Empirical Economic Bulletin, Spring 2009, Volume 2 by Bryant University

EEB--UNDERGRADUATE ECONOMICS JOURNAL f

f �'-'"•fl.'IC

,NI

--.....

t �--· ___ ,

EMPIRICAL ECONOMIC BULLETIN

Hit

I •

THE CENTER FOR CLOBAL AND RECIONAL ---ECONOMIC STUDIES BRYANT UNIVERSITY

SPRING 2009, VOLUME 2

EDITOR: Ramesh Mohan DESIGN & LAYOUT: Ramesh Mohan and Rebecca Marcus MISSION STATEMENT: The Empirical Economics Bulletin is the undergraduate journal for the Department of Economics, Bryant University. The journal is produced in conjunction with the Bryant Economic Undergraduate Symposium. The focal point of the Symposium is on training undergraduate students in the art of writing, presenting, and publishing empirical research papers on a range of socio-economic and economic topics. The Symposium’s primary emphasis is empirical studies with policy relevance. Students are then able to publish their empirical papers in the Empirical Economics Bulletin. An objective of the Economics’ Department, Bryant University is to train students to conduct quantitative economic data analysis and to present the results in a coherent and meaningful way. This objective is met through having the Symposium and the publication of the Empirical Economics Bulletin. The first issue was in Spring 2008 and has been published annually with original work from an array of student authors. SUBMISSION GUIDELINES: Students may submit their socio-economic and economic work here: Ramesh Mohan at rmohan@bryant.edu. Limit one submission per author. Each submission should have a title page with the title; name of author; abstract; keywords; JEL classification; author’s email. Previously published work is not accepted. The reading period is September 1 to December 1. Copyright reverts to author upon publication. Any questions may be directed to Professor Ramesh Mohan at rmohan@bryant.edu. © 2009 Empirical Economics Bulletin

Empirical Economic Bulletin, Spring 2009, Vol. 1 Empircial 2

Table of Contents The Effects of Aging on Inflation, Andrew Stone .......................................................................................... 1 Effects on Globalization on Income Inequality in High Income Coutnries, Craig Bradford .……………………. 15 The Relationship Between Housing Prices and Macroeconomic Factors in Spain, Garret Fitzgerald …….. 33 The Effect of Crime Rates on Housing Prices: A Hedonic Study, John Paul Goncalves ………………………….. 51 State Level Comparison of Factors Contributing to Rising Inpatient Hospital Costs, Jeff Fontaine ……….. 69 The Economic Benefits of Education as a Return to GDP Per Capita, Johnathan Brown, ……………………… 82 Are Educated Women Less Likely to Get Married?, Qian Jiang ………………………………………………………….. 102 The Effect of Tax-Burdens on Foreign Direct Investment: A Cross-Sectional Look at Developing Economies, Kevin Hauver …………………………………………………………………………………………………………………………………….. 117 An Empirical Analysis of the Impact of Home Foreclosure on the Crime Rate: Evidence in Atlanta, GA, Luis G.Acevedo ………………………………………………………………………………………………………………………………………… 131 The Effect of Aid Dependency and Quality of Institution in Alleviating Poverty in IDA Countries, Mahah Mirza …………………………………………………………………………………………………………………………………………………. 148 The Effect of Domestic Investment, Economic Growth and Human Development on Foreign Direct Investment into China, Michael Paolino …………………………………………………………………………………………….. 164 The Impacts of Energy Efficiency and Consumption on GDP in the Euro Area, Justin T. Getts …………….. 186

Empirical Economic Bulletin, Spring 2009, Vol. 1 Empircial 2

The Effect of Aging Populations on Inflation Andrew Stone*

Abstract: This research uses panel data of 66 high and middle income countries to examine the inflation effects of demographics. The primary objective of this research is to show that consumer prices change based on the amount of retired persons in a nation. The double log model used in this paper also includes interest rates and uses data from 1991-2007 in order to demonstrate the most recent demographic shifts. The results showing inflationary pressure coming from young retirees compliment the Modigliani life-cycle hypothesis that this age group consists of net consumers, while the deflationary pressures provided by the working aged population justifies their placement in the category of net savers.

JEL Classification: E31, J10 Keywords: Inflation, Age, Demography * Bachelor of Science in Business Administration: Economics, Bryant University, 1150 Douglas Pike, Smithfield, RI 02917. Phone: (413) 230-4456 Email: astone1@bryant.edu

_____________________________ The author thanks The World Bank and United States Census Bureau for providing data and gratefully acknowledges the encouragement and guidance from Dr. Ramesh Mohan, Professor of Economics, Bryant University.

Empirical Economic Bulletin, Spring 2009, Vol. 2

1.0 INTRODUCTION As the baby boomer generation begins to age, the United States and other governments across the globe have begun worrying about the impact of the population being largely elderly. Social programs are usually the first cause of concern; these programs increase government spending by billions as the elderly population increases just slightly. This research aims to ask the question, should central banks also worry about the aging population as they try to implement monetary policy? After all, the wealth we acquire as humans can only take us so far, and must be either spent or inherited at death. In addition, there comes a time when working becomes impossible, but spending is still necessary, causing the elderly to be net consumers. These two realizations lead to the conclusion that inflation should increase as the elderly share of the population increase. This study looks into this issue through an ordinary least squares regression analysis for 66 high and middle income nations. The regression compares the inflation as measured by consumer prices over time with the age groups present in their populations. The aim of the research is to show that inflation rates are correlated to the amount of the population that is made up of net consumers, and that this group is primarily individuals older than 65. Unlike previous studies, this research will also include interest rates and their affects on inflation into the analysis. This inclusion may allow researchers to find that the age of a population does not matter, but instead that fluctuation in interest rates cause inflation. The organization of this paper seeks to point out the theoretical background of demographic effects on macroeconomic variables and then explain the recent literature relating to these demographic effects on inflation. The empirical model is then outlined and the results from the data analysis are presented to show how age has affected inflation over the past two

Empirical Economic Bulletin, Spring 2009, Vol. 2

decades. Finally, the implications of this study as it relates to future monetary policy for developed nations are explored. 2.0 TRENDS Figure 1: World Population by Age Shares

Demographic shifts play a very large role in economic vitality due to the different impact each age group has on the market. A shift from a large middle age population to a large elderly population will create a much different economy resulting from varying behaviors and production capabilities. The world is currently undergoing an increase in elderly population as evidence by Figure 1. This figure also shows that this shift will continue into the future as life expectancies increase due to better medical technologies. The effects of this shift are two-fold. First, the shift has large demand-side effects because of the changes in consumption behavior

Empirical Economic Bulletin, Spring 2009, Vol. 2

that occur when an individual hits retirement age. Second, the supply-side of the market is affected as retirement signals the end of productive life that begins when an individual decides to retire. Modigliani and Brumberg’s (1954) life-cycle hypothesis states that an individual’s consumption and savings needs to be carried out over a lifetime and will therefore influence that individual’s actions. Based on this principle, we should expect to see working aged people (those aged 24-65) become net savers in order to prepare for their necessary consumption in retirement. Using demand-side analysis of this issue, the economy should then see downward price pressure from an increase in savings when the group of working age citizens is largest. Using the lifecycle hypothesis we can see that the spike in working aged citizens has suppressed inflationary pressures in the economy. As the population ages, as we will see in the future, and the percentage of net savers in an economy begins to fall, inflationary pressures are expected to return and begin pushing prices up. The supply side effects of demography shifts involve the assumptions of the productivity of the working aged population. Here, it is assumed that older working aged individuals (those aged 45-64) have gained valuable experience and are able to produce more with the same amount of inputs. Therefore, an increase in this population should lead to more supply in an economy, which allows the economy to meet the demand for goods at a lower price. Once again, inflationary pressures have disappeared with the appearance of a large working population. However, once this population moves into retirement their productivity disappears and the supply diminishes, reinstating the inflationary pressures. Both of these theoretical frameworks indicate that inflationary pressures will occur in an economy with a large non-working population. However, because non-workers are typically defined to be anyone under the age of

Empirical Economic Bulletin, Spring 2009, Vol. 2

24 or over 65, it should be specified that non-workers under 24 typically consume much less than the elderly due to lower health care costs and the absence of life savings. Thus, inflationary pressures are much greater from those aged 65 or older. 3.0 LITERATURE REVIEW Many studies have been done on demography shifts with respect to macroeconomic indicators in recent years as economists have come to realize the significance of the baby boomer generation. Bloom et al. (2008) predict that the shift in age structure that will occur over the next half century will not lower GDP growth per capita. Their research attempts to replace the misconceptions on age structure and labor force participation rate and show that aging populations will not decrease this rate enough to cause structural shifts in productivity growth. Faruqee (2002) analyzes the large aging populations of the U.S and Japan and claims that this demographic shift will have large macroeconomic implications, most notably a current account deficit stemming from social security outlays. Additionally, Faruqee predicts a decrease in per capita income as effective labor decreases due to this large elderly population. Navaneetham (2002) uses a study of South Asian nations to view the effect of age on GDP per capita growth rates and finds mixed results for the age shares under 50. However, the age share of the population 50-64 years shows a significant positive result with respect to GDP per capita growth rate. He indicates that these results likely come from the experience and productivity of workers in this age group. He finds in Singapore that the age group 65 years and older has a negative effect on GDP per capita growth, but has a positive effect in the Philippines. These mixed results may indicate that elderly age groups are more productive in some economies than in others.

Empirical Economic Bulletin, Spring 2009, Vol. 2

The specific research in the inflation category has been less common, but influential nonetheless. Bruer (2002) finds that Swedish inflation is closely related to that country’s age structure, placing a large pressure on Swedish economic policies in the coming years. Farvaque et al. (2006) also find results in OECD countries indicating that populations aged 64-75 have a significantly inflationary pressure on an economy. Additionally, due to the large share of working age populations in these nations, these researchers also find that central bank independence has not been the primary factor in keeping inflation rates low, but instead find this effect comes from the deflationary pressures of large working age populations. Lindh and Malmberg (2000) find that the age structure in OECD countries has a significant impact on inflation. Precisely, that net savers aged 64 and younger reduce inflationary pressures on an economy, while those aged 65-74 increase the inflationary pressures in an economy. The study discussed in this paper closely follows the Lindh and Malmberg model. 4.0 EMPIRICAL METHODOLOGY 4.1 Regression Model The regression model used here is based largely off that of Lindh and Malmberg (2000) where age structure is broken up into categories resembling the consumption trends of the lifecycle hypothesis. These structures are able to model the effects on inflation of youth consumption (ages 0-14), young adult consumption (ages 15-29), working age consumption (ages 30-44), experienced worker consumption (ages 45-64), young retiree consumption (ages 65-74) and elderly consumption (ages 75 +). The double log model used is able to smooth out business cycle trends in inflation that are not pertinent to the analysis and provide an accurate estimation of demographic changes on inflation. Additionally, inflation (lnCPI) is measured by a

Empirical Economic Bulletin, Spring 2009, Vol. 2

consumer price index where prices for the year 2000 is the base value in each case. The panel data provides 1207 observations for 66 countries from the years 1991-2007. The regression model is as follows: lnCPI = β0+β1lnCPIt-1 + β2IntRate + β3Age1 + β4Age2 + β5Age3 + β6Age4 + β7Age5 + β8Age6+ε where the Age variables (1-6) are the natural logs of the amount of the population in that age division. Age1 consists of ages 0-14, Age2 ages 15-29, Age3 ages 30-44, Age 4 ages 45-64, Age5 ages 65-74 and Age6 ages 75 and above. The independent variable lnCPIt-1 is a lagged variable of inflation that uses the natural log of the price index from the previous year. Some past research (Gali and Gertler, 1999) has indicated that lagged inflation may not be effective past one quarter, however the data available for this research is in annual increments and captures the inflation rates for each quarter in aggregate. In this manner, inflation for the previous period continues to be a relevant variable for the current period’s inflation according to the sticky price model of the New-Keynesian Phillips Curve. All variables use a log form of their respective values with the exception of interest rates (IntRate) which fluctuate largely in response to perceived business cycles of an economy and are therefore a large cause of inflation in the short term. By estimating the model in this way, the short term variables affecting inflation have been fully represented in interest rates or eliminated in the natural log values of the additional independent values. There is some evidence that this model may contain a large amount of multicollinearity between the age groups inside a nation’s population (Lindh and Malmberg, 2000), specifically resulting from the dependency of the consumption of the youth population on that of the older

Empirical Economic Bulletin, Spring 2009, Vol. 2

populations. The tests used by previous research proves the logical reasoning that youth populations have no income stream and are therefore dependent on their working parents to fund their consumption behavior. The exclusion of Age1 from the model helps to strengthen this regression, leaving the model to appear as so: lnCPI = β0+β1lnCPIt-1 + β2IntRate + β4Age2 + β5Age3 + β6Age4 + β7Age5 + β8Age6+ε 4.2 Data The data attained for consumer prices and interest rates is from the World Banks’ World Development Indicators (WDI) data set and includes data for the time period 1991-2007. The population data that has been split into the corresponding age groups by the researcher comes from the U.S. Census Bureau’s International Data Base (IDB). 5.0 EMPIRICAL RESULTS Table 1: Expected Sign

Independent Variable Expected Sign

Empirical Economic Bulletin, Spring 2009, Vol. 2

lnCPIt-1

IntRate

Age2

+/-

Age3

+/-

Age4

Age5

Age6

The expected signs from this regression have been already been briefly mentioned and are shown in Table 1 but should be emphasized here as their estimates are provided. The independent variable lnCPIt-1 is expected to provide a positive sign as the rise in price from consumer goods in one year carry over to raise the prices of other consumer goods in the following period. In order to battle this inflation central banks use their control over monetary policy and attempt to maintain consumer prices. One tool at their disposal is the interest rate, which the banks raise to attract savings rather than consumption in an economy. Thus, we expect inflation to fall as interest rates rise, providing a negative sign for the coefficients of interest rates. Demographic shifts within a nation have been shown to affect inflation rates with respect to the consumption patterns of individuals. This research is especially concerned with the inflationary trends expected from retirees. Therefore, the expectation is that working age populations (Age4 in this model) will have negative relationships with inflation, while retired age populations (Age5 and Age6) will have positive coefficients. Lindh and Malmberg (2000) and Farvaque et al. (2006) recognized the effect of younger citizens (Age1, Age2, Age3) on inflation as being relatively insignificant, with no consistent sign.

Empirical Economic Bulletin, Spring 2009, Vol. 2

Table 2: Regression Results

Independent Variable lnCPIt-1 IntRate Age2 Age3 Age4 Age5 Age6 Β0 Adjusted R-squared: .9583

Coefficient (Standard Error) .8055*** (.0058) -.0017*** (.00016) -.0158 (.0240) -.0431 (.0394) -.0039 (.0465) .1681*** (.0417) -.1021*** (.0245) .4206*** (.0262) ***, ** and * denote significance at the 1, 5 and 10% level respectively

The results of this research are provided in Table 2 and offer very little departure from previous research. This is able to show that time and country specific variables have very little effect on inflation and it is instead demographic composition within a nation that affects its inflationary pressures. Because of this, the results hardly differ from previous research even though two different decades and a large set of countries are used. The expected signs are found for both the interest rate variables and lagged inflation variables, granting further evidence to their contributions to inflation levels in a country. The expected sign is also found for the young

Empirical Economic Bulletin, Spring 2009, Vol. 2

retiree variable (Age5) which demonstrates that young retirees are participating as net consumers following their exit from the workforce. One deviation from expectation is the coefficient value for the elderly population (Age6), which appeared negative. This was somewhat unexpected, however not an isolated incident in research of this type. Studies from both Lindh and Malmberg, (2000) and Farvaque et al. (2006) found the same effects from this segment of the population. This coefficient seems to contend that consumption is low from this part of the population and that this group is creating deflationary pressures on the economy. Reasons for this sign change have been explored in terms of the retirement-consumption puzzle (Smith, 2004) indicating that retirees do not plan accordingly for the significant decrease in income upon retirement and are therefore forced to scale back their consumption patterns. However, that theory provides an explanation for decreased consumption upon retirement, which this study rejects based on inflationary pressures coming from the newly retired age group. Therefore, the retirement consumption puzzle may actually affect those aged beyond retirement who have expended much of their private retirement savings. This depletion of funds subsequently causes older retirees aged beyond 75 to cut back on consumption in order to maintain financial safety. Additionally, elderly persons often find themselves unable to take care of themselves or enjoy a consumptionfriendly lifestyle, resulting in excessive payments to health care providers. This would indicate that the basket of goods used for CPI in this regression would not best represent retirees over the age of 75 who purchase a much different basket of goods, including nursing care, medical supplies and prescriptions. The coefficients of the middle aged populations were ambiguous and showed no significance to the effects of inflation in this model. Explanations for this may include large differences in consumption patterns displayed by each individual within these groups. After all,

Empirical Economic Bulletin, Spring 2009, Vol. 2

these segments are the largest percentage of the population and are comprised of many different types of individuals. Some individuals within this group earn high incomes and save proportionally large amounts while some earn high incomes and spend large amounts. The same holds true for those earning low incomes. Thus, the net effects of all of these combined seem to be insignificant within the inflation framework. Still, the regression model has been able to explain 95% of the data as measured by its adjusted R-squared value, indicating that inflation’s computation may come largely from demographic effects and government monetary policy. 6.0 POLICY IMPLICATIONS This research has pointed out that the monetary policy from these governments, as measured by the interest rate, has a significant impact on the amount of inflation in an economy. For this reason, inflation targeting policies have been introduced by many central banks as the key indication of their success. However, the research has also shown that demographics can upset the inflation rates in an economy through a population reaching retirement age. So, if the health of the elderly population increases and we see consumption patterns begin changing due to a continuation of active lifestyles, inflationary pressures may arise from a larger segment of the population. For example, the group aged 75+ is currently helping bring deflation, but with the advent of new medical technology this group may be reduced to only those aged 80+. Therefore central banks need to be aware of demographic trends as they relate to inflation if these banks wish to remain in control of their inflation targeting policies. Viewing this research through a social lens, one sees a disparity in the inflation pressures of the age groups separated in the research. As previously mentioned, we should observe deflationary pressures from the experienced working age group as they increase their savings to prepare for retirement. However, we see their effect on inflation as insignificant, a possible

Empirical Economic Bulletin, Spring 2009, Vol. 2

problem due to economic inefficiencies in the economy, where this age group is being enticed to spend rather than save. Much of this could be due to social programs (such as Social Security and Medicare in the United States) which supplement income and benefits for the retired. Therefore, rather than save on their own, retired persons are able to count on taxpayers to pay for their retirement living. Does Social Security contribute to inflation? This fundamental question is not the goal of this research, but directly relates to the policy decisions stemming from this research and should be asked by policy makers. 7.0 CONCLUSION The goal of this study was to show the impact of demography on inflation among high and middle income countries through a time frame that had not previously been studied. The results of this research found very similar inflationary effects as the past studies, namely that those recently entering into retirement (aged 65-74) create great inflationary pressures into the economy as their consumption stream outweighs their earned income potential during that time. Also, the very elderly (aged 75+) contribute deflationary effects into the economy from inactive lifestyles and lower consumption streams. However, the largest effects on inflation come from interest rates and the inflation of the period prior as consumer prices rise to meet an equilibrium demand. These results indicate that the economic forces led by the baby boomer generation currently reaching retirement may be very different than anything previously seen.

Empirical Economic Bulletin, Spring 2009, Vol. 2

BIBLIOGRAPHY Bloom, D., Canning, D. and Fink, G. (2008) “Population Aging and Economic Growth” Program on the Global Demography of Aging, Available at: http://www.hsph.harvard.edu/pgda/Working%20Papers/2008/PGDA_WP_31.pdf Bruer, M. (2002) “Can Demography Improve inflation Forecasts? The Case of Sweden” Uppsala University Working Papers, Available at: http://www.nek.uu.se/pdf/wp2002_4.pdf Faruqee, H. (2002) “Population Aging and Its Macroeconomic Implications: A Framework for Analysis” International Monetary Fund, Available at: http://www.imf.org/external/pubs/ft/wp/2002/wp0216.pdf Farvaque, E., Hericourt, J., and Lagadec, G. (2006) “You Have To Be Old To Be Wise: Inflation, Demographic Trends and Central Bank Independence” Available at: http://mpra.ub.uni-muenchen.de/13076/ Lindh, T. and Malmberg, B., (2000) “Can Age Structure Forecast Inflation?” Journal of Economics and Business 52: 31-49. Modigliani and Brumberg, (1954) "Utility Analysis and the Consumption Function: An Interpretation of Cross-Section Data," Post-Keynesian Economics. 383—436. Navaneetham, K., (2002) “Age Structural transition and Economic Growth: Evidence from South and Southeast Asia” Centre for Development Studies, Available at: http://www.cds.edu/download_files/337.pdf Smith, Sarah (2004) “Can the Retirement Consumption Puzzle Be Resolved? Evidence From UK Panel Data” The Institute for Fiscal Studies, Available at http://www.ifs.org.uk/wps/wp0407.pdf

Empirical Economic Bulletin, Spring 2009, Vol. 2

Effects of Globalization on Income Inequality in High Income Countries Craig Bradford Abstract: This paper examines the relationship between technology, trade liberalization, and financial globalization in incomes inequality, focusing on high income countries. We find that technological progress has a smaller effect on income inequality in high income countries than in middle and low income countries. It is also found that increases in the percentage of workers in the services industry decreases Income Inequality by a significant amount. The GINI index is used to measure the level of income inequality, and the Chinn-Ito index is used to measure the level of openness to capital investment.

JEL Classification: O33, F16, J31, G18, F41

Keywords: GINI, Technology, Trade, Financial Globalization

Department of economics, Bryant University, 1150 Douglas pike, Smithfield, RI 02917. Phone (603) 547 0459 email: cbradfor@bryant.edu

Thanks to Professor Ramesh Mohan for your wisdom and guidance

Empirical Economic Bulletin, Spring 2009, Vol. 2

1.0 Introduction Inequality has been on the rise over the past few decades, not only in developing Nations but also in developed countries. The previous theory was that developed nations would reach a level of grow that would cause these inequalities to level off. Many have blamed the rising level of globalization to be the key factor in causing this rise. The flight of low-skilled jobs to countries with much lower wages rates has been blamed for this inequality in developed nations. Many protectionist supporters have rallied for more trade barriers to protect low-skill jobs. To devise correct policy, and create a more shared wealth among the population the true causes of inequality must be calculated. Wide income gaps cause vast differences in social welfare and overall living conditions. It is also thought that inequality can slow economic grow due to the fact that all economic opportunities may not be used to their full potential, by not allowing capital and labor to equalize. Inequality also causes a larger percentage of the population to be open to poverty conditions during economic downturns and recessions. As seen in many countries inequality can cause uprisings among to the impoverished, against policy makers and globalization. There has been a great deal of research done on this topic and the past. The recent work on investigating technological advancements and income inequality is what drives the basis of this research. The major finding in these papers is that recent technological advances in telecommunications and international trade have the greatest effect on income inequality. The study found that these advancements have driven income disparities across a study focuses on various income nations. This paper is focused on high income countries that are at the forefront of technology and education advancement. Technological differences between high income and low income countries are great, so it is possible that the exclusion of these countries in an empirical study will make considerable differences in the effects they have on income inequality. This paper examines not just the effects that trade globalization and the effect it has on inequality but also the roles of financial globalization and technological change have on inequality in High income countries. This paper discovers that trade liberalization and financial liberalization have contradicting effects on inequality. Trade liberalization has a tendency to decrease inequality while financial liberalization, in the form of FDI tends to increase inequality. It was found that technology changes are the major factor affecting the increases in inequality.

Empirical Economic Bulletin, Spring 2009, Vol. 2

Technology seems to be increasing the premium on higher skilled labor. Supporting this observation was the ease of access to education. It is observed that this access to education caused decreasing effect on Inequality. The most significant finding is that higher employment is the service industry has the greatest effect on income inequality. The major contribution of my paper is that it focuses deeply the effects of income inequality in high income countries instead of using a sample of countries from all income levels. This will in turn be more beneficial when looking at GINI problems in such countries as the United States. This Paper is organized into 5 sections. The next section outlines some of the recent trends that income disparities have shown in recent years, and some brief data to support some of the current assumptions on the subject. The Literature review section outlines much of the previous research that has been done on the subject of income inequality and GDP per capita, since it has been shown that there is a close relationship between the two. The data and empirical analysis section goes over the model that was created for the research done in this paper, and outlines the variables and limitations within the model. The conclusion section points out the policy implications that the model implies. 2.0 Trends: Figure 1

Figure 1 shows that the GINI index has been increasing steadily over the past few decades. This is a phenomenon that has been plaguing economist for many years. Researchers have come up with a few major theories about why this is happening. One interesting fact is that

Empirical Economic Bulletin, Spring 2009, Vol. 2

the top one percent of households received 21.8 percent of all pre-tax income in 2005, which is almost double what that same figure was in the 1970’s.1 This number is quickly approaching record high level that was set in 1928 when this figure was a 23.9%. During this same time period the top earning 5 percent of American families saw their real incomes increase by 81 percent (figure 2); while the bottom earning 5 percent of families saw their real incomes drop by 1%. Many of these drastic changes are blamed on the equally drastic increases in executive compensation. On average an American CEO earned 411 times as much as an everyday American worker in 2005, compared to 107 times as much in 1990. Top American Executives also make twice as much domestically as the top executive from France, U.K., and Germany. Figure 2

By the Numbers

Empirical Economic Bulletin, Spring 2009, Vol. 2

Figure 3

Another relatively accepted explanation for this trend is the increasing returns to education, and the relatively stable level of graduation rates. The data currently available shows that workers who have a college degree or above have a real median income that has been increasing, but those who do not have a degree have been decreasing significantly over the past years. Data also shows that college graduates real income has been increasing, but the relative number of college graduates has remained the same over the past 30 years. Higher levels of education, in general creates a higher level of income. As a result, people who cannot afford higher education, or have opted out of partaking in higher education generally receive lower wages. Recently the high demand for highly skilled and highly educated workers has increase the equilibrium wage level for people who have higher education levels. It is believed that this phenomenon has pushed high skill wage levels up but left low skill wages relatively unchanged. This idea is depicted in figure 4, where it can be seen that the wages of people with college degrees has increased more than people without college degrees.

Empirical Economic Bulletin, Spring 2009, Vol. 2

Figure 4

Source: Smith, Carl. “Why increasing returns to education are not a good sign.”

Some economist and the general population blame increase in Globalization and trade liberalization as a major source of income inequality. As we all know there have been significant advances in trade and communication internationally. Some of the contributing advances are technological strides with internet, and satellite communication, as well as trade advances through various free trade agreements. Some studies have shown that trade and foreign direct investment into low income countries has decrease the demand for low skilled labor in high income countries, sending that demand abroad. Many studies have found that international trade is insignificant when compared to other technological factors such as automation. Robert Lawrence discovered that low skilled jobs have been replaced by machines in wealthy nations, which has driven down the number of low skilled workers that can be affected by competition from low income countries. The graph below shows a color contrast between countries and their GINI index levels (figure 5). Countries that are redder have higher inequality levels, and countries that are more green have lower income inequalities.

Empirical Economic Bulletin, Spring 2009, Vol. 2

Figure 5

Source: “A Dollar a Day, Finding Solutions to Poverty”,

Developing countries have faced the same wage inequalities, as the global level of wage inequality increases. These countries have seen significant increases relative to the rest of the world, but these increases start from a significantly higher starting point. After China opened its protected trade barriers they encountered record level increases in income inequalities. In China there is a huge difference between the wages earned by the urban population and the wages earned by the rural population. India is also experiencing the same level of income disparages. Many critics of open markets and globalization blame the implementation of aggressive market oriented reforms, for these intense increases in income differentials. High levels of income differentials cause many problems. It can cause extreme poverty for many low income jobs such as those in agriculture. Extreme poverty is very prevalent in many countries where the rich and incredibly wealthy while the poor suffer. Large income differentials can also cause civil unrest. As seen in many African countries income disparages cause frequent uprisings and political unrest. The stability of many nations is shaken by new rising regimes that continually fight for power within the country. These factors make income inequality a very important aspect to study and keep track of.

Empirical Economic Bulletin, Spring 2009, Vol. 2

3.0 Literature Review Though the effect of trade liberalization and Income inequality is a relatively new concept and extensive research is yet to be done, many papers have analyzed the effect on GDP from the liberalization or protection of trade. GDP and income inequality are affected in similar ways by many different factors, so a look at the existing research on this topic is relevant to the discussion of income inequality. Many of these studies look at single events where trade reforms are drastically changed, and what effects those events had on GDP. Two somewhat recent articles of this type are Nash and Thomas (1991) and Papageorgiou et al (1991). These two papers look at a single year of trade liberalization and observe higher GDP growth after trade liberalization. Both papers conclude that trade liberalization leads to increased levels of GDP growth. The main issue involved with these two studies is that the single liberalization events involved much more than just trade transformations. A major external factor that was not taken into consideration was the fact that most of these trade liberalizing events also involved a regime change from Communist to an arguably more efficient Capitalist economy. Researchers have also suggested that trade liberalization in developing economies, such as the economies of many African countries are detrimental to long-term economic growth. Aksoy (2006) suggests that liberalizing trade reforms in the end leads to deindustrialization, and that most developing countries have indeed lightened their trade regimes. This deindustrialization is cause by already established industrial job being lost to countries that have set up factories with a more established and efficient production process. These international factories have the benefit of economies of scale and more experience that put the domestic factories out of business since they can offer a cheaper product to the people in that country. This causes the workers of that country to revert back to even lower skilled jobs that pay less and therefore causes significant GINI and HDI disparages. Subsequently the even greater problem is that in many of these countries there are more factors to take into consideration than just trade reform. The GINI Index is widely accepted as a measure of income disparages between economies. Studies have taken place that found the GINI index is either on the rise or has no trend in any direction at all. Anand and Segal (2008) suggest that there is no significant evidence that supports any trend in the reduction or increase in the GINI for since 1990. The paper

Empirical Economic Bulletin, Spring 2009, Vol. 2

concludes that there are too many methodological and data uncertainties and gaps to conclude any significant results from GINI data. This insufficiency of information mainly relates to countries where the data is much harder to collect and measure accurately. Some countries, such as those in Africa and India have very large black markets for labor and other goods. This makes raw data inaccurate because much of the transactions that take place in these countries go unnoted. These numbers have to be estimated, and therefore leads to inaccuracies that could make the GINI coefficient somewhat inaccurate. It is suggested that part of what causes income inequalities is the flight of low skilled jobs to regions where the wage rates for these jobs is significantly lower (Moss and Harrington 2006). When these low skill jobs leave the United States it creates an increased unemployment gap where these subsequently laid off workers use to have employment. According Moss and Harington, the GINI Index of inequality has seen an overall rise over the past three decades. Some economists attribute this rise in inequality on the social policies of Eastern Europe. Inequality in the United States has been particularly pronounced. From 2000 to 2003 the GINI index for the United States has increased from 38.8 to 46.4. In the recent past incomes at the upper end of the spectrum have been increasing as income for the lower end has seen a decline. We have seen this trend since about 1979. One explanation for this trend is the increasing returns to education. The data shows that workers who hold a college degree or higher, real median income had increased, but had decrease for those who do not have college degrees. They attribute this increase in the GINI index to three factors Globalization, technology and cultural norms. They suggest that globalization makes it easier and cheaper for jobs to more to lower wage areas. Technology increases communication capabilities so that data can move to and from these areas with greater ease. The final factor cited in the paper are cultural norms which refer to the obstacles put in place to prevent the flight of jobs to and from a particular country. Das (2008) uses the theory of convergence to explain his trends in the GINI on the global level. He suggests that the technological influence of the United States of America caused technological and productivity growth in developing nations. This spillover of technology is also aided by subsequently better government institutions. The investment by corporation to create profitable and useful infrastructure to use for business purposes speeds up the advancement of

Empirical Economic Bulletin, Spring 2009, Vol. 2

that individual country and suggests that it will eventually cause the economy to grow to a global economic mean growth rate. The recent rise in global Income inequality is often attributed to 3 different factors ( Jaumotte, et. al. 2008). The three factors are technology trade liberalization and financial globalization. By using a select group of diverse countries the study finds that the factor that creates the largest amount of income inequality is technological advancements. These advancements in technology are closely attributed to foreign direct investment. These advancements increase the premium on skills and tend to substitute away low-skill jobs. Increases in demand for higher skilled jobs and workers are created by technological advancement therefore increasing the inequality of income. Jaumotte finds that this effect from technology is more significant in lesser developed countries, since the use of technology in developed countries is much more widespread and integrated already. Small increases in technology have less of an effect as its influence increases at a decreasing rate. The study also finds that trade globalization is associated with a reduction in inequality that is offset by financial globalization and foreign direct investment. 4.0 Data and Empirical Analysis 4.1 Definition of variables This model is based on Jaumotte (2008) where the study finds that technological innovation has the greatest effect on income inequality. The empirical analysis of that study was based on a selection of high, middle, and low income economies. The study conducted in this paper differs in a few ways when compared to Jaumotte (2008). The following model is somewhat more simplified than the model created in Jaunotte (2008). This study also looks exclusively at economies that are rated as high income by the World Bank. The empirical model is as follows. (For more detail more detail on variables “Variable Description, Explanation and Source” in the appendix)

Empirical Economic Bulletin, Spring 2009, Vol. 2

Lngini= β0 +β1 Chinn-Ito + β2 ln(EX+IM) + β3 lnIndustry + β4 lnagri + β5 lnservices + β6 lntech + β7 %educ The countries studied in this report are all countries that were ranked as high income countries by the World Bank. The World Bank ranks 63 countries as high income earners, but only a select 32 countries were used due to gaps in data, particularly the dependant variable GINI index rankings. This is understandable since the GINI index is a relatively new indicator and much of the infrastructure and information needed to calculate the GINI is not yet readily available. The Chin-Ito data for foreign direct investment openness was taken from the 2007 publishing of financial openness rankings. The data used in this study was an average of data from 1995-2000. If there were discrepancies in the data for those years, the average of the available years from 1995-2000 were used. To account for the amount of trade openness this study uses the total amount of goods and services exported, as a percentage of GDP plus the total amount of goods and services imported, as a percentage of GDP. The reason net export data is not used in this situation is for the reason that we are looking at total trade not balance of trade. A country that imports and exports significantly more than any other country may have a low net export if both exports and imports values are relatively the same. Agricultural, Industrial, and Services employment data was taken from the World bank Economic indicators database. The values were created by averaging the % of total employment of 1995-2000 for each sector. The impact of technology on Income inequality is represented by ICT investment as a percentage of GDP, and the data was averaged for the years 2000-2005. Human capital advancement in this model is represented by the % of GDP invested in education and is an average of data from 1998-2000. In this model the GINI index is used as the dependant variable in a GLS linear regression equation. The GINI index data is from 2007, as it is has a lagging reaction to the independent variables used in the equation.

