 960

20 80

10 12

200 960

Self-Consumption

440

(20)

(10)

(200)

(20)

(10)

(200)

r/3

(20)

(12)

(240)

(20)

(12)

(240)

420

300

250

100

of Inventory

of Current Output (CO) Sales (S) Ending Inventory

(le)

Change in Inventory

(i)

_32

HAKCHUNGCHOO

business. The contention of Abraham is one of those casual observations often made by an outsider, the implications of which were not carefully assessed with respect to its applicability to a developing country by the followers. 2) Imputed Income From House Ownership Despite many attempts to compare farm household incomes and the wage earners' household incomes, no one has yet pointed out the inclusion of imputed income from the house ownership for city dwellers and its exclusion in the calculation of farm income in Korea. 18 This inconsistency in treating imputed income from house ownership and rentals is an upward bias to the city household income. An effort to adjust for this difference in the farm income would require an amount of estimation that is beyond the scope of this study. However, for the sake of making a comparison, one should recognize and allow for the imputed income from the house ownership and rentals included in the city wage earners' household income. As shown in Table 10, such imputed incomes are estimated to represent around 10 percent of household income. This percentage, of course, suggests an upper ceiling for such adjustment. Rural dwelling conditions in general are somewhat inferior to those in cities. If the imputed income from the ownership of farm housing is estimated, it would give a lower ceiling for adjustment. While recognizing these differences, no further attempt is made in this analyses except to compare income, excluding imputed income from house ownership and rentals. 3) Selecting Consumer Price Index as Deflator It is conventional to compare city and farm household incomes over an extended period in real terms. The question of choosing an appropriate price index as a deflator has received little attention from analysts in the past, at least in Korea. In past studies, one of the three consumer price indices - city consumer prices, farm prices paid, and farm consumer prices- has been used as a deflator, without due regard for its analytical implications and statistical problems. 18. Office of Statistics, Ministry of Agriculture and Forestry, Guidelines for Classificationsin Farm Household Survey, 1977, pp. 41-55. (in Korean).

URBAN-RURAL INCOMES IN KOREA

133

Table 10 - Imputed Income from Housing Ownership and Rentals of City Wage Earners as Percentage of Household Income, 1963-79

Year

Household Income (I) (thousand won)

Imputed Income (2) (thousand won)

Imputed Income as Percentage of Household Income (2)/(1) (%)

1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979

80.2 97.2 112.6 161.5 248.6 286.0 333.6 381.2 451.9 517.4 550.2 644.5 859.3 1,151.8 1,405.1 1,916.3 2,629.6

8.3 9.4 11.2 20.5 30.5 30.7 37.8 43.1 51.8 60.5 65.6 71.2 72.8 92.5 134.2 182.2 292.6

10.4 9.7 10.0 12.7 12.3 10.7 11.3 11.3 11.5 11.7 11.9 11.1 8.5 8.0 9.6 9.5 11.1

Source: EconomicPlanrdngBoard, Annual Report on the Family Income and Expenditure Survey, 1980, pp. 34-50.

The use of the city consumer price index income of the farm household implicitly assumes

to derive the real identical consump-

tion patterns of city and rural dwellers. Such an assumption would result in compared real incomes of the two types of households being identical to the results of comparing current price incomes of the two. Therefore, this practice is not only insignificant for comparative purposes, but also distorts the real incomes of the farm households. What is more startling is the use of the consumer price index of Seoul as the deflator for the farm household income in some publications. 19 Farm income in real terms derived in such a manner would inevitably

further

deviate

from reality.

19. For example, see: Economic Planning Board, Handbook Economy, 1979, p. 182.

of Korean

134

HAKCHUNGCHO0

The index of farm prices paid is compiled in a country to determine the agricultural price parity. This index includes, in addition to the prices paid for farm household goods and services, the prices of farm supplies and intermediate inputs, including wages, rentals and fees. The relative weights for these major components do vary according to places and time. As shown in Table 11, the weights of 562.7/1,000; 353.7/1,000; and 83.6[1,000, are assigned to household goods and services; farm supplies and intermediate inputs; and, farm wages, rentals and fees, respectively. This deflator is somewhat inadequate because of the inclusion of items other than household consumption goods and services. It is particularly inappropriate in the Korean context since the prices of nonconsumption farm goods and services increased at relatively rapid rates during the period of 1962 to 1979 as observed in Table 11. The most appropriate deflator is undoubtedly the farm household consumption price index. However, this price series compiled by the Korea Federation of Agricultural Cooperatives is not statistically compatible with the city consumer price index compiled by the National Bureau of Statistics at the Economic Planning Board for at least two reasons. First, the coverage of commodity items in the farm consumer price index is quite limited to the extent that it is neither representative of nor compatible with the city consumer price index. As shown in Table 12, the farm consumer price index in 1979 is derived from the prices of 94 items while the city consumer price index is derived from theprices of 349 items. Of major expenditure categories, only the electricity and heating category seems to be more or less compatible in terms of the number of items surveyed and the relative weights assigned. Thus, the farm consumer price index lacks representativeness as compared with the city consumer price index and is more vulnerable to sampling errors. Second, it is inevitable that due to the differences in the patterns of consumption and living, the same expenditure items will have different weights. However, the relative weights given to such expenditure items as food, clothing, and miscellaneous expenses show such extensive differences that the question of compatibility of the two indices arises. As shown in Table 12, the weight of miscellaneous expenses on 24 items in the farm consumer price index accounts for almost half total weight. Thus, fluctuation would be

URBAN-RURAL INCOMES IN KOREA

135

Table 11 - Rates of Increase of Korean Farm Prices Paid by Major Categories, 1962-79 (In Percent)

Year

Farm Prices Paid (1,000.0)

1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979

10.8 11.3 27.0 33.5 12.0 7.8 11.9 14.5 13.7 12.9 13.3 9.2 31.0 23.9 24.9 17.1 30.0 13.7

Farm Input Supplies (353.7)

Farm Consumer Goods (562.7)

Farm Input Services (83.6)

14.4 0.7 16.7 62.3 10.6 3.5 4.9 21.7 8.9 15.7 15.7 15.1 22.0 17.8 35.3 19.7 43.1 - 4.2

9.9 11.1 29.4 18.4 12.3 10.2 15.5 9.4 15.4 10.6 10.8 5.0 38.7 26.7 18.1 14.0 18.3 22.4

11.1 26.3 34.7 13.2 14.3 17.6 22.2 21.7 24.0 19.9 17.5 10.6 30.6 28.7 26.3 24.3 41.3 50.8

20.7 10.9 21.2 19.5 17.9

16.2 12.2 19.9 18.2 16.5

19.9 21.1 22.4 38.8 24.2

Average Rates of Increases: 1962-66 1967-71 1972-76 197%79 1962-79

18.9 12.2 20.5 20.3 17.1

Source: Miaistry of Agricultureand Fisheries, Yearbook of Agn'culture and Forestry Statistics, 1980.

quite sensitive to the representativeness of selected items and their changes. These problems in the price indices seem to have been responsible for the divergence in the movements of the two price indices in recent years. During the period of 1970 to 1973, the rate of increase in the farm consumer price index was slightly lower than that in the city

consumer

price index

as noted

in Table

13. However,

the farm

Table 12 - Assignment of Relative Weights of Major Expenditure Items of City and Farm Consumer Price Indices City Consumer Price Index Number of Cornmodity Items Surveyed Food

Farm Consumer

Price Index

Number of ComWeight

modity Items Surveyed

Weight

131

458.0

110.1

98.1 (55.2)

56.0

46.0 (25.9)

Clothing

92.5

175.4 (98.0)

Miscellaneous

283.4

455. l (256.1)

349

1,000.0

Housing Electricity

(126.8)

and

Heating

Total

225.4

1,000.00

$ oaree: Economic Planning Board, Annual Report on the Price Survey, 1979 Ministry of Agriculture and Fisheries, Yearbook of Agriculture and Forestry Statistics, 1980. Note: F'_ures in parentheses indicate the original weights given in the index of farm prices paid.

(562.7)

c_ O

URBAN-RURAL INCOMES IN KOREA

137

consumer price index increased by about 4 percentage points faster rate than the city consumer price index during the period of 1974 to 1978. This recent divergence in the two price indices can be attributed largely to relatively miscellaneous items and clothing

rapid increases in the prices of expenditures of farm households.

Doubts may be raised about this divergence in the rates of increases in the price series in view of the lack of compatibility of the two price series. If such a suspicion is warranted, then it would also distort the income comparison of the two types of households in constant prices, significantly against farm households in the later period. Table 13 - Comparison of City and Farm Consumer Price Increases by Major Expenditure Categories, 1970-73 and 1974-78 (In Percent) Annual Average of 1970-73

Annual Average of 1974-78

City

Farm

Difference

City Farm

Difference

All Commodities

5.0

4.2

0.8

15.4 19.2

-3.8

Food and Beverage

5.5

0.0

17.8 21.2

Clothing

5.5

4.0

1.5

13.2 20.0

-6.8

Housing

4.6

11.3 14.3

-3.0

Light and Heating

4.3

4.1

Miscellaneous

4.3

4.5

-0.1 0.2 -0.2

3.4

17.0 13.6

3.4

13.1 20.1

-7.0

Source: Derived from Economic Planning Board, Annual Report on the PriceSurvey, and Ministry of Agriculture and Fisheries, Yearbook of Agriculture and Forestry Statistics, 1980.

4. Effective Level of Living Incomes Compared

Estimated

and

The preceding examination of the theoretical and empirical problems of a conventional income comparison of the city wage earners' and farm households underscores a need to make appropriate adjustments with regard to data constraints. Therefore, we will first present the estimation procedures of adjusted income called the effective level of living income. This adjusted income will be compared both in time series and in cross section analyses.

138

HAKCHUNGCHOO A. Adiustment

Procedures

Because of the lack of theoretical consensus and data limitations, it is practically impossible to fully adjust the income of both city wage earners' and farm households. One alternative is to derive second-best approximations. The effective level of living income is defined as gross income adjusted for cost of living differentials, differences in number of household members, and direct tax and levy burdens between the two types of households as the first approximation. (We will call this series Estimation I.) Another alternative estimation of the effective level of living income is to have a further adjustment by excluding from the city household income the imputed incomes from the housing ownership and rentals so as to make the city household income more consistent with the farm household income. This will be referred to as Estimation II. In adjusting the differences in urban and rural costs of living and the sizes of households, it is inevitable to use the adult equivalent scales of the 1974 U.S. poverty study shown in Table 8, since no such data for Korea are available. There remains, of course, an empirical question as to whether or not data of an advanced country for a particular reference group can be applied to an analysis of the total population of a developing country. This adjustment factor, however, seems to be consistent with the contention of this paper for two reasons. First, the adjustment factor is derived not from the national average, but from the lowest income class of the United States. Second, as indicated by Table 6, the urban-rural cost-of-living differential for the U.S. is slightly smaller than that for Korea. Thus, the use of this factor seems to provide a rather conservative second-best adjustment. A cross-section comparison of the effective level of living income is also attempted here in order to assess the effect on intrasectoral inequalities of the two types of households being compared. The reference year for the cross-section analysis is 1976, a benchmark year for distribution of income analysis in Korea. 20 Because imputed incomes from house ownership and rentals by city wage earners' households by income class were not readily available, this cross20. Hakchung Choo, "An Estimation and Analysis of Size Distribution of income in Korea: Over Time and By Sector," KoreaDevelopment Review, Vol. I, No. 1, pp. 22-43, (in Korean).

URBAN-RURALINCOMESIN KOREA

139

section comparison is limited to the effective level of living income and is compatible to Estimation II of the time-series analysis. B. Time-Series Comparison The results of the adjusted income comparison in terms of effective level of living incomes for the two types of households are summarized in Tables 14 and 15. As these tables indicate, standardization by the adult equivalent scale does not seem to significantly affect the comparison of nominal gross and disposable household incomes in contrast to that by the per capita current income comparison shown in Table 2. By adjusting for the cost-of-living differentials, the ratio of effective level of farm living income to that of city wage earners' households increases by about 3 percentage points in the early 1960s, by about 16 percentage points in the late 1960s, and by about 10 percentage points in the early 1970s. Further adjustment, allowing for differences in direct tax and levy burdens, improves the ratio of income comparison in favor of farm households by about 1 to 3 percentage points. Estimation II gives an even better comparative picture of farm households against city wage earners' households, as shown in Table 15. The difference between Estimations I and II is explained by the ratio of imputed incomes from the house ownership and rentals shown in Table 10. In either case, it is safe to say that there is no evidence of widening income differentials between the two types of households, if not the contrary. The income disparity against farm households may have posed a problem during the period of around 1967 to 1970, but it seems to have been resolved with, among others, the agricultural price support policy, the increase in agricultural productivity _lue to improved seeds, increased use of fertilizer and pesticides, and the shift in agricultural labor to non-agricultural sector. A final observation concerning these time-series comparisons concerns the seemingly wide fluctuations in the ratios of compared incomes over time. These fluctuations are attributable primarily to harvest conditions and the upper income ceiling that were imposed in the city household income and expenditure survey until 1976 and changed intermittently during the period of this analysis. For example, the steep declines in the ratio of urban to rural income in 1967 and 1968 were due to the bad harvests in these years, coupled

X O Table 14 - Farm Household Effective Level of Living Income as Percentage _ation I)

Ad_ti Equivalent Income

Farm • Household (1,000 won)

City Household (1,000 won)

Effective Level of Living Income

Parity RaUo (City = 100.0)

Farm Household (1,000 won)

City Household (1,000 won)

Parity Ratio (City= 100_0)

of City Wage Earners' Household

Disposable

Farm Household (1,000 won)

Adult Equivalent

City Household (l,000 won)

Income

Parity Ratio (City= 100.0)

Income, 1963-79

Disposable Effective Levet of Living Income Farm Household (1,000 won)

City Household (l,000 won)

Parity Ratio (City= 100.0)

1963 1964 1965 1966 1967

93,2 125.7 112.2 130.2 149.5

80.2 97.2 112.6 161.5 248.6

116.2 129.3 99.6 80.6 60.1

95.5 127.5 117.3 137.1 160.5

80.2 97.2 112.6 161.5 248.6

119.1 131.2 104.2 84.9 64.6

91.2 122.9 109.1 126.8 146.9

77.5 94.7 109.6 157.3 240.1

117.7 129.8 99.5 80.6 61.2

93.5 124.7 114.1 133..5 157.7

77.5 94.7 109.6 157.3 240.1

120.7 131.7 104.1 84.9 65.7

1968 1969 1970 1971 1972

179.0 217.9 255.8 356.4 429.4

286.0 333,6 381.2 451.9 517.4

62.6 65.3 67.1 78.9 83.0

196.1 239.7 280.9 393.1 479.2

286.0 333.6 381.2 451.9 517.4

68.6 71.9 73.7 87.0 92.6

175.9 213£ 252.5 352.2 425.1

269.2 315.4 361.4 427.4 491.3

65.3 67.7 69.9 82.4 86.5

192.8 234.9 277.3 388.5 474A

269.2 315.4 361.4 427.4 491.3

71.6 74.5 76.7 90.9 96.6

1973 1974 1975 1976 1977 1978 1979

480.7 674.5 872.9 1,156.3 1,432.8 1,884.2 2,227.5

550.2 644.6 859.3 1,151.8 1,405.1 1,916.3 •2,629.6

87.4 104.6 101.6 100.4 102.0 98.3 84.7

534.I 751.3 972.4 1,282.3 1,523.1 2,012.3 2,416.8

550.2 644.6 859.3 1,151.8 1,405.1 1,916.3 2,629.9

97.1 116.6 113.2 I 11.3 108.4 105.0 91.9

474.3 665.7 860.2 1,136.7 1,406.1 1,852.6 2,180.6

520.3 634.8 838.8 1,116.4 1,369.0 1,870.7 2_573.8

91.2 104.9 102.6 101.8 102.7 99.0 84.7

527.0 741.6 958.3 1,260.6 1,494.7 1,980.7 2,369.9

520.3 634.8 838.8 1,116.4 1,369.0 1,870.7 2,573.8

101.3 116.8 114.3 112.9 109.2 105.9 92.1

_I_ f'3

O O

Table 15 - Farm Household _

Adult Equivalent Income

1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 I978 1979

Farm Household (l,000won)

City Household (1,000won)

93.2 125.7 112.2 130.2 149.5 179.0 217.9 255.8 356.4 429.4 480.7 674.5 872.9 1,156.3 1,432.8 1,884.2 2,227.5

71.9 87.8 101.4 141.0 218.2 255.2 295.8 338.2 400.1 457.0 484.6 573.5 786.5 1,059,2 1,270.9 1,734.1 2,337.0

Effective Levelof LivingIncome

lZarity Farm Ratio (City= Household 100.0) (l,000won) 129.6 143.2 110.7 92.3 68.5 70.1 73.7 75.6 89.1 94.0 99.2 117.6 111.0 109.2 112.7 108.9 95.3

Level of IAvi_gIncome es Peaeentageof City WageEarnexs' Household Income, 1963-79 (Estimation ID

95.5 127.5 117.3 137.1 160.5 196.1 239.7 280.9 393.1 479.2 534.1 751.3 972,4 1,282.3 1,523.1 2,012.3 2,416.8

City Household (1,000won) 71.9 87.8 101.4 141.0 218.2 255.2 295.8 338.2 400.1 457.0 484.6 573.5 786.5 1,059.2 1,270.9 1,734.1 2,337.0

Disposable Adult Equivalent Income

Parity Farm Ratio (City= Housekold 100.0) (I,000won) 132.8 145.2 115.7 97.2 73.6 76,8 81.0 83.1 98.3 104.9 110.2 131.0 123.6 121.1 119.8 116.0 103.4

91.2 122.9 109.1 126.8 146.9 175.9 213.6 252.5 352.2 425.1 474.3 665.7 860.2 1,136.7 1,406.1 1,852.6 2,180.6

City Household (l,000won) 69.3 85.3 98.4 136.8 209.7 238.4 277.6 318.4 375,6 430.8 454.7 563.8 766.0 1,023.8 1,244.1 1,699.7 2,286.3

Disposable Effective Level of LivingIncome

Parity Farm Ratio (CRy= Household 100.0) (1,000won) 131.6 144.1 110.9 92.7 70.1 73.8 77.0 79.3 93.7 98.7 104.3 118.1 112.3 111.0 113.0 109.0 95.0

93.5 124.7 114,1 133,5 157.7 192,8 234,9 277.3 388,5 474,4 527.0 741,6 958.3 1,260.6 1,494.7 1,980.7 2,369.9

City Household (I,000won)

Parity Ratio (City= 100.0)

69.3 85.3 98.4 136.8 209.7 238.4 277.6 318.4 375.6 430.8 454.7 563.8 766,0 1,023.8 1,244.1 1,699.7 2,296.3

134.9 146.2 116.0 97.6 75.2 80.9 84.5 87.1 103.4 110.1 115.9 131.5 125.1 123.1 120.1 116.5 103.2

Z _O [-_ r,_ _. O

142

HAKCHUNGCHOO

with the adjustment of the upper income ceiling in the city household income and expenditure survey from 500 thousand won to 1.5 million won in 1967. Because of the income ceiling adjustment, the 1967 and 1968 city household incomes are evaluated higher than the previous years when farm incomes were lower than usual. More gradual upper income ceiling adjustments were made in 1969, 1973, and 1975, before the ceiling was eliminated in 1977. If the distorting effects of these factors arc taken into account, the disparity ratios of the effective level of living income of the two types of households should not deviate as much as shown in Tables 14 and 15. C Cross-Section Analysis In order to assess the effects of adjusted effective level of living income on the urban-rural income differential by income classes, a cross-section analysis for 1976 is attempted and the resulting comparison is presented in Table 16. The year 1976 was chosen for this analysis because it is one of the years least affected by distortions caused by the upper income ceiling and the most recent of the three benchmark years mentioned earlier. This cross-section comparison is comparable to Estimation II of the time series analysis. According to the result, adjustment from the nominal adult equivalent income to the disposable adult equivalent income seems to somewhat adversely affect the size distribution of farm income. Farm households with an annual income of less than 800 thousand won pay more or less the same amount of direct taxes and levies as city households. However, farm households with an annual income of more than 800 thousand won pay relatively less than city households as their income increases. Hence, the level of income parity of farm households compared against city households increases as the level of income increases. The comparison of effective level of living income by income class shows more encouraging results than that of income adjusted by difference in family size and direct tax burdens. In terms of effective level of living income, the lower the farm household income is the higher the income parity compared with the city household becomes. The only exception is the highest farm income class that shows a slight decline by such adjustments. Another noteworthy fact is that the percentage distribution of farm income is more symmetrical than that of city income as shown

Table 16 - Feen HousehoM Effective Level of Living Income as Pexeeatage of City Wage Earners' Income by Income Classes, 1976 Adult Equivalent Income Annual Income of:

Farm Household (1,000 won)

Effective Level of Living Income

City Household (1,000 won)

Parity Ratio (City=100.0)

Farm Household (1,000 won)

City Household (1,000 won)

Parity Ratio (City=100.0)

Lem thee 500,000 Won

360.4

409.1

88.1

468.5

409.1

114.5

500,000 to less than 800,000 Won

664.0

717,3

92.6

746.3

717.3

104.0

800,000 to less than 1,100,000 Won

949.8

1,021.3

93.0

1,045.7

1,021.3

102.4

1,100,000 to less than 1,400,000 Won 1,400,000 to teas than 1,700,000 Won

1,248.1 1,545.2

1,329.9 1,653.1

93.9 93.5

1,382.9 1,687.4

1,329.9 1,653.1

104.0 102.1

1,700,000 Won ot mote

2,437.3

2,756.2

88.4

2,359.3

2,756.2

85.6

Disposable Adult Equivalent Income

Disposable Effective Level of Living Income

Â˘0 ;_

Percentage of Householda

Annual Income of: Farm Household (1,000 won)

City Household (1,000 won)

Parity Ratio (City=100.0)

Farm Household (1,000 won)

City Household (1,000 won)

Parity Ratio

Farm

City

(City=100.0)

Household

Hou_old

Â˘/_ Pq

Less than 500,000 Won

358.6

406.6

88.2

466.7

406.6

114.8

12.6

14.2

500,000 to less than 800,000 Won 800,000 lo less than 1,100,000 Won

658.0 939.6

711.2 1,006.4

92.5 93.4

740.3 1,035.5

711.2 1,006.4

104.1 102.9

24.4 22.5

30.1 22.7

1,100,000 to tess than 1,400,000 Won

1,231.8

1,300.8

94.7

1,366.6

1,300.8

105.1

14.5

13.3

1,400,000 to less them 1,700,000 Won

1.517.5

1,593.7

95.2

1,659.7

1,593.7

104.I

8.9

6.8

1,700,000 Won or more

2,374.4

2,586.8

91.8

2,296.4

2,586.8

88.8

17.1

12.9

144

HAKCHUNGCHO0

in the table. For example, the percentage of relatively low income farm households with an annual income of less than 800 thousand won is 37 percent while the same ratio of the city households is 44 percent in 1976. Assuming that the rapid pace of industrialization and urbanization will be sustained, the percentage of city households to all households will increase, and a high portion of households would be concentrated in the lower income brackets. For this reason, the crux of the income distribution problem lies not in inter-sectoral inequality, e.e., widening urban-rural income differentials, but in intra-sectoral inequality, i.e, worsening income distribution among urban households. 5. Summary and Conclusion The long-prevailing notion of widening urban-rural income differentials in Korea seems to be based on evidence that has a number of conceptual and empirical problems. However, without carefully examining the comparability of the surveyed incomes of the city wage earners' and farm households, the farm household income was further adjusted downward, allowing for the sampling bias and the capital gains from farm inventories. Such adjustments have resulted in further evidence of accentuated urban-rural income differentials during the past two decades of rapid growth. If the concept of income is used in distributive studies as a proxy to represent the level of living and welfare, a direct comparison of the income before tax is quite inadequate. It needs to be adjusted for the urban-rural differences in cost-of-living, in family size, and in the direct tax and levy burdens. In adjusting for the difference in family size, an adult equivalent scale, rather than the conventional equal weight, should be employed. The income and price data used for this purpose lack comparativeness in a number of ways. The farm household survey sample is somewhat upward biased because of the exclusion of the agricultural laborer households and farm households with very small landholdings, and because of the over-representation of farm households with large landholdings. On the other hand, the city wage earners' household survey is somewhat downward biased due to the imposition of a maximum income ceiling which was intermittently adjusted until 1976. However, little notice is given to the fact that the city household survey has upward bias due to a higher representation of high-paying

URBAN-RURAL INCOMES INKOREA

145

job categories and lower proportions of unemployed and underemployed. 21 Adjusting for the sample bias of the farm household survey without doing the same for the city household survey is inconsistent. The adjustment on capital gains from farm inventories is found to be an oversight by a foreign observer, which was accepted indiscriminately by the domestic followers without rigorous examination of its implications. Another omission of the farm household survey is imputed income from house ownership and rentals, which is included in the city wage earners' income. The use of a deflator in deriving farm incomes in constant prices also requires scrutiny to determine its validity in terms of the relative weights given to expenditure categories, and the number of items surveyed. Because of these theoretical and empirical problems involved in the conventional comparison of the city wage earners' and farm household income, these two types of incomes were reexamined in terms of effective level of living. The adjustments were made for cost-of-living differentials, family size, the direct tax and levy burdens and the omission of income from house ownership. Due to the lack of theoretical consensus on the adjustment procedures and existing data constraints, the adjusted effective level of living incomes are not the best, but second-best approximations were adequate to bring the income concept closer to what was intended at the outset. In terms of disposable effective level of living income, there seems to be no visible trends of widening income differentials between the two types of households, except for noticeable income gaps during the period 1967 to 1970. Some of the erratic fluctuations in the income ratios of the two are attributable primarily to harvest conditions and intermittent changes in the income ceiling imposed. Contrary to the evidence presented by a number of agricultural economists, farm households enjoyed higher levels of disposable effective level of living incomes than the city households throughout the 1970s. What is more significant is that in the reference year 1976, the effective level of living incomes of relatively lower income classes 21. HakchungChoo,"Some Sourcesof RelativeEquityinKoreanIncome Distribution:A HistoricalPerspective,"IncomeDtstrfbution,Employmentand EconomicDevelopmentin Southeast and East Asia_Vol. I, JapanEconomic ResearchCenterandthe CouncilforAsianManpowerStudies,July,1975, pp. 71-72.

146

HAKCHUNGCHOO

among farm households increased much more than higher farm household income brackets as well as the corresponding income classes among city households. In addition, a higher proportion of farm households is concentrated in the middle income brackets than their city counterparts. In the light of these evidences, the arguments for widening urbarvrural income differentials in Korea is found untenable. Finally, two implications from this study deserve mentioning. First, the core of the distributive problems is not inter-sectoral, but intra-sectoral inequality. The inter-sectoral income comparison by average defies the notion of dispersion or distribution. In a rapidly industrializing country like Korea, the center of gravity of the economy shifts to the non-agricultural sector as rapidly as its pace of industrialization. The growing inequalities within urban households and among the self-employed must be recognized as a higher priority policy issue than the inequality between city and farm households in a nation, considering that only about a quarter of the population is in the agricultural sector and this percentage is expected to decrease even further in the near future. Second, there is an unfortunate tendency among analysts to be careless with the data and statistics used for given analytical purposes, especially in developing countries. Without meticulous evaluation of the available basic data and statistics, the inherent biases and limitation can distort the positivism of the empirical analysis. Irresponsible use of existing imperfect and biased statistics should be cautioned against and should in no way be excused by shifting responsibility to the producers of statistics. Analysts must account for imperfect statistics when drawing their conclusions.

REFERENCES Abraham,W. I. (1976), "Observationson Korea'sIncome Distributionand the Adequacyof the StatisticalBase," MimeographedPaper,KoreanDevelopmeritInstitute. Ban, Sun Hwan (1979), "Determinantsof Farm Income and Its Distribution," HakehungChoo, ed., Income Distribution and Its Determinants in Korea, Vol. I, KoreaDevelopmentInstitute. (In Korean). Choo, Hakchung(1980), "EconomicGrowthand Income Distribution,"Human Resources and Social Development in Korea, Korea DevelopmentInstitute.

URBAN-RURAL INCOMES IN KOREA

147

(1979), "Redistributive Effects of Agricultural Tax-cure-Subsidies," Korea Development Review, 1 (3). (In Korean). Economic Council of Japan (1973), Meazuring Net National Welfare of Japan. Kang, Bong Soon and Moon, Pal Yong (1977), "An Analysis of Determinants of Farm Income," Korean Development Institute Working Paper 77-09. (In Korean). Kim, Kwang Suk and Kim, Dal Young 0975), "The Effects of Households Size, Structure and Income on Expenditure Patterns," Korean Development Institute Working Paper 75-10. Kravis, Irving B. et. al (1975), A System of International Comparisons of Gross Product and Purchasing Power, Baltimore: Johns Hopkins Press. Meerrnan, Jocob (1979), Public Expenditure in Malaysia: Who Benefit_ and Why, England: Oxford University Press. Perkins, Dwight H. et. al (1980), Rural Development, Studies in the Modernization of the Republic of Korea: 1945-75, Cambridge: Harvaxd University Press. $ametz, A. W. (1968), "Production of Goods and Services: The Measurement of Economic Growth," E. B. Sheldon and W. E. Moore, eds., Indicators of Social Change, Russel Sage Foundation. Singh, Belvir and Nagar, A. L. (1973), "Determination of Consumer Unit Scales," Econometrica, 41 (2). de Wulf Luc Henry (1975), "Fiscal Incidence Studies in Developing Countries: A Survey and Critique," International Monetary Fund Staff Papers, 22.

ON THE USE OF LINEAR PROGRAMMING FOR FAMILY PLANNING RESOURCE ALLOCATION Rolando A. Danao*

1. Introduction The scarcity of family planning resources in developing countries has led researchers to suggest the use of operations research techniques to allocate these resources in order to optimize the objectives of a family planning program. The objective could be that of minimizing births (or maximizing averted births) given the resources or of minimizing costs given target birth rates. Essentially, a family planning program can be regarded as a production problem with its various activities as inputs and averted births as output. This makes it easy to formulate the family planning program as a resource allocation problem. Consequently, the mathematical programming model has been suggested as the appropriate model to use. Although nonlinear programming models have been proposed e.g., Gould and Magazine (1971), the resource allocation models that have been developed for existing family planning programs are linear. Resource modelling for family planning is fairly recent (Reinke suggested it in 1970) and initial attempts at modelling usually begin with the linear case. Also, linear programming has been successful in solving other resource allocation problems. This paper looks at linear programming (LP) models that have been developed for specific family planning programs and suggests modifications that could (a) give averted births more accurately and (b) reduce the size of the model. An accurate determination of *University of the Philippines School of Economics. 149

150

ROLANDOA. DANAO

averted births is important for policy making while model size reduction could mean savings in computational expenses. 2. Review of Existing LP Models for Family Planning Resource Allocation Early resource allocation modelling for family planning were mostly theoretiad discussions on the applicability of the mathematical programming model to the family planning program. Reinke (1970) presents a simple hypothetical static model which illustrates the formulation of a two-contraceptive method family planning program into a linear programming problem. His model maximizes averted births subject to cost and maximum acceptance constraints. Correa and Beasley (1971) give a basic static model whose objective is to maximize prevented pregnancies subject to a cost constraint. Several modifications of this basic model are presented to show how nondemographic objectives can be incorporated in the model. Both models do not include physical resource (manpower, materials) constraints and do not discuss the estimation of the parameters of the model. Going beyond these theoretical considerations, Lawrence, Mundigo, and ReVelle (1972) developed a linear programming model of resource allocation specifically adapted to the Honduras national family planning program. This model is simple in the sense that it involves only one contraceptive method (pills). But it has several important features that make it a workable model. First, it is a dynamic model which includes changes in demographic variables such as age and fertility of the women of reproductive age and program changes such as continuation and dropout. Second, it distinguishes between costs of a beginning contraceptor and the costs of a continuing contraceptor and allows these costs to change over time. Third, it includes among its constraints not only the budget constraints but also the physical recources available to the program over the planning horizon. Fourth, it also includes a crude birth rate constraint that indicates the desired crude birth rate at the end of the planning horizon. The complete model describes the entire program and minimizes total female births 1 over the planning horizon. 1. The use of female births as decision variablessimplifiedthe model. As the model is intended to be used for planning horizons that extend beyond fifteen years, the number of females that enter the reproductiveStoup is an importantvariablein the model.

LINEARPROGRAMMING AND FAMILYPLANNING

151

This model, therefore, can evaluate the feasibility of program targets given the projected resources and if these targets are feasible, the model gives the optimal strategy for achieving those targets. Building on this model, Lawrence and blundigo (1975) developed another model that incorporates the features of the Honduras model and includes two contraceptive methods (pills and IUDs), the parity of women of reproductive age, and switching to another method. This model was adapted to fit the national family planning program of the Dominican Republic. Following these models, a similar model for the Philippine Family Planning Program was constructed (Pemia and Danao, 1978) that was sl_ecifically designed to evaluate the 5-year population program of the Population Commission. It has the essential features of the Dominican Republic model although it includes five contraceptive methods (pills, IUDs, rhythm, condom, sterilization). Because the planning horizon is less than 15 yeats, the females born during the planning period would not enter the reproductive age group; hence, they will not affect the fertility of the women under consideration. Consequently, the decision variables used were total births. A new feature of the model is the elimination of the crude birth rate constraint. Both the Honduran and the Dominican Republic model contain a crude birth rate constraint which states that the crude birth rate at the end of the planning horizon is less than or equal to a specified target crude birth rate. The presence of this constraint implies that if the target cannot be achieved with the given resources, the LP will simply give no feasible solution without any guide as to what can be achieved by those resources. On the other hand, the absence of the crude birth rate constraint leaves an LP that has an optimal solution from which the optimal crude birth rates achievable by the available resources can be computed. Hence, two things are accomplished: (a) the crude birth rates achievable by the given resources are determined and (b) these achievable crude eoirth rates can indicate to the decisionmaker how far he is from his targets. 3. The Objective Function Minimizing

Births or Minimizing

Costs?

The implementable models discussed in Section 2 minimize births over a planning horizon. This objective function was criticized

152

ROLANDOA. DANAO

by Haran (1979) who claims that "in the context of a family planning program in a developing country, it would seem more appropriate to minimize the total cost of the program subject to constraints on desired levels of birth rate so that the feasibility of achieving the targeted birth rate as well as the minimum cost required to achieve such a reduction in birth rate can be evaluated". Haran's choice of the objective function is a legitimate one but not appropriate for many developing countries. In many of these countries, birth rates are very high and financial resources are very low. The desired birth rates are so far from the existing birth rates that resources are exhausted before the desired birth rates are achieved. Consequently, one's objective i_ simply to maximize the output (fertility reduction) with the limited resources. If the desired birth rate is not far from the existing birth rate and hence, is achievable, then cost minimization becomes an appropriate objective. This happens when the family planning program has already made substantial progress that the actual birth rate is not far from the desired birth rate. In fact, if the desired birth rate has been achieved, one would want to determine the minimum cost that would sustain it. Minimizing Births or Maximizing Averted Births? Conceptually, minimizing births is equivalent to maximizing averted births. The Honduras, Dominican Republic, and the Philippine models have birth minimization as the objective function. We suggest, however, that averted birth maximization has several advantages. First, we note that when viewed as a production problem, the number of averted births is the output of the program. In fact, averted births per money unit (or the cost of an averted birth) has been used as an indicator of the cost-effectiveness of a family planning program. In the birth minimization model, this information is given by the dual problem. We claim, however, that the averted births per money unit given by the dual problem underestimates the actual value. This can be seen by means of a simple example. Consider a one-period two-method model that minimizes births subject to a budget constraint: Primal Problem:

Minimize

_0N

+ _lXl

+ /_2X2

(1)

LINEAR

PROGRAMMING

AND

FAMILY

PLANNING

subject to

ClX 1 + c2X2

153

_-- b

(2)

N + X 1 + X2 = w N, XI,

where 30

(3)

<_ 0

proportion of noncontraceptors one period,

who bear a child in

number of noncontraceptors,

proportion of method m contraceptors child in one period, m = 1, 2,

number of method m contraceptors, m = 1, 2,

cost of providing service to a method m contraceptor in one period, m = 1, 2,

b w

= =

budget for the period, number of women under consideration.

who bear a

Note that the equality constraint is an accounting identity which is included to preclude obtaining the trivial optimal solution N = X 1 =X 2 =0. The dual of the preceding problem follows: Dual Problem:

Maximize

bY 1 + wY2

subject to

(4)

Y2 _

(5)

elY1

+ Y2 _ 32

(6)

c2YI

(7)

<----0, Y2 free.

We note that the primal problem has a feasible solution N = w, Xt = X2 = 0 and since its objective function is bounded below by zero it has an optimal solution (N*, X_, X2_. It follows from the Duality Theorem (Bradley et al. [ 19771 ) that the dual problem has an optimal solution (Y_, Y2*)that satisfies

154

ROLANDOA. DANAO

oN*+81X+82X-bZf

+wr .

<8)

Since the primal objective function denotes births, then each term on the right hand side oft(8) denotes births. The variable Y_ must, therefore, be expressed in'births per money unit, while Y_ must be expressed in births per woman. If we assume that there is at least one noncontraceptor, i.e., N* > 0, then the first dual constraint is binding, i.e., Y2 = lo. Therefore, wY_= wSo, the number of births from all women under consideration in the absence of contraception. Hence, b Y_

-80 w + _0N* + _1 X_ + [32X_

8o (N*+X_+x2*)+8oN*+81X_

+_2X_

= (-80+81)Xf+<-8o+82)x. = + -th)x* + -82)x2"1 We have already noted that Yt * represents births per money unit. Also, Y1 * _ 0. These suggest that Y1 * represents averted births per money unit. This is confirmed by the last equation above since the right hand side represents averted births. But note that the right hand side shows that the averted birth per method m contraeeptor is _o - 13m, i.e., the dual problem assumes that if the contraceptor were not contracepting, her fertility is that of the noncontraceptor. In reality, this is not the case. The eontraceptors tend to have higher fertility than the noncontraceptors since they have a greater motivation to accept contraception. 2 This implies that the dual problem underestimates the number of averted births; hence, also the averted births per money unit. 4. Reducing Model Size A dynamic linear programming model rapidly increases in size as the number of contraceptive methods is increased and the planning horizon is lengthened. Consider the LP whose objective is to minimize births and call it the MINBIRTH model. Its objective function can be written as 2. The Philippine data reveal that the pre-acceptanceage-speclficmarit_d fertility rates of contraceptorsare higher than the age-specificmaritalfertility ratesof allmarriedwomen of reproductiveage. (Lain&1977)

LINEARPROGRAMMING AND FAMILYPLANNING

155

_Z _tNit n b Xlmt b + _Jmt c Ximt) c + _.,ZZ (_mt i t imt where

Nit =

(9)

yearly births per noncontraceptor of age i in year t, number of noncontraceptors of age i in year t,

/_Lt =

yearly births per beginning contraceptor of age i using method rn in year t,

X_rnt =

number of beginning contraceptors method m in year t,

/_nt = yearly births per continuing i using method m in year t, X_tmr=

of age i using

contraceptor of age

number of continuing contraceptors using method m in year t.

of age i

For the age range of 15-44, i.e., 15 < i g 54, there are 30t noncontraceptor variables and 58mr contraceptor variables. If there are five contraceptive methods, i.e., m = 5, the total number of decision variables is 320t. Let us now consider the LP whose objective is maximizing averted b/rths and refer to this model as MAXAVERT. Its objective function can be written as _'_-" i m _t where

[_tmt

_imt

+ 째_imtffiirnt]

(lO)

O_imt

number of averted births per beginning contraceptor of age i using method m in Year t,

O_mt

number of averted births per continuing contraceptor of age i using method m in year t.

Note that MAXAVERT's decision variables do not include the noncontraceptor variables. Consequently, the decision variables are reduced by 3Or. The absence of noncontraceptor variables in MAXAVERT also has an implication in reducing the number of constraints. The equality constraint (3) in the Primal Problem of Section 3 is an accounting identity. This identity is incorporated in the MINBIRTH model in order to preclude obtaining the trivial optimal solution where all decision variables are assigned zero values. For each age i and time t, there is one accounting identity. Hence, there are 30t

156

ROLANDOA. DANAO

such accounting identities MAXAVERT model.

which

are no longer necessary

in the

If sterilization is one of the contraceptive methods offered by the program, more simplifications can be made because of the exceptional characteristics of this method. It is the most cost-effective method in the long run since it has 100 percent continuation, 100 percent effectiveness, and no recurring costs. A dynamic model therefore, will choose to sterilize contraceptors as an optimal solution. Since this is not acceptable, maximum sterilization acceptances constraints have to be incorporated in the LP model. In this case, the model will first assign contraceptors to sterilization until the maximum limits are reached before assigning them to the other methods. However, as shown in Pernia and Danao (1978) no sterilizations are assigned on the last year of the planning horizon. This arises from the fact that on the last year of the planning horizon, the future effects of sterilization are not captured, making sterilization the least cost-effective method. These observations suggest that sterilization should be treated separately. Administratively, the family planning deeisionmaker determines the sterilization program over the planning horizon based on considerations like capacity of sterilization clinics and social acceptability. He then allocates the amount needed for this program. Computationaily, given this program, a simple pre-LP routine can be made to calculate the yearly population of sterilization contraceptors together with the number of averted births. The removal of the sterilization method from the LP model will further reduce the number of variables by 58t and the number of constraints by 30t. 5. The Dual Problem The dual variable corresponding to a resource constraint can be interpreted as the shadow price of each unit of the resource. In the MINBIRTH model, the shadow price corresponding to the budget constraint turns out to be nonpositive (as shown in Section 2), which is not intuitively appealing. However, it was shown that this nonpositive dual variable represents averted births per money unit. But as pointed out earlier, the number of averted births deter_ mined by the dual problem underestimates the actual number of averted births; consequently, it also underestimates the shadow

LINEARPROGRAMMING AND FAMILYPLANNING

157

price. The MAXAVERT model, on the other hand, counts the actual number of averted births and the shadow price corresponding to the budget constraint is nonnegative whose interpretation as averted births per money unit is straightforward. Using the example in Section 3, its MAXAVERT version is written as follows: Primal Problem:

Maximize

otis

subject to

cj Xl

Ot2 X 2

+ c2X2

x ,x2 Dual Problem:

Minimize

subject to

ClZ _ _1

cez _2 z_0. It is easily seen that bz must represent averted births; hence, dual variable z must represent averted births per money unit.

the

6. Conclusion The objectives and activities of a family planning program nearly fit the mathematical programming model. The choice of the objective function depends on the immediate goals of the program as well as the extent of its resources. For programs that want to maintain a certain birth rate, minimization of costs is a proper objective. On the other hand, if the program has to reduce high birth rates with limited resources, the minimization of births or the maximization of averted births would be the objective. We have shown that in this case, maximization of averted births has the advantage of giving a more accurate estimate of the cost per averted birth which is directly obtained from the dual problem. Moreover, the maximization of averted births does not require the inclusion of noncontraceptors as decision variables. This reduces the number of decision variables as well as the number of constraints thereby effecting some savings in computer time.

158

ROLANDO

A. DANAO

REFERENCES Bradley, S. P., Hax, A. C., and Magnanti, T. L. (1977), Applied Mathematical Programming, Reading, Massachusetts: Addison-Wesley. Correa, H. and Beasley, J. D. (1971)"Mathematical Models for Decision-Making in Population and Family Planning," American Journal of Public Health 61. Gould, F.J. and Magazine, M. J. (1971), "A Mathematical Programming for Planning Contraceptive Deliveries," Socio-Economic Planning 5. Haxan, E. G. P. (1979),"An in Family Planning 13-20.

Optimization Model for the Allocation Program," Socio-Economic Planning

Model Science

of Resources Science, 13:

Laing J., et el. (1977), Final Report on the 1974 National Accepter Survey. Population Institute. University of the Philippines. Lawrence, C. E., Mundigo A., ReVelle C. S. (1972), "A Mathematical Model for Resource Allocation in Population Program," Demography, 9: 465-483. Lawrence, C. E., and Mundigo A. I. (1975), "REALPOP: A Mathematical Model for Resource Allocation in Population Program - Results from a Test in the Dominican Republic," Studies in Family Planning, 6 (3). Pernia, Ernesto M. and R. A. Danao. (1978), "Cost-Effectiveness Analysis and Optimal Resource Allocation: The Philippine Family Planning Program," Report prepared for the Population Center Foundation, Manila and the International Development Research Centre, Canada. Reinke, W. A. (1970), "The Role of Operations Research in Population Planning," Operations Research 18.

AN AUXILIARY MODEL FOR QUANTIFYING THE SOCIOECONOMIC IMPACT OF A DEVELOPMENT PROJECT Jos_ Encarnaci6n, Jr. *

1. Introduction A development project is typically designed to advance only one or a few of the objectives that concern development planner_. Different projects specialize, so to speak, in the pursuit of different objectives, and a project may even have a negative impact on another area of concern. (A rural road might, for example, increase productivity in the area but also increase urban unemployment by facilitating rural-urban migration.) With sufficient data and correct model formulation, one could calculate the impact of each project on all the areas of concern. For this purpose it would be usefill to distinguish between: (a) relationships among variables that are specific to projects, and (b) relationships that are common to all projects. With a model of (b) in hand, impact analysis of a project could focus on (a) and then make use of the results already available from (b). A model of (b) is then auxiliary to (a). This paper gives a partial specification of (b) using the 1973 National Demographic Survey (NDS), which data are incomplete for the purposes of comprehensive project impact estimates. 2. Data and Notation The data are from the 1973 NDS of over 8,000 households. Our sample size of 3,196 was obtained by selecting households satisfying *University of the Philippines School of Economics. The author is indebtedto Ms. Elizabeth Jacintowho did the computationsfor this paper. 159

J'OSEENCARNACION, JR.

i60

the following criteria: the family is nuclear or extended vertically to the younger generation ooly; household head is male, working, and his noncash income (if any) is less than t_1,000 annually; the wife married only once and age is between 15-44 years; and information is provided on all the variables listed below. AGn

AM CEB CMR CND DCM DLR DMW

DRC DWP EWm

MWk

PWRj

1if wife is in age-group n, 0 otherwise, wheren = 4ifageis 15-19years 5 if age is 20-24 years 6 if age is 25-29 yeats 7 if age is 30-34 years 8if age is 35-39 years 9 if age is 40-44 years -- age of marriage of wife, in years = number of children born live = CND + CEB = number of children born live and now dead = 1 if CND > 0, 0 otherwise = 1 if rural residence, 0 otherwise = 1 if wife is a migrant whose place of residence is the same as in 1965 and different from place of birth, 0 otherwise = 1 if wife is Roman Catholic, 0 otherwise, --- 1 if wife is working, 0 otherwise = 1 if wife has educational level m, 0 otherwise, where m = 0 for no schooling 1 for one to four years of school 2 for five to seven years of school 3 for. one to three years of high school 4 for high school graduate 5 for one to three years of college 6 for college graduate = 1 if wife is in category k, 0 otherwise, where k = 0 for DMW = 0 1 for DMW = 1 and agricultural residence 2 for DMW = 1 and nonagricultural residence = 1 if wife is incategory ], 0 otherwise, where ] -'- 0 for DWP = 0 1 for DWP = 1 and place of work is at home 2 for DWP -- 1 and place of work is away from home

SOCIOECONOMIC IMPACTOF DEVELOPMENT PROJECTS

YHi

161

1 if husband's annual income is in category i, 0 otherwise, where i = 1C for cash income less than P1000 and noncash income (if any) less than P1000 2C for tb1000-2999 cash income and noncash income (if any) less than P 1000 3 for_3000-4999 cash income 4 for _5000-6999 5 for f'7000-9999 6 for_10000 and above

YICi defined the income YI¢= lifYI¢lC 2 if YW2C 4ifYI¢3 6ifYt¢4 8 ifYW5 11 ifYW6

same way as YHi but with respect to wife's = = = = = =

1 1 1 1 1 1, in thousand pesos.

Individual income data in the 1973 NDS axe reported only in brackets and family income as such is not given. We therefore do not use a family income variable as this would have involved too many categories or else a summing of individual incomes by taking the midpoints of categories as estimates of individual incomes. While the latter procedure is of course possible (cf. Canlas and Encamaci6n, 1977), it makes income data appear more precise than may be warranted. The means of the variables in the s_nple axe given in Table 1. Table1 - Meansof Variables AG4:0.0185 AGS: 0.1302 AG6:0.1990

CND: 0.4143 DCM:0.2663 DLR: 0.6815

EW3:0.1126 EW4:0.0726 EWS:0.0379

PWR2:0.1805 YH1C:0.3457 YH2C:0.4562

AG7:0.2362 AG8:0.2280

DMW:0.2700 DRC: 0.8483

EW6:0.0660 MWO:0.7300

YH3:0.1270 YH4:0.0291

AGg: 0.1881 AM: 19.666 CEB: 4.8276 CMR:0.0671

DWP:0.2447 EWO:0.0685 EWI: 0.2735 EW2:0.3689

MWI: 0.1549 MW2:0.1151 PWRO:0.7553 PWRI: 0.0641

YH$: 0.0217 YH6:0.0203

162

JOSEENCARNACION,JR. 3. The Model

This is based on an earlier paper (Encamaci6n, 1982) which presented a model of choice where wife's fertility and herlabor force participation are determined simultaneously by her educational level, husband's income, and other variables. Briefly, the model implies the existence of "threshold values" for wife's educational level, husband's income and family income, such that the qualitative effect of a variable changes when it passes the thresholds. Figure 1 illustrates. In the upper panel of Figure 1, number of children C is measured on the vertical axis while family income Y and wife's educational level E - Y and E are assumed to be perfectly correlated for purposes of a simple diagram - are measured on the horizontal axis. Natural fertility or capacity to bear children Ck increases with Y and E for reasons of better health and nutrition; number of child deaths Cm decreases for the same reasons. The number of children desired _C° falls with E and Y for a variety of reasons. What would then be observed for the number of children born would be the curve abCb, and the number of surviving children the curve cdC °. These two variables are thus nonmonotonie functions of E and Y, whose qualitative effects change at the threshold value Ec *. In the lower panel, the proportion of wife's time spent at market work t is measured on the vertical axis while E and husband's income Yh are measured on the horizontal axis. The wife's wage rate depends on E and we assume that the curve c'd'e_t ' indicates what is required of t if minimum consumption standards for the family are to be met. On the other hand, the curve ee't o indicates what t would be if the wife's choice were not required to satisfy consumption standards. With this requirement, the observed t would be the curve c'd'e't °. The threshold value Et* defined by the intersection point e' is such that E¢ * _< Et* under relatively weak assumptions: We thus have a roughly V-shaped curve for wife's labor force participation rate and an inverted V-shaped curve for her fertility as functions of education and income variables. Estimates of these two relationships are given in eq. (1) below and eqs. (2)-(5)set out in Table 2 (t-values in parentheses underneath regression coefficients). (1)

DWP= 0.3949 - O.0400EWO (-0.83)

- 0.1642 EW(-3.97)

1. In the earlierpaper cited above,it was shown that Ec* somewhatstrongerassumptions.

= Et*

under

SOCIOECONOMIC

IMPACT OF DEVELOPMENT

PROJECTS

°1 Ck

I I

•

I I E©*

E,Y

I I I

Ly h

_, FIGURE I

163

164

JOSE ENCARNACION, JR.

- 0.2078

EW2 - 0.1722

(-5.15)

(-3.92)

- 0.1900

EW4 + 0.2685

(-4.09) + 0.0521

(5.63) 0.0541

YH1C-

(0.96) - 0.0120

EW3

EW6 YH2C

(-1.02) YH3 - 0.0189

(-0.22) -2 R - 0.084

(-0.29)

YH4 - 0.0675

YH6

(-0,95)

Eq. (I) is essentially in accordance with the model, with a trough for EW at EW2 (five to seven years of schooling), although them is apparently an aberration cient is less than that

at EW4 (high school graduate) of EW3. Since the dependent

whose coeffivariable is a

dummy for actual employment rather than labor force participation, however, it is possible that high school graduates arc simply getting much less employment than they want. A trough seems to occur for YII at YH2C (where a poverty line could be drawn), but t-values for the YHvariables are all weak. Table 2 -- Resression Equations for CEB (2) const.

11.543

(3)

(4)

11.457

11.539

(5) 11.454

-0.2962 (-34.01 )

-0.2960 (-33.94)

-0.2956 (-33.88)

-0,2954 (-33.81 )

AG4

-6.1980 (-25.14) -4.7382 (-41.33)

-6.1899 (-25.04) -4.7555 (-41.36)

-6.1966 (-25.13) -4.7369 (-41.32)

-6.1854 (-25.02) -4.7528 (-41.34)

AG6

-3.0293 (-30.63)

-3.0392 (-30.66)

-3.0276 (-30.61)

-3.0362 (-30.63)

AG7

-1.3509 (- 14.47)

-1.3539 (- 14.50)

-1.3517 (- 14.47)

-1,3544 (- 14.49)

AG9

1.0110 (10.25) 0.3835 (1.84)

1.0189 (10.32) 0.3291 (1.54)

1.0105 (10.23) 0.3688 (1.76)

1,0181 (10.31) 0,3238 (1.51)

AG5

EWO

SOCIOECONOMICIMPACT OF DEVELOPMENT PROJECTS

165

Table 2 (Continued)

(2)

(3)

(4)

(5)

EWl

0.6199 (3.52)

0.5600 (3.07)

0.6055 (3.42)

0.5547 (3.04)

EW2

0.4615 (2.67)

0.4012 (2.26)

0.4505 (2.60)

0.3976 (2.34)

EW3

0.4606 (2.43)

0.3981 (2.07)

0.4591 (2.42)

0.4023 (2.09)

EW4

0.0913 (0.45)

0.0533 (0.26)

0.0929 (0.46)

0.0587 (0.29)

EW6

-0.0844 (-0.41)

YH1C

-0.0160 (-0.08)

-0.0941 (-0.45)

0.1035

-0.0288 (-0.14) 0.0901

(0.44)

(0.38)

0.2065

0.2016

YH2C YH3 YH4

(0.89) 0.1095 (0.46) -0.1025 (-0.36)

(0.89) 0.1.134 (0.48) -0.1025 (-0.36)

YH6

-0.3176 (-1.03)

-0.3046 (-0.98)

DRC

0.1936 (2.13 )

0.1888 (2.07)

MW1

0.2561 (2.87)

0.2461 (2.76)

MW2

0.1094 (1.07)

0.1070 (1.03)

DMW DWP

0.1939 (2.13)

0.1894 (2.08)

0.1956 (2.71) -0.2548 (-3.31)

0.1898 (2.62) -0.2465 (-3.19)

PWRI

-0.3104 (-2.38)

-0.3061 (-2.34)

PWR2

-0.2308 (-2.62)

-0.2193 (-2.47)

0.562

0.563

166

JOS_EENCARNACION,JR.

In Table 2, eqs. (2)-(5) for number of children born live are set out as columns; (2) omits the YH, MI_ and PWR variables while the other three equations similarly omit some of the variables listed in the first column. In all four equations, apeak forEICis seen atEW1 (one to four years of schooling). A peak for YHis apparent at YH2C in (3) and (5) even though t-values are weak. (The model calls for family income here and not husband's income, but we use the latter as a rough proxy.) Of interest are the dummies for religion (DRC), migrant status (DMW) and current employment (DWP), which are all significant. Apparently, looking at (2) and (3), being a Catholic adds about 0.2 children to a couple, as also being a migrant, while being employed reduces family Size by 0.25. A closer look at the migrant and employment variables shows finer detail. Eqs. (4) and (5) use MW1, MW2, PWR1 and PWR2 in place of the more crude DMW and DWP. Here it is migrants to agricultural areas (who are likely to have come from other agricultural areas) who have higher fertility, while other migrants exhibit a smaller increase not significantly different from zero. This would be consistent with the model if one considers that agricultural migrants probably improve their livelihood relatively more than do other migrants. (Cf. Encarnaci6n, 1977, for similar suggestive results in Southeast Asia.) As for the employment dummy variable, a breakdown of this to PWR 1 and PWR2 gives results that appear to go against usual expectations. Here we find that ceteris paribus, wives who work at home have apparently less children than those whose place of work is away from home. A possible explanation may be that wives working at home find it difficult to hold down a job away from home because of poorer health; their working at home and having less children would then be due to the same set of circumstances. 2 As the CEB equations following equation: (6)

AM-_ 21.575-

involve age-at-marriage AM__,we have the

1.9811EWO(-4.45)

1.9765 EW1 (-5.15)

2. However, it should be noted that if the regression coefficients of PWR1 and PWR2 are treated as means in a standard test of the difference between two means, we fred that the difference between them is not large enough to reject the null hypothesis.

SOCIOECONOMIC IMPACTOF DEVELOPMENT PROJECTS

167

- 1.9149 EW2 - 1.4952 EW3 (-5.13) +O.0831EW4+ (0.19)

(-3.70) 2.1397EW6-O.7416DLR (4.91) (-4.63)

-2 R =0.101

This is quite in conformity with usual expectations: AM is higher with higherEW and is lower for rural women. Finally, the NDS data permit the estimation of several equations concerning child mortality. Eqs. (7) and (8) below have CMR, the ratio of child deaths to children ever born, as a function of educational level and, in the case of the latter equation, of husband's income (as proxy for family income) as well. Its relationship to these (7)

CMR = 0.0395 + 0.0775 EWO + 0.0423 EW1 (5.04) (3.21) + 0.0281 EW2 + 0.0158 EW3 (2.17) (1.10) -2 - 0.0028 EW4- 0.0175 EW6 R = 0.023

(-1.18) (8)

F = 13.76

(-1.13)

CMR = 0.0291 + 0.0692 EWO + 0.0347 EW1 (4.38) (2.54) + 0.0221EW2 + 0.0111EW3 (1.66) (0.77) - 0.0062 EW4 - 0.0155 EW6 + 0.0251 YH1C (-0.41) (-0.99) (1.41) + 0.0115 YH2C+ 0.0167 YH3 (0.65) (0.93) - 0.0023 YH4 - 0.0009 YH6 (-0.11) (-0.04)

- 2 R = 0.025

F = 8.34

two variables is generally monotonic as one might expect. Perhaps more useful, however, axe eqs. (9) and (10), where the dependent variable DCM is a dummy equal to one if a child has died. DC_ could be interpreted as the probability (approximately) of a child death as a function of the variables on the right-hand side. It could then serve as a crude proxy for health. (9)

DCM = 0.7637 - 0.0233 AM - 0.3851 AG4 (- 11.54) (-6.75) - 0.3069 AG5 - 0.2099 AG6

168

JOS_.ENCARNACI(3N, JR. (-11.66)

(-9.19)

-- 0.1060 AG7 + 0.0734 AG9 + 0.1608 EWO (-4.90) (3.21) (3.38) + 0.0985 EW1 + 0.0629 EW2 (2.42) (1.58) + 0.0322EW3 (0.73) (10)

- 0.0162EW4(-0.35)

-2 0.0485 EW6 R = 0.122 (-1.02)

DCM = 0.7530 - 0.0231 AM - 0.3976 AG4 (-11.44) (-9.96) - 0.3116AG5 (-11.81)

- 0.2145 AG6 (-9.38)

- 0.1076 AG7 + 0.0747 AG9 + 0.1288 EWO (-4.98) (3.27) (2.64) + 0.0684 +EI짜1 + 0.0383 EW2 (1.63) (0.94) + 0.0117 EW3 - 0.0313 EW4 - 0.0371 EW6 (0.26) (-0.67) (-0.77) + 0.0587 YItlC + 0.0229 YH2C (1.08) (0.43) -2 + 0.0272 YIt3 - 0.0398 YH4 - 0.0915 YH6 R = 0.124 (0.49) (-0.60) (-1.28) probability (approximately) of a child death as a function of the variables on the right-hand side. It could then serve as a crude proxy for health. 4. Using the Model With due caution, one can use the regression equations reported above 3 for purposes of estimating the impact of a development project on some variables of concern: fertility, labor force participation, and health. Accepting the usual interpretation of crosssection regression results as long-term relationships among the 3. These are ordinary least-squaresestimates since the model can be taken as reeursive.

