Annals of Sri Lanka Department of Agriculture. 2007.9:51-58.
ANALYSIS OF FOOD CONSUMPTION PATTERNS IN SRI LANKA WITH SPECIAL REFERENCE TO ENERGY INTAKE R.M. HERATH1, H.U.WARANAKULASURIYA2, K.G. THILAKARATHNA3 and J.A.T.P. GUNAWARDENA1 1 Socio Economic and Planning Centre, Department of Agriculture, Peradeniya 2 Progress Monitoring and Evaluation Unit, Department of Agriculture, Peradeniya 3 Department of Census and Statistics, Colombo
ABSTRACT The pattern of food consumption in Sri Lanka has undergone dramatic changes over the last tow decades. This study examines the variation in food consumption pattern and economic parameters determining the demand for the period 1980-2002. The cross sectional household data obtained from Household Income and Expenditure Survey Reports (1980/81, 1985/86, 1990/91, 1995/96 and 2002) published by Department of Census and Statistics was used for the analysis. Analysis was done using AIDS model. The study reveals that food consumption pattern is shifting from starch food to non-starch food. Yet, rice is the cheapest source of energy. Own price elasticities of rice, wheat, pulses, meat, fish, milk and eggs were -0.6788 -0.6743, -0.4231, -0.9732, -0.9321, -0.9632, -0.6321 respectively. Own price elasticities and expenditure elasticities for rice, wheat, pulses, meat, fish, milk and eggs were significant (P<0.05). Sri Lankan consumers are more responsive to price changes in rice as compared to those of wheat flour. Cross price elasticity of rice (0.2567) with respect to wheat is significant (P<0.05) implying that demand for rice is affected by price of wheat. Therefore, reducing subsidy on wheat in order to increase the demand for rice is an effective strategy to increase income of paddy farmers. KEYWORDS: Energy, Food consumption pattern, Price elasticities.
INTRODUCTION Ensuring food and nutrition security through achieving self sufficiency, especially, in rice has been one of the prime policy objective of successive governments in Sri Lanka since independence. One of the most important nutritional aspects is intake of energy. Energy intake varies by income group, age group and among urban, rural and estate sectors and links to food consumption pattern. Sri Lankan food consumption patterns have been undergoing dramatic changes over the last two decades with increasing consumption of meat and wheat flour, and decreasing consumption of rice (Alderman and Timmer, 1980; Bogahawatte and Kaillasapathy, 1986). Food consumption patterns are associated with income, relative availability and accessibility to food, relative prices of different food items, urbanization and lifestyles of the people. Undoubtedly, many factors have influenced Sri Lankan food consumption pattern. Understanding the changes in food consumption pattern is important
52 HERATH et al.
for future food policy planning for the country. This study examines the variation in food consumption with special reference to gross energy intake (the nutrient intake from food) from different food items at national level and economic parameters determining the demand for food. A number of authors have studied the food consumption pattern and estimated the economic parameters for different food items. Ratnayake et al. (2004) estimated demand parameters for rice and wheat in Sri Lanka using time series data. Thudawe (2002) estimated food demand parameters for eleven food groups using cross sectional data extracted from a sub-sample of 900 units, randomly selected from Western Province data of National Income and Expenditure surveys for years 1990/91 and 1995/96. The first study was limited to only two food items and the second study to a single Province. The present study uses the sample of National Income and Expenditure Survey 2002 for the whole island and estimates demand parameters for seven food items. METHODOLOGY Data This study has used the data of the Household Income and Expenditure Survey (HIES) conducted by Department of Census and Statistics (DCS) every five years. Data of this survey is built up on reported consumption and expenditure. Accordingly, any food used in a processed or prepared form, such as rice flour or wheat flour, appears as a category different from unprocessed form. Therefore, necessary adjustments should be made to estimate total consumption of any food. Further, consumption by household members of different age classes is averaged in per capita consumption calculation in HIES. Despite the above mentioned defects in HIES data, they are the most complete, consistent sets of data available on consumption patterns in Sri Lanka collected from a large sample over enough geographical and temporal diversity to capture significant variance in the relevant variables. Therefore, published HIES data of surveys of 1990/91, 1995/96 and 2002 were used to make deductions on changes of food consumption pattern in this study. The food items reported in the surveys were grouped into eight food groups: rice, wheat, pulses, meat, fish, egg, milk and others such as coconut, sugar and vegetable. For estimating the economic parameters, the disaggregated household level data of the HIES of 2002 were used. The data used were household total expenditure on food, price, value and quantity of food consumed by the household for a period of one week within the year.