Empirical Economic Bulletin, Spring 2009, Vol. 2

4.2 Empirical results Table 1 Independent Variable

Expected sign

Chinn-Ito

lnEX+IM

lnIndustry

lnagri

LnServices

Lntech

%educ

The expected signs for the regression model have already been covered briefly and are summarized in the table 1. The variable Chinn Ito is expected to increase Income inequality because it is hypothesized that a country with a high openness of foreign investment is open to adverse effects that increase the wealth of the higher income laborers, while exporting jobs of lower income laborers. The level of trade a country participates in, which is represented by the exports plus imports variable is expected to be positive. The reasoning behind this prediction is that if a country decides to take part in a high amount of trade they are more likely to purchase cheaper products from other countries, which in turn puts the same companies and industries out of business in that country. The Employment in industrial jobs is predicted to have a negative impact of the GINI index. If a country were to create more industrial jobs the wages of low income workers would increase and catch up to the wages of high income earners. The Variable Agriculture is also predicted to be negative. If a country were to make more of their own agricultural products the incomes of the low skill laborers would also increase in the same way the industrial sector wages would increase. Employment in the services sector is expected to increase wage inequality by having a positive effect on the GINI coefficient of that particular country. As employment in the services sector increases, there are more people earning these high wages which in turn increase the income gap. Investment in technology in predicted to have an increasing effect on income inequality. As technology improves it will create more cheap opportunities to export jobs to low income areas, and therefore lower low-skill wages to a lower global level. Finally, education investment in expected to increase the GINI coefficient due to the theory of increasing returns to education.

Empirical Economic Bulletin, Spring 2009, Vol. 2

Table 2 Variable Chinn-Ito lnEX+IM lnIndustry lnagri LnServices Lntech %educ R-squared: 61.4%

Coefficient P-value (standard error) .08869 .013 (.0329) -.05456 .330 (.0548) -1.4988*** .000 (.3404) -0.20708*** .001 (.0538) -3.2771*** .000 (.8000) 0.2855* .100 (.1667) -0.05683** .022 (.0231) ***, ** and * denote significance at the 1, 5 and 10% level respectively

All of the variables in the model proved to be significant except for the country’s level of trade and technology. Trade levels within the countries looked at in this analysis yielded a pvalue of .33 which is highly insignificant. When exports and imports were run separately in the regression the coefficients tended to cancel each other out; exports creating a negative effect on the GINI index, and imports creating a positive effect. Investment in technology is completely statistically insignificant in this regression. This is inconsistent with the results found in Jamoutte (2008), which found that technology not only has a statistically significant result, but it has the greatest effect on increases in income inequality. It was found that the impact of technological advancement gives companies more ability to use telecommunications to send work to lower wage areas in order to cut costs. As high income countries are at the forefront of technological advancement it is understandable that any new advances in communications and technology would have little impact on wage inequality. The results of the OLS regression, in table 2 provide a few surprising results. First off, the variable of the Chinn-Ito Index, though its effect somewhat small is positive instead of the predicted negative reaction. This finding is parallel with the findings in Jaumotte (2008). The

Empirical Economic Bulletin, Spring 2009, Vol. 2

level of trade a country takes part in also departs from the previous hypothesis that it would create a negative impact the GINI index level. Employment in all three sectors creates negative coefficients, meaning that increasing employment in all of these sectors will decrease income inequality. This negative reaction was predicted for all three variables except employment in the service industry. Employment in the service industry also has the largest effect on income inequality, which is a surprising finding. The fact that all these variables have a negative coefficient leads us to think of further research into the effect of total unemployment levels and its effect on income inequality. Technology investment and education spending both have the coefficient sign that was predicted for each value. The results found in this empirical analysis are somewhat inconsistent with other papers that look at countries from all income groups. It is found that employment in the services sector actually reduces the overall income inequality in a high income country. This makes sense when we look at increasing returns to human capital. As low wage/ low skill workers start to gain the human capital needed to work at the services level, the gap between low income earners and high income earners will reduce. As low wage earners start to become less and less prevalent, and start to “catch up” with service wage earners. The impact of education on the GINI Index also differs from other papers. Education in developing nations creates huge strides in economic advancement and income levels, but education in High income countries works much in the same way that technology does in the sense that education in these countries are at the forefront of their global peers. Foreign direct investment is proven to have very little effect on a country when it comes to wage inequality. 5.0 Conclusion The model creates in this research was created to find the true cause or causes for the high increase in Income inequality over the past decades. This model proves that High income countries have very different needs when compared to lower income and developing nations. The most significant finding is that increasing employment in the services industry will lower income differentials by the greatest amount. This statement makes a lot of economic sense. As a country looks to increase GDP and income, over time laborers need to become more and more profitable. When laborers start to move from lower paying industrial and agricultural jobs, to higher paying

Empirical Economic Bulletin, Spring 2009, Vol. 2

service jobs there is an immediate increase in income inequality stemming from a select few working the higher wage service jobs. This same situation has been notoriously blamed for creating high levels of inequality, but there is one subsequent movement that most fail to take into consideration. In these developed trade economies and as the rest of the industrial and agricultural sectors start to move to these higher paying service jobs, the income gap is once again closed since more and more people are earning service wages. This phenomenon is comparable to a bell-curve movement and is pictured in figure 6. Figure 6 Increased Employment in Services and its Effect on GINI

As laborers start to move up the first portion of the bell curve income inequality increases until the peak is reached. During the decline on the other half of the bell curve everyone starts to make high paying service wages and a reduction in overall income inequality is reached. This phenomenon could explain why over the past few decades The United States has seen such an increase in its GINI levels. The United States is undergoing the initial first stage of the bell-curve, so income inequality is increasing drastically. Laborers should worry less about losing their low skill jobs to lower wage countries, and focus more on advancing their own skills to higher paying jobs that take the place of their low income exported jobs. If this theory is true then the United States should start to see slowing GINI growth levels, followed by a zero growth period at the peak of the curve which would be followed by decreasing income inequality levels.

Empirical Economic Bulletin, Spring 2009, Vol. 2

High income economies should give workers more incentives to move to service jobs, by making training such as college cheaper and more available. Federally funded College tuition should be increased to entice people to increase their education and human capital, so that they are more able to work at higher wage service jobs. This will help citizens transition more quickly from the industrial sector to the services sector. Policy should also be put in place to lower unemployment to decrease the GINI level by an even greater amount. The results from this model suggest that the overall percentage of employment has a large impact on income differentials. If working age laborers do not have jobs this increases the polarization of wages between high and low income by an large amount. High income countries should also not use protectionist policies to save jobs from being exported. These countries should let these low paying jobs be exported, in exchange for high skill and high wage jobs that will take the place of those exported jobs. This theory is also supported by the fact that countries that have higher trade levels also experience lower income inequality levels since they are more able to export the low wage jobs abroad while at the same time increasing their own average wage level domestically.

Empirical Economic Bulletin, Spring 2009, Vol. 2

Appendix: Variable description, explanation and source Variable Lngini

Explanation Gini index (2007)

Chinn-Ito

FDI investment openness (avg. 1995-2000) Amount of total international trade (% of GDP, avg. 1995-2000) employment in industry (% of total, avg. 19952000) employment in agriculture (% of total, avg. 19952000) ICT investment (%of GDP, avg. 2000-2005) ICT investment (%of GDP, avg. 2000-2005) % of GDP invested in education (avg. 1998-2000)

lnEX+IM

lnIndustry

lnagri

LnServices

Lntech

%educ

Empirical Economic Bulletin, Spring 2009, Vol. 2

source CIA World Factbook Chin Ito Index 2007 publish World Bank Economic indicators World Bank Economic indicators World Bank Economic indicators World Bank Economic indicators World Bank Economic indicators World Bank Economic indicators

Bibliography “A Dollar a Day, Finding Solutions to Poverty”, <http://library.thinkquest.org/05aug/00282/over_measure1.htm> Anand, S., Segal, P. (2008) “What do we know about global Inequality?” Journal of Economic Literature 46: 57-94 Askoy, Ataman. (2006) “Growth before and after trade liberalization” International Monetray Fund, IMF Working Paper. Das, G. (2008) “Does Trade and Technology Transmission Facilitate Convergence? The Role of Technology Adoption and Reducing the Inequality of Nations” Journal of Economic Policy Reform 11: 67-92 Hartman, Chris. “By the Numbers”. Inequality.org 2008 Jaumotte, F., Lall, S., Papageorgiou, C. (2008) “rising income inequality: technology, or trade and financial globalization?” International Monetray Fund, IMF Working Paper. Available at: http://www.imf.org/external/pubs/ft/wp/2008/wp08185.pdf Moss, D., Harrington, A., (2006) “Inequality and Globalization”, McGraw-Hill Primis. Harvard Business School Cases- Business and Government pp 96-102 Papageoriou, D., Michaely, M., Choksi, A. Lliberalizing foreign trade. Cambridge, M.A.: Blackwell And Oxford, 1991 Riezman, R., Whalley, J., Zhang, S. (2005) “metrics capturing the degree to which individual economies are globalized” Working paper, available at: http://search.ebscohost.com/login.aspx?direct=true&db=ecn&AN=0905951&site=ehost-live Smith, Carl. “Why increasing returns to education are not a good sign.” Modeled behavior. 20 Nov. 2006. Mar 2009 < http://modeledbehavior.blogspot.com/2006/11/why-increasing-returnsto-education-are.html>

Empirical Economic Bulletin, Spring 2009, Vol. 2

The Relationship between Housing Prices and Macroeconomic Factors in Spain

Garrett Fitzgerald1

This paper examines the effects of certain macroeconomic variables (i.e mortgages, inflation, and employment) on the price of houses in Spain. An error correction vector autoregressive (ECVAR) model is used to model the impact of the macroeconomic variables on real housing prices. Variance decompositions will be analyzed to determine the extent to which these variables have an effect on housing prices in Spain.

JEL classification: E32; R20; G00; G11 Keywords: housing, Spain, inflation, employment, mortgages

Bryant University, 1150 Douglas Pike, Smithfield, RI 02917. Phone (413) 586-2529, Email: gfitzger@bryant.edu

_____________________________________________________________________________________________________________________________________

The author would like to thank Dr. Ramesh Mohan, Department of Economics, Bryant University, for aiding in the development of this paper.

Empirical Economic Bulletin, Spring 2009, Vol. 2

1.0 INTRODUCTION The aim of this paper is to take several macroeconomic variables (interest rates, employment, and inflation) and model the impact of these variables on real housing prices in Spain. The goal is to examine the relationship between these variables and analyze any interesting indications, to use empirical tests to support basic economy theory surrounding the relationships between the variables and provide policy implications. This subject is important because the entire world is now in the throes of recession due to the burst of a global asset bubble. With many financial institutions reeling from losses due to risky loans to home borrowers, it may be helpful to analyze the world’s most extreme housing scenario: the Spanish housing bubble. The Spanish housing bubble is different in many regards from the United States housing bubble. In Spain, housing is the dominant and most widely held asset, and is a substantial portion of the country’s GDP. Spanish housing data has been criticized for being unreliable and biased, so this analysis will discover whether the data can hold up to a series of tests and produce results that make sense. Readers will learn of the various events leading up to and after the bust of the Spanish property bubble and may be able to draw important implications from such analysis. The remainder of the paper is organized as follows: Section 2 will present a review of literature, Section 3 presents the empirical analysis and results, and Section 4 will provide a conclusion and policy implications. 2.0 TREND The Spanish economy has experienced robust growth since the mid-1990s which has mainly attributed to the housing sector. From 1995 to 2008, Spanish housing prices have risen 190%, making the U.S housing market pale in comparison (Desmond, 2008). From 1999 to 2005, prices for apartments and houses accelerated at an annual rate of 15% in Spain. The housing market has been integral component of the Spanish economy. Housing related industries compose 20% of the economy in Spain, as opposed to 10% in the U.S (Wharton, 2007). Spain has one of the highest rates of homeownership in the world at approximately 80%, higher than the estimated 67.8% homeownership rate in the United States. 1 Adding to this, the 1

2005 Spanish Survey of Housing Finances. http://www.bde.es/informes/be/ocasional/do0810e.pdf

Empirical Economic Bulletin, Spring 2009, Vol. 2

Spanish tax and legal system favor homeownership. Eviction is often an expensive and slow process for owners. Much of the lending done in Spain is related to the housing sector. According to Martinez (2008), 60 percent of the credit granted to the private sector residents is related to the activity in the real estate sector. Housing is the most important asset held by households, representing 66.1% of the total value of households’ real assets and 58.9% of the total value of all their assets. There are several unique characteristics of the Spanish housing bubble which are important to note. Robust housing price appreciation was coincidental with equally robust increases in housing supply; between 2000 and 2007, homes built averaged 600,000 annually, a number which exceeded the annual figures for Germany, France, the U.K and Italy combined. Also, adjustable (i.e, variable) rate mortgages (ARM) account for nearly all of the stock of mortgages (approximately 98%). This may be due to the expectations of low interest rates for the future due to the primary objective of the European Central Bank to fight inflation and bank’s credibility for doing so. Also to note, home equity withdrawals are a rare occurrences in Spain. In contrast to the United States, Spain was relatively secure from the wake of the sub-prime crisis, as the regulation of securitized assets in Spain was conservative. As opposed to the situation in the United States, where securitization of mortgages was performed to transfer risk, in Spain securitization was related mainly to funding purposes. Causes for the Spanish Housing Bubble In 1999, members of the European Union formally introduced the euro and subsequently the monetary policy of all members was under the control of the European Central Bank. Spain’s membership in this monetary union implied a sharp reduction in real interest rates. The reduction in real interest rates substantially decreased the cost of capital for investment, and lowered the discount factor for any expected future payoffs which should increase asset prices as a result (Ayuso et al., 2006). The reduction in real interest rates bolstered borrowing and heightened expectations for nominal stability of interest rates which allowed for the lengthening of maturities for mortgage borrowing from 10 to 28 years between 1990 and 2007 (De Lis et al., 2008).

Empirical Economic Bulletin, Spring 2009, Vol. 2

Demographic factors were also a significant factor. A generation of baby-boomers composed the majority of the population, who, with their accumulated savings, were able to take advantage of the ongoing housing bubble. Also, an explosion of immigration that took place in the last ten years provided an abundance of cheap, unskilled labor for the construction sector and an increased demand for housing. As of 2005, foreigners accounted for 7% of the population, a statistic which has by estimates reached approximately 10% as of 2008 (Ayuso et al., 2006). Additionally, speculation was fueled by an extraordinary influx of foreign-investors (those seeking to speculate, purchase vacation or second homes, or both). Credit expansion among households increased substantially in time with declines in the household saving ratio. In 2005 the household saving ratio was around four percentage points lower than the average over the first half of the previous decade, net financial saving (∆ in financial assets minus ∆ in financial liabilities) has fallen dramatically, and private, non-financial sector debt had risen to 160% of GDP, more than twice the 1995 ratio (Ayuso et al., 2006). In Figure 1, average house prices have risen higher than historical norms in the past 10 years, and have only just begun to decline. The author would like to point out it has been widely assumed that official figures from the Ministry of Housing underestimate the true extent of the current Spanish housing correction. The data used by this paper, published by the Ministry of Housing (MiV), is provided by the Association of Official Appraisal Agencies in Spain (ATASA). The fact that estimates are based on home appraisals presents a serious bias in the data; appraisal agencies are pressured to overvalue homes when banks wish to lend out more money, and pressured to undervalue homes during economic hard times. Adding to this problem, a Bloomberg article investigated the fact that many banks have ownership equity stakes in homevaluation agencies (Sills et al., 2008). Thus, it is highly likely that the housing data may not be the best representation of housing price evolution, though it does capture the trend.

Empirical Economic Bulletin, Spring 2009, Vol. 2

Figure 1: The Average Price of Housing per Meter Squared, 1995-2008

Source: Ministerio De Vivienda Martinez (2008) argues that over the past two years the rise in interest rates was a major cause of the bursting of the housing bubble, along with the over-valuation of house prices, the excess housing supply and the growing debt levels of Spanish families. The EURIBOR interest rate, which is commonly used for as the base interest rate for mortgage lending, remained very low during the 2003-2005 period. The ensuing housing price appreciation, along with a rise in stock markets, bolstered household and corporate net wealth. At the onset of the financial crisis, the EURIBOR is shown to begin a sharp downward trajectory as the European Central Bank responded to the financial turmoil. Figure 2 is graph of the EURIBOR, which provides the base rate for mortgage rates, implies a negative correlation with the price of housing.

Empirical Economic Bulletin, Spring 2009, Vol. 2

Figure 2: EURIBOR 12-month Interest Rate: 1999-2009

Source: Banco de Espana Figure 3 shows the movements of the Spanish Consumer Price Index from 1995 to 2008. In the beginning of the series, we see the decline of inflation. At that time, Spain had its own central bank, and was implementing monetary policies in order to begin the transition into a European Monetary Union. Upon formation of the EMU in 1999, inflation begins an upward trajectory and stays within the range of 3.50%-2.50%, until peaking at 4.50% in early 2008. The movements of inflation show a modest correlation with the upward trending housing data, though it is expected the initial downward trend of CPI may prove to be a source of problems in our results.

Empirical Economic Bulletin, Spring 2009, Vol. 2

Figure 3: Consumer Price Index from 1995‐ 2008

Source: Banco de Espana Figure 4 shows the trend in employment by millions from 1995 to 2008. Graph 5 shows the registered number of unemployed, in millions, by branch of activity from 2006 to the current quarter. Despite a large increase in the labor supply which stemmed from significant flows of immigration and an increase in female participation rates, the average national unemployment decreased from 22.9% in 1995 to a low of 8.26% in 2007, before climbing to 9.50% in early 2008. In January 2009, the number of unemployed people rose by nearly 200,000, a 6.4% increase over December and a 47% increase since January 2008 (Burnett, 2009). An article for the New York Times described how Spain suffers more than other European countries in terms of employment because of the vast oversupply of low-skilled jobs and the lack of investment in retooling the country’s rigid labor market. The government intends to pressure banks to free up liquidity as well as introduce a $13 billion public works program. Recent projections of the national average unemployment rate by the end of 2009 are approximately 16%. Figure 4: Employment in millions from 1995-2008

Empirical Economic Bulletin, Spring 2009, Vol. 2

Source: Banco de Espana

Figure 5: Registered Number of Unemployed (in millions) by Branch of Activity, 20062009

Source: Banco de Espana

3.0 LITERATURE REVIEW

Empirical Economic Bulletin, Spring 2009, Vol. 2

The workings of this paper will take example and guidance from several others. Apergis (2003) provides this paper with a blueprint for empirical analysis. The paper by Apergis analyzes the dynamic effects of macroeconomic variables on the price of new houses sold in Greece. The empirical methodology utilizes unit root tests, cointegration analysis, an error correction model, variance decompositions, and impulse responses which will all be utilized later. A paper by Ayuso, Blanco, and Restoy (2006) provides analysis of house prices and real interest rates in Spain. The paper analyzes the contribution of interest rates to the recent course of housing prices in Spain. Ayuso, Blanco and Retory create a refined asset pricing model that shows that changes in the discount factor cannot fully explain the trend of house prices in Spain, providing support for the inclusion of other macroeconomic variables. A paper by Kearl (1979) examines the influences of inflation on housing, mortgages and construction by constructing a model. Though this model cannot quantify unspecified uncertainties in economic conditions, the model supports evidence that inflation distortions have significant effect on the housing prices and thus the demand for housing. Katja (2006) tests for the existence of real estate bubbles using traditional unit root tests under the assumption that the dividend yield ratio in stock markets is comparable to the rentprice ratio in the housing market. The paper provided valuable insight into recent developments in housing markets and unit root based testing. 4.0 EMPIRICAL RESULTS AND ANALYSIS The empirical analysis was carried out using quarterly data from 1995 to 2008. The variables used in the empirical analysis are housing price (HP) obtained from the Ministry of Housing in Spain, the mortgage interest rate (calculated as the weighted average of 3-year mortgage loans) obtained from the Bank of Spain website (INT), prices measured by the consumer price index from the National Institute of Statistics (CPI), and employment measured in millions (EMP). The housing price data is the average price per meter squared of appraised housing in Spain, deflated by dividing it by CPI. The empirical analysis was performed using Eviews 6.0.

Unit Root Testing

Empirical Economic Bulletin, Spring 2009, Vol. 2

The variables were tested for unit root non-stationarity by using unit root tests proposed by Phillips and Perron (1987). If a time series has a unit root, then the time series is said to be non-stationary. Stationarity is a stochastic (in other words, random) process whose characteristics such as mean and variance do not change over time or position. A unit-root would imply that the variance of a time series would increase to infinity over time. The Philips-Perron test was chosen because the test is nonparametric with respect to nuisance parameters, allowing for a wider class of time series models in which there is a unit root. The method also has advantages when there are moving average components in the time series. The results are reported in Table 1. When second differences were used, unit root non-stationary could be rejected at the 1% significance level, suggesting that the variables HP, INT, and CPI are I(2) variables or integrated in the first order, which means that it takes one first difference of the series to make it stationary.

Table 1: Unit-Roots Without trend

(Philips- Perron)

Variable Levels

First

With Trend Second

Levels

First

Second

Differences Differences Differences Differences HP -0.76 -2.29 -7.15*** -1.1 -2.36 -10.02*** INT -1.16 -3.32** -6.02*** -0.98 -3.18 -5.83*** EMP -2.04 -2.29 -5.37*** 1.24 -2.52 -5.56*** CPI -2.06 -3.35** -4.69*** -2.19 -3.33* -4.56*** Figures in under first differences denote t-statistics. Note: *,**,*** depict each variable as a 10%, 5%, and 1% confidence level.

Cointegration Analysis

Empirical Economic Bulletin, Spring 2009, Vol. 2

The possibility of cointegration among these variables was examined next. Cointegration is property of time series variables; if two or more series are non-stationary, but a linear combination of them is stationary, the series are said to be cointegrated. We can test that there is a statistically significant connection between all the macroeconomic variables by testing for a cointegrating vector. The results are presented in Table 2. Allowing for linear deterministic trend in the data, and an intercept and trend in the cointegrating equation, both the trace eigenvalue test statistic and max eigenvalue test statistic indicate the presence of two cointegrating vectors. The following is the vector autoregression estimate. There appears to be evidence in favor of the existence of a common cointegrating vector among the variables under examination. Table 2: Unrestricted Cointegration Rank Test (Trace and Maximum Eigenvalue) Hypothesized No. of CE(s)

Trace Statistic

0.05 Max-Eigen Critical Value

Statistic

0.05 Critical Value

None * 115.7173 63.8761 49.98748 32.11832 At most 1 * 65.72986 42.91525 31.01759 25.82321 At most 2 * 34.71227 25.87211 24.18815 19.38704 At most 3 10.52413 12.51798 10.52413 12.51798 Trace test indicates 3 cointegrating eqn(s) at the 0.05 level Max-eigenvalue test indicates 3 cointegrating eqn(s) at the 0.05 level * denotes rejection of the hypothesis at the 0.05 level An Error Correction Model As cointegration is confirmed, we go ahead and estimate an error correction vector autoregressive (ECVAR) model. A vector error correction model is a restricted VAR designed for use with non-stationary series that are known to be cointegrated. The model is specified so that it restricts the long-run behavior of the endogenous variables to converge to their cointegrating relationships while allowing for short-run adjustment dynamics. The cointegration term, known as the error correction (EC) term, gradually corrects deviation from the long-run equilibrium through a series of partial short-run adjustments. Our ECVAR model will involve four variables (real housing prices, mortgage interest rates, CPI, and change in unemployment) along with a 2-lag specification. Interest rates are expected to have a negative sign, CPI is

Empirical Economic Bulletin, Spring 2009, Vol. 2

expected to have a positive sign, and employment is expected to have a positive sign. The ECVAR estimations provided the following results:

∆HP =

0.782∆HP(-1)

- 0.317∆HP(-2)

- 14.69∆INT(-1)

- 0.384∆INT(-2)

[ 4.40448]

[-1.38186]

[-122552]

[ -0.03194]

[ -1.69010]

+ 0.131∆EMP(-2)

- 11.59∆CPI(-1)

- 13.22∆CPI(-2)

+ 0.045∆EC

[ 4.37221] R-squared Adj. R-squared

[-0.80215]

[-0.58731] 0.774241 0.698989

0.07∆EMP(-1)

[1.79170]

EC is the error correction term. Figures in brackets denote t-statistics. The results received are somewhat disappointing at first glance. The variable of interest rates, when lagged once, was shown to have a negative effect on housing prices, as basic economy theory suggests. When INT is lagged twice, the variable changes into a positive sign. The variable of employment achieves the expected sign at the second lag. The variable of inflation does not have the expected sign at first or second lags. The negative and statistically significant coefficient of the EC term implies a significant adjustment to disequilibrium deviations of housing prices from their optimal level determined by the long-run (cointegrated) housing price equation. There are several reasons the results might not have met expectations. If one carefully examines the trends of our variables in the previous section, we can see how some problems may arise. Inflation, as well, was volatile and experienced several steep declines over the length of the time series, which may have affected its relationship with housing prices. Interest rates on the other hand have our expected signs and are negatively related to housing prices.The employment time series looks extremely positively correlated with housing prices, yet a first lag returns a negative sign. Variance Decompositions

Empirical Economic Bulletin, Spring 2009, Vol. 2

A variance decomposition shows how much variation is produced in an endogenous variable by a shock to a specific variable. In other words, variance decompositions provide information about the relative importance of each random innovation in affecting the variables in the VAR. Each column gives the percentage of the forecast variance due to each innovation, with each row adding up to 100.

Table 5: Variance Decompositions Variance Decomposition of HP: Period S.E. HP 1 4 8 12 16 20

15.94554 53.59956 124.4781 225.9642 348.9765 503.1644

100 68.08932 56.11677 70.68651 76.53153 80.09069

INT

CPI

0 28.38433 25.39279 19.10614 16.51407 14.2412

0 1.26368 15.98766 9.271156 6.408209 5.173149

EMP 0 2.262673 2.502779 0.936191 0.546191 0.494962

The results from Table 5 suggest that, up to 4 quarters, shocks to the housing mortgage rate account for more variation in real housing prices than variation produced by shocks to employment or inflation. Over a longer period, up to 20 quarters, shocks to both the housing mortgage rate and inflation account for more variation in real housing prices. The variable with the highest explanatory power over the variation in real housing prices is the housing mortgage rate, the second being inflation, and third being employment. Impulse Responses

Empirical Economic Bulletin, Spring 2009, Vol. 2

Figure

Response of HP to Cholesky One S.D. Innovations 200

150

100

-50

-100 2

6 HP

10 INT

12 CPI

EMP

6 An impulse response traces the effect of a one-time shock to one of the innovations (variables) on the current and future values of the endogenous variables. In other words, it shows the dynamic behavior of a variable due to random shocks in other variables. Table 6 shows the response of housing prices to shocks in the mortgage interest rate, CPI, and unemployment. The impulse response functions are based on 500 Monte Carlo replications. These impulse responses are sensitive to the ordering of the variables, so different ordering might be in order to provide support, if any, to the results. In Table 6, as basic economic theory would suggest, a positive (higher) shock to the mortgage interest rate indicates that as the cost of financing a house purchase increases, the

Empirical Economic Bulletin, Spring 2009, Vol. 2

demand, and thus the price for housing falls. In response to a shock in mortgage interest rates, housing prices reach their steady state level at around 6 quarters. A positive (increase) shock to inflation leads to higher housing prices; the impulse response does not show housing prices reaching their steady state level within 20 quarters. Finally, a positive shock in unemployment initially shows a decrease in housing prices (and thus demand for homes), while by 7 or 8 quarters housing prices continue to rise.

5.0 CONCLUSION AND POLICY IMPLICATIONS The main objective of this paper was to analyze the dynamic effects of certain macroeconomic variables (the mortgage interest rate, inflation and employment) on real housing prices. The results suggest that real housing prices do respond to specific economic variables. Variance decompositions show that the mortgage interest rate is the variable with the highest explanatory power over the variation in real housing prices, followed by inflation, then employment. Impulse responses show that a positive shock in the mortgage interest rate depresses real housing prices and will eventually lead to higher housing demand, while a positive shock in inflation and employment will increase real housing prices and will lead to lower housing demand. After analyzing the results of our empirical tests, it is apparent that the most important factor for determining policies relating to housing are mortgage interest rates. It has been argued that low inflation environments will cause people to overvalue long-duration assets, i.e, a money “illusion” (Brunnermeir and Julliard, 2007), which would tend to lead to asset bubbles and busts. Due to the focus of the European Central Bank on maintaining a low inflation environment, such an argument might imply that Spain must find a solution to the natural tendency to overvalue long-duration assets. Economists have discussed the susceptibility of the Spanish economy to economic cycles. The integration of Spain into the European Monetary Union (EMU) and the ensuing reduction in real interest rates and increased expectations for monetary stability can be cited as one significant source of the housing bubble. The Bank of Spain, with its monetary defenses taken away, had no defense against the easing of interest rates by the EMU.

Empirical Economic Bulletin, Spring 2009, Vol. 2

It may important to note the conflicts of interest faced by political officeholders that may have resulted in failures to recognize or act against the housing bubble. The fact that housing is the biggest asset of the voting base, and that the construction sector produces a large amount of employment will be on the mind of any political incumbent when making policy decisions, or running for a second term. Policy Implications Policy implications would have to take into consideration that Spain no longer has control over interest rates. Without the control over interest rates, Spain could concentrate on collecting more reliable data on housing, tax incentives, reducing economic dependency on low skill labor and increasing regulation. Spain could improve the integrity of their data. One of the problems encountered during the creation of this paper was accessing even marginally reliable data on housing prices. An official Spanish Housing Price Index that is derived from actual housing transaction prices is published by the Institute for National Statistics yet only has data from 2007. The housing prices used in the time series were taken from the Ministry of Housing in Spain, whose housing data has been criticized due to the fact that it is based off of appraisals. In Spain, home appraisal agencies are often owned by banks! The potential for bias is obvious. Spain needs to provide adequate, reliable data in order to better understand and prevent housing bubbles. Spain could remove tax incentives which promoted house ownership, such as, the income tax deduction for owning a house. The tax deduction makes housing more valuable in comparison to other investment decisions, and to renting. This tax incentive without question aided in the creating one of the highest home ownership rates in the world ( approximately 81.0% in 2005). Spain must reduce its economy dependence of low skill labor and increase regulatory oversight over immigration and abuse of illegal labor. The existence of cheap, illegal immigrant labor had aided in overdevelopment of real estate projects, has expanded the black market, and has cut legal residents out of work. Also, the predominance of low skill labor in Spain’s economy makes the economy more susceptible to volatility. When times are good, low skill employment will be in high demand, and when times are bad, low skill workers will quickly find

Empirical Economic Bulletin, Spring 2009, Vol. 2

themselves out of work. Spain has been noted for its rigid labor market; this needs to be remedied. Spain must clamp down on the current state of appraisals/valuations. As mentioned earlier, there is an unmistakable conflict of interest, as banks have been known to own ownership stakes in the home-valuation agencies they use. It is only recently, in December 2007, that a law was instituted that limited the amount of business appraisers can take from clients with ownership stakes (such as banks).