SOCIOECONOMIC IMPACTOF DEVELOPMENT PROJECTS

169

variables, the procedure would simply be the following: Calculate the changes in the "independent" variables resulting from the project; then use the regression equations to estimate the changes in the dependent variables. The latter changes are then imputable to the project as its impact. For example, suppose that one long-term effect of a project is to raise male family heads' incomes in the region frOm YH1C to YH2C. From eqs. (1), (5) and (10), the coefficients of these two dummies and the differences between them are given in Table 3. Accordingly, we obtain estimates of a reduction in wives' labor force participation, an increase in births and a decrease in child deaths by multiplying the last column in Table 3 by the number of families involved. Comparability among projects can then be had by expressing the estimates per peso of project costs. Table3 - Coefficieats of YHIC andYH2C Equation

YH1C

YH2C

Difference

(1) DWP

.0521

-.0541

-.1062

(2) CEB

.0901

.2016

.1115

(10) DCM

.0587

.0229

-.0358

If a project affects other "independent" variables, similar computations can be made and then added to get the total impact of a project, since the effects of the independent variables are additive in the regression equations. 4 Several observations might be made regarding the estimates so obtained. First, they are not predictions of changes between the present and the future (after the installation of a project), since some changes will occur with or without the project. One is here simply estimating the difference that a project makes, cetefis paribus. Second, the estimates have to do with long-term, not short-term, 4. It is sometimes of interest to calculate the relative contributions of the independent variables to the variation of the dependent variable in a regression equation, especiallywhen the independent variablesare "mixed" as in eqs. (9) and(10); see the Appendix.

170

JOS_ ENCARNACI_)N, JR.

results. Finally, it is on the basis of some model which one considers correct that one justifies any particular interpretation of statistical observations - one cannot discuss the latter in a theoretical vacuum. This last observation would not be worth mentiorting were it not that statistical data are sometimes erroneously thought to be capable of "establishing" a causal relationship.

Appendix Relative Contributions of Mixed Variables to the Variation of a Regressand Consider a regression equation whose regressors include classi- " ficatory as well as ordinary scalar variables. A classificatory variable is essentially a vector that has as many components as there are different (mutually exclusive and exhaustive) categories in the classification. For example, one might estimate a regression equation that explains employees' salaries in terms of length of service (a scalar), occupation (a classificatory variable), etc. One might then want to estimate the relative contributions of the explanatory variables to the variation of the dependent variable. Handling this problem by beta coefficients is well known when the explanatory variables are all of one kind, either all scalar or all classificatory. There seems, however_ to be no convenient reference that discusses this matter when the explanatory variables are mixed, i.e. when they include both kinds. This expository note might therefore be of some use. I Let x - (x o, x 1.....

XK)

where Xk = 1 for an individual (or

observation) if it belongs to category K k (k = 0, 1, . . . , K) of classification x, xk = 0 otherwise, and _k=0 Xk = 1. More precisely, for any given individual i, Xk f = 1 if i is in category k, 0 otherwise, and _'k=O Xki = 1. To each i thus corresponds xi = (Xoi, Xli ..... XKi ). Suppose it is appropriate to explain y in terms of x, z, u and v by means of a regression equation, where z is another classificatory variable (z 0, z 1..... z i while u and v are real variables. (Discussion of more than two variables of either kind would be straightforward.)

SOCIOECONOMIC IMPACTOF DEVELOPMENT PROJECTS

171

We calculate K (1)

y'=c+Y.,a_1

xk+ZlbTZl+p(u-ff)+q(v-P

)

where the aft, bh* p and q are the regression coefficients and y' is the predicted y. As usual, overbars denote means. Note that x 0 and z 0 are omitted in (1) in order to have determinate coefficients (Suits 1957). We want to express (1) in the form K J (2)

yW=_ + Z0 ak Xk + _0 biz/+ p(u -- _) + q(v -- V)

where x o and z 0 are included, and the ak and bl measure the effects on an individual's y resulting from its belonging to k of x and to j of z, respectively. It is to be noted that the ak and b/, which might be called category effects (Encamaci6n 1975), are measured from V. For suppose that for an individual i, xki = 1 for a particular k and zli = 1 for a particular ]. Then I

Y- + ak + bI + p(ui - _) + q(vi - _)

so that ak and bi are simply added on to y. From least squares properties, K

(3)

c = 7-

using (1),

_1ak _k -- _ bl z/ - p(_--- u--')- q(V- -_) K

=Y--_akX

_-k--_b_.

But c is also the predicted y for an individual satisfying x 0 = 1, z 0 = 1, u = ff and v = V. Therefore K

(4)

a0 = - _..a_Â˘Zk 1 J

(5)

b0 -- - E1 b_.

Further, if an individual satisfies x k = I (k :/: 0), z 0 = 1, u -- if, v = v, the predicted y is c + aft. Since we already know from (3)(5) that

172

JOS_. ENCARNACION, JR.

(6)

c = 7+ a 0 + b 0

we have c + a_ = V+ (a o + a_) + boso that (7)

ak = a o + a_

k = 1, ....

The b1 are similarly determined. Substituting (6) in (1), i (8)

Y = Y+ a° + b° + Y'Ia_ xk 4- _1 b/ z/+ p(u - g) + q (v - _-) K = Y+a°

+b°

+ _1 (ak-a°)x_

4- _1 (b]-

bo)z /

+ p(u - u-) + q(v - v-) K

=Y+ao (1- Zxk)4- Zak xkl4-b°(1-- Zz/) J + _ b/z/+ 1

p(u - g) + q(v - 9).

But 1 - _1K xs = x 0 and 1 - E_ z/= z 0 ; hence (2). We note for later reference that x k = ns.]n, where ns. is the number of individuals for which xst = 1 and n is the total number of individuals. Also, as one might expect, n (9)

hffil_ Sffi0 _-" as Xkh/n

= _0 as nk./n

ffi_0 ak £-k = 0

i.e., the mean Zff as xs = 0 (in the same way that the mean p(u - u-), say, is zero). For, multiplying (7) by nk., summing both sides and then adding no. a o to the results, K

0 nk. as =ha 0 4-_1 ns. as which,

in view of (4), gives (9). H

The standard

motivation multiple

for calculating the partial beta coefficients of regression is to be able to compare the relative

SOCIOECONOMIC IMPACTOF DEVELOPMENT PROJECTS

173

contributions of the explanatory (scalar) variables to the variation of the dependent variable (see, e.g, Ezekiel and Fox 1959, p. 196). Accordingly, the variables are standardized to zero means and unit variances, so that their beta coefficients become directly comparable. Similarly, the beta coefficients discussed by Morgan et al (1962) perform the same function in the ease of classificatory variables. Our problem is to see whether all the beta coefficients in a regression with mixed variables arc directly comparable. Write

(10)

_[ix/(x)+[izg(z)+[iu

u-----_-_+t% v-_ su

which is to be equivalent to (cf. (2))

(11)

y'--y-_

_OakXk

-I. _JobIz!

+ p(u-u)

q(v--V)

where sy is the standard deviation of y, etc., (12)

flu ffi P su lsy

which is the textbook for [iv, and

(13)

definition

of a partial beta coefficient, similarly

K 2 [Ix = (7-'0 ak nk./(n - 1))1/2

from Morgan et al. (1962). The functions f(x) and g(z) are implicitly deemed by the equivalence of (10) and (11) and the definitions of the [i's. It is clear that if [iu >/_v, u contributes more than does v to the explanation of y variation. Our object is to show that f(x), say, standardizes x essentially in the same way that (u- u-)/ Su standardizes u, so that all the beta coefficients are then directly comparable. From (10), (11) and (13), for individual t,

174

(14)

JOS]_ ENCARNACI(3N,

JR.

_kK=o a k xki

f(xi) =

(_K= 0 a_ nk./(n -- 1))1/2 from which

(15)

f=0

a 2 x2i

Z"h=, Y-'_=oal Xk2h/(n-

since cross-product terms vanish and xki -- x_i (because xki = 0 or 1 and _,x__0 xki = 1). But (ut -- _-)2

(ul -- _)2

(16) S2

_=1

p2(Uh

_)21(n-

corresponds precisely to (15), the only difference being that while one can factor out p2 in (16), which Kof 2course does not affect the ratio, it is not possible to factor out Z0 ak in (15), which pertains to a vector. The key observation is that x being a classificatory variable, _g_=0a_ xkt is the analogue of p(ui - u-)and both have zero means. This completes our task, and all the beta squares may then be ranked to indicate the relative contributions of their corresponding variables to the explanation of y variation. REFERENCES Canlas, D. B. and Encarnacion, L (1977), "Income, Education, Fertility and Employment: Philippines 1973," Philippine Review of Business and Economics, 14 (2): 1-27. Encarnaci6n, J. (1975), "Income Distribution in the Philippines: The Employed and the Self-Employed," in Income Distribution, Employment and Economic Development in Southeast and East Asla, Tokyo: Japan Economic Research Center, pp. 742-775. .... (1977), "Population and Development in Southeast Asia: A Fertility Model," Philippine Economic Journal, 16:319-340. .... (1982), "FertiliW Behavior and Labor Force Participation: A Model of Lexicographic Choice," in Research in Population Economics, ed. J. L. Simon and P. Lindert, Greenwich, Conn.: JAI Press. Ezekiel, M. and Fox, K. A. (1959), Methods of Correlation and Regression Analysis, 3rd ed., New York: Wiley. Morgan, J. N. et aL (1962), Income and Welfare in the United States, New York: M_e(___Vaw-Hill, Appendix E. Suits, D. B. (1957), "Use of Dummy Variables in Regression Equations," Journal of the American Staffstical Association, 52:548-551.

PHILIPPINE DEMOGRAPHIC DEVELOPMENT: PROBLEMS AND PROSPECTS Ale]andro N. Herrin *

I. Introduction The population of the Philippines grew from 7.6 million in 1903 to 48.1 million in 1980, an increase of more than sixfold in eight decades. From 1903 to 1939, the average annual intereensal growth rate was around two percent. In the postwar period from 1948 to 1970, the average annual intercensal growth rate rose to three percent. The rapid growth in the 1950s and 1960s was unprecedented, arising largely from the rapid decline in mortality without a compensating decline in fertility. Fertility, however, began to decline in the 1970s, so that between 1970 and 1980, the average annual intereensal growth rate declined to 2.7 percent. This rate, however, is still high by international standards. The increased concern regarding the adverse consequences of uncontrolled population growth on economic and social development has led the government at the turn of the 1970s to adopt an official population policy whose main focus was the control of the rapid population growth through fertility reduction. The main population program was the family planning program. In 19_/8, a Special Committee to Review the Philippine Population Program recommended that the role of the family planning program be broadened to emphasize not only fertility reduction but also the overall welfare of the family, and that the population program be designed on a broader scale and be fully integrated in the *Professorof Economics,Universityof the PhilippinesSchool of Economics. 175

176

ALEJANDRON. HERRIN

development plans of the country. The recommendations of this Committee provided the basis for the thrust of the national population program since then, as reflected for example in the Philippine Population Program Medium Term Plan 1981-1985. In more recent months, however, questions were raised regarding the performance of the population program with respect to fertility reduction, as well as the future role of the family planning program in the overall development strategy. The glaring omission of a description of the family planning program in the recently prepared national development plan (1983-1987) has led to questions as to whether there has been a shift in population policy especially with respect to fertility reduction. The recent announcement by the Prime Minister to review population planning in the Philippines perhaps also suggests some ambivalence among planners regarding population policy, strategy and program emphases. These recent developments suggest a need to review Philippine demographic development as a basis for identifying issues in population planning for the 1980s. This paper reviews the trends in the basic demographic processes of fertility, mortality and population movement. Space limitations preclude presentation of the basic data described in this paper. These data, however, are compiled and presented in Herrin (1981) and Concepcion and Smith (1977). If the volume of past and current Philippine demographic research is any indication, it would appear that issues related to mortality and population movement have been neglected in favor of fertility and family planning. Hence, it is essential to put these two processes back into the population picture if population planning is to involve more than just fertility reduction. As will be evident in the subsequent discussions, there still exist serious information gaps regarding demographic trends in the Philippines. The next three sections of this paper examine the trends in mortality, fertility and population mobility and attempt to identify emerging problems as well as prospects for the 1980s. The last section concludes. 2. Mortality National Trends Analysis of available mortality estimates in the Philippines reveals a historical pattern of (a) gradually declining mortality during the

DEMOGRAPHIC DEVELOPMENT

177

early part of the century up to the beginning of the Second World War; (b) rapid decline during the early postwar years up to 1960; and (c) slackening of the decline thereafter up to the mid-1970s. The dimensions of these mortality changes can be described as follows. The crude death rate decline from the 1904-1905 level of 26.8 deaths per 1,000 population to 23.3 in the 1936-1941 period. This represented a decline of 13 percent in 34 years or 0.10 deaths per 1,000 per year. From 1948-1950 to 1960, the crude death rate declined by 41 percent, that is from 21.6 to 12.8 per 1,000, or 0.8 deaths per 1,000 per year. In this period of rapid mortality decline, life expectancy at birth rose from 42.5 to 52.8 years, a difference of 10.3 years in a little over a decade. From 1960 to 1975, however, the crude death rate decline from 12.8 to 8.7 per 1,000, representing a decline of 32 percent or only 0.27 deaths per 1,000 per year. Life expectancy at birth, on the other hand, rose to 59.4 years or only 0.44 years annually over the 15-year period, compared to 0.94 years annually during the 1950s. Thus, there appears to be a slackening of the rate of mortality decline in the more recent period. Moreover during this 15-year period, infant mortality declined from 113 infant deaths per 1,000 births to 76 in 1975. Overall, the levels of life expectancy achieved in the Philippines in 1975 is still about 10 years lower than that achieved by the industrialized countries in 1960, while the Philippine infant mortality rate in 1975 is still some 2.6 times higher than the average found in industrialized countries in 1960. These rough orders of magnitudes suggests that the Philippine mortality conditions still leave much to be desired. Areal and Social Group Differentials The decline in mortality during the postwar period was not uniform throughout the regions and provinces of the Philippines nor among various social groups. The 1970 estimates provided by Flieger, Abenoja and Lira (1981), (the most recent estimates available for regions and provinces[ ) reveal that between regions, the age-standardized etude death rates varied from close to 10 per 1,000 in the more economically advanced and urbanized regions of Central Luzon and Southern Tagalog (including Metro Manila) to as high as 16 per 1,000 in Western Mindanao. Infant mortality rates varied from 75 in Southern Tagalog (including Metro Manila) to 135 in Northern Mindanao. Finally, life expectancy varied from a low of 47.5 years in Western Mindanao to a high of 60.2 in Central Luzon. In view of the

178

ALEJANDRON. HERRIN

lack of comparable up-to-date data on mortality differentials by areas, it is difficult to determine the extent to which such differentials have narrowed in the 1970s. In view of the relatively slower pace of mortality decline in the more recent period, it is not likely that such differentials have narrowed considerably by 1980. Data on mortality differentials by social groups are also hard to come by. In a recent study of infant mortality by selected socioeconomic characteristics of parents based on the 1978 Republic of the Philippines Fertility Survey (RPFS), Esclamad, de Guzman and Engracia (1982) reveal large infant mortality differentials by mother's education. Infant mortality for the period 1973-1977, ranged from 105 for women with no schooling, 73 for women with primary schooling, 54 for women with intermediate schooling, 41 for women with high school, and 37 for women with college education. Rural-urban differential was also noted: 45 versus 66 infant deaths per I000 births. Additionally, infant mortality was higher for higher parity women and for women with closely spaced births. Slackening of the Rate of Mortality Decline A disturbing aspect of the historical mortality trend noted earlier is the slackening of the rate of mortality declines in the more recent period, i.e., from the 1960s on. An earlier study (Zablan, 1977) suggested that a threshold level of mortality may have been prematurely reached at a life expectancy at birth of less than 60 years. The slackening of the rate of mortality decline, however, is not unique to the Philippines. A recent international study revealed that in many developing countries, there has been a discernible slowing down of progress in mortality reduction relative to what would have been expected on the basis of European experience. (Gwatkin, 1980). Gwatkin suggests that, in general, this worldwide slackening in mortality gains may be due to the diminishing capacity of public health and medical care program to deal with the evolving disease patterns and to the general slackening of socioeconomic development in the 1960s and 1970s relative to that of the early postwar period. In the Philippines, one may specifically associate the rapid mortality declines in the immediate postwar period on the one hand, to the development of the rural health program in the mid-1950s, notably the establishment of the Rural Health Units as the vehicle for the introduction of modern health technology and the construction of potable water supply systems in the rural areas; and on the

DEMOGRAPHIC DEVELOPMENT

179

other hand, to the generally rapid economic growth during the 1950s. The impact of these two broad sources of mortality decline, however, appears to have lost their force by the 1960s as the data on mortality trends suggest. Several reasons may be advanced. First, in the health field, problems of staft'mg the health units, lack of medicine and supplies, and inadequate travel funds for rural health personnel increasingly imposed severe limits to rural outreach efforts. (Jacinto, 1969). Public health expenditures tended instead to increasingly concentrate in urban areas emphasizing curative rather than preventive medicine. As a result, further mortality reduction in the rural areas especially among infants tended to slow down leading to a slower improvement in life expectancy. Second, the rapid economic growth in the 1950s generated by the narrowly-based import substitution industrialization began to lose steam in the 1960s. The concomitant relative neglect of domestic agriculture and the consequent lag in food production contributed to the slowing down of overall growth. Real per capita consumption expenditures, which grew by 3.9 percent during the 1940-1960 period, rose by only 1.5 percent in the 1960-1970 period, and 2.0 percent from 1970 to 1979. In addition to this slowing down of per capita growth of personal consumption expenditures, income distribution has either worsened or at least has not significantly improved. Under these circumstances, the momentum of rapid mortality declines in the 1950s could simply not be sustained. In addition, the unevenness in the distribution of health services and the differential access to such services by households have not been conducive to reducing areal and household mortality differentials. Problems and Prospects In spite of the rapid gains in mortality reduction during the postwar era, infant mortality is still high and life expectancy at birth is still low compared with the levels achieved by developed countries. Moreover, significant mortality differentials still exist between regions and provinces and between social groups, reflecting both inequalities in the distribution of income on the one hand, and of health facilities and services on the other. In addition, progress towards mortality reduction appears to have prematurely slowed down in the more recent periods. In view of the fact that freedom from premature death is a direct component of human welfare, steady improvements in life expect-

180

ALEJANDRON. HERRIN

ancy up to levels approximating those found in the more developed countries should be a major concern of any development strategy. Because future mortality trends are going to be tied up more closely with socioeconomic development than in the past, programs now being implemented to increase employment, food production and incomes especially among the poorest segments of society can be expected to have substantial impact on future mortality reduction. But progress along these lines can be rather slow. In the meantime, can the potential for further mortality reduction be enlarged through selective health interventions at a cost not much greater than that currently allocated to the health sector? Past experience with "direct interventions" provide cause for guarded optimism (Gwatkin, Wilcox and Wray, 1980; Williamson, 1982). The potential contribution to mortality reduction of selective health interventions will invariably depend on the nature of such intervention and on its coverage. The nature of such interventions in turn will depend on what is known about the prevailing mortality patterns in specific areas and of the determinants of these patterns. Unfortunately, this is one area of mortality studies in which information is most unreliable in view of the inherent difficulty in assessing the precise cause of death and of incomplete coverage of the current monitoring system. Nevertheless, it may be instructive to look at the evolving mortality pattern from cause of death statistics compiled by the Health Ministry for what they suggest in terms of demographically significant health problems. Data at the national level, for example, reveal that the percentage of all deaths due to pneumonia and tuberculosis are still the highest in 1977 as they were in 1946, both accounting for 26 percent of all reported deaths. Gastroenteritis, nutritional deficiencies, bronchitis and related respiratory diseases accounted for another 11 percent of all reported deaths in 1977. With respect to infant deaths, pneumonia accounted for 26 percent of all reported deaths in 1977 with gastroenteritis, nutritional deficiencies, bronchitis and related respiratory diseases accounting for another 22 percent. Progress along these lines within the context of a restructured and expanded health care delivery system can be expected to provide added impetus towards further mortality reduction in the future. At the local level, there is need to determine the demographically significant health problems, and to determine whether the health services currently in place effectively address these problems. For

DEMOGRAPHIC DEVELOPMENT

181

example, can current programs focused on certain causes of death such as neonatal tetanus and maternal mortality be justified on the basis of the demographic significance of these causes of death? Would not the same resources be applied more efficiently through programs that will address the more significant causes of deaths such as respiratory diseases via a curative program at the village level (since these diseases are difficult to prevent), e.g., through greater availability of antibiotics? (See Williamson, 1982). Needless to say, population planning in the area of mortality reduction requires much more information on trends in mortaiity_ on changing patterns of cause of death, and on the impact and costeffectiveness of current health interventions than has so far been available. 3. Fertility Fertility

Trends

Data on long-term fertility trends reveal that the moderate decline occurring since the 1950s has accelerated in the 1970s. From a level of 50 or mote births per thousand population at the turn of the century, the crude birth rate remained fairly constant during the first half of the century. Since the 1950s, fertility began to decline, gradually reaching around 46 births per 1,000 in 1960 and to around 40 per 1,000 in 1970. A somewhat faster decline occurred in the 1970s. The National Census and Statistics Office estimated the crude birth rate to be 35 per 1,000 in 1975 and 33 per 1,000 in 1980. This level for 1980 is still about twice that of the average for industrialized countries. Estimates of total fertility rates based on survey data further reveal the decline in fertility for the most recent period. Total fertility rates deciined from 6.5 births per woman in 1958-62 to 6.3 in 1963-67 to 5.9 in 1968-72 and 5.2 in 1973-77. Estimates for 1980 plac_ the total fertility rate at 4.5. (Engracia, 1982). Data constraints limit attempts to map out the incidence of fertility decline in the Philippines. Nevertheless, a broad view of the pattern of fertility transition can be obtained from an analysis of the available cross-section information, specifically the data on areal fertility differentials and fertility differentials by individual and household socioeconomic characteristics. Total fertility rates by broad geographic areas in 1973-77 reveal

182

ALEJANDRON. HERRIN

increasing fertility as one moves from Metro Manila to other urban areas and to rural areas: 3.9, 5.6 and 5.3, respectively, after the rates were standardized for marital status. The time trend in total marital fertility rates from 1958-62 to 1973-77 reveal that the fastest decline occurred in Metro Manila and other urban areas: 21 and 12 percent declines, respectively. In contrast, fertility decline in the rural areas was relatively small (6 percent over the 15-year period), all of it occurring only in the last five-year period. Total fertility rates by region suggest that the most rapid fertility decline occurred in the more highly developing and urbanizing regions. Although regional fertility differentials have narrowed somewhat in the more recent period, significant differentials still exist. At around 1975, five regions had total fertility rates less than 5.0 births per woman, ranging from 3.1 to 4.9. The top three regions with the lowest fertility are the urbanized regions of Metro Manila, Central Luzon and Southern Tagalog, together accounting for 34 percent of the national population in 1975. The two other relatively low fertility regions are Ilocos and Central Visayas. In contrast, the remaining seven regions (all of the 4 Mindanao regions, two of the Visayas and two of Luzon) still exhibit high fertility in 1973-77, ranging from 5.0 births per woman in Western Visayas to 6.0 in Bicol. These regions include those among the least urbanized (Cagayan, Bicol, Eastern Visayas, Western and Central Mindanao), having the highest incidence of poverty (Cagayan, Bicol, Eastern Visayas and Northern Mindanao), and the highest infant mortality rates. Within regions, the incidence of fertility change can be gleaned from cross-section evidence on children ever born to. ever-married women age 45-49 by individual and household characteristics obtained from the 1978 Republic of the Philippines Fertility Survey (RPFS). The data show that, after standardizing for age at first marriage, lower fertility tends to be associated with women of higher educational attainment, living in urban areas and whose husbands are engaged in non-agricultural skilled or semi-skilled occupations. The major proximate determinants of fertility change in the Philippines are changes in nuptiality patterns and changes in marital fertility, the latter mainly through contraception. Although induced abortion is known to be resorted to by both single and married women as a means of fertility control, the extent of its use is not accurately known.

DEMOGRAPHIC DEVELOPMENT Changing Nuptilaity

183

and Its Correlates

Marital status distribution by age obtained from census data from 1903 to 1975 and from the 1978 RPFS reveals a significant trend in delayed marriage for females. The percentage single increased steadily especially among younger women ages 15-19 and 20-24 years over the seven decades, i.e., from 74 and 33 percent, respectively, in 1903 to 87 and 51 percent, respectively, in 1975. The singulate mean age at marriage (SMAM) has risen from 20.9 years in 1903 to 23.2 yeats in 19.75. The recent data from the 1978 RPFS showed a mean of 24.4 years.With the exception of Sri Lanka and Singapore, the 1970 level is the highest reached in South and Southeast Asia at the corresponding period, while the 1975 level is slightly lower than Taiwan (23.3), South Korea (23.7) and Japan (24.3), at the corresponding period (See Smith, 1980). The timing of changes in nuptiality patterns can be described as follows. Of the total change in SMAM from 1903 to 1975, 61 percent occurred by 1960, another 22 percent occurred during the 1960s, and 17 percent occurred during the 1970-75 period. The accelerating trend in age at marriage is evident from the implied average annual increase during the various periods, i.e., 0.02 year during 1903-60, 0.05 year during 1960-70, and 0.08 year durin$ 1970-75. The data for 1978, based on the 1978 RPFS, suggest that the trend in marriage patterns is continuing. An interesting aspect of Philippine marriage pattern is the high proportion of women (around 6 percent) who remained single at age 50 or so, which has remained more or less constant over the seven decades. Similar proportions in South, Southeast and East Asia are much lower (See Smith, 1980). Cross-section data for 1973 and 1978 on socioeconomic differentials of age at marriage suggest that increasing age at marriage is associated with urban residence, higher education of the woman, non-agricultural employment of women and female work participation before marriage. A recent analysis of the sources of nuptiality change conducted by Smith (1978) concludes that: (1) the traditional dimensions of Filipino social structure - ethnolinguistic and religious - do not seem to have induced nuptiality differentials of any moment; (b) some of the most sizable nuptiality differentials in the past are areal differentials growing out of regional histories, especially with respect to age, sex and marital status selective migration; many of these

184

ALEJANDRON. HERRIN

differentials have diminished sharply and are likely to disappear in the future; and (c) in the more recent period, overall nuptiality patterns are related to the expanding role of females in the three important interrelated social processes of urbanization, the rise of mass education, and growth of the non-agricultural force. How much of the observed fertility decline can be attributed to changes in the marriage pattern? A standard decomposition applied by Concepcion (1980) to the crude birth rate change between 1960 and 1975 reveal that of the crude birth rate decline of 24 percent, only 16 percent was accounted for by nuptiality change, while marital fertility change accounted for 78 percent. Changes in the age structure accounted for the remaining 6 percent. In absolute terms the decline in the crude birth rate during the 1960-75 period attributable to nuptiality change was only 1.8 births per 1,000 compared to 8.7 births per 1,000 attributable to marital fertility change. On an annual basis, the average decline in crude birth rate during the 15-year period due to nuptiality was only 0.12 births per 1,000. It will take a drastic change in nuptiality patterns to produce a much larger impact on fertility than what has been observed during the past one and a half decades. In view of the already late age at marriage among Filipino women, such drastic change is not likely. Hence, future fertility declines must necessarily rely on significant changes in marital fertility through contraception, given society's views regarding abortion. Contraceptive

Prevalence and lts Correlates

On the basis of accumulating data on contraceptive prevalence, the following broad trends can be noted: from a rate of 16 percent in 1968, it rose to 24 percent in 1973, then to 37 percent in 1978. In 1980, contraceptive prevalence is estimated at 42 percent. What is noteworthy in the data is the increased use of the "less use-effective" methods (rhythm, condoms, withdrawal), relative to the "highly use-effective_' clinical methods (pills, IUDs and sterilizations). In the 1978 and 1980 estimates, only one-third of current users were using the more effective methods. By way of comparison, in Thailand, which had approximately the same level of contraceptive prevalence in 1975 as the Philippines had in 1978, 85 percent of current users were using the more effective methods (Knodel and Nibhon, 1978). The proximate determinants of contraceptive use include such

DEMOGRAPHICDEVELOPMENT

185

demand-related factors as family size preferences and attitudes toward contraception, on the one hand, and of supply-related factors such as knowledge of contraception and access to contraceptive services and supplies, on the other. These demand and supply factors are in turn determined by socioeconomic, cultural and environmental factors, and by the direct effects of family planning program efforts. The evidences on the trends in the proximate determinants of contraceptive use are described below. First, using data from the 1978 and 1980 Community Outreach Surveys (COS), Herrin and Pullum (1981 ) found that at least in areas where concentrated family planning program efforts are being implemented, there is evidence of a decline in family size preferences as measured by the percentage of currently married, non-pregnant and fecund women aged 15-49 of given parity who stated they wanted no more children. Furthermore, this decline in family size preferences is clearly associated with the increased use of contraception in general, and of modern contraception in particular, i.e., pills, IUDs and sterilization. The changing demand for children, as inferred from these data is understandable in terms of the changing structure of costs and benefits of children, at least as perceived by parents. Data on these are fragmentary. However, the value of children studies, notably those conducted by Bulatao (1975; 1978; 1979a; and 1979b) for the Philippines reveal that among the various value domains within which children are considered, the "instrumental-assistance value," which includes financial help expected from children, old age security, help with household chores, and caring for other children, are found to be highly salient as well as among the most central values. On the other hand, financial cost, though less salient than worries of child rearing, appeared more central and ranked first in importance among the disvalues. In addition, differential value patterns appear to be associated with fertility control. Parents who were categorized as "low parity limiters" in contrast to the "high parity nonlimiters" tend to (a) place less va_ue on "help in housework" and "financial, practical help" as advantages of children, (b) consider cost of education as among the disadvantages of children, (c) consider three children to be of some financial burden, and (d) think less of an only child as undesirable because of mortality risk. Both sets of parents, however, tend to equally value children for their help in old age (Bulatao, 1979a).

186

ALEJANDRON. HERRIN

Data limitation does not allow a mapping out of the incidence of "low parity limiter" parents by residence and socioeconomic characteristics. Nevertheless, one can reasonably infer that socioeconomic change associated with greater opportunities for education, alternative satisfactions and roles, and declining mortality affected the demand for children, and hence for effective contraception, leading to actual changed fertility. The basis for such inference is the accumulating results of multivariate analyses which find that critical levels of education, income, female labor force participation, and child survival significantly influence completed fertility (e.g., Encarnacion, 1973; 1975; Canlas and Encarnacion, 1977: Harman, 1970; Paqueo and Angeles, 1979; Paqueo and Fernandez, 1979). Secondly, although the measures used are not strictly comparable, data on approval of contraception suggest a favorable trend. The percentage of married women of reproductive age 15-44, (MWRA) who said they approve of doing "something to avoid getting pregnant too often or to plan the number of children they have" was 58 percent in 1968 and 63 percent in 1973 (Laing, 1979, p. 5). In the 1978 Community Outreach Surveys (COS), the question was reworded to refer specifically to "the use of modern family planning methods like pills, IUDs, or condom," while in the 1980 COS, the reference to condoms was changed to sterilization (Laing, 1979; Laing, 1981b, p. 81). The corresponding percentages for the respective periods were 67 and 72. In spite of the favorable trend, 26 percent of MWRA still disapproved of modern family planning methods in 1980, of which half said they disapproved strongly. Approval of modern family planning methods was clearly associated, among other things, with increasing educational attainment, urban residence, and exposure to communications about family planning, either through interpersonal sources, through mass media or through field workers. Thirdly, awareness of specific contraceptive methods has risen markedly since the population program began. By 1980, practically all (99 percent) of MWRA have heard of at least one method of contraception. Additionally, awareness levels for modern contraception (i.e., pills, IUDs and ligation) are above 90 percent (Laing, 1981b, Table 9, p. 12). Data from the 1978 RPFS reveal almost uniformly high levels of awareness by age of woman and by family size (NCSO, et al., 1979, p. 124). Finally, it is obvious that the availability of contraceptive

DEMOGRAPHICDEVELOPMENT

187

supplies has significantly improved with the program than without it. Nevertheless, there are problems of increasing access to contraceptive supplies and services due to cost and logistical considerations. Relative Impact of Socioeconomic Factors and Family Planning Efforts on Contraceptive Prevalence. How well do the above changes explain the observed increase in contraceptive prevalence? Laing (1981a) has recently conducted a multivariate analysis of the correlates of contraceptive prevalence in the outreach areas in 1980. This is a very important study for the Philippines because it attempts to disentangle the effects of "demand" or socioeconomic variables and the "supply" of family planning methods, precisely the kind of information that is needed to clarify the relative importance of general development and specific interventions on contraceptive prevalence, and by extension on fertility rates. His results suggest that while socioeconomic variables accounted for a large proportion of the variation in clinical contraceptive prevalence in the Barangay Supply Point (BSP) areas, program variables also accounted for a significant portion of the total variation. The percentage of variance explained by the selected variables was 43 percent, of which 63 percent was explained by socio-economic factors, and 37 percent by program factors. With respect to overall prevalence, the percentage of variance explained by the selected independent variables was 39 percent, of which 70 percent was explained by socioeconomic factors, and 30 percent by program (both clinic and outreach) factors. These results can readily be interpreted in the following light. The socioeconomic variables are expected to be related to such demand factors as family size preference and attitudes toward contraception. BSPs with high proportions of household heads having high levels of education, non-farm or urban occupations and high incomes are expected to be areas which have high proportions of households having smaller family size preferences and more favorable attitudes toward contraception. Mass media exposure would also tend to be higher among these areas, and hence, awareness and knowledge of contraception will tend on the average to be greater. Program variables, on the other hand, can be interpreted as supply-related factors which increase the flow of information and the accessiblity of couples to contraceptive supplies and services, and hence reduce the effective

188

ALEJANDRON. HERRIN

cOst of contraception, prevalence.

thereby

leading to increased

contraceptive

Problems and Prospects On the basis of the foregoing analysis, one may summarize the forces underlying the observed fertility trends. Fertility levels remained high from the early part of the century up to the 1950s. Whatever fertility decline was evident, it was due mostly to the changing marriage patterns, first associated with the effects of selective migration, and later associated with the effects of increased education, urbanization and female participation in the labor force. The fertility decline that became perceptible in the 1960s accelerated in the mid-1970s, and was due almost entirely to the decline in marital fertility. Fertility declined more rapidly among highly educated parents with high incomes and living in the urbanized areas of Metro Manila, Southern Tagalog and Central Luzon. This rapid decline in fertility is understandable in terms of the responses of parents to the changing perceived costs and benefits of children associated with urbanization, industrialization, educational advancement, etc., and to their locational advantage in terms of access to contraceptive information and supplies provided by both program and non-program outlets. Further fertility decline in these areas can be expected from the impact of development and family planning programs already in place through some diffusion process, perhaps along the Western-type demographic transition, But what are the prospects for fertility decline for the rest of the population? Are high levels of Western-type modernization a necessary condition for fertility decline? If so, then nothing needs to be done in terms of selective interventions. But can the extent and pace of broad-based socioeconomic development expected in the decades ahead sufficient to depress fertility of the magnitude necessary for replacement fertility levels be achieved by the year 2000 or thereabouts, the level implied by ourrent population projections upon which the present development plan (1983-1987) is based? Most likely not, Recent experience and thinking, however, provide some cause for optimism. While Western-type modernization may be a sufficient condition for sustained fertility decline, it need not be a necessary condition. In a recent reappraisal of theories of fertility decline, Freedman (1979) suggested hypotheses that have important policy significance for population planning in less developed countries, such

DEMOGRAPHIC DEVELOPMENT

189

as the Philippines. These hypotheses are: "(1) that subsets of objective development alterations, much smaller than those that characterized the West, can provide motivations for lower fertility today; and (2) that under modem conditions, ideas and aspirations for a different way of life transcending what is actually available are also important in motivating lower fertility" (Freedman, 1979, p. 65). The experience of several countries cited by Freedman, as well as some evidence in the Philippines (Herrin, 1979) suggest that the above hypotheses merit serious consideration by policy makers and development planners. They suggest that "the idea and practice of family planning can sweep an LDC population far more quickly than was previously imagined possible." What is clearly needed are more information on the demographic impact of current development interventions to provide the basis for selective development interventions in the future. That the family planning program accelerated the fertility decline since the mid-1970s is inferred from the increased use of effective contraception in areas where program efforts have been more intensive, thus demonstrating the strategic role of the family planning program in fertility reduction. Further, progress may be expected from the program already in place. However, there are concerns regarding the extent to which the program can expand its rural outreach. Logistics, organization, availability of supplies and cost figure prominently in these concerns (Lieberman and Herrin 1981). In addition, the rapid growth of population in the 1950s and 1960s is now being reflected in increased numbers of women of reproductive ages who are now exposed to child-bearing risks. The slackened tempo of urbanization in the postwar period as will be described later, means that a large proportion of these women are to be found in the less accessible rural areas than in the urban areas. Program efforts to provide contraceptive supplies and services will, therefore, be expected to be much greater in the 1980s than in the prior decade. A final concern is with regards the pattern of contraceptive use. Although contraceptive prevalence has reached as high as 42 percent in 1980, two-thirds of this is due to the use of less effective methods (i.e., condom, rhythm, and withdrawal). An issue arises as to whether the program needs to exert greater efforts to push the use of the "'more effective methods" (i.e. pill, IUD, sterilization) or to accept the prevailing use patterns and simply accommodate to the actual

190

ALEJANDRON. HERRIN

demand for various types of contraceptive methods. Pushing the "more effective methods" however entails greater costs and requires greater organizational capacity. If the current method mix is to be accepted, then efforts must be made to make these "less effective methods" more use-effective. Such efforts may require changes in the structure and programming efforts of the family planning program (e.g., more intensive instruction in the use of the more efficient rhythm methods). 5. PopulationMovement lnternal Migration That substantial population movements have occurred in the Philippines is evidenced by the fact that in 1960, 13 percent of the population of all ages (representing 3.4 million persons) were living in a region different from the one in which they were born, and in 1970, this lifetime migration figure was 13.2 percent or 4.8 million persons. During the period between 1960 and 1970, 5.0 million persons or 14 percent of the population of 1970 resided in municipalities different from those they had lived in 1960. Of these, 51 percent crossed regional boundaries, 14 percent crossed provincial boundaries within their region, while 34 percent crossed municipalities within their provinces. Information on lifetime interregional migration up to 1960 point to the predominance of long distance frontierward movements. The major streams include those from Ilocos and Central Luzon to Cagayan Valley, and from Ilocos, Western and Central Visayas to Mindanao. Relatively smaller in volume but significant nonetheless are the rural-to-urban streams, namely those from the Ilocos, Bicol, Western and Eastern Visayas to Metro Manila. During the period 1960 to 1970, while long distance flows continued to be a major component of the total migration pattern, movement to the metropolitan region has replaced migration to the frontiers. Thus out of 11 major streams (migration of 50,000 or more ioeople), 7 were to Metro Manila. The regional origins of these major streams are Ilocos, Central Luzon, Southern Tagalog, Bicol, Western Visayas, Eastern Visayas and Southern Mindanao. The other major streams included the counterstreams from Metro Manila to Central Luzon and Southern Tagalog, and the movement from Central Visayas to Northern Mindanao and Southern Mindanao. Between 1970 and

DEMOGRAPHIC DEVELOPMENT

191

1975, interregional migration followed more or less the same pattern observed in the 1960,1970 period, with the predominance of migration into Metro Manila. The changing pattern of interregional migration is further revealed in the sex composition of the migrants. The early frontierward movements were male-dominated, while the later metropolitanward movements tended to be female-dominated. The predominance of female migrants in the more recent period is suggested by data from 1970-75. Of the total interregional migrants (887,910), 53 percent were females; while of the total migrants to the metropolitan regign (263,058), 60 percent were females. Of the demographic processes, migration is perhaps more sensitive to changes in social and economic opportunities. The historical migration pattern is therefore understandable in terms of the opportunities that were available in particular periods of Philippine economic history. These opportunities first took the form of land availability in the frontier areas of Cagayan Valley and Mindanao. Migration into the frontier came largely from areas with a relatively poor agricultural resource base such as the Ilocos and the Visayas, and from areas with high agricultural densities and agrarian unrest such as Central Luzon. In the more recent period, economic and social opportunities took the form of urban jobs and education, thus the predominance of rural-to-urban migration especially to Metro Manila. The concentration of wage employment in urban areas, especially in Metro Manila can be traced to the effect of industrialization policies of the early postwar period and the relative neglect of subsistence agriculture in that same period. The rapid industrialization in the early postwar period, however, did not lead to significant labor absorption in modern sector jobs due to the increasing capital-intensity of production techniques being used. This in turn was a result of the capital-intensity bias in the structure of incentives provided by the early industrialization program. This limited employment absorptive capacity of the modern sector contributed to the recent outflow of temporary workers abroad, as described below. International

Migration

As a component of population growth, international migration is of minor significance compared to fertility and mortality. However,

192

ALEJANDRON. HERRIN

there is increasing concern regarding the recent international flow as this has involved large numbers of skilled and semi-skilled workers. By piecing together several studies of international migration, one can characterize the dominant patterns of Filipino emigration as follows. The first is the outflow of migrant workers in the early part of the century, especially to the United States to work in the plantations in Hawaii and California. This flow involved large single males coming mostly from the llocos region. Subsequent migration up to the 1950s have been small in magnitude involving among others the migration of relatives of both sexes after the war. The second major pattern is the heavy outflow of Filipinos from various regions of the Philippines and of all ages (migration of entire families) after 1965, significantly associated with the U.S. Immigration and Nationality Act of 1965 which abolished the old national origins quota system. A large part of these emigrants were professionals, technical and allied workers together with their families. The third major pattern of emigration involves the contract workers or temporary migrant workers whose destinations tended recently to be the Middle East, Europe and Asia. The growth of this temporary migration of workers has outstripped that of permanent migrants, with seamen accounting for around two-thirds of the number from 1972 to 1977. This recent migration is mainly a response to the great demand for skilled and semi-skilled manpower in the international labor market, as well as to the limited employment absorptive capacity of the modern sector in the country. Observes AbeUa (1977): "The overseas labor market has become a very significant absorber of Filipino manpower. The number of Filipinos who found jobs overseas during the period 1972-1976 represent about ten percent of the additions to the labor force during that period. The manufacturing sector managed to absorb only an additional 240,000 workers between February 1972 and August 1976, whereas the overseas labor market absorbed 169,321 workers during the same period from January 1972 to December 1976, excluding permanent emigrant workers." (p. 98). Urbanization:

A Note

A recent study suggests that compared to the historical experience of Western countries as well as the contemporary experience of Southeast and East Asian countries, Philippine urbanization has proceeded slowly (Pernia and Paderanga, 1980). The urban popula-

DEMOGRAPHIC DEVELOPMENT

193

tion grew from 13 percent in 1903 to 33 percent in 1975 while the tempo (urban-rural growth difference) has slowed down from 3.4 percent during the 1918-1939 period to 0.45 percent during the 1970-75 period. Philippine urbanization, however, has been marked by increasing primacy. In 1975, the population of Metro Manila, consisting of four chartered cities and 13 municipalities, has 5.8 times the population of the next three largest cities. This pattern of urbanization is understandable in terms of the rapid growth of the total and rural population during the postwar era and the differential development of regions arising from the spatial biases of public policies, especially those associated with the early narrowly-based import substitution policies. Problems and Prospects The broad overview of the historical patterns of Philippine population movements suggests several policy questions for which there is yet very little information. First, the emerging femaledominated urbanward migration suggests questions regarding the implication of this trend for the economic and social status of the female migrants. Evidence showing the most common urban occupations of migrant and urban-born men and women based on the 1973 National Demographic Survey data suggest a pattern of "extreme occupational differentiation" by migrant status and sex (Eviota and Smith, 1979). Native working women tend to occupy the high prestige occupations, while working female migrants occupy the service oecupations, mostly housekeepers, cooks, maids, etc. The segregation is sharper in the Metropolitan area among in-migrants from rural areas, and among recent migrants. This differentiation is even sharper when compared with males. Educated male migrants tend to occupy white collar and craftsmen occupations, while relatively educated women migrants are still predominantly in the service sectors. The picture improves among female migrants with some exposure to college or university - 75 percent of female working migrants are in white collar employment; however, these are mainly in clerical and sales categories, and still 20 percent with college education are in the service sectors. Secondly, the old issue of balanced regional development takes on an important new twist, namely, unbalanced urbanization. This refers to the slow pace of urbanization on one hand, but of increasing primacy on the other. This raises the old question

194

ALEJANDRON. HERRIN

regarding the implications of rapid urban growth in one or very few places for the adequate provision of social and economic services including employment and housing. In addition, it raises questions regarding implications for the development potentials of the outmigration regions. Thirdly, the old issue regarding the implication for Philippine development of the brain drain, i.e., of the emigration of skilled professionals to the advanced countries, acquires a new twist in the more recent temporary outflow of workers. Questions regarding its implication for the national economy, on the one hand, and on the adequacy of protection and welfare services of these workers in the area of destination require more systematic studies. One may expect increased population mobility as development proceeds. However, the type of development pursued with its implicit and explicit spatial biases, can be expected to have significant impact on where people will move. In spite of the rural development efforts in the 1970s, one can expect continued and heavy movement of the population in Metro Manila, its environs and in a few selected industrializing regions of the country in search for wage employment, educational opportunities and occupational mobility. This is so because with the vanishing of the frontiers, these areas will continue to be the centers of economic activities in the 1980s, and partly because the effects of the rural development thrust and of the program of industrial dispersal in the 1970s will not be strongly felt until a few decades in many regions of the country. The rapid population growth in the postwar period is now being translated into a rapidly growing labor force, as children born in the 1950s and 1960s now seek a more permanent place in the labor force. The continued limited labor absorption capacity of the modern industrial sector, and the increased availability of employment abroad are expected to increase the flow of temporary migrants so noticeable in the recent period. 6. Conclusion Demographic trends and development efforts will be more closely interrelated in the 1980s than in the past decades. The simple transfer of health and medical technology from the advanced countries can no longer be relied upon to effect significant declines in mortality as it did in the early postwar era without concomitant economic and social change in the rural areas and among the poor

DEMOGRAPHIC DEVELOPMENT

195

and without adaptive innovations in technology and organization in effectively addressing emerging patterns of demographically significant health problems in specific areas. Likewise, short of massive socioeconomic transformation, further fertility declines will now have to increasingly rely on strategic aspects of development that can lead to changes in the perception of parents regarding the costs and benefits of children on the one hand, and on the reduction of the effective cost of contraceptive methods through an energetic family planning program, on the other. Finally, with the disappearance of the agricultural frontier, population movements will now be increasingly tied to the differential industrial activity between regions. All these factors suggest a greater need for integrating population factors into development planning. Needless to say, great effort will be needed to operati0nalize such integration. Evaluation of current programs as well as pilot-testing of alternative schemes will be needed to determine the most cost-effective means of reducing fertility and mortality. Similarly, the spatial implications of development programs needs to be more explicitly taken into account in future planning. How all these can be done is the challenge for policy makers, planners and social science researchers in the years ahead.

REFERENCES Abella, M. I. (1977), "Current Trends in External Migration," Philippine Labor Review, 2 (3): 95-104. Bulatao, R. A. (1975), The Value of CMldren: A Cross National Study. Vol. II: Philippines. Honolulu, Hawaii: East-West Population Institute, EastWest Center. Bulatao, R, A. (1978), "The Value of a Filipino Child: Pleasure and Profit Against Cost and Concern," in Population o1' the Philippines, ESCAP Country Monograph Series No. 5, Bangkok: United Nations. Bulatao, R. A. (1979a), "On the Nature of the TransStion in the Value of Children," Papers of the East-West Population Institute, No. 60-A. Honolulu, Hawaii: East-West Center. Bulatao, R. A. (1979b), "Further Evidence of the Transition in the Value of Children," Paper of the F_ast-WestPopulation Institute, No. 60-B. Honolulu, Hawaii: East-West Center. Canlas, D. B. and L Encarnackm, Jr. (1977), "Income, Education, Fertility, and Employment: Philippines: 1973," Philippine Review of Bmffness and Economics, 14 (2): 1-28.

196

ALEJANDRO

N. HERRIN

Concepcion, M. B. and P. C. Smith. (1977), "The Demographic Situation in the Philippines: An Assessment in 1977," Papers of the East-West Population Institute, No. 44. Honolulu, Hawaii: East-West Center. Concepcion, M. B, (1980), "Changing Fertility in the Philippines: When, How, Why, " Forthcoming in Declining Ferility in Developing Countries, ed. W. Parker Mauldin. Encarnacion, J. Jr. (1973), "Family Income, Educational Level, Labor Force Participation and Fertility," The Philippine Economic Journal. 12 {1-2): 536-549. Encarnacion, J. Jr. (1975), "Family Income, Education, Labor Force Paxticipation and Fertih'ty," in A Demographic Path to Modernity, ed. W. Flieger and P. C. Smith, Quezon City: University of the Philippines Press, pp. 190200. Engracia, L. (1982), "Demographic Profile," in Population and Development in the Philippines, (forthcoming). Eviota, E, and P. C. Smith. (1979), "The Migration of Women in the Philippines," Paper presented for the Working Group on Women in Cities, Honolulu, East-West Population Institute. Flieger, W., Abenoja, M. K. and Lira, A. C. (1981), On the Road to Longetivity: 1970 National, Regional and Provincial Mortality Estimates for the Philippines, Cebu City: San Carlos Publications. Flieger, W., Koppin, B. and Lira, C. (1976), "Geographical Patterns of Internal Migration in the Philippines: 1960-1970," UNFPA-NCSO Population Research Project, Monograph No. 5, National Census and Statistics Office. Freedman, R. (1979), "'Theories of Fertility Decline: A Reappraisal," in Hauser, P. M. World Population and Development: Challenges and Prospects, ed. New York: Syracuse University Press, pp. 63-79. Gwatkin, D. R. (1980), "Indications of Change in Developing Country Mortality Trends: The End of an Era," Population and Development Review, 6 (4): 615-644. Gwatkin, D. R., Wilcox, J. R. and Wray, J. D. (1980), "Can Health and Nutrition Interventions Make a Difference," Monograph No. 13, Washington, D.C.: Overseas Development Council. Harman, A. J. (1970), "Fertility and Economic Behavior of Families in the Philippines," Santa Monica, California: The Rand Corporation. I-Ierrin, A. N. (1979), "Rural Electrification and Fertility Change in the Southern Philippines," Population and Development Review, 5: 61-86. Hen-in, A. N. (1981), "Population and Development in the Philippines: A Status Report," ASEAN-Australia Population Project: Population and Development Dynamics and the Man/Resource Balance. Herrin, A. N. and Pullum, T. W. (1981), "An Impact Assessment: Population Planning II," Report prepared for the Commission on Population, Republic of the Philippines and the United States Agency for International DeveloPment , Philippines. Jacinto, M. L. (1969), "Health and Medical Service in the Seventies," Philippine Population in the Seventies, Concepcion, M. B. ed., Manila: Community Publishers, Inc.

DEMOGRAPHIC

DEVELOPMENT

197

Knodel, J. and Nibhon, D. (1978), "Thar3and's Reproductive Revolution," International Family Planning Perspectives and Digest, 4: (2). Laing, J. E. (1979), "Currently Relevant Findings and Implications from Family Planning Program-Oriented Research at the U.P. Population Institute," University of the Philippines, Population Institute. (mimeo). Laing, J. E. and Alcantara, A. N. (1980), "Final Report of the 1976 National Acceptor Survey," University of the Philippines, Population Institute. Lainf, J. E. (1980), "Major Findings from the 1978 Community Outreach Survey," University of the Philippines, Population Institute. Laing, J. E. (1981a), "Effects of the philippine Family Planning Outreach Project on Contraceptive Prevalence: A Multivariate Analysis," Studies in Family Planning, 12: 367-380. Lalng, J. E. (1981b), "Family Planning Outreach in the Philippines: Final Report on the Community Outreach Surveys," University of the Philippines, Population Institute. Lieberman, S. S. and Herrin, A. N. (1981), "Population and Development in the Philippines: Observations on USAID's Country Development Strategy Statement," Report prepared for the United States Agency for International Development. National Census and Statistics Office, et al (1979), World Fertffity SurveyRepublic of the Philippines Fertility Survey, 1978 First Report, Manila. Paqueo, V. and Angeles, E. (1979), "An Analysis of Wives' Labor Force Participation in the Philippines and the Threshold Hypothesis," Journal of Philippine Development, 6 (1): 1-27. Paqueo, V. and Fernandez, J. (1979), "An Empirical Analysis of a Disequilibrium Model of Fetility Behavior and the Threshold Hypothesis: 1973 Pernla, E. M. and Paderanga, C. W. Jr. (1980), "Urbanization and Spatial Development in the Philippines: A Survey," Survey of Philippine Development Research, Manila: Philippine Institute, for Development Studies. Smith, P. C. (1975), "The New Nuptiality: Recent Evidence of Delayed Maxrlage and a Projection to 2000," Philippine Sociological Review, 23 (1-4): 7-20. Smith, P. C. (1978), "Trends and Differentials in Nuptlality," in Population of the Philippines, ESCAP Monograph Series No. 5, Bangkok: United Nations. Smith,

P. C. (1980), "Asian Marriage Patterns in Transition," Journal of Famffy History, Spring 1980. (Also East-West Population Institute, Reprint No. 118). Williamson, N. E. (1982), "An Attempt to Reduce Infant and Child Mortality in Bohol, Philippines," Studies in Family Planning, 13 (4): 106-117. Zablan, Z. C. (1977), "The Prospects of Mortafity in the Year 2000," Technical Report No. 2. 1. Population, Resources, Environment and the Philippine Future: A Final Report, VOl. III-2.

CULTURE AND CORPORATE PERFORMANCE IN THE PHILIPPINES: THE CHINESE PUZZLE G. L. Hicks and S. G. Redding*

1. Introduction The work of Harry Oshima is rare insofar as it has broken out of the disciplinary mould in which many scholars feel constrained to operate, and has brought into account factors from a range of disciplines. In particular, he has faced up to the inevitability of taking culture into account when explaining economic behaviour. This paper pays tribute to the quality of such a lead, by following it. His more recent interest has been in the effects of culture, religion and work ethic on labour force behaviour. This is not our direct theme, for we are more concerned with managerial behaviour and its results, but the main thrust of the analysis is in line with his thinking. Our interest is in the cultural make-up of the 250 largest Philippine companies and their performances as distributed culturally. More specifically, we have attempted to update and extend the pioneering study of Yoshihara (1971) which was the first to examine such questions empirically on a comprehensive scale. We may perhaps preview the paper by indicating that significant differences in the pattern of management of a firm's financial resources are evident between Chinese and non-Chinese in the commercial sector. To the best of our knowledge, such empirical support for differences of this kind is rare, and perhaps, so far, unique. Further, it has been gathered in a context where Chinese and *The authors would like to acknowledgethe invaluableassistanceprovided by Miss Elizabeth Ong in Manila, and Mr. Alfred Kam in Hong Kong, in gatheringand processingdata for the study. 199

200

G.L. HICKSAND S. G. REDDING

non-Chinese are operating in the same environment (except for the factor of anti-Chinese discrimination which makes the findings even stronger), and where the ceteris paribus assumption can be said to hold. Our contention will be that the results seriously undermine any assumption by economists that socio-cultural variables can be left out of account in explaining economic behaviour. This study is exploratory and may pose more questions than it answers. The data used are indicative rather than explanatory, and closer studies of business behaviour in the Philippines are needed in order to follow up some of the leads indicated. It is hoped that research of such a nature may be stimulated, as the general field of managerial behaviour among the Overseas Chinese is still only poorly researched. This applies also to indigenous management thoughout the Southeast Asian region, and the contrast with Japan is stark in this respect. The paper will proceed by considering briefly the still sporadic, and so far unfruitful, flirtation of economics and the sociological disciplines handling culture. We shall then present a brief resume of Yoshihara's fmdings, before presenting and analysing the results of a new study based on 1979 company performance data in both the manufacturing and commercial sectors of the Philippine economy. 2. Economics and Culture A recent review by Jamieson (1980) has pointed to the long-dominant perspective of industrialism in the field of economic development. It is not necessary in this context to present the arguments for and against the convergence hypothesis in detail except perhaps to remind ourselves of the central argument that a technological imperative supposedly acts upon organizations to rationalize and bureaucratize their behaviour. Evidence is accumulating that, in Asia at least, this is not the case. For instance, in the case of Japan, Hayashi (1978) has brought into account the proposition that organizations tend to retain their pseudo-gemeinschaft nature. For the Philippines, Dannhaeuser (1981) has demonstrated that the rationalization of distribution channels to be expected with development does not in fact take place. Instead, a more loosely organized channel develops and proves to be effective. An important contributory factor in both these instances is the set of local cultural values; in the first case, those so commonly described

CULTUREAND CORPORATEPERFORMANCE

201

for Japan as affecting organizational life, and in the second case, the 'Chinese way' of doing business, a factor which we shall return to in more detail. The difficulty which economists have had in embracing such factors as parts of their models, is explained partly by the necessary division of academic labour and the consequent drawing and defending of boundaries which tends to set one group against another. A further factor has been the possibly comforting feeling of apparent certainty (or at least greater certainty) inherent in the capacity to use quantitative methods. Meanwhile, the sociological arena has been plaqued with conflict and doubt. Jamieson's resume (1980, p. I0) is perhaps useful here. In discussing the problem of incorporating cultural causes, he notes: Economics, that disciplineprincipally concernedwith economic behaviour and performance,must have found itself particularlyreluctant to enter such conceptual confusion. By the 1930s the Anglo-Americantradition of economics had largely ceased to be interested in the effects of cultural and institutional factors on economic activity. The greatdebatesabout method in economics, of historical induction versus deduction, individualizingversus generalizing,descriptiveeconomics or an economics that searchesfor laws and patternsraged in the 1880s, particularlyin Germany,but were largely over by the end of the first quarterof the twentieth century. Although the institutionalist school, led by men like Veblen, still stressedthe importance of institutional factors in economic analysis, the majority of professional economists had turned their back on these factors and stressedthe benefits to be gainedby ignoring such non-economicvariables,at least for purposes of analysis. The successof this approachcomparedwith the performanceof some of the other social sciences has been notable. In particular, Keynes showed how useful an analysiscould be which basedits reasoningupon very small amounts of primarydata about humanexistence. The theoretical eli. ruination of such cultural factors allowed a rigorous def'mitionof purely economic concepts, and paved the way for the introduction of some very powerfulmathematical techniques. In this manner, the situation to which economics has been led, is now such that the leading textbooks in the field (e.g., Samuelson) pay only the slightest attention to socio-cultural variables. In some cases, a kind of puzzled acknowledgement of their importance is given but because of the puzzlement, they are consigned to the category of residuals or explanations of last resort. In some cases also, the ignoring of such elements when making predictions has

202 resulted upon.

G.L. HICKSAND S. G. REDDING in prognostications

which are embarrassing

to look back

In this context, empirical evidence of connections between socio-cultural variables and economic behaviour may prove useful. Although not presented as part of an overall model, it is anticipated that refinement of existing models may be stimulated. 3. Yoshihara's 1971 Study Yoshihara's work was carried out against a background question of whether the process of industrialization with its inherent rationalism and economic calculation was possible in certain cultural contexts. He studied the 1968 performance data of 254 Philippine manufacturing corporations, extracting them from the top 1000 companies and restricting his sample to companies with sales of over 5 million pesos. Out of these, 166 were domestically owned, and he proposed that 100 were under Filipino control and 66 under Chinese-Filipino control. Another category of foreign-owned companies contained 65 subsidiaries and 23 non-subsidiaries. Analysing the rate of return being achieved, he noted that the median rate of return was 21.5 percent for the subsidiary group, 15.3 percent for the non-subsidiary one, 7.5 percent for the Filipino group and 8.9 percent for the Chinese-Filipino corporations. Introducing an important caveat that the figures for Filipino and Chinese-Filipino companies might be fictitious, Yoshihara noted two contributing factors. First, "a corporation is not a purely economic institution whose sole aim is to maximize the rate of return, its main purpose is to increase the welfare of family members." It is thus necessary to allow for "unnecessary expenditure often lavished on family members occupying high positions in corporations." The second factor is the under reporting of earnings, for which there is high incentive. It is likely that the same elements are in force for the period of our own study, using 1979 data, and the same caveats will necessarily apply. The thrust of the Yoshihara study was industrialization in a developing country and, after examining issues relating to technology transfer, scale of operations, and the country's desire to remain a politically and culturally distinct entity, he concluded that success

CULTUREAND CORPORATEPERFORMANCE

203

did not lie in duplicating the industrial experience of developed countries. Our own concern is less with national economic poliey than with the processes occuring within the economy itself and we now turn to the question of relative performance in various culturally-defined sectors. In pursuit of an interest in the effect of socio-cultural variables on economic behaviour, the main focus will be on businessmen of Chinese extraction, compared to others. 4. The Research Design From the list of 1000 largest corporations provided in the Business Day Special Report (1980), the largest 259 were chosen, across all sectors. Of these, 140 were in manufacturing, 74 in the commercial sector, with smaller numbers in mining, services, utilities and agriculture. Information from the Securities and Exchange Commission provided details of share ownership and identity of directors. Parallel with these data, a range of company financial performance figures were extracted from the Business Day Special Report and allowed the use of the following: Paid-up capital Sales rank Sales volume Income Assets Equity Income to equity ratio Income to sales ratio Gross profit to sales ratio Number of days accounts receivable Inventory turnover rate Liquidity ratio Liabilities to assets ratio Liabilities to equity ratio Fixed assets to total assets ratio Sales growth index (based on 1972) Complete data were not available in every case, but in the data processing, companies where data were missing on a variable were excluded from the analysis.