ANALYSIS OF FOOD CONSUMPTION PATTERNS IN SRI LANKA 53
Model Although consumer theory is developed from decision perspective of an individual consumer or consuming household, it is usually applied empirically in per capita or per household terms to aggregate market data. The present study used per household term for the analysis. The linear Approximation of Almost Ideal Demand System developed (LA/AIDS) by Deaton and Muellbauer (1980) was used for the estimation of demand parameters. As Moschini (1998) pointed out the AIDS model automatically satisfies the adding-up restriction and with simple parametric restrictions, homogeneity and symmetry can be imposed. Following equation presents the basic functional form of LA/AIDS used for the study. wi = αi + ∑ γij ln (pj) + βi ( ln (χ) - ∑ wj ln (pj) )+ μi j j i= 1,……7 Where, wi is the budget share of food group i,. pj is the price of good j, x is the total expenditure of the goods in the model, wj is budget share, μi, is random disturbances assumed with zero mean and constant variance. αi, γij and βi are respective parameters. The adding up restrictions requires ∑wi = 1 for all J; ∑αi = 0, ∑ γij = 0 ∑ βi = 0 The homogeneity restriction is satisfied if an only if for all j; ∑ γjk =0 k The symmetry is satisfied by γij = γij The iterative seemingly unrelated regression procedure (ITSUR) in SAS was used for parameter estimation. The homogeneity and symmetry were imposed in the estimation. RESULTS AND DISCUSSION Energy intake and its sources Table 1 shows national average per capita daily energy intake and contribution of different food commodities at five points of time from 1980 to 2002. The total calorie intake per capita per day shows a slight increase from 1982 kilo calories in year 1980 to 2078 kilo calorie in year 2002 with year to year variation.
54 HERATH et al. Table 1. Year
1980/81 1985/86 1990/91 1995/96 2001/02
Per capita daily energy intake and contribution from different food commodities. Total energy Intake (kilo calories) 1982 2138 2080 2095 2078
Percent energy intake from Rice
Whea t
Pulse s
Meat
Fish
Milk
Egg
Other
51 46 47 46 45
13 15 19 22 19
7 6 7 5 5
1 1 2 3 4
3 4 5 6 7
4 4 3 2 1
Neg Neg Neg Neg Neg
21 24 17 16 19
Source: Household Income and Expenditure Survey, 2002.
Rice is the major source of energy and rice and wheat together account for around 65% of daily energy intake. The percentage intake of energy from rice dropped from 51% in 1980 to 46% by 1985/86 and remained around that figure thereafter, whereas wheat flour consumption has increased from 13% in 1980 to 19% by 1990/91 and remained around that figure thereafter. There was a gradual decline in percentage energy intake from milk and pulses whereas intake of energy from meat and fish has gradually increased. Percentage energy intake from animal sources (meat and fish) markedly increased from 4% in 1980/81 to 11% in 2001/02. The other food items such as coconut, sugar, vegetables were also important sources of energy. These food items accounted for about 22% of energy intake in 1980 and dropped to 19% in 2002. Apparently, the reduction of the percentage energy intake from rice and pulses has been compensated by increased consumption of wheat, meat and fish. The per capita daily expenditure for different energy sources is shown in Table 2. Rice, the main source of energy, has accounted for the largest proportion of expenditure of food during the period from year 19802002. However, the expenditure shares on this commodity during this period shows a rapid downward trend and the expenditure share for wheat flour also has decreased, but at a slower rate. The expenditure shares on animal product such as meat, milk and fish have increased correspondingly to decrease of rice and wheat flour. This may be related to increase in per capita consumption of these commodity groups (except milk) and increase in relative prices of these food items. Energy associated with one Rupee expenditure on food commodities is given in Table 3. There was 83% reduction in the energy associated with one rupee expenditure during the period from 1980-2002. Of the food commodities, rice was the cheapest source of energy during this period. Wheat flour and pulses were second and third cheapest energy sources respectively.