BIBLIOGRAPHY Anonymous, (2007), “Real Estate Collapses, and Spain Trembles”, Universia Knowledge@Wharton, http://www.wharton.universia.net/index.cfm?fa=viewfeature&id=1341&language=english Apergis, Nicholas, (2003), “Housing Prices and Macroeconomic Factors: Prospects within the European Monetary Union”, International Real Estate Review, Vol. 6 No. 1: pp 63-74. Arellano, Manuel and Bentolila, Samuel, (2009), “Quien es responsible de la burbuja inmobililaria?” El Pais.com,http://www.elpais.com/articulo/semana/Quien/responsable/burbuja/inmobiliaria/elpepu

Empirical Economic Bulletin, Spring 2009, Vol. 2

econeg/20090222elpneglse_2/Tes, Translated into English: http://www.creditwritedowns.com/2009/02/spain-who-is-responsible-for-the-propertybubble.html Ayuso, J. and Blanco, R. and Restoy, F.,(2006), “House prices and real interest rates in Spain”, Banco de Espana Ocassional Papers, 0608. Brunnermeir, Markus K. and Julliard, Christian, (2006), “Money Illusion and Housing Frenzies”, C.E.P.R Discussion Papers, 6183. Burnett, Victoria, (2009), “Spain’s Unemployment Rose Sharply in January”, The New York Times, < http://www.nytimes.com/2009/02/04/business/worldbusiness/04specon.html?_r=1> De Lis, Santiago Fern and Herrero, Alicia Garcia, (2008), “The Housing Boom andButs in Spain: Impact of the Securitisation Model and Dynamic Provisioning” < http://www.allbusiness.com/economy-economic-indicators/economic-policy/11760429-1.html> Katja, Taipalus, (2006), “A global house price bubble? Evaluation based on a new rent-price approach”, Bank of Finland Research Discussion Papers, 29/2006. Kearl, J.H ,(1979), “Inflation, mortgages, and housing”, Journal of Political Economy, 87, 1-29. Lachman, Desmond, (2008), “The Housing Pain in Spain”, New York Sun. Accessed 03/17/09. http://www.nysun.com/opinion/housing-pain-in-spain/72501/ Martinez, Maria Teresa Sanchez, (2008), “The Spanish Financial System: Facing up to the Real Estate Crisis and Credit Crunch”, European Journal of Housing Policy, 8, 181-196. Peter C.B. Phillips & Pierre Perron, (1986), “Testing for a Unit Root in Time Series Regression”, Cowles Foundation Discussion Papers, 795R. Sills, Ben and Duarte, Esteban, (2008), “Spain’s Inflated Home Values Infect Mortgage Bonds”, Bloomberg.com < http://www.bloomberg.com/apps/news?pid=20601170&refer=home&sid=avLtlLCD0wcQ> Spain Real Estate Index, [source for housing price data] < http://www.spainrei.com/Spain-Property-Prices.shtml>

Empirical Economic Bulletin, Spring 2009, Vol. 2

The Effect of Crime Rates on Home Prices: A Hedonic Study

John Paul Goncalves1

Abstract: The hedonic regression model in this study is from a Florida based study that has been applied on a national level. The results of this research will indicate the most significant variables which support the overall effect on the price of a home. The emphasis of this study is to examine the overall impact of crime rates on average home prices in America’s state capitals.

JEL Classification: R21, R22

Keywords: crime rates, home price, state capitals

Department of Economics, Bryant University, 1150 Douglas Pike, Smithfield, RI 02917. Phone: (401) 578-4307. Email: jgoncal1@bryant.edu

Empirical Economic Bulletin, Spring 2009, Vol. 2

1.0 Introduction Crime is a challenge populations have faced since the creation of the first villages in this world. Basic theory shows that villages, towns, or cities that have minor or no existent form of laws or punishment system will experience more crime than pockets that have introduced these safety measures. The United States of America, a very civilized high income country, has sought for low crime. With the investment of trillions of dollars into introducing and creating police academies, barracks, law enforcement systems, judicial systems, and other precautionary investments, safety and lower crime rates should be ensured. Yet, only in the year 2005 did the US experience a decrease in its total crime rates. Considering these ‘precautionary investments’ have been in place since the US won its independence it is easy to state that the system is inefficient and that marginal costs greatly outweigh the marginal benefits. Should America worry about this inefficiency when crime rates have been diminishing? Yes, the US should be concerned. Why? Well, as known, the US and the rest of the world are currently experiencing recessionary-like shocks. Furthermore, historically, crime wanes during periods of economic growth and surges during economic downturns.1 Therefore, many cities across the US are possibly on the verge of a major crime wave. How legitimate is this historical fact? As economists, academics, criminologists, and demographers argue, crime is caused by the economy, unemployment, racism, and poverty.2 For that reason the likelihood of a crime wave occurring in the US is plausible. A crime wave is of course unsafe and problematic for the public, but at the same time very costly. For years, police chiefs have argued that safer cities are better for business, increase tax revenue and help property values. Consequently, if crime increases, state capitals must cope with deteriorating communities which lower property value, which then lower the amount of tax revenue and new business creation. For instance, James Larsen, a professor at Wright State University, found that home sales prices located within one-tenth a mile of a sex offender

1 2

Gabriel Kahn, “Top Cops in Los Angeles says cutting crime pays.” WSJ, November 29, 2009 Gabriel Kahn, “Top Cops in Los Angeles says cutting crime pays.” WSJ, November 29, 2009

Empirical Economic Bulletin, Spring 2009, Vol. 2

dropped 17% in value.3 Moreover, due to US crime rates, it is expected that property values across the country will decline $1.2 trillion in 2008.4 All in all, the justification that, an increase in crime rates will decrease property values can be made. Hence, Hellman and Naroff’s (1979) and Rizzo’s (1979) verification that crime has a trivial impact on home prices is also true. For these reasons, this study should similarly resemble the facts established above, and the previous results of researchers like Hellman, Naroff, and Rizzo. Additionally, this national level study will introduce a first look into the determinants of house prices using weighted variables such as cost of crime indexes. The structure of this study is as follows. Section two introduces crime trends in the US. Section three introduces the idea behind cost of crime and its imperative purpose in this study. A literature review follows in section four to help support the overall idea and methodology. Section’s five and six establishes the data, empirical methodology, results to the empirical methodology, and an analysis on the results. The remaining sections include: implications with the study, and concluding remarks.

2.0 Crime Trends Figure 1

This figure shows total reported crimes, both violent and property, in the U.S. Total crime has diminished in the long run, but has remained steady in recent years. Source: FBI Crime in United States (CIUS) index (2008)

3 4

Lisa Scherzer, “Three Home Value drains to Avoid.” WSJ, June 16, 2008 Emily Frielander, “Cities dealing with rise in abandoned properties.” WSJ, January 28, 2008

Empirical Economic Bulletin, Spring 2009, Vol. 2

Figure 2

This figure shows the total reported crime and the average property value of single family dwellings in U.S. state capitals. Even though crime has diminshed in the long run, as shown in figure 1, it still does have an effect on the average house price. Source: Census Bureau, FBI CIUS index, and Trulia.com (2008)

Figure 3

This figure breaksdown the total reported crime rate in each geographical region of the U.S.

Empirical Economic Bulletin, Spring 2009, Vol. 2

Source: FBI CIUS index (2008)

3.0 The Cost of Crime Normal crime indexes create an equilibrium that does not exist. A weighted index using the cost of crime will remove this equilibrium. Normally, when police are comparing public safety across jurisdictions, the total number of offences reported in a jurisdiction will be divided by the population of that jurisdiction. But how does this public safety figure truly measure the differences of safety between jurisdictions? For instance, in 2007, town A, with a population of a 1000, reported 100 murders while town B, with a similar population, reported 100 burglaries. In the end, both towns have a crime index of .1 or 10% (100 murders or burglaries/1000 population). How is this index an accurate assessment of public safety differences when burglaries and murders are proportionate? Luckily, this problem is simple to fix. By weighting the index through the figures from Cohen et al. (1995), the true implicit costs will be recognized and the bias in which murders and burglaries are equal will be removed. Town

Population Reported Offences

Normal Crime Index

Cohen cost index

New Index level

1000

100 murders

.1 or 10%

$2,740,000/murder

$274,000,000

1000

100 burglaries

.1 or 10%

$1,500/burglary

$150,000

After applying the Cohen et al. index to the earlier example, the equilibrium between the two towns is dramatically altered. Due to the fact that Cohen et al. found murders to obtain a more significant impact on victims than a burglary, town A now appears to be a hazardous jurisdiction compared to town B. Allen and Rasmussen’s (2001) study found similar results. By ranking cities by Cohen et al. (1995) index rather than the traditional index crime, Tallahassee’s metropolitan area improved from 5th in the nation to 54th and New York diminished from 73rd to 7th. Overall, this example clarifies the significance and reliability of Cohen et al. index. On the other hand, many

Empirical Economic Bulletin, Spring 2009, Vol. 2

economists and psychologists question this reliability. Butterfield (1996) criticizes the method behind the victim cost estimates because determining psychic costs seems impossible. Once again, the Cohen et al. victimization cost index will be used in the analysis of how crime effects house prices in America’s state capitals. The index will weigh the crimes in which the FBI has documented as reported in the year of 2008 in these state capitals. Although this paper does adopt this cost index as a more reasonable and preferred method of measure, it does not necessarily accept the authenticity of these specific estimates.

4.0 Literature Review Hedonic modeling breaks-down the dependent variable being studied into its essential characteristics, and then estimates values for each of these characteristics. This style of regression is very popular in real estate economics. Due to the fact that buildings are a heterogeneous good, the hedonic model approach focuses on the buildings characteristics, such as bedrooms, and lot size for a better interpretation of the total value or price of a building. Cohen (1995) and Rasmussen (1990) dispute that hedonic studies of housing markets show that the value of spatial differences in education, air pollution, property taxes, and other location specific attributes will be capitalized into house prices. Allen and Rasmussen (2001) created a hedonic model to measure the effect crime has on the average home price. This study was focused on the Jacksonville, Florida area, and included 2880 observations. With this size observation and the in-depth hedonic model, which has assistance from Cohen et al. crime index to create some weighted variables, Allen and Rasmussen were able to prove crime has a significantly negative effect on home prices in the Jacksonville area. Therefore showing the true importance a weighted crime index has in hedonic studies. This data source led to the creation of the hedonic model which is being used in this empirical study. Thaler (1978) approximated the impact property crime in Rochester, NY had on nearby home prices by using a cost of crime and implicit price model. Cohen (1990) concurs by stating it is necessary with hedonic models to use the cost of crime rather than index crime data to estimate the effect on home prices. Hellman and Naroff (1979) and Gibbons (2004) conducted studies to find the impact crime had on urban communities and property values. Overall, the two studies cause similar effects.

Empirical Economic Bulletin, Spring 2009, Vol. 2

For instance, Hellman and Naroff believe crime does cause variations in home prices which then reduce property tax revenue for communities. Meanwhile, Gibbons suggests the direct costs associated with property crime may discourage home-buyers, inhibit local regeneration and catalyses a downward spiral in neighborhood status; basically this lowers demand, in which lowers prices. All in all, deterioration from crime in urban communities leads to downward pressure on home prices. This study adds onto to the overall research started by Allen and Rasmussen by applying it to a national level. Most hedonic modeled studies on crime and real estate property values are based in one city or district. Unfortunately most urbanized large cities (which are the most commonly used) have higher crime rates than the millions of towns in America; therefore the ‘value’ found in Allen and Rasmussen’s can never be used to report the deterioration crime has on property values nationwide. This national level study will remove that bias and create a more consistent ‘value’ for the many towns in America.

5.0 Data and Empirical Methodology The data for this study comes from various sources. The Census Bureau provided data on average house prices for each of the fifty capitals in the United States in 2008. These average prices were then used to search homes that sold at that average price between January 1, 2008 and December 31, 2008 on Trulia.com. The remaining data points are placed under three different categories in the Hedonic model:

Pi = f (Si, Ni, Ci) Where: Pi = the selling price of the home Si = a vector of housing and lot characteristics Ni = a vector of neighborhood characteristics Ci = the number of crimes and the estimated cost of the number of crimes Ten homes were picked from each capital, and then averages of each characteristic of the ten homes were created into variables. These independent variables are sited under the housing and lot characteristics (Si). The variables are as follows: number of bedrooms and bathrooms;

Empirical Economic Bulletin, Spring 2009, Vol. 2

square footage of home and lot; age of home in years; and dummy variables for pool, fireplace, central air, fenced yard, gated community, and waterfront property. Zietz and Sirmans’ (2008) quantile regression on determinants of house prices supports these independent variables. For example, variables such as square footage, lot size, bathrooms, and floor type impact selling price, while other variables have a relatively constant effect on selling price. For better understanding please refer to table 2 and 3. Neighborhood characteristics were provided by the Census Bureau and Federal Bureau of Investigation (FBI). The independent variables under this vector include: population for each capital; percentage of population that is Caucasian, African American, and Hispanic; the proportion of population that is either “17 & under,” “18-24,” or “55 & over;” and the median income in each capital. Overall, this creates a total of eight independent variables for “Ni”. The reasoning behind creating three variables for age is because certain age groups commit more crimes than others. As Lynch and Rasmussen (2001) found in their study, the population between 18 and 24 years old is the most crime prone portion in the United States. Therefore it is expected that this group has a significantly negative effect on home prices. Furthermore, the median income level is expected to have a positively significant affect on home prices as the level increases. Income and personal property investment are believed to be positively related. Thus, if median income increases then investment increases, which improves the appraisal and sale price of a home. For better understanding please refer to table 2 and 3. The FBI provided the index of data on crime rates for each capital. The index consisted of counts on violent crimes (manslaughter/murder, forcible rape, robbery, aggravated assault) and property crimes (burglary, larceny-theft, motor-vehicle theft, arson). Questions might arise on why variables such as motor-vehicle theft were used when analyzing house prices. Well, home prices are not only valued through appraisals, structural components, and sale prices, but also through the laws of supply and demand. For instance, if a city has a 70% motor-vehicle theft possibility, no matter how well a home fits a buyer’s necessities, demand will decrease due to the greater chance of having their car stolen. As demand decreases the price of homes in the city will follow. That is why, when accounting for crime rates all forms must be taken into consideration. Moreover, Cohen et al. 1995 study provides the approximated cost for each form of crime listed above. The costs were then administered to the count of crimes each city experienced.

Empirical Economic Bulletin, Spring 2009, Vol. 2

This formed the weighted index which will also be used during the regression. As stated earlier, by using a weighted index it removes the bias a normal counted index creates. It is expected that the weighted index will be negatively related to home prices (i.e. higher cost of crime leads to lower home prices). For better understanding please refer to table 2 and 3. 6.0 Empirical Results and Analysis Table 1 reports the results from the hedonic regression model. The table includes two regressions; both of which are indistinguishable besides the removal of a fenced yard and gated community dummy, addition of percentage of population 18-24, and most importantly the removal of reported crime variables. The fenced yard and gated community variables were removed due to high triviality. The raw data hardly included any observations with these options. As shown in table 1, both variables have a negative effect on house prices. This is obviously false. As Allen and Rasmussen (2001) found in their study, fenced yards and gated communities are valued more in a high crime area because these attributes represent a barrier between the house and the street. Meaning, consumers are willing to pay a higher price for a home including these attributes, not pay lower. If the observations, included in this study, had a proper mix of fenced, non-fenced, gated community, and non-gated community properties, rather than strictly non-fenced and nongated communities, then the two variables would have been more effective. Overall, the variables were removed for regression 2. Differing from these variables is the percentage of population 18-24 years of age variable. As stated earlier, this was added onto regression 2 because Allen and Rasmussen’s study, along with reliable news sources, provided facts supporting that this age bracket commits the highest amount of crime in the United States. Therefore, it would be expected that 18 to 24 year olds would have a negative effect on house prices. Furthermore, table 2 supports the negative relationship. In fact, this age bracket reduces house prices by 1.44%. On a side note, as revealed in table 1, the independent variables that affect house prices in a statistically significant manner (meaning the 1%, 5% and 10% significance levels) in regression 1 are size of home (sq ft), age of home, central air, waterfront property, population, and median income. Little change occurred in regression 2; the statistically significant variables that affect house price are size of home (sq ft), lot size, age of home, pool, waterfront property, population, percentage of population that is Caucasian, and median income. Unfortunately, the weighted cost

Empirical Economic Bulletin, Spring 2009, Vol. 2

of crime variables were not found to be significant at either the 1%, 5%, or 10% levels, but the expected relationship to the dependent variable was proven true. As table 3 reflects, the two weighted crime index variables (property and violent) are expected to have a negative effect on house prices. Table 1 reports that property crime has a 3.99% or $8,841 depreciation on home values in these state capitals5; meanwhile, violent crime has a 2.67% or $5,916 depreciation on home values in the same state capitals. The outcomes of the expected relationship to the dependent variable are identical with Allen and Rasmussen’s study, but the ‘value’ from the regression output differs. They found property crime would decrease property values by $206 and violent crimes to reduce it by $145. Why such a noteworthy difference? Perhaps, due to Allen and Rasmussen (2001) focus on single family homes in the Tallahassee area, their mean home price is at a lower value due to observations sharing similar traits and buyer patterns. There mean level house value is $95,532 and this study’s is $221,592. At the national level, the observations are exposed to different characteristics that affect their appraisal value and sale price; such characteristics are home features, land and property values, buyer patterns and demand. In addition, by this study having a higher mean level, the percentage affect the weighted crime indexes have on home values is larger. On an end note, these two dollar ‘values,’ are superior estimates which may be interchangeably used on any home in America. 7.0 Implications with Study Alas, the two costs of crime weighted variables were not found to be significant in the regression output. It is believed that the low sample size and high amount of independent variables produced a problem with the degrees of freedom. All in all, significance was not apparent in many variables, two of which being the variables with the outright highest importance to the study. Even though the two variables did not show significance at the 1%, 5% or 10% level, they did have the predicted negative relationship to the dependent variable. As stated earlier, this can all be found in table 1 and table 3. Nonetheless, even with the variables not being significant, there is still enough evidence and support from other previous studies to support the outcome. Thaler (1978), Taylor (1995), Rizzo (1979), Lynch and Rasmussen (2001), and Rasmussen and Zuehlke (1990) all have found crime 5

The 3.99% taken from table 1 is applied to the mean selling price of home variable in table 4

Empirical Economic Bulletin, Spring 2009, Vol. 2

to have a negative impact on the average house price within a surrounding area; majority of the time the surrounding area has been a large city, just as this study is based on sizeable state capitals. 8.0 Conclusion In more recent times, the value of homes has been decreasing due to economic factors, never mind the many other external factors such as crime. Therefore if homeowners live in an area with high crime they have much at stake. In fact, with the use of macro-level data, this study finds that homes values can experience negative affects as high as $8,841 due to crime. A deplorable figure that could diminish appraisal levels, supply markets and/or demand markets of towns, states, or nations. Though this study provides several appealing results, this analysis was only for US state capitals, and it is possible the same results will not apply to other cities and towns nationwide or in other countries. This study could be replicated or manipulated to fit the needs of other future studies for similar hedonic studies, or weighted cost of crime analysis.

Empirical Economic Bulletin, Spring 2009, Vol. 2

References Butterfield, F. (1996) Prison: where the money is, New York Times, 2 June, p. 16. Cohen, M. A., T. R. Miller, and B. Wiersema. “Crime in the United States: Victim Costs and Consequences.” Final Report to the National Institute of Justice, 1995 Emily Friedlander, “Cities dealing with rise in abandoned properties.” WSJ, January 28, 2008 Gabriel Kahn, “Top Cops in Los Angeles says cutting crime pays” WSJ, November 29, 2008 Gibbons, Steve. “The Cost of Urban Property Crime.” The Economic Journal v. 114 22pp Nov. 2004 Hellman, Daryl and Joel Naroff. “The Impact of Crime on Urban Residential Property Values.” Urban Studies v. 16 7pp Feb. 1979 Lisa Scherzer, “Three Home Value drains to Avoid, WSJ, June 16, 2008 Lynch, Allen and David W. Rasmussen. “Measuring the Impact of Crime on house prices.” Applied Economics v. 33 1981-1989 Nov. 2001 Peek, Joe and James A. Wilcox. “The Measurements and Determinants of Single-Family House Prices.” AREUEA Journal vol. 19 31pp Nov. 1991 Rasmussen, D. W. and T. W. Zuehlke. “On the Choice of Functional Form for Hedonic Price Functions.” Applied Economics vol. 22 431-438 Mar. 1990 Rizzo, M. J. “The Cost of Crime to Victims: an Empirical Analysis.” Journal of Legal Studied vol. 8, 177-205 1979 Taylor, Ralph. “The Impact of Crime on Communities.” Annals of the American Academy of Political and Social Science vol.539 17pp. May 1995 Thaler, R. “A note on the Value of Crime Control: Evidence from the Property Market.” Journal of Urban Economics vol. 5 137-145 1978

Empirical Economic Bulletin, Spring 2009, Vol. 2

Zietz, Joachim, E. M. Zietz and G. Sirmans. “Determinants of House Prices: A quantile regression approach.” The Journal of Real Estate Finance and Economics vol.37 317-333 Nov 2008 All data sources are from the: Census Bureau, Federal Bureau of Investigation, and Trulia’s Real Estate website Table 1: Regression Results

Variable

Regression 1

Regression 2

Size of home(sq ft) Lot Size (sq ft) Bedrooms Bathrooms Log(age of home) Pool (dummy) Fenced Yard (dummy) Gated Community (dummy) Fireplace (dummy) Central air (dummy) Waterfront property (dummy) Population Percentage of Pop. Caucasian Percentage of Pop. African American Percentage of Pop. Hispanic Percentage of Pop. 17 & under Percentage of Pop. 18-24 Percentage of Pop. 55 & over Median Income Log(property crime) Log(violent crime) Log(cost of property crime) Log(cost of violent crime) Adjusted R-Squared F-Statistic

0.0000781** 0.00000152 0.008838 -0.057953 0.133269*** -0.067385 -0.033424 -0.01392 0.03597 0.077085* 0.234036** 0.000000205** -0.001957 -0.000522 0.004305 -0.001493

0.0000985*** 0.00000204* -0.008507 -0.034343 0.120889*** -0.09712**

-0.003971 0.0000124*** 0.026386 -0.150789 -0.036795 0.087093 0.810289 9.542349

0.04659 0.05958 0.191848** 0.000000194** -0.003858** -0.000421 0.005799 -0.00394 -0.014468 0.000999 -0.0000130***

-0.039917 -0.026689 0.816403 11.76572

Note: ***, **, and * denotes significance at 1%, 5%, and 10%

Empirical Economic Bulletin, Spring 2009, Vol. 2

Table 2: Variable Definition Variable

Definition

Size of home (Sq. ft)

The square footage of living area

Lot Size (Sq. ft)

The square footage of the lot the home is built on

Bedrooms

Number of bedrooms in the home

Bathrooms

Number of bathrooms in the home

Log Age of Home

The log of the age of home in years

Pool (dummy)

Whether or not the home has a pool

Fenced Yard (dummy)

Whether or not the home has a fenced yard

Gated Community (dummy)

Whether or not the home is in a gated community

Fireplace (dummy)

Whether or not the home has a fireplace

Central air (dummy)

Whether or not the home has a central air cooling system

Waterfront Property (dummy)

Whether or not the home is a waterfront property

Population

Total population of the state capital

Percentage of Pop. Caucasian

Percentage of population in state capital that is Caucasian

Percentage of Pop. African American

Percentage of population in state capital that is African American

Percentage of Pop. Hispanic

Percentage of population in state capital that is Hispanic

Percentage of Pop. 17 & under

Percentage of population in state capital that is 17 & under

Percentage of Pop. 18-24

Percentage of population in state capital that is 18-24

Percentage of Pop. 55 & over

Percentage of population in state capital that is 55 & over

Median Income

The median income of the population in the state capital

Log of Total Property Crime

Total reported property crimes in state capital

Empirical Economic Bulletin, Spring 2009, Vol. 2

Log of Total Violent Crime

Total reported violent crimes in state capital

Log of Total Cost of Property Crime1

Total cost of reported property crimes in state capital

Log of Total Cost of Violent Crime1

Total cost of reported violent crimes in state capital

measured accordingly with Cohen et al. cost of crime index

Table 3: Expected Sign Chart Variable

Data Source

Expected

Size of home(sq ft)

Trulia.com

(+)

Lot Size (sq ft)

Trulia.com

(+)

Bedrooms

Trulia.com

(-)

Bathrooms

Trulia.com

(-)

Log(age of home)

Trulia.com

(-)

Pool

Trulia.com

(+)

Fenced Yard

Trulia.com

(+)

Gated Community

Trulia.com

(+)

Fireplace

Trulia.com

(+)

Central air

Trulia.com

(+)

Waterfront property

Trulia.com

(+)

Population

FBI

(-)

Percentage of Pop. Caucasian

Census Bureau

(-)

Percentage of Pop. African American

Census Bureau

(-)

Percentage of Pop. Hispanic

Census Bureau

(-)

Percentage of Pop. 17 & under

Census Bureau

(-)

Percentage of Pop. 18-24

Census Bureau

(-)

Percentage of Pop. 55 & over

Census Bureau

(+)

Log(property crime)

FBI

(-)

Empirical Economic Bulletin, Spring 2009, Vol. 2

Log(violent crime)

FBI

(-)

Log(cost of property crime)

FBI/National Institute of Justice

(-)

Log(cost of violent crime)

FBI/National Institute of Justice

(-)

Table 4: Mean and Standard Deviation of variables used in housing market analysis Variable Selling Price of Home1 Log Selling Price of Home Size of Home (Sq ft) Lot Size (Sq ft) Bedrooms Bathrooms Log Age of Home Pool (dummy) Fireplace (dummy) Central Air (dummy) Waterfront Property (dummy) Population Percentage of Pop. Caucasian Percentage of Pop. Hispanic Percentage of Pop. African American Percentage of Pop. 17 & under Percentage of Pop. 18-24 Percentage of Pop. 55 & over Median Income1 Total Property Crime1 Total Violent Crime1 Total Cost of Property Crime1 Total Cost of Violent Crime1 1 measured in US currency ($)

Empirical Economic Bulletin, Spring 2009, Vol. 2

Mean 221592 5.307472 1799.447 12646.38 3.0212766 2.11702128 1.367905 0.276596 0.638298 0.489362 0.042553 274320.04 .7830426 .02810638 .1167447 .2770851 .1418936 .2125106 50381.42 14853 2429 20,547,932 134,140,043

Standard Deviation 104123.926 0.177271 524.4969 13967.96 0.6075382 0.56349806 0.467091 0.452151 0.485688 0.505291 0.20403 304173.92 .1480769 .0268249 .1211297 .03438202 .01357305 .02728981 7602.874 17667.237 2839.436598 27,613,128 171,033,669

Table 5: Determinants of House Prices Variables

Coefficient

Std. Error

t-Statistic

Prob.

Size of Home (Sq. ft) Lot Size (Sq. ft) Bedrooms Bathrooms Log Age of Home Pool (dummy) Fenced Yard (dummy) Fireplace (dummy) Central Air (dummy) Waterfront Property (dummy) Gated Community (dummy) Population Percentage of Pop. Caucasian Percentage of Pop. Hispanic Percentage of Pop. African American Percentage of Pop. 17 & under Percentage of Pop. 55 & over Median Income Log Total Property Crime Log Total Violent Crime Log Total Cost of Violent Crime Log Total Cost of Property Crime Constant

0.0000781 0.00000152 0.008838 -0.057953 0.133269 -0.067385 -0.033424 0.03597 0.077085 0.234036 -0.01392 2.05E-07 -0.001957 0.004305 -0.000522 -0.001493 -0.003971 0.0000124 0.026386 -0.150789 0.087093 -0.036795 4.888821

0.0000364 0.00000119 0.040737 0.036739 0.040489 0.041861 0.031623 0.030181 0.041373 0.089741 0.052839 8.75E-08 0.002303 0.006936 0.002281 0.005004 0.006093 0.00000237 0.188645 0.102065 0.098738 0.177647 0.630297

2.142413 1.272231 0.21695 -1.577432 3.291462 -1.609722 -1.056954 1.191839 1.863179 2.60791 -0.263448 2.346569 -0.850056 0.620775 -0.22874 -0.298429 -0.651732 5.239734 0.139874 -1.477377 0.882059 -0.207127 7.756372

0.043 0.216 0.8302 0.1284 0.0032 0.1211 0.3015 0.2455 0.0753 0.0157 0.7946 0.0279 0.4041 0.5409 0.8211 0.7681 0.521 0 0.89 0.1531 0.3869 0.8377 0

R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood F-statistic Prob(F-statistic)

Empirical Economic Bulletin, Spring 2009, Vol. 2

0.905144 0.810289 0.077212 0.137119 70.48065 9.542349 0

Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion Hannan-Quinn criter. Durbin-Watson stat

5.307472 0.177271 -1.9779 1.033144 1.622382 2.020402

Table 6: Determinants of House Prices using Weighted Index Variables Size of Home (Sq. ft) Lot Size (Sq. ft) Bedrooms Bathrooms Log Age of Home Pool (dummy) Fireplace (dummy) Central Air (dummy) Waterfront Property (dummy) Population Percentage of Pop. Caucasian Percentage of Pop. Hispanic Percentage of Pop. African American Percentage of Pop. 17 & under Percentage of Pop. 18-24 Percentage of Pop. 55 & over Median Income Log Total Cost of Violent Crime* Log Total Cost of Property Crime* Constant R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood F-statistic Prob(F-statistic)

Coefficient

Std. Error

t-Statistic

Prob.

9.85E-05 2.04E-06 -0.008507 -0.034343 0.120889 -0.09712 0.04659 0.05958 0.191848 1.94E-07 -0.003858 0.005799 -0.000421 -0.00394 -0.014468 0.000999 1.30E-05 -0.039917 -0.026689 4.955803

3.35E-05 1.15E-06 0.035681 0.035055 0.036856 0.035826 0.028197 0.038029 0.084128 8.30E-08 0.001841 0.006618 0.002227 0.004683 0.012028 0.006535 2.47E-06 0.072296 0.091495 0.492373

2.938795 1.766758 0.238418 0.979691 3.280007 2.710893 1.652292 1.566687 2.28043 2.34314 2.096232 0.876229 -0.18912 -0.8414 1.202889 0.152915 5.253894 0.552138 0.291703 10.06515

0.0067 0.0886 0.8134 0.3359 0.0029 0.0115 0.1101 0.1288 0.0307 0.0267 0.0456 0.3886 0.8514 0.4075 0.2395 0.8796 0 0.5854 0.7727 0

0.892237 0.816403 0.075958 0.155778 67.48243 11.76572 0

Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion Hannan-Quinn criter. Durbin-Watson stat

5.307472 0.177271 2.020529 1.233232 1.724264 2.052425

Note:* signifies Cohen et al. Cost of Crime Index

Empirical Economic Bulletin, Spring 2009, Vol. 2

State Level Comparison of Factors Contributing to Rising Inpatient Hospital Costs

Jeff Fontaine1

Abstract: This paper examines an array of factors that contribute to rising inpatient healthcare costs. The study utilizes existing information from previous studies and applies its methods to a 50-State comparison. Using state-level inpatient costs per day as the dependent variable, an ordinary linear regression (OLS) model has been used to determine which of the independent variables contributes significantly to the rising costs.

JEL Classification: I11 Key Words: Health Care

Bachelor of Science in Business Administration Undergraduate, Applied Actuarial Mathematics Concentration, Economics Minor. Bryant University, 1150 Douglas Pike, Smithfield, RI 02917 Email: jfontain@bryant.edu

The author gratefully acknowledges the help and guidance of Dr. Ramesh Mohan

Empirical Economic Bulletin, Spring 2009, Vol. 2

1.0 Introduction Healthcare expenditures noticed a significant increase over the time period 1998-2001. On average inpatient health costs rose 5.9% per year during this time period, which was twice the average annual rate of inflation of 2.9%. The largest sector of healthcare expenditures is inpatient health costs. Inpatient health costs are those that are incurred while patients are under hospital care. There have been many studies that have analyzed different factors that could lead to the increases in inpatient health costs. There are many different factors such as local area wages, income per capita, and physician market characteristics that have an effect on inpatient costs. The availability of hospital caretakers also contributes to rising costs as well. (Hay, 2002) This paper takes the research of previous studies and applies the on a 50-State comparison in 2006. Many of the variables are indeterminanible or rather unavailble due to their nature. Such unavailable variables include Treatment Patterns and Technology, Provider Market Structure and individual Demographics. Technology in hospitals can be thought of as new methods for surgeries or new mechanical tools that increase physican efficiency. Since technology varies by state and changes at a rapid rate, a data source for this information is unavaible and highly subjective. Health Insurance Products and Design also vary greatly by state and availability of data is scarce. Many different health insurance plans pay for different types of medical treatment for various diseases. Patients with diseases such as Cancer and Diabetes have intuitvely higher health costs, but certain health plans have different level deductibles on these diseases as well as many other popular ones. This paper provides an emprical assessment of the different factors that are believed to be responsible for abnormal growth in inpatient healthcare costs.

2.0 Trends The table below, Table 1, denotes the growth in different expenditure sections during the 19982001 time period. This table shows that inpatient health costs rose an average of 5.88% per year,

Empirical Economic Bulletin, Spring 2009, Vol. 2

which is the smallest growth compared to that of prescriptions and outpatient services. The research of Joel Hay has shown that inpatient expenditures are the number one component of health expenditures. Table 1: Expenditure Per Member Growth Rates (%) 1998-2001 Expenditure

Minimum

Maximum

Mean

Category Inpatient

Standard Deviation

.23

12.9

5.88

2.73

Prescriptions

7.3

14.8

11.08

1.28

Outpatient

7.18

23.14

14.96

3.53

Services

Services n = 51 for each category. Total expenditures per member per day include all members, including those who were not hospitalized. The average annual consumer price index inflation rate during this period was 2.9% (http://www.bls.gov). SD indicates standard deviation.