204

G.L. HICKSAND S. G. REDDING 5. Results

Table 1 presents a breakdown of the major sectors of the economy by culture of ownership. The two most significant sectors, both in terms of numbers of companies and sales volume, are the manufacturing and commercial sectors, which between them contain 214 of the 259 largest companies. In the manufacturing sector, some bias is given to the data with the inclusion of two American oil companies whose combined sales were 9,974 million pesos. Yoshihara's data were for 254 companies all in manufacturing, as opposed to our own 140. It is clear that his research went down to a smaller size of company than our own, but in terms of proportions of cultural ownership, little appears to have changed. He reported Chinese-Filipino companies as opposed to Filipino in a ratio of 66 to 100, If we include Spanish with Filipino in our sample, the ratio is 47 to 67 (or 70 to 100). Although the samples are not exactly comparable, there is no noticeable shift in proportions during the ten year gap between the two studies. If anything, the Chinese may have increased their influenced slightly in this sector. The commercial sector data are again somewhat distorted by two American companies in the petroleum field, whose distribution activities account for 6,712 million pesos out of the American total of 7,619 million. The remainder of the commercial sector, excluding those two companies, has sales of 12,586 million. Chinese-owned companies account for 7306 million or 58 percent of this total. Examining the sectors more closely, Table 2 presents the breakdown by industrial subgroup for manufacturing, and by Chinese as opposed to non-Chinese ownership. This indicates a Chinese-owned monopoly in tobacco, a small predominance (in terms of numbers of companies) in textiles and garments, and strong positions in vegetable/animal oils, food and beverage, paper products and basic metals. For the commercial sector, Table 3 indicates particular Chinese dominance in machinery supply, general retail, and chemical based products. Attention was next paid to analysing the financial characteristic of Chinese versus non-Chinese companies to see if it were possible to distinguish certain performance patterns which would indicate in simple terms how the Chinese manage, or at least what financial strategies are normal to them, and perhaps not to others.

TaMe l - Sales Volume and Industry Sector by Culture of Ownership a The top 259 Philippine companies. All figures in million pesos

Industry Sector

No. of companies included

No. of cos.

Spanish salesb

No. of cos.

Filipino

Manufacturing

140

10,039

Commercial

302

Mining Services

13 t4

106

Utilities e Agriculture

13 5

1 -

259

Total

136

10,586

No. of cos.

Chinese c

10,289

8,950

I4,729

2,020

3,341

7,305

7,619

761

6 9

2,400 3,303

2 1

171 174

2 1

873 615

3 2

N.A. N.A.

7 3

875 367

3 1

407 96

2 1

225 72

112

20,545

17,106

No. of cos.

American d

24,135

Source: See Text aThe control of a corporation by a certain nationality is determined by the majority of capdtalsubscription. bNaturallzed Filipinos descended from Spanish stock are included here. e.g., the fandlies Zobel, Ayala, Araneta, Elizalde, Ozaeta, Aboitiz, Soriano. qThisatural_ed Filipinos who are of Chinese ancestry, and who still bear their Chinese names axe classified as Chinese. figure inc2uded two American companies with sales of 9,974. eExduding government controlled organizations in power, conmnmicaflons and air transport.

No. of cos.

Others t_

q_ t_ po O _e _" _Z m

to O

206

G.L. HICKS AND S. G. REDDING

Table 2 - Manufacturing Industries (out of top 259 companies) No. of companies, Chinese and non-Chinese, by industrial sector

Industrial sector

Food + beverage Textile + garments Petroleum products Vegetable + animal oils Chemicals Wood + wood products Paper + paper products Rubber + plastic products Basic metals Non-metallic minerals Tobacco Alcoholic drinks Transport + heavy equipment Electrical machinery appliances, supplies Others

No. of Chinese owned

% in sector

No. of other owned

% in sector

% of total in manufacturing sample

l0 10 -

45.5 52.6 -

12 9 3

54.5 47.4 100.0

16.4 14.2 2.2

3 2

50.0 10.5

3 17

50.0 89.5

4.5 14.2

1I. 1

88.9

6.7

42.9

57.1

5.2

2 3 1 1 -

25.0 42.9 10.0 100.0 -

6 4 9 3

75.0 57.1 90.0 100.0

6.0 5.2 7.5 0.7 2.2

20.0

80.0

3.7

3 -

25.0 -

9 3

75.0 100.0

9.0 2.2

Source:See text.

The significant differences which are evident in the manufacturing sector are given in Table 4. From this table, it is possible to conclude that the Chinese-owned firm tends to be smaller in sales volume than the non-Chinese. It also operates on a smaller profit margin, ending with a net profit of 3.1 percent of sales as opposed to 4.3 percent. Its liquidity ratio is 1.24, as opposed to 1.34 for non-Chinese, indicating a slightly more cautious approach to finance, although the higher ratio of liabilities to equity is contradictory to this, explained perhaps by a very small equity base. A search for the key factors which distinguish the Chinese from the non-Chinese firm, was carried out, using discriminant analysis

CULTURE AND CORPORATE PERFORMANCE

207

Table 3 - Commercial Industries (out of top 259 companies) No. of companies, Chinese and Non-Chinese, by industrial sector

No. of Chinese owned

Industrial sector

Principal commodities General merchandise Primary industrial products General retail Auto vehicles + supplies Petroleum products Chemical based products Real estate and construction supplies Agricultural and industrial machinery None + office appliances + supplies Others Source:

%in sector

No. of others

%in sector

No. of total in commercial sample

4 4

44.4 36.4

5 7

55.6 63.6

12.3 15.1

1 9 3 -

33.3 69.2 27.3 -

2 4 8 1

66.7 30.8 72.7 100.0

4.1 17.8 15.1 1.4

66,7

33.3

4.1

38.5

"61.5

17.8

75.0

25.0

5.5

-66,7

2 1

25.0 33.3

2.7 4.1

See text

and multiple regression. Results are presented in Table 5. From these, it is evident that two major elements distinguish the Chinese firm, one is the use of debt and the other is the satisfaction with lower margins. In this part of the economy, however, differences between Chinese and non-Chinese firms are not strong and it is possible that the technological imperatives inherent in the manufacturing process are causing companies to conform to certain norms, deviance from which becomes perhaps dangerously inefficient. This influence applies less in the commercial sector where operational discretion is doubtless higher, and this emerges in greater distinctions between Chinese and non-Chinese f'ums. Table 6 presents the key discriminating variables and Table 7 gives the results of the discriminant and canonical analyses and the multiple regression. The Chinese-owned firm in the commercial sector is clearly smaller

in terms

of assets,

and of equity.

Its income

to sales ratio is

208

G.L. HICKS AND S. G. REDDING

Table 4 -- Manufacturing Sector Comparison of Chinese and Non-Chinese Financial Performance (in million pesos)

Non-Chinese Financial characteristic

Chinese mean nffi40

Sales Income

mean nffi94

203.9 + equity ratio

387.3

Probability a

1.25

0.15

0.006

0.116

1.45

0.10

Income to sales ratio

0.031

0.043

1.17

0.15

Liquidity ratio

1,24

1.34

1.16

0.15

4.96

3.05

1.35

Total liabilities

to equity

0,I0

Source: See text aAlternative calculations with logarithmic data produced no significantly different results.

Table 5 -- Manufacturing

Discriminant

analysis

Sector

Raos V a

Canonical discriminant functions

Corr

1. Total liabilities to equity

2.56

1. Income to sales

0.62

2. Liquidity ratio

4.06

2. Liabilities

to equity

0.57

3. Total liabilities to total assets

7.67

3. Liquidity

ratio

0.56

Multiple regression on Chinese share percentage

R square

1. Liquidity ratio

0.03

2. Liabilities to assetsratio

0.04

3. Liabilities to equity ratio

0.07

4. Gross profit

0.08

to sales ratio

5. Sales Source: See text aAll statistics referred to ate derived from the SPSS package.

0.10

CULTURE AND CORPORATE PERFORMANCE

209

only a third to that of the non-Chinese, due in part to the lower grow margin of 9 percent as opposed to 14 percent. The most dramatically discriminating variable is days of accounts receivable, which is 27 for the Chinese and 54 for non-Chinese. This much tighter financial control is also reflected in the inventory turnover rate which is no less than four times greater than that of the non-Chinese sector. As in manufacturing, the liquidity ratio is lower. It is, then, predictable that the key discriminating variables should emerge as days of accounts receivable and inventory turnover. Table 6 - Commercial Sector

Financial clxaracteristic

Chinese mean n=33

Non-Chinese mean nffi40

Probability a

1.72

0.05

Assets

55.96

Equity Income to sales ratio

14.03 0.009

48.09 0.027

1.41 1.03

0.10 0.15

0.091

0.141

1.79

0.05

Gross profit to sales ratio

157.32

No. of days accounts receivable 27.0

54.8

2.90

0.005

Inventory turnover rate

56.68

14.16

1.58

0.10

1.11

1.25

1.63

0.05

Liquidity ratio

Source: See text aAlteraativecalculationwith logarithmicdataprod,__.ce_ _ no ,i_if',_ntly differentresults.

6. Discussion There

can

doubt

as to

the

relative

success

the

Chinese-owned t-ran in the Philippine economy. In the commercial sector particularly, they remain the dominant force, and they are well represented in other sectors. One need hardly add that rids has been achieved in the face of long-standing legislation deliberately designed to exclude them (McBeath 1973), the details of which are unnecessary to consider here. The question is how rids has been achieved, and this carries

210

G.L. HICKSAND S. G. REDDING

with it two ancillary questions, namely, why other groups have not succeeded so well, and what the implications are for the study of Philippine economics. We may attempt an outline answer to the first question, but will leave the other two for others to ponder. The Chinese way of doing business, as we have already noted, is not widely understood, and often the subject of a degree of myth. Amyot (1973), in his close study of the Chinese of Manila cited five factors as contributory to Chinese economic success: hard work and frugality, the ability to work together, the availability of credit, business acumen, and a pioneering spirit. Dannhaeuser (1981, p. 590), in his study of Philippine wholesaling, included in his description of "the Chinese way," the following resume: .... [there is] a preferencefor autonomouspersonalizedowner-operated businesswhich set their own informalterms and a tendency to give, other factors being equal, fellow Chinesea slight personal edge over ...... kinship (the family) providesone of the more effective principles available to entrepreneurs for organizing their enterprises. A similar point can be made here with respect to ethnic identity and relations among firms ..... In essence the Chinese combine two things. They maintain business autonomy on each market levelwithin the context of givingverticaltrade preferences to one another largely on ethnic grounds. Omohundro's study (1972) of Chinese merchant culture in Iloilo noted the synonymous nature of the business and the household for the Chinese. Constituting only 2 percent of the city population, they nevertheless pay 35 percent of its taxes. Describing membership of the firm as being in concentric circles, he notes that it is in the inner circle of exclusively family members "where all the money is held, all decisions made, all responsibility shouldered, and where the profit accumulates." The money is held closely, and expenses curtailed. Expansion depends as much on available family talent as on capital. Looking at the same question from an historical point of view, Wickberg (1965, p. 121 ) concluded significantly: The Chinese were able to expandtheir economic influencebecause of three factors: liberalizedSpanishpolicies, new opportunities offered by the new export crop economy, and Chinesebusinessmethods. Of the three, perhaps the last wasthe most decisive. Historical influences of a different kind are discussed by McBeath (1973), who, in considering the prohibitions of the 1935 Constitution, says "The impact of the American colonial and

CULTURE ANDCORPORATE PERFORMANCE

211

Commonwealth period was to further entrench basic Chinese economic interests. This was particularly the case with commercial transactions." He further notes that any further compromise of these economic interests has been prevented by taking citizenship, using Filipino 'fronts' ignoring the law, and the use of bribes to prevent prosecution. This last element, that of a somewhat besieged community, is perhaps more important than it first appears, and certainly adds justification for the high levels of co-operativeness noted for Chinese business. Such co-operativeness is decisive in the key area of finance, a description of which is provided by Weightman (1960, p. 147) The Chinesemerchantsbelongto associationswhichare 'borrowingpools' and price-fixingagencies.No firmcan expectto continuein businessif, by unilateralaction, its prices,which are often fxxedby the associationconcemed,fluctuatewildly,orif it goesbackon its word.Withrespectto loans fromcompatriots,it wouldappearthat practicallyany Chinese(or at least any Hokkien)who doesnot havea recordasa poorriskandwhohas a plan of establishinga businesscan easily obtain a loanby merelypresentinga planof his intentionsand prospectsto a Chinesebankora richChinese.It need hardlybe addedthat the chamberof commerce,the bank or the interestedassociations,bend everyeffortto insure the successof the undertakingandthe returnof the loan. An example of this working is provided in Dannhaeuser's description (I 981 ) of the textile industry. Here the importers/wholesalers invested in the mills and have continued to support them with large orders. Trade terms are normally nondocumented and extremely flexible, guaranteed entirely by personal trust. It is not abnormal for provincial dealers to receive high credit from Manila at low interest rates, with loosely defined payback dates, which Filipinos and others would find it next to impossible to obtain. Such a system is supported by information sharing and the organizational fabric of trade and mutual aid associations, partnerships and surname bonds. Such descriptions of the Chinese approach to business as it is practised in the Philippines accord with other descriptions of general Overseas Chinese business forms (Redding, 1980). It is, however, a long step from describing such practices to defining which ones are the most crucial to success. In any case, the answer to that will depend, at least in part, on the context being studied. Relating back to the data presented earlier, it is possible to argue

212

G.L. HICKSAND S. G. REDDING

that they provide evidence that the Chinese way of doing business is, in fact, more efficient, at least in terms of its use of capital. As this is generally taken to be the ultimate criterion for business success (Ansolf, 1966), it takes on more significance than other variables. In arguing this, it is important to take account of the problem of using profit figures, and thence yield, a problem which Yoshihara acknowledged. The under reporting of profit in .the Philippines is likely to distort this criterion. Instead, it should be more revealing to look at criteria which indicate the way the capital is used, namely accounts receivable and inventory turnover. On both of these counts, signs of the careful husbanding of financial resources, the Chinese businessman is distinctly ahead of his competitors. For the commercial sector, where they dominate, the data present a picture which accords well with the previous descriptions of Chinese managerial behaviour and it need hardly be added that such corroboration in empirical terms is, to our knowledge, unique in the literature. The assets of the Chinese firms are only a third the size of firms in non-Chinese ownership, thus supporting the notions of small size associated with family identity. As previously noted, the expansion of the Chinese firm depends more on managerial talent within the inside circle than it does on the normal rationalities of corporate growth or the normal constraints of capital raising. The question of growth in the Chinese firm is still insufficiently understood and much more research will be needed before the constraints and rationale are understood, but the fact of some size limitation remains a common observation. Weightman (1960) noted that 'there seems to be a limit to the ability of the Chinese to organize on a large commercial scale," and he cited in support, Purcell's earlier comment that: they seem to lack the breadth and boldness of conception that would enable them to enter upon large enterprises as rivalsof the Europeans, but between the Europeans and the natives they have an assured position. The same element is reflected in the value of equity which is distinctly smaller in the Chinese case than for non-Chinese. Here again, the retention of equity ownership within a small family group is likely to be restricting. The resort to wider ownership would be uncommon. The normal constraints which this would introduce on capital raising are of course counterbalanced by the communal fundraising systems described earlier.

CULTUREAND CORPORATEPERFORMANCE

213

The net income to sales ratio is thought to be insufficiently reliable to take seriously, but insofar as the same under reporting of profit may occur across both sectors, Chinese, and non-Chinese, the much smaller Chinese f_ure may be taken as prima facie evidence for their commonly described business thrift and patience, the ability to persevere on the basis of a steady accumulation of small returns. Possibly more reliable is the gross profit figure, which supports the same contention, and this time at a higher level of probability (p = 0.05) as an indicator of Chinese/non-Chinese difference. The number of days of accounts receivable, at 27 for Chinese and 54 for non-Chinese, is the strongest differentiating element (p = 0.005). In simple terms, credit is controlled stringently. This may be partly a reflection of organizing ability in the finance area, and a strong focus on finance itself, and partly a result of the trust networks described earlier, in which default or late payment becomes a cause of social and business ostracism. It is also clear evidence that the communal funding system described earlier is not a cushion into which the inefficient businessman can relax. As already noted "the interested associations bend every effort to insure the success of the undertaking and the return of the loan". Similar conclusions could be made from the difference in inventory turnover rate, with Chinese business turning stocks faster and thus using capital more efficiently. Here again, is prima facie evidence for a higher level of concentration on an important aspect of management control The lower liquidity ratio also supports the other findings and suggests a smaller proportion of idle cash in a Chinese business than a non-Chinese. Such data, while adding support to the earlier description of Chinese business behaviour, also begin to explain the source of the efficiency which has led a group comprising no more than 2 percent of the population to take control of 58 percent of the country's commerce. The stark reality of this comparative efficiency, and its basis in culture, must provide grounds for questioning the economists' normally universaiistic assumptions about the distribution of business skill in a community. There are, we suggest, important questions posed by these findings. The question for economics is whether it is prepared to adjust its models to take more respectful and serious account of sodo-cultural variables. The question for applied economics when it is being used in the making of policy is whether it should treat all competing nations as the same, or same more significant than others in

214

G.L.HICKSAND S.G.REDDING

business terms. The mainstream economists, what one might call the Power-Sieat-Ranis-Little-U.N. School have been pushing for a more open economy, the promotion of labour intensive manufactured exports, and a reduction in the protection afforded to high cost, import substituting industries. Examples from the super successes such as Korea and Taiwan are often given. All this makes fine economic sense given the economists' implicit assumption about "equality of potential entrepreneurial ability," but we would hold that such an assumption is far from proven. If it could be shown that the assumption is more likely to be false than true, then it may well be that, although the Philippines has been following policies which in some pure sense are second (or third) best, it follows that the consequences of taking the economists' advice could have resulted in an even less favourable outcome. The world trade in manufactured goods for the developing countries is dominated by the so-called Post-Confucian cultures, (Korea, Taiwan, Hong Kong and Singapore) and it is at least possible that only the tiny minority of Chinese-Filipinos could compete in this environment. The question of Filipino business performance, as opposed to Chinese, is the other side of the same coin, and we have proposed that others should consider it. Some tentative views have already been expressed (e.g., Redding, 1978) on sources of weakness, but much closer research attention is needed in this area. It has always been possible to gather support, from people making intuitive judgements, for the idea that Chinese business acumen was high. It has not until now been possible to know empirically just how such acumen is expressed in managerial behaviour. This study of the largest enterprises in the Philippines will hopefully have begun to fill in this large gap in the understanding of Asian management.

REFERENCES Amyot, J. (1973), The Manila Chinese: Familism in the Philippine Environment, Quezon city: Institute of Philippine Culture, Ateneo de Manila University. Ansolf, H. I. (1966), CorporateStrategy, Longon: Penquin. Business Day Special Report: The 1000 Largest Corporations (1980), Quezon City: EnterprisePublication.

CULTURE AND CORPORATE

PERFORMANCE

215

Dannhaeusen, N. (1981), "Evolution sad Devolution of Down-ward Channel Integration in the Philippines, "Economic Development and Cultural Change, pp. 577-595. Hayashi, K. (1978), "Corporate Planning Practices in Japanese Multinationals, "Academy of Management Journal, 21 (2): 211-226. Jamieson, I. (1980), Capitalism and Culture: A Comparaffve Analysis: of British and American Manufacturing Organizations, Farnborough: Gower Press. McBeath, G. A. (1973), Political Integraffon of the Philippine Chinese, California: Center for South and Southeast Asia Studies.

Berkeley,

Omohundto, J. T. (1972), Chinese Merchant Culture in lloilo City, unpublished Ph.D. Dissertation, University of MichiSan. Putchell, V, (1951), The Chinese in Southeast Asia, Oxford: Oxford University Press. Redding, S. G. (1978), "Beneath the Western Veneer," Asian Business and Industry, 14 (7). .... (1980), "Cognition as an Aspect of Culture and Its Relation to Management Processes: An Exploratory View of the Chinese Case," Journal of Management, 17 (2). Samuelson, P. (1980), Economics, New York: McGraw-Hill. Weightman, G. H. (1960), The Philippine Chineae: A Cultural of a Marginal Trading Community, unpublished Ph.D. dissertation, Cornell University. Wickberg, E. (1965), The Chinese in Philippine Life, New Haven: Yale University Press. Yoshihara, K. (1971), "A Study of Philippine Manufacturing Corporations," The Developing Economics, 9 (3): 268-289.

LONG RUN TRENDS IN THE GROWTH OF DEVELOPING COUNTRIES (Closing the Gap Between Developing and Industrial Countries) Helen Hughes*

During their peak development periods in the 19th and early 20th centuries, the industrial countries grew for the most part at only 1 to 2 percent per year. Their growth was barely positive in the 1920s, and it was negative in the 1930s. Growth rates for the rest of the world were even lower during this time. The high rates of growth achieved worldwide during the last 30 years (Table 1) have thus been as unprecedented as they were unexpected. I From a long run perspective, growth has even been high since 1974, particularly for developing countries. Developing and industrial countries represent a broad income spectrum ranging from the lowest income countries of Africa, South of the Sahara to the high income countries of Northern America, Western Europe and the Middle East. Indeed, the classification of countries into industrial and developing groups or according to any other criteria is necessarily arbitrary. "Levels of development" range *The World Bank. I am gratefulto my colleaguesin the Economic Analysis and ProjectionsDepartmentof the World Bank for many insights, to Husein Goksal for helping with the computations,and to Nancy Wallfor assistancein editing this paper. Theviews are,however,my own and shouldnot be attributed to the WorldBank, I. Economists substantially underestimatedthe growthpotential of developing countries duringthe 1950s. See D. Morawetz, Twenty-t_ve Yearsof Economic Development, 19.$0-75, The World Bank,Washington,D.C., 1977. The 1970's petroleum price increaseswere accompaniedby another spate of unreliable forecaststhat predictedthat it wouldbe impossibleto transformthe savings of the petroleum countries into productive investment elsewhere, and slow growthby petroleum importingcountries. 217

218

HELENHUGHES

from very poor, undeveloped countries such as Bhutan or Chad to very highly developed industrial countries. Income and development levels do not always coincide. The petroleum exporting countries of the Middle East have high per capita incomes, but in terms of development capacity they are still developing countries. Some very poor countries, notably India, have large and sophisticated industrial sectors so that they are in some sense "newly industrializing" or "semi-industrial" countries. Regional differences within countries cut across income and developmeht lines. The Sao Paulo region of Brazil is more developed than Southern Italy. In the spectrum of development generally, growing numbers of the semi-industrial, higher income countries such as Argentina and Singapore, which are still developing countries, have more in common with industrial than with very low income countries. For analytical purposes, it is, nevertheless, worth grouping countries by categories such as levels of development, income, endowment of natural resources, geographical region, or some combination of these. This paper postulates that developing countries as a group have been catching up with the industrial countries during the last 30 years, and that the "closing of the gap" between poor and rich countries has been accelerating. A new gap, between the slowly and rapidly growing developing countries, may be identified, but although the slowly growing countries are largely very poor countries, this is not altogether, so. Some very poor countries have grown rapidly, and some relatively high income countries have lagged behind. The essential factors in growth and catching up have been countries' domestic policies and the liberalization of the international economy. These are discussed in Sections 1 and 2, respectively. The paper concludes with a brief review of long run production trends. 1. Long Run Growth Trends Many factors have been identified at one time or another as principal causes of rapid growth for the last 30 years. These include the vast improvements in technology, the development of "human capital" through mass education, the acceleration in investment and capital accumulation, the structural shifts which took a significant proportion of the work force out of low productivity agriculture into

LONGRUN GROWTHTRENDS

219

higher productivity manufacturing and service activities, the movement of women from household to remunerated work, and also the various flows of temporary and permanent immigrants. The remarkable freeing of trade and capital flows and to some extent of migrant flows (discussed in Section 2) has clearly been critical. But the major force underlying these trends was the focus on social progress. The 1930's economic crisis had prompted changing social perceptions, and these led to changed approaches to policy-making. These were strengthened by World War II. The unemployment and social waste that had prevailed for centuries became unacceptable. Wartime economies had demonstrated that national economic management could be used in essentially market-oriented systems, and greatly improved policy formulation and administrative capacities were developed in the transition to peace time economies. During the 1950s and 1960s, most industrial countries were "catching up" with levels of productivity and living standards that the United States had achieved by the late 1940s. Their progress until the end of the 1960s is well known. Although the United States has also grown very rapidly by historical standards, some European countries overtook it during the burst of economic activity that culminated in the boom of 1972-73. Japan, itself only recently a developing country, became a leading industrial power and continued to grow relatively rapidly. For other industrial countries, including the United States, the period of easy growth was over by the late 1960s, so that new goals and policies were needed. Declining rates of productivity growth, rising rates of inflation and unemployment marked the transition, and the lack of appropriate policy responses led to "stagflation". When governments tried to tackle the 1971 downturn by traditional, mainly employment-oriented policies, the industrial economies on both sides of the Atlantic overheated. The petroleum price increases of 1973-74 greatly exacerbated the industrial countries' difficulties in the subsequent cyclical downturn, bringing the halcyon years to an end. For the rest of the 1970s, the industrial countries nevertheless grew at some 3 percent or faster than at any time before the 1950s. 2 2. The growth rates of industrial countries in the 1970s appear,moreover, to be understated in relation to the 1960s by the emergence of the "black" economy. The increasing emphasis on environmental improvements such as pollution controls similarly leads to understatement. Onceany initial expenditure

220

HELEN HUGHES

The socialist countries of Eastern Europe found it relatively easy to plan postwar reconstruction and to rebuild their economies on the basis of heavy industry after World Wai- II. In the 1950s, they may have grown even faster than the industrial market economies. By the 1960s, however, it became clear that central planning could not deliver living standards as high as those of market economies, and in the 1970s, the centrally planned economies' growth began to falter so that the official growth rates (Table 1) seem suspect. The failure to use prices that would realistically reflect scarcities, the absence of markets to establish such prices, the failure of most public enterprises to organize production efficiently and economically, and the sheer difficulties of detailed central planning resulted in most of the East European countries lagging severely behind the industrial countries in productivity and living standards, and even failing behind the more rapidly growing developing countries. Table 1 - Growth Trends in GNP and GNP Per Capita (Average Annual Percentage Growth Rates in 1980 US$ at Official Exchange Rate) GNP

Industrialmarket economies Developingmarket economiesa European centrally plannedeconomies Centralsurplus petroleum exporters

GNPPer Capita

1950-60

1960-70

1970-80

1950-60

1960-70

1970-80

4.2

5.2

3.2

3.0

4.1

2.4

5.0

6.1

5.1

2.7

3.5

2.7

n.a.

5.5

n.a.

4.6

n.a.

8.3

n.a.

4.2

aExeludingcapitalsurplus petroleum exporters, South Africaand China. Source: WorldBank data. The developing countries' internal pressures for "catching up" with the industrial countries were, in contrast, stronger than ideology, and they were endorsed internationally by the spread of social welfare ideals. Once poverty became unacceptable in a national is completed, the improvements yield a continuing recorded output in the GNP sense.

service without increasing

LONGRUN GROWTHTRENDS

221

context, inequality among nations also became a national concern, leading to substantial international support for developing countries' growth. But the developing countries represent 75 percent of the world's population and some 150 of its 180 or so countries and territories. Although they performed remarkably well overall and are beginning to close the gap between themselves and the industrial countries, their experience was very diverse. Regional per capita income differences indicate the wide variations among individual countries' growth trends. 3 Differences existed at the end of World War II. They increased considerably over the next thirty years because most of the low income countries grew only very slowly, while the higher income developing countries were, like some more developed countries, taking part in the "catching up" process. The process continued in the late 1970s. Growth was not, however, merely a function of the level of development. Some very poor countries grew so rapidly that they overtook higher income countries. Korea and the ivory Coast are notable examples. Differ3. The use of official ratherthan purcha_ng power exchangerates, of Las. peyres indices not corrected for quality and preference changes to measure prices, and of indirectly derivedPaasche indices to measure output, not only distortscomparisonsbetween levels of income among countriesat a giventime, but also understatestrue growth rates, particularlyof such countries as the Federal Republicof Germanyand Japan which caught up rapidlyin the 1950s and 1960s, and of the rapidlygrowingdevelopingcountries. See IrvingB. Kravis, Zoltan Kenessey, Alan Heston and Robert Summers,A System of International Comparisons of Gross Product and PurchasingPower, The World Bank, The Johns Hopkins University Press (Baltimoreand London), 1975 and IrvingB. Kravis,Alan Heston and Robert Summers,International Comparisonsof Real Product and Purchasing Power, The WorldBank,The JohnsHopkinsUniversity Press (Baltimore and London), 1978, for an indication of the distortions introduced by the use of official exchange ratesin international comparisons. Such distortions ate not removed by the use of domestic currencygrowth rates to estimate past growth (see Irving B. Kravis, Alan W. Heston and Robert Summers, "Real GDP Per Capitafor Morethan One HundredCountries," The Economic Journal, No. 88, June 1978, pp. 215-242) for these also under. estimate the growth of countries moving from simple, non-monetized to industrialeconomies. Ongoingwork by Kravis,Heston and Summers is seeking to identify the characterand magnitudeof such underestimationof growth. The catching up process which is underestimatedby growth rate measurements, whether in local or international currencies, is fully captured only when countries revalue their currenciesagainst those countries with which they ate catdting up.

222

HELEN HUGHES

ential income levels and growth rates grouped by regions (Tables 2 and 3).

emerge

when

countries

are

Table 2 - Developing Countries' Population and Per Capita Income by Region, 1980

Population

GNP

PerCapita Income $

Million

Percent

Billion$

Percent

1871

55.7

1474.8

59.8

790

152

4.5

466.6

18.9

3070

249 183

7.4 5.5

453.2 227.3

18.4 9.2

1820 1240

34 991 262

1.0 29.5 7.8

28.9 208.1 90.7

1.2 8.4 3.7

850 210 350

CapitalDeficitPetroleum Exporters

482

14.4

520.9

20.7

1060

CapitalSurplusPetroleum Exporters

0.8

199.5

8.1

7390

PetroleumImportersa SouthernEurope Latin Americaand the Caribbean East Asiaand Oceania MiddleEastandNorth Africa South Asia Sub-SaharanAfricaa

People's Republicof China Total

977

29.1

280.6

11.4

290

3357

100.0

2465.8

100.0

740

aExeludesSouth Africa. Source: WorldBankdata.

Capital Surplus Petroleum

Exporters

This group of countries, all small in population terms, combines very high per capita incomes with only modest levels of development. Some of these countries only achieved economic independence in the 1950s, and all grew rapidly in the 1960s when their exports of petroleum were expanding fast. During the 1970s, as their incomes rose, they sought to come to grips with their unique development problems. They had not yet acquired the administrative capacity to easily manage the high growth strategies which would be necessary if they were to enjoy continued high incomes when their petroleum and gas resources would be exhausted. They boosted their domestic absorptive

capacity

by massive

immigration

from Turkey,

Egypt

and

LONG RUN GROWTH TRENDS

22|

Table 3 - Developing Countries' GNP and Per Capita Growth by Re_ (Average Annual Percentese Growth Rates in 1980 US$ at Official Exchange Rate) GNP

PetroleumImportersa SouthernEurope Latin America& Caribbean East Asia& Oceania MiddleEast & North Afri_ SouthAsia Sub-SaharanAfricaa CapitalDeficitPetroleum exporters CapitalSurplusPetroleum exporters

GNPPerCapita 1960-70 1970-80

1960-70

1970-80

5.8 7.2 5.3 7.g 5.0 4.2 4.2

5.0 4.5 5.9 8.3 5.6 3.3 2.9

3.4 5.7 2.7 4.9 2.4 1.8 1.7

2.7 2.9 3.4 5.7 2.7 1.1 0.0

6.5

5.4

3.8

2.7

n.a.

8.3

n.a.

4.2

Yemen

Republic,

aExdudes South AfricaandChina Source: WorldBankdata.

small

Middle

Eastern

countries

Lebanon and Syria, and nevertheless had to invest

such

the

Pakistan, India and Bangladesh. heavily abroad in order to absorb

income productively. However, industrial countries limited their

restrictions opportunities

They their

on such investment in for productive invest-

ment, while inflation eroded the value of their capital and income. "Conservationist" petroleum production and export policies to match consumers' energy conservation followed. Kuwait, for example, can meet its current economic and social obligations from its investment income without exporting any petroleum at all. Some of these countries, led by Saudi Arabia, have nevertheless maintained high petroleum flows, notably during the Iran-lraq war, and thus moderated OPEC price increases. Countries with ample petroleum resources, and low domestic absorptive capacities, naturally are interested in maximizing long run gains from petroleum by keeping prices so low that substitutes are not developed prematurely though not so low that substitutes would not be developed at all. As a consequence of the latter energy shortages, economic and political crises might develop. Some of the capital surplus petroleum e_xporters were also interested in protecting their foreign investments

224

against economic feasible.

HELENHUGHES

and political crises, but unfortunately

this was not

Southern Europe Some of the countries in this grouping - notably Greece, Spain and Portugal - have at one time been relatively wealthy. However, industrialization passed them by, so that by the 1950s, they had become relatively low income countries, illustrating what happens to countries that do not adjust to the growth around them. In the 1960s, with improving economic policies, they started to catch up as developing countries, with per capita income coming close to those of lagging industrial countries such as Ireland, and even above these as in the case of Spain. In the 1970s, their catching up impetus floundered for several reasons. Their economies have traditionally been closely linked with those slowly growing countries of Western Europe. Greece has now joined the European Economic Communities and some others are expected to do so. The Southern European economies were similar to those of European industrial countries in economic and social organization, so that they too found it difficult to adjust to changing world economic conditions. Yugoslavia's particular policy problems arising from federalism and socialist management policies grew as living standards rose. The severe economic management problems which came to a head in Turkey were, however, akin to those of other middle income developing countries, with heavily protected inward oriented production, poor fiscal discipline, an inefficient public sector, and repressed financial sectors. Southern Europe is the only developing country region that has to date discovered very little petroleum or gas and has poor coal reserves. Petroleum imports had to be high, but exports were not competitive enough to respond adequately to the severe balance of payments pressure created by the rising petroluem prices. Moreover, previously high immigration to Western Europe was declining sharply, so that workers' remittances diminished as petroleum prices rose. Even in Southern Europe, however, a country that had appropriate policies could continue to grow rapidly. Malta, which managed to limit price rises in spite of its dependence on petroleum imports so that tourism and merchandise exports flourished, had an average annual GDP growth rate of some 10 percent during the 1970s.

LONG RUNGROWTHTRENDS

225

Latin America and the Caribbean Latin America has had a longer history of political independence than most developing countries. Argentina and Chile were, together with Australia and New Zealand, among the highest income countries in the world in 1900, and they had taken their first steps toward industrialization by the 1930s. Latin America is richly endowed with natural resources, including petroleum. In addition to the pf -ipal petroleum exporters (Venezuela, Mexico, Ecuador, and Peru), Argentina and Bolivia are substantiaUy self sufficient in petroleum, Chile and Colombia have some petroleum resources, and even Brazil has petroleum and coal. The islands of the Caribbean are well developed in terms of human resources, have had relatively high income levels, and are well located to exploit North American product and service markets and tourist potential. But despite a heavy emphasis on industrialization after World War Ii, overall economic growth in Latin America and the Caribbean has been relatively slow. The region was heavily influenced by inward-looking economic policies. It was generally believed that developing countries were unable to compete internationally because "the rules of the game" in the international economy were skewed against developing countries; that public enterprises were socially more desirable than private ones; and that industrial development was superior to agricultural growth. In spite of a great deal of commitment to socialism, there was little practical attention to poverty alleviation. The policies that reflected these views led to capital-intensive and high-cost industrialization at the expense of agricultural development, low employment growth and chronic balance of payments difficulties. There was little public finance discipline, with consequent monetary and exchange policy difficulties as inflation took the place of taxation. Large inflows of capital into economies with highly protected manufacturing, and restricted f'mancial sectors accentuated difficulties with resource mobilization and allocations. 4 In the Caribbean, social policies have moved ahead of productivity gains so that high wages make many of the islands uncompetitive. It was difficult for most countries in this 4. For a detailed examinationof this and related issues, see W.M.Corden, Chapter2 and Ronald I. McKinnon,Chapter 3 in J. Cody, H. Hughesand D. Wall eds., Policies for l_rl_trial Ptogre_ in Deve3opingCountries, WorldBank, OxfordUniversityPress,1981.

226

HELEN HUGHES

region to take advantage of the liberal international economic trends. Those that attempted to do so did it rather inefficiently by offsetting domestic protection by high export incentives. The petroleum rich countries, like the region's other mineral rich countries before them, experienced a bonanza in the 1970s, but found it difficult to take advantage of their high incomes because of the inefficient structure of their economies, particularly as living expectations rose ahead of petroleum receipts. Paradoxically, many of the petroleum importing countries restructured their economies to varying degrees as a result of the shock of the petroleum price increases. Combined with growing internal industrial competitiveness in the larger economies, Latin American countries emerged from the 1970s with less bias against agriculture and greater international competitiveness. Overall, then, the Latin American and Caribbean countries, petroleum importers as well as exporters, grew faster in the 1970s than in the 1960s. East and Southeast Asia The East Asian market economies have been the outstanding development performers: Korea, Taiwan and Hong Kong which, with Singapore, form the remarkable "gang of four" that have been the fastest growing developing economies in the world. Hong Kong and Singapore are city states with little alternative to export-oriented manufacturing development, but Taiwan and Korea have demonstrated the value of paying attention to all sectors and seeking a balance between domestic and externally oriented policies. Growth has led to declining fertility and this has translated high income growth into rapidly rising standards of living. The growth of the Southeast Asian countries has also been relatively rapid. They all have rich and varied resource bases so that petroleum exports are not dominant even in petroleum-exporting Indonesia and Malaysia. The Philippines is known to have petroleum and Thailand has gas. The Southeast Asian countries have also pursued relatively balanced economic policies, developing agriculture as well as industry, and exports as well as domestic production, and they now have well developed infrastructure, agriculture, manufacturing and modern service sectors. The East and Southeast Asian countries thus not only grew faster than almost all other economies in the 1950s, but they maintained their lead by increasing their growth rates in the 1970s.

LONGRUN GROWTHTRENDS

227

Middle East and North Africa Petroleum and phosphates have made contributions to growth in North Africa. The resource poor countries (and also Egypt) have benefited from mig,'ation opportunities within the region. But performance has otherwise been disappointing. Industrial growth has been slow, very capital intensive and by and large not competitive with either the old industrial countries or the newly industrializing countries of Latin America and East and Southeast Asia. Public ownership of manufacturing enterprises has been largely unsuccessful in dealing with the risks, uncertainties and changes inherent in manufacturing production. Despite some privileged access to European markets under special bilateral country arrangements, agricultural exports have been constrained by the European Community's Common Agricultural Policy, and such constraints seem likely to worsen with the Community's enlargement. Egypt's situation has improved markedly in recent yeats with the re-opening of the Suez Canal and the export of petroleum. With the exception of Israel, however, the countries of the region have not done justice to their natural endowment and potential. China China, the largest of the developing economies, is still an enigma, not only to foreign observers, but to its own analysts. Chinese data now available are so dubious that generalizations are almost impossible. It appears that China, like the European centrally planned economies, found the first stage of putting a war torn country on its feet relatively easy, but it has now begun to face the more difficult problems. Its technology is very outdated, its infrastructure, particularly transport, is very poor and it has practically no distribution or monetary system. Its physical planning and enterprise management are not suitable to its modernization objectives. It will have to become a major exporter in order to purchase modem technology, but its current raw material exports, including petroleum and coal, will be needed for rapid domestic industrial growth. The export emphasis will therefore have to be on the production of labor intensive manufactures. Hong Kong and Macao ate already serving as bases, but a substantial expansion is likely to take a decade or so. China is not likely to emerge fully into the international economy until the 1990s.

228

HELEN HUGHES

SouthAs_ The large poor countries of South Asia came to independence with some relatively well developed human resources, infrastructure and production facilities, but with very low overall productivity levels. Some regions, notably Bengal, are probably the poorest in the world. India's large size and heterogeneity made it difficult to create a meaningful political consensus for rapid growth. Bangladesh, like Pakistan, has had acute political problems from its inception, Development has therefore been very slow and very painful, though not without major progress in some sectors. The 1970s were a turning point in agriculture. Inadequate monsoons in the early 1970s led to very poor harvests and low growth rates, but the effects of changes in agricultural policy in India during the late 1960s were being felt. In the mid 1970s, India achieved self-sufficiency in food, and from 1974 to 1980, its growth averaged 5.6 percent despite fluctuating climatic conditions. Infrastructure development, particularly in transport and energy, however, has been slow. The manufacturing sector is large and well develeloped but production has grown at only 4.5 percent a year. High protection and intensive regulation of industry, combined with a great deal of public ownership, has resulted in inefficiency despite the opportunities for competition in the large domestic market. Although India also has a restricted financial sector, in contrast to Latin America, it has followed conservative fiscal and monetary policies, so that it has not been plagued by inflation. Pakistan's agricultural development has also been considerable, and even Bangladesh may now be approaching self-sufficiency. The smaller countries of South Asia also have difficulties, even if these are sometimes of their own making. In contrast to the countries that are catching up, Sri Lanka, for example, is falling behind. During the last 30 years, its per capita income has fallen from about 60 percent to 16 percent of Malaysia's. Sub-Saharan Africa The sub-Saharan countries are the poorest, least developed and most prone to political problems. These underlie many of their economic failures. They lack human resources, physical and social infrastructure, and productive capacity. Productivity is almost universally low. The region is well endowed with agricultural resources and with petroleum and other minerals, but the develop-

LONGRUN GROWTHTRENDS

229

merit of these has presented problems which inexperienced governments have found very difficult to handle. Where agricultural growth takes the form of enlarging cattle herds, it often leads to desertification. The countries tied to the French franc are quite severely handicapped by the overvaluation of their currencies in export markets. Their incomes are, of course, overstated. The Lome Agreement by no means fully offsets the protectionist effects of the European Community's Common Agricultural Policy. The beneficial effects of the General Scheme of Preference and the Lome Agreement are frequently severely limited, particularly by "rules of origin" and similar regulations. As in Latin America the resource rich couhtries have been among the most troubled. Some, such as Zaire and Zambia, have experienced problems in absorbing their resource rents. The region's export performance has accordingly been dismal and balance of payments problems are endemic. Petroleurrr, imports are not large, but represent a high proportion of total imports, and of export income because the latter is negligible. A handful of countries - the Ivory Coast, Malawi, Kenya, and recently Botswana, Lesotho and Mauritius - have nevertheless been able to achieve high rates of growth. Their experience is worth studying. This is undoubtedly the region with the most difficult development problems. An Overview The developing countries' growth processes are as varied as their cultural, political and economic evolution. While some have found it relatively easy to achieve the political cohesion, human and physical resource accumulation necessary to "take-off" on the development path, others are still far away from rapid progress. All developing countries faced the same international economic stimuli in the 1950s and 1960s, and constraints in the 1970s, but those with well managed economies were able to take advantage of international trade, capital, and labor movement opportunities, whereas those that had inappropriate domestic policies could not. In the 1970s, for various reasons, the developing countries as a group grew more rapidly than the industrial countries. They have a strong growth motivation to improve living standards. They are better endowed with petroleum resources than industrial countries, even when the capital surplus petroleum exporters are excluded from consideration. Those petroleum importing countries which were

230

HELENHUGHES

growing strongly were competitive enough to be able to take advantage of the petroleum exporters' rapidly expanding markets which became a source of further growth. They were able to increase their previously low penetration of industrial country markets. The newly industrializing developing countries provided growing markets for other developing countries, particularly for primary products. With strong export performance, many developing countries were thus able to take advantage of high international liquidity to supplement their savings. The labor market opportunities that opened up in the petroleum exporting countries helped several low income countries to break balance of payments bottlenecks. The economic growth process has been accompanied by greatly improved living standards. The population explosion of the 1950s and 1960s followed a rise in nutrition and health standards with consequent declines in mortality rates. Where economic growth has been rapid, fertility rate declines appropriate to a "demographic transition" have followed. Total developing country population growth is now declining, although total population will continue to grow, adding some 2 billion people to the developing countries' population during the next 20 years. Life expectation has risen and infant and child mortality have declined. Primary education participation and literacy have become widespread in all but the poorest and most isolated countries or regions. Access to higher education has also grown, sometimes ahead of the demand for it. Other social indicators - calorie intake, quality of housing, access to potable water, sanitary facilities and so on - vary considerably from country to country that it is not possible to make any meaningful international generalizations. It is clear that millions of people perhaps half of the population - are still living in extreme poverty in the low income countries of sub-Saharan Africa and South Asia, and that there are also pockets of very great poverty in some of the higher income countries. Northeast Brazil, the Altiplano areas of other Southern Latin American countries, and Central America are examples. However, wherever there has been rapid and sustained growth, poverty has generally diminished sharply.5 5. Data on income distribution (usually household survey based) are as poor as those on physical indicators of standards of living and of little use for temporalinternational comparisons.A combination of physicalmadincome dis-

LONGRUN GROWTHTRENDS

231

Examination of the experience of a variety of countries shows that some apparent advantages have not led to radical improvements. Conversely, some characteristics thought to handicap the growth process are shown to be of lesser importance. Being small, "landlocked" or "sealocked", or lacking natural resources all limit development options, but are not serious barriers to growth. Indeed, having rich natural resources can present problems, particularly if complementary human resources are inadequate. Unrealistic expectations generated by the existence of mineral wealth, for instance, may encourage fiscal policies which ultimately inhibit growth. Many richly endowed countries have not grown consistently or rapidly. Large land areas can be handicaps in spite of the gains from economies of scale that major domestic markets provide. India, Pakistan, Bangladesh, Indonesia and Nigeria have had problems because of their large and heterogeneous political bases. In contrast, in the liberal trading environment of the last 30 years, a small domestic market does not necessarily preclude rapid growth. A good location can help, but whereas some small Mediterranean countries have taken advantage of their proximity to Europe, most of those in Central America and the Caribbean have failed to take advantage of the United States' market. Some island economies such as Malta and Mauritius are growing strongly, others are not. Being landlocked is a handicap if surrounding countries are poor. However, the world's highest per capita income country, Switzerland, is landlocked. Malawi and Paraguay have done relatively well in comparison to other countries in their regions despite being landlocked. Large injections of aid in per capita terms, e.g., in Talwan and Korea, and access to international capital markets, particularly in combination, have helped many countries grow rapidly, but others have failed to achieve sustained growth despite large aid and commercial capital inflows. Breaking through balance of payments constraints is critical, but the claims made for "outward oriented" industrialization strategies and export growth have sometimes exaggerated the importance of exports in development. Some countries have broken the balance of payments constraint mainly as a result of tribution data suggeststhat poverty has been reducedin terms of absolute and relative numbers in most developingcountries, the main exception being in countries which have experiencedwar, and that standards of rivinghave grown most in countriesthat have grownrapidly.It is impossibleto makeadjustments about the impact of growthon income distributionover time or among countries.

232

HELENHUGHES

emigration leading to significant receipts of remittances. The rapidly growing countries have indeed focused on exports, but they have also introduced social infrastructure policies to overcome human resource constraints, invested in appropriate physical infrastructures, broken rural production bottlenecks, and followed industrialization paths that lead to internationally competitive costs of production. Their common characteristic is a policy balance. 6 The specific components of such a policy balance vary according to the circumstances of the country, but the objective is simultaneous development of agriculture and industry, of production for export and for the domestic market, and of social and physical infrastructures. An important element of the policy balance has been the maintenance of open economies, especially for the smaller countries. International competition has helped develop competitive export lines but it has also ensured that production for the domestic market is efficient and competitive. Countries which develop industrial production predominantly for a protected domestic market, have let vested interests grow up, which seek to perpetuate their positions, resist competition, change and adjustment. In a liberalized trading environment, competitive pressures force continuous adjustment and lead to the development of a flexible and dynamic economy, a major benefit which has generally been found to exceed the costs arising from fluctuations in the international economy. Monetary, fiscal manpower and other macro-economic policies are also important components of policy, determining price stability and hence the climate for saving and investment, the utilization of factors of production, and the extent to which special sectoral policies are needed to stimulate growth. Generally, the more balanced and successful the macro-economic policies are, the less need there is to intervene sectoraUy by production licensing, subsidies and similar policies. That is, bureaucratic interventions can be limited to important issues and they can thus be more appropriately targeted and executed. The development of social and physical infrastructure in the framework of an open economy contributes to development of 6. For a more detailed account of the developing countries' experience in the 1970s. see FrancisX. Colaco, "The International Economic Environment and the Developing Countries, 1960-90," Word Bank, EPDIT Division Working Paper, 1980/14.

LONGRUN GROWTHTRENDS

233

competitive production and of human resources, and in itself is a stimulus growth. Reasonably equitable social policies not only lead to the accumulation of human capital but also hasten the transition from the short life expectancy/high fertility pattern to the slower population increase which improves the growth of per capita income. A few countries, including Singapore and Malaysia, have been successful in accelerating "trickle down" distribution effects by careful attention to access to public goods. The majority, however, paid little attention to welfare. Some, like Sri Lanka, have traded off growth against equity. Some countries achieved neither growth nor equity. In all the rapidly growing countries, governments played an important rc_lein investment in the social and physical infrastructure, and in following policies that made it possible for markets to function. Hong Kong and Singapore, for example, often regarded as laissez falre economies, had very considerable public works, education, health and housing programmes which, together with fiscal, monetary and labour policies, stimulated private sector growth. Indicative planning has played an important role in the development of Korea, Singapore and Taiwan. Growth-oriented policies are hard to implement successfully without both a degree of political cohesion and some administrative capacity. The latter in turn requires the development of skills and human capital. Political cohesion demands consensus on the objectives of economic development and the means of achieving them. The rapidly growing countries have in fact avoided the shibboleths of the right as well as of the left. A narrow dedication to ideology has had a poor growth record. This does not mean, however, that the objective of efficiency in production is incompatible with equity in distribution, unless of course production decisions are overriden by distributional considerations. There is, evidently, no one path to rapid economic growth. Countries must be selective and choose the one that suits their particular economic, social and political conditions and levels of development. If they do so, they can complement their domestic resources by taking advantage of the liberal international economy that has come into existence in the last 30 years to accelerate their growth. If they do not, even minor international economic fluctuations could exacerbate their domestic problems.

234

HELENHUGHES

2. International Linkages The liberalization of the flows of trade and capital after World War II, and a return to the high volumes of migration reminiscent of the nineteenth century, contributed to a rapid spread of technology, and gave countries an opportunity to exploit economies of scale and specialization beyond their own borders, thus enabling the developing economies to avoid balance of payments constraints and augment their savings through trade and borrowing. In trade, the principal liberalization impetus came through the General Agreement on Tariffs and Trade (GATT) and the series of multilateral tariff and other protection reductions negotiated under its auspices. The post World War II reductions in capital movement controls culminating in the end of fixed exchange rates played a parallel role in capital markets. The Movement of Labor The industrial countries of "new settlement" were traditionally countries of immigration, and still have relatively low density of settlement. The United States and Canada remain the most popular countries of immigration, attracting migrants from the developing countries, particularly from Central and Latin America. The end of the colonial era and associated political settlements also led to some migration to other industrial countries, principally France and the United Kingdom. With full employment in industrial countries in the 1960s, a new type of short-term immigration began to flow to the rapidly growing countries of northwestern continerttal Europe. Germany and Switzerland were the principal host countries for these "guest workers". Flows of workers among developing economies were also increasing, mainly from slowly to rapidly growing countries. Migration was growing in Africa and Latin America in the 1960s, and the rapid economic growth of the low population petroleum exporting countries of the Middle East led to a new wave of migration from neighboring states and from South Asia. The United States legally admits some 400,000 people annually but estimates of illegal immigrants, mainly from Latin America, who now reside in the United States, range from 2 to 6 and more million people. There were about 6 million temporary migrants in Europe in the mid 1960s, with as many dependents. Estimates suggest that about 2 million people have migrated in Africa (including to South

LONGRUN GROWTHTRENDS

235

At'flea) and 3 million in Latin America. Immigration into the Middle East involves some 2 to 4 million. To the individuals concerned, the benefits of such migration evidently exceeded the costs. There have always been more applicants than places for short- and long-term migration, and the pressure for illegal migration has been very strong. In the case of the United States, it has been successful. Although migrants have, for the most parts, come into the work force at the lowest prevailing wages and for the least attractive jobs, both wages and jobs were more attractive than those in their home country. The European host countries also pay substantial social security benefits to the migrants, either together with their wages or as a lump sum on their return home. While living conditions for temporary workers are usually inferior to those enjoyed by the local population, they are rarely as bad as, and are often much better than those in the home country. The workers accumulate considerable savings and they usually return to their home ountry with new skills and sufficient capital to improve their living standards markedly. For permanent migrants, the industrial economies offer an unequalled opportunity to catch up with the industrialized countries' living standards, usually within a generation. Workers' remittances have been important in breaking the balance of payments constraints in the developing countries bordering the Mediterranean, some Caribbean and Latin American countries, and Middle Eastern and South Asian countries. In 1977, workers' remittances and net private transfers contributed some $20 billion to the foreign earnings of the developing country. Rapid growth in remittances to the countries bordering the Mediterranean took place during the 1960s. More recently, there has been spectacular growth in remittances to the non-oil developing countries of the Middle East. For Yemen People's Democratic Republic, Yemen Arab Republic and Jordan, remittances in 1979 were by far the largest source of foreign exchange earnings. Table 4 shows other countries where remittances were a major source of foreign exchange (i.e., between 10 percent and 100 percent of merchandise export earnings). They assume particular importance in countries such as India, Bangladesh and Egypt which have not been able to formulate and implement appropriate policies to stimulate exports. 7 7. Gurui_ri Swamy, "International Migrant Workers' Remittances," World Bank Staff WorkingPaper, No. 481, August 1981.

236

HELEN HUGHES

Table 4 -- Magnitude of Remittance Inflows into Major Labor-Exporting Countries, 1978-79 RemittanceInflow $ million

Remittanveinflow as %of merchandise exports

Yugoslavia Greece Turkey Spain Cyprus Portugal Morocco Tunisia

2938 989 1012 1752 69 1689 763 220

43 30 77 13 18 69 51 25

MiddleEast

Egypt Sudan

1762 69

89 12

As/a

India Pakistan Bangladesh

1022 1303 115

15 77 21

Aft/ca

UpperVolta Marl Benin(1977)

66 31 24

60 33 17

Europeand North Africa

Source: G_ushi Swamy, "InternationalM_ant Workers'Remittances: IssuesandI_ospects,"WorldBankStaff WorkingPape_No. 481, August1981.

All the effects of migration are not positive. It is argued that emigration draws away the relatively skilled and enterprising workers, relatively few of whom return to inject skills, enterprise and capital back into the economy. In the country of destination, immigration delays teehnalogical change, investment, and increases in productivity. If true, migration thus contributed to the competitiveness of developing country exports in the 1960s and 1970s by avoiding the substitution of capital for labor through new technology. There are, moreover, social, and political costs in countries of emigration and immigration, and these, particularly in countries of immigration, tend to become paramount. Immigration into Europe was slowed down in the early 1970s by the host countries for social reasons and similar trends may now be observed in the Middle East. Thus, although there seems litre doubt that migration is a very efficient - perhaps the most efficient - means of raising living standards for poor people, it is likely to continue to be

LONG RUN GROWTH TRENDS

237

limited, and probably will grow relatively slowly in the future. Trade, capital and associated technology flows will have to continue to be the core of international economic relations. Trends in Trade The

industrial

two-thirds

market

economies

of world trade throughout

growth of trade generally rapid nomies took up centrally planned

still

accounted

the 1960s

for

and 1970s,

almost and the

among them was the dominant component in the trade growth of the period. The developing econearly a third, leaving a very small share for the economies. 8 (See Table 5).

Table 5 - Country Composition of World Exportsu 1960, 1970 and 1978 (Percen0

Industrial market economies European centrally planned economies

1960

1970

1978

Developing market economies (of which capital surplus petroleum exporting countries)

(4)

(2)

(5)

People's Republic of China Total

b 100

1 100

US$ billion

150

1,594

aMetchand_e trade and nero-factorsen'ices. Non-factorservicedata for the centrally plannedeconomieshad to be estimatedfrom verypartialinformation. bless thanI pezeent Source:WorldBankdata. The commodity composition of trade has changed sharply. Table 6 on merchandise exports indicates that an increase in the share of manufactures in the 1960s was followed in the 1970s by an increase in the share of fuels. The growth rate of global exports peaked in the late early 1970s with growth rates of 7.5 percent for 1960-70

1960s and and about

8. For further detail, see A. Schwartz, E. Lutz, I. Jacobsen, and A. Lee, 'World Trade Flows by Origin and Destination, 1970-90: A Functional Analysis," World Bank, EPDIT Division Working Paper No. 1980/8

238

HELEN HUGHES

6.5 percent in the 1970-79. The capital surplus petroleum exporting developing economies' exports increased rapidly in the 1960s when the volume of petroleum consumed and traded internationally increased sharply. Thus, they maintained their share in petroleum price increases in the 1970s. The other developing economies' share of total trade, however, declined, largely reflecting the decline in their share of primary product exports. Supply constraints rather than fluctuating or low commodity prices were thus commodity export problems for many developing countries. Countries like Colombia, Ivory Coast and Malawi did well in spite of fluctuating prices and were able to expand output of primary products, or move into processing activities. Table 6 - Composition and Growth of World Merchandise Exports by Commodity Groups, 1965, 1970 and 1978 (Percent)

Agricultural products

1965

1970

4.5

7.4

Metals and minerals

1978

Real Average Annual gIowth rates 1965-78

Fuels

4.0

Manufactures a

8.9

Total

100

6.9

US$ billion

190

310

1,120

astandardIndustrial TradeClassification5 to 9 minus68. Note that this understates the growth in trade in manufacturesbecauseprocessedrawmaterialswhichareincreasingly t_adedareincludedin SITC categories0-5. Source: WorldBankdata. Latin American

countries

in particular

lost a substantial

their agricultural temperate climate exports to the industrial economies in the 1950s and 1960s. This was largely due excessive protection of manufactures. The introduction

share of market to their of the

European Common Agricultural Policy and agricultural protectionism in the other industrial market economies, however, was also a

LONGRUN GROWTHTRENDS

239

factor in restricting the markets for a number of products, notably sugar and beef. Latin A_erica also lost some of its share of mineral exports to other developing countries. Exports of petroleum and manufactures did not grow rapidly enough to make up for the export losses in agricultural products and minerals. Thus Latin America's share of world merchandise exports feU from 12 percent in 1950 to 8 percent in 1960 and to about 5.5 percent in the 1970s. The neglect of trade during the 1950s and the 1960s was part of the policy mix that resulted in most Latin American countries' growth lagging substantiaUy behind that of other countries at similar and lower levels of development. Except for petroleum, Middle Eastern and African countries also had relatively poor export performance during the 1950s and 1960s. The South Asian countries' shares of world exports also fell during the period. The East and Southeast Asian countries, in contrast, increased their share of trade rapidly from decade to decade. Although the most rapid expansion of trade from the 1950s to the 1970s was among industrial countries, trade between developing and industrial countries also grew rapidly. At the end of the 1970s, the industrial countries accounted for some 75 percent of developing country markets, and the developing countries represented about 25 percent of industrial country markets. Intradeveloping country trade, which had lagged in the 1950s and 1960s behind north-south trade growth, began to grow more rapidly in the 1970s with the growing demand in petroleum exporting countries and the relatively strong performance of the more rapidly growing developing countries. The newly industrializing developing countries provided a market for raw materials, and about a third of their exports of manufactures went to other developing countries. 9 (See Table 7). The developing market economies' share of world exports of manufactures rose from about 10 percent to 20 percent between 1960 and 1977, growing at about 15 percent a year between 1960 and 1973, and at more than 12 percent per year even in the second half of the 1970s. Exports of manufactures began on a small scale in the 1950s when Israel and Hong Kong were the only significant exporters. Talwan followed in the early 1960s and then Korea, 9. O. I-lavrylyshynand M. Wolf, "Trade Among DevelopingCountries Theory, Policy Issues, and PrincipalTrends," WorldBank Staff WorkingPaper No. 479, August1981.