ANALYSIS OF FOOD CONSUMPTION PATTERNS IN SRI LANKA 55
Table 2. Per capita daily expenditure on different food commodities. Year Total Per cent expenditure on expenditure Rice Whea Pulse Mea Fish Milk on food t s t (Rs/capita/d ) 1980/81 5.42 31.62 7.03 2.38 1.76 8.53 1985/86 7.91 25.29 7.85 3.64 2.56 9.83 1990/91 16.13 25.73 5.99 4.85 3.37 10.04 1995/96 26.31 21.17 5.57 3.66 4.39 11.12 2001/02 47.75 19.30 5.47 3.08 3.11 9.34 Source: Household Income and Expenditure Survey Reports.
3.39 4.30 5.15 7.15 7.90
Egg s
Othe r
0.75 0.91 1.10 1.10 0.86
44.54 45.62 43.78 45.86 50.95
Table 3. Energy associated with one Rupee daily expenditure on food commodities. Year
1980/81 1985/86 1990/91 1995/96 2001/02
Total energy per rupee expenditure (kilocalories) 924.89 632.20 284.21 205.89 157.06
Percentage of total energy associated with one rupee expenditure Rice
Wheat
Pulse s
Meat
Fish
Milk
Egg s
Other
21.97 19.19 23.38 20.09 13.04
7.68 10.16 9.82 13.73 9.07
2.67 5.31 9.67 10.53 7.66
0.31 0.50 0.69 1.12 0.48
1.58 1.96 2.29 3.26 2.95
2.30 2.34 3.62 3.82 4.40
0.06 0.07 0.12 0.13 0.09
63.43 60.48 50.42 47.31 62.32
Source: Household Income and Expenditure Survey Reports.
The changes in per capita food consumption in quantity term are given in Table 4. The per capita rice consumption decreased from 318g per day in 1980 to 285g per day in 2002 and the consumption of wheat flour increased from 78g per person per day in 1980 to 105g per person per day in 2002. The change in consumption of rice and wheat explains substitution effect between these two. The increase in wheat consumption may be due to tendency towards diversifying staple food diet with the changing life styles coupled with urbanization and increasing per capita income and availability and accessibility to wheat flour based products in different forms in the market. Recognizing the importance of wheat flour as a major source of energy, government has been importing and distributing wheat flour at subsidized price. However, increasing wheat consumption will be a threat to domestic rice production resulting in shrinking demand for rice. Per capita consumption of pulses declined by 43 percent from 1980 to 2002. Another interesting phenomenon is that consumption of non starchy food (meat and eggs) has increased. Increase in consumption of these non staple foods may have also lead to decrease in consumption of rice. Theoretically, as income grows to a certain level, per capita rice consumption increases with consumers substituting the coarse grains and root crops with rice. As income increases beyond a threshold, people can afford to have a high cost balanced diet containing food that provides more proteins and vitamins such as vegetables
56 HERATH et al.
fruits, fish and livestock products. From that income threshold, per capita rice consumption starts declining. Table 4. Per capita daily consumption of different food commodities; 1980-2002. Food % change from 1980 % change 1980/ 1985/ 1990/ 1995/ 2001/ from 1980/81 to 2001/02 81 86 91 96 02 Rice (g) 318 307 301 281 285 -10 Wheat flour (g) 78 90 85 115 105 35 Pulses and nuts(g) 10 8 7 6 5 -43 Meat (g) 5 7 8 NA 10 90 Fish (g) 33 33 25 NA 35 5 Milk (ml) 18 16 10 5 2 -87 Egg- (No/month)* 1 1 2 3 3 91 NA - Data not available. Source: Household Income and Expenditure Survey Reports; 80/81, 85/86, 90/91, 95/96, 2001/2002.
Results of econometric model Table 5 shows the uncompensated price elasticities and expenditure elasticities obtained from LA/AIDS models. Own price elasticities for rice, wheat pulses, meat, fish and eggs are significant at (P<0.05) and show expected negative signs. These results imply, when own prices of these commodities increase the demand for these commodities decrease with relevant magnitudes. The cross price elasticities for rice and wheat are positive and significant at (P<0.05). The cross price elasticity of wheat with respect to rice is only 0.03 which implies that a 1% increase/decrease in rice price will result in only 0.03% increase/decrease of wheat demand. In other words, wheat demand is not so responsive to price of rice. In contrast, the cross price elasticity of rice with respect to wheat price is 0.26 which means that a 1% increase/decrease in wheat price will result in 0.26% increase/decrease in rice demand. Therefore, manipulation of wheat prices could be effectively used to expand the demand for rice. All expenditure elasticities have expected positive sign and the elasticities of rice, wheat, meat, fish and milk are significant (P<0.05) indicating that these commodities are normally good.