(Hay, 2002) The figure below, Figure 1, depicts the growth of physician expenditures since the early 1980’s. Physician expenditures are positively correlated with inpatient health costs and have significantly contributed to the overall growth in inpatient health costs during this time period. (The Lewin Group, 2002)

Empirical Economic Bulletin, Spring 2009, Vol. 2

(The Lewin Group, 2002) Table 2 is consists of high technology drug categories that were selected to explain increases in technology of popular drug categories. The increases in technology of each respective drug category is positively correlated with inpatient health expenditures. It requires large amounts of capital for research in development to develop new technologies for these drugs, and that is reflected by the increase in inpatient healthcare expenditures.

Empirical Economic Bulletin, Spring 2009, Vol. 2

Table 2: New Patient Growth Rates for High-Tech Drug Categories 1998-2001

(Hay, 2002) 3.0 Literature Review Hospital inpatient expenditures over the time period of 1998-2001 accounted for over 34.2% of total health care expenditures, making it the highest contributor of health care expenditures. In 2001, per capita inpatient hospital spending increased by over 7%, nearly three times higher than the previous year. Also in 2002, a 6 month study has shown that the growth in total health care expenditures has slowed down from 10% growth to about 8.8%. Even though there is a reduction in growth, it is still significant in this time frame, seeing as how in the 1994-1998 era there were reductions in total health care spending as high as 5.3%. Many different variables have been factored into regression models in order to explain what drives health expenditures. Such variables include: population growth, aging population, disease, trends in private and public health care coverage, percentage of the population uncovered by health insurance, hospital business issues, new technology, labor costs, legislation, geographic variation, and many more. These variables all contribute to total health expenditures, but the task of explaining the significant increase in costs over the 1998-2001 time period still lies at hand. A list of what is believed to be the most significant factors contributing to this increase is as follows; workforce

Empirical Economic Bulletin, Spring 2009, Vol. 2

shortages and costs, new technology costs (including drugs), retreat from tightly managed care, legislation changes to private and public health spending, and shifts in hospital business directions. (Forest, Goetghebeur, & Hay, 2002) Nurse labor shortages have been thought to contribute a significant amount to inpatient health costs. For every 1% gap between the supply and demand of nurses, in this case shortages, a .5% to 1% increase in hospital inpatient expenditures per capita was noticed. Nurses constitute 44% of total inpatient health costs, so clearly the availability of nurses in the labor pool is significant to this study. A double-log univariate regression was used in 2000 to show that for each 1% shortage of nurses in the labor forced constituted a .96% increase in inpatient expenditures. Over the time period of 2000-2005 the Health Resources and Services Administration (HRSA) has determined that the nurse shortage will increase by 40%, which will lead to further increased inpatient expenditures. Per capita disposable income is also positively correlated with inpatient daily expenditures. For each 1% increase in per capita disposable income, an increase as much as 2% in inpatient expenditures is noticed. This is also the same for hospital and physician office wage levels, meaning that for each 1% increase in this category, an increase as high as 2% in inpatient expenditures is noticed. (Hay, 2002) The relative importance of cost driver categories can be split into nine different categories. These categories are in order from least to most important are: health care regulation, health status, provider operating costs, physician supply, treatment patterns and technology, provider market structure, general price inflation, demographic and economic conditions. These findings are consistent with the above research. However this research does not account for the relevance of the nursing shortage and how it contributes to health costs. However this research demonstrates how physician supply and costs are more significant to inpatient health costs as compared to those of the nurses. This research has different perspectives on each category. Health status, healthcare regulation, and health insurance product does not appear as significant as expected. (The Lewin Group, 2002)

Empirical Economic Bulletin, Spring 2009, Vol. 2

4.0 Data and Empirical Methodology 4.1 Data The data that was used in this research came from a variety of sources, both independently obtained and acquired from previous research. This paper is a cross-sectional study of the 50-US States including the District of Columbia. Data acquired regarding per capital income for each individual state was acquired from the Bureau of Economic Analysis (BEA). The US Department of Commerce news release provided per capita income for 2006-2008. The Kaiser website was used for obtaining the bulk of the data used for this research. Kaiser State Health Facts provided information for the following; inpatient daily expenses, birth rate per 1000, total health spending, percentage of health spending on subdivisions of physician services, drugs and other medical nondurables, nursing home care, dental services, home health care, medical durables, and other personal care, percentage of population in different age categories, as well as percentages of the population with different types of healthcare. 4.2 Methodology In this study an ordinary linear regression model (OLS) was used in order to determine which variables chosen contribute significantly to the daily inpatient health expenditures per state. Multiple regressions were used in this study with different variables added and deleted from each respective model. The primary model uses LOG DAILY INPATIENT EXPENDITURES (LDIE) as the dependent variable. The independent variables in this model are; LOG BIRTH PER 1000 (LBPT), LOG PER CAPITA INCOME (LPCI), TOTAL HEALTH SPENDING (THS), HOSPITAL CARE (HC), PHYSICIAN AND OTHER PROFESSIONAL SERVICES (PPS), DRUGS AND OTHER MEDICAL NONDURABLES (DOMN), NURSING HOME CARE (NHC), DENTAL SERVICES (DS), HOME HEALTH CARE (HHC), MEDICAL DURABLES (MD), OTHER PERSONAL HEALTH CARE (OPHC), CHILDREN 18 AND UNDER (CHILDREN), ADULTS 19-64 (AD19-64), ADULTS 65-74 (AD65-74), ADULTS 75+ (AD75+), EMPLOYER HEALTH INSURANCE (EHI), INDIVIDUAL HEALTH INSURANCE (IHI), MEDICAID (MEDICAID), MEDICARE (MEDICARE), OTHER PUBLIC HEALTH INSURANCE (OPHI), and UNINSURED (UNINSURED).

Empirical Economic Bulletin, Spring 2009, Vol. 2

The original regression equation has the following form: LDIE= - β0(LBPT) + β1(LPCI) - β2THS + β3(HC) + β4(PPS) + β5(DOMN) + β6(NHC) + β7(DS) + β8(HHC) - β9(MD) + β10(OPHC) - β11(CHILDREN) - β12(AD19-64) - β13(AD65-74) β14(AD75+) + β15(IHI) - β16(MEDICAID) + β17(MEDICARE) - β18(OPHI) - β19(UNINSURED) ε The second regression equation eliminated variables in order to obtain a more effective model. All age groups except Adults over 65 were eliminated from the equation. The log of birth per 1000 , total health spending, dental services, home health care, and other personal health care are eliminated as it is believed they are not significant. The new regression equation is: LDIE= β0(LPCI) - β1(HC) - β2(PPS) - β3(DOMN) - β4(NHC) – β5(MD) – β6(AD65-74) – β7(AD75+) + β8(IHI) – β9(MEDICAID) + β10(MEDICARE) - β11(OPHI) - β12(UNINSURED) + ε The third regression equation eliminated per capita income and medical durables to obtain a more effective equation resulting in: LDIE= - β0(HC) – β1(PPS) – β2(DOMN) – β3(NHC) – β4(AD65-74) – β5(AD75+) + β6(IHI) – β7(MEDICAID) + β8(MEDICARE) – β9(OPHI) - β10(UNINSURED) + ε These regression equations were broken down further into the following three regression equations.

Empirical Economic Bulletin, Spring 2009, Vol. 2

Regression 1 The regression below represents the core regression. Daily expenditures, the dependent variable, is determined from the four variables below. COEFFICIENT

P-VALUE

ADULTS 75+ -3.253679**

0.0195

0.484132

0.4444

0.766672***

0.0004

0.663805

0.1531

DOMN LPCI PPS *,**,*** Represents significance at 5% 2% 1% Levels, respectively Regression 2 This regression has added in the age 65-74 age group as well as the population enrolled in Medicare. Many coefficients came out as expected, however only two variables were significant. COEFFICIENT

P-VALUE

ADULTS 65-74 -3.886553

0.0350

-1.859417

0.2192

0.654370***

0.0016

0.353678

0.4336

0.687694

0.3041

2.81E-08

0.1334

ADULTS 75+ LPCI PPS DOMN MEDICARE *,**,*** Represents significance at 5% 2% 1% Levels, respectively

Empirical Economic Bulletin, Spring 2009, Vol. 2

Regression 3 This regression includes the entire adult age group, as well as the percentage of population enrolled in Medicaid and those Uninusured as well. COEFFICIENT

P-VALUE

ADULTS 75+ -1.181944

0.4596

-3.812058*

0.0461

0.680594

0.5047

0.525503

0.2896

1.004489

0.1471

0.609721***

0.0092

-3.54E-09

0.8937

2.06E-08

0.5317

ADULTS 65-74 ADULTS 19-64 PPS DOMN LPCI UNINSURED MEDICAID *,**,*** Represents significance at 5% 2% 1% Levels, respectively

5.0 Empirical Results Abbreviation

Description

Expected Sign

Hospital Care

PPS

Physician and other Professional

Services

Empirical Economic Bulletin, Spring 2009, Vol. 2

DOMN

Drugs and Other Medial Non‐

Durables NHC

Nursing Home Care

‐

AD65‐74

Adults 65‐74

AD75+

Adults 75+

LPCI

Log Per Capita Income

EHI

Employer Health Insurance

‐

IHI

Individual Health Insurance

‐

MEDICAID

Medicaid

‐

MEDICARE

Medicare

OPHI

Other Public Health Insurance

‐

UNINSURED

Uninsured

Hospital Care represents the proportion of each state’s health expenditures spent in hospitals. The expected sign is positive because research has shown that daily hospital costs increase if a state spends a high amount in that sector. Physician and other Professional services represent the proportion of health expenditures spent towards physicians and other professionals. The expected sign is positive, because research has shown that Physician wages are going up significantly during this time period, and an increase in their wages represents a higher cost to patients. Drugs and Other Medical Non‐Durables is the total amount spent on drugs. Research has shown that technological advances in drugs significantly increase the cost of these drugs, resulting in higher patient costs. Nursing home care is the amount spent by each state in nursing homes. There is little research on this topic, but the expected sign is negative. The reason for this is intuitive, if senior citizens are in nursing homes which have nurses and

Empirical Economic Bulletin, Spring 2009, Vol. 2

other health professionals on duty, and then their overall inpatient hospital costs should go down as they are not at the hospital. The expected sign of the population age groups of 65‐74 and 75+ are both positive. Research has shown that the older population, those over 75, is growing more rapidly than any other age group. As people become older they are more vulnerable to disease and thus need more hospital care, which is why inpatient daily expenses are expected to increase. The log of per capita income was taken and its expected sign is positive. Before the 1998‐2001 study, it was believed that as income went up, inpatient health costs would go down, however research has shown that these two variables are positively correlated. The different health insurances represent the number of persons in each state that have each different type of health insurance. Different health insurance plans have many different deductibles for a variety of patient needs. For EHI, IHI, and OPHI the expected sign is negative. Inpatient health costs should decrease if patients have health insurance, not just due to deductibles. For Medicare and Medicaid the expected signs are respectively negative and positive. The reason for this is that Medicare helps senior citizens buy prescription drugs as well as provide them with a certain criteria of health insurance. The reason the expected sign for Medicaid is positive is because if the population of a particular state tends to favor those with lower income, then their health status may be low as well. Those with higher income can afford better health care and thus would have lower health costs, as opposed to the poorer individuals who might tend to have higher inpatient health costs. The uninsured respresents the number of persons in the population without health insurance coverage. The sign is negative seeing as how those who do not have health insurance will incur more inpatient costs as they have no deductible.

6.0 LIMITATIONS This paper attempted to find a link between hospital costs and different types of insurances. There are many different types of variables that were not available to the author of this paper, such as hospital market structures, physician demands, and supply of labor for hospital care. These variables as well as many others have a significant impact on the empirical research. However due to the availability of this information, this paper took into account only the variables that were attainable through various research methods.

Empirical Economic Bulletin, Spring 2009, Vol. 2

6.1 POLICY RECOMMENDATIONS The independent research of this paper has suggested that there is a positive correlation between Medicare and inpatient health costs. This correlation infers that perhaps a more detailed healthcare plan should be in place for senior citizens in the recent future. With healthcare expenditures and technologically increasing, the aging population might not be able to afford these increase costs and thus would need a more efficient health plan. If there were a more positive correlation between those Uninsured and the number of citizens enrolled in Medicaid, then recommendations could be properly issued. However the outcome of this paper did not prove a positive correlation. If in the future this expected positive correlation is found, state regulators could pass different legislation and such that would accommodate the poor population in such hard economic times.

Bibliography Analysis, B. o. (2009). State Personal Income 2008. US Department of Commerce News Release , 1‐14. Forest, S., Goetghebeur, M., & Hay, J. (2002). Forces Influencing Inpatient Hospital Costs in the United States. BioMedCom Consultants Inc. Hay, J. W. (2002). Hospital Cost Drivers: An Evaluation of 1998‐2001 State‐Level Data. The American Journal of Managed Care , Volume 9. Kaiser. (n.d.). 50 State Comparisons. Retrieved April 1, 2009, from Kaiser State Health Facts: http://www.statehealthfacts.org/index.jsp The Lewin Group, I. (2002). Drivers of Healthcare Costs Associated with Physician Servies. The Lewin Group, Inc.

Empirical Economic Bulletin, Spring 2009, Vol. 2

THE ECONOMIC BENEFITS OF EDUCATION AS A RETURN TO GDP PER CAPITA Jonathan Brown Abstract: This paper examines many different factors of education, including the levels of education received, the expenditures per student as well as for each level of education, and the measurement of unemployed with said levels of education as to how it affects the levels of GDP per capita. What is consistent across each regression is that in fact, the average years of education received by the population will most closely have a beneficial effect on the levels of GDP per capita. What these regressions also show are tendencies to look more towards the future rather than the past. When considering unemployment, it didn’t matter much of how much was being spent on education but rather simply, what the literacy rates were for the population.

JEL classification: Keywords: Foreign direct investment, Capital taxation

Department of Economics, Bryant University, 1150 Douglas Pike, Smithfield, RI 02917.

Empirical Economic Bulletin, Spring 2009, Vol. 2

1.0 Introduction It has been heavily debated by many economists that education will only raise the levels of output per worker in the short run, but in the long run, will have little to no effect on the growth levels of that economy. This paper will examine random variables related to education, the amount of GDP invested in education, the labor force with certain levels of education attained and the amount of unemployed with certain levels of education attained. From this will be derived a regression analysis on how these variables affect the level of GDP per capita currently recorded from the year 2008. A total of 30 countries have been chosen at random including both developed and developing countries in order to figure out how these different economies respond to education levels through its GDP per capita. Using the World Development Indicators database, nine random variables were chosen as well as the average years of education for each country most current in 2008 in order to derive a regression equation that will indicate which variables have the greatest effect on GDP per capita, as well as which variables have a negative effect on GDP per capita. This paper specifically looks at data from the Twenty-First century as a means to better the immediate future of the economies of the selected countries. Data has been collected from the years of 1999 to 2005 to be analyzed as a time series affecting the current GDP per capita recorded from 2008. One of the strongest educational systems in the world can be found in the United States where they lead the world in average years of education with twelve years. What this means is the average adult over the age of fifteen has completed both primary and secondary levels of education, earning their high school diploma. The use of colleges and universities is also crucial in developing strong skills that can be later used in the labor force, making each student have a concentration in a specific subject or even two subjects when entering the job market. In the case

Empirical Economic Bulletin, Spring 2009, Vol. 2

for the United States, advancements in education are crucial in order to obtain a skilled job however there are variances among other countries that value education not quite as much. Sometimes the most effective way for a student to have a positive impact on the level of GDP per capita is to complete a basic level of education and then enter the labor force immediately. The average years of education will help to determine among these countries whom value education the highest and whom value immediate entry into the job market. As cases before have shown, key determinants of economic growth in the long run do not rely on education as it is only a short term means to increase the levels of income for that country. A key working paper that will be referenced throughout this paper is titled, “Literacy and Growth,” written by Serge Colombe and Jean-Francois Tremblay. The rest of the paper is organized as follows: Section 2 gives trends in both education and growth over the years. Section 3 gives a brief literature review. Data and estimation methodology are discussed in section 4. Finally, section 5 presents and discusses the empirical results. This is followed by a conclusion in section 6.

2.0 Trends in Education and Growth Education has been becoming more of a concern as every new day passes. As a child growing up in the United States, education is greatly stressed as the key to the future and the path that will help to develop careers later in life. In fact, education has become such an important tool of the individual that it is no longer acceptable in highly developed countries like the United States to merely obtain a high school diploma. With technology advancing to more and more intricacies, it is becoming almost a requirement for students to go on to a tertiary level of education to become masters of a certain area of study. However, this is not consistent across

Empirical Economic Bulletin, Spring 2009, Vol. 2

every country of the world, but it is becoming more and more evident that overall, education is being treated more seriously and a tool that could lead to lessening the gap between developed and developing countries.

Figure 1: Growth of per Capita GDP

Source: Education and Economic Growth (2007)

Figure 1 represents the growth of per capita GDP for nine major regions of the world over a fifty year period. With the exceptions of Russia and Africa, every region of the world has shown a long term growth in its GDP per capita.

Empirical Economic Bulletin, Spring 2009, Vol. 2

Figure 2: Net Enrollment in Primary Education

Source: Education and Economic Growth (2007)

Education in general is being understood as an important role in developing human capital and as shown above, the net enrollment in almost every region in primary education has seen an increase in numbers. There has been either little or no change in the US and Western Europe as they have been atop the highest percentage of enrolment for many years but an encouraging image is the other regions of the world that are increasing their percentages of enrolment. It is becoming ever more important with more and more technological advancements that these other regions of the world become better educated simply in order to understand what these new technologies accomplish as well as bringing some form of contribution to these advancements in technology within their own regions.

Empirical Economic Bulletin, Spring 2009, Vol. 2

Table A: Children Out of School

Source: Education and Economic Growth (2007)

Table A above represents the entire world of children who are not a part of the educational system for each year between 1999 and 2004. It has been reported that data from 2007 has shown that there are around 77 million children that are not enrolled in schools, varying little from the number shown for 2004. It also gives the numbers of children who are not in some form of primary school either. Although these numbers are high, the trend since the turn of the new millennium gives hope for the futures of the worlds’ economy. There has been a steady decline in the numbers reported showing that education is being identified as an important indicator for economic development across the entire world.

Empirical Economic Bulletin, Spring 2009, Vol. 2

Figure 3: Educational Attainment of Adult Population

Source: Education and Economic Growth (2007)

Figure 3 depicts the educational attainment of the adult population. It is taken from the year 2004 and includes all of the adult population with the average number of years those adults had spent in the educational system. Apart from what this paper has used as average number of years currently found for the countries used, we can see that the levels of education are actually a little bit higher than what was found. What this can tell a researcher is that it is including every level of education, for example the extra years spent in a college or university, or that for those other countries, either pre-school education is included in the statistics or that those countries have longer minimal requirements for children in the educational systems, reflected in the numbers for the entire adult population. The data that was found and used in this paper however has the United States ranked as the leader in average years spent receiving an education, which is 12; expressing that on average the entire population has received at least a high school diploma.

Empirical Economic Bulletin, Spring 2009, Vol. 2

Figure 4: Private internal rates of return for university level achievement

Source: Education and Economic Growth (2007)

Figure 4 represents the return that the populations of countries have from the advancement of their education into tertiary levels of education, or the advancement into a college or university. It is also broken down to specifically show the return that both males and females of the selected countries experience as a result of this investment.

3.0 Literature Review In the paper by Coulombe and Tremblay (2006), they use a time series from 1960-1995 measuring the literacy levels of the labor markets. It uses 14 OECD countries and considers literacy levels as an investment in education. The measure of human capital based on literacy scores tells us more for the relative growth of countries rather than using years of schooling.

Empirical Economic Bulletin, Spring 2009, Vol. 2

Overall, literacy scores have significant, positive effects on growth paths and long run levels of GDP per capita and labor productivity. One more year of schooling increase aggregate labor productivity by about 7%. Investment in human capital for women is actually more important than that for men. In turn, increasing the average tests scores becomes a more accurate measure of human capital than schooling data would because it is more comparable across countries. However at the same time it could be distorted by migration and the depreciation of human capital over time. One should be cautious about an open economy as convergence of human capital is the driving force behind the convergence of GDP per capita during the economy’s transition to the steady state. Imbalances of human and physical capital could also result. In the future, other analysis could include comparisons of the performance of human capital based on literacy tests with those based on schooling data for growth in sub-national economies. Islam, Wadud and Islam (2007) use a multivariate causality analysis on the relationship between education and growth, specifically for Bangladesh. This paper also includes both capital and labor as variables which also shows bidirectional causality between education and growth. This bidirectional causality can be defined in three categories including income driving education to grow, education causing income to grow or both education and income causing each other to grow simultaneously. It was originally thought that it would be one or the other and never determined that it could be possible to occur simultaneously in fact. In order to determine the trend, one should use an income equation and an education equation that are both statistically significant at the 1% level. As a result of the paper’s finding, if the analysis is confined to 19842003, then there is actually no evidence of a long term relationship between education and growth specifically for Bangladesh.

Empirical Economic Bulletin, Spring 2009, Vol. 2

Stevens and Weale (2003) provide data from the early 20th century that is more representative of thoughts today about the relationship between education and economic growth. It analyzes the role of education in facilitating the use of best-practice technology. Overall, the living standards have risen greatly since the 1800’s and can be linked to the advancement of education. Education is needed for people to benefit from scientific advancements as well as being able to provide contribution to that advancement. Levels of income do in fact depend on the levels of education and education should be looked at as an investment decision into human capital. In the end, there is no conclusive evidence that returns to education will vary more than 6%-12% based on previous studies. There is evidence however that education is needed as a means to make good use of available technology respecting that returns to education diminish with levels of development. Miller (2007) says that overall, schooling is necessary for industrial development. Schooling from the 19th century generates cognitive, behavioral and social knowledge which in turn causes organization. Schooling is necessary but it is not the driving factor behind industrial development however individuals and societies clearly gain from investments made into schooling. The specific form of education system is an indispensable component of an industrial growth society. Making investments in all elements of the schooling system and making people attend those schools is necessary but not a sufficient condition for expanding the GDP. The participation in education has steadily increased in 2007, however there are still 1 in 5 adults in the world’s population that do not have minimum literacy skills as well as 77 million children who are not enrolled in the schooling system. It has been proven that each additional year of schooling will raise the income of that individual about 10% here in the United States. In an OECD area, the long term effects of one more year of schooling on the output is between 3% and

Empirical Economic Bulletin, Spring 2009, Vol. 2

6%. Future predictions make it possible for the relationship between what people know and the wealth of a society to become stronger and clearer in the near future.

4.0 DATA and Empirical Methodology 4.1 Definition of Variables The basic model used in this model uses GDP per capita as the dependant variable and bases findings from the literature review to determine other possible influential variables that will have a direct return on that GDP per capita. The model chosen is as follows:

GDP (PPP) = β1 + β2(AVGED) + β3(LIT) + β4(LFPRI) + β5(LFSEC) + β6(LFTER) + β7(PUB) + β8(UNPRI) + β9(UNSEC) + β10(UNTER) + β11(EXPPRI) + β12(EXPSEC) + β13(EXPTER) + ε

GDP (PPP) is the GDP per capita for the year 2008 and all variables are part of a time series of data collected from the year 1999-2005 from 30 various countries around the world. Independent variables consist of twelve variables obtained from various sources. Appendix A provides the data source, acronyms and descriptions of selected variables. First, AVGED represents average years of education received for selected countries. Second, LIT represents literacy rate as a percent of total population. Third, LFPRI represents % of the total labor force with a primary education. Fourth, LFSEC represents % of the total labor force with a secondary education. Fifth, LFTER represents % of the total labor force with a tertiary education. Sixth, PUB represents public spending on education as a total (% of GDP). Seventh, UNPRI represents the unemployed with primary education (% of total unemployment). Eighth, UNSEC

Empirical Economic Bulletin, Spring 2009, Vol. 2

represents the unemployed with secondary education (% of total unemployment). Ninth, UNTER represents the unemployed with tertiary education (% of total unemployment). Tenth, EXPPRI represents the expenditure per student for primary education (% of GDP per capita). Eleventh, EXPSEC represents the expenditure per student for secondary education (% of GDP per capita). Lastly, EXPTER represents the expenditure per student for tertiary education (% of GDP per capita).

4.2 Data This study uses results collected yearly from the period of 1999 to 2005 for thirty different countries around the world. Data was primarily obtained from the WDI Online database as well as a few other worldwide data sources. Summary statistics for the data are provided in Table 1.

Table 1: Summary Statistics Std. Variable

Obs.

Mean

Dev.

Min

Max

GDP (PPP)

30 28653.33 12940.60 4400.00 55600.00

AVGED

8.73

1.96

4.90

12.00

LIT

96.44

4.63

82.40

99.70

LFPRI

29.83

16.23

3.43

69.17

LFSEC

41.26

15.44

12.50

73.43

LFTER

24.60

10.16

7.00

49.00

PUB

5.38

1.15

3.83

8.00

Empirical Economic Bulletin, Spring 2009, Vol. 2

UNPRI

40.27

15.52

1.17

72.00

UNSEC

40.97

12.52

14.43

65.29

UNTER

15.99

9.56

2.80

44.57

EXPPRI

18.19

4.71

11.00

27.60

EXPSEC

22.68

6.18

10.20

36.00

EXPTER

35.57

12.49

14.50

67.00

5.0 Empirical Results The primary objective of this particular study was to find out which specific variables related to education will have a positive return to the levels of GDP per capita. Initial tests showed many variables to be skewed or highly correlated and therefore many regressions had to be performed. In general, it would be expected to find that both the literacy rate and the average years of education would yield a positive return to GDP per capita. Those who are unemployed would be expected to hurt GDP per capita and have a negative coefficient and expenditures the government makes on education would also be a negative factor of GDP per capita. We would also expect to find that those in the labor force would have a positive return and the higher the level of education, the more positive the coefficient will be.

Empirical Economic Bulletin, Spring 2009, Vol. 2

Table 2: Regression using Education Coefficient Std. Error t‐Statistic Prob. AVG_EDU 2962.459 1166.44 2.539744 0.0187 EXPSEC 1556.34 277.3267 5.611936 0.0000 EXPTER 293.1553 140.0034 2.093916 0.048 LFTER 995.1515 340.9446 2.918807 0.008 PUB ‐5670.973 1874.559 ‐3.025231 0.0062 UNSEC ‐175.7432 133.6736 ‐1.314719 0.2021 UNTER ‐631.0387 276.4473 ‐2.282673 0.0325 C ‐19605.61 7032.182 ‐2.787983 0.0107 R‐Squared 0.809161 Adj R‐Sq 0.748439

This regression proved to be the most statistically significant out of any test that was run and also reflected the highest R2 value of 0.809161. In this test however, there was one statistically insignificant variable, UNSEC or unemployed with a secondary education. Every other variable was statistically significant at the 95% confidence interval.

Table 3: Regression using Literacy

LIT EXPPRI EXPSEC EXPTER LFPRI LFSEC LFTER C R‐Squared Adj R‐Sq

Coefficient Std. Error t‐Statistic Prob. 1784.662 650.316 2.744301 0.0118 ‐268.95 510.7842 ‐0.526543 0.6038 499.861 459.5825 1.087642 0.2885 302.0047 167.5816 1.802136 0.0852 117.0582 163.4525 0.716161 0.4814 11.80037 164.8751 0.071572 0.9436 480.4644 192.1392 2.500606 0.0203 ‐176446.4 61250.62 ‐2.880728 0.0087 0.760721 0.684586

In Table 3, instead of using the average years of education, the literacy rate was used. Also, every variable related to expenditure on grade level and the labor force level was used.

Empirical Economic Bulletin, Spring 2009, Vol. 2

Public expenditure on education was also omitted. What was found to be odd was that the EXPPRI had a negative coefficient whereas EXPSEC and EXPTER both had positive coefficients but only the EXPTER was determined to be statistically significant. This would appear to be odd because one would think that the more expenditure per student, the less the GDP per capita would be for a country. Another regression that was run used average education, as well as the public expenditure per student. Instead, this regression looked at the relationship that unemployment plays and at what levels of education those people are at. However, the regression proved to be statistically insignificant. In order to fully test the insignificance, instead of using the average years of education, the literacy rate was used. Results are as follows:

Table 4: Unemployment, Literacy and Public Expenditure Coefficient Std. Error t‐Statistic Prob. LIT 2043.064 412.1833 4.956688 0.0000 PUB 2550.001 1507.396 1.69166 0.1037 UNPRI 219.8154 156.4291 1.405208 0.1728 UNSEC 98.74788 184.0801 0.53644 0.5966 UNTER 336.964 189.4369 1.778766 0.0879 C ‐200394.3 42710.79 ‐4.69189 0.0001 R‐Squared 0.686657 Adj R‐Sq 0.621377

In Table 4, it is proven that the literacy rate is statistically significant when also considering the unemployed levels of education and the public expenditure on education. It was also very odd to find that all three groups of the unemployed would result in a positive coefficient when it was assumed that the more people who are unemployed, the less is being

Empirical Economic Bulletin, Spring 2009, Vol. 2

contributed to the GDP per capita. In this regression, there could be a possibility for skewed data as the public expenditures variable and the set of unemployment variables all have positive effects of the GDP per capita. One would assume that these would lower the levels of GDP per capita which lead to the belief that in some way there are highly correlated variables. Also, use of the literacy variable might also lead to varying results as average years of education were used for the first regressions. After observing the results of several regressions, there has been evidence that the overall most important variables are those that involve the tertiary level of education. These variables turned out to continually have the greatest effect, whether negative of positive, on the levels of GDP per capita. What one can assume from these findings are that as the world continues to advance, it will and has been more important to obtain not only some level of education, but some level of tertiary education in order to more effectively contribute to the levels of GDP per capita. This goes along with the fact that more education is needed in order to aid in the advancement of technology which in turn aids the advancement of long term positive economic growth. Out of some possible policy implications, the strongest would most likely be to invest more into the tertiary levels of education and to try and limit the amount of unemployed who have achieved the tertiary levels of education. Both the primary and secondary levels of education had showed positive returns however they were not as impactful as that for tertiary levels. It may also mean that overall, the current levels of both expenditures and unemployed are at sufficient levels that are not drastically affecting the levels of GDP per capita either negatively or positively.

Empirical Economic Bulletin, Spring 2009, Vol. 2

Other possible policy implications are that countries begin to invest more into the quality of education and continue to raise the importance of obtaining an education. The country itself must understand how important schooling is and allow for the development of those students to create a stronger, more productive economic entity. Although not always the most positive, it has been proven that primary and secondary levels of education are important for GDP per capita, but it seems as if what countries are doing right now is on the correct path towards economic growth.

6.0 Conclusion

This paper examines many different factors of education, including the levels of education received, the expenditures per student as well as for each level of education, and the measurement of unemployed with said levels of education as to how it affects the levels of GDP per capita. What is consistent across each regression is that in fact, the average years of education received by the population will most closely have a beneficial effect on the levels of GDP per capita. What these regressions also show are tendencies to look more towards the future rather than the past. When considering unemployment, it didn’t matter much of how much was being spent on education but rather simply, what the literacy rates were for the population. As holds true with the working papers discussed in the literature review, education must be looked at as an investment into the future and as time moves on, it is becoming more and more critical to obtain some level of tertiary education rather than just obtaining a basic level or even the completion of high school. Trends have proven that there is more awareness of the importance of education for

Empirical Economic Bulletin, Spring 2009, Vol. 2

long term development primarily because as the developed countries become more technologically advanced, in order for the developing countries to try and keep up they must teach their youths how to use that technology to their own economic benefit. What can be done to improve upon future analysis is to analyze more in depth the variable of education and how exactly it contributed to society. How it is felt as a return into both human and physical capital can both greatly affect how the levels of education can affect the levels of GDP per capita. Also, there could be problems affecting the results by using the average years of education for some of the regressions and using the literacy rate for the others. There could be elasticity issues in how incremental changes are experienced for each variable. This could help to create more significant influences from the variables on the levels of GDP per capita. Other studies should also look into how education is considered an investment in human capital for the future and how it helps to stimulate innovation. If it is true that tertiary levels of education have been the most significant, then how much is this advanced level of education an important factor as we live our lives during the new technology age. In any case, it is important to realize what will help to positively affect the state of the economy both in the long run and in the short run as well.