240

HELEN HUGHES

Table 7 -- Regional Structure of Merchandise Trade Flows, 1970 and 1979 Industrial Market Economies

Developing Market Economies

Centrally Planned Economies

72 68

24 27

4 5

100 100

67 68

25 27

6 4

100 1O0

24 23

9 13

65 63

100 100

70 67

24 27

5 6

100 100

World

Industrial Market Economies 1970 1979 Developing Market EconomieÂ˘ _ 1970 1979 Centrally Planned Economies b 1970 1979 World 1970 1979

alncludescapitalsurpluspetroleumexportingcountries. blncludesPeople'sRepublicof China. Source: UnitedNationstradedata.

Singapore

and the

Southern

European

countries

in the mid 1960s.

Colombia was the f'trst of the Latin American countries to attempt to facilitate non-traditional exports by a change in exchange rate policies combined with high export incentives. Brazil followed. By the early 1970s, enthusiasm for exports of manufactures had become the new conventional wisdom. Although Hong Kong, Talwan, Korea and Singapore continued to dominate the developing countries' exports of manufactures and even increased their share in the 1970s, more than 30 developing countries were exporting over $100 million of manufactures annually by the late 1970s. While some relatively advanced industrial countries such as India continue to lag in exports of manufactures, many newcomers are emerging. Products exported have also widely diversified. Traditional products such as processed food, textiles, footwear and toys still dominate, but there is considerable expansion in primary product

LONGRUN GROWTHTRENDS

241

processing, basic materials such as chemicals and steel, mechanical components and machinery of various types, and in consumer goods such as china, pottery and glassware. Industrial countries found it more difficult to adjust to imports from developing countries than to increased mutual trade with them. The developing countries faced protectionist measures against their exports almost from the start despite the industrial countries' trade liberalization movement of the 1950s and 1960s. In fact, there is a continuous history of protectionism against developing countries from the time the first "developing" country, Japan, came on the market with a large volume of low cost products in the 1920s. It is salutary to recall that the trade diversion policies pursued by the United Kingdom, France and the United States in their home and colonial markets in the 1920s were precursors to the protectionism of the 1930s, and contributed to Japan's involvement in World War II. Protectionist measures against Japan were continued as soon as it came back into world markets in the 1950s, and were extended to include the rapidly expanding East Asian exporters by the 1960s. Industrial countries' difficulties in adjusting to developing country exports of manufactures reflect, in part, the particular characteristics of those exports. Developing countries tend to come in at the bottom end of the market with very low cost products, reflecting their per capita income and correspondingly low wage scale. Once production and marketing problems are overcome by the developing country or transnational corporation entrepreneurs, â&#x20AC;˘ exports can be increased rapidly because of ample labor supply. For the same reason, it takes time for costs of production to rise, although the growth process eventually does increase per capita incomes and force countries to move into other, less labor intensive products. But this may take a long time, particularly in relatively large countries where pockets of poverty may remain for long periods. In any case, other developing countries whose incomes are still low can come into the export market, again with low costs and prices. Also, in contrast to trade among the industrial countries, the initial impact of developing country exports tends to be on entire industries such as clothing, textiles and footwear, rather than on individual ftrms which might anyway diversify successfully. Further, these industries tend to employ the least skilled workers, often women who are denied entry into skilled trades and, as secondary income earners, have low geographic mobility. Whereas adjustment is

242

HELEN HUGHES

relatively easy within industry groups such as metal working, glassware, or pottery, which produce a range of products, it is difficult when a whole, large industry employing marginal workers is affected. Adjustment is also easier within a transnational corporation's organization. Japan has been particularly successful in utilizing its large corporations for the purpose of adjustment. The corporations can bear the costs covering as they do a range of activities. But when a relatively large industry is being phased out, the least efficient and financially weak firms employing the least skilled and enterprising workers are usually the last ones left producing the traditional product. If such firms are geographically concentrated, adjustment becomes particularly difficult because whole regions begin to decline.

In the 1950s and 1960s, when the industrial countries' growth was rapid, the adjustment process worked quite well, even in clothing, textiles and footwear where the labor force characteristics that make adjustment difficult were concentrated. The developing countries' expansion went almost unnoticed in some industries, with the industrial economies' production in those industries being almost phased out. Some articles of sports equipment in the USA are an example. The difficulties of adjustment, however, became greater as growth in the industrial countries slowed. Developing country exports of manufactures still have a relatively low share in industrial country markets (Table 8) averaging less than 3.5 percent of apparent consumption of manufactured goods. The share of industrial production is about the same (3.3 percent). Market penetration grew by about 8 percent annually from 1970 to 1979 (7 percent from 1976 to 1979). The displacement impacts of this growth, together with the impact of developing country exports to third markets, were much smaller than those resulting from increasing trade among the industrial countries, or from changes in technology, consumer tastes, petroleum price increases and the recession. 10 But trade with developing countries is politically easy to 10. For further detail, see the summary of a study of developing country penetration of industrial country markets in Helen Hughesand Jean Waelbroeck, "Trade and Protection in the 1970s: Can the Growth of Developing Country Exports Continue into the 1980s?," The WorldEconomy, Vol. 4, No. 2, June 1981.

Table 8 - Share of Imports in the Apparent Consumption of Manufactured in Industrial Countries, 1970-1979.

Share in Apparent Consumption Developing Country Imports

All Imports

of:

1970

Growth of Import Shares Developing Country Imports

All Imports

Goods

Developing Country Imports

All Imports

1979

1970-79

Australia Canada

20.8 26.1

2.1 1.2

24.3 36.3

5.3 2.3

2.8 3.0

12.0 6.6

EEC

18.9

2.7

29.6

4.7

5.3

6.4

Belgium France

61.7 12.1

5.6 2.1

70.7 16.0

4.6 2.9

1.2 3.3

-0.5 5.0

Germany

19.3

2.3

30.3

4.5

5.2

8.4

Italy Netherlands United Kingdom

15.1 41.2 16.3

2.1 4.9 3.3

29.3 53.6 27.2

4.8 9.0 5.5

7.3 3.0 6.3

9.3 7.2 4.1

Japan Sweden United Total

tates

4.5

1.3

5.7

2.3

1.8

5.5

31.0

2.8

38.7

4.1

2.3

3.5

5.4 10.6

1.2 1.7

9.6 16.8

3.0 3.4

6.6 5.1

11.3 8.1

Source: Helen Hughes and Jean Waelbroeck, "Trade and Protection in the 1970's: Can the Growth of Developing Country Exports Continue in the 1980's?" The World Economy, June 1981.

to UO

244

HELENHUGHES

isolate and so it has often had to bear the brunt of the industrial countries' willingness to adjust. Such unwillingness is of course shortsighted. The long run experiences of countries such as Greece, Spain, Portugal and Argentina, as well as the more recent experiences of countries such as the United Kingdom, indicate that the longer adjustment is put off, the costlier it becomes. Trade in non-factor services (tourism, contracting, shipping, insurance, etc.) is becoming an increasingly important component of world trade accounting for more than 20 percent of world exports (Table 9). Developing countries have been successful in service exports, notably in tourism and contracting services. 11 Table9 - Non-FactorServices:Sharesin Total Exports,1970, 1978 (Percent) 1970

1978

IndustrialMarket Economies

DevelopingCountriesa World

25 20

22 18

alneludescapitalsurpluspetroleumexportingcountriesbut excludesChina. Source:WorldBankdata.

Debt as an Engine of Growth At the end of World War II, the former colonial and neocolonial powers and other industrial countries acknowledged some responsibility for helping the developing countries extricate themselves from extreme poverty, whether for reasons of conscience, humanity or political interest. They began to asist the newly emerging countries through grants, loans on concessional terms and technical assistance. Aid flows grew rapidly in the 1950s (in part as colonial grants were transformed into aid) reaching some $10 billion or 0.4 percent of the industrial countries' GNP in the early 1960s. Thereafter, they grew 11. A Sapir and E. Lutz, "Trade in Non-Factor Services: Past Trends and Current Issues," Word Bank Staff Working Paper No. 410, August 1980, p. 35 and A. Sapir and E. Lutz, "Trade in Services: Economic Determinants and Development Related issues," World Bank Staff WorkingPaper No. 480, August 1981.

LONGRUNGROWTHTRENDS

245

more slowly, about 4 percent a year, and while they reached about 1 percent of GNP in Sweden and the Netherlands, for industrial countries as a whole, flows dropped to about 0.35 percent of GNP. The centrally planned economies' contribution to aid has been negligible, but the capital-surplus OPEC countries began to make contributions in the early 1970s, and these became substantial after 1975, reaching more than 3.5 percent of their GNP. The aid impetus flagged in the 1970s, but there were some improvements in its quality. Less aid was given in kind (food aid, etc.). There was a move from loans to grants, particularly for low income countries, and to "forgiveness" of interest and amortization on old loans. The developing countries' successful development effort as well as humanistic concerns led to a somewhat greater concentration of aid in the lower income countries. The tying of aid, which reduced its value to the recipients, however, was continued by most donor countries. Private capital flows to developing countries were very low, consisting largely of short term trade credits. It was widely thought that developing countries were not creditworthy for longer term capital, but as capital flows were freed from controls in the industrial countries, private long term capital began to flow quite early to the higher income and rapidly growing developing countries. Private direct investment, that is, in equity form, was the first major form of the flow. Many transnational corporations had links with the former colonial countries, mainly in the exploitation of natural resources, transport, power and other public utilities. But some of these corporations had been closely associated politically with .the foreign rulers, and almost all had appropriated high rents from their exploitation of mineral and other natural resources, and monopoly gains from public utilities. The developing economies wished to appropriate these rents and other gains for themselves, and the transnationals' investment moved to manufacturing and allied banking, insurance and advertizing activities that were beginning to flourish with high protection and other incentives. Some countries such as Korea, followed Japan in industrializing almost entirely without foreign investment. In others, principally those in Latin America, Southeast Asia and Africa, foreign investment became an important component of industrialization although usually it accounted for only a small proportion, usually less than 25 percent, of total investment in manufacturing, in all except sub:

246

HELEN HUGHES

Saharan African countries. Thus, despite disinvestment in minerals, net direct investment flows continued to grow at some 3 to 4 percent, accounting for the bulk of private flows and about a fifth of total flows to developing countries in the 1960s (Table 10). Table 10 -- Composition of Net Capital Flows to Developing Countries, 1960-62 and 1976-78 (Percent) 1960-1962 Official development assistance

34 (20) (7) (7)

56 (14) (12) (29)

100

Othernon-concessional, mainly official Private non-concessional Direct investment Export credits Financial flows Total US5 billion US$ billion in 19785

1976-1979

Note: The figures for 1976-78 coverflows fromOPECand the centrally planned economies as well as from the DAC and India, keland, israel, Luxembourg,Spain, and Yugoslavia. For OPEC,the 3-yeaxtotals werean estimated $13.6 billionin net bilateral and multilateral aid and an estimated $4.4 billion in net non-coneessionalflows,together representing about 9 percent of the entire net flow. For the centrally planned economies,the f_ures were $2.5 billionand $3.2 billion, respectively,or about 3 peIcent of the total. Comparableinformation on OPECand CPEflows is not availablefor the 1960-62. Source: OECD,Development Cooperation,(Paris) passim.

In the 1960s, export credits also began to be available to the more rapidly growing countries. Some were part of aid flows, but others were from private sources, though sometimes granted in conjunction with aid funds for such projects as hydro-electric installations. Some were associated with direct investment, and some were obtained independently. Export credits gave developing countries access to additional funds, but often at a cost. In essence, such credits were tied to given suppliers whose technology, while efficient for capital rich economies, was often costly for labor intensive incentives

ones. Government interest by the industrial countries,

subsidies, were often

used as export only available for

LONGRUNGROWTHTRENDS

247

such products. Developing countries having become exporters of capital goods themselves have had to match such subsidies. Moreover, by the mid 1960s, export credits were being pushed out to the developing countries in such volume that they led to severe repayment problems and "debt crises" in a number of countries. Toward the end of the 1960s, when both the exporting and receiving governments became more experienced in the use of export credits, these problems were slowly overcome. Their use has been revived as the industrial countries began to experience relatively low growth. However, export credits still involve hidden costs for developing countries. Those using them to import face the additional costs of being tied to the suppliers offering the credit, and those who must compete by giving credit as exporters are forced, in effect, to provide subsidies to their clients. In the late 1960s, the faster growing developing countries began to have access to the then rapidly expanding international capital markets. The loosening of capital flows among industrial countries, the excess of savings over investment in some of these countries, and the particular circumstances that led to the creation of the Eurocurrency markets, came just as the growing developing countries were ready and able to borrow directly from financial institutions. Long established commercial banks and other financial institutions in the industrial countries provided the bulk of the intermediation services required, but developing country financial centres also began to spring up in Places such as Singapore. There were considerable shifts from direct investment, to avoid difficulties over equity shares, and export credits, to avoid tying, to purely financial flows. Then in the 1970s, the increase in petroleum prices led to accelerated growth of the international capital market. Many developing countries and the European centrally planned economies took advantage of greatly increased international liquidity to cushion their adjustment to the petroleum price increases. The international capital market was thus able, again quite unexpectedly, to "recycle" the high savings of petroleum exporters and other countries by moving them to countries that, being in the early stages of the catching up process, had low capital stocks and high marginal productivity of capital. The rate of growth of total net private flows to developing countries was, however, about the same: 7 percent a year in the 1970s as it had been in the 1960s. With the move toward positive, and even high interest rates at the end of the 1970s as inflation expectations caught

248

HELENHUGHES

up with reality, new trends are emerging with a move back into export credits and direct investment. It is ironical that the complaints of the developing economies' lack of access to capital markets, prevalent in the 1950s and 1960s, should by the mid 1970s have turned into a fear that they had accumulated too much debt! Since 1975, it has been argued that the developing countries have borrowed so much that they, or a large number of them, will go bankrupt, one setting off others in domino fashion. It is further claimed that this will shake the world economy and bring on another financial collapse like that of 1929, a view particularly favored at the end of 1979. The banks that have lent to the developing countries are said to be "overexposed" in developing countries so that they too will go bankrupt, in domino fashion, contributing to the collapse of the entire international financial structure. It is argued that even if an international crisis of this dimension does not occur, disaster is nevertheless around the corner because the "developing economies" and "the banks" are now so vulnerable that the international capital markets will stop functioning or, put more crudely, "will stop recycling the petrodollars" and industrial country savings. It is by definition true that, having acquired access to capital markets, the developing countries have acquired debt liabilities. The question that should be asked is whether these debts are excessive in relation to the productivity in social as well as financial terms of the projects they have financed, and the national income growth of the countries concerned. This is the "solvency" equivalent of a private enterprise's borrowing. It has, of course, been greatly eased by inflation which transferred income from lenders such as the petroleum rich countries to borrowers. A second legitimate question relates to the management of the debt and its servicing, and related balance of payments issues. This is a "liquidity" issue. The developing countries again had these problems eased because the initial reaction to accelerated inflation was to allow interest rates and other costs of borrowing to lag behind inflation. The gradual shift to positive interest rates at the end of the 1970s led to a build up of interest rate obligations for the early 1980s, but these will stabilize, or decline if inflation drops. The liquidity of the private market has made for smooth refinancing facilities in contrast to the difficulties of rescheduling debt from official sources, so that, for strong borrowers with well established

LONGRUN GROWTHTRENDS

249

relationships with lending institutions, the maturity terms of debts have not been of very great concern. The growth of workers' remittances and the continuing growth of exports has meant that debt service obligations could be met. Thus, although the developing countries' debt has been increasing - they owed nearly $400 billion at the end of 1979 - for developing countries as a group, there is no "debt problem". The ratios of debt to exports and to GNP increased in the 1970s, but they are still relatively low. Debt to reserve ratios have been falling as heavy borrowers have increased their reserve levels for debt management purposes. Debt service ratios have risen, but this in part reflects the shift from direct to indirect financing. Capital service ratios that include direct inv_tment service have hardly risen at all. Banks are interested not in getting their money back, but in eaminginterest on it, and refinancing opportunities therefore continue to be favorable. It is the interest service ratio tha.t is most relevant to the developing countries' situation, and this is still very low (Table 11). Borrowing is highly concentrated in a relatively small number of middle income countries, with 31 countries accounting for 80 percent of total debt and 90 percent of the debt service. Among these, Brazil ($52 billion), Mexico ($34 billion), Spain ($20 billion) and Korea ($15 billion) are the largest borrowers, accounting for almost a third of all debt. With the exception of Korea, they are also among the largest recipients of private investment. Other countries with debt of more than $10 billion include Algeria, Egypt, Indonesia and Venezuela. These are all petroleum exporters. India, with a debt of $16 billion, almost all on concessional terms, is the only large low income borrower. Countries such as Zaire, Jamaica and Turkey, which have run into severe debt servicing problems, are not large borrowers. Their debt problems are really signs of general economic management problems which lead to low returns on investment and balance of payments difficulties. Most of the large borrowers have gained access to international capital markets because of their domestic growth and their export performance, and their management of reserves and other aspects of the balance of payments. In favorable circumstances, even relatively high debt service ratios of 30 percent or more are manageable. It should be noted that countries such as Canada and Australia sustained higher rates of borrowing in the late 19th and early 20th centuries. Debt, indeed, has also stimulated growth

250

HELEN HUGHES

Table 11 - Developing Countrya Debt and Debt Service Indicators, b 1970-1979 (Percent) Indicators

1970

1974

1977

1978

1979

Debt/GNP

Debt/Exports

100

108

321

186

270

271

220

Debt/Reserves Debt Service Ratio c

Interest Service Ratio

Capital Service Ratio d

110

250

314

366

Memo Item Total Debt (U.S. billion)

aIndudes alldeveloping countriesexcept (i) the capitalsurpluspetroleum exporters;and (ii) countries for which tellable time seriesdata are not available (Afghanistan,Bahrain,Botswana, Burundi, Comoros, Guinea, Iran, Iraq,Lebanon, Lesotho, Liberia,Maldives,Nepal, Papua New Guinea and South Africa). bTotal developing country debt to total GNP, exports and reservesof all developing countries. CRedemption and interest payments as percentages of merchandiseexports, non-factor servicereceiptsand factor services(including emi_ant remittancesand other privatetransfers)earnings, dThe debt serviceratioplus profits and dividends(includingprofitson directinvestment)in the numerator. Source: World Bankdata.

indirectly.

Countries

that

have large debt must continue

to grow and

to export. They have to avoid the costs of highly protectionist policies. 12 The banks operating in the international sphere have also been sound on the whole. Most refinancin_ arrangements have been made on mutually beneficial terms, and losses have been negligible. The developing countries accounted for about half of all international borrowing in the late 1970s, but their borrowing represented only a small proportion, certainly less than 10 percent of total global capital flows, if the internal industrial market economies' capital flows are 12. N. Hope, "Developments in and Prospects for the External Debt of Developing Countries," World Bank Staff Working Paper No. 483; September 1981.

LONGRUN GROWTHTRENDS

251

included. Excessive government guaranteeing has led to some marginal lending, for example, in sub-Saharan Africa. Some uneconomic lending has also been evident where the existence of strong political forces increased the likelihood of support for countries in trouble. But such borrowing has been marginal to the system as a whole. Particular countries and banks may be expected to run into trouble in the future, as they have in the past. Capital market forms will undoubtedly continue to evolve. Nevertheless, the overall situation is healthy and favorable to developing country growth. 3. The Long Term Outlook Economists are congenitally myopic, always fighting yesterday's battles. This is the reason why the notion persists that the gap between industrial and developing countries continues to widen, primarily because of the international environment's hostility to developing countries. The use of official rather than purchasing parity exchange rates also helps to perpetuate the myth, Translating the 1960 to 1980 growth trends from official exchange rate into purchasing power parity, and extending the trend toward the year 2000, indicates in fact that the "gap" has stopped growing. For some developing countries, narrowing of the gap started in the 1950s and 1960s. The 1970s' continuing acceleration of many developing countries' growth was the turning point for many others. This does not, of course, mean that income disparities between rich and poor countries are diminishing rapidly enough, or that serious development problems do not remain for developing countries. The case for direct assistance to very poor countries on a liberal and significant scale remains as strong as ever. Given the role of the international markets in facilitating the growth of countries wanting to take advantage of international trade, labor and capital flows, .the case for maintaining and further liberalizing the international economy remains even stronger. Overall, the outlook for developing countries seems more optimistic than it was 30, 20 and even l0 years ago (Table 12). Although the developing countries' per capita incomes would still be substantially below those of the industrial countries even in the 1990s, a considerable number of developing countries would be "catching up", and some would be overtaking the slowly growing industrial countries. In a reversal of the situation existing as recently

252

HELEN HUGHES

1960,

produced

than

in developing

half

the

countries

world's

by the

output

year

Table 12 - Shares of World Population (Percent)

Centrally planned industrial economies

Market industrial economies

is likely

2000. and GNP

Capital surplus petroleum exporting economies

Developing economies

Population

1960

1978

2000

GNPat

official

exchange

rates

1960

1978

2000

GNPat

purch_ing

power parity exchange

rates

1960

1978

2000

Source: Based on the World Bank data for the 1950s, 1960s and 1970s and data derived from Irving B. Cravis, Alan W. Heston and Robert Summers' project on purchasing power parity (see "Real GDP for More Than One Hundred Countries," The Economic Journal, June 1978) and Robin Marris' pzoject on "Catch-up slow Down or Convergence? Statistical Observations on 25 Years of World Economic Growth in the Light of Kravis Numbers'. The 1978-2000 figures are e_trapolations of 1960-1978 trends. Note that the centrally planned economies' trends ate derived f_om official data which may be subject to revision.

LONG RUN GROWTH TRENDS

253

REFERENCES Cody, J., Hughes, H. and Wall, D. eds. (1981), Policies for Industrial Progress in Developing Countries, World Bank, Oxford University Press. Colaco, F. X. (1980), "The International Economic Environment and the Developing Countries, 1960-90," EPDIT Division Working Paper 14, World Bank. Havrylyshyn, O. and Wolf, M. (1981), "Trade Among Developing Countries Theory, Policy Issues, and Principal Treands, ' World Bank Staff Working Paper No. 479. Hope, N. (1981), "Developments in and Prospects for the External Debt of Developing Countries," World Bank Staff Working Paper No. 483. Hughes, H. and Waelbroeck, J. (1981), "Trade and Protection in the 1970s: Can the Growth of Developing Country Exports Continue into the 1980s? The World Economy, 4: Kravis, I. B., Kenessey, Z., Heston, and Summers, R. (1975), A System of International Comparisons of Gross Product and Purchasing Power, The World Bank, Baltimore and London: The Johns Hopkins University Press. Kravis, I. B., Heston, A. and Summers, R. (1978), International Comparisons of Product and Purchasing Power, The World Bank, Baltimore and London: The Johns Hopkins University Press. .... (1978), "Real GDP Per Capita for More than One Hundred Countries," Economic Journal, 215-242. Marris, R. (), "Catch-up, Slow down or Convergence? Statistical Observations on 25 Years of World Economic Growth in the Light of Kravis Numbers," Morawetz, D. (2977), Twenty-five Years of Economic Development, 1950-75, The World Bank, Washington, D.C. OECD, Development Cooperation, Paris Sapix, A. and Lutz, E. (1980), "Trade in Non-Factor Services: Past Trends and Current Issues," World Bank Staff Working Paper No. 410. .... (1981), "Trade in Services: Economic Determinants and Development Related Issues," World Bank Staff Working Paper 480. Schwartz, A., Lutz, E., Jacobsen, I. and Lee, A. (1980), "World Trade Flows by Origin and Destintion, 1970-90: A Functional Analysis," World Bank EPDIT Division Working Paper No. 80-8. Swamy, G. (1981), "International Migrant Workers' Remittances," World Bank Staff Working Paper No. 481.

RELEVANCE OF THE EXPERIENCES OF JAPAN TO CONTEMPORARY ECONOMIC DEVELOPMENT Shigeru Ishikawa*

1. Introduction History is a set of multi-sided, once-and-for-all events. So is a particular experience of any one country in the past. If the interest in such an experience is historical, the totality of that experience matters. If the interest stems from the specific purpose of acquiring some hint or clue for identifying the interrelationship of factors behind some particular development issue of a contemporary developing country or if the purpose is to derive some lesson in formulating an effective policy to resolve that particular development issue, one has to choose from the history of another country, a certain specific experience which involved a similar issue, and to focus attention to some causally related facts and factors which were responsible for having brought about that particular issue. In order to identify these facts and factors however, the other facts and factors involved in the same experience, which originated from the conditions particular to that country in the past, especially those which came from the natural environment, initial conditions and international environment, must be effectively controlled. Based on previous studies on contemporary economic development or on the experience of prewar Japanese economic develop*Institute of Economic Research, Hitotsubashi University, Tokyo. The original version of the paper was presented at a regular Japan Research Centre Seminat, S.O.A.S., University of London, on April 14, 1980 under the title of"Economic Development and the Experience of Japan arid East Asia." The comments made on it by Professors Hla Myint, R.J. Fisher, K.N. Raj, Konosuke Odaka and Yukihiko Kiyokawa are highly appreciated.

255

256

SHIGERUISHIKAWA

ment since the early Meiji era, 1 it seems increasingly important to keep in mind the difference of the above two approaches to historical experiences. The present paper aims mainly at elaborating this point. In the following, three different approaches to what should be "relevant" Japanese experiences to contemporary economic development are elaborated by referring to various examples, and the merits and demerits of each of them are discussed. Section 2 takes up an approach which considers the entire prewar development. Section 3 shows the approach to specific success stories as "relevant" without paying due attention to the various conditions particular to Japan. Section 4, whether successful or not, pays attention to certain specific experiences to explore really relevant factors, by controlling the particular conditions of Japan. Most of the concrete examples are taken from the author's works.

2. A Stylization of Japan's Development Process Japan's experience in prewar economic development has, since the early 1960s, been increasingly referred to in development literature as relevant for contemporary developing countries attempting to "learn by reference". There are three reasons for this: (1) Among the countries which so far succeeded in economic development, Japan is the latest "late comer" country, and the economic level and st.'ucture in her development stage retained the most number of similarities to those of today's developing countries in terms of per capitaincome, factor endowments, industrial structure, technology gap, and some other criteria; (2) for the developing countries in Asia, the Japanese experience, in particular those relating to the development of rice agriculture, constitutes a more relevant and more easily applicable experience than that of other developed countries; and (3) the entire process of Japan's economic development is by now well documented and even statistically outlined with a fair degree of precision and detail. 2

1. IshJkawa,S. (1967) and Ishikawa,S. (1981). 2. The most important statistical study is published in 13 volumes of Ohkawa, I(., Shinohara, M. and Umemura, M. (eds) (1965). The last 3 volumes are forthcoming. For the summarized version in English, see Ohkawa, K. and Shinohara, M. (eds) (1979).

RELEVANCEOF JAPAN

257

These reasons are quite legitimate and hence reference to the Japanese experience is usually made, per se. The reference to it becomes useful and even desirable. Some caution must be exercised, however, in taking the reference value of Japan's story of success in resolving developmental issues and in its applicability to contemporary development problems. If this careful attitude is lacking, the reference to various success stories in Japan is misleading, and sometimes even dangerous. In the case where a success story refers to the successful transformation of an underdeveloped economy to a developed one or to similarly successful events of a comprehensive scope, one should not overlook the fact that in Japan, unique initial conditions and international environment were favorable to success, and these conditions and environment are easily identifiable. The reference to such a success story of a comprehensive scope is even useful, if it is stylized as a logical sequence of certain factors under certain clearly defined constraints. Moreover, such stylization leads to the formulation of a special theory of economic development. It may then become possible for one to refer to such a theory as a standard process of comparison with which to analytically evaluate particular development processes of a contemporary developing country. In fact, some of the existing development models were formulated in a similar process and played a similar role for contemporary economic development. Examples are (1) A.W. Lewis' "classical" model of dualistic economic development with unlimited supplies of labor 3 which is based on the stylization of experience in British industrialization in the 18th and 19th centuries; (2)A. Gerschenkror_'s hypothesis of "relative backwardness", or of "late comer" countries' industrialization 4 based on the comparative experience of industrialization in the Western European countries; (3) G.A. Feldman's model of heavy industrialization under centralized economic planning 5 based on the Soviet Russian planned economic development; and (4) M.H. Watkins, J.W. McCarty, D.C. North and others' "Staple Theory" of trade and development 6 based 3. Lewis, A. W. (1954) and Lewis, A. W. (1958). The development model by John Fei and Gustav Ranis is formulated essentially according to the Lewis model. Fei, J.C.H. and Ranis, G. (1961). 4. Gerschenkron, A. (1962). 5. Summarized by E. Domar (1957), ch. 9. 6. Watkins, M. H. (1963); McCarty, J. W. (1964) and North, D. C. (1955).

258

SHIGERUISHIKAWA

on the experience of the early period of development of Canada, Australia and United States where the economy was developed first on the basis of expansion of natural resource-based export production and where the manufacturing sector was finally built up through the backward linkage effect. 7 Unfortunately, the Japanese development experience does not seem to have been stylized and formalized as a development model from the point of view of relevance to contemporary economic development. There were a number of serious attempts, of course, to summarize the Japanese experience in internationally comparative terms on the basis of the analysis of carefully collected and processed statistical data. 8 These works however, seemed to identify those aspects of Japanese economic development which were unique as compared to those of other Western industrialized countries, and to explain why such uniqueness arose, rather than relate the Japanese experience to contemporary development issues. Nevertheless, the existing data and studies are already fairly sufficient for an attempt at the stylization of the prewar Japanese experience and to pinpoint its relevance to contemporary developing countries. A tentative stylization and some of its essential components are given below. A. The basic conditions

of the economy

1. The economy consisted of two institutionally different sectors: a 'traditional' and a 'modern' sector. 9 Broadly, these 7. In connection withthe development model of the Staple Theory, H. Myint's "Vent-for-Surplus" model might be refered to (1971), ch. 5. The model is based on the experience of the opening-up of land frontier for export production in the small natural-resource rich and sparsely populated countries hke Thailand and Burma since the mid.18th century. This e_perience, however, lacked the process in which natural resource development led to the creation of the modern industry sector, or at least to the increase in land productivity. The Vent-for-Surplus model accordingly lacks the formalization of an endogenous mechanism leading to industriahzation such as the Staple Theory. 8. Among those specifically notect are Ohkawa, K. and Rosovsky, H. (1973), Shinohara, M. (1962) and Fujino, S. (1965). 9. The production units of the traditional sector consisted of individual families whose major objective was maximization of family-welfare. Often, member families of the same villagecommunity were treated in the same way as quasi-members of the same family. The production units of the modern sector consisted of capitalist firms whose emphasis was maximization of capitahst profits.

RELEVANCEOF JAPAN

259

two sectors corresponded to the agricultural and nonagricultural sectors, respectively. Natural resources were not rich. In the earlier period of development, capital was scarce, labor was abundant and even in considerable surplus. 1째 The rate of population growth was low by the contemporary developing countries' standard. The economy was mostly open and exposed to foreign trade.

B. The process of economic development 1. During the whole prewar period, technological progress and its geographical diffusion in the traditional (agricultural) sector brought about a steady increase in labor productivity and, correspondingly, the ratio of the amount of agricultural products which were marketed to the modem sector increased substantially, although the intersectoral commodity terms of trade did not change substantially in favor of agriculture until World War I. After that, however, they changed substantially despite the large influx of rice from Korea and Taiwan. 2. Side by side with the increasing inflow of agricultural products, surplus labor in the traditional labor was supplied to the modem sector almost without limit at least until World War I. This led to the expansion of the modem sector at a very high growth rate, while maintaining a very large relative share of non-labor income judged by international standards. 10. The definition of "surplus labor" in this paper follows the conventional one, namely that part of the labor force in the farm sector that can be removed without reducing the total amount of output produced, even when the amount of other factors is not changed (but by some reorganization of the inter-farmstructure). This state of"surplus labor" arisesin micro.economic terms when in the farm sector, at least some farms' willingnessto work more at the existing wage rate or marginal income is not satisfied because of the insufficiency of demand. Sen, A.K. (1966) and Ishikawa, S. (1967) ell. 4. In this state, it is quite likely that the marginal productivity of labor in the farm households is lower than the market wage rate as indicated in Ryoshin Minami's (1973) definition of "disguised unemployment" or "unlimited labor". Meanwhile, Harry Oshima's (1958, 1981) well known def'mition of "surplus labor" purely in terms of seasonal unemployment is very interesting. Recently, he even associated an essential part of his development model of monsoon Asia with this surplus labor concept. While I agree with this conception to a large extent, I do not suggesthere a reconciliation of these two approaches.

260

SHIGERUISHIKAWA

3. The choice of industries and technologies in the modern sector was such that there was a gradual and evolutionary improvement and upgrading of the industrial and technological structure. It also involved an alternating process of import substitution and export promotion in each of the successively expanding industries. Some of the important aspects in the above styhzation are taken up again in later sections. 11 3. Initial Conditions

and the Role of Agriculture

The presentation of a success story such as Japan's was misleadLag mainly with respect to particular events or aspects of less comprehensive scale. Two examples are shown below from the discussion of the Japanese experience during the 1950s and the 1960s. First, there was an argument 12 that Japanese agricultural development since the Meiji era took place on the basis of (1) Small family size farms and (2) technical progress made possible by both varietal improvement and increased application of quick-pay-off purchased fertilizers. Therefore, in contemporary developing countries in South and Southeast Asia, agricultural development as a

11. Tile kind of model anticipated by this stylization is a dualistic development model of industrialization such as Lewis' model described above. But it differs significantly from Lewis' model in the following two points: (i) As a necessary condition for the development process to proceed smoothly, it puts a greater emphasis on the productivity increase of the traditional agricultural sector, and for this to be promoted, on the changes in the terms of trade in factor of that sector, even in the stages where surplus labor remains. The existence of surplus labor itself implies the potential unlimited supplies of labor from the traditional sector, but this potentiality is only realized when that sector is capable of supplying the amount of food (wage goods) to feed the labor thus supplied. In other words, it is essential that the ratio of food output that can be marketed increase at the same time as the surplus labor is supplied. (ii) As another necessary condition relating to the process of development, the process of capital accumulation in the modern sector is to be formalized in greater detail in terms of the choice of industries and techniques. Thisis necessary essentially because the process of Japanese economic development was typically a process under the open economic system exposed to foreign competition. 12. See for instance, Ogura, T. (1963), Part IV which was written by the experts who participated in the FFHC of FAO's Expert Meeting on Japanese Agricultural Development in January 1963.

RELEVANCEOF IAPAN

261

prerequisite of economic development should be made possible, institutionally, on existing small-farming system even without land redistribution and financially, by inexpensive methods of technical change. In the 1970s and thereafter it became apparent that these arguments and, in particular, the policy suggestions based on them were too simplistic to be realistic. Detailed surveys of the existing studies as well as additional studies brought out the following points as effective counter-arguments. (i) Flood control, irrigation and drainage projects are essentially the precondition for yield increase in rice-based agriculture by way of introducing biological and chemical innovations; (ii) these land infrastructure projects usually require a huge amount of resource costs and (iii) while in Japan such land infrastructure were already completed in elemental, basic forms for nearly all rice lands before the Meiji Restoration, this was not so in most developing countries in South and Southeast Asia. (iv) Diverse patterns of complicated land ownership and tenurial relationship existed in South and Southeast Asia; interfamily distribution of both land ownership and operational holdings in most areas were much more skewed.13 Second, there was an influential argument which is closely related to the f'wst argument during the 1950s. In prewar Japan, in particular during the Meiji era, the agricultural sector supplied on a net basis a significant amount of savings to the emerging modern industry sector, thereby contributing to the success of her industrialization. The land tax was considered among the major instruments of this saving transfer out of agriculture in the earlier period. In the early stage of industrialization where agriculture was the only major sector the source of investments for funds should derive from the savings surplus to be generated in the agricultural sector. Otherwise, industrialization cannot successfully proceed. This should also be the case in industrialization of contemporary developing countries. 14 As against this argument, a debate on the size and direction of saving or, more generally, resource transfer between agriculture and nonagriculture, to be stricter, between the farm and nonfarm sectors in the process of contemporary economic development began in the 13. lshikawa, S. (1967), ch. 2. 14. See for instance, Johnston, B. and Mellor, J. (1961).

262

SHIGERUISHIKAWA

mid-1960s. It was hypothesised by critics that in the contemporary developing countries where basic land infrastructure in agriculture was yet to be built up, the farm sector had to be supplied with capital required for it from outside. If food and agriculture would not become a bottleneck impeding deliberate industrialization, the required capital must be larger, especially because a higher, post-war rate of increase in agricultural population tended to make the ratio of the amount of agricultural products that was marketed to the nonagricultural sector smaller than expected. Therefore, it was considered possible that once a deliberate attempt at industrialization was started and continued in the contemporary developing countries, the intersectoral resource transfer had to turn sooner or later to the net inflow from industry to agriculture. Major factors in this net outflow were either or both of the change in the intersectoral commodity terms of trade in favor of agriculture and the government investment in land infrastructure in agriculture. In the same line of argument, it was held that if it had been statistically confirmed that the net intersectoral resource flow in prewar Japan was, as claimed, a net outflow from agriculture, that would mean that agriculture in Japan contributed not only to the substantial amount of unlimited supplies of labor to industry (i.e., both actuallabor and the constantly cheap wage goods to feed it, 1s at the same time), but also at least part of the funds to employ such labor freely. The reason why this contribution was made possible was the specially favorable conditions surrounding Japan, e.g., the initially existing land infrastructure investment, the successful technological progress in agriculture achieved on that basis and the low rate of growth of agricultural population. 16

15. For simplicity of discussion, the agricultural raw materials are assumed away as part of the agricultural products supplied to the non-agricultural sector. 16. Ishikawa, S. (1967), ch. 4. The debate has been made from the beginning on the basis of the preparation of precise statistical concepts and framework on the intersectoral resource flow. For instance, the sectoral division of agriculture and nonagriculture or farm or nonfarm, capital transfer and more comprehensively, resource transfer comprisingboth capital and current transfers, and the resource transfer at current prices arising from the interseetoral commodity transactions or file resource transfer at constant prices (i.e., the resource transfer taking into account the terms of trade effect also).

RELEVANCEOF JAPAN

263

By now, many economist seem to agree that the policy suggestion of the 1950s advising the contemporary developing countries to extract as much saving surplus as possible from agriculture, was too simplistic, although careless statements favoring the same policy suggestion arestill heardfrequently. Therearealsoa few statistical works on thesubject. Itisworthwhiletonotehowever,thata study of Indiafortheyearsbetweentheearly1950sand theearly1970s estimatedthatthe net resourceflow of agriculture was an inflow in the early1950s at both currentand constant1960-61prices. 17 Then itturnedintoan outflowand theamount increased, butafter the rnid-1960s, the periodwhen a new agricultural development policyassociated with the Green Revolutionwas enforced,the direction ofnetflowturnedtoan inflow. Taiwan's case of intersectoral resource transferwas also estimatedsystematically fortheperiodbetween 1895-1960.18The resultindicates invariably net outflowsfrom agriculture forthe entireperiodof 1911-15to 1956-60at both currentand constant (1935-37)prices. Thisprovidedadditional evidenceasidefrom that of Japan thatagriculture shouldbe a net resourcecontributor for industrinlization. Itshouldbe noted,however,thatduringtheprewar years,Taiwan'sexternaltrade mostly with metropolitanJapan exhibited exportsurplusinvariably and itsamount was suchthatthe net resourceoutflowout from agriculture was nearlyentirely, or more than,counterbalanced, therebyleavingthe amounts of net resourceoutflow remainingwithinthe nonagricultural sectorof Taiwan almost negligible, whether thesebe positive or negative. PrewarTaiwan'scaseseemstorepresent atypical patternofresource transfer out of agriculture in a colonial typemonoculturecountry, rather than a typicalpattern in the course of deliberate industrialization. 19

17. Mundle, S. (1981). 18. Loe, T. H. (1971). 19. Two sets of figuresin Lee's estimates areimportant in this connection. I. Relation between agriculture'snet resource ouflow (A) and the economy's export tradesurplus(B): T$ million at 1935-37 value. 1911 -15

1916 -20

1921 -25

1926 -30

1931 -35

1936 -40

264

SHIGERU ISHIKAWA

Systematic efforts at estimating intersectoral resource transfer for prewar Japan has also been carried on. 20 So far, however, only the estimates at current prices are nearly completed, which indicate that large proportions of agricultural savings were outflows on a net basis. Due to the non-availability of reliable price indices constituting the intersectoral terms of trade index, the estimates at real prices have yet to be completed. 21

1. A 2. B 3. A-B

50 11 9

62 41 21

60 60 0

59 65 -6

89 91 -2

90 90 0

II. Composition of net resource outflow of agriculture: current T$ million 1. Net resource outflow 2. Net current transfer 3. Of which, land rent 4. taxes and fees 5. Net capital transfer

22 6

42 9

50 15

57 16

55 17

-2

-5

-9

-1

Panel II indicates that most of the net resource outflow of agriculture at current prices was realized through the payment of land rent by the farmers to the absentee landlords. Net capital transfer (net saving transfer) was negligible and even negative. These findings are in sharp contrast to the case of prewar Japan where (i) net capital transfer was positive and very large in the sense that the net capital transfer (plus taxes and fees) occupied about a half or more of total agricultural savings (plus taxes and fees) was negative and if minus taxes and fees was also negative and as large as to substantially offset the net capital transfer to the amount of net resource outflow (at current prices) very small: See Mundle, S. and Ohkawa, K. (1979). 20. By the group led by Kazushi Ohkawa. The latest version is published by Mundle, S. and Ohkawa, K: (1979). To be mentioned also is Teranishi, J. (1976), which concentrates on the estimates of the intersectoral resource flow from the financial statistics, confining the scope of resource flow to the capital transfer (plus tax payment) at current prices. 2i. This means that of the two components of net resource flow of agriculture, namely, one, due to the terms of trade effect and two, due to the balance of intersectoral commodity trade at current prices, the amount of the second component and, hence, the direction and scale of the entire net resource

RELEVANCEOF JAPAN

265

Upon reflection, it seems that the debate later tended to lost its substantive content from the point of view of economic development, with too much attention paid to the numerical results of the estimates. There was insufficient attention paid to the issue of net resource flow between agriculture and nonagriculture, an issue which arose when two kinds of products, one requiring the use of limited land resources for their production and the other not, are exchanged in the product market. More specifically the issue flow are not yet finally confirmed. It may be noted further that while more appropriate price indices for evaluating the terms of trade effect are yet to be constructed, the existing observation of a seculartrend of change in the terms of trade in favor of agriculture (in particular that which emeged after around 1918) made in Ohkawa, K. and Shinohara, M. (1973) oh. 12 would not be changed drastieally. If so, it is likely that the net resource transfer due to the terms of trade effect was an inflow into agriculture and it more than offset the nel resource outflow by current commodity transactions in many prewar years. In evaluating the terms of trade index, the choice of appropriate base year always poses a problem. In this sense, the following calculation of the real resource flow by taking the average of 1888-92 price indices as 100 is nothing but a hypothetical one for simple illustration. But it may suggest the range of the above. Real net resource outflow of agriculture in million yen at 1988-92 prices: E M 1. Real net resource outflow ( Pe _ ) - 52.6 2. Of which due to the interseetoral trade balance 57.5

152.0 6.7

E-M

(

l"e

)

3. due to the terms of trade effect

-I I0.I

-158.7

M Pe [-_---( I/-_--- I)] Notation: E = Exportatcurrent prices, M = Importatcurrent prices, Pe = price indices of agriculturalproducts marketed, Pm= price indices of manufactured products purchsaed. Notes:

Data for E and M are taken from Mundle, S. and Ohkawa, K. (1979). Pe and Pm are taken from Ohkawa, K. and Shinohara, M. (I979).

266

SHIGERUISHIKAWA

involving competing claims of capitalists (profits) and landlords (rents) for the surplus created in the capitalist (modern) sector, and the consequences of the rivalry can be traced to the changed in intersectoral terms of trade. When the resource interflow outside the market mechanism assumes a significant weight, this should not be overlooked. However, the primary channel of resource transfer is still that which is realized through relative price adjustment in the product market. 22 4.

Controlling

the Conditions

Particular to Prewar Japan

Cases of the "relevant" Japanese experience in economic development are mainly the ones that suggest possible critical factors which work behind current developmental issues and which contemporary developing countries are struggling to identify to formulate effective measures. The experience of the country on the basis of which universal causes and effects of a particular issue can be reasonably identified, by controlling the conditions particular to Japan. In fact, the experiences investigated in the two sections above are intended to become useful experiences ultimately. In as far as the total experience of Japan's prewar economic development is concerned, what was presented was an attempt at stylization of the basic conditions of the economy and the processes of development. But this suggests that once thay are formalized as a model, they would serve as useful analytical instruments. As for the "success stories", an attempt was made to identify the effect of the initial and other conditions particular to Japan and on that basis it was argued that what were often held as the caused for the successes are not relevant to the developing countries. But in these cases, too, it would probably be possible to find out the underlying factors and their interrelation that are relevant to these developing countries. In these sections, three other cases of Japanese experience will be shown, and for each case an attempt is made to control the effects of the particular conditions of Japan and thereby identify relevant factors and relations underlying each experience. As will be seen, these cases are different from the two cases in the previous section in 22. Lewis, A. W. (1954). Also, see John Robinson's discussion on two kinds of prices which depend upon the two different bases of specialization, respectively: one, natural endowments and the other, economies of scale. Robinson, J. (1960).

RELEVANCEOF JAPAN

267

that the policy suggestions usually made in reference to these experiences seem broadly yalid, but not necessarily relevant and specifically or rigorously useful. The effort of effectivity controlling the particular conditions become all the more necessary. Yet, it is not necessarily easy and efforts need to be continued. Case 1: Large scale absorption of labor in flee-based agriculture. An earlier inter-Asian comparative study on labor utilization in both single rice cropping and total agricultural production indicated that the per hectare labor input of Japan for the entire period after 1860 and before the 1950s, and to lesser extents those of other East Asian countries were significantly larger, often 4 to 5 times, than those in South and Southeast Asian countries. 23 This finding seems to have been shocking to some economists in the countries of South and Southeast Asia who were concerned about the employment problem which was becoming acute. Since the early 1960s it has been increasingly recognized that with the present rate of increase in population and labor force it would not be possible for most of these countries to begin absorbing a substantial portion of incremental labor force in the urban modern sector within the next decade or two, even with a significantly raised growth rate of that sector. An investigation of the practical possibilities and the policy measures for increasing the labor absorption in agricultural production in these countries was started several years ago and is still under way. 24 Studies of conditions special to Japan necessarily pointed to the irrigation facilities that were widely constructed before the Meiji Restoration. More interestingly, it has led to a consideration of the difference in international environment which can be explained by a kind of "relative backwardness" hypothesis in agriculture. There is indeed a clear difference in the choice of techniques in rice and 23. Ishikawa,S. (1967) ch.3. 24. The investigation here denotes the ILO-ARTEP (Asian Regional Team for Employment Promotion) project on Labor Absorption in Asian Agriculture which was started when K.N. Raj was director of ARTEP. I was requested to write the "issues" paper for the project under the title of "Labor Absorption in Asian Agriculture - An 'IssuesPaper" which was later incorporated in lshikawa, S. (1981), oh. 1. The project has so far produced four publications: Bardhan, P.K. et. al. - (1978), ILO-ARTEP (1979A), ILO-ARTEP (1979B) and ILOARTEP(1980).

268

SHIGERUISHIKAWA

other crop production that could be made in prewar Japan as the earlY-comer country, with respect to modern rice production, and that can be applied presently in the countries of South and Southeast Asia as the late-comer countries. This is because, first, in Japan when deliberate effort was began to increase per hectare crop output, most of the yield-increasing inputs were labor-using. But with the progress of industrialization and urbanization, technological changes took place which replaced labor-using inputs gradually by labor-saving inputs such as chemical fertilizer, electric pumps, pedal and motorized threshers and even power tillers; second, for the countries of South and Southeast Asia even though the yield levels were substantially lower than that in the early Meiji era (one-half or less for rice, for instance), the above yield-increasing and labor-saving inputs were in many cases, the only alternative inputs, because their prices were much cheaper than the existing inputs and some of the traditional yield-increasing and labor-using inputs in prewar Japan were simply not available now. This suggests the possibility that the scope for reducing the large differentials in labor absorption may not be as large as it first appeared to be. It may be said, however, that the crucial stage of "relevance" study has just began. A number of important aspects of Japanese experience which remain even after controlling these particular conditions are noticeable. (i) the mechanization devises such as electrical pumps after the 1920s and pedal and motorized threshers after the 1930s, which were conventionally considered labor-replacing in its technical nature, often brought about a labor-using effect. This was made possible by using these devises to increase the rate of land utilization, such as the expansion of the area of rice land with a second crop. (ii) the characteristic of the organizational structure of agriculture in which farm households were small-scale and fairly equally-distributed in terms of both ownership and operational holdings was certainly conducive to increased labor input in land. These are contrasted with the organizational structure of agriculture in contemporary developing countries of Asia where operational holdings are more less bipolarized. In some countries a large number of landless laborers exist. Mechanization for profit has taken place mostly by the richer classes in the agricultural sector. Hence, mechanization has been seldom used to increase labor absorption. (iii) the importance of increasing irrigation facilities and choosing appropriate technologies relating to them and (iv) the importance

RELEVANCE OF JAPAN

269

of encouraging a close cooperation between local farmers and local agricultural experiments for effective diffusion of better fanning techniques in Asia which will increase labor absorption. The problem is how to add more cases in the list of "relevant" experiences and synthesize them for useful policy suggestions. Case 2: The role of the small- and medium-sized f'_ms in industrial development. Since the first national factory census was made in 1909, the observed size structure of Japanese factories was continuously skewed toward the small (0-50 employees) and medium (51-400 employees) sized groups. To be more precise, observing for instance the proportion of total number of employees apportioned in each of the size classes with the number of employees 20-49, 50-99, 100-499, 500-999 and 1000+ (persons) for the years between 1909 and 1958, the weight of these size classes did not differ very much from one another and the relative weight of each site class did not change with the passage of time. This is in sharp contrast to countries like India and Pakistan where the weight of the largest size class transcends that of other size classes. In other developing countries in Asia, there is also a tendency for the larger sizes to be lopsided. In the cottage size class with employees below 20 persons, there is a clear tendency, common to many countries, for its weight to decrease with the growth of per capita output. Hence this size class was not included in the above comparison.2s Moreover, in Japan these smaller sized factories have significantly smaller capital equipment per worker and labor productivities, far lower than those of the larger sized factories. Due, however, to the considerably lower wage levels prevailing there, they could maintain their competitive power vis-a-vis the larger sized factories. The usual argument on the basis of the above observation is that Japan's experience indicates the crucial importance of smaller and medium industry development in the development process, both for industrialization by capital-savingmethods and for relieving the unemployment problem, which the contemporary developing country should learn. This is most probably a proper advice, especially because of the unfavorable initial conditions of the contemporary developing 15. Ishik_wa, S. (1967), ch. 5.

270

SHIGERUISHIKAWA

countries such as rapid population growth and low capital endowment. This advice is weak, however, in that it lacks a concrete prescription, based on Japan's experience, of what steps should be taken for its implementation. In order to seek such a prescription, one must first disentangle many factors involved in bringing about the macroeconomic behaviors of the small and medium enterprises as prescribed above. Here, only a few points can be made on the reasons for the long-term persistence of small and medium sized enterprises in Japan. (i) The persistence of the smaller sized enterprise is essentially a market phenomenon arising from the equilibrium labor and capital allocation among the different sectors and among the different size groups in the manufacturing sector. It is by no means culturally or sociologically determine. (ii) The parallel expansion of the smaller sized groups and larger sized groups is explained by their complementary relationship in sub-contracting, long term sales-purchase agreements or their cooperation on the management level. This relationship is found to be stronger, as the size group becomes smaller, as the series of postwar comprehensive surveys of small and medium firms in Japan indicate. 26 But it should be noted that this relationship tends to arise and develop only after the size of the product markets for individual industries reaches certain minimum levels which enable the manufacturing firms of respective industries to capture sufficient economies of scale. These levels are not yet attained in most of the industries in the countries of South and Southeast Asia. (iii) As a factor which is related to the particular international environment surrounding the prewar industrial development, it should be noted that the prewar state of industrial technology development among industrial powers was such that the range of products for which the smaller sized firms in Japan were capable of adapting the imported technologies and even the products themselves to become competitive producers, was much wider than the range presently allowed for the smaller sized firms in South and Southeast Asia. This is to be discussed next. Case 3: The technology nology.

development

26. Ishikawa,S. (1981), ch. 4.

stage and appropriate

tech-

RELEVANCEOF JAPAN

271

In previous studies on prewar Japanese experience with regard to choice of appropriate technologies in processing and the machinery industries, it was observed that in any prewar period, alternative production technologies actually chosen or attempted to be chosen consisted of the following seven catagories: 27 (i) The kind of foreign technologies which the government or firms tried to introduce, but were not successful mainly because of insufficient human technological knowledge. 28 (ii) The kind of foreign technologies which were transplanted under protective policies. (iii) The kind of foreign technologies which were adapted locally in the labor-intensive direction by replacing ancillary equipment with labor or with even primary equipment redesigned to economize on capital investment required. (iv) The kind of foreign technologies whose product design and market demand conditions, and hence whose capital and technological capability requirement, were made much smaller. (v) The kind of foreign technologies which were outmoded in developed countries but were still appropriate given factor endowment and the prevailing factor price ratio. (vi) The kind of foreign technologies which were established as domestic or naturalized technologies through channels (ii) -

(v). (vii) The indigenous technologies which survived due largely to the fact that the consumer maintained special preference over the products made from these technologies, Of these, categories (iii) and (iv) and the course of technology development in which category (i) was steadily converted to category (iv) through development of categories (iii) and (iv), were considered as relevant experience to contemporary developing countries. Among many examples for category (iii) were mining facilities in the Meiji period and improved hand-reeling of silk machines in the Suwa 27. Ibid. 28. It is assumed that a country's technological capability consists of two components: (i) objective technological knowledge either embodied in machinery and equipment or in written documents and drawings and (ii) human or subjective technological knowledge obtainable by formal and informal educa. tion and by experience.Broadly, thereis a significantcomplementary relation. between the two components.

272

SHIGERUISHIKAWA

district made of ceramic kettlers and wooden frame (1875). Those for category (iv) were power looms for single breadth cloth (1887), three wheeled trucks (1910s) and small-sized passenger cats (1933). In the light of these experiences special attention was focused on the recent events in Thailand and the Philippines regarding the design and manufacture of extremely simplified, adaptive models of imported power-tillers, 29 and some spread of low-cost four-wheelers of the Asian Utility car type. 3째 However, given the conditions particular to Japan, this view appears somewhat simplistic. Firstly, the following three points should be noted in connection with the above Japanese experience. (i) With the exception of the cases which emerged in the early Meiji period, the cases of process adaptation in category (iii) and product adaptation in category (iv) were realized in the modern larger-scale factories as a last measure to establish those industries whose technological sophistication was perhaps very high for Japan in respective periods after various efforts were taken to establish them by ordinary means relating to categories (ii) and (v). They were therefore by no means "minor improvements" without major devotion to the existing technological capability. (ii) The level of industrial technologies which was achieved in the Japanese machinery industries by the end of the prewar period was not very low as compared with the level of industrial technologies achieved in the western industrial powers at that time. Most of the western technologies which were then considered as "matured" were technologically mastered. 31 Exceptions were the frontier technologies such as those involved in the production of airplanes, passenger cars and machine tools, 32 for which domestic markets were secured by deliberately making the products "lower in quality but cheaper in prices", hence non-competitive vis-a-vis foreign products. 29. Ishikawa (1981), Postscript to ch. 4. 30. UNIDO(1978). 31. See a study of the development of Japan's industrial machinery industry according to the well-known Catching-up Product Cycle Pattern of industries (or Flying Geese Formation Pattern) in Mitsubishi Keijai Kenkyujo (1963), Part III. 32. See an analysis of the technology experts on the progressof Japanese industrial technology. See Hoshino, Y. (1956).

RELEVANCEOF _APAN

273

(iii) Many reasons are conceivable behind the above performances. But one should note a steady progress of formal education. This refers not only to the high enrolment ratio of primary school pupils which started from 28 percent around 1873 and soon after the turn of the century reached 98 percent. The system of middle and higher level education for technicians and engineers was steadily expanded in response to the demand increase. The effect of "learning by experience" in production was firmly based on human technological knowledge fostered by formal education in various dimensions. 33 Thus, by way of elaboration, the process of the rise and development of the machinery industries in prewar China may be compared in some respects with that of Japan. As a framework of comparison, a fairly universally observable sequential pattern can be noted in the rise and development of the machinery industries in the "late comer" countries. 34 In this pattern, (1) the machinery industries started with the firms or workshops in the factories which were engaged in maintenance and repair of the imported machinery and equipment, (2) each of the machine industries progressed with the production of increasingly sophisticated replacement parts; (3) the development of such machinery industries in a number of branches led to the emergence of a market for replacement parts, tools and materials. Many specialized replacement parts maker also arose to cater to this market, (4) finally, the machinery industries capable of producing or even adapting the machinery arose in succession. Following this framework, it is interesting to note that in both China and Japan the machinery industry started as the maintenance and repair section of the ship-building industry around the same period; in China around 1866 and in Japan 1861) 5 Furthermore, 33. The number of students in the system of middle and highereducation in industrial technology increased especially rapidly in three occasions: (i) around the Russo-Japanese War, (ii) after World War I and (iii) after the beginningof the Sine-Japanese War in 1937. 34. Ishikawa's preface to Odaka, K. (1982). This refers to the pattern which arose as a spontaneous process of the development. There wasanother pattern which arose when the establishment of the machinery industries took place as the state enterprises, which was not dealt with here. 35. For China, the year of the establishment of Fachang Machine !_aetory. For Japan, the year of establishment of Nagasaki Iron Works, the forerunner of Nagasaki Shipyard, Mitsubishi Shipbuilding Co. The establishment of this Iron

274

.SHIGERU ISHIKAWA

on the basis of the experience of the ship-building, textile machines and some other branches of the machinery industries, the following may be said: 36 in China, the time intervals between the start of the industry as a repair and maintenance section and the manufacturing of the machinery by its capability were in general very long, and meanwhile the proportion of the number of factories in the industries that were specialized in repairing activities were continued to dominate. 37 In Japan, in contrast, the time intervals were very short and the major activities were soon turned into the manufacturing proper. 38 Another important contrast is that in China, the "learning by experience" on the production workers' throughout knowledge. machinery

weight of level was

very large in the formation of human technological This was exhibited in the establishment of a new industry in which not only the skilled workers but also

even the managers were supplied by the spillover effect of the previously existing industries. The increasing sophistication of the replacement parts manufactured in these industries was largely due to the learning by experience effect. Nevertheless, the manufacturing of spinning machines (1928), complete set of cotton power looms (1921), oil engines (1910) and diesel engines (1919) was made possible in the larger-scale factories and under the leadership of

works was intended to make regular repair of the warship Kanrinmaru bought from Nethedand. Before, whenever repairing became necessary the steam ships had to return to Europe. Mitsubishi Shipbuilding Co. (1957), pp. 119-121. 36. The scope of the machinery industries compared is limited, mainly because the nature of the data source of the Chinese side is confined here to that of the history of the machinery industry in Shanghai. Chinese Soeiai Science Academy, Economics Institute (1979). 37. As for the ocean-going shipbuilding industry, the establishment of the manufacturing section had to wait for the establishment of the People's Republic of China (while a few ocean-going vessels were manufactured at the state-owned shipbuilding firm called Jiangnan Zhizao Ju, the management authority was in fact delegated to foreign advisers). If the inland shipbuilding is taken out, the situation was different. Inland shipping started expanding especially after the 1960s centering around Shanghai and the surrounding Lower Yangtze Delta area. Correspondingly, repairing and manufacturing of steam engines also began. As for the textile machines and engines, see the text discussion: 38. See Nihon Sangyo Kikai Kogy0-Kai (1964), Ishii, TI (1979) and Sanbei, T. (1941). _ "

RELEVANCE OFJAPAN

275

educated engineers. These factories produced a number of different machine products at thesame time, thus using the available capacity of production more efficiently. In Japan, the factories starting the manufacture of ship-building, coal, mining and textile machines were mostly of the latter-type, and experience mainly of the engineers and technicians played an important role. The above f'mdings suggest very strongly the importance of formal education in different levels as a basic means of raisinghuman technical knowledge. Technologies of categories (iii) and (iv) cannot be obtained, however, simply on the basis of learning by experience in production. Next, there is another point which should be noted regarding the prewar Japanese experience, namely that industrialization and the accompanying development of technological capabilities in prewar Japan took place at the technology development stage of the Western industrial powers which is usually related to the First Industrial or Steam Power Revolution from the end of the 18th century and the Second Industrial or the Electric Power Revolution from the end of the 19th century.39 Though with a considerable lag, Japan was able to achieve these two Revolutions largely before World WaxII. In contrast, the efforts at industrial and technological development of contemporary developing countries have been made at the technology development stage of industrial powers which is regarded as the Third Industrial Revolution initiated by the introduction of automatic control in the production process. And yet, on the basis of the technology development stage of their own industries many of these countries are still entering the Second Industrial Revolution. An implication of this difference in international environment in technological advance is that even for the developing countries, the relative importance of formal education especially in the higher level becomes greater and that of experience less, for acquiring human technological knowledge. As far as production operations are concerned, the requirement of human capability for manual workers tends to decrease, but the human capability required of technicians and of engineers for mainten_ce and repair of the greatly sophisticated Plant _d equipment tends to increase significantly.