ANALYSIS OF FOOD CONSUMPTION PATTERNS IN SRI LANKA 57
Table 5. Uncompensated price and expenditure elasticities. Food item
Price elasticities Meat Fish
Milk
Eggs
Expenditure elasticities
-0.2312 (0.1198)
-0.2234 (0.1123)
-0.0124 (0.0065)
0.4563* (.1243)
0.0325 (0.0271)
0.1342 (0.0781)
0.2423 (0.1234)
0.0324 (0.0163)
1.2453* (0.5432)
-0.4231* (0.1905)
0.3212 (0.1746)
0.9342* (0.4321)
0.7421 (0.3821)
0.0213 (0.0156)
0.9321 (0.4765)
-0.0421 (0.0214)
0.6321* (0.2987)
-0.9732* (0.3847)
0.3421* (0.0923)
0.2345 (0.1257)
0.3231 (0.1625)
1.2378* (0.5132)
0.3462 (0.1742)
-0.4532 (0.2467)
-0.3231 (0.1627)
0.4321* (0.2012)
-0.9321* (0.3456)
0.3432 (0.1836)
0.4213 (0.2003)
0.5623* (0.2113)
Milk
0.3342 (0.1682)
0.0231 (0.0126)
0.0321 (0.0267)
0.2312 (0.1267)
0.4321 (0.2678)
0.9632 (0.3816)
0.3421 (0.1821)
0.9832* (0.35620
Eggs
0.0213 (0.0112)
0.0123 (0.0068)
0.0232 (0.0127)
0.3233 (0.1627)
-0.5321 (0.2342)
-0.0321 (0.0175)
-0.6321* (0.2638)
0.8452 (0.4786)
Rice
Wheat
Pulses
Rice
-0.6788* (0.1984)
0.0304* (0.0132)
-0.0332* (0.0144)
-0.1321 (0.0751)
Wheat
0.2567* (0.1133)
-0.6743* (0.2362)
-0.0234 (0.0128)
Pulses
0.0932 (0.0467)
0.1322 (0.0662)
Meat
0.8734 (0.4387)
Fish
* significant at (P=0.05).
CONCLUSIONS The general trend of changing food consumption pattern during 1980-2002 had been a decrease of per capita rice consumption and a substantial substitution of rice by wheat with reduction of pulse and milk and an increase in meat and egg consumption. There is a considerable reduction in percentage intake of energy by rice that has been effectively substituted by increased consumption of wheat, and complemented by the increase in energy intake from meat and fish. The percentage expenditure on rice and wheat declined substantially, probably due to the drop in relative prices of both commodities and reduction of per capita consumption of rice. The reduced percentage expenditure on these two commodities has been spent for increasing percentage expenditure on animal products and other foods. The high expenditure elasticity of wheat implies that wheat consumption is likely to increase with the increase of per capita income. The estimate of cross price elasticity of rice with respect to wheat price indicates that reducing the subsidy on wheat may help increase the demand for rice. This study shows that proper planning is necessary for demand and supply expansion of rice, to improve farm economy. ACKNOWLEDGEMENTS
58 HERATH et al.
Authors are thankful to the Department of Census and Statistics, Sri Lanka for providing necessary data to carry out this study. REFERENCES Alderman, H. and P. Timmer. 1980. Consumption parameters for Sri Lankan Policy Analysis. Sri Lanka Journal of Agrarian Studies, (2) 1-3. Bogahawatte, C. and K. Kaillasapathy. 1986. The consumption Patterns of Food in Sri Lanka with Special Reference to Protein- Calorie Nutrition, Agricultural System, (22): 289-303. Deaton, A. and J. Muellbauer. 1980. An Almost Ideal Demand system American Economic review 70: 312-326. Department of Census and Statistics Various years Household Income and Expenditure Survey, Department of Census and Statistics, Sri Lanka. Moschini, G. 1998. The Semiflexible Almost Ideal Demand System European Economics review 42: 346-64. Rathnayake, I.M., N. Priyadharshana and J. Weerahewa. 2004. Structural Change in Demand for Cereals in Sri Lanka, Tropical Agricultural Research, 16: 292-304. Thudawe, I. 2002. Food Demand and Energy Adequacy: Implications for the Poor. Sri Lanka Economic Journal, 2(2): 73-114.