Appendix A: Variable Description and Data Source Economic Variable

Description

Empirical Economic Bulletin, Spring 2009, Vol. 2

Source

GDP (PPP) AVGED LIT LFPRI

LFSEC

The 2008 World Factbook

GDP per capita (2008) Average years of education received Literacy rate as a percent

Nation Master Online United Nations Development Report

% of total labor force with primary education % of total labor force with secondary education

WDI ONLINE

PUB

% of total labor force with tertiary education Public spending on education, total (% of GDP)

UNPRI

Unemployed with primary education (% of total unemployment)

WDI ONLINE

UNSEC

Unemployed with secondary education (% of total unemployment)

WDI ONLINE

UNTER

Unemployed with tertiary education (% of total unemployment)

WDI ONLINE

EXPPRI

Expenditure per student for primary education (% of GDP per capita)

WDI ONLINE

EXPSEC

Expenditure per student for secondary education (% of GDP per capita)

WDI ONLINE

EXPTER

Expenditure per student for tertiary education (% of GDP per capita)

WDI ONLINE

LFTER

WDI ONLINE

Bibliography •

Education and Economic Growth: from the 19th to the 21st century, Miller, (2007)

Empirical Economic Bulletin, Spring 2009, Vol. 2

100

http://www.cisco.com/web/strategy/docs/education/Education-and-Economic-Growth.pdf •

Education and Economic Growth, Stevens and Weale, (2003)

http://www.niesr.ac.uk/pubs/dps/dp221.pdf •

Literacy and Growth, Coulombe and Tremblay, (2006)

http://0-web.ebscohost.com.helin.uri.edu/ehost/pdf?vid=5&hid=21&sid=1fad8aae-8700-4e58b851-454070c9c812%40sessionmgr2 •

Relationship between Education and GDP growth: a multivariate causality analysis for Bangladesh, Islam, Wadud and Islam, (2007)

http://economicsbulletin.vanderbilt.edu/2007/volume3/EB-07C30001A.pdf •

http://0-ddpext.worldbank.org.helin.uri.edu/ext/DDPQQ/member.do?method=getMembers&userid=1 &queryId=6

•

http://www.photius.com/rankings/economy/gdp_per_capita_2008_0.html

•

http://www.nationmaster.com/graph/edu_ave_yea_of_sch_of_adu-education-averageyears-schooling-adults

Empirical Economic Bulletin, Spring 2009, Vol. 2

101

Are Educated Women Less Likely to Get Married? Qian Jiang1

Abstract: In the past, there are some research indicate that highly educated women marry less. Because women tend to face the success penalty, however, a new analysis of U.S. census data indicates that--despite cultural messages to the contrary--the success gap, in which better educated women marry less, is actually shrinking. Using 2008 Current Population Survey, this paper utilized Probit regression to analyze how increase of women's educational attainment can influenced marriage and other aspects such as financial well-being. In addition, this study tries to estimate the best levels of education for women to be more likely to married. JEL Classification: J1, I21

Keywords: Marriage; Women; Education

Department of Economics, Bryant University,1150 Douglas Pike, Smithfield RI02917

Email:qjiang@bryant.edu.

Empirical Economic Bulletin, Spring 2009, Vol. 2

102

Introduction Today, more people realize the importance of good education. Over the years, social

views have changed; the role of women have transformed in the labor market as well as in the family. Nowadays, women are encouraged to attain higher levels of education and enjoy a successful career instead of marrying young and becoming housewives. Evidence suggests that, up to a point, an additional year of schooling is likely to raise an individual’s earnings about 10 percent (Krueger, 2005). However, several studies show that women face a conflict between their roles in the two worlds. Many research suggested that there is the “success penalty” or the disadvantage to women in the marriage market (Rose, 2003). While male are more likely to find a wife or have a family if he is successful, women seems to be the opposite. According to Matsui (2004), highly educated women, working full time/ part time or students, and living in a larger city tend to delay marriage. Also, Rose’s (2003) study pointed out several source of the penalty such as the “female hypergamy” and other psychological behaviors of male chauvinistic. The aim of this paper is to examine whether higher education level would disadvantage women in marriage. If education does disadvantage women in marriage, would different racial background disadvantage women at a different level?. This study using econometric examination predict the “magic number” years of education that is best for marriage. The structure of this study is as follows. Section two introduces marriage trends in the US. Section three provides literature reviews that are related to this topic. In section four data and empirical methodology used in this study are presented. Section five includes empirical results, followed by conclusion.

Empirical Economic Bulletin, Spring 2009, Vol. 2

103

II.

Trend The population in the United States is becoming more educated, but significant

differences in educational attainment lie within age, sex, race, and origin. In 2003, over fourfifths (85%) of all adults 25 years or older reported they had completed at least high school; over one in four adults (27%) had attained at least a bachelor’s degree. Study done by the census bureau also shows that the younger population is more educated than the older population (Census, 2007)

Figure 1: Education Attainment of the Population 1947-2003

Sources: U.S. Census Bureau. Current Population Suvry

Figure 2 shows that since 1991, the proportion of young women enrolled in college has exceeded the enrollment rate for young men, and the gap has widened over time. In 2005, about 43 percent of women ages 18 to 24 were enrolled in college, compared to 35 percent of young men. This represents a major shift in the gender balance at U.S. colleges and universities. In 2005, women make up the majority (54 percent) of the 10.8 million young adults enrolled in

Empirical Economic Bulletin, Spring 2009, Vol. 2

104

college. Several reasons have been cited for this crossover: gender differences in academic achievement (girls generally do better in high school than boys), changes in societal values, and a shift in women’s expectations for higher education and future employment.

Figure 2: Proportion of 18 to 24 year old Men and Women Enrolled in College, 1967 -2005

Source: U.S. Census Bureau Colleges need to balance women’s advantage in enrollment rates against their disadvantage in the post-college labor force. Women’s earnings, relative to those of men, have not kept up with their gains in educational attainment. In 2005, the median weekly earnings for women working full-time were $585, compared with $722 for men. The tradeoff that women make between education and marriage seems to be going away. In 1980, a woman with three years of graduate school was 13 percent less likely to be married than a woman with only a highschool diploma. By 2000, that gap shrank to less than 5 percent (Rose, 2003).

Besides deferring marriage age, we also see a trend in declining marriage rate. Figure 3 shows the median age at marriage, for men and women from 1890 to 2002. In 1890 the median age was relatively high, about twenty-six for men and twenty two for women. During the first

Empirical Economic Bulletin, Spring 2009, Vol. 2

105

half of the twentieth century, the typical age at marriage dropped. By the 1950s, it had reached historic lows, roughly twenty three for men and twenty for women. Women today are marrying substantially later than they ever have.

Figure 3: Median Age at Marriage, 1890-2002

Source: U.S. Bureau of the Census About half of young adults live with a partner before marriage. Cohabitation is far more common today than it was in the early or mid-twentieth century. Cohabitation today is a diverse, evolving phenomenon. For some people, it is a prelude to marriage or a trial marriage. For others, a series of cohabiting relationships may be a long-term substitute for marriage which would in terms decrease the marriage rate (Cherlin, 2005). In addition to change in marriage age, we also see a trend in changes in demographic. Figure 4 shows that since 1997, U.S. marriage rate has been declining while the population base increased.

Empirical Economic Bulletin, Spring 2009, Vol. 2

106

Figure 4: U.S. Population and Marriage Rates 1997 to 2006

Source: U.S. Bureau of the Census III.

Literature Review There are many reasons to believe that education can improve women’s well being, even

for women with very limited labor force participation. In particular, evidence suggests that education reduces mortality (Lleras-Muney, 2002), increases the cognitive ability of women’s children (Murnane, 1981), reduces the incidence of criminal activity (Lochner and Moretti, 2001), aids in overcoming addiction (Sander, 1995), and improves the health of women’s children (Thomas, Strauss, and Henriques, 1991 and Currie and Moretti, 2002). Goldin (1992) presents evidence that attending college may have improved the marriage outcomes of women who attended school in the 1960’s and 1970’s. Rose (2003) discussed the source of women’s success penalty, i.e. female hypergamy, where women tend to marry up, thus lack of suitable husbands for women in a higher social class. On the contrary, men tends to “go down a step to take a wife” because “a woman from a more distinguished family than her husband may consider herself superior and act haughtily toward him”. Hypergamy with respect to education can lead to a success penalty as it tends to disadvantage women at the top of the distribution. Similar to this idea was Becker’s positive

Empirical Economic Bulletin, Spring 2009, Vol. 2

107

assortative mating theory. He suggests that people with similar traits tend to marry. Which means if one is a college graduate, his/her spouse would be more likely to be a college graduate as well. Juhn and Murphy (1997) revealed that men with high wages tend to marry women with favorable labor market prospects, because marrying a college educated spouse is associated with significantly higher family income. Also, women who attended college were much more likely to marry college educated husbands. According to Rose (2003), there is indeed a tradeoff between motherhood and marriage for women with more than a college degree. Around 81.5 percent of women with 16 years of education were mothers at age 40-44, while only 63.4 percent of women with a professional degree or doctorate had children. The difference in black and white marriage rates lies primarily at the lower end of the education distribution. Lefgren and McIntyre (2006) found that beyond high school, education is associated with reductions in the probability of ever having been married. Beyond college completion, additional education is associated with fewer and less stable marriages.

IV.

Data and Methodology This study used data from the 2008 Current Population Survey. The data set contains

information on marital status, race, education attainment, metro level, as well as total personal income. Using 29,904 sample size, this research explore the relationship between education attainment and marriage outcomes for women between the ages of 35-45 who are currently residing in the U.S. Table 1 shows summary statistics. Nearly 87.85% (26,273 women) have been married at least once, only 12.14% (3631 women) have never been married. 77.48% of women are in the labor force and 22.52% of

Empirical Economic Bulletin, Spring 2009, Vol. 2

108

women are not. The average income for women in our sample is $23,479.83. Education in this study is reported in categories as opposed to years. For our analysis, we combine all individuals who complete less than or equal to 12th grade (receive no diploma). We also group together all individuals who are high school graduate or attended college but did not earn a bachelor or received associate degree. Finally, we combined masters and professional degrees. Table1: Summary Statistics for 2008 Current Population Survey Variable Age Income and Wages In Labor force Low education (less than12th grade) High school graduates College graduate Master and professional degrees Total observations

Mean 40 (3.14) 23,479.83 (32,132.28) 0.7748 (0.44) 0.076 (0.265) 0.512 (0.499) 0.053 (0.223) 0.358 (0.479) 29,904

Min 35

Max 45

688,117

Using the 2008 CPS data, we begin to analyze the relationship between education and marriage outcomes. As mentioned earlier, education is reported in categories as opposed to years. For this reason, we focus on the marginal change in marriage outcome associated with moving up to the next education category. Prior studies have largely utilized multiple regression, probit, or logit models to analyze statistical relations between marriage and other explanatory variables (Johns, Yang, & Chen, 2003). Because of the discrete nature of the dependent variable in this study, ordinary least squares regression would be an inappropriate model.

Empirical Economic Bulletin, Spring 2009, Vol. 2

109

The standard ordered probit model is widely used to analyze discrete data of this variety and is built around a latent regression of the following form: Ŷ = x’β + ε

(1)

where x and β are standard variable and parameter matrices, and ε is an error term. Using the ordered probit model, we included the flowing explanatory variables: metropolitan of residing city (Metroi), race (Racei), education level (Educi), and income level (Inci).

where

Y*I = β0 + β1Metroi + β2 Racei +β3 Educi + β4 Inci + ε

(2)

Y*I = unobserved marital status Yi = observed marital status Yi = 0 if Y* ≤ 0, indicating the woman is not currently married Yi = 1 if 0 ≤ Y* < µ1 , indicating the woman has at least been married once One possible difficulty in interpreting the results of parameters estimated using equation (2) involves the use of Metro and Inc as explanatory variables. Nevertheless, we find that Income is an important variable to examine because some studies have shown that income level has close correlation on marriage rate. From a practical standpoint, the primary rationale for placing the income level in the model is that woman with higher income tend to have higher position jobs which require a higher level of education, consequently defering the marriage age. Also, the opportunity cost for women in high level position is much higher than other women; therefore, they are less prompt to leave their high paid job to get married and start a family. V.

Empirical results Table 2 reports the empirical results of regression (2). The likelihood ratio chi-square of

1279.56 with a p-value of 0.000 shows that our model as a whole is statistically significant, as compared to model with no predictors. However, coefficient in a probit model does not have a

Empirical Economic Bulletin, Spring 2009, Vol. 2

110

significant meaning, thus we conducted the marginal effect test on regression (2). Marginal effect results are shown in table 3. Table 2: Regression results Variable

Coefficient

P >| z|

.0133361

Standard Error .0098953

Metro

1.35

0.178

Age Incwage White Asian Black MP CG HS cons

.0295089 -2.30e-06 .3655167 .2299523 -.4859689 .2484719 .293801 .2301687 -.4189307

.0030805 2.80e-07 .0583832 .0725908 .061616 .0373603 .0545458 .0353469 .1396479

9.58 -8.19 6.26 3.17 -7.89 6.65 5.39 6.51 -3.00

0.000 0.000 0.000 0.002 0.000 0.000 0.000 0.000 0.003

(*) LR chi 2(9) = 1279.56, prob>chi2 = 0.000

From table 3, we conclude that at 5% significance all variables are significant, except Metropolitan level (Metro). Age, white, Asian, master degree or professional degree (MP), college degree(CG), and high school education (HS) are positively related with marital status, while income level (Incwage) and black has negative relationship. Dy/dx represents the marginal effect of the regression. The probability of a women being married at age 35-45 would increase 0.25% if she move to a metropolitan area and 0.56% higher for every year she gets older. For a white female, chances of getting married at age 35-45 is 7.9% higher, and for an Asian female it is 3.8% higher.Probability of a female who holds a master or professional degree being married at least once at age 35-45 is 4.5% higher, 4.7% higher if she has a bachelor degree, and 4.3% higher if she is a high school graduate. For female ages between 35 and 45, the negative coefficient for income level and for black women suggest that a negative relationship between marriage

Empirical Economic Bulletin, Spring 2009, Vol. 2

111

outcomes. We can associate income variable with women’s opportunity cost to give up high paying jobs and start family. The negative relation for black female could be due to the increase of incarceration rate for black men and an increase in black men dating and marrying outside of their race (Ali, 2009). So it makes even harder for black female to find a husband. Table 3: Regression results for Marginal Effect Variables

Dy/dx

Standard Error

P > |z|

Metro Age Incwage White Asian Black Mp Cg hs

.0025258 .005589 -4.35e-07 .0794416 .0381904 -.1139563 .045128 .0471297 .0437882

.00187 .00058 .00000 .01433 .01043 .01723 .0065 .00726 .00675

0.178 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000

*** *** *** *** *** *** *** ***

Note: dy/dx is for discrete change of dummy variable from 0 to 1 *, **, and *** indicates significance at the 10%, 5%, and 1% level, respectively.

Table 4 shows the predicted probabilities when variables are set to specific values. As we see from the table, for a black woman with a master or professional degree, chances of being married at age 35 is 78%, while 82% chance of being married if she only has a college degree, which confirm the success penalty that Rose (2003) suggested in her study. A black woman would have a higher chance of being married at age 35 if she is a college graduate than if she holds a master degree. However, if the same black woman has only a high school degree, her chance of being married at age 35 decreases to 77.4%, which means a low education has negative effect on marriage outcome. If the black woman is 45, her chance of being married is higher than when she was 35, however the probability of being married does not vary much across different levels of education. This is not surprising because women would be more likely

Empirical Economic Bulletin, Spring 2009, Vol. 2

112

to be married at least once at age 45 than 35. Same trend for both white and Asian women, however with a lower degree. Table 4: Predicted Probabilities of Marriage and Education Master or Professional Degree Black White Asian College Graduate Black White asian High School Graduate Black White Asian

0.7881 0.9028 0.9267

0.8632 0.9444 0.9597

0.8209 0.9217 0.9419

0.8876 0.9565 0.9690

0.7743 0.8944 0.9199

0.8527 0.9389 0.9554

Table 5 illustrates women’s financial wellbeing and education in the United States. The incomes reported are the predicted incomes for women of the given education category. The change in incomes is the marginal increase associated with increasing one level of educational category. According to the empirical results, women who have never been married tend to make more money than married women in the job market given similar education background. Moreover, women who have never been married receive greater benefits than married women when educational attainment increases. For married women, on average, income increase by $10,3078 when education level increases from bachelor to master or professional degrees. On the other hand, income for women who never married increases by $19,207, nearly doubles, by going from bachelors to master or a higher degree.

Empirical Economic Bulletin, Spring 2009, Vol. 2

113

Table5: Women’s Financial Wellbeing and Education Married Women Income Loweduc HS CG MP

7942.083 17544.02 23342.11 33720.65

Change Married Women Income

Never Married Women Income

Change Never Married women Income

+9601.937 +5798.09 +10378.54

7156.493 19128.6 25519.24 44726.49

+11972.107 +6390.64 +19207.25

Table 6 summaries the probability of women married at age 35 and at age 45. We notice that with the same income, women who are older are more likely to be married or at least married once. For a women with master degree or higher, and at age 45 has less chances of being married. However, women with a college degree are more likely to be married at age 45, probably because most women in this sample at 45 are currently married or at least married once. Pursuing education takes time; therefore most women would have to defer their timing on marriage. Also, women with a low educational level (less than twelve years of education) are less likely to be married.

Table 6: Probability of Women Getting Married Same Income Level MP CG HS Loweduc

Empirical Economic Bulletin, Spring 2009, Vol. 2

Age 35 -1.95e-06 (1.05e-06)*** .1547215 (.0665018)* -.1347805 (.1323798) -.0986707 (.0618746)*

Over Age 45 -2.51e-06 (9.08e-07)*** .0922271 (.0747227)* .2649926 (.1653271) .0426073 (.068662)*

-.0289546 (.1149206)

-.5116825 (.1129493)

114

VI.

Conclusion Using the 2008 CPS data, this paper examines whether higher educational attainment

would disadvantage women in the marriage market. Due to different limitations, the paper was not able to conclude on the exact effects of education on marriage outcomes. The result is measure in the likelihood of being married once or never been married. The evidence in this paper suggests that education is correlated with women’s well-being in the marriage market. There is little indication that highly educated women are being disadvantage. College graduate seems to have the highest percentage of being married at least once at age 35-45. On the other hand, the more schooling a woman receive; she is more likely to be married at least once. Women who did not receive at least twelve years of school have a higher probability of not married for all races.

Empirical Economic Bulletin, Spring 2009, Vol. 2

115

References Ali, P. (2009). Black Women Have Higher Chance of Remaining Unmarried. Retrieved 2009, from THe Frisky: http://www.thefrisky.com/post/246-black-women-have-higher-chance-ofremaining-unmarried/ Census, U. B. (2007). Current Population Survey, 2007 Annual Social and economic . Retrieved from U.S. Bureau of Census: http://www.census.gov/apsd/techdoc/cps/cpsmar08.pdf Duncan, J. S. (2002). How does mother's education affect child height? Journal of Human Resources . Goldin, C. (1992). Understanding the Gender Gap: An Economic History of American Women. Johns, T. R., Yang, C. W., & Chen, P.-y. (2003). An Ordered Probit Model for Understanding Student Grade in Operations Management. Juhn, C. a. (1997). Wages inequality and family labor supply. Journal of Labor Economics . Krueger, A. B. (2005, December 11). What's the Return on Education. New York Times . Lleras-Muney, A. (2002). The relationship between education and adult mortality in the United States. National Bureau of Economic Research . Lochner, L. a. (2001). The Effect of Education on Crime: Evidence from prison inmates, arrests, and self-reports. National Bureau of Economic Research . Matsui, M. (2004). Searching for Mr. Right: The duration of remaining single based on evidence from Japan. Munane, R. J. (1981). New Evidence on the Relationship between mother's education and children's cognitive skills. Economics of Education Review . Rose, E. (2003). Does Education Really Disadvantage Women in the Marriage Market. University of Washington Institute for Economic Research Working Paper No. UWEC-2003-15 . Sander, W. (1995). Schooling and quitting smoking . Review of Economics and Statistics .

Empirical Economic Bulletin, Spring 2009, Vol. 2

116

The Effect of Tax-Burdens on Foreign Direct Investment: A Cross-Sectional Look at Developing Economies

Kevin Hauver

Abstract: This paper investigates the effect that certain aspects of the tax burden have on foreign direct investment in developing economies. Using data from 35 select countries, the paper uses an OLS regression model to determine the impact that various taxes, both on individuals and corporations, can have on FDI. The paper concludes that corporate tax rates are not a statistically significant factor for determining FDI inflows into a host country, but that indirect tax rates are. This is likely due to the use of ‘enterprise zones’, which offer favorable indirect tax rates to companies that choose to operate in a certain region of a host country.

JEL classification: F21, F23, H24, H25, C31 Keywords: Foreign direct investment, Capital taxation

Department of Economics, Bryant University, 1150 Douglas Pike, Smithfield, RI 02917. Email: khauver@bryant.edu. Phone: (978) 870-8706 ______________________________________________________ The author would also like to thank Dr. Ramesh Mohan, professor of economics at Bryant University, for all of his help and assistance. Empirical Economic Bulletin, Spring 2009, Vol. 2

117

1.0 Introduction Foreign Direct Investment has played a vital role in developing countries around the world. Poor countries often have low savings rates and thus, they must rely on FDI for the investment needed to lift them from the depths of poverty. Many nations use cuts in the corporate tax rates to encourage foreign firms to invest in their economy. This study aims to discern the effect that this corporate tax rate, as well as the value-added tax, has on the location of FDI. It tries to quantify exactly what impact that these tax rates have on the composition of FDI as a percentage of the host country’s GDP. The connection has important implications for macroeconomic policy. Presumably, developing countries who could benefit from additional FDI would desire to have it constitute a significant portion of their GDP, at least until they have the sufficient savings to sustain domestic investment. As such, these countries often use corporate and other tax cuts to attract foreign capital. Whether or not these taxes have a significant impact on FDI largely impacts the merit of these tax cuts. This study aims to fulfill two research objectives, one that is shared with many papers on the topic and one that is not. Primarily, it aims to determine whether or not corporate and indirect taxes play a significant role in attracting FDI to developing countries. Several papers on this topic have concluded that lower tax rates do in fact correlate with increased FDI inflows. However, these papers have focused primarily on developed countries, or developing countries in a specific region. The study of taxes and FDI with relation to developing countries as a whole is currently a void in the literature. This study aims to define whether or not taxes play a significant role in bringing capital to developing countries, countries which arguably need it most. In addition to focusing solely on developing countries, this paper also aims to quantify the effects that indirect tax rates have on foreign direct investment. International firms such as KPMG have entire practices based on “advising on the indirect tax consequences of entering new markets,”1 so it is likely that indirect tax rates are a significant factor in the investment decisions of multinational firms. This paper aims to go beyond the current literature and examine crosssectionally how both corporate and indirect tax rates affect investment decisions. The rest of the paper is organized as follows: Section 2 discusses the trends of this research topic. Section 3 gives a concise literature review. Section 4 outlines the empirical model 1

http://www.kpmg.com/Global/WhatWeDo/Tax/GlobalIndirectTax/Pages/default.aspx

Empirical Economic Bulletin, Spring 2009, Vol. 2

118

used. Section 5 presents and analyzes the empirical results. Lastly, section 6 follows with a conclusion. 2.0 Trends

Figure 2- 1: FDI Flows, 1999-2006

For the past ten years, foreign direct investment has been on the upswing in certain regions of the world. This stream of investment has taken several different forms. Firms in highly developed countries such as the United States continue to look for ways to improve their financial performance, and one of these ways is to cut labor costs. The abundance of cheap labor in developing countries has led to a flood of outsourcing in the past decade. Much of this outsourcing results in FDI in other countries, often in the form of factories or other physical capital. Other times, multinational corporations looking to diversify or expand seek to acquire controlling shares of foreign companies. These acquisitions are another source of FDI for developing countries. An increase in overseas operations, as well as increased investment in other companies outside the home nation has characterized the business climate of the twentyfirst century. FDI was at a peak in 2000, and experienced steady decline in the next few years. This was likely due to the September 11 attacks, which induced fear in many international investors. Despite this setback, the global economy made a recovery, and FDI continued to grow after reaching a low in 2003. FDI flows finally reached their 2000 levels again in 2006, and we

Empirical Economic Bulletin, Spring 2009, Vol. 2

119

expect to see increasing FDI in the years to come. Figure 2-1 on the previous page shows the amount of global FDI from 1999 to 2006. One can see from Figure 2-1 that FDI flows to developing countries are a relatively small portion of the overall foreign investment. Nevertheless, foreign direct investment is an important source of capital for developing countries. It is crucial in boosting their economy and their domestic savings are often not enough to support investment without it. Because of this, many countries are increasingly opening their borders to support FDI. One of the ways in which countries are differentiating themselves from their neighbors to attract FDI is through the corporate tax rate. Multinational corporations looking to minimize their overall tax burden are often willing to shift operations to countries with lower corporate rates, and as the economy becomes more globalized, this tax rate is becoming more of a potential competitive advantage. A survey by the accounting firm KPMG, which tracked corporate tax rates in 86 countries from 1993 to 2006, found that “the survey has recorded a consistent and dramatic reduction in corporate tax rates over that 14-year period.”2 Figure 2-2 shows the overall downward trend that global competition has had on corporate tax rates:

Figure 2-2: Average Corporate Tax Rate, 1993-2006 Source: www.thetaxfoundation.org

http://www.taxfoundation.org/blog/show/1978.html

Empirical Economic Bulletin, Spring 2009, Vol. 2

120

Global competition for increasingly mobile capital has undoubtedly driven down corporate tax rates. But the corporate tax rate is just one of several taxes that could potentially influence the location decision of multinational firms. This paper also aims to discern the effect that the value-added tax (or sales tax in some countries) has on the FDI inflows in a particular country. Another trend that we seek to analyze for the purpose of this paper is that of indirect taxes. Since the paper seeks to identify both corporate and indirect tax rates as factors in investment decisions, it is prudent to examine the global trend for indirect taxes in addition to the trend in corporate taxes.

Figure 2-3: Indirect Taxes by Region, 1992 and 2002 Source: http://www.newint.org/features/2008/10/01/facts-tax As one can infer from the graph above, indirect tax rates have not changed by nearly the same magnitude as corporate tax rates have. In fact, while corporate tax rates have declined substantially over this period, indirect tax rates have increased in all the surveyed regions. This makes sense economically, as many countries have likely resorted to increasing indirect tax rates to offset declining corporate tax revenues. However, it implies that countries are not likely using indirect tax rates as a competitive advantage for attracting foreign investment, as they are with corporate rates. Countries likely do not view indirect taxes as a significant factor for investment, and thus are very willing to substitute indirect taxes for corporate taxes. However, although not Empirical Economic Bulletin, Spring 2009, Vol. 2

121

adjusted on a competitive basis, multinational firms may still consider indirect taxes a factor for their investment decisions. This study aims to find whether or not that is true. 3.0 Literature Review The empirical research on this topic comes to a general consensus that corporate income taxes have a statistically significant effect on foreign direct investment, whether at the state level or the country level. Gropp and Kostial (2000) argue that tax regimes, including corporate tax rates, significant affect FDI inflows for a foreign country. They also note the trend that the competitive market for FDI causes countries to systematically lower their corporate tax rates and thus erode the tax base. Bellak et. al (2007) also agree that changes in the corporate tax rate are negatively correlated with FDI inflows. They also say that changes in the tax rate become less effective at attracting FDI as a nation’s infrastructure becomes more developed. Desai and Hines Jr. (2001) were rare in that they also studied the effect of taxes other than the corporate income tax. ““Taxes other than income taxes significantly affect the pattern of income production by multinational firms by altering their investment and transfer-pricing incentives. (Desai and Hines Jr, 2001)” They also said that governments are likely competing actively for FDI using their other tax rates as differentiation. Mooij and Ederveen (2001) took the analysis a step further and calculated that a 1% chance in corporate income tax rate corresponds to 3.3% decrease in FDI inflows. They also argue that marginal or average tax rates have more of an impact on investment decisions than the statutory base rate does. Cummins and Hubbard (1994) further discredit the notion that taxes do not affect international investment decisions. In fact, the authors argue, ““Tax parameters influence foreign direct investment in precisely the ways indicated by neoclassical models.” Hines Jr. (1993) was unique in that it looked at investment decisions for individual U.S. states. It found that the same patterns found in country-level data held true for states, pointing out that high corporate income tax rates had a negative effect on a state’s investment flows from other states. Egger at. Al (2007) had findings consistent with most empirical studies, in that ““Unilateral tax rates significant affect the production and location decisions of multinational firms.” However, the paper goes on further to explain that bilateral tax rates (the tax rate of the host country in relation to the investing country) are also very significant for investment decisions. Overall, the general consensus is that corporate income tax rates do in fact have a statistically significant effect on foreign investment decisions.

Empirical Economic Bulletin, Spring 2009, Vol. 2

122

4.0 Data and Empirical Methodology 4.1 Definition of Variables FDI%GDP = β0 + β1FDI+ β2GDPCAPITA + β3CORPTAX + β4INDIRECTTAX + β5PROJGROWTH +β6INFLATION + β7OPENNESS This empirical model is based on one used by Gropp and Kostial (2000) where OrdinaryLeast Squares (OLS) regression model used to determine the impact of seven variables on a country’s FDI to GDP ratio. In this empirical model, FDI represents the flow of funds from all foreign countries to a particular host country in the year 2007. The transfer of funds to foreign affiliates takes one of three forms: equity capital, inter-company debt, and reinvested earnings. The definition of FDI in this paper is consistent with the IMF definition of FDI flows. FDI as a percentage of GDP is used rather than the absolute value of GDP inflows. This is so that the coefficient of the independent variables does not change with the scale of the country being considered. This is necessary because of the cross-sectional nature of the data. For example, if the absolute value was being used, the regression might show that a 1% change in the corporate income tax rate changes FDI by $500 million. This doesn’t make sense in a country like Ecuador, where the total FDI for the year was only $178 million. For this reason, FDI as a percentage of GDP is used. Independent variables consist of seven variables obtained from various sources. Appendix A provides data source, descriptions, and expected signs for the variables. The FDI of the host country is used to control for economies that already have substantial investment. These countries have proven stable and profitable for FDI and thus are likely to attract more. GDP per capita is used for a similar reason, to control for larger countries that have more market opportunities and are thus likely to attract FDI. CORPTAX is the first focal variable of the study. It measures the corporate tax rate that the host country imposes on corporations. The rate levied on income accrued to foreign corporations is used, if it differs from the domestic rate. INDIRECTTAX is the tax rate levied on purchases or consumption within the host country, either a sales or a value-added tax. Tax data was obtained from Deloitte International Tax Source. PROJGROWTH is the projected growth rate in the country’s economy for the 2008, the year following the investment year. This data was obtained from the IMF’s World Economic Outlook Database, October 2007. INFLATION is lagged one year, and it is included to show the presumably negative effect that rampant inflation has on the outlook of investors. OPENNESS is Empirical Economic Bulletin, Spring 2009, Vol. 2

123

measured by imports and exports as a percentage of GDP. It is included to control for the fact that countries considered more “open” are inherently more appealing for foreign direct investment. Data for GDP, Inflation, and trade was obtained from the IMF’s World Economic Outlook Database, October 2007. 4.2 Data The data to determine the above variables was obtained from various sources. The data for FDI inflows to the sampled countries was obtained from the United Nations Conference on Trade and Development (UNCTAD) web site, specifically the FDISTAT program. Data for GDP, projected growth, and inflation were obtained from the International Monetary Fund’s World Economic Outlook reports from various years. GDP for 2007 and inflation rates from 2005 were obtained from the October 2008 reports, and projected growth for 2007 was obtained from the October 2006 report. The October 2006 report was chosen because the 2007 growth rate projected at the end of 2006 would have been consistent with the projections firms would use to estimate growth in various countries, from which they could potentially choose as a location for investment. Corporate and indirect tax data for most countries was obtained from Deloitte’s International Tax Source database, at http://www.dits.deloitte.com/Default.aspx. 5.0 Empirical Results This paper uses an Ordinary-Least Squares (OLS) model to determine the correlation between a country’s foreign direct investment inflows and its corporate and indirect tax rates. Table 1 shows the results from this model. Contrary to what was expected, corporate taxes are not a statistically significant factor for FDI composition in developing nations. The model does show some linkage between corporate taxes and FDI, but not enough to be statistically significant. Indirect taxes, though, are indeed statistically significant. However, we expected that the indirect taxes would carry a negative sign for the coefficient, but instead the results show indirect taxes are positively correlated. This could be due to the fact that countries with higher overall indirect rates are better able to offer foreign businesses incentives to locate in so called ‘enterprise’ zones. These are certain areas, designated by the host country, which have much lower indirect tax rates (sales or VAT) than the rest of the country. The fact that the indirect tax rate shows a positive coefficient could indicate that countries actively soliciting FDI have a higher national indirect rate but make better use of enterprise zones. Overall, the data shows that every 1% increase in the indirect tax rate corresponds to a 0.43% increase in the FDI/GDP ratio. Empirical Economic Bulletin, Spring 2009, Vol. 2

124

Another possible explanation for the positive coefficient is the relation between corporate and indirect tax rates. If a country is using indirect tax rates to lure foreign investment, it is likely that they are competing on a corporate tax basis as well. This means that as countries slash their corporate tax rates to lure investment (reflected by a negative coefficient in our model), they might increase indirect tax rates to offset the revenue loss. This possible strategy explains both the negative coefficient for corporate taxes and the positive coefficient for indirect taxes. However, although it explains the correlation between indirect tax rates and FDI, it does not necessarily explain the causation. The data does not indicate whether the FDI is actively drawn by indirect tax adjustments, or if the higher indirect tax rates are simply a reaction to reductions in other taxes. In addition to the focal points of corporate and indirect taxes, this study also analyzed a number of control variables to better understand what drives foreign investment decisions, as well as to avoid any omitted-variable bias. These variables did not prove statistically significant, with the exception of openness, which was significant down to the 1% level. For the selected sample, a 1% increase in openness corresponded to a 0.532% increase in the FDI/GDP ratio. Considering openness reached as high as 386% percent in this sample, this shows the power that maintaining open trade can have for attracting investment. The projected growth rate of the country showed some linkage, but not enough to be statistically significant. This indicates that for a developing country, the most important factors that correlate with foreign investment inflows are indirect tax rates and openness, closely followed by corporate tax rates and projected growth.