39. Hoshino,Y. (1956)andMinami', R.('1976).

276

SHIGERUISHIKAWA

A question, however, is how does the above difference affect the choice of industries and technologies in the process of contemporary industrial development. In terms of the seven categories of technologies observed above in connection with the Japanese experience, it seems that the scope of category (i) increases; in other words, the chances to reduce it by various efforts in connection with categories (ii) through (v), especially (iii) and (iv), tend to be smaller. Instead, the potential role of foreign direct investment as a measure to reduce the scope of category (i) would increase, a role which was rarely observed in Japan and hence not listed as a special technology category. In the light of this inference, it is worthwhile to take note of the recent experience of the Republic of Korea. 4째 First, it is observed authoritatively that many of the principal industries in her rapid industrialization use technologies that can be characterized as mature, in that "the mastery of well-established and conventional methods embodied in equipment readily available from foreign suppliers is sufficient to permit efficient production." In the case of new industries for which more sophisticated technologies were required, e.g., the ship-building industry and the integrated steel industry, only licensing and turnkey arrangements were relied upon. Except for electronics and certain chemicals, exclusive reliance on direct foreign investment was not observable. Second, despite this, the rate of growth of exports has been remarkable. In the case of machinery, the rate of export growth was especially rapid, and in 1975 the ratio of machinery exports to total machinery output reached as high as 32.3 percent. Moreover, 62 percent of total amounts of machinery exports in the same year was directed to the U.S., Japan and West Germany. This suggests that the potentials for the industrial powers to import labor-intensive and technologically simple machinery from the developing countries are expanding. It is not yet certain, however, how far this Korean experience is suggestive of the future course of most of other developing countries where industrial and technological development is lower. What is certain from the above studies is that in order to really understand the relevance of the Japanese experience in the appro, priate technology issue, a much more intensive study than thus far conducted is necessary. ---_i_-Westphal, L., Rhee, V. W. and Purseil, G. (1981). See also Odaka, K. (1982), the chapter written by Chuk Kyo Kim and Chul Hee Lee.

RELEVANCE

OF JAPAN

277

REFERENCES Bardhan, P.K. et. al (1978), Labour Absorption in Indian Agriculture: Some Explanatory Investigations, Bangkok: ILO-ARTEP. Chinese Social Science Academy, Economies Institute (1979), Shanghai Minzu Jigi Gonaye (National Machinery Industries in Shanghai), Shanghai, Dhonghua Shuju. Domar, E. (1957), Essays in the Theory o/ Economic Growth, New York: Oxford University Press. Fei, J. C. H. and Ranis, G. (1961), Development of the Labor Surplus Economy: Theory and Policy, Homewood, Illinois: Richard D. Irwin. Fujino, S. (1965), Nihon no Keiki Junkan (Trade Cycles in Japan) Tokyo: Keiso-Shobo. Gersehenkron, A. (1962), Economic Backwardness in Historical Perspective, Cambridge, Massachusettes: Belknap Press of Harvard University Press. Hoshino, Y. (1956), Gendai Nihon Gi/utsushi Gaitsetsu, (Brief Discussion on Technology History in Modern Japan), Tokyo: Dai-Nihon Tosho Co. ILO-ARTEP (1979), Employ men t Expansion in Indian Agriculture, Proceedings of a National Seminar Held in Bang#lore, 26-28 February 1979, Bangkok. ILO-ARTEP (1979), Employment Expansion in South-Asian Agriculture, Proceedings of a Seminar Held in Dacca, Bangladesh, 20-23 November 1979, Bangkok. ILO-ARTEP (1980), Labor Absorption in Agriculture: The East Asian Experience, Bangkok. Ishii, T. (1979), "Toyoda Sakiehi to Shokki Gijutsu no Hatten (Sakiehi Yoyoda and the Development of Weaving Machine Technology," Hatsumei, 76: (1-6). Ishikawa, S. (1967), Economic Development in Asian Perspective, Tokyo: Kinokuniya. lshikawa, S. (1981), Essays on Technology, Employment and Institutions in Economic Development: Comparative Asian Experience, Tokyo: Kinokuniya. Johnston, B. and Mellor, J. (1961), "The Role of Agriculture in Economic Development," American Economic Review, 51 : 571-581. Lee, T. H. (1971), Inter-Sectoral Capital Flows in the Economic Development of Taiwan 1895-196 O, Ithaca: Cornell University Press. Lewis, A.W. (1954), "Economic Development with Unlimited Supplies of Labour," Manchester School of Economic and Social Studies, 22: 139-191. Lewis, A.W. (1958), "Unlimited Labour: Further Notes," Manchester School of Economic and Social Studies, 26: 1-32. McCarty, J.W. (1964), "The Staple Approach in Australian Economic History," Business Archives and History, 4-1. Minami, R. (1973), The Turning Point in Economic Development: Japan's Experience, Tokyo: Kinokuniya.

278

SHIGERU ISHIKAWA

Minami, R. (1976), Doryoku Kakumei to Gi/utsu Shimpo (Power Revolution and Technological Progress), Tokyo: Tokyo Keizai Shimpo Co. Mitsubishi Keizai Kenkyujo (1963), Nihon Sangy_ Kikai-Kogy-d no Seichff to Kozo (Growth and Structure of the Industria/ Machinery Industries in Japan), Tokyo: Nihon Kikai-Kogyo Rengokai. Mitsubishi Shipbuilding Co. (1957), Sogy_ Hyakunen no Nagasaki Z-csensho (A Hundred Years History of Nagasaki Shipyard), Nagasaki: Nagasaki Shipyard. Mundle, S. and Ohkawa, K. (1979), "Agricultural Surplus Flow in Japan, 1888-1937," Developing Economies, 17: 247-265. Mundle, S. (1981), Surplus Flows and Growth Imbalances, New Delhi: Allied Publishers. Myint, H. (1964), The Economics of the Developing Countries, London: /Hutchison. Myint, H. (1964), Economic Theory and the Underdeveloped Countries, London: Oxford University Press. Myint, H. (1975), "Agriculture and Economic Development in the Open Economy," in Agriculture in Development Theory, ed. L. G. Reynolds, New Haven: Yale University Press, Nihon Sangyo Kikai Kogyo-kai (1964), Nihon Sangyo Kikai Kogy_ no Hatten Katei (The Development Process of Japanese Industrial Machinery Industries)Tokyo: Nihon Kikai-Kogyo Rengokai. North, D. C. (1955), "Location Theory and Regional Economic Growth," Journal of Political Economy, 249-51. Odaka, K. (1982), The Motor Industry in Selected Asian Countries, A Study of Ancillary Firm Development, Singapore: SingaPore University Press (forthcoming). Ogura, T. (ed) (1963), Agricultural Development in Modern Japan, Tokyo: Japan FAO Association. Ohkawa. K., Shinohara, M. and Umemum, M., eds., (1965), ChokiKeizai Tokei Estimates of Long-Term Economic Statistics of Japan since 1968), Tokyo: Tokyo Keizai Shimposha. Ohkawa, K. and Rosovsky, H. (1973), Japanese Economic Growth: Trend Acceleration in the Twentieth Century, Stanford: Stanford University Press. Ohkawa, K. and Shinohara, M., eds., (1979), Patterns o/Japanese Economic Development: Quantitative Appraisal, New Haven: Yale University Press. Oshima, H. (1958), "Underemployment in Backward Economies: An Empirical Comment, "'Journal of Political Economy, 65. Oshima, H. (1981), "A Preliminary Model for Studying the Secular Growth Robinson, J. (1960), "The Philosophy of Prices," in Collective Economic Papers, Vol. Two, Oxford: Basil Blackwell. Sanbei, T. (1941), Nihon Mengy_ Hattatsu Shi (The History of Development of Japanese Cotton Industries), Tokyo: Kei-Shobo. Sen, A. K. (1966), "Peasants and Dualism with or without Surplus Labour," Journal o/Political Economy. 74:425 - 450.

RELEVANCE

OF JAPAN

279

Shinohaxa, M. (1962), Growth and Cycles in the Japanese Economy, Tokyo: Kinokuniya. Teranishi, J. (1978), "The Pattern and Role of Flow of Funds between Agriculture and Non-agriculture in Japanese Economic Development," in Japan's Historical Development Experience and the Contemporary Developing Countries: lssues for Comparative Analysis, Tokyo: International Development Center of Japan. UNIDO (1978), The Manufacture of Low-Cost Vehicles in Developing Countries, New York: United Nations. Watkins, M. H. (1957), Essays in the Theory of Economic Growth, New York: Oxford University Press. Westphal, L., Rhee, Y. W., and PurseU, G. (1981), "Korean Industrial Competence: Where It Came From," World Bank Staff Working Paper No. 469.

THE IMPACT OF SRI LANKA'S ECONOMIC REFORMS IN 1977 ON EMPLOYMENT AND INCOME DISTRIBUTION H. N. S. Karunaflake*

1. Introduction After 1956, Sri Lanka swung away from the relatively free economic arrangements that had prevailed in the Island from the preand post independence period to one of increasing restrictions. Thereafter, economic policy tended to become more and more introspective. The government that was elected in 1956 viewed economic policy as essentially state-controlled. For twenty years thereafter, from 1956 to 1977, the economy was subject to increasing restrictions and controls. Economic policy tended to give weight to the public sector and opportunities for the expansion of private business enterprise were, to a large extent, influenced by political considerations because the government which was largely influenced by socialist and left wing doctrines took hold of administration during these two decades except during the five year period 1965-1970. In the twenty-year period from 1956, welfare considerations, along with restrictions and controls, dominated economic policy and the objective of successive governments was to increase transfers to all segments of the population. This tended to progressively curtail the volume of resources available for investment. The main negative aspects of these policies were increasing unemployment and underemployment, recurring shortages and a growing black market, reduced economic activity in most sectors and very little evidence of *CentralBankof CÂŠylorL 281

282

H.N.S. KARUNATILAKE

diversification and structural change in the economy. On the credit side, increasing welfare transfers to the population helped not only to raise real income levels and cushion the people from the impact of inflation, but also contributed to higher standards of health, medical care, nutrition and education. The physical quality of life improved considerably in this period but this happened at the expense of economic growth, while unemployment increased progressively reaching the highest levels since the country became independent. 2. Economic Reforms The United National Party government came into power in 1977 and realized that policies which previous governments followed in the past twenty-year period have resulted in economic stagnation and social discontent especially among the unemployed youth. Major changes in economic management had to be introduced to reactivate a largely dormant economy and to exhaust the potentials for rapid development through private enterprise. The package of measures worked out by the new government represented a rejection of the earlier inward-looking policies almost in their entirety. This involved dismantling exchange controls, import licensing and other restrictions on foreign trade and investment that had prevailed over a long period of time. The S.L.F.P. 1 government which held office from 1970 to 1977 had not been too enthusiastic about private foreign investment although substantial inflows of official aid had taken place during this period. The new government not only welcomed foreign aid but also private foreign investment. The government was convinced that the resources that could be mobilized internally would not be adequate to support the large volume of investments that had been planned in the public and private sectors. The government decided to concentrate its own investment effort in three major areas. They were all designed to rapidly stimulate economic activity and employment. The new administration gave foremost priority to opening up new avenues of employment. In order to rapidly build up the infrastructure for raising agricultural productivity and settling people on new land, the 1. The Sri Lanka Freedom Party or coalitionsled by the party have been in power from 1956 to 1965 and from 1970 to 1977. The United National Party has been in office from 1965 to 1970 and from 1970 onwards. Elections are due in mid-1983 at the end of a six-yearperiod.

SRI LANKA'SECONOMICREFORMS

283

government proceeded with the Mahaweli development programme, the largest scheme undertaken in Sri Lanka, begun by the preceding administration. The object was to review the Mahaweli river basin development program, which should have normally taken 30 years, into a much shorter period of 6 years by selecting for implementation three of the main projects that were in the original development program. In order to encourage private foreign investment, it decided to set up an Investment Promotion Zone which was devoid of restrictions, controls, taxes and customs duties and welcomed the opening of branches of foreign banks. It also embarked on an ambitious housing and construction program in the urban and rural sectors. The more important components of the economic measures were the relaxation of controls, the liberalization of trade, the renewed emphasis on the free market system with minimum state interference and the readjustment of the exchange rate. The preceding dual exchange rate was replaced with a unified rate and the rupee was sharply devalued in November 1977. The principal object in liberalizing the economy was to provide a more favourable climate for foreign investment because the trade and exchange restrictions that had prevailed for more than 20 years had kept away a lot of private foreign investment. Whereas many South East Asian countries had prospered in the sixties and seventies with the inflow of overseas investment, foreign investors during this period had felt that Sri Lanka had not provided the right climate. Furthermore, due to very little investment in the private sector, employment opportunities grew very slowly, thus creating an increasingly large backlog of young job seekers, most of whom had at least a secondary education. Much of the new employment generated before 1977 was in the public sector, mainly in the state-owned manufacturing and service corporations. The gem trade and tourism were the only active areas in the private sector, but these did not offer substantial employment prospects for the large numbers who were in the labour market. The liberalization of the economy, however, had unfavourable repercussions on domestic industry and some industries, notably the handloom textile industry. The light engineering sector also found it hard to survive. Nevertheless, the government continued to pay a lot of attention to agricultural development. Investment programs and large scale river valley development schemes were designed to vastly improve existing irrigation facilities and to provide hydro-electric

284

H.N.S. KARUNATILAKE

power for industry. The incentives provided for the development of agriculture which included government guaranteed credit to farmers, subsidized inputs, guaranteed prices for farm products and the freer availability of agricultural equipment and machinery all contributed to maintain the impetus that had been generated earlier to progressively increase the output of domestic agricultural crops particularly paddy yield. 3. Increase in Employment Opportunities to seek employment abroad were provided to those who could not find work at home. The Government, therefore, adopted a multi-faceted approach to solve the unemployment problem. One, it created employment opportunities as quickly as possible within the country and two, it gave options to those who were more enterprising to seek employment abroad. In the seven years preceding 1977, travel abroad was considerably restricted and this was ostensibly because foreign travel involved a loss of foreign exchange. Large numbers who were qualified and had options of getting employment abroad, particularly in the Middle East, could not leave the country because the issue of passports was restricted and various other emigration formalities had to be done. After the increase in oil prices and the resulting rise in investment activity in the Middle East, Sri Lanka was perhaps the only country, particularly between 1973 and 1977, that did not encourage its skilled workers to seek employment in he Middle East. The neighboring countries such as India, Pakistan and Bangladesh had fully exploited these opportunities. By withdrawing the exit permit system and making provision for the free issue of passports, government liberalized foreign travel and encouraged Ceylonese to seek employment abroad. This immediately tended tO relieve pressure on unemployment and it has become a very important source of foreign exchange for the country. Remittances from those employed in the Middle East have increased from year to year. Furthermore, the income distributive effects of overseas employment have indeed been very favourable. The breadwinners of a very large number of families who have found employment abroad have made it a point to remit their earnings which have helped sustain their families and dependents. In fact, it was during the years 1980 and 1981 that the largest numbers have sought employment abroad. Thus, the data in the Consumer Finance and Socio-Economic

SRI LANKA'SECONOMICREFORMS

285

Survey 1978/79 would heavily underestimate the unemployment figure. However, the Consumer Finance and Soeio-Economic Survey 1981/82 which is currently being conducted by the Central Bank will give a more accurate picture of the new employment opportunities that have resulted from the economic reforms. Before 1977, although migration for employment overseas took place on a limited scale, this had no effect on income distribution and it did not help raise income levels because it was only those who belonged to the upper class, mostly professionals, who left the island in search of employment. Their employment abroad had a negative effect on local income generation and foreign exchange reserves because not only did they retain their earnings overseas but they explored every avenue of taking the maximum amount of foreign exchange out of the country. In the case of the large number of workers who found employment in the Middle East, the converse was true. Every cent that they earned was ultimately remitted to Sri Lanl_a. This became a major avenue of augmenting Sri Lanka's foreign exchange reserves and gave individual families adequate funds not only for their sustenance but also for investment mainly in land, buildings and small industry. The major economic gains of the present administration have been in the area of investment, employment and income generation. The policy of the government has been to trade off a higher level of income, employment and production for a reduced level of subsidies and welfare payments. Before 1977, any curtailment of subsidies would have brought large numbers to the brink of starvation because incomes were not rising and new opportunities of employment were not increasing. Since 1977, the government has been committed to a policy of confining subsidies to the most needy and with this objective, it has introduced a scheme of food stamps which are given only to the poor segments of the population. About half the population today gets subsidized food and kerosene oil through food stamps. 2 In 1977, when the term of office of the previous government came to an end, unemployment had been conservatively estimated at a very high figure of 1.2 million or about 24 percent of the 2. In March1982, thescope of the food stamp schemewaswidenedbecause of the sharplyrisingcost of livin_ Now almost, two thirds of the populationare entitled to food stamps.

286

H.N. $. KARUNATILAKE

workforce. The picture has changed considerably since then. Between 1973 and 1978/79, unemployment fell from 24 percent in the former year to 14.8 percent in the latter period. Some 280,000 job opportunities were created in the public and private sectors between 1978 and 1980. This figure, of course, is conf'med mainly to the formal sector and excludes a large number of additional jobs that have been available in the non-formal private sector, particularly in domestic agriculture and in self-employment. Quite apart from new jobs for wage earners, self-employment has greatly reduced the volume of unemployment. The extensive opportunities for engaging in trade and business have given considerable scope to self-employment in production, in retail trade, or in the service sector. An estimated 125,000 persons enter the labour force each year and the rate at which new employment has increased has not only helped provide employment for the heavy backlog that existed in 1977, but has also given opportunities to a large number of new comers in the labour market. The advertisements in the press everyday indicate that for certain types of skills, there are still a large number of vacancies. Unemployment today is structural because those who a_re in search of jobs do not have the aptitudes and the skills. In the vocational and skilled categories specifically, there are still labour shortages and this is partly due to the exodus of technically qualified persons to the Middle East and other countries. The Consumer Finance and Socio-Economic Survey 1978/79 shows distinct gains in employment in practically every sector. Not only has there been a substantial broad base of employment opportunities but the volume of employment in each area, whether in new economic activities or in existing sectors, have increased significantly. In 1978/79, the total all island percentage of the employed population was 32.4. This is an increase of 6.6 percent over that of 1973. An increase of the same order of magnitude was found in all sectors of the economy and in the different districts and Zones. For the Island as a whole, the percentage of the workforce employed was 83.3 while a third was in regular employment. Another third was in self employment or in the unpaid family worker category. The rapid changes that have taken place in the economy resulting from new investments in almost every sector and the greater activity in communications, services, transportation, small industry and retail sales sectors have demonstrated that the economy has a high labour

SRI LANKA'SECONOMICREFORMS

287

absorptive capacity not only for workers with skills and initiative but also for the less educated workers who are mainly in search of blue-collar jobs. Such opportunities, in particular, have increased rapidly in the construction, transport, tourism and banking sectors. Those with less education have shown a willingness to adapt themselves to job opportunities which fall into the blue collar category. However, despite the relatively large supply of labour that has been available, wages for blue collared employees have been high because of the steady demand for skilled labour. Data from 1978/79 Consumer Finance and Socio-Economic Survey shows that 37 percent of those who have jobs were in regular employment, 26 percent had found casual jobs and 23 percent fell into the self-employed category. The ratio of those with regular employment was greater in the case of the relatively more educated groups whereas the incidence of unemployment appeared to be higher in other categories. 4. The Urban, Rural and Estate Sectors The sectoral distribution of unemployment shows that the highest rate of unemployment has consistently been in the urban sector. The 1978/79 survey gives a figure of 19.7 percent. Unemployment has been less important in the rural sector where the percentage of unemployed was 14.6. Traditionally, the urban sector has had the highest level of unemployment and this is mainly due to structural factors. In 1963, the percentage of unemployed in the urban sector was 17.8 percent, in 1973, the figure nearly doubled to 32.1 percent, and in 1978/79, it has decreased sharply to 25.7 percent. Yet, it remains above the figure in 1963. A major reason for the sharp increase of unemployment in the urban areas in the early seventies was primarily the inactivation of the private sector which in earlier periods had provided a substantial part of employment in the urban sector. Furthermore, since the bulk of private investment has been in the urban sector, the relatively slow rate at which investment took place up to 1977 tended to reduce the availability of new job opportunities. Furthermore, the controls and restrictions that had prevailed for over two decades had restricted the opportunities for business in the retail and wholesale trade, construction and transport, which have been major private sector activities.

288

H.N.S. KARUNATILAKE

The increase in employment in the urban sector has been brought about largely by the steady expansion of business resulting from the economic reforms and liberalization of 1977. Many stagnant areas in the urban sector, particularly construction, wholesale and retail trade, transport, banking and financial services, expanded considerably generating job opportunities and in turn, resulted in higher levels of investment in other sectors. The estate sector has continued to offer the highest proportion of regular employment to members of the workforce. Unemployment has always been least in the estate sector 3 where the percentage was relatively low at 5.6. The estate sector has consistently had a low level of unemployment because the estate working population consisting largely of expatriate Indians tended to diminish due to the repatriation of workers to India. Although the estate sector has a very high birth rate, work continues to be available for both adults and children because children over 14 years of age are allowed to work. Especially in the better tea plantations at higher elevations, many estates have experienced labour shortages, the absence of villages in the vicinity means that the local population have not been able to replace or supplement expatriate workers. In the estate sector, self-employed and unpaid family worker categories were almost non-existent while in the rural sector, most the employment consisted of self-employment or fell into the unpaid family worker category. Between 1973 and 1978179 in the rural sector, unemployment decreased from 24.5 to 14.6 percent. The liberal economic policies have contributed to the development of the rural sector in many ways. Greater opportunities are available today for the cultivation of a wide variety of crops, and the large new irrigation schemes particularly Stage 1 Phase 1 of the Mahaweli program have now assured cultivators in a considerable part of the Dry Zone with water _throughout the year which makes possible cultivation of two crops. It is important to note that the decrease in unemployment in the rural sector has taken place despite the migration of people from the urban to the rural sector. In the past three years particularly, there has been a tendency for people to stay in the villages or for people in 3. The estate or the plantation sector has a working population of about 1.3 million persons and consists mostly of tea, rubber and coconut plantations. The bulk of the workers on the tea plantations are of Indian origin.

SRI LANKA'SECONOMICREFORMS

289

the urban areas to move into the rural sector and the suburban areas because of the very high cost of living in the cities. The underemployment rates in urban, rural and estate sectors were 12.8, 23.5 and 33.4 percent, respectively. One contrasting feature was that the estate sector which recorded the lowest unemployment rate of 5.6 percent had the highest underemployment rate of 33.4 percent. A comparison of the degrees of underemployment between 1973 and 1978/79 shows that in the urban sector, underemployment has, in fact, decreased with the increase in the average number of days worked from 41.68 in 1973 to 47.76 days in 1978/79 during a period of 60 days. In the rural sector, it has also decreased with the average number of working days increasing from 38.59 to 41.54 days in a two month period. In contrast, underemployment in the estate sector has increased slightly with work in a 60-day period falling from an average of 40.66 to 37.10. For the Island as a whole, the number of working days in a 60-day period has increased from 39.43 in 1973 to 42.14 in 1978/79. 5. Unemployment,

Education

and Age

The survey showed that the incidence of unemployment was relatively significant among educated youths who have had formal education without any technical or specialized knowledge. The proportion of the workforce with GCE (Ordinary Level) and (Advanced Level) employment qualifications was 28 and 36 percent, respectively. On the other hand, the corresponding ratios for those with no schooling or with primary education were less than 4 percent and 7 percent, respectively. This has been the pattern even in 1973-1977 when the overall level of unemployment reached the peak figure of 24 percent. The data no doubt indicate that unemployment has been caused mainly by the inflexible aspirations of those who have received a secondary and higher education and who have invariably sought after white collar jobs in the government sector. Furthermore, the lack of technical and other skills have debarred them from certain categories of jobs that needed skills. The survey disclosed that the percentage of the workforce employed in the categories of 'no schooling illiterates' and 'no schooling literates' was higher than in any other category. But the increase in employment over the 1973 estimates in these two categories was less than 5 percent. However, between 1973 and

290

H.N.S. KARUNATILAKE

1978/79, a substantial improvement in employment was found both in those who had secondary education and those who had passed the GCE Examination. Here, the increase in employment was more than 15 percent. In the case of those who had passed the degree, gains in employment were very much higher than in any other category. One of the major problems in the period before 1977 was the large number of graduates who could not find employment. But with the economic changes that were brought about in 1977, these problems were, to a large extent, overcome and in the five year period since 1973, graduate employment increased by 11 percent. Furthermore, in the case of graduate employees, a significantly high proportion of them in the workforce was in regular employment rather than in casual employment. The unemployed classified by age, showed that there was near full employment in the higher age groups. The incidence of unemployment among youths between 14 and 25 years was still very high relative to the older age groups, even though this age group experienced a substantial reduction in unemployment. A comparsion with the 1973 data showed that the fall in the incidence of unemployment among the secondary school drop-outs and those who had passed the GCE (Ordinary Level) was 42 percent. Among those who had received a post secondary school education and among graduates, unemployment had decreased substantially from 16.2 percent in 1973 to 5.3 percent by 1978/79, which is a fall in unemployment for this category by 67.5 percent in a five year period. 6. Impact on Income Distribution The data obtained from the Consumer Finance and SocioEconomic Survey over a twelve-month period in 1978/79 shows certain significant changes in the household size, size of spending unit, number of income receivers and dependents. All these indicate that incomes, wages and employment have shown favourable trends after 1977. Over the years, the average size of the household has been decreasing from 5.75 persons in 1963 to 5.62 persons in 1973 and to 5.40 persons in 1978/79 and the decline in the average household size could be partly attributed to a fairly significant drop in the birth rate. As seen in Table 1, the average number of spending

SRI LANKA'SECONOMICREFORMS

291

units in a household had increased between 1963 and 1978/79; in 1963, the average number of spending units in a household was 1.08, in 1973, 1.05 and in 1978/79, 1.10. A spending unit consists of one or more persons who are members of a household and who share expenditure on major items. The main reason for the reduction in the number of spending units in a household in 1973 was the progressive decline in employment opportunities during this period. Table 1 -- AverageSize of Household,Numberof SpendingUnits, Income ReceiversandDependentsPerHousehold,1963, 1973 and 1978/79. Average Averagesize of household Average number of spending units per household Averagenumber of income receiversper household Averagenumber of dependents perhousehold

1963

1973

1978/79

5.75

5.62

5.46

1.08

1.05

1.10

1.56

1.44

1.63

4.19

4.18

3.83

Source: ConsumerFinance andSocio-EconomicSurvey 1978/79. The increase in spending units between 1973 ad 1978/79 shows that employment opportunities have been increasing and that there have been more income earners in each family. There has been an increase in the number of income receivers per family from 1.56 in 1963 to 1.44 in 1973 and to 1.63 in 1978/79. This is consistent with the increase in spending units. Invariably, an increase in income receivers tends to add to the number of spending units. At the same time, the increase in employment opportunities has also made a fairly significant reduction in the average number of dependents per family. In 1963, the number of dependents per family was 4.19, in 1973, 4.18 and in 1978/79, 3.83. All these figures reflect not only the growth of new income and employment opportunities but also the broad base of income earning options open to members of the households. The comparison of income distribution between the four Consumer Finance and Socio-Economic Surveys that have been

292

H.N.S..KARUNATILAKE

conducted so far in 1953, 1963, 1973 and 1978[79, respectively, become meaningful because the definitions of spending unit and income receiver remained unchanged. 4 Table 2 gives the estimate of mean and median incomes of income receivers in the four surveys examined. However, these incomes are not comparable as they are not expressed in constant rupee values. For comparability, these incomes will have to be deflated by a price index. However, in the absence of a satisfactory price index, this has not been possible. One important consideration is the sharp increase in money prices in 1973 and 1978/79. Even in 1973, there was a fair degree of inflation in the country following the sharp increase in the price of crude oil. However, in 1973, basic essential goods were available at subsidized prices and to some extent, the consumer was insulated from inflation. Between 1973 and 1978/79, mean incomes have gone up by more than 250 percent from Rs 228 in 1973 to Rs 616 in 1978/79. In 1978/79, although the country experienced substantial inflation, the worst inflationary pressures began after 1979, particularly in 1980 and 1981. In this context, the increase in incomes between 1973 and 1978/79 would have a greater degree of comparability than with the money incomes that would be recorded in the survey that is at present being conducted for 1981/82. The median monthly income between 1973 and 1978/79 rose from Rs 180 in 1973 to RS 408 in 1980. The figures in Table 2 show that the mean income increased faster than the median income moving towards greater concentration of income. One major drawback of the present economic policy package is the rising disparity between income growth and distribution.

4. However,in themost recentsurveyundertakenin 1978/79, amore rigorous definition of income receiverwas made. If an individualreceivedanincome "belowRs 30 for a month and Rs 180 for the six.month reference period, he was not treated as an income receiver. However, there was one major difference the 1978/79 survey was conducted in four rounds spread over a period of one year to enable the seasonality of incomes to be recorded unlike in the previous ConsumerFinance Surveyswhere the seasonality of income was not recorded. In 1973, the income for the two months as well as the six months immediately preceding the survey period was sought. The longer reference period of six months was retained in the current survey mainly to capture the incomesof a seasonalnature suchas incomes frompaddy farming.

SRI LANKA'S ECONOMIC REFORMS

293

Table 2 - Consumer Finance Survey Estimates of Average One Month Income (Rs) per Income Receiver, 1953, 1963, 1973 and 1978/79 Year Average

1953

1963

1973

1978/79

Mean Income

108

134

228

616

180

408

Median Income

Source: Consumer Finance and Socio-Economic Surveys. Using the Gini Coefficient it will be seen from Table

to indicate 3 that the

the distribution of income, Gini Coefficient has risen

significantly from 0.41 in 1973 to 0.49 in 1978/79, an indication of a significant increase in the inequality of income distribution. This takes the situation back to the pattern of income distribution which prevailed in 1963 when the Gini Coefficient was 0.49. The distribution of income analyzed according to deciles also shows that the income shares of the lower deciles decreased while those of the higher decfles, especially the highest, moved up significantly. Table 4" shows that the redistribution of income from 1953 to 1963 leaned towards the middle income receivers at the expense of the richest 1/10th and the poorest 2/10ths. Between 1963 and 1973, there has been a significant redistribution in favour of the lower and the middle income groups at the expense of the highest decile. Table 3 - Consumer Finance Survey Estimates of the Gini Coefficient for Income Receivers, 1953-1978/79 Year

Gini Coefficient

1953

0.50

1963

0.49

1973

0.41

1978/79

0.49

Source: Consumer Finance and Soeio-Economic Surveys.

294

H.N.S. KARUNATILAKE Table4 - Percentageof Income Receivedby Deciles All Island Percentageof Total Income

Decile

1953

1963

1973

1978/79

Lowest Second Third Fourth Fifth Sixth Seventh

1.51 3.56 3.56 4.37 5.71 6.31 7.94

1.17 2.70 3.56 4.37 5.55 6.82 8.98

1.80 3.17 4.38 5.70 7.10 8.75 10.56

1.20 2.56 3.57 4.80 5.84 7.29 9.13

Eighth Ninth Tenth

10.39 14.16 42.49

11.46 16.01 39.24

12.65 15.91 29.98

11.25 15.27 38.99

Source: Consumer Finance and Socio-Economic Surveys. The situation that prevailed in 1978/79, however, was quite different. With the exception of the highest decrie which gained 9 percent over a five-year period, all other decries experienced a reduction in their proportionate shares of total income. Taking the lowest decile, its share dropped from 1.80 percent to 1.20 percent. For the next decile, from 3.17 to 2.56 percent, the third decile from 4.38 to 3.57 percent. The lowest decrease was in the 9th decile, where the figures descended from 15.91 to 15.27 percent. An evidence of the less favourable distribution in incomes is seen in Table 5 which shows the mean income per income receiver by decile as between 1973 and 1978/79. While it is true that incomes increased in all deciles, relatively speaking, there is a much higher percentage increase in the higher than lower decries. The mean income for the lowest decile moved up to only 101 percent while in the 8th to 10 decile, increase ranged from 141, 161 and 254 percent, respectively. A remarkable mean income change is viewed in the 10th decile where the figure climbed up to as high as Rs 2,411.03 in 1978/79. It is premature at present to pass judgment upon these distribution trends particularly for the years 1978-79. These are the

SRI LANKA'S ECONOMIC REFORMS

295

Table 5 - Mean Income Per Income Receiver by Deciles All Island, 1973 and 1978/79

Deciles

Mean Income per Month (Rs) 1973

Mean Income per month (Rs) 1978/79

Percentage change from 1973 to 1978/79

Lowest

41.17

82.79

101

Second

72.03

161.04

123

Third

99.79

227.64

128

Fourth

129.67

296.65

129

Fifth

161.28

369.26

129

Sixth

198.97

452.60

127

Seventh

240.38

565.37

135

Eighth

287.88

693.83

141

Ninth

361.95

945.80

161

Highest

681.61

2411.03

254

All Island

227.50

616.19

171

Source: Consumer Finance and Socio-Economic Surveys.

years when new economic economic system and have during the initial stage of up to the higher deciles. invest their money in the

policies have been introduced into the produced results. It is a general case that economic growth, incomes tend to move This is true because entrepreneurs who economy are given financial incentives to

boost their industries which results in substantial surpluses for reinvestment in other profitable ventures. The sharp increase in income, particularly in the higher deciles, are incomes that accrue as high profits as a result of economic liberalization and the freedom to undertake whatever kind of business one wants.

TECHNICAL CHANGE AND FACTOR UTILIZATION: THE CASE OF TAIWAN, 1952-1980 Shirley W. Y. Kuo*

The rate of technical progress measures the rate of increase in output that is not caused by the increase in inputs but by the advancement in the efficiency of production. This concept has many other names. "Change in total factor productivity," "change in productive efficiency", "change in output per unit of input", "residual", etc. are the typical ones. The purpose of the present study is to analyze the rates of technical change of the non-agricultural sectors in Taiwan for the period of 1952-1980. During the last three decades, Taiwan economy grew very rapidly. But how has technical change contributed to this rapid growth? Has the contribution of technical progress played the same role in the different periods? How have the economy's leading industries behaved during the period? What role have the domestically produced and imported intermediate inputs played in the course of development? These are the main questions we would like to focus on in this study. Two types of measure will be used. They are the value-added production function developed by Solow I and the gross production function inclusive of intermediates based on an input-output

* Central Bank of China in Taipet,Taiwaa. 1. Robert M. Solow, "Technical Change and the Aggregate Production Function," The Review of Economics and Statistics, Vol 39 (Aug. 1957). 297

298

SHIRLEYW. Y. KUO

method. 2 The value-added production function is in the form of Equation (1) and the gross production function in the form of Equation (2): (1)

h(t)/(L,K)

(2)

H(t) F(L,K, xa, xm ),

value-added

labor

capital stock

gross output

where

xcl =

domestically-produced

intermediate

xm =

imported intermediate

inputs

time

h(t) and H(t)

inputs

= technical change.

Although two types of production function are used, theoretical framework will be applied on them, namely:

the same

(1) the production functions are assumed to exhibit constant returns to scale; (2) the necessary conditions for a producer equilibrium hold, so that factors are paid the value of their marginal products; and (3) quantities of output and input entering the production functions are identified with real products and real factor inputs. Data used in the two measurements are different. For the valueadded production function, the national income data are used. The manufacturing sector is to be examined only at the aggregate level, and observations are yearly series for the period 1952-1980. For the gross production function, the input-output data deflated into con2. Evsey D. Domar, "On the Measurement of Technological Change," Economic Journal, Vol. 71 (1961); Mieko Nishimizu and Charles R. Hulten, "The Sources of Japanese Economic Growth: 1955-71," The Review of Economics and Statistics, Vol. 60 (Aug. 1978); Mitsuo Saito, General Equilibrium and Price (Tokyo: Sobunshia, 1975); Tsunehiko Watanabe, "A Note of Measuring Sector Input Productivity," Review of lncome and Wealth Vol. 17 (Dec. 1971).

TECHNICALCHANGEAND FACTORUTILIZATION

299

sistent domestic constant prices are used. 3 The measurements are done for 46 disaggregated non-agricultural sectors, and observations are for the four years 1961, 1966, 1971 and 1976. Accordingly, the period of observation by this approach is 1961-1976. 1. Technical Change Assessed by a Value-added Production Function, 1952-1980 In this section, technical changes in the non-agricultural, manufacturing, and services sectors of the Talwan economy for the period 1952-1980 will be measured by a Solow type of production function. The production function in a Cobb-Douglas form can be written as where where

(3)

Y =

A o egt K" L 1-c'

value-added (net of indirect taxes)

base of natural logarithm

the rate of technical change

year

K L

capital labor

capital share

The period of observation is divided into two sub-periods, 1952-61 and 1961-80. 4 The results (Table 1) show that the rates of technical 3. Among four tables used in this study, 1966 and 1971 are originaltables and 1961 and 1976 are extension tables. The quality of the 1971 table is superior to that of 1966 as the census in 1966 is considered to be biased. All four tables are consistently deflated in 1971 domestic constant prices. Here, consistent means the elimination of the distortions caused by tariffs and import restriction, etc. p_Â˘ _ / are deflated first by nominal rates of protection at commodity level for each of the 58 sectors with some service sectors as exceptions. After this is done, domestic price indices at industry level are applied to deflate p_ _ and P_lx1_. Some service sectors and value-added terms are deflated by GDP deflator. In the deflation of price changes, 1971 is used as the base year; however, the 1971 table is also deflated by nominal rates of protection for consistency. 4. The reasons for using 1961 as the demarcation year are: a) The second monetary reform was successfullyachieved by 1961, as the

300

SHIRLEY W. Y. KUO

Table 1 - Rates of Technical Change and Their Contributions to the Growth Rates in the Non-agricultural, Manufacturing, and Service Sectors, 1952-1980. (In Percent)

Sector and Period Non-agricultural

Rates of Technical Change

5.0 3.1

1961-1971 1971-1980

1952-1961 1961-1980

Contributions by Technical Change to the Growth Rate

Sector

1951-1961 1961-1980

Manufacturing

Growth Rates of Value-added

8.3 11.1 5.1 1.3

53 12 12.2 9.9

30 9

Sector 6.5 3.4

1961-1971 1971-1980

12.1 14.4 6.4 1.1

42 7 18.1 10.5

20 7

Services Sector 1951-1961 1961-1980 1961-1971 1971-1980

4.0 3.1

6.9 9.4 4.6 1.6

51 17 10.3 8.3

34 15

Sources: Directorate-General of Budget, Accounting and Statistics, Exe_tive Yuan, National Income of the Republic of China (1980); Shirley W.Y. Kuo, "Labor Absorption in Taiwan, 1954-1971," Economic Essays, VoL 7 (Taipei: National Talwan University, Graduate Institute of Economics, November 1977); Directorate-General of Budget, Accounting and Statistics, Executive Yuan, Yearbook of Labor Statistics, Republic of China (1980).

TECHNICALCHANGEAND FACTORUTILIZATION

301

progress in the non-agricultural, manufacturing, and service sectors are all much higher in the fh,st period than in the second. Their contributions to the growth of the respective sector were accordingly much higher in the f'trst period than in the second. Technical progress explained approximately 50 percent of the growth in the first period, but explained only 12 percent of the non-agricultural, 17 percent of services, and 7 percent of manufacturing growth in the second period. It can be seen that the slowdown in the technical progress started in the 1960s, and took a change for the worse in the 1970s. In fact, the technical change in manufacturing registered a very meager rate of 1.1 percent for the period 1971-1980. The slowdown in the rate of technical progress in the manufacturing sector deserves our attention. It is caused by the fact that during 1971-1980, the capital stock grew at a much higher rate than the value-added, 14.0 percent vs. 10.5 percent. At the same time, labor grew at 6.2 ,percent, thus making the weighted average of the growth rates of capital and labor very close to the growth rate of the value-added in manufacturing. In short, the assessment by a Cobb-Douglas production function shows that the economic growth in the 1950s was largely attributable to technical progress. The contribution of technical change decreased in the 1960s. In 1971-1980, the high rate of manufacturing growth was largely attributable to the big amount of investment and rapid labor absorption, namely, more factor utilization than technical change. average annual rate of price inflation came down from 10.5 percent in 1952-1960 to 2.0 percent in 1961-1965 and 2.9 percent in 1966-1970. b) In 1961, the multiple exchange rate was abandoned, and the simple exchange rate becameeffective. c) The real wage rate, having remained nearly fLxed,beganto riserapidly after 1961. d) The rate of labor absorption into the non.agriculturalsector, having kept pace with the increase in total population, started to exceed population growthrapidlyafter 1962. e) The rate of investment in the manufacturingsector accelerated after 1961. f) The average propensity to save out of Net National Product was increased from the percentage of 4.5 in 1951-1959 to 8.0 in 1963 and 12.0 thereafter. It can be referred that a fundamentalchange in savingcapability occurredbetween 1960 and 1963.

302

SHIRLEY W. Y. KUO

Table 2 -- Growth Rates of Capital Stock, Labor and Value.added in Manufacturing, 1952-1980. (In Percent) Growth Rates of Value-added

Period

1952-1961 1961-1980

12.1 14.4

1961-1971 1971-1980

Growth Rates of Labor 2.7 6.6

18.1 10.5

Growth Rates of Capital Stock 8.7 14.8

7.0 6.2

15.5 14.0

Sources: Sameas Table 1. Note: Capitalstock is referredto capitalinuse.

In order to have a clearer look on manufacturing, we shall use another approach in terms of a gross production function to assess the technical changes of manufacturing and manufacturing subindustries for this later period. 2. Technical Change Assessed by a Gross Production Function, 1961-1976 A. The Model In this section, the rates of technical change of the non-agricultural, manufacturing, and services sectors of the Talwan economy over the period 1961-1976 are measured by a gross production function. For the model, the following notations will be used: n X

= =

the number of sectors in the input-output table domestic production

the amount of domestically produced input i that is used in the production of output j

Xi]

the amount of imported input i that is used in the production of output

, LL

the amount ,of labor used in the production of outputi _. ,: , ,. : ., ... , , .,_ , ,

xi/

: . â&#x20AC;˘

TECHNICALCHANGEAND FACTORUTILIZATION

303

a aii

the amount output j

of capital used in the production

the value share of domestically produced input i in the/th industry's gross output

the value share of imported industry's gro_s output

the labor income share in the jth industry's gross output

the capital income share in the ]th industry gross output

the amount industry

constant

the rate of technical change

year

pff

the domestic price index of output j

the import price index of output ]

of indirect

input i in the jth

taxes

paid by the /th

A gross production written as follows:

function

d ai/a

x/- Ao/e '! i= 1x,! (4)

n laff + l= _, 1aÂ˘

in a Cobb-Douglas

n i= 1

a/T

n x,7

L! r!

form can be

(/= 1, 2,...,n) + l/ + kI=1

or in logarithm form,

(5)

lnX! = InAol + gjt + i= _ 1 a_zl + lnxt_ n + i T., .m.lnx II.m.+ l/lnLi + k! InK1 = 1 a Zl

(j=

1, 2, . . . ,n)

where the term lnAo I + gjt measures the status of technology. Writing this technology term as a dependent variable In GI, we have

304

SHIRLEY W. Y. KUO

equation particular

(6) to express year:

the status

of technology

of industry

/ in a

(6)

lnGj = lnAo/ + gjt = lnX/ - i, iX__1 a.a.,,lnxff

- i=_ 1 a qm lnxiT-

lilnI.j-g/lnK/

q--- 1,2 ....

,n)

Since no yearly time series of I-O data are available, we can only calculate the rate of technical change by a discrete comp_krison. AO The growth rate of any variable Q can be expressed as _2 which is equal to AlnQ. change, we therefore have

Using g to denote

the

rate

of techQnical

g] = AlnG i n (7)

= AlnXj-

i _ 1 a.d. q lnxff-

- l/lnL i - k] InK/

_, ain_lnx /] rn i=l

(l = 1, 2,...,

Equation (7) is the one used to calculate the rate of technical change in this study. 5 The results obtained as combined twenty non-agricultural sector's observations, forj = (12 - 5 7), are shown in Table 3.

5. The double deflation technique for an I-O table is still an uncultivated area. The approach applied in the double deflation of the four tables in the Taiwan case follows the design of Dr. Larry Westphal and the late Professor T. Watanabe. Although four years' Xii .s in real terms were made and used for decomposition of sources study elsewhere (Shirley W.Y. Kuo, "Economic Growth and Structural Change in the Republic of China," World Bank, 1979, mimeo), an intensive analysis on their "residuals" may not be the most appropriate. Since a study of technology change is a study solely focusing on the change of residuals, any errors due to deflation may come up to a share, too big a weight in the change so as to obscure the true picture. Therefore, in this study, the xi] .s other than those of 1971 are estimated via equilibrium conditions of profit maximization as shown in equations (A1) and (A2). d (A1)

d a Pt

Table 3 - Rate of Technical Change in the Manufacturing Industries. (In Percent)

(Number of Industry I-0 Classification)

Rate of Technical Change

by Industry

,4 1961-71

1971-76 Z

1. (l 2+13+14+15+16+17+18+19+20)

Food, beverage, tobacco

0.9

2.3

2. 3. 4. 5. 6. 7.

Textiles and footwear Wood, furniture Paper, printing, publishing Leather Rubber Chemicals

2.6 2.8 2.9 1.6 3.4 4.3

1.7 2.1 0.6 1.1 "1.0 -0.2

Petroleum and coal products Non-metallic mineral products

0.4 3.4

1.7 2.2

10. (40) 11. (41+42+43)

Basic metal Metal products

2.2 2.5

0.9 1.5

12. (44) 13. (45+46+47)

Machinery Electrical machinery

4.7 6.0

1.4 5.0

14. 15. 16. 17. 18. 19.

(48) (49) (50) (51+52+53) (56) (54+55)

3.6 8.1 2.4 2.6 1.1

1.4 3.6 0.3 1.1 2.9

20.

(57)

Transportation equipment Miscellaneous Manufactures Construction Electricity, gas and city water Wholesale and retail trade Transportation, warehousing and communications Services

3.9 4.1

2.6 0.6

(2!+22+23+24+25) (26+27+28) (29) (30) (31) (32+33+34+36+37)

8. (35) 9. (38+39)

Â˘3

P'I

O t_

Sources: Based on recomp_led and deflated Input-Output Tables of 1961, 1966, 1971 and 1976.

306

SHIRLEYW.Y. KUO

As seen in Table 3, rates of technical change estimated by the gross production function show the same tendency as that estimated by the value-added production function: slower rates of technical change in 1971-1976 than in 1961-1971. Out of the twenty non-agricultural industries, only three industries had higher rates of technical change in 1971-1976. They are food-beverage-tobacco, petroleum-coal products, and wholesale-retail trade. Chemical industry has a negative rate of technological change possibly due to heavy investment in petrochemical industry both in the public and private sectors for the production of high-valued, new petrochemical products, yet time-lagging products. The deterioration in technical progress during 1971-1976 in the machinery, electrical machinery, and transport equipment is noteworthy, for those industries have been the leading industries in manufacturing. In Table 4, the rates of technical change are aggregated into the non-agricultural, manufacturing, and service sectors. In this aggregation, domestic productions Xj are used as weights. These aggregated magnitudes show the following characteristics: 1) The rates of technical progress 1961-1971 than in 1971-1976.

in all sectors

were higher in

m a,7 l-tj)x,

(A2)

xi/ --

m Pt

where

r, /2

ir-1 _" Pl xq +z-I = Pt xij

+ Wi + Rj + 1"/

indirect taxes paidby industry j

compensation for labor used in industry ]

compensation for capital used in industry j

In the calculation,1971 dataofa_/, aim, z1, ki and tj are used. In the calculationo_ these out elasticities,indirecttaxes are excluded. Other variables, Xi, Li, Ki, Pi, Pi, and p_/are the respective calculation year's figures. Changesin the 1961-1971 are measured based on the observations of 1961-1966 and 1966-1971.

TECHNICAL CHANGE ANDFACTORUTILIZATION

307

2) The service sector had a higher technical progress than manufacturing in 1961-1971, but the orderwas reversed in 1971-1976. 3) Heavy manufacturing always had a higher technical progress than light manufacturing. The difference, however, was much greater in 1961-1971. This was similar to Japan's case in the period 1955-1963. The United States also had a higher technical progress in the heavy industry than fight industry in 1946-1957, although the difference between heavy and fight industries was smaller than the cases of Japan in 1955-1963 and Taiwan in 1961-1971. (See Table 4 and 5). Regarding the deterioration of technical progress in 1971-1976, a few points should be noted. First, during this period, capital increased at a very high annual rate of 16.2 percent. It is our general understanding that rapid capital investment will increase productivity. However, _he situation was the other way. One possible reason is that much capital was invested in the heavy industry during this period- the typical ones were electricity (particularly in nuclear power), steel mill, shipyard and petrochemical industries. It is probable that investment was already done but outputs had not yet been fully produced. In other words, during this period, production in heavy industries might greatly lag behind investment. Second, in 1974 and 1975, the Taiwan economy experienced a serious recession due to the oil crisis and worldwide recession. The growth rates of the value added in these two years were 1.1 percent and 4.2 percent, respectively, dropping from the previous 12.8 percent in 1973. However, the number of employed did not decrease very much because the decline in demand was mostly adjusted through hours of work and wage change rather than through the number of workers. Thus, the growth rate of labor shown in the statistics was not as much affected by the recession. Third, there was a significant deterioration in the speed of development of leading industries in 1971-1976. By leading industries, we mean the six manufacturing industries which ranked in the top six as judged by the growth rates of gross output, exports and employment during the 1960s. They were electrical machinery, transportation equipment, textiles, leather and miscellaneous manufacturing. The rates of technical change of these six industries all decreased in 1971-1976. Among the six leading industries, the product share of leather was trivial, and the contents of miscellaneous manufacturing too sundry. Thus, only four industries, electrical machinery, machinery,

308

SHIRLEY W. Y. KUO

Table 4 - Rate of Technical Change in the Non-Agricultural Sectors of Talwan, ROC. (Based on I-0 measurement.) In Percent Rate of Technical Change

Sector

1951-71 Non-agricultural sector Industrial sector

2.9

Services sector

3.I

Manufacturing sector Light manufacturing Heavy manufacturing

2.9

1971-76 1.9

2.8

1.9 1.6 2. i

2.1 4.1

2.0 2.3

Sources: Same as Table 3. Note: Light manufac_ring incJndes food, beverage and tobacco, textiles and footwear, wood and furniture, leathez, basic metal, metal products, and miscellaneous manufactuzes. Heavy manufacturing includes paper, printing and publishing, rabbet, chemicals, petroleum and coal products, nonmetallic mineral products, machinew, electrical machinery, and transportation equipment.

Table 5 - Rates of Technical Change, Japan and the U.S. (Based on I-0 measurement,) in Percent Japan (1955-1963)

U.S. (I946-1957)

Sector Gross Output Manufacturing

sector

1.72

Light _anufacturing Heavy manufacturing Services sector

Value Added 1.93

0.95 2.20

1.71 2.04

4.07

Source: Mitmo Saito, General Equilibrium and Price (Tokyo: Sobunshia, 1975) p. 88. Note: The estimations are based on an Inpat-Outpat Model.

TECHNICALCHANGEAND FACTORUTILIZATION

309

transportation equipment, and textiles will be taken up for further observations. The growth rates of gross output, value added, and exports of these four leading industries are all smaller in 1971-1976 with no single exception (Table 6). We also notice the tremendously rapid expansion of electrical machinery industry in 1961-1971 and the slowdown of its expansion in 1971-1976. The relative rapidity of growth of these four leading industries can be measured by the ratio of the growth rate of each industry to the average growth rate of the manufacturing average. A significant decline in these relative growth rates in 1971-1976 shows a weakened leading force of the four leading industries in 1971-1976. The relatively faster growth of the four industries in the 1960s increased significantly their shares of gross output and value added in manufacturing during 1961-1971, from 22.7 percent to 37.1 percent in the case of gross output, and from 22.9 percent to 41.3 pereent in the case of value added, respectively. However, the relative deterioration in the growth of the four industries in the 1970s made their shares in manufacturing even smaller compared to those realized in 1971. The past success of manufacturing development in Taiwan was characterized by product cycles: t'wst, food processing, then, textiles, and then, electrical machinery and transportation equipment as the leading industry. In the early period before 1971, textile industry successfully took the place of the declining food processing. However, the evidence shown in Table 7 seems to indicate that the three leading industries, electrical machinery, machinery, and transportation equipment, which emerged in the 1960s, were not able to satisfactorily replace the outgoing old industries (including textiles) in the 1970s. Since technical progress and rapid growth of output influence each other, the slowdown of the growth rates in the leading manufacturing industries acted unfavorably to the advancement of technology in the 1970s. The inclusion of intermediate inputs in the assessment of technical change enables us to decompose the sources of output growth into the following free categories: technical change, domestically produced intermediate inputs, imported intermediate inputs, labor inputs, and capital input. Contributions by each category of these sources can be identified respectively through each term which appeared on the right hand side of equation (7). These contributions are summarized in Table 8.

t*J

Table 6 -- Growth Rates of Gross Output, Value Added and Exports of the Four Leading Industries. (In Percent) Growth Rates of Gross Output

Industry

Growth Rates of Value Added

1961-1971

1971-I 976

1961-1971

Textiles and footwear

28.9

17.0

19.7

Machinery

29.5

10.4

Electrical machinery

49.6

Transportation equipment Manufacturing average

Growth Rates of Exports

1971-1976

1961-1971

1971-1976

7.2

33.3

22.9

18.6

13.8

38.0

20.8

26.2

36.1

12.9

71.9

30.2

28.4

16.5

23.3

12.1

50.5

28.6

19.6

15.0

15.7

11.4

24.4

21.4

Sources: G_nss output and Exports: based on the _._ompfled Input-Output data, at 1971 constant prices. Value added: Dkectotate-Geaetal of Budget, Accc_mti_g and Statistics, Executive Yuan, National Income of the Republic of China, 1961, 1971 and 1976.

_t O

TECHNICAL CHANGE AND FACTOR UTILIZATION

311

Table 7 - Changes in the Shares of Leading Industries in Manufacturing. (In Percent) Shares of Gross Output Industry 1952

1961

1971

1976

1980

17.0

16.9

21.5

18.3

14.9

Machinery

1.5

1.9

2.8

2.9

2.6

Electrical machinery

0.7

1.8

9.4

10.3

11.5

0.7

2.1

3.4

4.0

6.2

19.9

22.7

37.1

35.5

35.2

Textiles and footwear

Transportation equipment Sum of the Above Four Industries

Sources: Directorate-Generalof Budget, Accounting and Statistics, Executive Yuan, NationalIncome of the Republic of China,variousyears.

The main conclusions

which

emerge from Table 8 are:

1) The contributions by technical change to the output growth for the non-agricultural sector decreased from 18.5 percent in 1961-1971 to 15.0 percent in 1971-1976. The contributions by technical change to output growth for the manufacturing sector did not change much in the two periods, accounting for about 15 percent. However, for the service sector, the contributions to output growth by technical change decreased to 22 percent in 1971-1976.

from 30 percent

in 1961-1971

2) Intermediate inputs were the dominant source of output growth, explaining about 60 percent for the non-agricultural sector and 70 percent for the manufacturing sector. The growth of the domestically produced intermediate inputs for the manufacturing use contributed about 45 percent of the manufacturing growth, while that of imported intermediate inputs, 25 percent in both periods. 3) The light manufacturing sub-sector showed a different pattern as compared with the heavy manufacturing sub-sector in that light manufacturing had a much larger contribution by domestically produced intermediate inputs than imported intermediate inputs, 52.7 percent vs. 21.5 percent in 1961-1971 and 51.1 percent vs. 18.2

Table 8 - Sources of Output Growth (Percentage Distribution, 1961-76)

Due to Technical Progress

Sector

Due to Growth in Labor Input

Due to Growth in Capital Input

Due to

Growth in

Domestic Intermediates

Imported Intermediates

1961-1971 Non-agricultural Manufacturing

sector sector

18.5

9.5

12.1

39.5

20.4

14.8

6.1

9.7

44.4

25.0

Light manufacturing

12.9

5.5

7.4

52.7

21.5

Heavy manufacturing

i4.8

5.8

10.9

38.8

29.7

29.8

23.1

17.3

23.1

6.7

15.0

11.0

16.5

38.6

18.9

14.0

6.7

11.3

43.3

24.7

I.ight manufacturing

14.6

6.6

9.5

51.1

18.2

Heavy manufacturing

13.7

7.1

13.1

36.3

29.8

21.6

31 .I

36.5

8.1

2.7

Services sector

1971-1976 Non-agricultural Manufacturing

sector sector

Services sector Souroes: Same as Table3.

Note: In aggregation, the sha_w,sof gross output of the sub-industries in the I-0 classification are used as weights.

TECHNICAL CHANGE AND FACTOR UTILIZATION

313

percent in 1971-1976. On the other hand, the contributions of the domestic and imported intermediate inputs in heavy manufacturing were much closer, 38.8 percent vs. 29.7 percent in 1961-1971, and 36.3

percent vs. 29.8 percent in 1971-1976. 4) The contribution of the growth in capital input to the output growth in 1971-1976 was much larger than that in 1961-1971, accounting for 16.5 percent vs. 12.1 percent for the non-agricultural sector, 11.3 percent vs. 9.7 percent for the manufacturing sector, and 36.5 percent vs. 17.3 percent for the services sector. The larger contribution of capital growth in the services sector was attributable to the implementation of "the ten major projects", through which a larger amount of investment was made to transportation in 1974-1979. From the above observations, we may conclude that the rates of technical change in the Taiwanese non-agricultural sectors were not the same for different periods. During the 1950s, technical progress explained about one half of the economic growth. However, it declined to about 20 percent in the 1960s and to around 15 percent in the 1970s. The high rate of economic growth in the 1970s was mostly attributable to the high rate of growth in capital and produced intermediate inputs. The implementation of the ten major projects through which a large amount of investment was made to infrastructure and heavy industries contributed greatly to this growth.

REFERENCES Domar, Evsey D. (1961), "On the Measurement of Technological Change," Economic Journal, 71. Kuo, Shirley W.Y. (1979), "Economic Growth and Structural Change in the Republic of China," World Bank (mimeographed). Nishimizu, Mieko and Hulten, Charles R. (1978), "The Sources of Japanese Economic Growth: 1955-1971," The Review of Economics and Statistics, 60. Salto, Mitsuo (1975), General Equilibrium and Price, Tokyo: Sobunshia. Solow, Robert M. (1957), "Technical Change and the Aggregate Production Function," The Review of Economics and Statistics, 39. Watanabe, Tsunehiko (1971), "A Note on Measuring Sector Input Productivity," Review of Income and Wealth, 17.