Empirical Economic Bulletin, Spring 2009, Vol. 2

125

Table 1 – Regression Results Variable

Coefficient

Standard Error

t-Statistic

Probability

INDIRECTTAX

0.431319**

0.168550

2.559002

0.0183

CORPTAX

-0.126026

0.131126

-0.961107

0.3474

OPENNESS

0.05327276*** 0.01835575

0.02902238

0.0085

PROJGROWTH

0.689471

0.515859

1.336551

0.1957

INFLATION

0.072827

0.183438

0.397011

0.6954

GDPCAPITA

-.0000970

0.000165

-0.587125

0.5634

FDI2007

-.0000379

0.000048

-0.845875

0.4072

-5.768501

6.275358

-0.919231

0.3684

*,**, and *** indicates significance at the 10%, 5%, and 1% level, respectively. R-squared

0.594225

Adjusted R-squared

0.458967

6.0 Conclusion The results of this model draw conclusions that are substantially different than what the literature would predict. Most of the literature concluded that the corporate tax rates are a significant factor for international investment decisions. However, this paper concludes that for these countries, corporate tax rates are not a statistically significant factor in determining FDI inflows. However, the paper concludes that indirect tax rates are a significant factor for determining FDI in developing countries – a similar finding to that of Desai and Hines Jr. (2001). We also find that a developing country’s level of openness is very important for determining FDI inflows. The major implication made by this paper is that international corporations consider different factors when investing in developing nations than they do when considering developed nations. Developed countries are generally stable, have low, predictable growth rates, and are very open to trade. Thus, when investing in developed countries, firms are likely to look more at Empirical Economic Bulletin, Spring 2009, Vol. 2

126

corporate tax rates, which can have a significant factor on their total profits. This idea is backed by the majority of the literature on this topic. However, this paper provides some insight into the decision factors for investment in developing nations. The model indicates that the two most important factors for determining FDI inflows in a given country are indirect tax rates and openness. Openness is immediately intuitive, as firms are likely to invest in countries that make it easy to trade and do business internationally. The correlation between indirect tax rates and FDI, however, is not quite as apparent. It is possible that countries with higher indirect tax rates are those who levy high value-added taxes nationwide but make effective use of enterprise zones to lure foreign investment. This would explain the puzzling positive correlation between indirect tax rates and FDI as a percent of GDP.

6.1 Policy Implications The findings in this paper have considerable implications for economic policy. It finds that corporate tax rates are not statistically significant for determining the proportion of FDI inflows into a country. Developing countries often rely on foreign investment to sustain capital expansion that their domestic population is unable to support on its own. Economic theory would indicate that corporate tax rates would be a significant decision factor; the fact that it isn’t has implications for the economic strategy of developing nations. Instead, the paper suggests that these countries should focus first open making their country as open to trade as possible, as this has a huge impact on FDI (5% of GDP for every 1% increase in openness). Also, the paper suggests that countries may be able to raise their overall indirect tax rates, and offer businesses incentives via enterprise zones. This allows governments to have greater flexibility in making certain locations seem attractive for investment, and the data suggests that it is an effective strategy.

Empirical Economic Bulletin, Spring 2009, Vol. 2

127

Appendix A: Variable Descriptions Acronym

Description

Data Source

Expected Sign (+/-)

FDI%GDP

FDI

Foreign Direct Investment as a percentage

See sources for “FDI” and

of the host country’s GDP in 2007

“GDPCAPITA”

Foreign Direct Investment inflow to the

United Nations Conference on

host country in 2007, expressed in United

Trade and Development

States dollars

(UNCTAD) web site –

“FDISTAT” GDPCAPITA

Gross Domestic Product per capita in the

Data for GDP and GDP per

host country in 2007, expressed in United

capita obtained from the IMF’s

States dollars

World Economic Outlook

Reports CORPTAX

The statutory corporate income tax rate

Deloitte International Tax

levied on foreign corporations in the host

Source and

country in 2007. Expressed as a percentage.

www.doingbusiness.org for

selected countries. INDIRECTTAX

PROJGROWTH

The indirect tax rate (either sales or value

Deloitte International Tax

added, depending on which is used) in the

Source and

host country in 2007. Expressed as a

www.doingbusiness.org for

percentage.

selected countries.

Projected growth rate in the host country’s

IMF World Economic Outlook

gross domestic product in 2008. Expressed

Reports

as a percentage. INFLATION

The inflation rate in the host country in

IMF WEO Reports

A measure of the host country’s openness to

Trade Data obtained from

trade. Measured as total exports plus total

Correlates of War Project Trade

imports, all over gross domestic product.

Data Set Codebook.

2005. Data is lagged two years to show delayed effects of inflation. Expressed as a percentage. OPENNESS

Expressed as a percentage.

Empirical Economic Bulletin, Spring 2009, Vol. 2

128

Appendix B: Summary Statistics

Variable Observations Mean Std Dev Min Max FDI%GDP 30 6.00 6.28 0.28 28.91 FDI 30 $15,646 $19,277 178 83521 GDPCapita 30 $8,494 $7,801 828.85 35,162.93 CorpTax 30 25.85 6.49 15 35 IndirectTax 30 14.55 5.76 0 22 ProjGrowth 30 5.57 1.60 2.7 10 Inflation 30 5.82 4.06 0.7 17.1 Openness 30 89.6400 70.0700 23.04 386.55

Empirical Economic Bulletin, Spring 2009, Vol. 2

129

Bibliography Becker, Johannes, Clemens Fuest and Thomas Hemmelgarn. "Corporate Tax Reform and Foreign Direct Investment in Germany - Evidence From Firm-Level Data." CESifo Working Paper (2006). Bellak, Christian, Marcus Leibrecht and Joze Damijan. "Infrastructure Endowment and Corporate Income Taxes as Determinants of Foreign Direct Investment in Central and Eastern European Countries." LICOS Discussion Paper Series (2007). Cummins, Jason G and R Glenn Hubbard. "The Tax Sensitivity of Foreign Direct Investment: Evidence from Firm-Level Panel Data." NBER Working Paper Series (1994). De Mooij, Ruud A and Ederveen Sjef. "Taxation and Foreign Direct Investment: A Synthesis of Empirical Research." CPB Discussion Paper (2001). Desai, Mihir A and James A Hines Jr. "Foreign Direct Investment in a World of Multiple Taxes." National Bureau of Economic Research (2001). Egger, Peter and Doina Maria Radulescu. "Labour Taxation and Foreign Direct Investment." CESifo Working Paper (2008). Egger, Peter, et al. "Bilateral Effective Tax Rates and Foreign Direct Investment." Oxford University Center for Business Taxation (2007). Gropp, Reint and Kristina Kostial. "The Disappearing Tax Base: Is Foreign Direct Investment Eroding Corporate Income Taxes?" European Central Bank Working Paper Series (2000). Heady, Christopher. "The 'Taxing Wages' Approach to Measuring the Tax Burden on Labor." CESifo Working Paper (2003). Hines Jr, James R. "Altered States: Taxes and the Location of Foreign Direct Investment in America." NBER Working Paper Series (1993). Slemrod, Joel. "Tax Effects on Foreign Direct Investment in the U.S.: Evidence From a CrossCountry Comparison." NBER Working Paper Series (1989).

Empirical Economic Bulletin, Spring 2009, Vol. 2

130

An Empirical Analysis of the impact of home foreclosure on the crime rate: evidence in Atlanta, GA

Luis G. Acevedo1 Abstract: Over the past 28 years the United States has seen its share of prosperity and hard times. At times there have been significant increases in the number of subprime mortgages issued in the United States. Studies have shown that the number of foreclosures is highly correlated to the number of subprime loans issued. Another main issue that seems to occur with the abundance of loans and the spike in foreclosures is that crime rates tend to also increase during these times. The two major crimes that will be taken into account are violent crime and property crimes. I use foreclosure, income, and crime data from various databases from 1980 to 2008. The purpose of this paper is to determine the relationship between United States foreclosure rates and crime rates. Results from simple regression models suggest that the increase in foreclosures is significantly related to crime level increases across the United States. I conclude with the discussion of implementing social and educational programs to help the less fortunate that are on the verge of losing their homes. JEL classification: E21; E51; G21; G33 Keywords: Foreclosure, Property Crime; Violent Crime; Unemployment Rate; Atlanta

Bryant University, 1150 Douglas Pike, Smithfield, RI 02917. Phone: (401) 261-4949, Email: Lacevedo@bryant.edu

The author thanks Dr. Ramesh Mohan, Department of Economics, Bryant University, for

Empirical Economic Bulletin, Spring 2009, Vol. 2

131

valuable discussions and personal guidance throughout this research. The author also gives thanks to Dr. Peter Nigro, Department of Finance, Bryant University, for assistance on foreclosure data. I.

Introduction The outlook of the United State economy has been gloomy at times over the last couple

of decades, during prosperous time we witness business that is booming and it is evident that there are legal opportunities to make money. On the contrary, during periods of declining growth we have seen that the number of foreclosures has increased during this time. The abundance of subprime loans during the years from 2002 to 2004 lead to a record number of no-documentation loan originations, these loans are the main reason why the subprime lending market crashed. Studies have shown that the higher the degree of foreclosure tends to more negative aspects of a community (Immergluck and Smith 2006). This analysis reports specifically on foreclosure activity in Atlanta, Georgia, between January 2006 and August 2008. A foreclosure in itself has many other impacts on the community and there are usually many other impacts of this event. For instance, usually many homeowners have be forced to foreclose their homes if they have recently lost a job and are not able to cover their mortgage payments, that is another main reason for foreclosures because of divorces. Divorces are a key factor in determining foreclosure because this typically leads to only one partner having the responsibility of paying a mortgage, household incomes usually decrease substantially and the probability of foreclosure is greater. In other instances, foreclosures occur because of influential factors of the economy, such as interest rates. The subprime mortgage debacle brought with it a huge number of Hybrid-Adjustable rate mortgages that begin at low rates but then hit reset dates where the interest rate jump to a capitalization rate (typically 6.00%) plus an index rate such as the 3 month LIBOR (London Inter-Bank Offering Rate). When this

Empirical Economic Bulletin, Spring 2009, Vol. 2

132

adjustment occurs it causes the mortgage payment to increase drastically which is a main reason why default rates usually increase as well. The outcome of foreclosure has been known to negatively affect the community in other ways as well. For instance, if there is a foreclosure in a neighborhood the homes in a close proximity to the foreclosed home will have their value drop by 1.5%. Another negative aspect is that when these homes are foreclosed they typically become abandoned and this usually leads to other problems that society sometimes faces. Over that past couple of years (2007-2009) there has been a sever increase in the number of defaulting mortgage holders and as stated before this only worsens the problem for the economy. The purpose of this case study is to answer questions on the relationship between the foreclosure rates and crime rates in a community, in this case more specifically the Fulton County in Georgia. II. Literature Review Immergluck and Smith (2006) study shows that the higher the number of foreclosures in a community then the more adversely affected a communities home prices will be. It also states that as the number of homes that become unoccupied typically means that in these areas we will see that there are more individuals that are unemployed and the more people that will be looking for jobs. This typically is also highly correlated to the high number of crimes in a community. Raphael and Winter-Ebmer (1998) study investigate the affects of unemployment on the community. Usually people that become unemployed are more likely to commit crime because of the fact they have lost a source of their income. With this said, it is apparent that a high unemployment rate is bad for society, individuals have too much time on their hands and when they are not closely watched they may be motivated to commit a crime if they have trouble finding employment.

Empirical Economic Bulletin, Spring 2009, Vol. 2

133

Yamada (1985) goes into further detail the previous two papers that were reviewed; this study shows that a deteriorated labor market shows that there is a higher number of crimes that will be committed. The study used was a vector auto regression model, using time-series data. This study was important because of the fact that it shows how much the crime rates will be affected by a decrease in the quality of the labor market. Cerro and Meroni (2000) study was very similar to this analysis because of the fact that it took a look at some of the determinants in the crime rate in Argentina from 1990-1999. For instance, in Atlanta poverty typically leads to higher levels of crime in a community. The study done by Cerro and Meroni concluded that income inequality and the unemployment rates were the most significant factors were found to have the most positive and significant effects on the crime rate.

III. Trends Atlanta is located in central Georgia; it has seen its population growth become one of the highest in the United States in the last couple of years. One reason that there have been ample job creations in various sectors such as: finance, service industry, and small businesses. Atlanta’s location is prime because if it’s mild temperature and metropolitan area. It is home to HartsfieldJackson International Airport, which is known as "the nation's busiest airport in terms of flight operations.” The Atlanta police department does a great job of keeping the community safe and an even greater job of keeping the crime records organized for detailed statistics. Uniform Crime Reports is a database that the city of Atlanta has online, which tracks the number of crimes, and detailed information of where, when, and what occurred at a crime scene. This is a great asset of the police department because they are able to see which area criminals are more willing to

Empirical Economic Bulletin, Spring 2009, Vol. 2

134

commit crimes and they are also able to determine which areas they should allocate more police force to. The city of Atlanta is broken down into six different zones. The two figures below show exactly how the zones are broken down and where exactly they are located.

Figure 1 – Atlanta Police Department Zone Visual

Source: Atlanta Police Department The breakdown of Atlanta’s police force is divided into six patrol zones, the first zone is also known as Western Atlanta. The second zone is Northern Atlanta and is probably the biggest of all six zones area-wise. The next zone, zone three is located in Southern Atlanta. The fourth zone is the second largest biggest on and is located at the Southwest corner of Atlanta. Zone five is the most populated as it is Downtown Atlanta. And lastly zone six is Eastern Atlanta. Mortgage default rates have been increasing over the last couple of years and this has had a great deal of affect on the number of foreclosure that has taken the United States by storm. Currency we can see that the number of foreclosure has dampened the US housing markets and has caused the prices to decrease from all-time highs. The average home price in October 2004 was $264,540 where in today’s markets a single-family home comes at an expense of $180,100

Empirical Economic Bulletin, Spring 2009, Vol. 2

135

in December 2008. The subprime mortgage mess has also affected many US investment banks and investment firms that have long positions in subprime-backed securities. These mortgagebacked securities, which are basically bonds with the lowest rating, had the subprime loans as their underlying collateral. As these mortgage loans started being delinquent the value of these securities started to decrease drastically to the point where many banks started losing money. The reason why we have seen the economy so deeply affected by these defaults is because many of these loans were packaged into collateralized debt obligations (CDO). These CDOs were sold to banks and they consisted of pools of many subprime mortgages. When the mortgage payer would make a payment it would go directly to the holder of the Mortgage-backed security and that would be their investment cash flow. As stated the problem with this is since the mortgage holders were defaulting many of the investment banks were coming up short on cash. These losses were even more magnified because of the fact that many of these investment banks were highly leveraged, some were even levered up to 50 times the amount they had in assists. So little losses on these CDOs would mean a great deal, and the amount of money lost was tremendous. The fact that the number of Foreclosures in the Atlanta area keeps increasing, and is estimated to peak sometime in 2010 leads researchers to believe that home prices are going to continue to come down. This is bad in a way because many of the loyal home owners that have been paying their mortgages down for most of their life have seen their 401Ks have been decreasing as well as the amount of equity that they have built in their house has been lost. The only positive aspect of this type of real estate bust is that when the price hit rock bottom there will be an abundance of home properties on the market. This leads many investors to believe that this may very well be the “time to buy”. The problem that has also affected the opportunity to buy in our current economic crisis is that the United State economy is also going through a

Empirical Economic Bulletin, Spring 2009, Vol. 2

136

“credit crunch”, this means that the availability for investors to borrow money has become a lot harder because of the fact that many people have been defaulting on their mortgages. This crisis is one of the main reasons why the economy has had trouble rebounding. The typical US consumer has felt a great deal of pressure on his/her wallet which has directly affected the level of consumer spending. With this said, many s try to turn to banks to borrow more money in order to try to get by, and as stated before many of these banks are stricter than they were before because they themselves are low on cash. To conclude this thought, that can be said is that the US economy is in big trouble. The Unemployment rate is usually used as a key determinant in estimating crime rates because as history has shown as the unemployment rate increases people have less income coming into their pockets and they also have more time on their hands. Because of this fact some of these distressed workers take into consideration committing actions that may be unlawful, especially if that is the only means of getting their hands on some cash. Although the unemployment rate is not highly correlated with the violent crime rates, it is definitely more closely related to property crimes (Rafael & Winter-Ember, 2001). This study shows evidence that the drop in property crime rates during the 1990s in the United State was attributed to the lower unemployment rates. The effect of the change in the wage from month to month has mostly affected business’ profits, crimes such as burglary, larceny and auto theft are typically the property crimes that tend to spike the most during periods of increased unemployment. During the 1990s Atlanta had tremendous growth economically, it is home to more that 24 Fortune 1000 companies (Ex. Home Depot, UPS, Bellsouth, and Delta Airlines). During the period of 1990-2000 more than 1.1 million new residents moved to the Atlanta area, which has attracted new businesses to open in this area. Many feel that the reason why Atlanta has been

Empirical Economic Bulletin, Spring 2009, Vol. 2

137

able to do well during recessions is because they have a strong labor force and many companies that are financially strong, and so Atlanta typically suffers less from slowdowns in the US and is able to get back on track quicker than most states. Figures from the Bureau of Labor Statistics in 2000, estimate that during the time period 1998-2025 metropolitan Atlanta is expected to create 1.8 million high-paying tech jobs in light of the growth in the business sector. The Atlanta economy’s biggest sectors are services, manufacturing, and trade. Average household income for a family in Atlanta in 2008 was $55,939 according to the census bureau. The population growth is increasing at alarming rates and how job creations in Atlanta have been plentiful. A important topic that many community officials have looked into is that the crime rates in the respective areas where there are higher levels of income will not be as affected as the zip codes that are the least wealthy will not be as effected when it comes to foreclosures and unemployment spikes. Crime is most elastic with to wages in sectors that use low-skilled labor these results were determined by looking at sector specific wages (Doyle, Et Al, 1999). They also found that income inequality has no significant effect on crime rates. It is easy to disagree with this point because of the fact that because my studies have shown that in the lower income areas of Atlanta the crime rates are highly positively correlated to foreclosure, unemployment and income levels. Consider the scenario, if the crime rate in Atlanta were to increase by 20% what would this most likely be attributed to? This study looks at the results show that the percentage increase of crimes would be higher in the low-income communities rather than in the wealthier ones.

Empirical Economic Bulletin, Spring 2009, Vol. 2

138

Figure 2 – Total Percentage of Population Getting arrested in Atlanta (Month-to-Month)

Source: Author Compilation Violent crimes are defined as an incident where the offender uses or threatens to use violent force upon the victim. Criminal homicide, forcible rape, robbery, and aggravated assault are all considered violent crimes. These types of offenses spiked and hit a peak during 1992. The number of violent crimes has steadily decreased since then however it became an issue again in 2004 when rates turned and started rising. When it comes down to national rates Atlanta falls under the worst part of the country when it comes down to violent crimes – the South, made up 42% of the United States total violent crimes registered. It has also been found that younger people are more likely to commit more violent crimes than older people. Individuals in their thirties, forties and fifties made up 41 percent of the people arrested for violent crime, where as 33 percent of all violent crimes in the United States were in their twenties. People of the age of sixty comprised only 1 percent of violent crimes. Lastly these studies have also shown that 62% of all violent crimes happen to be aggravated assaults.

Empirical Economic Bulletin, Spring 2009, Vol. 2

139

Figure 3 – Property Crime vs. Violent Crimes in Atlanta (Jan 2006 – Mid 2008)

Source: Author Compilation

Property crimes are considered to be unlawful activities that involve taking money or property, but it does not necessarily involve force or threat of force against a victim. Property crimes include burglary, larceny, theft, motor vehicle theft, arson, shoplifting, and vandalism. The affect of unemployment and foreclosures tend to escalate the rates in property crimes, this is the case be many people find themselves at home with less money in their pockets and more time on their hands. Some of these individuals take these problems into their own hands even though this might encounter coming across other people’s property. The individuals that have been hit hardest by the downturns in the economy are typically the ones that tend to search to find a source of income. One of the reasons why property crime rates increase during times of higher

Empirical Economic Bulletin, Spring 2009, Vol. 2

140

foreclosure rates, higher unemployment rates, and typically recessions because many business owners try to lower their costs and spend less money in security systems and hiring security officers. This could be one of the reasons why property crime rates increase IV. Empirical Models The two equations that I will be working with are to determine the level of crime rates in Atlanta. The two dependent variables are Property Crime Rates and Violent Crime rates. Equation1.1: Model: Property Crimes = β0 + β1*Unemployment Rate + β2* Income + β3*ANoFpZ + β4*PoPgA + β5 *Violent Crimes Equation1.2: Model: Violent Crimes = β0 + β1*Unemployment Rate + β2* Income + β3*ANoFpZ + β4*PoPgA + β5 * Property Crimes

Table 1 – Summary of the variable signs for Property Crime Model Variable

Expected Sign

Income

‐; studies have shown that more robberies occur in areas of lower income, if income increase there tends to be higher security and therefore fewer crimes occur.

Unemployment Rate

+; The higher the unemployment rates goes means that there will many people trying to find alternate way of making money.

# of Foreclosures

+; as more homes get foreclosed one means that there are typically less employed individuals in the community.

% of Population getting arrested

+; the more people getting arrest means that crimes rates will be increased.

Violent Crimes

‐; study have shown that as the number of violent crimes increase usually make the area less desirable, and the police force usually become strengthened and the overall occurrence of crimes, property crimes more specifically will decrease.

Empirical Economic Bulletin, Spring 2009, Vol. 2

141

V. Data The data for this paper was obtained from various sources, the majority of the data came from the Uniform Crime Reports Database of the Atlanta Police Department. The Bureau of Labor Statistics (BLS) and the government website for the census were also key contributors. Foreclosure data and demographics data of Atlanta were provided by Peter Nigro from Bryant University, RI. The explanation of the data is below: Variable

Description

Data Source

Property Crime

Property crimes include burglary, larceny, theft, motor vehicle theft, arson, shoplifting, and vandalism.

Atlanta Police Department – Uniform Crime Reports Database

Violent Crime

Violent crimes consist of criminal homicide, forcible rape, robbery, and aggravated assault.

Atlanta Police Department – Uniform Crime Reports Database

Unemployment Rate

The unemployment rate is percentage of the population that is currently unemployed.

Bureau of Labor Statistics (BLS)

Average HH Income

The average gross amount of money that a working household bring home every year.

Bryant University Professor of Finance: Peter Nigro

# of Foreclosures

The number of properties that get taken over by a bank because the homeowner can no longer afford it.

Bryant University: Peter Nigro

% of Population Getting Arrested

The number of arrests made monthly divided by the total population.

Atlanta Police Department – Uniform Crime Reports Database and the demographics report

Empirical Economic Bulletin, Spring 2009, Vol. 2

142

a. Regression Results Dependent Variable Explanatory Variables Unemployment Rate Average HH Income # of Foreclosures % of Population (Arrested) Violent Crime

Property Crime in Atlanta (1996‐1998) Model 1

Violent Crime in Atlanta (1996‐1998) Model 2

0.074726

-0.009265

(0.1143)*

(0.9324)

-0.64374

-2.067541

(0.0970)**

(0.0148)**

0.058055

0.039188

(0.0021)**

(0.3962)

1.4074

2.804591

(0.0000)**

(0.000)**

-0.332181 (0.000)**

Property Crime

-1.687113 (0.000)**

Method Sample Size R‐Squared

OLS 32

0.980525

0.871236

* Significant at 15% confidence level **Significant at 10% confidence level From the regression results, where the dependent variable is property crime, we are able to see that almost all of the variables had the expected signs, only the unemployment and the average household income variables were different from what was expected. Both average household income and violent crimes had negative signs. The coefficient for income was 0.64374 which leads us to believe that as the income levels increase in a city the number of property crimes should decrease because there is higher security and the potential of getting caught is greater. Many high net worth individuals can afford to buy security that protects them from property damage or theft, this is something that may not be available to protect the people who have less income and cannot afford extra security. On the other hand we can also see that

Empirical Economic Bulletin, Spring 2009, Vol. 2

143

violent crime is greater than the amount of property would decrease. Consider the scenario, if there was a city where a tremendous amount of violent crime are the odds are that people are going to move away and that the police will strengthen their police force there. So the marginal opportunity of committing a property crime will be highly increased which could deter potential thieves. The unemployment rate, foreclosure rates, and the percentage of the population getting arrested all had positive signs. The correlation between foreclosure rates is simple to understand, if the number of foreclosure in a community is increasing this will mean that some people could have quite possibly been living in an area that they could not afford which in a sense could make them vulnerable to committing a property crime. For instance, a person may want what others’ have and if the only means of attaining it may be by committing an offense, this is likely to stop some people and for others it really does not. Unemployment is similar to the foreclosure’s expected affect on the property crime rate. As workers have become laid off and lose their jobs they will most likely still have some of the fundamental expenses (home, car, and food) but will have a significantly lower income. In the meantime many of these people look for other ways to make money and to make ends meet. I believe that this is the key factor on why the crime rate is highly related to the unemployment rate. The above regression also shows that all the factors chosen were important in the equation - they all play a significant part in the increase or decrease of the property crime rate. Almost all of the factors fall into the 10% or 15% range which means that there is a 85% or 90% chance that they play a role in the amount of property crime in Atlanta. Since all the variables are important, this makes the accuracy of the regression more viable. The study went further to also see the affects of these variable and their impact on determining the level of violent crimes in Atlanta. For the most part many of the variables affect

Empirical Economic Bulletin, Spring 2009, Vol. 2

144

the violent crimes in the same way that property crimes were affected. Both average household income and the unemployment rates had negative coefficients which means that they are inversely related to the level of property crimes. At any point in time when these variables are increasing, then the level of violent crime should be decreasing. For household income this is fairly easy to understand because as a family has a higher level of income they can afford to buy protection and security, therefore it should lower the level of violent crimes. On the other hand, when the unemployment rate is increasing our results show that the number of violent crimes will decrease, this can be thought as people are willing to commit small crimes because they are not considered life threatening or felonies. Foreclosure rates are positively correlated because as the rate of foreclosures increases then there is a downward pressure on home prices and these cities are known for having high crime rates because some of these foreclosed homes become abandoned. The percentage of individuals getting arrested is positively correlated because if there is more people committing crimes then the proportion of the population not committing crimes will therefore decrease. The number of property crime is negatively correlated; this is easy understood by comprehending that if the number of property crimes is increasing then what typically happens is that the police force gets stronger which will deter individuals from committing violent crimes. VI. Conclusion When people think of Atlanta many individuals think of its crime rates and think the worst-case scenario and how things are unsafe. What they forget to look at is all the positive that Atlanta has to offer, this city has great opportunities and the job growth and its metropolitan area is one of the most potent in the United States. However this analysis looks at which issues affect the crime rates and how viable they are. We see that some variables are more significant than others but

Empirical Economic Bulletin, Spring 2009, Vol. 2

145

what is most important is that they give us a greater picture how these fundamental these variables are. The results are that the property crime rate is more elastic in regards to the home foreclosure rates than violent crimes. In conclusion, the regression is considered accurate and viable and should hopefully prove to help us understand why foreclosure rates and other factors affect the crime rates. VII. Proposals There are various programs that the city of Atlanta might want to implement after seeing how foreclosures affect crime rates. I will provide them with a proposal on a plan to keep foreclosure as the last possible option, and if they do end up foreclosing there will be programs available to start life over and think positively. Another proposal may be to implement a larger police force during economic downturns where the number of foreclosures increases. By doing so I believe that the city of Atlanta will have an easier time on trying to cracking down on crime. My studies have shown that the rate of unemployment increases during periods of recession, so one way to combat this loss of jobs is to train and hire some of these unemployed individuals and create neighborhood patrol officers. This would be a great opportunity to increase the police force to combat with the higher crime rates that usually rise due to the number of foreclosures and unemployed people in the area.

Empirical Economic Bulletin, Spring 2009, Vol. 2

146

References Cerro, A. M. & Meloni, O., 2000. “Determinants of the crime rate in Argentina during the '90s”, University of Chile, Department of Economics. Volume 27, Pages 297-311. http://econ.uchile.cl/public/Archivos/pub/6a59c032-d7eb-40bd-a86d-b3522a2507d5.pdf Immergluck, D., and G. Smith, 2006. The external costs of foreclosure. The impact of single-family mortgage foreclosure on property values. Housing Policy Debate 17:57-80. Steven Raphael & Rudolf Winter-Ebmer, 1998. "Identifying the Effect of Unemployment on Crime," University of California at San Diego, Economics Working Paper Series 98-19, Department of Economics, UC San Diego. U.S. Department of Justice, Bureau of Justice Statistics, Key Crime and Justice Facts at a Glance. (2008) - http://www.hoover.org/research/factsonpolicy/facts/8112882.html U.S. Department of Justice, Bureau of Justice Statistics, Key Crime and Justice Facts at a Glance. (2008) - http://www.hoover.org/research/factsonpolicy/facts/8112882.html Yamada, T., 1985 “The Crime Rate and the Condition of the Labor Market: A Vector Autoregressive Model.” National Bureau of Economic Research

Empirical Economic Bulletin, Spring 2009, Vol. 2

147

The Effect of Aid Dependency and Quality of Institution in alleviating poverty in IDA countries Mahah Mirzaa

Abstract:

Poverty alleviation has been a topic of much discussion ever since the Millennium Development Goals were set by World Bank in the 1990s. This paper examines to what extent aid dependency and quality of institution affect 31 IDA countries in achieving the first of eight MDG. The study provides evidence that although these IDA countries experience a decrease in poverty, it may not be enough to meet MDGs by 2015. Aid dependency increases the poverty level whereas some Quality of Institution indicators such as control of corruption, rule of law and regulatory quality have greater impact in reducing poverty than other indicators.

JEL Classification: E02, F34, I32, M38 Keywords: Aid Dependency, Measurement and Analysis of Poverty, Government Effectiveness, IDA, Institutions a

Bryant University, Class of 2010, 1150 Douglas Pike, Smithfield, RI 02917, USA

Phone: (401) 719-8820. Email: mmirza@bryant.edu.

The author thanks Dr. Ramesh Mohan for his continuous support and guidance throughout the research and compilation of this paper. The author would also like to thank Mr. Michael Paolino and Mr. Andrew Stone for their feedback, comments and criticism that helped the author in improving this paper.

Empirical Economic Bulletin, Spring 2009, Vol. 2

148

1.0 INTRODUCTION Poverty has been an issue of much concern especially in the late 90s when the millennium development goals were set. World Bank has classified extreme poverty as those who live on an income of $1.25 per day1 in 2005 prices. As of 2005, 1.4 billion people live in extreme poverty (Poverty Analysis, 2009). Countries affiliated World Bank and IMF have since worked towards achieving this goal. As of 2008, World Bank reports that they have underestimated the number of people living in extreme poverty (World Bank Press Release 2008). As a result, many studies have been conducted to analyze the effect to what extent economic growth has succeeded in alleviating poverty. This paper explores the effect that aid dependency and quality of institution has had on poverty alleviation and concentrates on 31 IDA countries between the years 2000 and 2007. The countries used in this study are listed in Appendix C. This research was written with a particular focus on IDA countries. World Bank has a separate organization called IDA, International Development Association, which deals with the “world’s poorest countries and helps them attain interest-free loans and grants for programs that boost economic growth, reduce inequalities and improve people’s living conditions” (What is IDA?, 2009). IDA countries face the worst of extreme poverty, and as a result these countries became the focal point of this study. As of 2008, there are 168 such IDA members; however, due to unavailability of data some countries were excluded from this study. The rest of the paper is organized as follows: Section 2 summarizes the current trends in poverty in IDA countries based on World Development Report and similar studies. Section 3 outlines the empirical model and describes its variables. The results of this study are discussed in section 4 which is followed by conclusion and policy recommendation in section 5.

Poverty Analysis, (2009). Overview - Understanding Poverty, The World Bank Group, 2009

Empirical Economic Bulletin, Spring 2009, Vol. 2

149

2.0 TRENDS According to World Bank, “living standards have risen dramatically over the last decades ...but wide regional disparities persist” (Poverty Analysis, 2009). The World Bank website reports that the proportion of people living in extreme economic poverty has dropped from 52 percent in 1981 to 26 percent in 2005. Although this statistic may seem encouraging, in Sub-Saharan Africa over the same time period, “in absolute terms, the number of poor people has nearly doubled,” from 200 million to 380 million (Poverty Analysis, 2009). In the following chart World Bank provides data on where the world stands in terms of the percent of people living below $1 or $2 a day in 2004 and where the world is projected to stand in 2015. Figure 1: Share of People living on less than $1 or $2 a day in 2004, and projection for 2015

Source: World Development Indicators As of 2004, Latin America and Caribbean, Europe and Central Asia, Middle East and North Africa and South Asia are on target. “In middle-income countries, the median

Empirical Economic Bulletin, Spring 2009, Vol. 2

150

poverty line for the developing world—$2 a day in 2005 prices—is more relevant. By this standard, the poverty rate has fallen since 1981 in Latin America and the Middle East & North Africa, but not enough to reduce the total number of poor” (Poverty Analysis, 2009) . The East Asia and Pacific region has made remarkable progress in halving their percent of people living in poverty over 14 years. It is expected that by 2015, only 2.8% percent of this region will be living below the poverty line. South Asia and Sub-Saharan Africa had started with similar percentages of people living below the poverty line, at 43% and 46.7% respectively. In 2004, South Asia has managed to stay on track at 30.8%, but the same is not true for Sub-Saharan Africa region. In a span of 14 years, this region has only managed to bring 5.6 percent of the population below the poverty level. World Bank expects that this region will be 8% short of its original goal of 23.4% in 2015. It is difficult to predict if Sub-Saharan Africa will continue to experience a fall in the percent of people living below the poverty line given its poor track record at reducing poverty, especially compared to the rest of the world. Figure 2: Top 10 recipients of the net increase in net ODA, 2002-06

Source: DAC database and staff estimates

The figure above shows that seven of the top ten aid recipients between 2002 and 2006 are African countries. In addition, “Donors did make encouraging commitments to IDA ($25.1 billion for 2008-2011), as well as to the concessional windows of other regional development banks and the Global Fund for AIDS, TB and Malaria (GFATM)” (Global Monitoring Report, 2008). Furthermore, “Aid for health has scaled up

Empirical Economic Bulletin, Spring 2009, Vol. 2

151

dramatically with support from over 100 traditional and non-traditional entities amounting from $6.8 billion in 2000 to nearly $17 billion in 2006.” The trends for Quality of Institutions show that “Developing countries…have made progress in strengthening development strategies and institutional frameworks for implementation” (Global Monitoring Report, 2008). When countries perform well in a sound environment, they receive scaled-up aid from IDA. However, Global Monitoring Report states that in recent years, a significant portion of the increase in ODA has been concentrated in a few countries.

2.1 Literature Review Economic literature on Aid, Policy Implication, Growth and Poverty Reduction show that on average, aid has little impact on growth. Collier and Dollar (2001) concludes that “actual allocation of aid is radically different from poverty-efficient allocation” of aid. However, a study conducted by Burnside and Dollar (2000) shows that aid has a positive impact on growth in a good policy environment. This also shows that the quality of government policy, institution, transparency, accountability and control all come into play in determining the effectiveness of aid. Collier and Dollar (2001) in agrees with Burnside and Dollar (2000) that the allocation of aid that has maximum effect on poverty depending on the level of poverty and the quality of policies. Easterly (2003) has an interesting take on the Burnside and Dollar model. He supports but Burnside and Dollar paper but says that economic growth depends on investment as a share of GDP, and therefore determines the quality of the investment – similar approach to that of Burnside, Dollar and collier. In a separate study Barro (1991) finds that political instability – measure of quality of governance, is inversely related to growth and investment. Based on his findings, Barro speculates that the extreme poverty and below-average growth rate conditions in Sub-Saharan Africa can be attributed to poor governance. Knack (2000) states that the amount of aid is a determinant of the quality of governance. According to Knack, “higher aid levels erode the quality of governance” through poor control of corruption and the rule of law.

Empirical Economic Bulletin, Spring 2009, Vol. 2

152

3.0 DATA AND EMPIRICAL METHODOLOGY

3.1 Definition of Variables The model used in this study is based on a Collier-Dollar’s simple growth model that explains the relationship between growth and aid dependency and government policies:

(1) In this equation by Collier-Dollar, Growth, G, is modeled as a function of five independent variables: exogenous conditions (X), the level of policy (P), the level of aid dependency relative to GDP (A), the level of aid squared (A2) and the interaction of policy and aid. The equation above has been adjusted to take extreme poverty, aid dependency and quality of institution into account.

(2)

The dependent viable in the equation, Pov, is the log of poverty (percent of population living below the poverty line) and is expressed as a function of the independent variables: GDP per capita growth, inflation (GDP deflator), Age dependency ratio (dependents to working-age population), Aid (as a percent of GNI) and Quality of Institution, QOI, for which CPIA transparency, accountability, and corruption in the public sector rating (1=low to 6=high) and World Governance Indicators: Voice and Accountability, Political Stability, Government Effectiveness, Regulatory Quality, Rule of Law, Control of Corruption have been used.

3.2 Data The data for this study was obtained from World Development Indicators and World Governance Indicators which is publicly available on World Bank’s website. The seven Quality of Institution variables are run separately in seven equations to avoid any

Empirical Economic Bulletin, Spring 2009, Vol. 2

153

error of correlation. The seven variables, defined by World Governance Indicators, are presented in Appendix A. It must be mentioned here that the data is not completely comparable from country to country. Often due to unavailability of data the number of countries has been cut short from 168 to 31, as mentioned earlier. Of the 31 countries, 18 countries are from the Sub-Saharan African region, three from each of the following regions: South Asia, East Asia & Pacific, Europe & Central Asia, and Latin America & Caribbean regions, and one from the Middle East & North Africa region. Table 1 below shows the effects of GDP, Inflation, Age Dependency Aid dependency and quality of institutions on poverty. The significance level of each variable is expressed in parenthesis. Results that have significance level of 90% or better are indicated by: *, 95% or better with ** and 97.5% or better with ***.

Empirical Economic Bulletin, Spring 2009, Vol. 2

154

Table 1 Effects of Aid Dependency and Quality of Institutions Equation

Constant

1.5698 (0.0000)***

1.3674 (0.0000)***

1.3463 (0.0000)***

1.3574 (0.0000)***

1.4161 (0.0000)***

1.2767 (0.0000)***

1.3074 (0.0000)***

-0.0390 (0.0103)***

-0.0419 (0.0057)***

-0.0475 (0.0019)***

-0.04481 (0.0031)***

-0.4335 (0.0083)***

-0.4373 (0.0023)***

-0.0381 (0.0111)***

0.0016 (0.3159)

0.0026 (0.0865)*

0.0030 (0.0455)**

0.0021 (0.1567)

0.0025 (0.1153)

0.0033 (0.0242)***

0.0033 (0.0354)**

0.2965 (0.1139)

0.2346 (0.2048)

0.2201 ((0.2239)

0.2916 (0.1172)

0.2532 (0.2124)

0.2767 (0.1131)

0.2608 (0.1516)

0.0083 (0.0225)***

0.0091 (0.0174)***

0.0095 (0.0117)***

0.0077 (0.0326)**

0.0077 (0.0427)**

0.0092 (0.0095)***

0.0080 (0.0261)**

GDP per capita growth (%)

Inflation, GDP deflator

Age Dependency Ratio

Aid (as a % of GNI)

CPIA

Voice and Accountability

-0.0702 (0.1428)

-0.057 (0.1886)

Control of Corruption

Political Stability

Government Effectiveness

Rule of Law

Regulatory Quality

Empirical Economic Bulletin, Spring 2009, Vol. 2

-0.1043 (0.0895)*

-0.0471 (0.1276)

-0.0008 (0.9577)

-0.1148 (0.0250)*** -0.1207 (0.0982)*

155

Significance Level: 90% or better *

Empirical Economic Bulletin, Spring 2009, Vol. 2

95% or better **

97.5% or better ***

156

4.0 EMPIRICAL RESULTS The results of all seven regression equations show that there is a consistent effect of independent variables on the dependent variable. The results reveal that increase in GDP and higher quality of institution tends to reduce poverty level whereas higher inflation, age dependency ratio and aid dependency tends to increase poverty level. The effect of each variable matches the expected effect on the dependent variable. The expected signs of the regression model are indicated in Appendix B. The independent variable GDP per capita growth shows with very high significance level, 97.5% or better, that poverty level can be reduced through higher GDP. Aid dependency also shows some statistically significant results, 95% or better in terms of its effect on poverty. Of all the regressions run, the Quality of Institutions variables: Control of Corruption, Rule of Law and Regulatory Quality are statistically significant with significance levels of 90% or better and 97.5%. On the other hand, the governance indicator Government Effectiveness has almost no significant impact on the poverty level. Each of the seven regression equation show, with 95% significance level or better, that these IDA countries experience an increase in their poverty level when they increase their aid dependency. Given that 18 of the 31 countries used in this study are Sub-Saharan African countries, the results are likely to reflect the effect of Sub-Saharan countries compared to other regions. See Appendix C.

Empirical Economic Bulletin, Spring 2009, Vol. 2

157

5.0 CONCLUSION AND POLICY RECOMMENDATION It is evident from the results that aid dependency does very little to decrease poverty. High levels of aid, from foreign donors or even their governments, do not necessarily ensure that the lives of the poor improve. This may be because aid does not trickle down to the extreme poor, but is rather consumed along the bureaucratic process. This raises the question: Should World Bank, IMF, OECD countries stop pouring billions of dollars into developing countries, particularly Sub-Saharan African countries? Through effective rule of law and control of corruption these IDA countries do experience a decrease in their poverty level. Results from this study and trends from other studies show that even though the poverty level has decreased, the Sub-Saharan African countries are still lagging behind. In order to meet the MDGs by 2015, countries in Sub-Saharan Africa must engage in transparency in their government policies and ensure political stability. It can, therefore, be suggested that World Bank work closely with Transparency International in setting stricter guidelines to qualify for financial aid. However, aid is making these IDA countries very dependent on donors. World Bank should also think of ways to slowly pull out of providing aid to the IDA countries, but this has to be done carefully otherwise the IDA countries will be severely hit if all donors pull out all at once. Sub-Saharan Africa can also opt to follow policies implemented in South Asia. It was noted earlier that both Sub-Saharan Africa and South Asia began with similar percentages of people living below the poverty line in 1990. South Asia has seen their poverty level decline. Sub-Saharan Africa should follow South Asia’s foot-steps to lower their percent of people living below the poverty line. By closely monitoring their quality of institution, in addition to generating high levels of GDP it may be possible for Sub-Saharan Africa to cut down its poverty level by 2015 to meet MDG. It is also important to note that these conclusions and policy recommendations are based on the findings of this research. It would be possible to draw better conclusions about SubSaharan

African

Empirical Economic Bulletin, Spring 2009, Vol. 2

countries

could

included

the

research.

158

Appendix A: Acronym, Description and Data Source Acronym

Description

International Development Association – IDA organization that deals with the world’s poorest countries and helps them attain interest-free loans and grants for programs that boost economic growth, reduce inequalities and improve people’s living conditions. Millennium Development Goals - an eightMDG point road map with measurable targets and clear deadlines for improving the lives of the world’s poorest people that 189 world leaders promised at the United Nations Millennium Summit in 2000 to achieve by 2015. Country Policy and Institutional Assessment – CPIA means of rating countries against a set of 16 criteria grouped in four clusters: (a) economic management; (b) structural policies; (c) policies for social inclusion and equity; and (d) public sector management and institutions. Voice and refers to the extent to which citizens of a Accountability country are able to participate in the selection of governments, as well as freedom of expression and association in the media. Control of measures the extent to which public power is Corruption exercised for private gain, including petty and grand forms of corruption. Regulatory refers to the ability of the government to Quality formulate and implement sound policies. measures the extent to which agents have Rule of Law confidence in and abide by the rules of society, in particular the quality of police and courts. Government measures the quality of public services, the Effectiveness quality of the civil service and the degree of its independence from political pressures. Political measures the perceptions of the likelihood that Stability the government will be destabilized or overthrown by unconstitutional or violent means, including domestic violence and terrorism

Empirical Economic Bulletin, Spring 2009, Vol. 2

Data source

World Bank

End Poverty 2015 Millennium Campaign

World Bank

World Governance Indicators

World Governance Indicators World Governance Indicators World Governance Indicators World Governance Indicators World Governance Indicators

159

Appendix B- Variables and Expected Signs

Acronym

Variable Description

Expected sign (+/-)

(Dependent Variable)

Pov

Log of % Population living below the poverty line

GDP

GDP per capita growth

Inflation

GDP deflator – measure of prices

all

domestically

(-)

(+)

new

produced

goods and services in an economy1

AgeDep

Age Dependency Ratio

(+)

Aid

Aid (as a percent of GNI)

(-)

QOI

Quality of Intitution

(-)

Bureau of Economic Analysis

Empirical Economic Bulletin, Spring 2009, Vol. 2

160

Appendix C: Countries used in this study

Bangladesh Benin Burkina Faso Cambodia Cameroon Cape Verde Cote d'Ivoire Guinea Haiti Honduras Kyrgyz Republic Lao PDR Madagascar Malawi Mali Mauritania Moldova Mozambique Nepal Nicaragua Nigeria Rwanda Senegal Sierra Leone Sri Lanka Tajikistan Tanzania Uganda Vietnam Yemen, Rep. Zambia South Asia: 3 Sub-Saharan Africa: 18 East Asia and Pacific: 3 Europe and Central Asia: 3 Latin America and Caribbean: 3 Middle East and North Africa: 1

Empirical Economic Bulletin, Spring 2009, Vol. 2

161

BIBLIOGRAPHY Barro, Robert (1991).Economic Growth in a Cross Section of Countries. The Quarterly Journal of Economics. 407-443.

Burnside, Dollar, Craig, David (1997).Aid Policies and Growth. Policy Research Working Paper Series. 1777.

Collier, Dollar, Paul, David (2001).Can the World Cut Poverty in Half? How Policy Reform and Effective Aid Can Meet International Development Goals. World Development. 29, 1787-1802.

Collier, Dollar, Paul, David (2002).Aid Allocation and Poverty Reduction. European Economic Review. 46, 1475-1500.

Easterly, William (2003).Can Foreign Aid Buy Growth?. Journal of Economic Perspectives. 17, 23-48.

Knack, Stephen (2000).Aid Dependence and the Quality of Governance: A Cross-Country Empirical Analysis. Policy Research Working Paper Series. 2396.

Poverty Analysis, (2009). Overview - Understanding Poverty. Retrieved April 15, 2009, from World bank Web site: http://web.worldbank.org/WBSITE/EXTERNAL/TOPICS/EXTPOVERTY/EXTPA/0,,contentM DK:20153855~menuPK:435040~pagePK:148956~piPK:216618~theSitePK:430367,00.html

Press Release, (2008, August 26). World Bank Updates Poverty Estimates for the Developing World. Retrieved April 15, 2009, from World Bank Web site: http://econ.worldbank.org/WBSITE/EXTERNAL/EXTDEC/EXTRESEARCH/0,,contentMDK:2 1882162~pagePK:64165401~piPK:64165026~theSitePK:469382,00.html

Empirical Economic Bulletin, Spring 2009, Vol. 2

162

The World Bank Group, (2009). Fact Sheet: Scaling Up Aid: Opportunities and Challenges in a Changing Aid Architecture . Retrieved April 29, 2009, from Global Monitoring Report 2008: Fact Sheet - Scaling Up Aid: Opportunities and Challenges in a Changing Aid Architecture Web site: http://web.worldbank.org/WBSITE/EXTERNAL/EXTDEC/EXTGLOBALMONITOR/EXTGL OMONREP2008/0,,contentMDK:21710781~pagePK:64168445~piPK:64168309~theSitePK:473 8057,00.html

The World Bank Group, (2009). Governance Matters 2008, Worldwide Governance Indicators 1996-2007. Retrieved April 15, 2009, from Governance & Anti-corruption Web site: http://info.worldbank.org/governance/wgi/index.asp

The World Bank Group, (2009). International Development Association. Retrieved April 15, 2009, from IDA - What is IDA? Web site: http://web.worldbank.org/WBSITE/EXTERNAL/EXTABOUTUS/IDA/0,,contentMDK:212067 04~menuPK:83991~pagePK:51236175~piPK:437394~theSitePK:73154,00.html World Bank, World Bank Governance Indicators. Retrieved April 29, 2009, from World Bank Web site: http://www.worldbank.org/wbi/governance/pdf/Governance_Indicators_eng.pdf

Empirical Economic Bulletin, Spring 2009, Vol. 2

163

The Effect of Domestic Investment, Economic Growth and Human Development on Foreign Direct Investment into China

Michael Paolino1

Abstract This paper examines the relationship between foreign direct investment, domestic investment, human development, and economic growth, and uses the ordinary least squared technique, and a time series analysis using data for the period 1977-2007. The analysis seeks to answer the fundamental question of what impact Chinese domestic investment, economic growth, and human development has on foreign direct investment into the country. Research regarding foreign direct investment and its fundamental correlation with economic growth and domestic investment has produced varied results. This paper examines those variables, with the addition of human development variables, where research is partial. This paper seeks to expand on current research by examining the effects domestic investment, economic growth, and human development factors have on foreign direct investment into China. The results of the study indicate that when domestic investment and economic growth in China are increasing, foreign direct investment is attracted to the country. The human development variable that had the greatest impact on the amount of foreign direct investment into China was the literacy rate.

JEL Classification: P33 E24 O10 O11 O15 Keywords: Foreign Direct Investment, FDI, Economic Growth, Human Development, China

Bryant University, 1150 Douglas Pike, Smithfield, RI 02917. Phone: (401) 226-3748. Email: mpaolino@bryant.edu. 1

Empirical Economic Bulletin, Spring 2009, Vol. 2

164

1.0 INTRODUCTION China is one of the world’s fastest growing economies, and has attracted a large amount of foreign direct investment (FDI) over the last 20 years, leading developing countries as the largest recipient of FDI (Tang et al., 2008). Several factors of the Chinese economy and lifestyle can help to explain the substantial levels of FDI China has been experiencing. This paper examines the effects of economic growth and domestic investment on the amount of FDI China has received. In addition, this article also examines several human development variables that also seek to explain the high levels of FDI China has experienced over the last several decades. The goal of this paper is to enhance understanding of the effects the Chinese economy and its people have on levels of FDI into the country. Various studies have examined the effects of FDI and economic growth. However, this paper observes not only economic factors regarding levels of FDI, but human development issues as well, where research is limited. China was chosen for this research since it has been one of the world’s fastest growing economies, and as a result received considerable levels of FDI. Therefore, the effects of domestic investment, economic growth and human development should be significant and help in understanding the correlation between these variables and FDI. From a policy perspective, it is crucial to understand the effect domestic activity, both economic and social, has on FDI. Since China has received considerable levels of FDI, examining domestic economic activity and human development in China should yield interesting results. High levels of FDI can help bring technology to the host country. China’s high levels of FDI could lead to domestic technological spillover, that is, foreign companies will bring new technology and train Chinese workers, improving the Chinese economy. The Solow model suggests technological growth is a fundamental factor in increasing long run economic growth. Therefore, it is clear that FDI into China will have a positive impact on the domestic economy. However, what factors attracted these high levels of FDI in the first place? This study examines the three factors that are believed to answer this question, domestic investment, economic growth and human development. Chinese domestic investment is the first variable thought to contribute to the amount of FDI the country attracts. If Chinese banks are investing heavily in domestic economic ventures,

Empirical Economic Bulletin, Spring 2009, Vol. 2

165

it is safe to assume there are many profitable businesses fulfilling Chinese demand. Domestic investment is measured by domestic credit provided by the banking sector, and domestic credit to the private sector. High levels of domestic investment indicate that individuals are opening new businesses and investing in their businesses. A high-quality domestic business environment should have an impact on the amount of FDI that flows into the country. Also, economic growth is examined when attempting to explain what brings FDI into China. Economic growth is measured by GDP per capita growth. When foreign businesses look to invest in China, they undeniably inspect how economic growth is in the country. Economic growth indicates that businesses in the country are doing well, and is a place businessmen may invest or start a business. In addition to understanding the correlation between FDI, domestic investment and economic growth, it is imperative, from a policy point of view, to be aware of the impact human development variables have on FDI into China. Investigating human development factors and their correlation to FDI is where this research paper differs from similar studies. Numerous studies have examined, with varying results, the impact of economic growth and FDI. Previous research has also found the relationship between domestic investment and FDI. However, this study looks at these factors, as well as human development variables and their relation to FDI into China. Human development in this study is measured by three key variables, infant mortality rate, life expectancy, and education (measured using literacy rate). Therefore, this paper seeks to answer several fundamental questions that differ from previous studies. The first is to conclude whether economic growth in China has an impact on FDI into the country. Second, what has been the effect of high levels of domestic investment on the amount of FDI into the country? Lastly, this study seeks to investigate whether human development factors in China affect the amount of FDI that flows into the country. The remainder of this article is organized as follows. Section 2 investigates trends in Chinese FDI, as well as historical economic growth, domestic investment, and human development. Section 3 is a literature review, and summarizes previous studies in this area of research. Section 4 discusses the empirical methodology and data used in this study. Section 5 reviews the empirical results of the paper, and section 6 contains a conclusion, and offers insight into potential policy implications based on the results of this study.

Empirical Economic Bulletin, Spring 2009, Vol. 2

166

2.0 TRENDS Attracting FDI has been a key pillar of China’s ‘opening up’ polices and economic reforms (Tang et al., 2008). Figures 1 and 2 show FDI in China and domestic investment and GDP for the years 1978-2003, respectively. At China’s initial ‘opening up’ period, inflows of FDI were low. FDI varied from .05 billion Chinese yuan in 1983 to 1.3 billion Chinese yuan in 1984. From 1984 until the early 1990s, FDI increased at an average rate of approximately 30 percent each year. However, the total amount of FDI was still small and remained as low as 40 billion Chinese yuan until 1992. In 1992 Chinese leader Deng Xiaoping made his famous ‘southern tour,’ where he promoted his economic reforms, including developments that led to FDI liberalization. During the 1997 Asian financial crisis, the Chinese government further liberalized their FDI policy. For example, the Chinese government eliminated the FDI project approval requirement. China joined the World Trade Organization (WTO) in 2001, and this marked a new era of FDI liberalization. As figure 1 demonstrates, China’s FDI inflows increased dramatically from the year 2000 to 2001, moving from 337 billion Chinese yuan to 388 billion Chinese yuan. This catapulted China into becoming the largest FDI host country in the world, attracting 437 billion Chinese yuan of FDI in subsequent years (Tang et al., 2008). Domestic investment and GDP in China, shown in figure 2, display similar trends to that of FDI for the same time period. Economic growth in China has increased exponentially since the major reforms of 1978, growing at an average annual rate of 9 percent (Tang et al., 2008). Compared to the rest of the world, China has shown considerable economic growth. The parallel growth of GDP and FDI can not be ignored. China’s international trade increased considerably from 36 billion Chinese yuan in 1978, to 5,138 billion Chinese yuan in 2002. These trends suggest that such a substantial increase in international trade is associated with large inflows of FDI. Similar to the rapid growth in GDP and FDI, China’s domestic investment shows a significant increase, growing at an average rate of 20 percent from 1978 to 2003 (Tang et al., 2008).

Empirical Economic Bulletin, Spring 2009, Vol. 2

167

Figure 1: Trends in Chinese FDI

Figure 2: Trends in Chinese GDP and Domestic Investment

Source: China Statistical Yearbook Figure 3 graphs the growth rates of FDI, domestic investment and GDP in China during 1978-2003. FDI growth reached several peaks. In 1980, FDI growth reached 168 percent and in 1983 it reached 132 percent. In 1991 and 1992, Chinas FDI growth rate reached 161 percent. The growth rate of domestic investment peaked in the same or subsequent years as FDI, however, well below the FDI growth rate. GDP showed a similar growth trend to FDI and domestic investment, reaching its peaks in 1984, 1987 and 1993. Figure 3: Growth Rates of FDI, DI and GDP 1978-2003

Source: China Statistical Yearbook Generally, the three figures presented demonstrate that both FDI and domestic investment display an upward trend, matching the economic growth trend of GDP during the period 1978 to 2003 (Tang et al., 2008). Although trends in China’s FDI, GDP and domestic investment show a

Empirical Economic Bulletin, Spring 2009, Vol. 2

168

strong positive relationship, this paper seeks to verify empirically that there is a strong underlying relationship connecting these variables. In addition to the FDI and economic trends in China over the last several decades, this paper will also observe trends in human development. The three human development factors that this paper examines are life expectancy, infant mortality rate, and literacy rate, a proxy for education. The Human Development Report publishes the human development index (HDI) which looks beyond GDP to a broader definition of well-being. According to their website, the index “…is not in any sense a comprehensive measure of human development. It does not, for example, include important indicators such as gender or income inequality and more difficult to measure indicators like respect for human rights and political freedoms. What it does provide is a broadened prism for viewing human progress and the complex relationship between income and well-being” (Human Development Report, 2007). This broad look at human development is sufficient to investigate the trends in human development in China over the period being examined. Figure 4 shows China’s HDI since 1975. As can be seen, China was consistent with other Eastern Asian countries until 2002 where it passed the average East Asian countries HDI. Although China’s HDI recently surpassed other Eastern Asian countries, it still lags significantly behind the trends of European and OECD countries. Chinese growth rate of HDI, however, seems to be greater than most other countries examined. Understanding trends in HDI over the last several decades is important because this paper is investigating the impact of human development factors on FDI into China. Overall, China is middle of the road in terms of human development measured by the HDI when compared to other countries around the world. Among richer European and OECD countries, however, China is still lagging behind, although it is growing at a faster rate than most other countries.

Empirical Economic Bulletin, Spring 2009, Vol. 2

169

Figure 4: China’s Human Devlopment Index

Source: Human Development Report, 2007 In addition to the Human Development Index, the trends in the variables used in this paper are examined in the charts below. Figure 5 is a graph of Chinas literacy rate since 1977. The data shows that the literacy rate (a proxy for education) has increased from 70 percent of the population being literate in the 1970’s, to over 90 percent literacy rate in the late 1990’s and 2000’s. This indicates that the population, in general, is becoming more educated. Figure 6 shows the life expectancy at birth of Chinese citizens. Life expectancy was as low as 65 years in 1977. The graph shows that life expectancy has increased over the last several decades, and has reached 72 years old, slightly below the United States average, as of 2007. Finally, figure 7 shows the infant mortality rate in China since 1977. In 1977, as many as 45 infants per 1,000 live births were dying. As of 2007, that number was as low as 21 infant deaths per 1,000 live births. This significant decrease in the infant mortality rate indicates China’s medical advancements and improvements in child care. Understanding the trends in these three variables is important to appreciate the results of this study. Since the late 1970’s China has made significant advancements that have increased their life expectancy, increased education in the country, and decreased the infant mortality rate. These are factors that investors interested in starting a business in China may look at before doing so. These investors will see noteworthy

Empirical Economic Bulletin, Spring 2009, Vol. 2

170

advancement in Chinese human development factors over the last several decades. However, have these improvements in Chinese standards of living, as well as increases in economic growth and domestic investment lead to more FDI into the country? Whether these three factor influence the level of FDI into China is the fundamental question this paper seeks to answer, therefore, it is crucial to understand the trends in these variables over the time being studied. Figure 5: Literacy rate, adult total (% of people ages 15 and above) Literacy Rate 100.00 90.00

Literacy Rate (%)

80.00 70.00 60.00 50.00 40.00 30.00 20.00 10.00 0.00

Year

Figure 6: Life expectancy at birth, total (years) Life Expectancy

74.00

Life Expectancy (years)

72.00 70.00 68.00 66.00 64.00 62.00 60.00

Year

Empirical Economic Bulletin, Spring 2009, Vol. 2

171

Figure 7: Mortality rate, infant (per 1,000 live births)

2007

2005

2001

1999

1997

1995

1993

1991

1989

1987

1985

1983

1981

1979

50.00 45.00 40.00 35.00 30.00 25.00 20.00 15.00 10.00 5.00 0.00 1977

Infant Mortality Rate (per 1,000 live births)

Infant Mortality Rate

Year

Source: These three graphs are authors’ compilation using World Development Indicators data

3.0 LITERATURE REVIEW There has been much research done with the goal of examining the effects foreign direct investment has on a host country. This paper, however, seeks to explain the factors that contribute to foreign direct investment initially. That is, what are the main factors that cause international businesses to invest in a foreign country? Research in this area is limited, which is why this paper seeks to expand on current text. There are, however, several key papers that make significant contributions to the research topic discussed in this paper. First, Fedderke and Romm (2006) have conducted similar research to what is examined in this study, except they observed South Africa as opposed to China. They used data from 1956-2003, and find that, “Reducing political risk, ensuring property rights, most importantly bolstering growth in the market size, as well as wage moderation, lowering corporate tax rates, and ensuring full integration of the South African economy into the world economy all follow as policy prescriptions from our empirical findings.” They find that growths in market size as well as integration into the world economy are both important factors for determining levels of FDI

Empirical Economic Bulletin, Spring 2009, Vol. 2

172

into South Africa. Since they are examining South Africa, their determinants vary slightly from this study. China is clearly a globalized nation, therefore, this factor may be less important in China, yet may be crucially important when establishing determinants of FDI in South Africa. Shah’abadie and Mahmudie (2006) find that domestic investment and economic growth, as well as human capital, are key factors in determining FDI. They state that, “The results of the studies approved that FDI depends on…domestic investment…economic growth…and human capital,” and these factors “…have a direct and positive impact on FDI in Iran.” Their study focuses on Iran, using data from 1959-2003. Uppenberg and Riess (2004) present a dilemma regarding FDI and domestic economic growth, which they refer to as the growth-FDI nexus. They state that “…while a strong positive correlation between inward FDI and economic growth exists…it is not clear whether the causality runs from FDI to growth or vice versa…and growth-enhancing policies in general are more promising than specific support for FDI.” They are able to conclude that economic growth in general is a more important determinant of FDI than specific policy strategies attempting to boost FDI. While their study examines Europe, their findings regarding the growth-FDI nexus are vital. Uppenberg and Riess are able to confirm empirically that domestic economic growth is a key variable in determining factors influencing FDI inflows into a country. Herrmann and Gast (2008) find a complementary relationship between investment and trade. However, since they use relatively recent data, and also observe OECD countries, it is difficult to make the parallel between their findings and the finding of this paper. However, their contribution is important because the research in this paper did not include trade factors, which Herrmann and Gast have shown are important factors for FDI inflows, at least for OECD countries from 1991-2001. Rodriguez and Pallas (2008) find that human capital plays an important role in determining FDI. They examine Spain, using data from 1993-2002, and are able to conclude that “…the evolution of human capital…play[s] a very important role in attracting flows of FDI.” Rodriguez and Pallas’s contribution is important because they confirm that human capital factors play an important role in attracting FDI in Spain. Hong (2008) finds interesting and perhaps contradictory results regarding FDI determinants in China. Hong uses an 11-year panel dataset on FDI and urban characteristics across Chinese cities in his paper, and he finds that “Cheap labor plays an increasingly important

Empirical Economic Bulletin, Spring 2009, Vol. 2

173

role [in attracting FDI], but labor quality…lose[s] [its] significance.” In other words, cheap labor is more important than quality of labor. This is interesting because the findings in this paper, discussed later, are contradictory to Hong’s conclusions that quality of labor is not as important as the price of labor. Ang (2008) suggests that real GDP is found to have a significant positive impact on FDI inflows, and there is evidence that growth rate of GDP exerts a small positive impact on inward FDI. Ang’s findings are significant because this paper uses GDP per capita growth as a variable for determining if domestic economic growth has an effect on FDI inflows into China. Although Ang is using data from 1960-2005, and also examining Malaysia, his findings that real GDP has a greater effect on FDI inflows than the growth rate of GDP does, are noteworthy. Perhaps the results of the research in this paper could be strengthened if real GDP was used as opposed to the growth rate of GDP. Although, since Ang is studying Malaysia, it is possible GDP growth is in fact a better choice of variable. Either way, Ang’s findings and variable choice is worth mentioning. In addition, he offers several policy implications regarding FDI inflows, which are discussed in section 6.0. Additional research on this topic done by Naude and Krugell (2007) suggests that geography does not play a significant role in determining levels of FDI. In their paper, they find that geography does not seem to have a direct influence on FDI flows to Africa. Since a variable for geography was not included in this paper, it is important to justify this by pointing to other papers that have concluded geography as not a significant determinant of FDI inflows. Overall, recent research in the area of investigating determinants of FDI inflows matches the method and findings of this paper closely. With a concrete understanding of past literature on the topic, the conclusions and contributions of this paper will become more evident.