LABORING POOR OF JAPAN IN THE EARLY STAGES OF INDUSTRIALIZATION Toshiyuki Mizoguchi *

1. Introduction To break off from a dormant economic situation, most devdoping countries embark on introducing industrial concepts and techniques within their economic systems. But with the continuous growth of industries, a new and more serious problem crops up - the rise of informal sectors in large cities. In the early stage of industrialization, a large number of people move from the rural to the urban sector to seek new jobs. While this pulls down disguised unemployment in the rural sector, modern industries do not have enough abilities to offer job opportunities for all of these people. As a result, they assemble in some specific area around large cities and make up the so-called informal sector. 1 Such a pattern can be observed in developing countries in Asia, Africa and South America. 2 Japan also had such an experience before the Second World War. Since the late 18th century, the area of the informal sector increased in large cities like Tokyo, Osaka or Nagoya. While it is true slums existed before 1868, when Japan started her modernization through *Hitotsubashi University, Tokyo. The Research was financially supported by the Grant-in-Aid for Scientific Research [Type-B 54049] of the Ministry of Education, the Government of Japan.

1. The definition of informal sector varies by authors. But the term is used here only to indicatethe activitiesby residentsin the slum area. 2. An eminent survey on the researchesof informal sectors in these areas wasdone by Torii-Tsumita(1981) 315

316

TOSHIYUKIMIZOGUCHI

the Meiji Restoration, these were different from the slums at present. In this period, the Tokugawa Government segregated small numbers of people in order to appease the fret of farmers under the feudal system. Thus the informal sector had its origin in political or sociological reasons rather than economic situations. While the political segregation was officially abolished by the Meiji Government, social segregations remained afterwards. Thus, some of the informal sectors of Japan had unique characteristics by international standards. This is known as the traditional type informal sector â&#x20AC;˘ (TIS). It is also true that the informal sector had grown in the process of industrialization like the cases of recent developing countries. The slum area spread not only around the TIS, but settled in new districts in the suburbs of cities. The residents were mainly composed of migrants. In this sense, their characteristics Were different from the TIS and should be compared with those found in recent developing countries. These are to be called the new type informal sector (NIS) in the following description. Since the number of residents in the informal sectors increased in prewar Japan, the Government of Japan attempted to introduce anti-poverty policies. 3 These policies were initiated mainly for the creation of job opportunities for the residents although minimum assistance was given to those who were unable to work, particularly the physically and mentally handicapped. In other words, â&#x20AC;˘ the principles underlying such policies were based on the concept of self-reliance rather than on the modern concepts underlying the system of social securities. This line was kept not only before World War II but also in the 1950s. It was only in the 1960s that a significant development of social securities was introduced. It is very important to note that problems around the NIS have been solved gradually although crucial problems remain around the TIS in view of social justice or political considerations. Since the latter is the unique problem for Japan by international standards, a study of changes for the NIS would be of interest. If the NIS was very similar to those in developing countries, there are some possibilities that experience could supply some information for future policies. An examination of the characteristic of the NIS in detail would be in order. 3. See Mizoguchi(1982).

JAPAN'SPOORIN EARLYSTAGES

317

For this purpose, some survey data collected by central or local governments as well as private institutions are available. 4 Among these is a large-scale survey entitled the Survey on Laboring Poor, [Saimin Chosa] conducted by the Social Bureau, Ministry of Domestic Affairs, the Government of Japan, in 1911, 1912 and 1921. This survey is very important in three aspects. First, the survey covered a relatively large number of households, and had data on various aspects of sociological and economic characteristics of the poor. Second, the survey was done in the period of the early stages of industrialization in Japan. Third, the survey covered all slums of Tokyo where most of the slums were composed of the NISfi 2. Characteristics of Surveys These surveys aimed to study the general characteristics of the poor living around (not necessarily in) slums who were known as Sairain who were a little different from the absolute poor Hinmin, being included in Saimin as their component. The 1911 survey was done for two slum areas of Tokyo (Shitaya and Asakusa), while the 1912 survey covered two other slum areas in Tokyo (Honjo and Fukaya) as well as slums in Osaka. If the information in these two surveys is combined, one can obtain the data for all slums in Tokyo, although there were some restrictions for the attempt because the questionnaires were not the same for these two surveys. The 1911 survey was composed of six independent reports covering (1) household characteristics, (2) housing conditions, (3) rent, (4) labor exchange institution, and (5) regular worker's characteristics. The last category was included to compare the informal sector with other workers. One should note that the survey tried to cover all households belonging to the informal sector. This method differs from the standard approach which utilizes random sampling methods. 3,047 households were counted as those belonging to the informal sector. The report supplied tables on the number of households classified according to characteristics of household heads such as age, sex, occupation, place of birth or level of education. The data provided a past record on these households 4.Miyaji (1981). 5. The TIS is distn_outedmainly in the western part of Japan, including Osaka,NagoyaandFukuoka.

318

TOSHIYUKIMIZOGUCHI

before they located in the slums. Family budget figures were also available. However, the data were classified according to family size only. The 1912 survey was very similar to the 1911 survey but the number of households covered decreased to 2,910. The total population in the informal sector was about 1 percent of the Tokyo population according to these two surveys. Nonetheless, the surveys were very detailed, covering a large number of samples as contrasted with those in recent developing countries. The 1921 survey was rather specific in its nature. The survey was conducted in two or three areas of the informal sector in Tokyo, Osaka, Kyoto, Kobe, Yokohama and Nagoya, covering only households with two or more family members. The major contribution of this survey was the detailed family budget survey for one month, although the data also gave other characteristics of households. It is said that the figures were relatively reliable because researches were done with the assistance of officers of the Social Bureau. If the figures for Tokyo were used, one of the drawbacks found in he 1911-12 survey can be overcome. It is also convenient that the family budgets were reported in each sample, permitting a more varied classification. In this paper, the 1911-12 survey was used to examine household characteristics in the Japanese informal sector in the early stages of industrialization. However, reliance was placed in the 1921 survey regarding the analysis of income and consumption relations. Since the characteristics would change in this decade, some problems exist about the comparability of these results. But for the international comparison, this would supply some interesting results. While these reports were published independently, a convenient reprint edited by Masumi Tsuda regarding the results of Tokyo can be used. 6 3. Performance

of the Informal Sector

For the purpose of analyzing the performance process of the informal sector, the family structure is examined. According to the reports on the informal sectors in recent developing countries, family structures are different from the national average, as suggested by Torii-Tsumita. 7 The population in the informal sectors is mainly 6. The reprint of reports was compiled in a book, with introduction by Tsuda. See Tsuda (1971). 7. Torii-Tsumita(1981), pp. 22-23.

JAPAN'SPOORIN EARLYSTAGES

319

composed of economically active persons, so the 20s age group occupies a large share in the employment sector of developing countries. The cohort structures decline sharply as ages increase because of the high death rate, and the sex ratio shows imbalance. According to the review by Torii-Tsumita, the average family size of informal sectors was 9.9 in Jakarta, 6.2 in Singapore and 10.1 in Kuala Lumpur. These are much higher than the national average in each country. In contrast to these, there were relatively small differences in family structures between the laboring poor and others in Tokyo of the 1920s. The average family size of the laboring poor sector was smaller than others. While the average family sizes were 3.462 in the 1911 survey and 3.579 in the 1912s, they were 4.01 for overall Tokyo and 7.7 for overall Japan, including the rural area. These suggest that families of the laboring poor in Tokyo were an u_rbannucleus type with a relatively small number of dependents in the 1910s. The cohort structures in the informal sector also showed less significant differences from other sectors. When the averages and the coefficients of variations for the informal sector in the 1911-12 survey as well as those in overall Tokyo are calculated, a difference could be found only in the average age for the male population, although the difference is completely inverse that found in developing countries. While the average age of males was 18.31 with its coefficient of variation, 0.75, in the laboring poor, they were 16.86 with 0.65 for overall Tokyo. The corresponding results for females were 17.92 with 0.79 for the informal sector and 18.31 with 0.68 for overall Tokyo. It is not easy to explain why the demographic characteristics of the informal sector are different between Tokyo and developing countries. However, there were data which may supply some information. Two kinds of motivations for the entry to the informal sector can be considered. One is the "'pull" factor in the informal sector, which includes the high wage rate, abundance of job opportunities or relatively low costs of living. Another is the "push" factor in the native land of the population living in the informal sector. There are some researches in this respect regarding recent developing countries. According to Yap's migration function, which includes the wage rates of origins and destination as explanatory variables, the income effects do not necessarily explain the migra-

320

TOSHIYUKIMIZOGUCHI

tion. 8 Torii's t'mdings derived from questions asked directly of migrants in Thailand suggested that major motivations of migration are not much concerned with income. 9 They moved from the rural to the urban sector because of push effects including the loss of cultivated land, etc. These are called the "Nee-Enclosure" by Torii-Tsurnita. The situation seems to be different in the case of the Tokyo laboring poor. The 1911 survey asked the place of birth for household heads of the laboring poor sector, and reported that only 32 percent of them was born in the rural area while 48 percent, in the Tokyo city zone. Since the 1911 survey researched relatively old slums of Tokyo, and a new ,slum area was covered by the 1912 survey, the percentages should be evaluated with some adjustments. One could say that the major origins of slum performance were not the push effects of the rural, like the case of Thailand. Other interesting questions were presented to migrants from other prefectures than Tokyo. The replies on the motivations leaving their native land in the 1911-12 survey also support the assumption that the pull effects would have been strong. About 59 percent of household heads in the 1911 survey reported that their major object of migration was to seek better employment or to have selfemployed jobs in Tokyo, while only 19 percent of them in the 1911 survey said that they fall into some trouble in their native land. The percentages were slightly different for the 1912 survey; the former was 40.9 percent and the latter was 34.8 percent. This means that a relatively large share of population in new slums felt the push effects from their native land in comparison with the old slums, but the push effects would not be as strong as that found in recent developing countries. If such propositions could be accepted, it is not surprising that the family structures of the informal sector were not much different from other sectors. 4. Occupation and Industry Torii and Tsumita wrote in their paper, "the Sector is called as 'informal' mainly because the type of occupation and goods and services produced in the sector are indefinite". In fact, previous 8. Yap (1977). 9. Torii (1976).

JAPAN'SPOORIN EARLYSTAGES

321

studies on the slums of recent developing countries proved that the conventional standards of classification were useless in the sector. For example, a large number of workers belong to miscellaneous groups if one adopts the standard system of occupational classifications. In other words, most of the people in the slums do engage in work in the informal sector. Further, some also reported that there were traditional and institutional barriers to self-employed jobs in the informal sector. Other characteristics unique to Tokyo's labouring poor can be found. A pioneer work by Masumi Tsuda pointed out that occupational distribution was more modem than what was anticipated. 10 This paper's conclusion also supports Tsuda's suggestions. In the 1911 and 1912 survey, there were tables on the number of family members classified by industries to which they belonged. The industries can be divided into four categories in which some unique occupational distribution can be found in comparison with recent developing countries (please see Table 1). The table gives the impression that the unemployment ratio was very high in Tokyo slums: the ratios were 9.9 percent for males and 29.2 percent for females in the 1911 survey and 14.8 percent for males and 33.4 percent for females in the 1912 survey. Since the 1911 survey covered mainly the old slums in Tokyo, it is not surprising that their ratios were lower than the 1912's, However, this seems to depend on the definition of unemployment. The labor force was defined as residents above 10 years old. There was some voluntary unemployment among children from ages 10 to 15 and among those above 60 years old. The 1911 survey gave the distribution of unemployed by ages and reported that 325 among 374 unemployed males belong to the age groups of 10-15 and above 60. If these are exlcuded, the unemployment ratio for males would be about 1 percent in the 1911 survey. Although such information cannot be found in the 1912 survey, the unemployment ratio would decline sharply if the same calculations were followed as was done for the 1911 survey. According to the survey by Squire as cited by Torii-Tsumita, unemployment is concentrated in young generations in recent developing countries, but this trend is not evident in the 10. Tsuda is a pioneer worker on the examination of the surveyand pointed out variousinteresting characteristicsof the laboringpoor. This writer owes very much to his contributions.See Tsuda (1972).

322

TOSHIYUKI MIZOGUCHI

Table 1 - Occupational Distribution of Family Members in Tokyo Slum in 1911-12 1911 Survey Male Female

1912 Survey Male Female

Manufacturing Formala Informal

499 795

302 1,187

874 521

1,226 728

697 1,523

714 373

494 1,056

64 246

374

1,062

512

I, 135

3,820

3,638

3,457

3,399

Others Formal Informal Unemployed b Total (above 10 years old)

aRegardingthe definitionof the formal andthe informal,see text. bWhilethe numberof unemployedis obtainedfrom the detailedtables for the 1911 survey,the figuresfor the 1912 survey are calculatedon the assumptionthatthexe wereno child-employeesunde_10 yearsold. Source:Tsuda(1971).

surveys. 11 Considering these together, the employment situation was much better in the 1910's Tokyo slums than recent developing countries', at least for males'. The unemployment ratio also declines for females when the age groups of 10-15 and above 60 are excluded, i.e., from 29.2 percent to 16.4 percent in the 1911 survey. But one cannot deny that the ratio was high even after this adjustment although one can infer that there was some voluntary unemployment for females. However, it should be noted that the unemployment ratios were nearly equal for the generations from the 20s to 50s for females. The next important topic is to examine whether family members in slums could find their jobs in the formal sector or were forced to remain in jobs in the informal sector because this might be the key to the solution of the poverty problem. For this purpose, jobs were classified into the formal and the informal. Such an attempt has been done by various authors, including the ILO missions for developing 11. Squire (1979) cited in Torii-Tsumita (1981).

JAPAN'SPOORINEARLYSTAGES

323

countries. Among them is the proposal by Sethuraman. 12 Jobs of family members can be classified into manufacturing and nonmanufacturing, and further divided into the formal and the informal. It is rather easy to classify the nonmanufacturing jobs. The following occupations are defined as informal: rickshaw drivers, unskilled construction dairy workers, petty traders, low grade artists, domestic servants and miscellaneous service workers. Since the data give only the kinds of occupation for the manufacturing industry, one cannot identify which occupation belonged to the informal. Thus, one can classify the work, in considering the general situation in the period, with the use of various kinds of information. For example, the jobs in the food processing industry are defined as informal because most of the work concern the small-scale rice mill industry. The formal sector includes metal, machinery, cotton and fiber, chemicals, printing and publishing and glass-producing industries. In Table 1, it is very impressive that the share of the manufacturing industry was very high: 41.3 percent for males and 71.1 percent for females. One should note that there were some documents which pointed out that major occupations of slum people in the 1880s consisted of jobs in the informal sector, like rickshaw drivers, ragpickers, dairy construction workers and petty traders. 13 The situation was very similar to that found in recent developing countries. However, industrialization in the late 19th century had changed the nature of slums. First, new industries absorbed young laborers who came from the rural areas and thus prevented the explosion of informal sector. Second, young generations in slums could find their jobs in the formal sector outside the slums. For example, the fiber and cotton industry presented job opportunities for females although their wage rate was very low. This explains a relatively large share of formal sector employment for females in Table 1. Third, owing to industrialization, there arose some self-employed jobs belonging to the informal sector of the manufacturing industry, like willowing. It is very interesting that the share of the formal sector is lower among household heads than among other family members. This suggests that some families would remain in the slums mainly because their household heads engaged in work in the informal sector although other family members could find jobs 12. Sethuraman (1977). 13. See Yokoyama (1949) as well as other documents cited in Tsuda(1972).

324

TOSHIYUKIMIZOGUCHI

outside the slums. Thus, new generations could leave the slums in the formation of their own households. The next topic concerns working conditions: i.e., the wage and labor hours. While the 1911 survey gave the tables classified by sexes and industries for the distribution of number of household members by monthly income and labor days (not hours), the 1912 survey showed their averages without any distribution data. Only the results of the 1912 survey are shown here because the following results are supported by the 1911 survey. According to the figures in Table 2, there were few differences in income between the formal and the informal sectors if one excluded that for females in the formal sector which depended on the low wage in the cotton and fiber industry. In addition, the levels of income in Table 2 were as high as those of regular workers outside the slums as compiled in the 1911 survey. It may be true that the workers covered by the 1911 survey belong to a relatively low income group in Tokyo, but it is very surprising that the family members in the slums had obtained the same levels of income. Tsuda pointed out that the income level of the rural classes in the early 1910s was as high as the low income classes in the urban. 14 These would explain why the "push" effects were not dominant in the migration in Japan. The labor days in the slums seem to be normal in the sense that they were similar to the regular workers in that period, although the conclusions may change if one could obtain data in working hours. 5. Income and Expenditure

Relations

As was noted in Section 2, reliance was placed on the 1921 survey regarding the income and expenditure relations because the survey had detailed family budget data on a household basis. However , it is a good policy to show here the figures shown in the 1911 survey as references. The 1911 survey gave the distribution of the number of households for expenditures on food, clothing, housing and miscellaneous items in each classified by family size. Since one can obtain the average of these expenditures by groups of family size, one can calculate the total consumption expenditures as the sum of these categories. The average monthly consumption expenditure per household was 20.30 Yen in the 1911 survey and 63.74 Yen in the 1921 survey. 14. See Tsuda0972).

JAPAN'SPOORINEARLYSTAGES

325

Table2 -- Avefap Income andLaborDays Pet Monthin 1912 Survey Income ('4/Month) Male Female

Income (Day/Month) Male Female

Manufacturing Formal Informal

13.4 12.9

3.3 3.7

25.7 25.0

25.1 24.4

Others Formal Informal

12.8 12.1

3.9 3.8

28.7 24.1

24.2 22.8

Total

12.7

3.7

24.5

24.6

Owing to the inflation caused by the World War I, consumer prices rose significantly between these two surveys. According to the calculation by Tsutomu Noda, the CPI rose from 100 to 213.2 in this decade. 15 Thus, the real average consumption level of the 1921 survey was 29.89 Yen in 1911 prices, which was about 50 percent higher than that of the 1911 data. But this does not necessarily mean that the consumption level of the informal sector rose significantly. First, while the 1911 survey covered all households including single member households whose income was generally low, the 1912 survey researched the households with two or more family members. Second, since the 1921 survey selected representative samples from all households and requested them to make the detailed family account books, one can infer that representative samples might be selected from the ones where household wives were relatively well educated. Third, it is generally known that downward biases can be found in the family budget surveys which use simple questions on expenditures in comparison with those adopted in daily family account books. Considering these, the differences in consumption levels between these surveys overestimate the rise of real income of households in the informal sector. First, the income distribution should be studied. The summary 15. See Noda'sestimatescompiled in Ohkawa(1967).

326

TOSHIYUKIMIZOGUCHI

table in the 1921 survey report gave the distribution of the number of households classified by monthly receipts. Since in the definition of the survey, 'monthly receipts' included the carry-over from the previous month and new debts in the month in addition to income, the table can be used only as an approximation. But the biases would not be so large as to affect seriously the conclusion. As shown in Table 3-A, the number of households concentrated in monthly receipt groups which were a little higher than the average receipts, 72.26 Yen. About 60 percent of households belonged to the groups from 50 to 80 Yen. Unlike the usual form of income distribution, it was symmetric or skewed a little to the right side. It is important to compare the distribution with the income levels of workers outside the informal sector. For this purpose, one can refer to an interesting survey called the Tsukishima Survey conducted in 1919-20 by the Bureau of Public Health, Department of Domestic Affairs. 16 This was promoted by Iwasaboro Takano, who was one of the pioneer workers in this field in the world. The survey researched social and economic characteristics of households of weli-tralned laborers living in the Tsukishima area of Tokyo. While the survey had various kinds of information, only the average income of these households was used here. Since the CPI in 1919 was neatly equal to that in 1921, one can compare the income level of well-trained laborers to the one of households in the informal sector: the average income level in Tsukishima survey was 72.92 Yen which is 22 percent higher than that of the informal sector. It appears that the differences are smaller than anticipated. One can also obtain another kind of income distribution data classified by disposable income per adult equivalent family member which were calculated through the reclassification of individual household data. In Table 3-B, distribution is shown in 10 income groups. While the number of households in the highest income group is smaller than other income groups, the data could be used as if the decile group data is in a broad analysis. It is interesting to note that the coefficient of variation of income is much smaller in Table 3-B than that in Table 3-A. This means that the standard of living was very similar among the households in the informal sector. One should note that about 75 percent of households in the informal sector had 16. Bureauof Public Health (1921). This report was reprinted in Oass/c Book Series on People's Livings, by Koseikanpublishedin 1971.

JAPAN'S POOR IN EARLY STAGES

327

the surplus in the income-expenditure relations and that above I0 percent had the higher household income than the average of welltrained laborers in the Tsukishima Survey. The situation seems to be different from the experience in recent developing countries where the income differences are large between the formal and the informal sector. It is very interesting to compare the composition of monthly receipts. Naturally, the primary income of household heads was the most dominant part. According to the Douglas-Long Law, the supply of labor of other families is determined by the level of household heads' income. 17 Thus, it is interesting to take the ratio of various receipts to the income of household heads as is shown in Table 4. While the ratio of wage type income of wives is relatively stable, that of 'other' income rises in the low income group, then declines Table 3 - IncomeDistribution Data of Tokyo Informal Sector in 1921 (A) Data by Monthly Receipts Range of Monthly receipts (yen)

Number of Households

Family Size

Monthly Receipts (yen)

Disposable Income (yen)

Consumption Expertditures (yen)

Saving Ratio (%)

3.7

28.03

25.61

26.30

- 2.7

30-

3.5

35.69

25.48

33.31

-30.7

40-

3.8

45.66

40.62

41.13

1.3

50-

4.2

55.03

48.40

49.07

1.4

60-

111

4.3

64.85

56.57

55.30

2.2

70-

4.5

74.24

64.86

59.56

8.2

80-

4.5

84.91

70.40

67.86

3.6

90-100

4.7

94.97

72.32

67.60

6.5

100-120

4.9

109.30

82.04

77.66

5.3

120-150

4.5

131.31

80.61

78.88

2.1

6.6

186.65

143.92

124.92

13.5

497

413

72.26

59.45

57.42

3.4

150Total or Average

17. See, for example, Long (1958).

328

TOSHIYUKI MIZOGUCHI

(B) Data Per Adult Equivele_t Disposable Income Group Number

Number of Households a

Number of Adult Equivalents

Per Capita (yen) Monthly Disposable ConsumpReceipts Income tion Expenditures

Saving Ratio (%)

3.71

15.42

9.78

13.42

-35.4

3.49

17.89

13.48

14.76

9.5

3.34

19.94

15.23

16.35

6.7

3.56

19.92

16.94

16.86

0.5

3.13

22.28

18.62

18.13

2.6

3.06

23.10

20.47

19.01

7.1

2.99

26.32

22.25

21.19

4.7

2.69

28.48

25.25

22.06

12.6

2.51

31.74

27.82

25.20

9.4

2.44

39.62

33.84

28.76

15.0

awe exclude twohouseholds whosefgures seem,to beunxeliable inthecalculation of Table 3-B. Source: Tsuda (1971).

afterwards as the household income increases. Since the major part of 'other' income is the wage type income of family members other than household heads and their wives, one can consider that this represents the pattern of labor supply. In low income households, the family members were forced to engage in jobs even if the wage rate was low. But in the high income groups, the family members went to work if they found the jobs in the formal sector. In this respect, the Douglas-Long Law could be applied. In fact, when one calculates the ratio of family members' income to the number of other family members with jobs, it increased sharply as the income of the household head rose. These suggest that the households of the lowest 15 percent or 20 percent of income groups in the informal sector belonged to the absolute poor group, but the others could afford their living to some extent. This is consistent with the previous suggestions from the income-expenditure relations. Further, it is important to note that the ratio for debts is small, above a half of households.

JAPAN'S POOR IN EARLY STAGES

329

Table 4 - Ratios to Income of Hoasehoid Beads by Kinds of Receipts (%)a Group Number

Wage of Wives

Other Income

Carry Over

Debts

Totalb

6.45

7.14

22.45

30.50

166.36

5.58

11.29

16.45

20.51

154.03

6.63

4.70

16.42

12.01

139.76

7.11

18.08

12.73

7.02

144.94

7.93

14.12

13.84

5.97

141.86

4.86

7.05

7.31

5.98

125,11

6.79

9.98

10.00

9.80

136.57

3.83

4.41

8.06

3.10

119.40

6,34

5.15

9.31

4.68

125.48

5.93

6.70

8.37

4.72

125.27

aTableis calculatedfrom data by dispo,ableincome per adultequivalentfamilysize& bTotal includesthe income of householdheads.

The 'other too

small

income'

to be treated

includes

the transfer

as an independent

income item

but its share was here.

Further,

the

expenditures for remittances were also small in the case of the Tokyo informal sector. Some recent reports inform that a large number of households in the urban informal sector send their money to the rural areas in recent developing countries. 18 In this sense, the informal sector in the 1920s Tokyo had a high degree of independence in comparison with recent experiences in developing countries. Finally, considering the composition of consumption expenditures shown in Table 5, the income elasticities calculated from the regression equations for individual data were added. It is well known that the Engel coefficients - the ratio of expenditure for food to the total consumption expenditures - is one of the important indicators of the standard of living. One can identify a household as poor when 18. Two papers are cited in Torii-Tsumita (1981). They are Bienefeld-Sabot (1971) and Jhonson-Whitelaw (1974).

330

TOSHIYUKI MIZOGUCHI

its Engel coefficient in _ broad study is above 60" percent. In this respect, about a half of households in the informal sector were in the poor situation. Further, Oshima suggested that the coefficients do not decline or even increase as income rises in the absolute poor group of households. 19 Table 5 shows that the coefficients were relatively stable from group (1) to (6). This result shows a larger share of the poor in the informal sector than that indicators. Judging from the income elasticities, the clothing were the most luxurious items. This is because the miscellaneous expenditures arc the most usual household groups, including those shown in Survey. These suggest that the cultural activities among the informal sector.

shown in other expenditures for very interesting luxurious in the the Tsukishima were restricted

Table 5 - Composition of Consumption Expenditures (%)a Group Number

Food

Accommodation

Clothing

Fuel & Light

Mi_ellaneous

64.5

7.7

4.0

8.8

15.0

62.4

8.4

4.6

8.4

16.2

60.9

7.8

6.1

8.8

16.4

61.2

8.6

5.7

8.3

16.2

61.1

9.3

6.9

7.1

15.6

62.8

7.9

6.4

7.4

15.5

58.3

7.2

8.2

7.2

19.1

58.7

8.5

8.6

7.7

16:5

57.7

7.5

9.4

7.1

18.3

54.1

9.0

13.8

6.5

16.5

0.795

1.825

Income Elasticities b 0.553

1.124

aSeeNote a of Table4. bThe income elasticitiesare calculatedwith use of regressioncoefficients of linear Engel functionsestimatedfrom individualhouseholddata.? meansthe regressionequation wastoo unstableto be usedhere. 19. See Oshima (1977).

JAPAN'SPOORIN EARLYSTAGES

331

6. Final Remarks The discussions in this paper could clarify the similarity and differences in household characteristics between the informal sectors in the 1910s and 1920s Tokyo and recent developing countries. Like recent developing countries, Japan suffered the explosion of the informal sector in the early stages of industrialization. However, Japan had solved the problem gradually through the process of industrializatio_ although there appeared some serious situations in the early 1930s when the Japanese economy fell into a recession, owing to the world economic crisis as well as in the mid-1940s when the Japanese economy suffered heavy damage from the defeat in the Second World War. This experience seems to give a bright perspective for the future of developing countries. In this respect, it is important as an actual problem to examine the causes of the differences of characteristics mentioned above. While there is limited information on the conditions of the informal sectors in the 1980s or 1890s Japan, one can infer that they were as serious as those found in recent developing countries. Most of the people in and around slums engaged in work in the informal sector, and their standard of living was much lower than that found in other areas. However, as years passed on, the relative income of the people in the slums had gradually risen and approached the relatively low income groups in the formal sector. In the late 19th century Japan, the rural sector had taken off from self-sufficient economy, and its income level was low, but was not too low to push out family members into the informal sector. Thus, the inflow of population to the informal sector was not remarkable after 1880s, although a significant inflow was found in the 1860s and 1870s owing to the economic and social confusions caused by the civil war, i.e., the Meiji Restoration. This was one of the differences of the Japanese experience from those of the recent developing countries. In this respect, one can say that the problems around the informal sector are the shadow of rural development problems. Second, one should note that the educational level of the people in the informal sector was relatively high. Since the late 19th century, the Government of Japan had tried to spread education at the primary level. These efforts were also emphasized for the slum areas. The literacy rate for the above 15 years old was about 85 percent for males and above 50 percent for females, according to the

332

TOSHIYUKI MIZOGUCHI

1911-12 survey. The rate was much higher if one restricts caiculation for young generations. Thus, most of the young people the slum area became the relatively high quality labor to employed in the informal sector. This suggests that the role education is very important in solving the problems around informal sector.

the in be of the

Third, one should note that Japanese central and local governments focused their policies to upgrade the standard of living in slum areas through self-reliance of these peoples, rather than the removal of slums. Their major policies were to supply the job opportunities with settlements of labor exchange institutions, some financial aids to improve housing or sanitary conditions in addition to the spread of education as mentioned above. It was only after the mid-1960s when Japan passed the turning point from a labor surplus to a labor scarce economy, that some local governments including Tokyo moved to eliminate slums. The experiences should be refered to when one considers the slum problems in recent developing countries.

REFERENCF._ Bienefeld. M. A. and Sabot, R. H. (1971), The National Urban Mobility, Employment and Underemployment and Income Survey in Tanzania, Dar es Salam University. Bureau of Public Health, Department of Domestic Affai_ (1921), Field Survey on Laborer's Households in T_ukishima Area of Tokyo City, [Tokyoshi Kyobashiku Tsukishima niokeru Jitaai Hokoku]. (In Japanese). Jhonson, G. E. and Whitelaw, W. E. (1974), "UTban-Rural Income Transfers in Kenya: An Estimated Remittance Function," Economic Development and Cultural Change, 22 (3). Ohkawa, K, and or. al. (1967) "Prices" [Bukkal, Long Term Economic StatiJffcs of Japan, 8, Toyo Keizai Shimposha. (In Japanese with English summary). Oshima, H. T. (1977), "Some Notes on Defining and EstimaKng Poverty Levels," Mimeographed Paper of the University of the Philippines. Long, C. (1958), The Labor Force under Changing Income and Employment, National Bureau of Economic Research. Miyaji, Mikio (1981), Bibliography on Researches of Poverty Problems in Japan, [Wagakurfi no I-Iinkon Mondai Kankei Bunk Mokutoku], IADRPHU. (In Japanese). Mizoguehi, T. (1982), "Anti-Poverty Policies and Social Securities in Japan: 1880-1980," Discussion Paper, IADRPHU.

JAPAN'S POOR IN EARLY STAGES

333

Sethuraman, S. V. (1977), "The Urban Informal Sector in Africa," International Labor Review, 116 (3). Squire, L. (1979), "Labor Force, Employment and Labor Market in Course of Economic Development," World Bank Staff Paper (336.), Torii, Y. (1979), "Labor Market in South East Asian Economic Development," Azia Keinzai, 19 (5). (Japanese). Torii, Y. and Tsumlta, Y. (1981). "Explosion of Informal Sector in Economic Development," Working Paper, Economic Development Project in Keio University. Tsuda, M., Reprint of Social Bureau, Domestic Affairs', Tables of Survey of Laboring Poor, [Saimin Tokeihyo], Keiso Shobo. (In Japanese). , Japanese Low Income Society, [Nihon no Kaso Shakal], Minoruba Shobo. (In Japanese). Yap, L. Y. L. (1977), "Attraction OfCities: A Review of Migration Literature," Journal of Developing Economics, 4 (4). Yokoyama. G. (1949), Low Income Classes in Japan, [Nihon no KasoKalkyff], Chuo Rodo Gakuen, [originally published in 1888]. (In Japanese).

AN ANALYSIS OF THE PERSONAL CONSUMPTION EXPENDITURES IN JAPAN, 1892-1967 Konosuke

Odaka*

1. Introduction The purpose of this paper is to present findings of a statistical analysis of the personal consumption expenditures in Japan during her experience of modern economic growth.1 It covers approximately five decades, omitting the extraordinary experience of World War H and the reconstruction period immediately following the War (1939-1953). The study represents an attempt to make an extensive use of the newly estimated historical statistics of the Japanese economy, commonly referred to as the LTES series. 2 The personal consumption data are based on Professor M. Shinohara's estimates and the national income data, on Professor K. Ohkawa's. Obviously, the scope of the present investigation is limited. It explores the rationale of the Japanese consumption behavior from a macro-economic point of view, while leaving untouched a detailed analysis of the pattern of consumption demand, either by commodity or by type of households (e.g., workers' vs. farmers'). *Hitotmbashi Untver_y, Tokyo. In undzrt_',__ tl_ lmm_t research, the wdtea was tmMflt_ greatly from the encottragemcnt and com_tmcti_ criticisms offered by Messrs. K_ush/Ohkawa, Kaguo Sato, Hugh Patrick, Ysmshi Toda and Edward Lincoln. Howevea, an their commemt_ have been incorporated in the pmseat vernon of the pap_. Necdlms to say, the wfil_r alo_ k respon_lÂ˘ for any temm_iug eum_.

1. Kuznets' concept. See Kuznets (1966, ch.l). 2. LTI_ stands for Esthnates of Long-Tom Economic Statistics of Japan since 1868, 14 vohnnes, edited by Kazus_ Ohkawa,lV[iyoheiShinoharaand MatajiUmcmura(Tokyo: T6yb Kcizai_ Sha, 1965-). 335

336

KONOSUKEODAKA

Nevertheless, it will prove to be of value in drawing some analytical observations; for one thing, there have been in the past only a handful of analyses of long-term consumption expenditures of the country, such as Gleason (1965), Ohkawa (1970) and Yoshihara (1977). It has been pointed out frequently that the Japanese savings ratio is exceptionally high compared with other major industrialized nations (e.g., Mizoguchi 1970, oh. 5). This is true not only from cross-sectional observation, but also from a comparison of historical records. The fact is born out easily by Figure 1, which relates annual real disposable incomes per capita in three representative countries to corresponding average propensities to consume (hereafter abbreviated as APC). A slight inverse association may be detected for all cases. In particular, note the wide range of the APC values for Japan in marked contrast to those for the other two countries. In the latter, APC has always been higher than 0.90 with only a single exception, the United States in 1941. By contrast, APC in postwar Japan (1954-1967) is as low as 0.84 on the average. While the figure is much higher for prewar Japan (0.92), it still belongs to the lower end of the range by the international standard. Moreover, it seems suggestive that the six prewar incidents where APC registered especially low values (less than 0.85) all correspond to the years of relatively high output growth in real terms. In explaining the unique pattern of consumption (or of savings), one may conceive of two possible approaches: behavioral and structural. The former, which may in turn be subdivided into two groups, seeks an answer to the difference in human behavior. On the one hand, one may argue that the behavioral criterion, such as utility maximization, differs from one country to another. Alternatively, one may regard the difference as a matter of degree, so that it may be reduced to the difference in functional specification or in parametric values of the model, while postulating the applicability of a uniform maxim. The second methodology emphasizes the importance of structural consideration in analyzing the economic behavior of a developing economy, since changing compositions of either consumption goods or of occupational structure are likely to influence the aggregate consumption expenditures. The components of consumer demand undergo constant transformations from one period to another, owing to changes in relative prices, shifts in personal tastes, and alternation

A_.Pm,ope_dty

I I_

" "0" " "

xxl_°x. o

•

OJO"

C/3 0

•

®_, • •

o • e

e • O J)

0 '_

•

(lOel-m_) (IOS_-Ie7_

U.K.

(I'I_Q-INS,1946-

U.9.

1W29-1941,1947-1_N_)

KHSB)

GO0

Jl.

, JOJ9 19:15

¢3

o1,41 KImy:

•

1sty

_' _I_

l'_ ,-1

D Per_._,'_I_mom IO_ ipeeeMelnceme k(_n Hle_,leeoym) ote: The data for both the U. K. and the U. S. have been converted (average for 1934-36)_as

reported in Starbtical

Abstrac¢

to yen by using the foreJ_

exchange

of the U.8., 1937, p. 272.

rates at New York City _o -..I

FIGURE AVERAGE

PROPENSITY

TO CONSUME IN THREE COUNTRIES

338

KONOSUKE

ODAKA

of old and new products. By contrast, the persistent demand for cheap, indigenous food and clothing in Japan has been highly instrumental in keeping the average consumption spending at a relatively low level (Rosovsky and Ohkawa 1961 ). Moreover, changes in occupational structure may affect the aggregate economic behavior to a significant degree. For instance, family budgeting in certain households, such as tenant farmers' and small shopkeepers', cannot be easily separated from business accounting; consequently, its behavioral pattern may be distinguishable from that of ordinary households. More specifically, the f6rmer may display a higher saving ratio than the latter, although such is not necessarily the case in post-World War II Japan. The present study is conffmed to an examination of the characteristics of macro consumption patterns over time. It will be assumed throughout that the identical set of functions may be meaningfully proposed for both Japan and other representative economies. There is some justification for this choice of methodology. To begin with, the contemporary economic historians generally agree that Japanese capitalism took root after the Sino-Japanese War. Put another way, the period chosen for the present analysis matches approximately that of modern economic growth, as suggested by Professors K. Ohkawa and H. Rosovsky (1965). Therefore, unless clear evidence is discovered to suggest to the contrary, there seems no a priori reason to postulate that Japanese households have been subject to irrational behavior, which is at variance with the universal economic principle prevailing in other industrial nations. On the other hand, the macro analysis is not suited for the analysis of structural change. Important as it is, the structural analysis has to be conducted by a separate research project of a much larger scale. In the following pages, some estimates of the personal consumption function will be presented, after considering several alternative formulations of the model. In addition, a few intercountry comparisons will be carried out in order to determine the extent to which the Japanese experience may be considered unique. In the final section, some concluding remarks will be offered. 2. The Models It is beyond the scope of this study to conduct comprehensive

JAPAN'SPERSONALCONSUMPTION EXPENDITURES

339

tests on the relative strength of competing theories of consumption. The choice of the models has therefore been dictated not only by their theoretical appeal but also by their ease of handling and the availability of relevant data. Of the representative economic models of consumer behavior, the following six have been chosen as likely candidates for use in the present investigation: A. Absolute income hypothesis B. Hypotheses allowing for variable preference field (1) Relative income hypothesis (2) Habit formation hypothesis C. Hypotheses not allowing for variable preference field (1) Permanent income hypothesis (2) Wealth effect hypothesis D. Liquidity-asset holding hypothesis. According to the Keynesian consumption theory, personal consumption expenditures at year t (Ct) is a linear function of the absolute level of personal disposable income at the same year (Yt) so that (1)

Ct=_+fJYt(cL>O,

1 >_>0).

This implies that as APC becomes smaller, the greater the level of income becomes. The data found in Figure 1 seem generally consistent with the expectation. Furthermore, an international comparison of historical experiences by Kuznets ( 1966, pp. "248-250 and 262-264) suggests that the APC's are by no means constant over time. Consequently, the above equation may still claim some validity, despite the classical discovery that the proportion of aggregate gross savings in GNP has been remarkably stable in the United States (Friedman 1957). A version of the relative income hypothesis may be expressed as Ct/Yt = f(Yt/Yo), where Yo stands fur the highest income level previously attained. Unfortunately, however, this functional form is not suitable for postwar Japan through the 1960s, for the level of her current income has constantly renewed its past record. A habit formation hypothesis may be represented by a specification of the form (2)

Ct =Â˘x+_Y t +'YCt_ 1 (1 >B>O,

1 >'y > 0).

340

KONOSUKEODAKA

This hypothesis has an intrinsic attraction, especially because of the relative importance of indigenous components in the Japanese consumption pattern. Moreover, it should be noted that the equation is essentially equivalent to an empirically tested version of the permanent income hypothesis, where permanent income is estimated as a weighted sum of past income streams. A variant of the equation (2) may be (3)

Ct = ot+ #oYt+ #1Yt_ l,

which can be derivedby combiningtheabsoluteincome hypothesis and an assumptionthatthe coefficient /3in the equation(I)isa linearly decreasing functionof thegrowthrateof Y sothat# --#0 + #1 - #I(AYt/Yt), where AY t -- Yt - Yt-1.3 This version of the hypothesis has an affinity to the view by Fujino (1972, p. 121) that the consumers' time preference is dependent on the rate of growth of real income. A simple-minded wealth-effect hypothesis may be expressed by a formula Ct = ot + #Wt + 3,Yt, where Wt stands for the wealth owned by the household at time t. By definition, Wt = 141 t_ z + Yt-1 Ct- 1, so that the above equation may be rewritten to yield (4) Ct=[3Yt_ 1 +(1-l_)Ct_ 1 +'rAY t. Note that (4) has no intercept and that the coefficients of Yt- I and Ct-1 add up to unity. However, it is also possible to interpret its generalized form, having a non-zero constant and no constraint on the coefficients, as a version of the lagged-income hypothesis. It is noteworthy that functions of types (2), (3) and (4) have a common property. Assuming, for instance, that Y grows at a constant rate (r) and that APC has reached a constant value (APC*), one can easily find that equation (4) yields APC* = (33+ 7r)[([3 + r) and dAPC*/dr = (7 - 1)#/(r + #)2<0 (since "r<l and /5>0). In other words, they too are consistent with the observation (in Figure 1) that the growth rate of income is negatively associated with APC. Finally, the last set of hypothesis asserts that an increase in asset holding will result, other things being equal, in a relatively high value of APC. By construction, this hypothesis seems more appropriate for structural analysis. In the present investigation, the liquid-asset 3. I owe this formulationto Brinner(1973, pp. 10).

JAPAN'SPERSONALCONSUMPTIONEXPENDITURES

341

holding hypothesis has not been subjected to empirical testing due primarily to the difficulty in procuring the necessary data. The unit of measurement presents some problems in actual estimation procedure. The subsequent analysis is based on real per capita data, where both consumption and income are divided f'trst, by the size of population and then by the consumer price index, which takes years 1934-1936 as the base. 4 For a commentary on the nature of the basic data, the reader is referred to the Appendix to the present paper. After a process of elimination, the following six functions have been selected for the analysis of the Japanese consumption behavior. Using small-case letters to stand for per capita real values and omitting the subscript t, the fitted regressions are: (5) (6) (7)

c = _1 + fllY + e c = a 2 + r2 Y + 3'2Y- 1 + e c = a 3 +/33y + "r3c_ 1 + e

(8)

CffiOt4 + fl4y_ 1 + _Â˘4C- 1 +_4 Ay + e

(9) ( 1O)

c=flsy_ 1 +_sc_ 1 + 8sAy+e c/y = a 6 + r6 (c_ 1[Y- t ) + ")'6(&v/y) + e

Among these, equations (9) and (10) may be interpreted as variations of the permanent (or expected) income hypothesis. As usual, e represents the disturbance term. 3. Results of Estimation The ordinary least squares method of estimation has been applied to the data, ignoring the possible repercussions of c on y. The results of the estimation are reported in Table 1. Clearly, equation (10) is inferior to others and consequently omitted from the subsequent discussion. Both equations (5) and (6) display not only the existence of significant serial correlations but also relatively poor performance in forecasting. From a purely technical point of view, therefore, the remaining three formulations (7), (8) and (9) seem preferable to others. In other words, it is def'mitely advisable to incorporate a 4. Ideally, one shouldconstruct time.seriesdata of standardizedconsumer units rather than using population. The use of real quantities presumes, of course,the non.existenceof money illusionin the long run.

t_ 4_ to

Table 1 - PersmmlConsumption Funclioa, lapin (1892-1938 and 1954-1967)

Equation Number

Dependent Variable

Constant

_2 Y

7 8

18.42

(93.57) 0.4120 (5.62) 0.3814

(2.86) 3.839

(9.95)

(2.03) C

0.857 (18.76)

Y-I

c-1

(c/y)_ 1

_y/y

0.7833

(9.66) 15.33 (8.95) 4.873

Forecastingc Capability (%)

Independent Variablesa

dard error of regression. C((F0recast)/(Actual) - 1)x 100.

1969

1970

0.993

0.97

7.93

0.93

0.59

1.58

0.995

1.08

6.65

0.30

0.70

1.75

0.998

2.00

4.67

-0.29

0.00

1.22

I_ 0.4059 (5.08)

_O O_ 0.5583 (10.58) 0.6343

0.4239

0.998

2.05

4.65

-0.29

(4.95) (7.85) 0.2246 0.7613 (4.88) (14.49)

(8.26) 0.4079 (7.83)

0.998

2.06

4.78

-0.09

0.424 0.72

0,05

0.3176

0.0786 -0.6065 (t.57) (-6.10)

avaIues of the coefficients have been rounded at the fifth place below the decimal point. Figures in paxentheses show _'_3d,UOS" i

1968

-0.tl 0.11 -

1.05 1.42 -

JAPAN'SPERSONALCONSUMPTION EXPENDITURES

343

long-run perspective in explaining the Japanese consumption behavior. It is well-known that the existence of the lagged variable in the equation (7) permits at least two possible interpretations: it represents either authenti_ lagged behavior of consumers or merely a combination of the absolute income hypothesis (equation (5)) and the autoregressive error term. It is easy to decide on this question by following a test proposed by Professor Z. GriHches (1967, pp. 33-34), which entails estimating a supplementary equation c = Â˘7 + _TY + 77Y- 1 + _7 c- 1 + e and checking to see (i) if 77 is negative and significantly different from zero, and (ii) whether or not 8777 = - 3'7holds.S The computation shows that the value of - V7 (significant) is approximately half of the value of _777. Hence the autoregressive error model is not acceptable as an appropriate interpretation of the result. Regarding equation (9), it can be noted that the sum of the coefficients of y_ 1 and of c_ t does not differ significantly from unity. 6 In terms of predictive capability, however, it seems slightly inferior to other two equations, (7) and (8). A natural question comes to one's mind: is it justifiable to take the entire fifty-one year period as a unit of observation, as has been done in Table 17 Is it not highly conceivable that the sociopolitical renovations after World War II have caused fundamental transformations in the pattern of consumption expenditures? To name a few examples, the rapid "Americanization" of lifestyle has been quite manifest after the War (but particularly after the mid-1960s), as testified by the increasing importance of dairy and wheat products in food consumption. The intensified rate of urbanization, together with the spreading of higher education and of the mass media must have made the general public more susceptive to demonstration effect than they had been previously. The high rate of technological progress in machinery and electric appliance industries has resulted in the decline in the relative prices of their products, which in turn has stimulated the purchase of consumer durables such as automobiles, color television sets, vacuum cleaners and washing machines. In a 5. If one addsto the equation(5) the assumptionthatet = pet_ 1 + _t, the supplementaryequation is easily obtained. Note this equationmay be reduced to the identicalformas our equation (8). 6. Standarderrorof 03s + 75) is 0.0078.

344

KONOSUKEODAKA

word, the age of the mass consumption has finally arrived. Furthermore, various social innovations such as land reform, the liberalization of labor movement, the introduction of the new social security system; etc., seem highly relevant in transforming consumer behavior after the War. There is.some empirical evidence of a change in the pattern of consumption. Professor T. Mizoguchi (1968) had once computed, on the basis of both time-series and cross-section data, itemwise expenditure elasticities of consumption for the prewar and postwar decades. 7 He had discovered, among other things, that: 1) The elasticity of food consumption rose during the period of 1878-1908, after which it declined until 1938. This suggests that the per capita volume of food consumption reached a saturation point by the turn of the century despite the low level of income that prevailed then. 8 Presumably this phenomenon reflects the rice-dominant diet of the day. The elasticity value jumped up after World War II before it started to go down again; in the early 1960s, however, it maintained a comparatively high level (greater than 0.5) in view of the improved standard of living. This reflects, as_argued by the author, the increasing "Westernization" of the postwar pattern of food expenditures; and 2) The income elasticity of cloth consumption was higher in the prewar period compared with that in the postwar days. Moreover, it has shown a downward trend in the postwar period, suggesting a decline in the relative price of textile products: another result of the technological progress in the postwar period. In view of the importance of the question raised above, a statistical test was performed in order to determine whether or not a structural change had in fact taken place after World War II. For this purpose, the method proposed by Professor G. ChOw (1960) was adopted. The results are significant at the one percent level for all the 7. In calculating the elasticities, Mizoguchi used total consumption expenditures in place of personal disposable income. The time-serieselasticities have been computed for successive ten-year periods for the prewar years (1878.1938), and three.year periods for the postwar years (1952-1964). 8. Mizoguchi (1968) warns, however, that the period prior to 1908 requires extra care in handling the consumption data, as elasticities of certain expenditure items for the period (1878-1908) are not only erratic but take highly unrealistic values.

JAPAN'SPERSONALCONSUMPTIONEXPENDITURES

345

equations tested ((7), (8), and (9)). 9 It may be concluded therefore that the parametric values of these regression equations have undergone significant changes after the War and that the consumption functions should be estimated separately for prewar and postwar decades. The results of separate estimation of equations (7) through (9) appear in Table 2. On the basis of this computation, Figure 2 depicts the predictive ability of the equation (8). Excepting the periods of wide fluctuations (such as 1906 and 1921-22), the performance of the equation is generally satisfactory. Similar conclusion may be reached for two other regressions. 1째 It may be observed here that the regression coefficients differ significantly between the two periods and that the relative weight of the constant terms is much lower for the prewar period. Furthermore, the coefficients of y_ 1 and c_ 1 in the equation (9) fail to add up to unity. 11 Consequently, there is difficulty in applying the wealth-effect hypothesis for prewar Japan. The last two columns of Table 2 report estimated values of long run marginal propensities to consume (LMPC) and average lengths of adjustment lags, respectively. The former was calculated on the assumption that y grew at a constant rate of r*, which is the observed trend value of y for the respective period (see Table 4). Obviously r* does not represent the steady-state rate of growth. However, it seems meaningless to calculate the value of LMPC for a stationary state where income ceases to grow (r ---0), as sometimes done. The average lengths of lag stand for mathematical expectations 9. F.statistics are found to be 6.43, 4.32, and 5.08 for equations (7), (8) and (9), respectively. Exactly the same procedure has been applied to two sub-periods in prewar Japan (1892 vs. 1905-1938), yielding totally negative results. In addition, a test for the autoregressive model has been performed; the resuits were again negative for both pre-and postwar years. 10. From a theoretical point of view, it would be definitely better to choose a specification that has the ability to explain the entire span of observation period without taking a recourse to the concept of a structural change. In fact, the use of log-linear versions of the consumption functions has given a negative result to the test of a structural change. Even in this ease, however, some of the parametric values are so far apart between the pre- and postwar periods. Consequently, confidence intervals of some coefficients (such as #3 and 73 in the equation (7) and 54 in the equation (8) are completely disjoint between the two periods. 11. Standard errorof (f15+ 75) is 0.0088 (prewar) and 0.0175 (postwar).

346

KONOSUKE

ODAKA

Table 2 - Personal Consumption

Function in lapan: Prewar (1892-1938)

and Postwar (1954-67) Average

Equation Period

Prewar

Postwar

Number

Independent Constant

4.265 (1.87)

4.236

Y-I

0,2433 (4.58)

Variables c I

0.7206

0.2469

(1.81)

(3.09)

(8.10)

(3.62)

0,2133 (2.74)

0.7793 (9,16)

0.2478 (3.54)

13.63 (5.08)

10,28 (2.43)

0,4748 (4.14)

0,4095 (2.77)

0,6128 (12.67)

0.2515 (3.03)

0.7078 (7.11)

0.6408 (11.37)

0.5854 (14.45)

LMPC'_ Lag (years)

0.987

2.08

4.42

0.809

2.51

0.987

2.08

4.47

0.808

2.54

_e r_ O Z t8 Z us

0,7153 (11.26) 0.2386

0.2682 (4.80)

0.986

2.06

4.59

0.899

0.999

2.07

2.65

0.779

0.37

0.999

2.19

2.64

0.781

0.40

0.999

2.66

3.17

0.814

aLong-tun masginaI propensity to consume. It has been computed as follows: LMPCfat the equation (7): _ (l+t)](l+r._) and LMPCfor the equations'(8) and (9): (fl4_t)/(l+t-_), Whete t is the growth rate Oa'endvalue) of pet captta personal disposable income (given in Table 4).

.$L ._

348

KONOSUKEODAKA

derived from the actual lag distributions, in conformity to the procedure suggested by Professor Griliches (1967). The average lag has not been computed for the equation (9), for the latter does not assume the lagged income hypothesis. It is clear from Table 2 that while the short-run marginal propensity (SMPC) in the equation (7) increased in the postwar period, the value of LMPC moved to the opposite direction. In addition, note that the average length of lag has been substantially shortened after the War. This, of course, is not surprising. The relatively cheap, indigenous components must have dominated the consumption demand in the prewar decades. 4. lntereountry

Comparisons

In order to identify the uniqueness, if any, of the Japanese experience, an inter-country comparison of personal consumption expenditures has been attempted. For this purpose, three economies have been chosen (United Kingdom, United States, and Canada) and exactly the same models as described above (equations (5) through (9)) have been estimated, t 2 The selection of these countries has been purely accidental, as it has been dictated by the ease of access to the data. The period of observation covers the years between the 1920s and the 1960s, excluding the World War II days. Table 3 reports the results of the computation. Just like the Japanese case, the estimated equations (5) and (6)indicate signiff cant autocorrelations. However, the predictive ability of equation (5) for the United Kingdom is surprisingly high. In the case of the United States, it is noticeable that equation (9) excels all the others in terms of forecasting capacity. The SMPCs obtained from equation (7) are much greater than that for prewar Japan, although they are roughly equivalent to the postwar Japanese values. The sums of the coefficients of y_ 1 and of c_ 1 in the equation (9) add up to unity for all the countries reported in Table 3.13 The magnitude of the/35 coefficient in the same equation, representing presumably the wealth effect on consumption, is greatest for Canada, followed by Japan (prewar), the United States, and the United Kingdom in this order. 12. In addition, the equation (10) has been estimated with no satisfactory results. 13. Standard errors of (f15+ 75) are 0.0036 for the U.K., 0.0095 for the U.S. and 0.0064 for Canada.

Table 3 -Personal

Country

Equation Number

Constant

Consumption

Function,

Independent Variables Y

Y-1

c-1

U.K., U.S. and Canada -2 R

Forecasting Capability (%)a

Period Covered 3' Z

6.149 (6.35)

6 U.K.

7 8

67.47 (4.90)

7 8 9

63.09

0.65

0.89

&29

0.29

0.58

0.988

0.59

0.89

0.21

0.31

0.80

0.993

1.17

0.70

-0.36

0.16

1.09

0.996

2.03

0.54

-0.58

-0.04

0.51

0.996

2.12

0.54

-0,47

0,996

1.51

18.58

-1.72

(6.83)0"7520 (0.92)0"1069 0.5399 0.3911 (7.52) (4.43) 0.2003 0.7744 0.7156 (2.17) (7.19) (10.62) 0.1343 (1.90)

0.8618 0.7039 (11.75)

(10.52)

0.8695 (75.95)

(12,70)

46.88 (3.46) 12.46 (0,62)

0.6927 (12.07)

0.15 -1.14

1921-38 &1948-62 (1913 prices

O'J

0.75

-1.42 r_

0.7384 0.1372

(4.90)

0,988

(49.74)

1.008 (I.I0)

6 U.S.

(5.76)5"865 2.863 (2.68)

0.8519

0.997

1.37

17.17

-1.80

-1.47

-1.81

1929-41 &

0.997

1.40

15.98

-1.50

-1.28

-1.75

0.998

1.66

14.85

- 1.07

-0.57

- 1.33

1948-42 (1947/49 prices)

0.997

1.73

14.64

-0.89

-0.33

-l,16

(2.29)

0.2630 (1.29)

0.2127 (3.13) 0.7070 0.6835 (3.02) (12.78)

0,1611 (1.36)

0.8280 0.6745 (6,48) (13.28)

'Table 3 (Continued)

,Country

Equation Number

Independent Variables Constant

_2 Y

Canada

59.86 (3.12)

0.9005 (58.05)

47.01 (2.69)

0.5513 (5.03)

14.58 (1.07)

0.4423 (6.89)

Y-

c - I

0.3677 (3.21) 0.5330 (7.22)

12.36 (0.89)

0.3749 (3.91)

0.6049 (5.72)

0.4861 (6.14)

0.3460 (3.85)

0.6460 (6.81)

0.4747 (6.10)

aExtrapolated to three-year peribd succeeding to the observation period.

Forecasting Capability (%)a

s 1

Period Coveted 3

0.990

0.87

36.70

-0.22

0.39

2.24

0.992

0.55

32.52

-0.14

1.23

1.26

1927-41 &

0.996

1.44

23.19

-0.10

1.53

0.67

1948-68 (1944/49

prices) 0.996

1.69

23.23

-0.11

1.43

0,76

O > I:'Â˘

0.996

1.73

23.16

0.16

1.75

0.99

0 '_-

JAPAN'SPERSONALCONSUMPTIONEXPENDITURES

351

The statistical test on equation (7) for the validity of the autoregressive error model indicates that the values of/37_ 7 for both the United States and the United Kingdom are different from those of -77 by not more than ten percent. This indicates that there is room for doubt on the validity of the interpretation that equation (7) represents the lagged-income hypothesis. 14 On the other hand, the interpretation seems acceptable for Canada, which has yielded the average lag of about 1.1 to 1.2 years. In order to make a comparable observation on the effect of growth rate on consumption, it is necessary to standardize for the value of r. Accordingly, several hypothetical consumption propensities have been computed in Table 4. Obviously, Japan ranks at the top in the actual annual rate of growth (trend value) of per capita personal disposable income, while Canada is at the bottom. The comparison is somewhat misleading, because the Great Depression appeared to have been more damaging to the Canadian income level than for other countries; on the other hand, Canada's growth rate after the recovery was very rapid indeed. Note that the rate of growth of prewar Japan is roughly level with those for the United States and the United Kingdom. In short, the effect of growth rate on personal consumption must have been most manifest for postwar Japan. Given the estimated equations (8) and (9), columns three through eight in Table 4 measure the extent to which a change in r affects the consumption propensities. It is clear that in all cases, an increase in the growth rate results in a decline in consumption propensities, as anticipated. It is also clear, however, that for the same rate of growth of income, Japan indicates consistently lower propensity values compared with other countries. On the other hand, the other three show highly comparable figures with each other when r is normalized. What would be an appropriate interpretation of these results? First of all, the high savings ratio in Japan is not solely ascribable to the high rate of growth. In the case of prewar period, the relative strength of the demand for traditional commodities should perhaps 14, If one assumes that the lag.adjustmenttheory is also applicableto both the United States and the United Kingdom, average lagsare 0.6 and 0.3 years, respectively. (A similar observation has been made by Yoshihara (1972) on the postwar U.S. economy).

352

KONOSUKE ODAKA

Table 4 -- Effects of Income Growth on Consumption Propensities

Country

Actual rate of Growth (r)a

LMPC, Derived from Eqtn. (8)

APC*, b Derived from Eqtn. (9)

r_-

actual

0.02

0.08

actual

0.02

0.08

Japan (prewar) Japan (postwar)

0.0229 0.0791

0.81 0.77

0.82 0.79

0.72 0.77

0.93 0.81

0.93 0.85

0.79 0.81

U.K.

0.0216

0.86

0.83

0.96

0.89

U.S.

0.0233

0.88

0,85

0.96

0.89

Canada

0.0146

0.91

0.90

0.85

0.98

0.97

0.90

aTrend values, as estimated for respective periods by fitting In y = o_+ _Jt-In this computation, the _ values for pre-and post-W.W, II years were estimated separately and then transformed to weighted averages. bComputed as (_ + gr) / (_ + r)_ except for the value for postwar Japan, where the formula (_ + _r) / (1 + r - j_) has been used in consideration of the fact that the coefficients of y_l and c_1 do not add up to unity in this case.

be taken into consideration. In the postwar scene, on the other hand, not only has the adjustment lag decreased in length, but the SMPC moved upward in comparison with the prewar experience. The low LMPC value in postwar Japan may perhaps be explained in reference to the extremely low wealth-income ratio in the Japanese households after the War. Special reference may be made to the condition of residential housing. I have estimated elsewhere (1971, p. 128) that the quality of Japanese residential houses in 1963 was approximately two-fifths of that of the United States (as of 1960) in terms of space and facilities. Furthermore, the ratio of household tangible assets to annual personal disposable income in Japan was 0.83 for the fiscal year 1970, which is about half of a similar figure for the United States in 1935 (1.64). 15 The maladjustment between the flow and the stock quantities may also simple-minded wealth-effect

be responsible for the failure hypothesis in postwar Japan.

the

15. See Japan Statistical Yearbook, 1972, p. 501 for the Japanese data, and Goldsmith (1956, p. 14) for the U.S. data.

JAPAN'S

PERSONAL

CONSUMPTION

EXPENDITURES

353

5. Summary The major findings of the present paper may be summarized follows:

1) The intertemporal movements in consumption expenditures in Japan during her modern economic growth may be best explained by the lag-adjustment hypothesis, rather than the absolute income hypothesis. Alternatively, one might interpret this as the relevance of a version of the permanent income hypothesis. 2) It seems that a structural change in the consumption pattern took place in Japan after World War II. 3) A comparison of consumption functions between prewar and postwar periods suggests that the former is characterized by an extremely low adjustment speed while the latter by a decline in the long-term marginal propensity to consume. 4) Japanese propensities to consume, having been normalized for difference in growth rates, rank low compared with those of the three countries examined (the United Kingdom, the United States, and Canada) _. To this extent, Japanese consumers have behaved differently from those in the other nations. 5) The Japanese economy experienced a serious setback in the early 1970s, notably by the Nixon shock and the two oil shocks. The average growth rate of real national product per capita (net of depreciation) significantly declined in the 1970s from 5.1 percent for 1970-1974 to about 1.8 percent in 1975-1979; even negative rates were registered in the calendar years 1974 and 1975 (-3.8 and -2.0 percent, respectively). Correspondingly, the average rate of growth in real personal disposable income per capita (y) decreased from 6.7 percent for 1971-1974 to 2.3 percent in 1975-1979. Surprisingly, however, APC kept its relatively low position throughout this period of transformation despite the slowing down of the economy; not only did it average to approximately 76.7 percent for the 1970-1979 period, but it actually went down from 77.8 percent for the In-st half of the 1970s (1970-1974) to 75.6 percent for the second half (1975-1979). This fact alone seems sufficient to cast serious doubt on the validity of the habit formation hypothesis as applied to Japan in the 1970s, although it may be still too early to give the final verdict on this matter. In any event, further work is urgently needed in order to explore the behavior of the Japanese consumers, and to determine if in fact

354

KONOSUKEODAKA

they behave differently from their Western counterparts. Especially pertinent in this regard is the study of the role of an asset variable, a factor virtually neglected in the present paper.16

Appendix Note on the Data Sources The following data sources have been consulted in carrying out the present investigation. Japan C: Miyohei Shinohara, Kofin sh_hi (Personal Consumption), LTES, Vol. 6 (Tokyo, 1967) for the prewar period and Economic Planning Agency, Yearbook of National Income Statistics for the postwar years. Y: Work sheets of Messrs. Kazushi Ohkawa and Nobukiyo Takamatsu for the prewar decades and Economic Planning Agency, Yearbook of National Income Statistics for the postwar years. C.P.I.: Noda estimates for the prewar years (Kazushi Ohkawa, et al., Bukka (Prices), LTES, vol. 8 (Tokyo, 1967)), and the Bureau of Statistics index (Prime Minister's Office) for the postwar decades. The pre- and postwar indices have been â&#x20AC;˘linked by the adjustment factor suggested by Tsutomu Noda in Bukka. Population: Estimates by Professor Mataji Umemura for the prewar period, and those of Bureau of Statistics for the postwar years. United Kingdom All the data have been taken from C.H. Feinstein, National Income, Expenditure and Output of the United Kingdom 1855-1965 (Cambridge, Cambridge University Press, 1972). United States C and Y: U.S. Department of Commerce, Office of Business Economics, The National Income and Product Accounts of 16. A study by Sato (1980, p. 197) is suggestivein this regard.

JAPAN'S PERSONAL CONSUMPTION EXPENDITURES

355

the United States, 1929-1965. Statistical Tables. A supplement to the Survey of Current Business (Washington, D.C., 1966). C.P.I.: B.L.S. index has been adopted from U.S. Department of Commerce, Historical Statistics of the United States, Colonial Times to 1957 (Washington, D.C. 1961) and U.S. Department of Commerce, Statistical Abstract of the United States (Washington, D.C., 1971 ). Population: U.S. Department of Commerce, Statistical Abstract of the United States, (Washington, D.C., 1971). Inclusive of military

personnel

abroad.

Canada All the data Expenditure October 31, used instead

come from Statistics Canada, National Income and Accounts, Historical Revision 1926-1971 (Ottawa, 1972). The implicit consumption deflator has been of the consumer price index (C.P.I.).

REFERENCES Brinner, R. (1973), "Neoclassical Wage Dynamics and Involuntary Unemployment," Unpublished Ph.D. thesis, Harvard University. Chow, G. (1960), "Test of Equality between Sets of Coefficients in Two Linear Regressions," Econometrica, 28: 591-605. Friedman, M. (1957), A Theory of the Consumption Function, Princeton: Princeton University Press. Fujino, S. (1972), Shotoku to bukka no kiso riron (The Foundations of Income and Price Theories), Tokyo: S&un Sha. Gleason, A. (1965), "Economic Growth and Consumption in Japan," in The State and Economic Enterprise in Japan. ed. W.W. Lockwood, Prince- ...... ton: Princeton University Press, pp. 391.444. Goldsmith, R. (1956), A Study of Saving in the United States, Vol. 1II, Princeton: Princeton University Press. Griliches, Z. (1967), "Distributed Lags: A Survey," Econometrics, 35:16-49. Kuznets, S. (1966), Modern Economic Growth; Rate, Structure, and Spread, New Haven: Yale University Press. Mizoguchi, T. (1968), "ShOhi k6z6 no ch6ki henka to kokusai hikaku yori mira tokushoku (Consumption Pattern in Japan and Its International Comparison),"Ikkyd ronsd(The Hitotsubaski Review), 60: 20-35. (1970), Personal Savings and Consumption in Postwar Japan. Tokyo: Kinokuniya.

356

KONOSUKE

ODAKA

Odaka, K. (1971), "Housing Conditions in JaPan and the United States: A Comparison," in Comparison of Levels of Living in Real Terms in Japan and the United States, Tokyo: Japan Society for the Promotion of Science. Ohkawa, K. (1970), "Chochikuritsu no ch61d bend6 (Long-Term Movement of the Savings Rate)," Keizai kenkyfi (The Economic Review), 21: 129-136. Ohkawa, K. and Rosovsky, H. (1965), "A Century of Japanese Economic Growth," in The State and Economic Enterprise in Japan, ed. W.W. Lockwood, pp. 47-92. Rosovsky, H. and Ohkawa, K. (1961), "The Indigenous Components in the Modern Japanese Economy," Economic Development and Cultural Change, 9: 476-501. Sato,

K (1981), "Senkan-ki Nippon no makuro keisai kik6 (The Macro-economic Mechansim of lnterwar Japan)," Keisai kenkyfi (The Economic Review), 32: 194-202. Yoshihara, K. (1977), "The Application of Alternative Demand Models to the Japanese Expenditure Pattern," Econometric Studies of Japan, eds. R. Kosobud and R. Minarni, Urbana: University of Illinois Press, pp. 7-48. (1972), "The Growth Rate as a Determinant of the Saving Ratio," Hitotsubashi Journal of Economics, 12: 60-72.

GROWTH AND EQUITY IN A SUB-REGIONAL ECONOMY: HAWAII, 1900-1976 Mitsuo Ono *

1. Introduction This paper continues the analysis of Hawaii's economic growth initiated by Professor Harry Oshima during the early '60s at the University of Hawaii. He noted that "once we know the growth rate of Hawaii, we should be able to begin the study of the various factors which contributed/to the growth of Hawaii. For example, the role of education, the role of capital formation, the role of technical progress, and others can be studied closely. These studies undoubtedly will uncover other factors pertinent to Hawaii's growth and will lead to other studies, as development economists are now discovering. The implications of the Hawaiian case for comparative analyses of the growth of nation may be far reaching" (Oshima and Ono, 1965, pp. XI-5). Indeed, the socio-economic development of Hawaii involved a remarkable interplay of natural resource utilization, trade and physical and human capital formation molded under varied cultural and institutional forces related to the islands' racially mixed population base.

*ResearchCorporation of the University of Hawaii.This paperis derived from a study sponsoredby the State of HawaiiDepartmentof PlanningandEconomicDevelopment(DPED).Viewsexpressedherein arethose of the author. Commentsby Mr.RobertSchmittand Dr. RichardJounof DPED are gratefully acknowledged.Also, the typing assistanceof Mrs.Chato Ramos is deeply appreciated. 357

358

MITSUO ONO

During the last hundred years or so, Hawaii has evolved from a sparsely populated territory to an economy which ranked among the top ten states in the Nation in the mid-70s, as noted in Table 1. Table 1 -- Hawaii's Ranking on Selected Socio-Eeonomie Indicators Hawaii's

Percapita personalincome (1977) Percapita money income(1977) Median family income (1975) Percentage of people living below poverty level (1975) Long-term unemployment rate (1976) Life Expectancy ( 1969-1971 ) Death Rate (Rate per 1,000 population) (1975) Median School Years Completed ( 1976) Percent High School Graduates (25 years +) (1976) Inter-state Mobility (Percentage moving) (1965-1970)

U.S.

Hawaii

$ 7,019 $ 4,838 $14,094

$ 7,677 $ 5,259 $17,770

RankAmong States 8th 9th 2nd

11.4 6.8% 70.85

7.9 8.9% 73.6

44th 12th 1st

8.9 12.5

5.9 12.7

50th Top 2nd grouping

66.6

73.0

- do -

11.5

18.5

5th

Source: U.S. CensusBureau,StatisticalAbstract, variousissues. What was this "catching-up" process? were the critical factors involved?

How did it happen?

What

Although there are a number of economic histories on Hawaii, they are primarily descriptive. There is a need to identify and analyze the key social and economic changes (including sources of growth) that had taken place in Hawaii during the past seventy-five years or so using data obtained from the system of economic and social accounts. In this regard, historians in Hawaii are fortunate in having a comprehensive and valuable reference material available: Robert Schmitt's Historical Statistics of Hawaii. In addition, the gross state product estimates for 1901 and 1958-1960, formulated by Professor Oshima, provided the computational base analyzed herein. Without these historical data sources, this statistical investigation would not have been possible.

GROWTHAND EQUITYIN HAWAII

359

A further contribution qlere is that although there have been many country studies analyzing sources of long-term economic growth, to this writer's best knowledge, no such studies have been attempted for a sub-regional island economy like Hawaii, the only state in the Nation which has a continuing set of articulated income and expenditure accounts beginning with 1958. 2. Analytical Approach The estimation methodology applied herein follows the Denison approach, (Denison, 1962, 1967, 1974, 1976). His method uses the simple Cobb-Douglas aggregate production function concept. Thus, assuming short-term competitive conditions, changes in output value are equal to changes in factor supply weighted by the factor share of output value plus changes in a residual factor index. A simple derivation follows: (1)

dY Y

- dK Y

dY -dK

(2)

_ dK

+ dL

L +

residual

(3)

dY Y

_ dK K

__dY dK

K Y

__dY dL

L + Y

residual

dY Y

_ dK K

dY/Y dK/K

+ d__L_L dY/Y L dL/L

+- dL Y

dY -dL

+ residual

d-Y

dL_L L

Also (4)

+ residual

Since dKdY_ Mp K = Profit Rate;_LY = MP L = Wage Rate MP (5)

Overall Growth Rate

dY Y

_ dK K

. K K Y

MP + d__L_L L

Growth of Capital Growth - Capital Stock" Share + of Labor

. L L Y

residual

Labor Residual " Share + productivity index

360

MITSUOONO Calculations

are made as follows:

Assume that the labor share of income is 73 percent. Then a one percent increase in labor quantity is associated with 0.73 percent increase in growth. Calculations show that a 1.3 percent increase in labor is related with 0.949 percent increase in growth. Assume that the long-term average annual growth was 3.8 percent. Dividing 0.949 percent by 3.8 percent provides the contribution of labor to the growth rate. Assumptions

used include the following:

Constant returns to scale with no interactions among variables, hence, each component is additive. Also, marginal products of factors are equal to factor returns which assume pure competitive conditions. Differential factor returns are wholly identified with the targeted sources. Also, no major economic structural changes are assumed for analytical periods. This is maintained by averaging short periods for analysis. As in all quantitative studies, the estimation techniques applied here which reduce raw data into summary measures will be Subject to some criticism. 1 However, in line with standard practice, detailed descriptions of the methods used to derive these results are available from the author so that future researchers can reproduce and further improve them.

1. For example, in the book review of Dertison's Sources by Abramovitz (1962), he noted that: "In spite of the author's careful and candid exposition, there is a certain air of reliability and precision about his estimates which the state of knowledge today simply does not support". Also, in a book review of Denison's Why Growth Rates Differ by Deane (1968), she wrote: "Denison's numbers raise some profoundly controversial issues. For these are not the kind of statistics that effectively prove or disprove anything at all, except the immense complexity of the task of defining or interpreting national growth rates. They raise more interesting questions than they answer and Denison is always explicit. Only the most naive reader could suppose that these numbers are intended to be taken at their face value". These comments are also applicable to the preliminary statistical results included herein.

GROWTHAND EQUITYIN HAWAII

361

3. Historical Overview of Hawaii's Growth Initially, the various stages of Hawaii's growth are presented help identify sources of long-term growth.

Period I - Agricultural Emergence (1866-1900) This period can be characterized primarily as the transformation period from an agricultural subsistence and trade economy to an agricultural industrial production economy. There were further developments in entrepSt trade such as expanded exports of forestry and agricultural products, e.g., sandalwood and services to whaling ships and related activities. From 1866 to 1900, annual sugar exports increased from 8,000 tons to about 350,000 tons, respectively. Rice and coffee exports in 1873 amounted to 470 tons and 130 tons, respectively. By 1900, the foundation of agricultural industrialization was already established. Period II - Agricultural

Expansion (1901-1935)

According to Professor Oshima's estimates, the State product in 1901 was about $38 million. Estimated population in 1900 was 155,500, of which 75 percent were living in rural places. Also, approximately 40 percent were native born. The balance was foreign-born, about 60 percent and 24 percent of whom were Japanese and Chinese, respectively. In 1900, approximately 75 percent of the 10 year-and-over population were gainfully working. About 63 percent were in agriculture, forestry and fishing. This would increase to 73 percent if the laborers in the "industry not specified" category were considered doing agricultural work. About 10 percent of the workforce were in manufacturing and mechanical industries, about 8 percent in transportation and communication, and about 7 percent in domestic and personal services. In 1909, the average annual payroll per worker in manufacturing amounted to $370 per year. Dividing this by annual hours worked (62 hours per week x 52 weeks) resulted in a 12 cent-hourly rate. In 1910, 319 farms harvested 8,000 tons of pineapple valued at $331,000, significantly less than the production of rice (503 farms reporting, 418,200 hundred weight harvested, sales valued at $1.1 million) and sugarcane (1,028 farms reporting, 4.2 million tons harvested, sales of $26.3 million). Also, in 1910, of the total outgoing shipment ($46 million), about 93 percent ($45 million) were sugar exports. The

362

MITSUOONO

emerging agricultural industrial expansion was based on sugar production. Living conditions of the population also improved. The life expectancy at birth during 1919-1920 was 48 years, a substantial increase from the life expectancy of 32 years during 1883-1886. During this period, the external sector (primarily agricultural exports) had the key economic role in expanding agricultural production. The high price of sugar (partially subsidized) on the Mainland yielded high profit rates which were then reinvested into capital equipment, research and development to improve productivity of sugar production and milling. According to Professor Oshima, "It seems clear that high profits as a source for financing the accumulation of capital in the sugar industry were the result of efficient management and low wages." (Oshima and Ono, 1965, pp. I-7). On the factor supply side, he adds that "while the business sector was accumulating capital embodying new technology, families were raising a large number of children despite low wages" (Oshima and Ono, 1965, pp. I-8). This is reflected in the median age of the population which dropped from 29.1 in 1910 to 23.8 in 1930. School attendance also increased. During this period, then, the expansion on the aggregate demand side (primarily moved by agricultural exports) was accompanied by "human capital" expansion on the factor supply Side. Period III - Military Base Emergence (1935-1949) The two previous periods were primarily focused on agricultural development and export sales, primarily sugar, pine_ipple, rice and coffee. The visitor industry was only meeting "selective" tourism needs (as contrasted with "mass" tourism) supporting a relatively small number of recurring visitors in a few expensive Waikiki hotels during the winter months. Between 1922 and 1935, estimated visitor expenditures ranged around $6-10 million, while the combined sales value of sugar and pineapple was about $90-100 million. Beginning in the late twenties to mid-to-late thirties, however, the military base complex in Hawaii emerged. For example, the military population increased from about 18,000 (1929) to about 30,000 (1940), or about 67 percent increase. From 1935 to 1940, the military resident population rose by 35 percent, from about 22,000 to 30,000. Value of military expenditures increased from about $35 million in 1935 to about $45 million in 1940. For the 1939-40 period, personal income

GROWTHAND EQUITYIN HAWAII was approximately $550.

363

$230 million, while per capita income was about

Period IV - War Time Expansion (1941-1945) This period was probably the most important in the current economic history of Hawaii which significantly sped up the transformation of the economy. Activities in this period were devoted exclusively toward meeting requirements of World War II. Defense employment increased from 6,925 in 1940 to 64,350 in 1945. Armed Forces ashore increased from 44,278 in 1941 to 300,328 in 1945. The impact of military expenditures which shot up from about â&#x20AC;˘$45 million in 1940 to about $800 million plus in 1945 was reflected in increased personal income which rose from about $250 million to about $754 million between 1940 and 1945. Per capita income increased from about $550 in 1939-40 to about $1,328 in 1945. Hawaii's Gross State Product (GSP) rose from about $309 million in 1940 to about $1,225 million in 1945. The population almost doubled from 478,000 to about 815,000 while civilian employment increased from 155,000 to about 208,000. 'War-time activity during this period resulted in a significant backlog of consumer demand and accumulated savings which influenced business and government policies in the next phase (Period V) of Hawaiian economic growth. Period V - Postwar Adjustment

(1946-1958)

The GSP increased by 63 percent between 1946 and 1958 (from $873 million in 1946 to $1,425 million in 1958). Personal income and per capita personal income increased concomitantly by 64 percent and 51 percent, respectively. Personal income expanded from $719 million to $1,180 million while per capita personal income increased from $1,312 to $1,983. The changes during this period were as follows: a) Population increased from 545,439 to about 605,356 between 1946 and 1958 (+11 percent). b) The civilian labor force increased from 190,297 in 1946 to about 210,400 in 1958 (+11 percent). c) Membership in unions increased from an estimated 37,000 in 1946 to about 60,000 in 1958 (+62 percent). d) Consumer prices rose approximately by 42 percent (58.3 in 1946 to 82.8 in 1958, (1967=100).

364

e) f) g)

k) 1) m)

MITSUOONO

Per capita personal income rose from $1,312 in 1946 to $1,983 in 1958 (+51 percent). Registered motor vehicles increased from 81,700 in 1946 to 202,434 in 1958 (+148 percent). The general excise tax base for contracting rose from $47.0 million to $174 million (+270 percent). In the same period, the construction cost index rose by about 52 percent. The number of telephones increased from 70,600 in 1946 to 176,750 in 1958 (+150 percent). Transpacific calls rose from 89,273 messages in 1946 to 326,669 messages in 1958. Hawaii originated calls rose from 50,471 in 1949 to 197,098 in 1958. Interisland telephone calls increased from 155,850 in 1947 to 375,000 in 1958. Increases in the retailing and service general excise tax bases were 60 percent and 122 percent, respectively, from 1946 to 1958. (Retailing, $416.3 million to $665.0 million; Services, $61.1 million to $135.7 million). Number of hotel units increased from 1,572 to 5,494 units (+249 percent). Hotel employment rose from 1,313 to 3,611 while its payroll increased from $2.7 million to $9.9 million. Government receipts expanded from $42.4 million to $119.2 million from 1946 to 1958 (+181 percent). The median years of school completed increased from 8.7 to 11.3 years. (+30 percent). The life expectancy of persons (at birth) rose from 69.5 years (1949-1951) to 71.6 (1959-1961) (+3 percent).

Although these indicators showed considerable overall improvements, there were still major economic transition problems, e.g., the prolonged shipping strike in 1958. More important, the gap in business activity caused by sharp reductions in military spending had to be balanced by additional income-employment generating industries. A major stopgap was the expansion in the construction industry. In the quest tbr new industries for Hawaii, the Industrial Research Advisory Council and the Economic Planning Coordination Authority worked together in identifying and planning for potential business growth areas, e.g., tourism development. The foundation laid in the postwar adjustment period prepared the statehood (1959) expansion phase that continued from 1959 through 1965.

GROWTHAND EQUITYIN HAWAII Period VI - Statehood

365

Expansion (1959-1965)

As planned, a significant economic tranformation that occurred in this period was the influx of mass tourism in Hawaii which boosted the number of overnight visitors from 172,000 in 1958 to 687,000 in 1964. Between 1959-1964, the average intended stay of visitors declined from 21 days to about 14 days. Estimated visitor expenditures in this period rose from $109 million to about $225 million. Hotel units and hotel employment rose from 6,802 and 3,699, respectively, in 1959, to 14,227 and 6,308 in 1965. Hotel payroll rose from $9.9 million to $23.4 million. Period VII - Service Industrialization

(1966-Current)

This period reflects the shift from agricultural to service industrialization. Relative to agricultural and military support production, the visitor industry has become the primary export earner for Hawaii. The export share of visitor expenditures rose by 33 percent (from 30 percent to 40 percent in 1967 and 1971, respectively) while the combined share of agriculture and military support declined by 30 percent (from 25 to 20 percent). The linkage between multi-unit construction (especially multiunit residential buildings such as condominiums and apartment dwellings) and tourism also strengthened during this period, especially on Oahu. This tie-up between the visitor industry and construction activity also expanded in the outlying islands. Period VIII - International

Development

(1971-Current)

This period is characterized by the expanding role of Hawaii as a base for international investment and technical assistance, especially for the Pacific basin countries. The linkage in construction and tourism, supported by non-Hawaiian financial flows (entrusted at 50-60 percent) further expanded, which also accelerated housing values upward. This trend continues to the present. The relative share of visitor expenditures to direct export earnings (covering four major industries) increased from 40 percent in 1971 to about 48 percent in 1976. Summary The growth of the Hawaiian economy begins with its highly favorable climate, key geographical location and bountiful resources.

366

MITSUOONO

The early industrial development of Hawaii starts with a "demandcreating supply" model related to the sugar industry. There was an assured market for sugar on the West Coast due to the "Gold Rush", the Civil War and the import tariff advantage. With relative low labor costs and a disciplined work force, sugar producers reinvested high profits into further production facilities, agricultural research and development. Cooperative arrangements among sugar producers in establishing a central experimental station to test better species and applying large scale (plantation) growing techniques eventually lowered production costs due to economies of scale and advance of knowledge. Labor unrest was limited. Human capital investments expanded in education, training and population increase. As productivity of agricultural industries increased, profits of local producers plus Mainland investment sources were channeled into industrial and mercantile activities, thus expanding the production base for additional profits and further investments. In the meantime, immigrant families were accumulating savings and investing in education, housing, land and in business activities. With a broadening base, income equity improved. Thus, the combination of application of advances of knowledge in better production techniques to obtain better yields, (for example, through better planting methods, selection of disease resistant strains, better irrigation systems, fertilizer applications and control of harmful insects), better managerial practices and diversification (economies of scale) together with inputs of a diligent, trained but relatively low paid labor force and reinvested capital resources created an economic environment highly suitable for significant improvements in Hawaii's agricultural and commercial development. This agro-industrial base was further augmented by inputs from increased urbanization and expanding military-base procurements. During the post-war adjustment period, the' "assured demandcreating supply" model transformed to a "supply-creating demand" model. The Hawaiian economy had a postwar excess production capacity during the late forties and early fifties before the statehood expansion. Intensive investigations were made to identify and plan for the development of exportable goods and services to fill the gap caused by military demobilization. Adequate investment outlets for accrued domestic capital funds built up during the war-expansion period had to be found. The labor movement also expanded and became more active in seeking better wages and working conditions

GROWTHAND EQUITYIN HAWAII

367

during this postwar adjustment period. With growing labor force from returning veterans and lack of work opportunities, the unemployment rate rose, although this was somewhat softened by higher (government subsidized) college enrollment of veterans and outmigration of students and workers to the Mainland. Business and government agencies expanded the development of new industrial potentials for Hawaii, including tourism, manufacturing for h-nportsubstitution and exports, and diversified agricultural activities, e.g., clothing, giftwares, flowers, fruits and macadamia nuts. This transition period was a critical one in Hawaii's economic history. Demand creation and business investment development went handin-hand. For example, the linkages among technological developments in transportation, commumcation, hotel construction, expanding visitor industry demand and business expansion were analyzed and coordinated by the Hawaii Visitors Bureau. This in turn further increased business activities associated with the visitor industry and soon. In summary, theeconomicdevelopmentof Hawaiiwasbasedon a favorable investmentclimate, including both physicaland human capital formation. In the early stage of Hawaii's specialized agricultural development, a primary demand-oriented model was applicable (where production followed export demand). Because of the small economic base, import-substitution industrial development was not feasible. In the later stages (especially after the postwar transition period), a primary supply-creating demand model appears to be applicable. During the past seventy years or so, factor productivity increased significantly due to a better trained labor force, some economies of scale and technological developments, resulting in more efficient transportation and communication facilities and equipment. 4. Analysis of Historical Findings The next step is to relate these growth patterns with quantitative summary measures. Growth Rate The real growth rate of Hawaii's State Income between 1900 and 1976 (slightly dampened by declining sugar prices and the 1975 recession lag) averaged 3.8 percent per annum. State income grew by

368

MITSUOONO

6.9 percent in current dollars which was reduced by rising prices, expanding by about 2.9 percent per annum. The per capita income growth rate was 1.5 percent while the per worker income growth rate was 1.9 percent. While population expanded at a 2.3 percent annual rate, employment grew at 1.9 percent per annum. For comparative purposes, the real U.S. Gross National Product (GNP) increased by 2.9 percent from 1909 to 1959. From 1909 to 1927, real GNP increased by 2.8 percent per annum (Denison, 1962, p. 21). From 1929 to 1969, real National Income increased by 3.3 percent (Denison, 1974, p.12). For Hawaii, the real State income increased by 4.3 percent annually between 1930 and 1970. These figures indicate that Hawaiian output increased at a faster rate than that of the Mainlan& The average unweighted labor return (including self-employed) share of state income from 1900-76 was estimated to be 73 percent while the capital return share was assumed to be 27 percent (residual). For the Mainland, Denison noted a 71 percent labor share of national income, 22.8 percent for reproducible capital and a 5.8 percent land share from 1909-58 (Denison, 1974, p. 260). During the 1900-76 period, the labor share of Hawaii state income grew from about 64 percent to about 81 percent, primarily because of increasing supplementary wage payments received by employees, including the self-employed. The same trend is also evidenced by Denison's findings (Denison, 1962, table 4) for the United States. From 1909 to 1957, the share rate increased from 60 percent to about 83 percent. From 1958 to 1968, however, it was reduced from 83 percent to 80 percent (Denison, 1974, table J-l). Contribution

of Labor Input

Between 1900 and 1976, persons employed in Hawaii increased from 90,172 to 390,303, a fourfold increase (about 1.9 percent per annum). The population increased at a 2.3 percent annual rate. The de facto population almost doubled from 429,100 in 1940 to 952,306 in 1976, with an annual average increase of 3 percent per annum. Employment during this same period increased from 188,232 to 390,303 (about a 2.8 percent annual increase). Between 1900-1960, the employment level almost tripled (2.3 percent per annum increase). For the United States, employment increased by 92 percent between 1909 and 1958 (Denison, 1962, p. 37). The average annual

GROWTHAND EQUITY1NHAWAlI

369

growth rate was 1.3 percent in 49 years. In this same period, the population including Armed Forces overseas increased from 90,492 to 174,141, a 92 percent increase or an annual average rate of 1.3 percent. These data indicated that Hawaii's employment levels were growing faster than that on the Mainland. The average annual number of hours worked per employed person in Hawaii declined by 40 percent from 3,182 to 1,920 hours per annum (a drop of 0.6 per annum from 1900 to 1976). In 1900, it was estimated that the average worker spent 61.2 hours per week working; but in 1976, the comparable rate was 40.0. The estimated number of weeks worked dropped from 52 to 49 weeks per annum. The total hours worked are over-estimated, however, since it is assumed that all workers are working full-time. Between 1910-60, the annual manhours worked in Hawaii dropped by 40 percent. For the United States, Denison estimated (from 1909 to 1958) that annual hours worked dropped by 25 percent from 2,704 in 1909 to 2,057 in 1958 (Denison, 1962, p. 37). Thus, rapid gains in employment benefits were accruing to Hawaii's workers. The overall net gain in hours worked increased by 2.7-fold between 1900 and 1976 (average worker hour index multipfied by the employment index) which reflected a 1.3 percent annual rate of increase in labor input. Using the estimation procedure, the labor input growth rate (1.3 percent) was multiplied by the average labor share rate (73 percent) to obtain a labor (0.95 percent) contribution rate per annum toward the growth rate. This meant that during the 1900-76 period, labor received an average share of 73 percent of state income. If labor costs remained fixed and with a gain of one percent of labor quantity input during this period, the state income should also increase by 0.73 percent, all other things remaining constant (including production methods and other activities). This also assumes that any new supply of factor inputs was fully absorbed into underutilized capacity. Hence, the data presented here relate to contributions of factor inputs toward improvements in long-term productive capacity (Denison, 1962, pp. 31 and 32). Also, as noted by Abramovitz (1962) "the growth Denison seeks is growth in capacity to produce or, as he sometimes calls it, potential production". Since a one percent labor quantity increase is related with a 0.73 percent increase in growth, then, a 1.3 percent labor quantity

370

MITSUOONO

increase is related to a 0.95 percent increase in growth. Dividing 0.95 percent (as derived above) by 3.8 percent gives the contribution of labor inputs (employment adjusted by hours) toward long-term Hawaiian productive capacity. This labor input contribution "explains" about 25 percent of the growth rate. This rate may be overestimated, however, since it was assumed that employment covered only full-time workers. These adjustments can be made later when more detailed data are available. There are many growth policy implications related to this summary measure. Thus, working hours and employment data can be divided into various demographic, social and economic groups. As noted by Denison (Denison, 1962, Chapter 6) employment changes are related with changes in population, labor force participation rates, employment rates, etc. Working hour changes are related with potential to actual average working hours per week, average weeks worked per year, and the like. In turn, these relate to impact of work stoppages, nutrition and health status of the population, propensity for leisure, use of leisure time, family/household work patterns, lost labor due to crime, and other activities and so on. Also, each of these opportunity costs can be related to public and private programs to reduce lost labor and to find cost-effective ways to _ncrease labor input. Simulation techniques can be applied to formulate different "what if" scenarios (under limiting assumptions) showing how different socio-economic activities can impact long-term Hawaiian productive capacity, e.g., what are the gains in retrieving lost labor from those unemployed or underemployed or in preventing institutional commitment through education and social rehabilitation. Variations in Labor Input If annual hours worked per person were kept constant from 1900-1976, the employment growth rate would show an increase of 1.9 percent per annum. Multiplying this rate by 73 percent gives 1.39 percent. Thus, during 1900-1976, the employment factor without accounting for lesser hours worked per worker would be accountable for about 37 percent of the 3.8 percent growth rate of long-term Hawaii State income capacity (instead of 25 percent rate found previously). Hence, the reduction in annual working hours per worker decreased labor input's contribution by about 32 percent.

GROWTHANDEQUITYIN HAWAII Contribution

371

of Education

Between 1900 and 1976, the median school year completed by the resident population increased from an estimated 3.6 years (related with school attendance) to 12.7 years. These data cover attainment of formal training only and do not cover additional human capital investments such as on-the-job training, special studies, and so on. The average rate of school year completed per person increased by 3.5 fold, about 1.6 percent gain per annum during the past 76 years. However, educational attainment gains must be related to economic value added. Thus, after making rough calculations, it was estimated that the real income gain associated with these school year changes was about fivefold (from $1,207 in 1900 to $6,399 in 1976). This resulted in an increase of about 2.2 percent per annum. Assuming that 50 percent Of this gain was attributed to variables such as intellectual ability, ll_me training, mass communication, and so on, approximately 50 percent of the 2.2 percent was allocated to gains from better household social environmental conditions. Denison (Denison, 1962, pp. 69-70) makes a 60 percent adjustment to reduce the school year differentials in economic contribution by education attainment. Applying the estimation methodology, 1.1 percent (education input) is multiplied against 73 percent (labor share of income) to obtain 0.8 percent (the growth rate attributable to education). Dividing 0.8 percent by 3.8 percent produces 21 percent, the contribution of formal educational attainment toward the long-term Hawaii State income-generating capacity. Denison initially found that educational attainment contributed 20 percent towards improvement in the U.S. growth rate during the 1909-1957 period (Denison, 1962, pp. 73 and 148). For 1920-1957, it was calculated that education contributed 23 percent of the growth rate but in a later study (Denison, 1974, p. 346), the figures were revised downward to 13 percent due to reductions in the number of days attended per school year and changes in census estimates. (Denison, 1974, p. 346). Given the limitations of these summary measures, how can these data be used for human investment policymaking2 Some of these issues are discussed by Denison (1962, Chapter 7). Unquestionably, adequate educational attainment of the population is a prime requisite for the expansion and improvement of Hawaii's long-term productive capacity. For each one percent school year completed, there appears to be a 1.4 percent gain in worker's

372

MITSUO ONO

income. However, if education is to be cost-effective, it must be matched by available jobs (through effective demand creation and investment) to absorb the growing supply of school graduates. The mismatch between educational output and work opportunities needs to be reduced or eliminated through appropriate educational program planning and evaluation. Another useful investigation is to analyze the potential gains derived from the "permanent" labor commitment of Hawaii's labor because of social and economic barriers toward full labor mobility. The overall productivity level of Hawaii's working population may be higher than that on the Mainland because of many reasons, one of which may be the possible trade-off between the enjoyment of Hawaii's "amenities" for better compensation on the Mainland for the same type of work. Contribution

of Reproducible

Capital

This component consists primarily of business structures and equipment, non-farm residential structures, inventories and net out-of-state returns. However, only producers' equipment and construction are covered herein. Capital stock yields service returns over a period of time. Thus, the dwelling unit provides protection from the weather, comfort, prevention from illness, and other amenities which help in improvhag the productivity of workers. The contribution of capital stock is computed assuming that stock and flows are proportionate with each other. This portion of the estimates is probably among the most difficult and tentative. The average annual real growth rate of construction activity was conservatively estimated at about 5 percent. The average (net) construction investment share to state income was computed at 6 percent for the period 1900-1976. Using the basic estimation formula, (6.0) (5.0) = 0.30 percent; 0.30 percent/3.8 percent = 8 percent. Therefore, 8-10 percent of the average Hawaiian productive capacity between 1900-1976 was attributed to construction activity. In the same manner, the contribution of producer's equipment input was computed as follows: (5 percent annual rate of increase) (5 percent share of total output value = 0.3 percent 0.3 percent/3.8 percent = 8 percent Therefore,

construction

and equipment]machinery

net input contri-

GROWTH AND EQUITY IN HAWAII

373

buted about 16 percent of Hawaiian economic growth between 1900-76. The relative contribution input of structures could be underestimated. Psacharopulos computed a 35 percent contribution of capital investment coveting the period 1950-60 (Psacharopulos, 1969, p. 43) which may be appropriate for that period. These estimates need further investigation. Denison shows contribution of private structure and equipment to growth of 15 percent for the period 1909 to 1957 (Denison, 1962, p. I48). For the early period 1909 to 1929, capital input's contribution was 24 percent. From 1926 to 1957, its contribution was 12 percent. His data include inventories and net foreign assets. Putting the above together, estimates for Hawaii indicated that labor, education and capital inputs contributed 25, 21 and 16 percent, respectively, toward Hawaii'slong-term productive capacity. The balance of 38 percent, discussed below, is associated with the contribution of "residual" productivity factors including advancesin knowledge, work ethics, technological applications, economies of scale, industrial structure, management practices, and the like. Contribution of "Residual" Productivity Factors Because of their tentative nature, these estimates are considered exemplary, subject to future /mprovement by researchers. To estimate their contributions to long-term output, proxy indicators were used based on an eleven year period (1966-76). It was assumed that the trend found in these data appliesover the analytical period. This, of course, is subject to further analyses. The proxy indicators were: (1) changes in professional employment, (2) changes in membership of professional societies, (3) number of companies engaged in research and development (R/D) and, (4) changes in the volume of Hawaiian originated transpacific calls made to the Mainland, assuming that communication expansion is correlated with technological developments. Between 1966-76, professional employment, membership in professional societies and the number of R/D finns increased by 10.5 percent, 5.3 percent and 11.5 percent, respectively. Output elasticities were 16 percent, 32 percent and 15 percent, respectively. That is, a one percent change in real output is associated with a 0.16 percent change in professional employ/nent, a 0.32 percent change in professional society membership and a 0.15 percent change in the number of R/D firms. The output elasticity of transpacific *ele-

374

MITSUOONO

phonic caUs was 0.24 percent. Averaging all four items together resulted in an output elasticity rate of about 0.22 percent. That is, there was a 22 percent contribution of "advance of knowledge" to output capacity. For the contribution of specialization toward productive capacity, it was assumed that increases in the number of telephones, adjusted byreductions of telephone rates, were associated with specialization and economies of scale. Rough calculations showed a 10 percent contribution of this factor toward Hawaii's production capacity. In summary, initial data on the contribution of production factors toward Hawaii's long-term economic growth were as follows: Sources

PercentContribution to Growth Rate

Labor Inputs CapitalInputs (Construction and Equipment) Education Inputs "Residual" Advanceof Knowledge Speciahzation and Scale Economies Others

25 16 21 22 10 6 100%

Changes in Personal lncome

Size Distribution

A number of available data series indicated that the family income distribution pattern in Hawaii had become more equal over the past 75 years. However, these data are fragmented and sketchy. The long-term trend toward improved living conditions in Hawaii are noted in Table 2. The above data indicate that the overall levels of living improved for certain types of families in Hawaii as reflected by the declining Engel coefficient (a key indicator of personal well-being assuming consistency of data) and the rising share of housing expenditures. Data also show the disparity in livelihood between the middleincome Caucasian families (mostly professional and managerial workers) and plantation workers in the 1930s. Based on the Engel coefficient, the living condition of middle-income Caucasian families was better than that of agricultural plantation workers. Data show that changes in income levels could be linked with occupational

GROWTH AND EQUITY IN HAWAII

375

Table 2 - Percent of Food and Housing Expenditures to Total Expenditures Percent of Food

Percent of Housing

"Representative" families - 1901

56.3

N.A.

Wage Earner's Families in Honolulu- 1910

50.4

18.1

Filipino Plantation Worker's Families- 1933-1934

45.5

12.6

Middle-lncome Caucasian Families In Honolulu, 1937

24.3

31.3

Wage Earner's and Clerical Workers' Famih'es in Honolulu - 1943

44.1

16.6

Office Workers' Families in Honolulu, 1951

29.8

27.5

Urban Families and Single Consumers on Oahu, 1961

25.9

25.6

Urban Families and Single Consumers on Oahu, 1972-73

20.0

33.5

Family Type

Souzce: HistoricalStatistics of Hawaiiandthe State of Hawaii 1980 Data Book.

changes. Professional,technical and kindred workers and the Manager and Administrators'group made up 14 percent of the employed in 1940.However, in 1970, thisratewas 23 percent.More important, the proportion of farm laborersand foremen among the employed was 28 percent in 1940. In 1970, thisratewas reduced to 2 percentof the employed. Using this indirectapproach, the 62 percent proportion of workers ten years old and over in agriculture in 1900 can bc related tothe comparable 3 percent rate in 1975. Unquestionably, this shiftfrom agricultural to non-agricultural pursuitsover the past 75 years has resultedin improved levelsof livingof workers in Hawaii. In the early 1900s, 25 percentof the totalpopulation(I54,000) were Hawaiians and part-Hawaiians(of which, presumably, most would be in the lower income categories). Few of them received significant incomes from land holdingsand other inheritances. The Caucasian group could be divided into two income-levelsectors,

376

MITSUOONO

immigrants (1) working in the management of mercantile trade activities and the sugar plantations, and (2) working as plantation laborers. The Caucasian group comprised about 17 percent of the total population, of which a small group could be included in the higher income category. The balance of the population (approximately 26,000 Chinese, 61,000 Japanese and others) were mostly immigrant agricultural workers. Thus, this population profile of the early 1900s would indicate a rather skewed two income sector distribution, a relatively small upper income sector and a very large lower income sector. If these data could be converted into a summary income distribution measure, it would show a high rate of relative income inequality during this early 1900 period. During the 75 years, income equity appears to have improved. For example, a good proxy measure of income equity is the percentage of housing units which are owner-occupied. In 1930, about 18 percent of housing units were owner-occupied. In 1976, the comparable rate was 45 percent. Another measure associated with long-term improvements in income distribution is the Engel Coefficient. It dropped from about 60 percent in the 1900s to about 20 percent in the mid-1970s. These reflect the broadening of the personal income base during the past seventy-five years. There are recent family income size distribution data from the Census Bureau and other sources which could be used to compute summary income distribution measures. Thus, Professor Oshima's standard deviation of quintile share can be computed for 1949, 1959 and 1975 (Table 3) (Oshima and Ono, 1965, p. VIII-15). Between the 26 year-period (1949-75), the standard deviation of quintile share declined by almost 25 percent. Within the same period real state income almost tripled, per worker income increased one and a half times and per capita income almost doubled. Key factors making for this equalizing trend are noted below. An important clue is the average number of earners per family, including working wives. Hawaii had a female labor force participation rate of 59.4 percent in 1976, the highest in the Nation (47.3 percent for the Mainland). Although per worker income is relatively lower compared with other states, Hawaii's family income is much higher because there are more earners per family. A simple model can be formulated to indicate this employment-income pattern in Hawaii. As noted, changes in output (supply) can be related with changes in the quantity of factors of production and changes in the

GROWTH AND EQUITY IN HAWAII

377

Table 3 - Standard Deviation of Quintile Shares, Hawaii, 1949, 1959, 1975 Quintile of Families

1949

Income Shares (in percent) 1959 1975

Lowest

20%

5.3

5.9

4.0

2nd

20%

11.0

11.6

11.0

3rd

20%

15.4

15.9

20.0

4th

20%

21.0

22.4

29.0

Highest 20%

47.3

44.2

36.0

Standard Deviation of Quintile Share (in percent) Hawaii U.S. Mi__i_dppi

14.7 13.9 17.0

13.3 13.3 16.3

11.1 12.4 (E) 15.2 (E)

E - Estimated Sources: Hawaii's Income and Expenditures, 1958, 1959 and 1960 and State of Hawaii's Data Book, 1978.

productivity of these factors. Changes in the demand for these production factors, in turn, depend on the income elasticities of demand. In Hawaii, as the demand for agricultural products, services and construction activities expanded, employment demand was met by a relatively well-trained, disciplined work force, willing and able to work. This resulted in overall rising incomes and improved living conditions for most of the people in Hawaii. An accessible and quality-driven educational system produced a broadly trained work force which could easily move into expanding job opportunities as they became available. With the mechanization of household operations, the prevalence of the extended family system and the drive toward better living standards, wives entered into the work force as opportunities arose. With more earners per family, family incomes rose, allowing the purchase of more consumer durables, in turn improving household work efficiency and allowing household members to work. As Professor Oshima noted (Oshima, 1961), higher incomes, economies of scale and consumer asset formation, â&#x20AC;˘together with the underlying institutional forces, raised household â&#x20AC;˘ saving rates. On factor demand in Hawaii, there was continuous

378

MITSUOONO

upgrading of the industrial and occupational employment structure. As noted, agricultural employment declined while output per worker significantly rose due to better technological application and managerial practices. For example, between 1950 and 1957, output and capital per person in the sugar plantations increased by 54 percent and 48 percent, respectively (MoUet, 1961). Accompanying all of these dynamic changes were repidly rising educational levels. The median school year completed by the population almost doubled from 6.9 in 1940 to 12.7 years in 1975. If the standard deviation of quintile share is computed to reflect the income distribution pattern in the early 1900's, a guess would be that it would be about double in 1900 than that in 1975 (a standard deviation of about 24 percent in 1900 relative to an 11 percent rate in 1975). Between 1901 and 1973, the Engel Coefficient declined by about 64 percent. Using elasticities between the EngelCoefficient and the Standard Deviation of Quintile Share between the 12 year period, 1961-1973, it was estimated that a one percent drop in the Engel Coefficient was associated with a 0.74 percent decline in the Standard Deviation of the Quintile Share. Thus, assuming that this relationship is applicable between 1901-1951, an approximation was made that in 1901, the inequality rate fluctuated around a 24 percent level. Summary of Findings A. Contribution of Factor Input Changes 1. Between 1900 and 1976, state income (SI)increased from $38 million (in current $) to $6,239 million. In the same period, the GSP deflator increased from 17.4 to 161.7 (1967= 100). Deflating the SI by these deflators resulted in SI values of $219.0 million and $3,852.5 n_illion (in 1967 $) in 1900 and 1976, respectively. These SI values showed an 18-fold increase, averaging about a 3.8 percent annual increase over the 76 years. This 3.8 percent annual average growth rate is the capacity benchmark used for all future computations. 2. Between 1900 and 1976, the average labor share of SI was 73 percent. The estimated labor input (employment adjusted for hours) showed a 1.3 percent average annual increase. Thus, (1.3%) (0.735) = 0.95 percent. Labor input then contributed

GROWTHAND EQUITYINHAWAII

379

about 25 percent of the average annual growthrate over the 76 years (0.95/3.8). 3. Between 1900 and 1976, the median school years completed for persons were 3.6 years and 12.7 yeats, respectively. Average income of employed persons increased from $1,621 to $4,123 (in 1967 $) between these two periods. The average annual growth rates were 1.7 percent and 2.2 percent respectively. Taking 2.2 percent and applying a 50 percent rate to isolate "pure" educational attainment contributions (excluding improvements in other factors such as training, ability, and so on) the contribution of education to the growth rate was 21 percent. ThuS, (1.1) (0.73) = 0.08; (0.08/3.8) = 21 percent. 4. Between 1900 and 1976, then, the quantitative (24 percent) and "qualitative" (21 percent) contribution of labor amounted to 45 percent. Denison found for the U.S. (1962, 148) for 1909-57 that the comparable figures were about 34 percent for employment and hours (0.97/2.89), 20 percent for education (0.58/2.89) and about 3 percent for better labor utilization, sex age composition changes and so on. 5. Between 1900 and 1976, the average annual rates of equipment and construction inputs into the Hawaiian economy were 5.0 percent and 4.8 percent, respectively. After weighing each of these two rates by 6 percent (after depreciation), the contribution of net construction investment and producer durable equipment expenditures were about 0.3 percent each per annum. These rates (0.3 percent each) were then divided by 3.8 percent resulting in the following contributionto the average annual growth rate between 1900 and 1976: Equipment Construction

= 8 percent = 8 percent 16 percent

Denison shows (1962, 148) that the contribution of capital inputs of structures and equipment to the growth rate during the period 1909-57, was about 15 percent (0.44/2.89). The net changes in impact of inventories and Hawaiian owned assets abroad on the long-term Hawaiian growth rate was

380

MITSUOONe

assumed to be relatively small. This needs to be analyzed, however. Thus, between 1900 and 1976, labor input changes accounted for 46 percent while capital input changes accounted for 16 percent of the long-term (76 years) improvement in Hawaiian production capacity. This leaves a residual of 38 percent to be explained in factor productivity increases due to advances of knowledge, economies of scale, shift from agriculture to nonagricultural employment, better labor utilization, and other reasons.

B. "R esidual" Productivity Applying output/factor

Factors elasticity rates, findings were as follows:

1. Of the residual (factor productivity), advance of knowledge and technological applications could have contributed 22 percent to Hawafi's economic growth. 2. Specialization could have contributed about 10 percent toward Hawaii's economic growth. â&#x20AC;˘ 3. The residual including reaUocation of resources (farm to nonfarm employment) and other variables was reduced to 6 percent for further analysis. C. Changes in lncorne Distribution

of Farnilies

1. In 1975, the Standard Deviation of Quintile Share (SDQS) was 11 percent, a 25 percent decline from the 1949 measure of 14.7 percent. 2. Using Engel Coefficient elasticity rates, it was guessed that the SDQS could be conservatively about 24 percent during the early 1900 period. Thus, between 1900-1976, the SDQS was cut about half, from a 24 percent rate to a 11 percent rate. 5. Concluding Remarks According to the "notional" estimates used herein to indicate what possibly could have contributed toward Hawaii's long-term growth capacity, one of the key contributors appears to have been technological applications. The critical role of technological applications in economic growth is widely accepted. In Hawaii, during the past 75 years, inventions, and improvements in sugar and pineapple production and processing have significantly improved productivity

GROWTH AND EQUITY IN HAWAII

381

within the sugar and pineapple industries. However, as evident by present trends, these early advantages are not permanent for Hawaii. Its primary resource for further development is still in its skilled manpower base. Further expansion of Hawaii's international linkage with the Pacific basin countries in technical and cooperative arrangements should result in greater and better opportunities for the people of Hawaii. In this regard, Hawaii's academic institutions play a key role for this further development. As Professor Oshima notes, further analysis is required to study the relationship between economic growth, technological and institutional innovatioas. This interaction is seen in the development of agricultural industrialization in Hawaii. Thus, with assured market demand (expectations of future growth), early industrial leaders were willing to invest not only in physical capital and in research and development, but also in human capital. The commitment of investment requires expectations of adequate future gains. This has led to entrepreneur decisions to take technological risks, given the skilled manpower base. This relationship between Hawaiian economic growth, assured market demand for its goods and services, human investment, institutions and technological innovations and applications needs further investigation and Hawaii is an ideal location for this research. Finally, as noted, this preliminary investigation is based on fragmented and rather inadequate data base. If this type of analysis is to be expanded and improved in order to project Hawaii's future economic growth patterns, the statistical foundation of these data needs to be further strengthened.

REFERENCES Abramovitz, M. (1962), Economic Review,

"Economic 762-782.

Growth

in the United

State,"

American

Christensen, L. R. and Jorgensen, D. W. (1972), "Measuring the Performance of the U. S. Economy, 1929-1969," Discussion Paper No. 265, Harvazd Institute of Economic Research. Crawford, D. (1933), Paradox in Hawaii, Stratford'Co. Dearie, P. (1968), "Book Review - Why Growth Economic Review, 980-982.

Rates Differ,"

American

Dcnison, E. (1962), $ourees o/Economic Growth in the United States, Supplementary Paper 13, Committee for Economic Development, New York.

382

MITSUO ONO

, (1967), ton D. C.

Why Growth

Rates

Differ,

Brookings Institution,

Washing-

____ 1969,

(1974), Accounting for United States Economic Growth, 1929Brookings Institution, Washington D, C. , and Chung, E. (1976), How Japan Economy Grew So Fast, Brookings Institution, Washington D. C. Hawaii State Department of Planning and Economic Development (1976) Hawaii's Income and Expenditures Accounts: 1958-76, Volumes I and II, Honolulu, Hawaii. Hawaii State Department of Planning and Economic Development (1978), The State of Hawaii Data Book. Honolulu. Hawaii State Department of Planning and Economic Development (1980), The State of Hawaff Data Book, Honolulu. Hitch, T. (1962), "A Brief Political, Social and Economic History of Hawaii," First Hawaiian Bank, Honolulu. , (1971), "Lessons Taught by Recent Hawaiian Hisotry," First Hawaiian Bank, Honolulu. Kendrick, J. (1961),Productivity Trends in the United States, Princeton: Princeton University Press. Kuykendall, R. S. and Day, A. C., (1976), Hawaii, A History from Polynesian Kingdom to American Statehood, Englewood Cliffs, New Jersey: Prentice Hall, Inc. Mak, J. (1978), "A Concise Economic History of the Hawaiian Islands," Economic Department, University of Hawaii, Honolulu. MoUet, J. A. (1961), Capital in Hawaiian Sugar: Its Formation and Relation to Labor and Output, 1870-1957, Agricultural BulletIn 21, University of Hawaii, Honolulu. Morgan, T. (1948), Hawaii, Century o/Economic Change, Cambridge: Harvard University Press. Oshima, H. T. (1961), "'Consumer Asset Formation and the Future of Capitalism," Economic Journal. Oshima, I4. T. and Ono, M. (1965), Hawaii's Income and Expenditures, 1958, 1959 and 1960, Volumes I, II and III, Economic Research Center, University of Hawaii, Honolulu. Psacharopolus, G. (1969), The Anatomy of a Rate of Growth: The Case of Hawaii, 1950-1960, Economic Research Center, University of Hawaii, Honolulu. Sehmitt, R. C. (1977), Press of Hawaii.

Historical

Statistics

of Hawaii,

Honolulu:

University

United

States Government Printing Office (1975), Historical Staffstics of the United States, Washington D. C. United States Government Printing Office (1976), Staffstical Abstract, Washington D. C. United

States Government from 1976 to 1987: XXII, Joint Economic

Printing Office 0977), "U.S. Economic Growth Prospects, Problems and Patterns," Volumes I th_ Print, 95th Congress, 1st Session, Washington D. C.

ASIAN URBANIZATION AND DEVELOPMENT: A COMPARATIVE VIEW Ernesto M. Pernia*

1. Introduction There has been a growing recognition that urbanization is a problem to be reckoned with during the 1980s and beyond. The reason is not because there is something wrong or unusual about urbanization per se which is a feature that goes with development. Rather, the reason seems to be that urbanization, like economic development itself, continues to be unbalanced spatially or concentrated in the metropolitan centers of LDCs. There is a sense that the great majority of the people outside these centers have been practically left out from the benefits of urbanization and development. There is also the common observation that the carrying capacity of urban centers is bursting at the seams, so to speak. These and other concerns appear to have become prominent in recent years. 1 This paper offers a comparative perspective on Asian urbanization in relation to development. An analysis of certain indicators of urbanization and spatial concentration may provide a clue to the particular aspects of the "urbanization problem" we should be concerned with. The focus of the paper is South, Southeast and East Asia, and the constituent countries exclude city-states (Hongkong *Professorof Economics,Universityof the PhilippinesSchoolof Economics. I am grateful to ProfessorHarry T. Oshima who gave useful suggestions.Researchassistancewasprovidedby RosarioGulinaoandtypingby Ana R. Aureo. 1. See, e. g., the reportof the U. N. EconomicandSocial Council(1978) andTodare (1979). 383

384

ERNESTOM. PERNIA

and Singapore) and countries in turmoil (Cambodia and Vietnam) or with inadequate data (Nepal). In addition, two centrally planned Asian countries, namely, the People's Republic of China (PROC) and the People's Republic of Korea (North Korea) are included to increase the range of experiences. 2 The trends for the different Asian regions are first presented'in the context of the world's more developed and less developed regions. Comparative data on the constituent countries in each of the regions are next shown. Then a modified urbanization-development model is proposed and subsequently tested empirically. The concluding section summarizes the findings and implications. 2. Asian Regions in Context According to data from the United Nations (1980), the world in 1980 was about 41 percent urban; more developed regions were 70 percent urban and less developed regions 30 percent urban. In absolute terms, these translate to 1,806 million urban population in the world as a whole, 834 million in more developed regions and 972 million in less developed regions. Against this background, one can situate the Asian regions in 1980 with the following statistics (from Tables 1 and 2):

SouthAsia SoutheastAsia CentrallyPlannedAsia East Asia

PercentUrban

UrbanPopulation (in millions)

22,0 22.7 26.1 72.5

201.1 61.4 241.4 112.9

The data readily indicate extremes in urbanization regions. At one end is East Asia which corresponds average for the more developed world, and at the South, Southeast and Centrally Planned Asia which mean for the less developed world and fax below the world as a whole. 3 The majority of Asia is thus

levels in these closely to the other end are fall below the average for the still relatively

2. professor Oshimahas written important treatises (1978, 1980, 1981) on the economic performance of, and prospects for, Asian countries. The present papercouldperhapsserveas a complement to those treatises. 3. The less developedworld averal_eis actually pulled up by Latin America whose urbanizationlevel is closer to the more developed worldthanto the lessdevelopedworld.

ASIANURBANIZATIONAND DEVELOPMENT

385

unurbanized, reflecting the low level of development in these regions. This is particularly true of South Asia and Southeast Asia which are less than a quarter urban. The relatively unurbanized status of Asia is the result of its slow pace of urbanization even in recent decades. This is contrary to the common impression that Asia has a problem of rapid urbanization. If anything, the problem seems to be more that Asian regions have been urbanizing rather sluggishly as evinced by the following comparative data (from Table ][) on rates of urbanization 4 (in percent) over three decades: 1950-60

1960-70

1970-80

South Asia SoutheastAsia CentrallyPlannedAsia East Asia

11.3 20.3 82.8 53.3

14.5 16.0 21.4 45.8

19.4 19.0 24.3 46.8

World

25.7

17.0

17.3

MoredevelopedRegions Less developedRegions

28.5 39.3

28.7 24.3

28.3 26.4

South Asia's rate or speed of urbanization has been the slowest and that of Southeast Asia has been practically the same especially in the 70s. These rates resemble the world average but are lower still than the mean for less developed regions, s Centrally Planned Asia's urbanization has been faster than South and Southeast Asia (unusually fast during 1950-60) and close to the less developed world average. The remarkable _performance is that of East Asia whose speed of urbanization has been over 50 percent faster still than the average for the more developed world. The pattern of urban population growth is quite the reverse. Southeast Asia manifests the highest rate of urban growth, approximating the average for the less developed regions, followed closely by South Asia. What is more striking is the pattern of rural population 4. Rate of urbanizationis here definedas the percentagechangein urbanruralratio rather than the change in proportion urban. The former measure is superiorbecauseit does not havean upperlimit of I. 5. There is also evidence to show that the rate of urbanizationin LDCsis not rapid compared to the hist_ experience of WesternoounUies(see Davis 1975, Pernia1976, Preston 1979).

386

ERNESTO M. PERNIA

growth. The growth rates for South and Southeast Asia are very high, relative to the average for the less developed regions as weU as for the world as a whole. But the real contrast is with East Asia and the more developed throughout the rural population

regions whose rural growth rates have been negative three decades. The comparative rates of urban and growth (from Table 2) are (in percent): 1950-60

1960-70

1970-80

Urban

Rural

Urban

Rural

Urban

Rural

South Asia Southeast Asia Centrally Planned Asia East Asia

33,5 47.3

20.0 22.2

42.8 48.7

24.6 27.9

47.6 52.3

23.7 28.3

7.1 7.7

37.9 32.9

13.5 -8.8

39.0 29.5

11.8 -11.8

World

39.8

11.1

33.9

14.3

33.4

13.8

22.7

-4.6

18.7

-7.5

48.3

19.1

49.3

18.2

More developed Regions Less developed Regions

95.5 41.5

27.6 59.6

-0.8 14.6

It is clear that in purely demographic terms, the high rate of rural population growth is slowing down the pace of urbanization in Asia (except East Asia) and in the less developed world (despite high urban growth rates). If the urban-rural growth difference (URGD) is computed, one would see the same interregional pattern as that for rates of urbanization (Table 2). 6 South

Asia

This region, as already mentioned, is predominantly rural. It was 16 percent urban in 1950 and even in 1980 only 22 percent urban. The countries in this region are among the lowest in terms of levels of income and their growth rates. Recent data on levels of urbanization, industrialization and GNP per capita countries (from Tables 1 and 5) are as follows: 7

for individual

6. URGD is also used to measure speed of urbanization. 7. Industrialization level is here indicated by manufacturing share of GDP since this is the most dynamic component of the industrial sector. Data on GNP per capita are taken from the World Bank (1980).

ASIANURBANIZATIONAND DEVELOPMENT

387

GNPper capita Urbanization Industrialization (1980) (1978) (1978) (1960-78 annual change) Bangladesh Burma India SriLanka Pakistan

11.2% 27.2 22.3 26.6 28.2

8% 10 17 23 16

90 US $ 150 180 190 230

-0.4% 1.0 1.4 2.0 2.8

The degree of concentration (proportion of urban population in largest city)_in these countries in 1980 ranged from a low of 6 percent for India to a high of 30 percent for Bangladesh (Table 3). Urban concentration has remained more or less stable in India and in Pakistan (at 21 percent) but has markedly risen in Bangladesh from 20 percent in 1960. By contrast, Sri Lanka has evinced remarkable deconcentration from 28 percent in 1960 to 16 percent in 1980 despite the presence of only one city of over 500,000 inhabitants. Southeast

Asia

The region as a whole has exhibited practically the same urbanization trend as South Asia although all countries are now classified by the World Bank as middle-income countries. Indonesia used to belong to the low-income group of countries until recently. Comparative data on urbanization, industrizliT.ation and GNP per capita for individual countries (from Tables 1 and 5) are shown below: Urbanization (1980) Indonesia Thailand Philippines Malaysia

20,2% 14.4 36.2 29.4

Industrialization GNPpercapita (1978) (1978) (1960-78 annual change) 9% 18 25 17

360 US $ 490 510 1,090

4.1% 4.6 2.6 3.9

The income levels as well as their growth rates are significantly higher in Southeast than in South Asian countries. Thus, if the link between urbanization and economic growth continues to hold, Southeast Asian countries would probably accelerate in urbanization in the coming years, at least relative to South Asian countries.