4.0 DATA AND EMPIRICAL METHODOLOGY 4.1 Definition of Variables This paper uses annual data from 1977 to 2007 provided by world development indicators online. Each variable consists of thirty-one observations. Summary statistics are

Empirical Economic Bulletin, Spring 2009, Vol. 2

174

provided in table 1. For variable description and data source, refer to appendix A, and for expected signs refer to appendix B. The model for this paper is based on a simplified version used by Fedderke and Romm (2006), with the omission and addition of variables thought to be specific to China, namely human development variables, where research is limited. Table 1: Summary Statistics Standard Variable

Observations

Mean

Minimum

Maximum

Deviation FDI

2.23

1.92

0.10

6.20

GDP

8.58

2.79

2.30

13.70

68.96

1.81

65.30

72.00

36.07

7.78

22.12

47.00

80.02

9.78

70.00

90.90

DCP

112.42

145.95

51.00

888.00

DCB

101.97

71.90

38.00

452.00

FDI = β0 + β1GDP + β2LE + β3MR + β4LR + β5DCP + β6DCB + E Where: FDI = Foreign direct investment, net inflows (% of GDP) GDP = GDP per capita growth (annual %) LE = Life expectancy at birth, total (years) MR = Mortality rate, infant (per 1,000 live births) LR = Literacy rate, adult total (% of people ages 15 and above) DCP = Domestic credit to private sector (% of GDP) DCB = Domestic credit provided by banking sector (% of GDP) Foreign direct investment, this papers dependent variable, is measured in terms of net inflows, as a percent of gross domestic product. Foreign direct investment is the net inflows of investment to acquire a lasting management interest in an enterprise operating in an economy

Empirical Economic Bulletin, Spring 2009, Vol. 2

175

other than that of the investor. It is the sum of equity capital, reinvestment of earnings, other long-term capital, and short-term capital as shown in the balance of payments. The FDI figures used in this study are net inflows in the reporting economy, divided by GDP. This study incorporates six independent variables, GDP per capita, life expectancy at birth, infant mortality rate, literacy rate, domestic credit provided by banking sector, and domestic credit to the private sector. GDP per capita growth is measured as an annual percentage, and is based on constant local currency. GDP per capita is gross domestic product divided by midyear population. GDP at purchaser’s prices is the sum of gross value added by all resident producers in the economy plus any product taxes and minus any subsidies not included in the value of the products. It is calculated without making deductions for depreciation of fabricated assets or for depletion and degradation of natural resources. Life expectancy at birth is defined as the number of years a newborn infant would live if prevailing patterns of mortality at the time of its birth were to stay the same throughout its life. The infant mortality rate is defined as the number of infants dying before reaching one year of age, per 1,000 live births in a given year. The literacy rate is defined as the percentage of people ages 15 and above who can, with understanding, read and write a short, simple statement on their everyday life. Domestic credit provided by banking sector is measured as a percentage of GDP and is defined as all credit to various sectors on a gross basis, with the exception of credit to the central government, which is net. The banking sector includes monetary authorities and deposit money banks, as well as other banking institutions where data are available (including institutions that do not accept transferable deposits but do incur such liabilities as time and savings deposits). Examples of other banking institutions are savings and mortgage loan institutions and building and loan associations. The final independent variable, domestic credit to the private sector, is also measured as a percentage of GDP, and is defined as financial resources provided to the private sector through loans, purchases of nonequity securities, and trade credits and other accounts receivable, that establish a claim for repayment.2

Variable definitions according to World Development Indicators Online

Empirical Economic Bulletin, Spring 2009, Vol. 2

176

5.0 EMPIRICAL RESULTS The empirical results of this study offer interesting insights into the effects domestic investment, economic growth and human development variables have on Chinese FDI. Table 2 contains the regression results for FDI. Of the six independent variables, four were found to be statistically significant. Domestic credit to the private sector was found to be significant at the 1% level. Domestic credit provided by banking sector was found to be significant at the 5% level. Both GDP per capita growth and the literacy rate were found to be significant at the 10% level. Additionally, life expectancy at birth and the infant mortality rate were found to not be statistically significant. Table 2: Regressions Results for Variables Impacting Foreign Direct Investment in China Variable Constant

Coefficient

T Score

-44.3481

-1.5972

(27.766) 0.1358*

GDP

(0.0778) 0.4144

2.9910

(0.0642) -0.1249**

DCB

2.0264

(0.0470) 0.1919***

DCP

1.2370

(0.0923) 0.0951*

1.0586

(0.3915) 0.1141

1.7462

2.5165

(0.0496)

R Square

0.7348

F Statistic

11.0821

Note: ***, **, and * denotes significance at the 1%, 5%, and 10% respectively. Standard errors in parenthesis.

Empirical Economic Bulletin, Spring 2009, Vol. 2

177

The first independent variable, GDP per capita growth, was found to have a positive coefficient that is significant at the 10 percent level. GDP was expected to have a positive coefficient before the regression was run. A positive coefficient was projected because the better the Chinese economy is doing, the higher its growth rate of GDP will be. Consequently, higher GDP growth indicates a strong economy, where foreign investors would like to invest. Therefore, increases in FDI should be observed with increases in GDP growth in China. The regression proved the expected sign correct, and a positive sign for GDP indicates FDI into China in fact depends on GDP per capita growth in the country. However, at a 10 percent level of significance, this variable was not as strongly linked to FDI as some of the other variables in the regression. Previous literature, such as that done by Fedderke and Romm (2006) show similar results to those found in this study. Fedderke and Romm conclude that growth in market size is a key determinant of levels of FDI in a country. This is consistent with the findings of this paper, which concludes economic growth (and consequently increase in market sizes) is a significant determinant of levels of FDI inflows into China. Ang (2008) finds that there is a difference between real GDP and the growth rate of GDP per capita, when investigating levels of FDI inflow. This is significant because this paper uses GDP per capita growth as a variable for determining if domestic economic growth has an effect on FDI inflows into China. Although Ang is using data from 1960-2005, and also examining Malaysia, his findings that real GDP has a greater effect on FDI inflows than the growth rate of GDP does, are noteworthy. Perhaps the results of the research in this paper could be strengthened if real GDP was used as opposed to the growth rate of GDP. Although, since Ang is studying Malaysia, it is possible GDP growth is in fact a better choice of variable. Either way, Ang’s findings and variable choice is worth mentioning. The results of the next three independent variables, life expectancy at birth, infant mortality rate, and literacy rate, can be examined closely. These variables were added to contribute to current literature regarding determinants of FDI into China. These three variables offer insight into the effects human development variables have on the amount of FDI inflows into China. First, life expectancy was found to have a positive sign, as predicted before the regression was run. However, this variable was not statistically significant. Similarly, the infant

Empirical Economic Bulletin, Spring 2009, Vol. 2

178

mortality rate, whose expected sign was negative, was also found not to be statistically significant. Lastly, the literacy rate was found to be significant at the 10 percent level. Of the three human development variables examined in this paper, only one, the literacy rate, was found to be statistically significant. Both life expectancy and the infant mortality rate were not found to be significant. This outcome suggests that businessmen interested in investing in China are not concerned with life expectancy or the infant mortality rate in China. Because these variables were added to the model to expand on current literature, this paper will offer possible explanations as to why these two variables were not significant, as well as why the literacy rate was found to be significant. Shah’abadie and Mahmudie (2006) offer a possible explanation to the outcome of this papers regression. In their paper, Shah’abadie and Mahmudie state that human capital has a direct and positive impact on the amount of FDI into Iran. Although their study focuses on Iran, the regression results of this paper suggest that the same fact regarding human capital is true in China. That is, human capital has a direct and positive impact on FDI inflows into China. Human capital is defined as the stock of skills and knowledge embodied in the ability to perform labor. According to this definition, literacy would be considered a human capital variable, while life expectancy and the infant mortality rate would not. This is a possible explanation as to why these two variables, life expectancy and the infant mortality rate, were found not to be significant, and the literacy rate was found to be significant. Because literacy is an aspect of human capital, as Shah’abadie and Mahmudie have concluded for the case of Iran, it will have a positive and significant effect on levels of FDI into China. Because literacy rate is a proxy for education, the higher the literacy rate is, the more educated the population is. Consequently, the more educated the population, specifically the workforce, the more productive the workers will be. Having educated and more productive workers is obviously something investor’s interested in starting a business in China will look at. Therefore, this explains why out of the three human development variables included in the model, only the literacy rate was found to be significant. In addition, Rodriguez and Pallas (2008) find that human capital factors play an important role in attracting FDI in Spain. Therefore, this paper concludes that human capital, specifically literacy, is a key factor in attracting FDI into China. This innovation matches the findings of previous studies examining other countries.

Empirical Economic Bulletin, Spring 2009, Vol. 2

179

It is also worth noting that the conclusions of Hong (2008) are that cheap labor is more important than quality of labor. As opposed to Hong’s findings, this paper concludes that literacy rate (a proxy for education) is in fact a significant determinant of FDI inflows into China. Hong finds that labor quality is not as important as cheap labor. Therefore, an educated workforce is not as important to investors as a cheap workforce. Perhaps Hong’s findings regarding the importance of a literate and therefore educated workforce explain why the literacy rate variable was found to be significant only at the 10 percent level, and not at 5 or 1 percent. The incongruity between Hong’s study of China and this study can perhaps be explained by the time periods being examined. Possibly, labor quality plays a more significant role today than it did in previous years, which is why it was found to be significant in this papers regression results, and not Hong’s. In either case, it is important to consider Hong’s findings when interpreting the findings of this paper. The last two independent variables, domestic credit to the private sector and domestic credit provided by the banking sector, which both measure levels of domestic investment, were found to have the greatest level of significance. Domestic credit to the private sector was found to have a positive coefficient, as expected, and was significant at the 1 percent level. Domestic credit provided by the banking sector also had a positive coefficient, and was significant at the 5 percent level. Domestic investment can be seen as a key determinant of the amount of FDI into China because if domestic investments are being made, this suggests the economy is strong and there are many business opportunities open. Shah’abadie and Mahmudie (2006) find that domestic investment is a key factor in determining FDI. They conclude domestic investment has a direct and positive impact on FDI. Although their study focuses on Iran, using data from 19592003, their conclusions are similar to this study’s findings concerning China. Therefore, this study concludes that similar to Shah’abadie and Mahmudie’s findings regarding Iran, domestic investment in China also has a direct and positive impact on FDI. As mentioned earlier, domestic credit to the private sector and domestic credit provided by the banking sector, which both measure domestic investment, were the most significant variables in this papers regression. More specifically, credit to the private sector was found to be significant at the 1 percent level. This high level of significance can be explained by the fact that credit to the private sector directly indicates how well businesses are doing, and if they are

Empirical Economic Bulletin, Spring 2009, Vol. 2

180

expanding. Credit provided by the banking sector also indicates how domestic Chinese businesses are doing. If Chinese businesses are taking out loans on credit to expand, this indicates that there are in fact new business opportunities in the market. Foreign investors may see this, and realizing the market is strong, want to invest in China to fulfill market demand. The overall fit of the regression, or how well future outcomes are likely to be predicted by the model, is measured by the R square number. The R square of this papers regression was 0.7348, suggesting that the variables in the model explain about 73 percent of the factors influencing levels of FDI into China. The T statistic is the estimated coefficient divided by its own standard error and it measures how many standard deviations from zero the estimated coefficient is. It is used to test the hypothesis that the true value of the coefficient is non-zero, in order to confirm that the independent variable really belongs in the model. All the statistically significant variables in the regression had a T statistic of greater than 1.7. The literacy rate, along with domestic credit to the private sector and domestic credit provided by the banking sector all had T statistics of greater than two.

6.0 CONCLUSION AND POLICY IMPLICATION The goal of this paper was to answer three fundamental questions: whether economic growth in China has an impact on FDI into the country, what effect high levels of domestic investment have on the amount of FDI into the country, and whether human development factors affect the amount of FDI that flows into China. The results of this paper verify that the economic performance of Chinas economy does in fact impact how much FDI flows into the country. Additionally, it has been demonstrated empirically that domestic investment in China has a considerable direct and positive impact on FDI. Finally, human development factors were examined to differentiate this study from others. The findings here were interesting. Human development factors are not as important as human capita. In other words, variables such as life expectancy or infant mortality were found not to be significant determinants of levels of FDI into China. However, the literacy rate, which measurers the education level of the population and workforce, was found to be significant. Therefore, human development factors in general do not explain levels of FDI into China, but

Empirical Economic Bulletin, Spring 2009, Vol. 2

181

human capital factors, such as the literacy rate, do. In summary, the literacy rate, and other human capital factors are variables that help explain levels of FDI into China. From a policy point of view, as suggested by Ang (2008), increases in the level of financial development, infrastructure development, and trade openness promote FDI. Alternatively, higher corporate tax rate and appreciation of the real exchange rate appear to discourage FDI inflows. In addition, higher macroeconomic uncertainty induces more FDI inflows. Therefore, if the goal is to increase FDI inflows into China, the country should promote financial development and develop its infrastructure. The results of this paper also conclude that the country should promote human development. Increased education will lead to increases in FDI inflows into the country. If China is looking to decrease levels of FDI, it should increase the corporate tax rate.

Empirical Economic Bulletin, Spring 2009, Vol. 2

182

Appendix A: Variable Description and Data Source

Acronym

Description

Data Source

FDI

Foreign direct investment, net inflows (% of GDP)

WDI

GDP

GDP per capita growth (annual %)

WDI

Life expectancy at birth, total (years)

WDI

Mortality rate, infant (per 1,000 live births)

WDI

Literacy rate, adult total (% of people ages 15 and above)

WDI

DCP

Domestic credit to private sector (% of GDP)

WDI

DCB

Domestic credit provided by banking sector (% of GDP)

WDI

Empirical Economic Bulletin, Spring 2009, Vol. 2

183

Appendix B: Variables and Expected Signs

Acronym

Variable Description

FDI

Foreign Direct Investment

GDP

Gross Domestic Product

Life Expectancy

Infant Mortality Rate

Literacy Rate

DCP

Domestic Credit to Private Sector

DCB

Domestic Credit Provided by Banking Sector

Empirical Economic Bulletin, Spring 2009, Vol. 2

Expected Sign

184

BIBLIOGRAPHY Ang, James B. "Determinants of Foreign Direct Investment in Malaysia." Journal of Policy Modeling 30, no. 1 (January 2008): 185-189. EconLit, EBSCOhost (accessed April 13, 2009). Fedderke, J. W., and A. T. Romm. "Growth Impact and Determinants of Foreign Direct Investment into South Africa, 1956-2003." Economic Modelling 23, no. 5 (September 2006): 738-760. EconLit, EBSCOhost (accessed April 13, 2009). Gast, Michael, and Roland Herrmann. "Determinants of Foreign Direct Investment of OECD Countries 1991-2001." International Economic Journal 22, no. 4 (December 2008): 509-524. EconLit, EBSCOhost (accessed April 13, 2009). Hong, Junjie. "WTO Accession and Foreign Direct Investment in China." Journal of Chinese Economic and Foreign Trade Studies 1, no. 2 (2008): 136-147. EconLit, EBSCOhost (accessed April 13, 2009). Human Development Report, 2007 Naude, W. A., and W. F. Krugell. "Investigating Geography and Institutions as Determinants of Foreign Direct Investment in Africa Using Panel Data." Applied Economics 39, no. 10-12 (June 2007): 1223-1233. EconLit, EBSCOhost (accessed April 13, 2009). Rodriguez, Xose A., and Julio Pallas. "Determinants of Foreign Direct Investment in Spain." Applied Economics 40, no. 19-21 (October 2008): 2443-2450. EconLit, EBSCOhost (accessed April 13, 2009). Shah'abadie, A., and A. Mahmudie. "Determinants of Foreign Direct Investment (A Case Study for Iran). (In Farsi. With English summary.)." Biquarterly Journal of Economic Essays 3, no. 5 (2006): 89-125. EconLit, EBSCOhost (accessed April 13, 2009). Tang, Sumei, Selvanathan, E. A. and Selvanathan, S.,”Foreign Direct Investment, Domestic Investment and Economic Growth in China: A Time Series Analysis”. World Economy, Vol. 31, Issue 10, pp. 1292-1309, October 2008. Available at SSRN: http://ssrn.com/abstract=1276772 or DOI: 10.1111/j.1467-9701.2008.01129.x Uppenberg, Kristian, and Armin Riess. "Determinants and Growth Effects of Foreign Direct Investment." EIB Papers 9, no. 1 (2004): 52-84. EconLit, EBSCOhost (accessed April 13, 2009). World Development Indicators Online

Empirical Economic Bulletin, Spring 2009, Vol. 2

185

The Impacts of Energy Efficiency and Consumption on GDP in the Euro Area

Justin T. Getts

Abstract: This paper analyzes the aggregate data of the Euro Area to determine how GDP per unit of energy is affected by the use of common energy sources. Time series data from 1980 to 2005 is used to show the change in how energy is used compared to the growth of GDP. It is revealed in this paper that the consumption of efficient forms of energy is highly correlated to GDP growth and the use of inefficient energy sources leads to less growth.

JEL Code: Q40, Q43 Keywords: Energy; Energy Efficiency; Energy Consumption; Euro Area; GDP

Department of Economics, Bryant University, 1150 Douglas Pike, Smithfield, RI 02917. Phone: (203) 804-8432. Email: jgetts@bryant.edu.

The author thanks Dr. Ramesh Mohan for help and guidance

Empirical Economic Bulletin, Spring 2009, Vol. 2

186

1.0 INTRODUCTION The energy crises in the 1970s, including but not limited to the 1973 oil crisis raised the prices of energy and oil. This had a negative effect on economic growth and called for a look into energy conservation. It was not known at the time whether energy conservation would help or hurt the economy. Some thought it would help the economy because people would spend less on energy and have more left to invest. Others thought that energy conservation would hurt the economy because tasks that need energy might need to be done slower or be given up altogether. Since the energy crises, there has been quite a bit of research interest on the Granger Causality between energy consumption and economic growth. Many have found causality to exist and some have even come up with policy implications on their findings. The main question these papers looked to answer was whether or not energy consumption or energy efficiency can stimulate economic growth. What these papers did not look at was if the use or conservation of certain energy types affected economic growth more than others. This paper will assume the causality found in previous works and look at the GDP per unit of energy use ratio in the Euro area and relate that to usage of specific types of energy in the same area. By studying different types of energy, policy decisions can be made as to where, if any, government funding in energy should be placed. It would make sense that the funding be placed in the most efficient and cost effective forms of energy. The goal of this paper is to recommend certain energy types and discourage others as to best affect the economic growth per unit of energy usage ratio. This paper proceeds as follows. Section 2 looks at current trends for the energy consumption based on energy sources and its relation to GDP. Section 3 briefly reviews previous studies on the topic and looks into current policies the EU implements regarding energy. Section 4 deals with the data used in the empirical analysis while Section 5 presents the results found. In Section 6, conclusions are drawn about the topic and possible policy implications are presented.

Empirical Economic Bulletin, Spring 2009, Vol. 2

187

2.0 TRENDS IN ENERGY AND THE EURO AREA Identifying what types of energy should be used is a topic worth investigating. The main sources of energy in the world include petroleum, coal, and natural gas. All of these sources are non-renewable meaning there is a limited supply of these sources. The development of other energy sources will be necessary once the supply of the three main energy sources run out. Some alternative sources of energy that are renewable include nuclear, hydroelectric, biomass, geothermal, wind, and solar power. The usage of these renewable sources has been increasing over the years but is still much less than the three main sources of energy. Figure 1 shows the amount of each source of energy used in the world. We can see that the consumption of oil, coal, and natural gas account for about 80 percent of the total energy consumption in 2006. Figure 1: Global Sources of Energy in 2006

Source: <http://news.cnet.com/i/bto/20080424/Energy_sources.gif>

Empirical Economic Bulletin, Spring 2009, Vol. 2

188

Because this paper is concentrated on the Euro area, it is shown in the next graph, Figure 2, the energy consumption by source in Spain which is part of the Euro zone. Spain has split up its energy consumption similar to that of the rest of the world with one exception. Spain uses less coal and more petroleum in providing its energy. The amounts of renewable energy sources are similar to the rest of the world in that they represent less than twenty percent of the total energy consumption. Figure 2: Sources of Energy in Spain

Source: <http://www.energytribune.com/live_images/euro_pie_feb_08.gif>

Figure 3 shows the energy consumption in relation to GDP. As you can see, most countries shown have decreased their energy consumption in relation to GDP from 2001 to 2002. This also means that their GDP per unit of energy ratio has increased as well. The countries with the most energy consumption in relation to GDP such as Lithuania and Estonia are the least efficient energy users. The countries with a lower ratio like Denmark and Austria have a much higher energy efficiency.

Empirical Economic Bulletin, Spring 2009, Vol. 2

189

Figure 3: Energy intensity of the economy Energy consumption in relation to GDP (tons of oil equivalents per 1,000,000 euro)

Source: <http://www.scb.se/Statistik/OV/OV0011/2003M00/OV0011_2003M00_DI_54_EN_Energy_EU.gif>

Even though most countries here are increasing their energy efficiency, some countries like Poland, Finland, and Portugal are doing just the opposite. These countries may need to look into some reforms in their energy policies to keep up with the changing energy trends.

Empirical Economic Bulletin, Spring 2009, Vol. 2

190

3.0 LITERATURE REVIEW The empirical relationship between economic growth and energy consumption for Greece from 1960 to 1996 done by (Hondroyiannis et al, 2002) is shown to be a relevant study. The results from this analysis do not show any short term correlation but does suggest a long term relationship between energy consumption and economic growth. Improvements in economic efficiency would promote economic growth and, in turn, improve energy consumption. Also, improvements in economic efficiency would help improve energy efficiency. Therefore, improved energy efficiency would not hamper economic growth and should be a good focus point for Greece to improve. It is assumed that the same policies would work for any middle size country. According to (Chapman and Suri, 1998), pollution is shown to follow an inverted-U path in relation to economic growth. The turning point of pollutant emissions to economic growth is either at $55,000 or at $224,000 depending on if international trade is configured into the analysis. This would imply that even the most developed countries have a long way to go to reach the turning point where more energy consumption does not yield a higher economic growth. Even though countries do not need to worry about reaching the peak of the Kuznets curve, many developing countries already have flattening returns to energy consumption. In the paper by (Cleveland et al, 2000), there are three different studies examined. The first study determines how much energy is actually taken out of fossil fuels in the United States. The second study looks to figure out if there is a causal relationship between energy consumption and GDP. The third, energy quality is looked at to determine if changes can be predicted in the energy/GDP ratio. The main point of this article is to show that countries are improving their energy efficiency by replacing low quality energy with high quality energy. Some low quality energies cannot be replaced, however, because it would not make sense to use certain energy sources for specific tasks. The causal relationship between energy consumption and GDP in G-7 countries is looked at by (Sari and Soytas, 2002). It is found that causality runs from energy consumption to GDP in these countries. They state their case that energy conservation in countries like Argentina could negatively impact economic growth in the same country. The causal relationship between energy consumption and GDP for Korea is examined by (Lee and Oh, 2003). The results of this article suggest there is a long run causal relationship

Empirical Economic Bulletin, Spring 2009, Vol. 2

191

between energy consumption and GDP in Korea. The methods used were a Granger Causality test with a VECM model. According to (Levine, Qingy, and Sinton, 1998), China is an example for other countries to follow in energy efficiency policies. China’s energy consumption has not grown as fast as their GDP meaning their energy efficiency has been getting better every year. This also shows that economic growth is possible with energy conservation acts in place. The paper also looks at the various energy conservation policies put in place by China and attempts to show why each one worked. Each policy is determined whether or not it will help other countries like it did for China. 4.0 DATA AND EMPIRICAL METHODOLOGY 4.1 Definition of Variables GDPPEt = β0 + β1COALt + β2DNGt + β3GEO_S_WEt + β4NUCELECt + β5HYDROt + β6PETROt + β7N_ELEC_IMPt + εt GDPPEt is the GDP per unit of energy use in constant 2005 purchasing power parity dollars per kilogram of oil equivalent at time t. Time t runs from the starting year 1980 to the ending year 2006. The sample from which data is taken from is the Euro area, or also known as the Euro zone. This area includes these 16 countries: Austria, Belgium, Cyprus, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, Malta, Netherlands, Portugal, Slovakia, Slovenia, and Spain. These 16 countries make up the aggregate data known as the Euro zone. The importance of the GDP per unit of energy has been downplayed by authorities in the recent past. It is, however, important for this study because it is one of the few variables that capture both GDP and energy consumption. Another name for the relationship between energy consumption and GDP is energy efficiency. Independent variables consist of seven variables all from the Energy Information Administration. Appendix A and B provide data source, acronyms, descriptions, expected signs, and justifications for using the variables. First, COALt is the consumption of coal from 1980 to 2006 in Quadrillion Btus. Coal is an inefficient form of energy that was used much more in the past as it is now. Second, DNGt is the consumption of natural gases from 1980 to 2006 in the Euro zone. Many homes are heated with natural gas and claims are that it is cleaner than heating oil. Third, GEO_S_WEt is the consumption of Geothermal, Solar, Wood, Wind, and Waste Electric power from 1980 to 2006. These sources of electricity are not used very often but many

Empirical Economic Bulletin, Spring 2009, Vol. 2

192

say are cheaper sources for the future. Fourth, NUCELECt is the consumption of nuclear electric power from 1980 to 2006 and is a main source of electricity. Fifth, HYDROt is the consumption of electricity from a hydroelectric source. The amount of hydroelectric power consumed over the last 25 years has been quite constant. Sixth, PETROt is the consumption of oil used in heating homes and operating vehicles. It the most used energy source in the Euro area. Lastly, N_ELEC_IMPt is the amount of electricity brought in from other regions minus the exports of electricity from 1980 to 2006. This is important because not every country supplies enough their energy to its citizens and need to buy it from elsewhere.

4.2 Data This study takes data from the World Bank and the Energy Information Administration for the Euro area. Summary statistics of the data are provided in Table 1. Table 1: Summary Statistics

GDPPE COAL DNG N_ELEC_IMP GEO_S_WE HYDRO NUCELEC PETRO

Empirical Economic Bulletin, Spring 2009, Vol. 2

Observations 26 26 26 26 26 26 26 26

Mean 6.669231 15.99388 14.07221 0.079186 0.49593 3.20004 7.662409 29.03564

Std. Dev. Minimum Maximum 0.530554 5.7 7.46 2.745098 12.73165 19.72517 2.998925 10.25982 19.60231 0.062515 -0.02502 0.188499 0.418403 0.157742 1.614721 0.22195 2.811622 3.812505 2.301608 2.434414 9.966801 1.635547 26.20626 31.45939

193

5.0 EMPIRICAL RESULTS 5.1 Non-renewable Energy Sources The first column of Table 2 represents the relationship between GDP/energy unit and the non-renewable energy sources like petroleum, natural gas, and coal. This relationship is important to look at because there currently is no direct substitute for these energy sources when it comes to heating homes and fueling cars. The results for this regression came out as expected. It was expected that natural gas and petroleum come out as either a positive variable or slightly negative because they are known to be more efficient energy sources than coal. The coefficient of coal was expected to be negative because coal is an inefficient energy source. All three variables came out to be significant at the 1% level showing that all three variables made an impact on the regression fit. The positive coefficient in petroleum shows that it is more efficient than coal which has a negative coefficient. 5.2 Electricity Added Shown in column 2, the results for the regression between GDP/energy unit and all energy sources both renewable and non-renewable appeared to be a better fit. The results still came out as expected with coal having a negative coefficient and new sources of electricity like nuclear, geothermal, and solar energy having a positive coefficient. The positive coefficient shows that these types of energy are more efficient than the non-renewable energy sources. One problem with this set of data, however, is that many variables are related and have a similar impact on the regression. Therefore, the variables DNG and HYDRO turned out to be insignificant. The use of these sources of energy has been relatively constant over the last 25 years in the Euro area. 5.3 Electricity Imports This regression in column 3 of Table 2 has added the variable N_ELEC_IMP to the independent variables to see if the amount of energy taken in from other areas affects the GDP/energy ratio. It has been found on a 1% significant level that imports positively affect the GDP/energy ratio. It is not known whether imports of electricity cause for a higher GDP or whether a higher GDP allows for countries to import more electricity. It makes more sense to be the latter.

Empirical Economic Bulletin, Spring 2009, Vol. 2

194

Table 2: Regression Results for the Euro Area

I. 9.943757*** (1.017234) -0.126417*** (0.022447) 0.100471*** (0.018353)

GDPPE II. 9.804803*** (1.235277) -0.126383*** (0.024040) -0.044580 (0.054113)

R^2

-0.091834*** (0.024272) 0.958923

0.530590** (0.223606) 0.002212 (0.124024) 0.088324*** (0.030604) -0.049383* (0.027740) 0.972574

III. 9.709387*** (0.896563) -0.161054*** (0.019254) -0.001285 (0.040561) 2.182604*** (0.513156) 0.391956** (0.165484) -0.006636 (0.090012) 0.056750** (0.023413) -0.042259* (0.020197) 0.986322

F-Statistics

171.1938

112.2976

185.4210

Number of Observations

CONSTANT COAL DNG N_ELEC_IMP GEO_S_WE HYDRO NUCELEC PETRO

Note: *** , **, and * denotes significance at the 1%, 5%, and 10% respectively. Standard errors in parentheses

5.4 Policy Implications For the better of the Euro Area, policy implications can be made in regards to energy consumption and economic growth. It has been shown in this paper that the use of efficient energy sources lead to a higher GDP per unit of energy use. Therefore, it is suggested that government funding towards the development of energy sources like geothermal, nuclear, and solar energy be increased. These sources of energy should be developed further to be made available as main sources of energy. Also, to discourage the use of inefficient energy sources, it is suggested that a tax be placed on the production and consumption of coal in the Euro Area. Coal can easily be replaced by petroleum or natural gas and action should be made to make this replacement sooner rather than later. The increase in use of efficient energy sources and the

Empirical Economic Bulletin, Spring 2009, Vol. 2

195

decrease in use of inefficient energy sources will stimulate economic growth and show the way to higher energy efficiency. 6.0 CONCLUSION This study looked at energy sources in the Euro Area and attempted to determine the effect of energy consumption on GDP. The hypothesis of this paper was that the consumption of the most efficient sources of energy would lead to a higher GDP per unit of energy consumed. Likewise, the consumption of the least efficient sources of energy would lead to a lower GDP per unit of energy consumed. The results of this paper show that coal, which is often considered an outdated energy source, is an inefficient form of energy. Other sources of energy, such as petroleum and natural gas, are much more efficient than coal and can be considered direct substitutes. The newest forms of energy such as geothermal, solar, and nuclear energy are shown to be very efficient energy sources. Even though these sources of energy are not direct substitutes of oil and natural gas, they still can be used to satisfy many types of energy needs.

Empirical Economic Bulletin, Spring 2009, Vol. 2

196

Appendix A: Variable Description and Data Source Acronym GDPPE

COAL

DNG

GEO_S_WE

NUCELEC

HYDRO

PETRO

N_ELEC_IMP

Description GDP per unit of energy use (constant 2005 PPP $ per kg of oil equivalent) Euro area Consumption of Coal in EU (Quadrillion (10^15) Btu), 1980-2006 Consumption of Natural Gas in EU (Quadrillion (10^15) Btu), 1980-2006 Consumption of Geothermal, Solar, Wood, Wind, and Waste Electric Power in EU (Quadrillion (10^15) Btu), 1980-2006 Consumption of Nuclear Electric Power in EU (Quadrillion (10^15) Btu), 1980-2006 Consumption of Hydroelectric Power in EU (Quadrillion (10^15) Btu), 1980-2006 Consumption of PETRO in EU (Quadrillion (10^15) Btu), 1980-2006 Consumption of Electricity Imports in EU (Quadrillion (10^15) Btu), 1980-2006

Empirical Economic Bulletin, Spring 2009, Vol. 2

Data Source World Bank

Energy Information Administration Energy Information Administration Energy Information Administration

Energy Information Administration

Energy Information Administration Energy Information Administration Energy Information Administration

197

Appendix B: Variables and Expected Signs Acronym

Variable Description

COAL

Coal consumption An inefficient and outdated source of energy

DNG

Natural gas consumption

Used in home heating

+/-

GEO_S_WE

Geothermal, solar, wind, wood, and waste electricity consumption

New, cheaper sources of electricity

NUCELEC

Nuclear electricity consumption

Source of electricity

HYDRO

Hydroelectricity consumption

Source of electricity

+/-

PETRO

Petroleum consumption

Energy used in home heating and cars

+/-

N_ELEC_IMP

Net electricity imports

Electricity consumed but not produced

+/-

Empirical Economic Bulletin, Spring 2009, Vol. 2

What it captures

Expected sign

198

BIBLIOGRAPY Chapman, Duane and Suri, Vivek, (1998), “Economic growth, trade and energy: implications for the environmental Kuznets curve”, Ecological Economics, 25 (2): 195-208. Cleveland, C., Kaufmann, R., Stern, D., (1999), “Aggregation and the role of energy in the economy”, Ecological Economics, 32 (2000): 301–317 Energy Information Administration: Official Energy Statistics from the U.S. Government <http://www.eia.doe.gov/> Hondroyiannis, G., Lolos, S., Papapetrou, E., (2002), “Energy consumption and economic growth: assessing the evidence from Greece”, Energy Economics, 24 (4): 319-336 Lee, K., Oh, W., (2003), “Causal relationship between energy consumption and GDP revisited: the case of Korea 1970-1999”, Energy Economics, 26 (1): 51-59 Levine, M., Qingyi, W., Sinton, J., (1998) “Energy efficiency in China: accomplishments and challenges”, Energy Policy, 26 (11): 813-829 Sari, R., Soytas, U., (2002), “Energy consumption and GDP: causality relationship in G-7 countries and emerging markets”, Energy Economics, 25 (1): 33-37 The World Bank Group: WDI Online <www.worldbank.org>

Empirical Economic Bulletin, Spring 2009, Vol. 2

199

Empirical Economic Bulletin, Spring 2009, Volume 2

Articles inside

Kevin Hauver

Are Educated Women Less Likely to Get Married?, Qian Jiang

Effects on Globalization on Income Inequality in High Income Coutnries, Craig Bradford

G. Acevedo

The Effect of Crime Rates on Housing Prices: A Hedonic Study, John Paul Goncalves

The Relationship Between Housing Prices and Macroeconomic Factors in Spain, Garret Fitzgerald

Mirza

The Effects of Aging on Inflation, Andrew Stone