388

ERNESTO M. PERNIA

Urban concentration (proportion of urban population in largest city) is very pronounced in the region, ranging from 23 percent in Indonesia to 69 percent in Thailand (Table 3). And this indicator has been steadily rising in all four countries, as can be seen below:

Indonesia Thailand Philippines Malaysia

1960

1970

1980

20 65 27 19

22 68 29 23

23 69 30 27

It may be hypothesized that the exceptionally marked urban concentration or primacy in Southeast Asian countries is not unrelated to the import-substitution industrialization strategy pursued by these countries in the 1950s and 60s. s This point will be discussed further and partial support for the hypothesis will be shown in subsequent sections. Centrally

Planned

Asia

This region includes two countries: PROC whose level of urbanization appears similar to some countries in South and Southeast Asia, and North Korea which resembles more the countries in East Asia than elsewhere. By World Bank income standards, PROC would be considered a low-income country and North Korea a middle-income country, as denoted by the following data (from Table 1): GNP per capita Urbanization (1980)

(1978)

(1960-78 annual change)

PROC

25.4%

230 US$

3.7%

North Korea

59.7

730

4.5

8. For a discussion of import-substitution policies widely adopted among Southeast Asian countries, see Myint (1972). While there has been a shift away from these policies, their spatial impacts probably continue to linger up to the present. An additional reason for the extreme urban concentration in Thailand may be a geographical one: the lack of good harbors in coastal areas to service big cities other than Bangkok.

ASIANURBANIZATION ANDDEVELOPMENT

389

Another point that may be noted is that the economic growth performances of both countries compare well with those of the high performers in Southeast Asia. The remarkable characteristic that seems to set these two countries apart from the other Asian countries is the relative absence of urban concentration. PROC exhibited only 6 percent urban concentration from 1960 to 1980 while North Korea had 15 percent concentration in 1960 which declined to 12 percent in 1980. It would seem that such relative lack of concentration is due to central controls on population movements. East Asia The countries in this region are among the great economic performers of the postwar era: Japan in the 50s and 60s, Taiwan in the 60s and 70s and South Korea in the 70s (see, e.g., Oshima 1980). The average annual growth rate of GNP per capita in these countries from 1960 to 1978 was in the vicinity of 7 percent (see also Table 4). It is not surprising, therefore, that they have also experienced very rapid urbanization rates of over twice those manifested by the other Asian countries. By 1980, more than half of the population in South Korea was urbanized, and over three-fourths of both Taiwan's and Japan's populations were urbanized. The growth rates of rural population in these countries have been negative for some time already. Data on 1980 degree of concentration show that 41 percent of South Korea's urban population is in Seoul, while for Japan, 22 percent is in Tokyo. The relatively low concentration in Japan may be attributed to its policy of regionalization of industrial development and more developed system of transportation and communication. The high concentration in South Korea may be partly explained by its heavy industrialization-cure-protection policy-in a way similar to the phenomenon in Southeast Asian countries. 3. Urbanization and Development The level of urbanization at a point in time, its pace over time, and the degree of concentration are indicative of the current and future scale of the urbanization problem. These are among the major indicators of concern relative to the urbanization issue. From the previous discussion of experiences acrossAsian regions and countries

390

ERNESTOM. PERNIA

within each region it appears that urbanization is closely related to economic development. What needs to be done now is to determine the principal correlates of urbanization. The Asian countries included in this study portray varied experiences and circumstances such that a cross-sectional statistical analysis should throw some light on the urbanization-development nexus. Specifically, what this crosssectional analysis should do is to identify the factors that account for the variation in urbanization levels and rates, as well as in the degrees of concentration across Asian countries. On the basis of standard development theory (e.g., Lewis 1954, Ranis and Fei 1961), it is commonly supposed that overall development of the economy as well as developments in both the agricultural and industrial sectors determine urbanization in a fundamental way. Agricultural development tends to release farm labor and population over time which are then attracted to the urban-industrial sector. Thus, it has been assumed that the speed of rural-urban transformation is directly conditioned by agricultural and industrial developments. This seems to be the traditional view. Recent data on LDCs, however, suggest that rapid population growth tends to retard the urbanization process. The relationship may be hypothesized to operate in two ways. In the first place, where overall population growth is high, it is usually pronouncedly higher in the rural sector than in the urban sector, and this has the direct demographic effect of dampening the rise in the proportion urban. In the second place, population growth terids to hamper economic development and thus, indirectly, the urbanization process itself. It therefore seems warranted to expand the standard urbanization-development model by adding the population growth variable. Concerning the degree of urban concentration, the paper's hypothesis is that it is also influenced by industrial development (or manufacturing activity) and population growth. In addition, degree of openness of the economy would play a crucial role inasmuch as importation of goods and services requires licenses and foreign exchange which are more easily obtainable in the capitar city. Likewise, most other support services for manufacturing are found in the metropolis. There is clearly then a strong incentive for industries and business concerns to locate in the capital metropolis which in most cases is also the capital port of the country. This is all the more so in developing countries where transportation and communications are deficient (Alonso 1968). The spatial coincidence of the capital

ASIANURBANIZATIONAND DEVELOPMENT

391

metropolis and the capital port is thus advantageous for manufacturing activity with its import requirements. As is known, importintensive industrialization characterized many Asian economics during most of the post-war era. Data, Notations,

and Results

The data employed in the regression exercise pertain to the South, Southeast, Centrally Planned and East Asian countries considered in the previous discussion. The data are reported in the most recent publications of the United Nations (1980) and the World Bank (1980) (see Tables 1 through 6). To increase the number of cases, cross-section observations were pooled for 1960, 1970, and 1980 (or 1978). The variable notations and their specifications are as follows: URB t

RURBt_I,

ffi level of urbanization

at time t, specified as proportion ur.b.an ) urban-rural ratio (or 1- proportion urban rather than simply proportion urban which has an upper limit of 1.

rate (or speed) of urbanization during some interval, specified as percentage change in URB.

CONC t

= degree of concentration at time t, specified as L 1 - L ' where L denotes the proportion of urban population in the largest city.

IND t

ffi industry share of GDP at time represents economic level.

t, which

GRAGt_I,

= average annual production.

GRMANt-

t,t

= average annual growth rate of manufacturing production.

GRPOPt- 1,t OPEN t

ffi average annual growth rate of population. = degree of openness of the economy, specified as the import share of GDP.

The regression results

growth rate of agricultural

correspond to three dimensions

of an

392

ERNESTOM.PERNIA

urbanization-development model explaining: (1) level of urbanization, (2) rate of urbanization, and (3) degree of concentration. 9 (1) URB = -1.249

(1') URB =

+ 1.669 IND - 0.732 GRAG + 0.234 GRMAN (4.683) (2.211) (0.917) R2 - 0.66

0.559 + 1.292 IND(3.494) - 1.129 GRPOP (2.146)

0.533 (1.685)

CRAG

GRMAN (1.178) R e -- 0.73

Equation (1) shows that level of urbanization is significantly conditioned positively by economic level (IND) and negatively by agricultural growth (GRAG).I 0 A 1.0 percent increase in economic level brings about a 1.7 percent change in urbanization level; on the other hand, a similar change in agricultural growth pulls down urbanization level by 0.7 percent. Manufacturing growth (GRMAN) has a positive effect on urbanization but is not significant. Equation (1') is an enhanced model with population growth (GRPOP) added as an explanatory variable. GRPOP has a significant negative influence on URB and the overall explanatory power,of the model increases from 66 percent to 73 percent. This result lends strong support to the hypothesis. The results for rate of urbanization (specified in semi-log form) are as follows: (2) RURB = 3.910 - 0.006 IND - 0.415 CRAG + 0.110 GRMAN (0.483) (3.716) (3:288) R 2 = 0.48 (2') RURB -- 5.270 - 0.021 IND - 0.313 GRAG + 0.097 GRMAN (1.632) (2.898) (3.220) - 0.556 GRPOP (2.351) R 2 = 0.61 Equation (2) parallels equation (1) but the dependent variable is expressed as speed of urbanization over time. Economic level (appropriately lagged as IND t_ i ) has the reverse sign as expected but is now insignificant. The negative sign simply means that urbaniza9. t-values are enclosed in parenthesesunderneathregressioncoefficients. I0. Both equations(I) and(I f) arein double-logformulations.

ASIANURBANIZATIONAND DEVELOPMENT

393

tion tends to slow down at higher economic levels. Agricultural growth (GRAG) continues to be negative and significant, and manufacturing growth now exhibits a significant positive effect. Equation (2') is likewise analogous to equation (I') with the added population growth variable (GRPOP) once again figuring importantly with its negative sign, and raising the explanatory value of the model by 13 percent. 11 The negative effect of agricultural growth on urbanization in all four regressions, though contrary to standard urbanization-development theory, seems to reflect absorption of labor in agriculture which would otherwise migrate to urban areas. The last regression results have to do with urban concentration (in double-log): (3) CONC = 1.914 - 0.055 URB + 0.682 GRMAN (0.203) (2.501)

(3') CONC=

1.020-

0.192 URB + 0.433 GRMAN(0.827) (1.761)

+ 0.889 OPEN (2.822)

+ 0.655 GRPOP (0.802) R 2 = 0.34 0.096 GRPOP (0.130) R 2 ---0.56

Among the independent variables in the previous equations, GRMAN and GRPOP were picked for both theoretical and statistical significance reasons (equation 3). URB (similar to 1ND) is included as a control variable but is not significant. Equation (3') shows that adding degree of openness (OPEN) raises the R 2 by 22 percentage points. All the signs are in accord with expectations although they are not significant for URB and GRPOP. The important thing to note, however, is the significance of the variable OPEN- a 1.0 percent increase in degree of openness raises urban concentration by about 0.9 percent. This result strongly supports the hypothesis that openness of the economy to the foreign sector is a strong incentive for concentration in the principal port and city of the country.

11. We also experimented with 2-SLS regressions to deal with possible simultaneitybias but the resultswerenot useful,.

394

ERNESTOM. PERNIA 4. Conclusion

Asia is still predominantly rural- a reflection of both its low level and pace of development. From within this vast region, however, East Asia has sprung forth as a great achiever (at least in a relative sense) in both urbanization and development so that it can now be better associated with advanced countries than with developing Asian countries. Whether or not South and Southeast Asian countries will follow the trajectory of East Asian countries would depend on many things. The empirical results of an expanded urbanization-development model suggest that, in addition to manufacturirtg activity and agricultural development, population growth plays a crucial role in urbanization.12 Population growth seems to result in a slowing down of the urbanization process. Hence, if population growth is going to decelerate in South and Southeast Asian countries, ceteris paribus, one could expect faster urbanization in the coming decades. Another important point to consider is that agricultural development appears to retard urbanization, perhaps because it allows for labor absorption in the rural sector which would otherwise migrate. This could be the effect of agricultural growth at low levels of economic development. It is possible that at higher levels, agricultural development would have the reverse consequence, as observed, for example, in industrialized countries. In any case, the negative relationship between agricultural development and urbanization observed for Asian countries lends further support to the notion that rural/agricultural development can reduce unwarranted migration to cities. Urban concentration or primacy seems moderate in South Asian countries but high and rising in Southeast Asian countries, including South Korea. It is virtually negligible in the Centrally Planned countries of PROC and North Korea for obvious reasons. There is no clear development-concentration relationship, however, even if the exceptional cases of PROC and North Korea are set aside. Countries like Thailand and South Korea have extremely high concentration ratios but differ substantially with respect to urbanization and development levels. Then there is India which has little concen12. Needless to say, one should be cautious about using the results of cross-sectionanalysisfor predictingfuturetrends.

ASIAN URBANIZATIONAND DEVELOPMENT

395

tration, and Bangladesh which is less urbanized and developed than India but has a moderate degree of concentration similar to Japan. It would seem, therefore, that there are other factors that account for urban primacy differentials (after allowing for measurement problems). The analysis suggests that degree of openness of the economy in addition to manufacturing growth, is a significant determinant of the primacy phenomenon. The reason behind manufacturing growth is known: manufacturing activity has invariably been concentrated in the metropolitan capitals of many Asian countries. The finding on degree of openness bears out the hypothesis that concentration in the metropolis is a response to the need to be near the principal port as well as to offices that issue licenses and foreign exchange, among other things. Thus, spatial concentration appears to be partly an unintended consequence of macroeconomic and growth policies in the past, salient among which was the now-famous import-substitution industrialization strategy. This point seems worth noting in the design of urbanization and spatial development policies for Asian countries.

Tal_e I - Urba,'dzationIndicators for Selected Asian Regions/Countries, 1950-1980 Region/ Country'

Percent Urban

Percent Change in Urban-Ruzal Ratio

U_ban-Rural Ratioa

1950

1960

1970

1980

1950

1960

1970

1980

1950-60

1960-70

1970-80

15.7

17,1

19.1

22.0

0.186

0.207

0.237

0.283

11.3

14.5

19.4

Bangladesh Burma India Sri Lanka Pakistan

4.4 16.1 16.8 14.4 17.5

5.2 19.3 17.9 17.9 22.1

7.6 22.8 19.7 21.9 24.9

11.2 27.2 22.3 26.6 28.2

0.046 0.192 0.202 0.168 0.212

0.054 0.239 0.218 0.218 0.284

0.082 0.296 0.245 0.280 0.331

0.127 0.373 0.286 0.362 0.392

17.4 24.5 7.9 29.8 34.0

51.8 23.8 12.4 28.4 16.5

54.9 26.0 I6.7 29.3 18.4

Southeast Asia

15.0

17.5

19.8

22.7

0.177

0.213

0.247

0.294

20.3

16.0

19.0

12.4 10.5 27.1 20.4

14.6 12.5 30.2 25,2

17.1 13.2 32.9 27.0

20.2 14.4 36.2 29.4

0.142 0.117 0.372 0.256

0.171 0.143 0.432 0.337

0.206 0.152 0.491 0.369

0.253 0.168 0.568 0.416

20.4 22.2 16.1 31.6

20,5 6.3 13.7 9.5

22.8 10.5 15.7 12.7

44.6

55.2

64.2

72.5

0.803

1.231

1.795

2.635

53.3

45.8

46.8

2!.4 50.2

27.7 58.0 62.4

40.7 71.3

54.8 77.0 78.2

0.272 1.008

0.383 1.381 1.659

0.686 2.484

1.212 3.348 3.596

40.8 64.6

79.1 49.7

76.7 44.8

SouthAsia

Z O

Induaesia Thailand Philippines Malaysia East Asiab South Korea Taiwan Japan

Centrany Planned Asia

11.3

18.9

22.1

26.1

0.128

0.234

0,284

0.353

82.8

21.4

24.3

PROC North Korea

11.0 31.0

18.6 40.2

21.6 50.1

25.4 59.7

0.124 0.450

0.228 0.672

0.276 1.003

0.341 1.481

83.9 49.3

21.0 49.3

23.6 47.7

29.0

33.9

37.5

41.3

0.408

0.513

0.600

0.704

25.7

17.0

17.3

52.5

58.7

64.7

70.2

1.107

1.423

1.831

2.350

28.5

28.7

28.3

16.7

21.8

25.8

30.5 0.201 0.280 l_portton urban _Ratio of urban population to rural popu_t_m or l-p_oportion urban. ORegional average for East Asia excludes Tarwan.

0.348

0.440

39.3

24.3

26.4

World More Developed Regions Less Developed Regions

_2_ eÂ˘

_. ,.1

Source: Table 2 of this paper.

t-' 0

,-t

Table2 - UrbanandRuralPopulalioa_, andGt_ow_b llxte4t: A_an Regto_s/Cotmtlies_ 195080 Urban Population (in millions)

Region/ Country South Asia l_n_t_desh Burma India Sri Lunka Pakistan "southeast Asia lndoneda _d Phifippines Malaysia East Asiaa South Korea Taiwan Japan

1950

1960

1970

Percent Growth of UrtmuPopulation

1980

1950-60 1960-70 1970-80

RuralPopulation (in mifiions) 1950

1960

1970

Percent Growth of Rural Population 1980

Urban-Rmal Growth Diffeaence

1950-60 1960-70 1970-80 1950-60 1960-70 1970-80

71.5

95.4

136.3

201.1

33,5

42.8

47.6

384.7

461.8

575.3

711.6

20.0

24.6

23.7

13.5

18.2

23.9

1.8 3.0 59.2 1.1 6.4

2.6 4.3 76.6 1.8 10.1

5.1 6.3 107.0 2.7 15.0

9.5 9.6 154.5 4.1 23.4

48.3 44.6 29.2 60.2 58.7

94.4 47.8 39,7 54,4 48.4

85.1 50.9 44.4 50,1 55.3

39.2 15.4 293.4 6.6 30.I

48.8 18.0 351.2 8,1 35.7

62.5 21.4 436.1 9.8 45.4

75.3 25.6 539.8 II,4 59.6

24.3 16.6 19,7 23.5 18.8

28,2 19..2 24.8 20.5 27.1

20.4 19.7 23.8 16.1 31.2

24,0 28.0 9.5 36,7 39,9

66.2 28.6 19.2 33.9 21.3

64.7 3L2 20.6 34.0 24.1

18.4

27.1

40.3

61.4

47.3

48.7

52.3

104.2

127.4

162.9

209.1

22.2

27.9

28.3

25.1

20.8

24.0

13,5 3.3 8.3 2.0

20.4 4.7 12.4 2.8

31.3 7.1 18.9 4.1

44.4 57.5 45.8 58.2

50.8 43.1 49,2 41.6

53.4 50.5 52.6 45.6

66.1 17.9 15,3 4.9

79.2 23.1 19.2 5.9

99,1 31.0 25.2 7.6

123.6 42.4 33.3 9.9

19.8 28.9 25.6 20.0

25.1 34.3 31.3 29.2

24.7 36.6 32.0 29.4

24,6 28.6 20.2 38.2

25.7 8.8 17.9 12.4

28.7 13.9 20.6 16.2

65.6

87.2

112,9

41,5

32.9

29.5

57,6

53.2

48.5

42.9

-7,7

-8.8

-11.8

49.2

41.7

41.3

4.3 6.8 ................. 42.0 58.8

12.8

20.9

57.4

86.6

63.9

16.0

17.8

18.6

17.3

11.5

4.2

-7.2

45.9

82.4

71.1

74.4

92.0

39.9

26,7

23.6

41,6

35.4

29.9

25.6 -15.0

-14,6

54.9

42.1

38.2

9,4 2.1 5.7 1.3 46,3

-15.4

Centratly_z_ed Asia

64.4

125,9

173.7

241.4

95.5

37.9

39.0

503.5

539.1

612_0

684.2

7,1

13.5

11.8

88.4

24.4

27.2

PROC Noxth Korea

61.4 3.0

121,7 4.2

166.7 7.0

230.7 10.7

98.2 39.9

37,0 64.4

38,4 53.8

496.8 6.7

532.8 6.3

605.1 6.9

677.0 7.2

7.2 -6.3 _

13,6 10.2

11.9 4.2

91,0 46.2

23.4 54.2

26.5 49.6

77A,1 1012.1 1354.4 1806.8

39.8

33.8

33.4

11.1

14.3

13.8

28.7

19.5

19.6

-0.8

-4.6

-7.5

28.4

27,3

26.2

14.6

19.1

18.2

45.0

29.2

31.1

World

t-,u,_ O "= _0 _Z _>

Regions 448.9 I_ Developed Regions 275.2

572.7

702.9

824,4

27.6

22.7

18,7

439.3

651.6

972.4

59.6

48.3

49.3

1776.9 1973.7 2255,8 2567.0 405,5

402.4

383.9

355.0

1371.4 1571.3 1871.9 2212.0

aRegional averagefor East Asia exdedes Talwan. Sotwce: United Nations, Panem# of Urban and Rural Poptdaff_n Growth. 1980, Annex II, Tables 48 md 49.

Table 3 -- Urban Concentration Indicators: Asian Countries, 1960-80 Percentage of Urban Population Country

In Cities of Over 500,000 Persons

In Largest City

Number of Cities ot Over 500,000 Persons

Index

Prtmacya 4t_

1960

1970

1980

1960

1970

1980

1960

1970

1980

1960

1970

1980

Bangladesh Burma India Sri Lanka Pakistan

20 23 7 28 20

25 23 6 20 21

30 23 6 16 21

20 23 26 0 33

39 23 31 20 50

51 29 47 16 52

1 1 11 0 2

2 i 19 1 6

3 I 36 1 7

0.80 1.56 0.68 4.85 0.88

1.00 1.81 0.56 2.17 0.95

1.20 1.89 0.46 1.92 0.99

Indonesia Thailand Philippines Malaysia

20 65 27 19

22 68 29 23

23 69 30 27

34 65 27 0

44 68 29 23

49 68 36 27

3 1 1 0

6 1 1 1

9 1 3 1

1.15 3.55 0,96

1.32 3.68 0.99

1.48 3.71 1.17

South Korea Talwan Japan PROC North Korea

35 42 ............ 18 20 6 6 15 13

1.07

1.52

1.49

22 6 12

35 42 15

38 41 13

41 44 19

5 38 1

7 47 I

9 65 2

1.25 0,72 1.00

1.35 0.72 0.85

i .48 0.71 0.73

aRat/o of populat/onof largestcity to the combinedpopulationsof the second,th/rdand fourthlargestc/t/es. Source:WorldBank, WorldDevelopment Report, 1980, AnnexTable20; andUnited Nations,Patternsof Urbanand RuralPopulation Growth, 1980, AnnexTables48 and50.

_, N *-aÂ˘

t_ _o

4_ O O Table 4 - Average Annual Percent Growth Rates of Population, GDP and Sectoral Production: Asian Countries, 1960-70, 1970-78 Cotmtr/

Population

GDP

Ag_icullu_e

Industry

Manufacturing

Services

1960-70 1970-78 1960-70 1970-78 1960-70 1970-78 1960-70 1970-78 1960-70 1970-78 1960-70 1970-78 Bangladesh Burma India Sril.anka Pakistan

15 2.2 2.5 2.4 18

2.7 2.2 2.0 1.7 3.1

3.6 2.6 3.6 4.6 6.7

2.9 4.0 3.7 3.4 4.4

2.7 4.1 1.9 3.0 4.9

1,6 3.6 2.6 2.3 1.9

7,9 2.8 5.5 6.6 10.0

5,9 4.5 4.5 3.0 4.8

6.6 3.3 4.8 6.3 9.4

5.3 4.2 4.6 1.2 3.5

3.8 1.5 5.2 4.6 7.0

4.7 4.2 4.6 4.3 6.2

Indonesia Thailand Philippines Malaysia

2.2 3.0 3.0 !9

1.8 2.7 2.7 2.7

3,5 8.2 5.1 6.5

7.8 7.6 6.3 7.8

2.5 5.5 4.3 -

4.0 5.6 4.9 5.0

5.0 11.6 6.0 -

11.2 10.2 8.6 9.6

3.3 Ii.0 6.7 -

12.4 11.5 6.8 12.3

8.0 9.0 5.2 -

8.7 7.4 5.4 8.4

SouthKorea Taiwan Japan

2.4 2.6 1.0

1.9 2.0 1.2

8.5 9.2 I0.5

9,7 8.0 5.0

4.5 3.4 4.0

4.0 1.6 1,1

17.2 16.4 10.9

16.5 119 6.0

17.2 17.3 ii.0

18.3 13.2 6.2

8,4 7.8 11.7

8.7 4.1 5.1

PROC North Korea

2.1 2.8

1.6 2.6

5.0 7.8

6.0 7.2

........ ........

Source: Wo_ld Bank, World Development Report, 1980, Annex Tables 2 and 17.

t_ r_ *4 O _Â˘

Table S - P_rcentage Distribution Country

Agriculture

of GDP: Asian Countries,

Industry

1960-78

Manufacturing

1960

1970

1978

1960

1970

1978

Bangladesh Burma

61 33

59 38

57 46

8 12

10 14

13 13

India Sri ].,anka

50 34

47 34

40 35

20 22

22 19

Pakistan

Indonesia ThaQand

54 40

47 28

31 27

Philippines Malaysia

26 37

28 32

South Korea Taiwan

40 28

Japan

1960

Services

1970

1978

1960

t970

1978

6 8

7 10

8 10

31 55

31 48

30 41

*-j

26 31

14 17

14 12

17 23

30 44

31 47

34 34

_, "-2

14 19

18 25

33 27

8 13

9 16

9 18

32 41

35 47

36 46

27 25

28 18

30 26

35 32

20 9

23 14

25 17

46 45

42 42

38 43

30 15

24 10

19 29

27 41

36 48

12 22

I8 33

24 38

41 43

43 44

40 42

,< t'-" O Z ,-]

Source: World Bank, World Development Report, 1980, Annex Table 3; and Worm Tables, 1980 (Second Edition), Table 4, pp. 392-395.

4_ O

4_ O tsd

Table 6 - Export and Import of GDP (in percent)

Exports of Goods and N.F.S.a Country

1960

1970

1977

Bangladesh Burma India Sri I_anka Pakistan

10.0 19.7 5:3 29.8 8.4

8.3 5.2 4.1 17.5 7.8

9.1 6.0 6.2 23.4 9.5

Indonesia Thailand

13.3 17.4

12.8 16.7

Philippines Malaysia

10.6 53.6

South Korea Taiwan Japan

3.4 11.1 11.0

Imports of Goods and N.F.S. a 1970

1977

9.3 20.7 8.3 32.8 15.0

12.5 8.7 4.7 19.7 14.6

15.7 10.0 7.1 20.7 19.4

21.6 21.5

12.6 18.9

15.8 21.5

18.8 27.0

19.1 43.8

19.0 50.3

10.4 40.8

19.4 39.2

22.5 41.9

14.3 29.5 10.8

35.6 53.5 13.1

12.8 18.6 10.5

24.1 29.6 9.5

35.6 47.8 11.4

aN.F.S,meansnon-factorservices. Source: WorldBank,WorldTables,1980 (SecondEdition), Table3.

1960

r_ t_

_0 Z

ASIAN URBANIZATION

AND DEVELOPMENT

403

REFERENCES Alonso, W. (1968), "Industrial Location and Regional Policy in Economic Development," Working Paper No. 74, Institute of Urban and Regional Development, University of California at Berkeley. Davis, K. (1975), "Asia's Cities: Problems and Options," Population and DevelopmentReview, 1: 71-86. Lewis, W. A. (1972), Southeast Asia's Economy: Development Policies in the 1970s. New York: Praegex Publishers. Oshima, H. T. (1978), "The Role of Manpower Development in Postwar Asian Differential Growth," Philippine Review of Business and Economics, 15: 37-58. .... (1980), "Perspectives and Prospects for Southeast Asian Growth in the 1980s and Research Issues," U. P. School of Economics, (mimeo). .... (1981), "Manpower Quality in the Differential Economic Growth Between East and Southeast Asia," U. P. School of Economics, (mimeo). Pernia, E. M. (1977), Urbanization, Population Growth, and Economic Development in the Philippines. Westport, Connecticut and London: Greenwood Press, Preston, S. H. (1979), "Urban Growth in Developing Countries: A Demographic Reappraisal," Population an d Developm en t Review, 2:195-216. Ranis, Gustav and Fei, J. C. H. (I961), "A Theory of Economic Development," American Economic Review, 51 : 533-565. Todaro, M. P. (1979), "Urbanization in Developing Nations: Trends, Prospects, and Policies," Working Paper No. 50, Center for Policy Studies, The Population Council. United Nations Economic and Social Council, Population Commission, Twentieth Session, (1978), Concise Report on Monitoring of Population Policies. E/CN.9/338.22 pp. 27-28. United Nations (1980), Patterns of Urban and Rural Population Growth, New York. World Bank, (1980), World Development Report, 1980, Washington, D.C.

GROWTHAND STABILITY IN SELECTEDNEWLYINDUSTRIALIZING AND SOCIALIST ECONOMIES Miyohei Shinohera â&#x20AC;˘ 1. Introduction The "postwar" period used to mean the immediate postwar years of confusion and reconstruction. The period of reconstruction was, however, followed by over twenty years of accelerated economic growth, both throughout the world and in Japan, at a rate more than double that of any prewar period. The space of over one-third of a century after the end of World War II, therefore, is no longer a short-term,but a truly long-term period. From the vantage point of the present, one can now analyze the growth patterns of the developing or the Third Worldcountries from a longer-term perspective, which is not feasible through short-term observations, and one can thus formulate more accurate prognoses for their future. There was a time when the epithet "developing" was nothing short of a euphemism for those countries characterized by chronic stagnation apparently beyond cure by aid from developed countries. But some of the developing countries have since detached themselves from the rest and jumped rapidly from low development to high growth, reaching the status of NICs- newly industrializing countries - or prospective NICs. The Third World has become an uneven group of nations, which has to be looked upon as a yet-tobe-advanced group of nations in which both stagnation and dynamism exist side by side. *Institute of Developing Ecoaom_

Japan.

4O5

406

MIYOHEISHINOHARA

Having gone through alternations of excessive growth and growth retardation, even the qualified NICs are now forced to plan for stable growth, that is, to try to strike a balance between growth and stability. Of importance to any country in any period, "growth and stability" is a matter of vital concern in the turbulent decades of the 1970s and 1980s. 2. Macroeconomic Balance and Inflationary Growth The Cases of Korea and Taiwan While the problem of macroeconomic balance is a first consideration in any discussion of "growth and stability," it is always the case, post facto, that aggregate demand equals aggregate supply. Given a closed system, it is also the case that investment = saving in ex post terms. Given an open system inclusive of external trade and the government sector, it is conceivably the case that: Investment + Current government expenditure + Exports = Savings + Tax revenue + Imports ex post, or again Investment = Savings + Government surplus + Excess imports. Investment here includes public investment, and government surplus and excess imports are restricted to current accounts. It should not be supposed, however, that all is well simply if these equations stand at equilibrium ex ante as well. The equilibrium, if attained beyond full employment income, causes inflation. If the resource-and-energy ceiling is below the full employment ceiling, full employment is unattainable, and there can be inflation side by side even with unemployment. To the left of point A in Figure 1, full employment income is below the resource-energy ceiling level income. So long as this condition is maintained, it is presumably the differences between demand for labor and full employment that will determine changes in wage levels, and in price levels as well. If the reality moves to the right of point A, prices can rise even below full employment and with the unemployment figure in the positive. When the government sets a high priority on reducing unemployment, people tend to have inflationary expectations and to behave speculatively. Under such conditions, actual rates of wage and price changes are not just a function of the unemployment rate but also a function of the anticipated inflation rate, causing the well-known

NEWLY INDUSTRIALIZING ANDSOCIALIST ECONOMIES GNP

407

full employment ceiling resource energy actualGNP

A ceiling

FIGURE FULL EMPLOYMENT

CEILING VS. RESOURCE

CEILING

Philips Curve to begin to shift upwards, greatly reducing the efficacy of Keynesian aggregatedemand policies. In this way, it is very important to see just at what point the ex ante equlibrium between aggregate demand and aggregate supply is attained. Apart from details, it will be of great importance in assessing the given conditions and in selecting policies for implementation to see whether such an equilibrium is attained (l) beyond the full employment point, (2) beyond the full capacity point, (3) beyond the resource-energy ceiling, and how these three are interrelated. Since the world economy is now located near the resource-energy ceiling, Keynesian aggregate demand policies, focusing on the reduction of unemployment, may naturally bring about inflation. Therefore, a problem arises whether or not inflationary or less inflationary growth will be appropriate. The cases of Korea and Taiwan present very interesting examples in thisregard. Incidentally, these three factors are, of course, not the sole limitations that restrict GNT growth. If, for example, the expected I

408

MIYOHEISHINOHARA

inflow of capital is not enough to finance the balance of payments in current account, the balance of payments ceiling may be a limiting factor on GNP growth in a short-term upswing. Policy-makers will have to pay close attention to how the balance of payments ceiling, full capacity ceiling, full employment ceiling, and resource-energy ceiling are mutually related. Turning to the case of newly industrializing countries in the Asian Third World, keeping in mind the points presented above, the Republics of Korea and Taiwan present a striking contrast. In the years 1962-1978, consumer prices increased by 8.9 times in Korea, but only 2.1 times in Taiwan. In comparison with Taiwan, Korea is characterized by "inflationary growth", The growth of money supply for the same period is again strikingly different, up by 154 times in Korea, and by 23 times in Taiwan. As for the personal savings ratio, it remained low in Korea under strongly inflationary conditions, averaging 2.4 percent in 1954-1960, and 2.8 percent in 1961-1970, but rising at last to 9.2 percent in 1971-1978, as shown in Table 1. Taiwan exhibited over the same period a steady upward trend from 4 percent to 12.4 percent on to 20.2 percent. Inflation rates thus lead to personal savings differentials, which in turn are fed back to inflation rates by way of varying rates of growth of money supply. Table1 -- PersonalSavingsRatio in KoreaandTalwan

Korea Talwan

1954-60 annualaverage

1961-70 annual average

1971-78 annual average

2.4% 4.0%

2.8% 12.4%

9.2% 20.2%

Sources:Korea- 1951-60;TheBankofKorea,NationallncomelnKorea,1978. 1961-78;EconomicPlanning Board,Handbookof KoreanEconomy, 1979. Taiwan- Directorate-General ofBudget,Nationallncomeof theRepublicof China,Dec.1979. Table 2 shows, for the same time frame, the sum of excess imports of goods and services and net payments abroad of factor income as a ratio of gross domestic investment. Although this ratio in Korea was reduced quite significantly from 69.5 percent to 23.5 percent since the end of the Korean war, Taiwan's performance from

NEWLYINDUSTRIALIZING AND SOCIALISTECONOMIES

409

41.1 percent down to -8.2 percent is still more impressive. As the personal savings ratio had already moved up to two-digit f_ares in and after 1963, the proportion of the external deficit over gross domestic investment began to drop dramatically from 34.7 percent in 1962, to 4.0 percent in 1963 and to -1.2 percent in 1964. Table2 - External1)eficit/C_rom DomesticInvestment Ratioin Korea&Tgiwan 1954-60 annualaverage

1961-70 annualaverage

1971-78 atmtuflaverage

Korea Taiwan

69.5% 41.1%

49.2% 11.9%

23.5% -8.2%

Sources:Korea- IMF,InternationalPb_mcidb_ttgt_" tiN, Yearbookof NatlonolAccountStatisticL Taiwan- Directorate-General of Budget,NationclIncome of theRepublicof China,Dec.1979. It is noteworthy, to say the least, that the personal savings ratio became higher in relative terms while dependence on the external deficit was correspondingly reduced in Talwan which followed the path of non-inflationary growth. 3. Asian NICs and Investment Cycle NICs, Korea in particular, are said to be in economic difficulties. Korea, which has been cited until recently as an exemplary case of rapid growth, has suddenly begun to go downhill. The rate of growth of real GNP and of mining and manufacturing output actually turned negative in 1980 (in percent):

Real GNP MMOutput

1976

1977

1978

1979

1980

15.1 29.8

10.0 19.9

11.6 22.9

6.4 11.7

-5.7 - 1.7

The sudden shift may be analyzed from various points of view. First, a measure of "stabilization crisis" is inevitable if an economy built on inflationary growth is to be normalized. The growth rate of M1 fell drastically between 1977 and 1980 (40.7,

410

MIYOHEISHINOHARA

24.9, 20.7, and 16.3 percent). But M2, which includes savings deposits, fell more slowly than M 1 (39.7, 35.0, 24.6, 2.6.9 percent). Both wholesale and consumer prices, meanwhile, presumably due to the impact of the second oil crisis, rose at accelerated rates (CPI: 10.2, 14.4, 18.1, and 28.6 percent). In other words, monetary policies cannot directly stabilize prices in an economy with built-in inflationary expectations. They may in fact incur the risk of inviting reduction in output. A negative growth of real GNP at -5.7 percent could not otherwise be accounted for. Second, it is important to realize that the Korean economy, which is an investment-led type of economy, is marked by relatively mid-term, cyclical changes. The ratio, A1 in Figure 2, of gross domestic fixed investment to GNP began to rise steadily from the mid 1960s, topping at 32.9 percent in 1979, well above the previous peak of 25.8 percent in 1969. There was another peak in 1975, which, however, was below the generalized trend of investment growth through the '60s and '70s as indicated by the straight dotted line, and thus may not represent an independent peak in any mid-term cycle of fixed investment. There is thus in effect a ten-year cycle between the last peak in 1969 and the latest in 1979, exactly in line with the Juglar Cycle of ten years' duration. In a dynamic economy, fixed investment on the upswing grows faster than GNP, so that once the investment boom cools down, there is a countervailing adjustment period of three to four years during which the investment ratio tends to be pushed down. A1 or gross domestic fixed investment includes public investment as well as investment in the private sector, while A2 represents private investment in plant and equipment. In the case of Korea and elsewhere, where the government plays a leading role, A 1 may be the better yardstick for explanation. The share of capital goods in total imports, B in Figure 2, reveals an interesting pattern: a trough at 13 percent in 1965, a peak at 36.5 percent in 1968, another trough at 26.0 percent in 1974, and then up to 33.9 percent in 1978, which, except for 1965, traced a path just about a year ahead of the one exhibited by the fixed investment ratio shown as A 1. It should thus be possible to predict that the downward trend in the share of capital goods- 33.9 percent in 1978, 31.0 percent in 1979, and 23.0 percent in 1980 - will prove to be an advance indication of a largely reduced fixed investment

, _

•

(

',,, >,

,'_

_,___. "_ '

_'_' to

o l

412

MIYOHEISHINOHARA

ratio in 1981. In this sense, the recession beginning in 1980 has a downward adjustment of a mid-term cyclical nature in relation to fixed investment. The cyclical changes, as shown in Figure 2, of the share of capital goods in total imports are still closer to the Juglar Cycle. Ten years between the peaks of 1968 and of 1978, nine years between the troughs of 1965 and 1974, represent almost a perfect replica of Juglar's mid-term cycle, more impressively than do the investment ratio cycles. Third, the gross fixed investment ratio of over 30 percent translates into a rather imbalanced growth pattern that is hard to sustain for long. The long-term uptrend in the investment ratio is thus likely to taper off sooner or later, to be subject thereafter to mid-term, moderate ups and downs. The threshold of 1980, as it were, may well prove to be a major turning point of the Korean economy. By that point in time, let it be noted, the Korean economy had already experienced a labor shortage and growing wage costs. Fourth, Korea was relatively immune from the impact of the first oil crisis. Net external debt accounted for 36.1 percent and 31.7 percent in 1974 and 1975 of gross domestic investment, dramatically above 16.6 percent and 9.2 percent for 1972 and 1973. Petro-dollars, in other words, were invited, in effect, to f'mance growing investment needs. In the wake of the second oil crisis, however, the Korean economy, with its already very high investment ratios, could not afford to accommodate additional stimuli..It could no longer hope to be immune to the adverse effects of global recession and inflation, nor to ignore the impact of the general deterioration in the international economic climate. Taiwan, as noted earlier, presents a case of less "inflationary growth" than Korea with "steady" growth in investment. This is not to deny, however, that Taiwan's economic growth has been nothing short of a dynamic one. There are fairly clear mid-term cycles, for example, in the share of capital goods in total imports: a 13.1 percent trough in 1952, a 27.9 percent peak in 1960, a 21.4 percent trough in 1963, a 34.7 percent peak in 1969, and a 28.6 percent trough in 1973. Intervals of eleven years between the troughs of 1952 and 1963, of nine years between the peaks of 1960 and 1969, and ten years between the troughs of 1963 and 1973, all combine to demonstrate the ten-year Juglar cycle. Figure 3 has been drawn up to provide the Taiwanese counterpart to the Korean case in Figure 2.

NEWLY INDUSTRIALIZING AND SOCIALIST ECONOMIES

Notes:

A : B :

413

Grossprivatefixed inveslm_t GNPratio. Shazeof capitalgoodsin total imports.

Sources: A - Di_ctorate-Generalof Budget,AccotmtingandStatistics,Executiv_Yuan, NaJ#o_ Income of the Republic of Ckiw, 1979; B - Councilfor EcollomicPI_ and DevelopmentEx_cuUveYuan,TatwanStatistleel De,,, Book, 1979. FIGURE3 FIXED INVESTMENTCYCLEIN TAIWAN

In terms of either imported capital goods or private fixed investment ratio, the In'st peaks were reached in 1960, followed by those in 1968 or 1969. The third expected peaks failed to materialize due to the oil crisis and outflows of foreign capital, but up till then, there had been mid-term rises at regular intervals of eight to nine years, culminating in the peaks, as in the case of Korea, in 1968 and 1969. The Korean peaks of 1978 or 1979 did not repeat themselves in Taiwan, presumably because of the overwhelming pressures of the oil cris_ Both the level of private fixed investment ratio and its rate of acceleration were higher in Korea than in Taiwan. Although no

414

MIYOItEI SHINOHARA

straightforward comparison is warranted between the private fixed investment ratios in the two countries, because the Korean data are inclusive of fixed investment by public corporations, it is noteworthy that, in 1974-1978, it grew from 19.2 percent to 25.4 percent in Korea but declined from 16.3 percent to 13.8 percent in Taiwan. It is no wonder that an economy marked, as in Korea, with dynamic development, should exhibit a tendency toward mid-term investment cycles as a symptom of entrepreneurial dynamism. After a rapid spurt under dynamic conditions, one is obliged to take a breather from time to time. The Japanese economy did more or less the same. What makes the Korean case interesting is that during the period of adjustment in 1969 through 1976, the investment rate simply levelled off, with slight ups and downs, instead of adjusting itself, as might be expected, downward. In the case of the private fixed investment ratio, all that happened was that the extent of the increase somewhat slowed down. This is presumably because the Korean economy at that time still had stronger growth inertia than the Japanese economy. Given its investment ratio already at excessive levels, the Korean economy will have to be adjusted downward in 1980s. The New Five-Year Economic Plan for 1982 through 1986 assumes an average fixed investment ratio of 30.8 percent, an all-time high that will be very hard to sustain. The share of capital goods in total imports, which anticipates the investment ratio, had already decreased from 31 percent in 1979 to 23 percent in 1980. It is indeed very significant that the two dynamic economies among the Asian NICs are found to be marked typically by mid-term investment cycles, which have been known to occur so far in Japan and the United States. Economists should feel intrigued by the further parallelism in Korea and Taiwan. 4. Inter-Sectoral

Imbalance and Socialist Economies

Among classical approaches to the business cycle, the aggregate gap between production capacity and effective demand has often been cited in accounting for changes in the business climate. The "undereonsumption" theory of Hobson, Keynes, etc. and the so-called "capital stock adjustment principle" follow this approach. Another approach tries to analyze business cycles in terms of sectoral balance rather than of supply-demand gaps, Spiethoff and Hayek follow this strategy and try to fred the essential trigger on downward

NEWLYINDUSTRIALIZING AND SOCIALISTECONOMIES

415

shift in business in "intersectoral imbalance" among consumer goods, investment goods, and intermediate goods sectors. The "overinvestment" theory in particular emphasizes such intersectoral balance. According to the underconsumption theory, underconsumption in the guise of "saving > investment" that occurs near the peak of a boom constitutes an important cause of the downward shift. According to the overinvestment theory, however, it is excess demand in the form of "investment > saving" due to credit creation that is considered to be the direct cause of sectoral imbalance and recession. In his "Investment Cycles in European Socialist Economies," A. Bait pointed to the occurrence of mid-term investment cycles in East European economies in 1951 through 1965. His points of emphasis lay, most significantly, on various manifestations of intersectoral imbalance in socialist economies, rather than on overall supply capacity in excess of effective demand, as might have been appropriate with reference to capitalist economies. Increasing social dissatisfaction at short supply of consumer goods, as well as the relative shortage of agricultural and raw materials production set back by heavy industrialization, etc. were the major concerns that Bait concentrated on. Indicators of fixed investment cycles in Hungary, Yugoslavia and China have been put together in Figure 4: net fixed capital formation (NFCF) over net material product (NMP) for Hungary, gross fixed capital formation (GFCF) over gross material product (GMP) for Yugoslavia, and the share of machinery and equipment in total imports for China. The dotted line for China shows the same for imports from Japan, apparently one year ahead of swings, both up and down, in total imports. In the case of Hungary, an interval of just ten years between the peaks of 1953 and 1963, was followed by twelve years until the next peak was reached in 1975. Troughs in 1957, 1966, and 1977, occurred every nine to eleven years. The cycles in Yugoslavia are somewhat shorter; eight years (1962-1970) between peaks, and another eight years between 1970 to 1978 or beyond. Troughs in 1957 and 1966 generally follow the Hungarian pattern. It is since the 1970s that the two economies have begun to behave differently, but the fact remains that cycles are not short-term but mid-tenn. In view of the parallel shifts in the investment ratio and share of

416

MIYOHEI

SHINOHARA

% 26

'75

24 22

Hungary NFCF

NMP

'77

'63

'78

/ Yugoslavia

14 _66 12

GFCF

'62

GMP

'59

'73 40

(el

'66

'80 26

Importsfrom Chinaof /* Math. 8 Equip. i_' "Total Imports from China

[d) 76

4"_ 30 25

," (a)

_=(b)

(¢)

/:'66 t IA

//,,.\/'

;/\

_ _

"1 I

% ),,'i

ImPorts from dal:_n

of Mach.& Equip+ Total Imports from Japan

1965

. 1970 .....

, ....

1975

FIGURE 4 FIXED INVESTMENT

"-.,': "/ ",'

/ /

V// _' '62

oi_._,,1955 ,. ............. 1960

:X \

, /

'_l

CYCLE IN SOME SOCIALIST

ECONOMIES

1980

NEWLY INDUSTRIALIZING AND SOCIALIST ECONOMIES

4]7

capital goods in total imports, as identified in the cases of Korea and Talwan, the share of machinery and equipment in total imports has been chosen as the indicator for China. But the investment ratio, obtained from Kan Chao's and other estimates of capital formation in Mainland China, fails to reveal a corresponding pattern. Either income estimates are unreliable and trade statistics are to be preferred, or changes in trade statistics do not significantly affect a oountry's investment ratio - it is yet to be decided which interpretation is correct. It is assumed here for the sake of argument that the share of machinery and equipment presumably reflects changes in the domestic investment ratio. Should this assumption prove to be inappropriate, the following comments will have to be revised accordingly. Tamio Shimakura at the Institute of Developing Economies in Tokyo used the share of machinery and equipment in China's imports from Japan, and discovered five year cycles, peaking in 1965, 1970, 1975, and 1980. He pointed out that every downward shift was associated with a major political incident. In contrast to "short-term political cycles" occurring in capitalist countries, the "Shimakura Cycle" indicates that Socialist China is characterized by relatively "mid-term political cycles." Let it also be noted that if the 1971 peak may be ignored as just a small rise in a short-term cycle, there is a good case for recognizing a ten year interval between the genuine peaks of 1966 and 1976. Apart from such inferences, the Shimakura Cycle provides an important suggestion on the interactions between economic and political affairs in socialist societies. Fixed investment enforced as planned and heavy industrialization carried too far tend to create an imbalance between agriculturaland industrial sectors, and between light and heavy industries. Let these be known as "intersectoral" or interindustrial" imbalance. If this imbalance is allowed to reach excessive proportions, there are likely to

internal

political

disputes

about

the

management

the

economic crisis and the locus of responsibilities, which in turn may trigger a period of power struggles. PoUtical instability adds to the seriousness of economic stagnation. Seen in this light, the process of economic contradictions due to excessive investment leading to political confrontation which in turn perpetuates economic stagnation can evolve as a matter of politico-economical dynamics in socialistsocieties.

418

MIYOHEISHINOHARA

China of course may not be the typical example, but it is still arguable that the Shimakura Cycle might not have occurred in the absence of the essential elements of such dynamics. It is important to remember that socialist planning strategies can alternate from heavy to light industrialization and vice versa. Excessive reaction to heavy industries tends to overemphasize light industries. Sewing machines and bicycles are produced in great numbers, causing iron and steel shortages. The Baoshan steel mill project, once postponed, thus enters its second phase of construction. Such a turn of events is an inevitable consequence of preoccupation with short-term phenomena that only makes for mid-term cycles in the investment ratio. In order for socialist planners to realize genuinely stable economic development, they should learn to rid themselves of short-term considerations and to work out, from mid and long-term perspectives, what constitutes a normal investment ratio and a balanced industrial structure suitable for stable growth. In addition, there should emerge an established set of political rules of the game, if power struggles are to be avoided, and if mid-term economic cycles are not be politicized. Both political and economic conditions have to be set right before socialism can be a genuine success in Asia. In her "Dialectics of Socialist Society: Polish Case," which has just been translated into Japanese, Jadwiga Staniszkis brilliantly described the changing process of movement and counter-movement in the economic, political, and social scenes in Poland. The overall background is provided by political-economic cycles that evolve as changing rates of investment anticipate alternating waves of totalitarianism and anti-totalitarianism. Investment actually exhibited negative growth in 1954-1957, 1969-1970, and 1976-1977. A few comments may be in order here by way of conclusion: (1) The theory of mid-term cycles has been denigrated in Japan as fatalism that fails to take sufficient account of policy impacts. But in Japan, the United States, Korea, Taiwan, and even in a few socialist countries, mid-term cycles could not be placed under control through policy measures. There is rather an undeniable tendency for policies and politics to be on the receiving end of the impact of dynamic investment cycles. It is high time, after one-third of a century of experience since the end of World War II, that a critical review of the possibility and efficacy of political control be initiated.

NEWLYINDUSTRIALIZINGAND SOCIALISTECONOMIES

419

(2) It is rather ironical that the outmoded theory of the business cycle that attaches importance to inter-sectoral and inter-industrial imbalance as a result of excess demand -investment due to credit creation in excess of saving- should have been found to be applicable to mid-term cycles under socialist regimes. One of the problems that still remain to be solved is to define how this outdated theory needs to be revised under the current world economic conditions. (3) While Keynesian thinking was prevalent, typologies of business cycles were deliberately shunned. Postwar textbooks on economic change frequently omitted reference to the waves of Kondratieff, Juglax, or Kuznets. It is very important at this point in time to reanalyze their waves from a new perspective free from the shackles of Keynesian ideology. (4) Cycles axe an attribute of dynamism, and growth is their shadow. The higher the dynamism of an economy is, the wider the amplitude of cycles in the investment ratio. The possibility, therefore, of controlling mid-term waves by aggregate demand management may diminish as dynamism increases. (5) Finally, two types of growth were distinguished, inflationary and non-inflationary, in the rapid development of NICs. The view is that the latter is to be preferred as a sounder strategy.

REFERENCES Bait, A. (1971), "InvestmentCycles in EuropeanSociafistEconomics," Journal of Economic Literature. Chao, K. (1974), Capital Formaffon in Mainland China, 1952-1965, University of California. Sasamoto, T. and Shimak-ura,T., cds. (1977), The Evolution of Japan-China Trade (in Japanese),Instituteof DevelopingEconomies. Staniszkis,J., The Dialectics of Socialist Society: Polish Case, Japanese translation published in 1981 based on the author'sunpublishedEnglishmanuscript,Iwanami,Tokyo.

DISTRIBUTION FLOW OF EDUCATION IN THAILAND Edita A. Tan and Wannaslri Nalyavitit* with Technical Note by Fe Lisondra

1. Introduction Income inequality has been a serious problem in a large number of developing countries including ASEAN members except Singapore. A high degree of inequality between .45 and .50 Gini has persisted in the Philippines, Thailand and Malaysia during a fairly high growth period of the past three decades. Recent attempts to solve the problem mainly through Malaysia's Bumiputra policy and the Philippine's land reform program have not made a perceptive impact so far. It might be argued that without applying drastic and consistent measures the inequality could be expected to remain serious. There are structural givens in underdeveloped economies which tend to perpetuate inequality and the poverty of those in the lower income brackets. The higher the degree of inequality in a low-income country, meaning the poorer are its poor, the more difficult it is to break the problem. The relative and absolute poverty of the lower income groups separates and locks them within their environment. Such environment tends to be socially (and economically) restrictive on individual choices so that opportunities for upward mobility are limited. This paper is an attempt to describe the process of intergenerational transfer of education capital under specified market condi*Univeraityof the Philippines School of Economics, and ThammasatUniversity. The authorswouddlike to acknowledgethe supportgivenby the Coun_ for Asian Manpower Studies (CAMS), the Kited/ ResearchInslitute and the ResearchUnit of the Facultyof Economics,ThammasatUnivex_ty. 421

422

EDITAA. TANAND WANNASIRINAIYAVITIT

tions affecting choice in a not atypical LDC country-Thailand. A constrained optimization model in education choice is used as the framework of analysis. The model explains how inequality in income and in education interacts intergenerationally and traces the change (or lack of it) in equality and poverty over time. The model is presented in Section 3 after a discussion of the state of inequality and poverty and their trend in Section 2. 2. Descriptive Analysis of the Distribution of Education Equity in the distribution of education means equality in educational opportunity. This has been defined in varying degrees of strictness. It means each child of given ability and aptitude has an equal chance to pursue the education and training that suit him best. In strict terms, the reference would be inherent ability and aptitude. There is an assumption that the distribution of inherent or inborn traits among the population is not affected by race, socio-economic, location and other environmental variables. Each child in the population has an equal chance to be born with a particular level of these inherent traits. Education of all forms - formal and non-formal - builds on these traits so that at later stages of life, it is not possible to distinguish inherent from acquired abilities. In fact, performance in IQ tests is found not to be constant over time but improves with acquired knowledge. In the looser sense, equality in educational opportunity is measured in terms of measured ability which has already been affected by environment, socio-economic background and quality of formal schooling. In the absence of information on the distribution of inherent traits in the population, equity in educational opportunities can be judged on the basis only of an a priori notion of equality in these traits. The judgment therefore, of degree of equality or inequality of educational opportunity in Thailand is based on this a priori distribution. The writers wish to believe that the distribution is equal, i.e., that each child of whatever background has the same probability distribution of possessing inherent traits as the probability distribution of the population to which he belongs. It means that he has a ten percent probability of being of superior scientific talent if the population has the same probability of producing this much scientific talent. Two sets of data are used to indicate the degree of inequality in the distribution of education in Thailand: a) attendance in school

EDUCATION IN THAILAND

423

and the highest level of formal schooling attained by those who left the formal school system; b) performance in achievement tests administered to grade III and senior high school pupils. The population was grouped according to variables that are most likely to affect the distribution of opportunity: family income, father's education, location - by region, whether rural (non-municipal) or urban (municipality, a. School First time

metropolis). Attendance some

broad

are discussed.

indicators

Table

of movement

1 shows

that

of distribution

the average level

over

of school

attainment has been rising from 1960 to 1975. This results from an increasing rate of attendance at each age group. Finer indicators of inequality are obtained from several sources: the 1975 survey of youth, the 1973 national achievement test of grade III pupils and the study by Nitongkorn and Vutisart of the achievements of senior high school students who were admitted to college in 1977. School attendance was very much influenced by location and socio-economic variables as shown by the following cross-tabulations In Table 2 attendance rate by location for each age range is given. As expected, children in the Bangkok Metropolis had the highest rate at each age range or at its corresponding schooling stage - lower

Table 1 -- Attendance Rate by Age Range 1960, 1975

Age Range

1960

4-6

Age Range at Any Program

1975

4-6

6.3

119.3

7-9

71.4

11-13 in grades 5-7

19.8

10-14

72.4

14-18 in high school and lower vocational schools

17.9

15-19

24.3

2.9

20-24

5.1

7-10 in grades 1-4

20-22 in college and technic,el schools

Source. NationalStatisticalOffice 1975 Surveyof Youth; Report of the UNESCORegional Advisory Team for Educational Planningin Asia Bangkok, 1965.

424

EDITA A. TAN AND WANNASIRI NAIYAVITIT

elementary, upper elementary, high school and post-high school. Attendance rates dropped between upper elementary to high school from 72 percent to 24 percent for all children, 90 percent to 60 percent for municipal children, and 70 percent to 18 percent for non-municipal children. For the last two stages, Le., high school and post-high school, the corresponding decline was 24 percent. Northeast, the poorest region, showed the lowest attendance rate at all levels and for both municipal and non-municipal areas. Only 1.2 percent of its 20-24 year old youths attended school, as compared with 22 percent for Bangkok. Those who stopped schooling are not likely to go back to school so that their highest attainment as per survey time might be taken to be their lifetime attainment. Some of these have completed their Table 2 - Rate of School Attendance of Children and Youth by Age, by Region, by Municipal/Village, 1975 Total

4-6

7-9

10-14

15-19

20-24

Kingdom I) Municipal 2) Non-Munlcipal

38.9 58.6 35.7

6.3 27.4 3.7

11.4 85.2 69.6

72.4 89.7 69.7

24.3 59.3 17.7

5.1 19.8 2.1

Bangkok 1) Municipal 2) Non-Municipal

56.8 59.2 46.2

21.2 25.6 6.1

83.1 85.0 76.0

89.8 91.0 84.2

57.6 61.0 40.5

21.7 24.0 9.7

Central 1) Municipal 2) Non-Municipal

42.2 57.1 40.6

9.8 29.6 8.1

74.2 84.2 73.2

76.3 88.2 75.1

53.3 53.3 23.3

4.8 11.9 3.9

Northern 1) Municipal 2) Non-Municipal

39.5 62.6 38.0

6.9 43.8 5.1

77.5 89.5 76.9

71.4 88.2 70.4

19.2 60.1 16.3

3.0 15.5 2.0

Northeast 1) Municipal 2) Non-Municipal

29.9 53.5 30.0

1.1 8.9 .5

63.2 82.4 62.5

61,6 86.0 60.7

12.3 55.7 10.3

1.2 14.1 .5

Southern 1) Municipal 2) Non-Municipal

44.3 58.8 42.5

6.1 27.6 3.9

72.2 86.4 70.6

85.1 89.0 84.6

33.6 59.4 30.1

3.8 11.7 2.8

Souzce:National8ta_ Tsblc CI.

Office(NSO),The C_ildrenand Youth Sun,ey, 1975.

EDUCATIONIN THAILAND

425

programs such as academic, teacher training and vocational The highest attainment of the current cohort over their life can therefore be estimated from their attendance rate. This estimate is given directly by the survey (please see Table 3). It is seen that majority of school leavers, even in the older ages of 20 to 24, completed lower elementary levels only while many very young out-of-school youths aged 7-9 have never attended school. This number amounted to 22 percent of the total population of this age range in 1975. Family background is another important determinant of school attendance. Tables 4 and 4a show attendance rate by age and by family income which were obtained from the raw data of the National Statistical Office (NSO). The effect of income is strongest in the three lowest income brackets to which 64 percent of Thai youths belonged. The difference in attendance rate increases as age or schooling level rises. The lowest income group, for instance, to which 25 percent of youth population belonged, had attendance rates of 67 percent, 62 percent, 24 percent and 10 percent for ages 6-10, 11-14, 15-19, and 20-24. At the highest income bracket the corresponding rates were 91 percent, 93 percent, 77 percent, and 38 percent. It must be noted that the attendance rate reported by the NSO is lower than in the cross-tabulation since the data consisted of unmarried children of the head of households only. The difference in attendance rate between the two reports reflects the effect of relationship and civil status on school attendance. This effect is shown more precisely in the empirical test of the model. Cross-tabulations with other socio.economic variables are not given here since they should have the same pattern as that of income. b. Distribution of Learning at Grade III and High School V. Th_ quality of education offered each child differs so that amount learned also differs. It also depends on his total environment. The educational content of environment varies substantially according to location and socioeconomic background. Schools differ in quality of instruction and available instructional materials. Family income exerts a further effect on learning by determining the physical and mental health of the child. It is not uncommon in poor situations in LDCs that children have not enough to eat. Hunger must have a direct or immediate impact on the rate of learning in school. Persistent malnutrition would have longer and more pro-

Table 3 - Percentage Distribution of Youth Not Attendhtg School by Age, H,_ghestLevel Attained, Municipal/Non-Municipal Zero Level

Age P,_.ge Total Kingdom 7- 9 10-14 15-19 20--24 Municipal 7--9 10--14 15-19 20-24 Non-munidpal 7- 9 1O- 14 15-19 20--24 Total Youth 7-- 9 % naa 1O--14 % na 15-19 % na 20-24 %ha % in munidpal

Lower Education

99.2 .5 .2 .3 I00.0 2.1 .2

Upper Elementary

Lower Secondary

Upper Secondary

Teache_ Training

.8 98.4 90.7

...... I.I 5.8

..... 2.6

83.8

5.3

5.0

2.4

4_ to

Academic