Agribusiness Review - Vol. 6 - 1998

Paper 11
ISSN 1442-6951

Income Strata and Meat Demand in Urban China

Haiou Cai, Colin Brown, Guanghua Wan and John Longworth*
School of Natural and Rural Systems Management
University of Queensland
St. Lucia Campus, Queensland, 4072
Australia
Tel: +617 33654805
Fax: +617 33659016

*Haiou Cai is a Research Assistant, and Colin Brown a Senior Lecturer, at the School of Natural and Resource Systems Management, The University of Queensland. Guanghua Wan is a Senior Lecturer of the Department of Agricultural Economics, University of Sydney, Australia. John Longworth is a Professor and the Executive Dean of Faculty of Business, Economics and Law, The University of Queensland.

Abstract

Rising meat consumption in China has focussed attention not only on China's livestock and grain policies, but also on its impact on international markets for meat and grain. Yet little is known about meat consumption in China. An LA-AIDS model employing purchase data segmented by income class is used to identify price and income elasticities for different groups of Chinese consumers. The magnitude of own price and income elasticities for ruminant meat exceeds the corresponding elasticities for pork and poultry. However, the response to own price and income changes is compounded by the interaction between household income and meat group.

Key Words: China, income strata, meat demand, consumption pattern, elasticities, LA-AIDS, policy.

• I. Introduction
• II. Income Growth and Disparity in China
• III. Price Changes and Price Subsidies
• IV. Meat Consumption Patterns
• V: A Linear-Approximate Almost Ideal Demand System Model and Estimation - Methods
• VI. Data Sources
• VII. Empirical Results
• VIII. Implications
• References

I. Introduction

Profound changes have occurred in recent times in the consumption and production of meat products in various Asian countries. This has been especially the case in China where meat consumption and production has increased almost three-fold during the 1990s. What happens in Chinese meat demand, livestock production and feed grain consumption is of paramount concern to international agricultural and food analysts (see, for example, Brown, 1995 ; Paarlberg, 1997 ; and Wu and Findley, 1997) . This concern has spawned a plethora of sector and trade impact analyses based on aggregate models and, at times, hastily cobbled together demand elasticities. This paper reports on a study seeking to further refine estimates of demand elasticities at the centre of these trade impact analyses.

There are few recent empirical studies on Chinese household meat consumption (see Table 1) . Previous studies have focused on either a broad group of commodities such as food, clothing and housing ( Fan, Wailes and Cramer, 1995 ; Lewis and Andrews, 1989 ) or on whole food groups such as grain, oil and meats (Chern and Wang, 1994) . Specific meat consumption and demand issues were covered only marginally. Furthermore, there has been no attempt to estimate demand elasticities for different levels of income. The conditional demand analysis in this study focuses on the meat subgroup of commodities including ruminant-meat (beef and mutton), pork and poultry-meat differs from the previous complete demand system studies. A priori, income disparity could be expected to bring about significantly different meat consumption and demand patterns. The specific focus on the meat subgroup enables urban household annual per capita purchase data for different income groups to be used to estimate the demand relationships.

The following section highlights the growth in household incomes in China since the 1980s along with the widening disparities in incomes. Changes in meat price levels and subsidies along with changes in meat consumption patterns are then discussed. These developments set the scene for the empirical analyses presented in the latter part of the paper.

Table 1 Selected Recent Studies on the Demand by Chinese Households for Various Meats

NOTES: AIDS stands for almost ideal demand system. LA-AIDS stands for linear approximate almost ideal demand system. LES stand for linear expenditure system. QES stands for quadratic expenditure system.

II. Income Growth and Disparity in China

From the mid-1980s to 1995, the Chinese economy has grown at an average rate of over 8 per cent a year, while the average per capita income growth for urban households has increased at a rate of 18 per cent per annum (SSB, 1996) . Thus, the living standards of the urban population have increased substantially. Associated with this, there has been a dramatic change in consumption patterns. According to official statistics, urban households, on average, spent around half of their total disposable income on food consumption during the last decade (SSB, 1996) . Meanwhile, the urban population has grown rapidly from 19 per cent of total population in 1980 to 30 per cent in 1995, and is expected to reach 50 per cent by 2020 ( SSB, 1996 and Lin et al ., 1996 ).

While the average growth of the overall economy and household income between 1985 and 1995 was substantial, the growth was not the same across regions or among individual households. Kuznets (1955) initially proposed that the relationship between income inequality and the level of economic development is likely to be U-shaped. That is, increasing inequality during the initial move to a modern economy gives way to more equality as a country approaches the more advanced stages of industralisation. Recently, studies have investigated Chinese income inequality. For example, Oi (1993) reported that rural-urban inequalities had narrowed since the 1978 reforms. However, Yang and Wei (1996) found a widening income gap between coastal and interior provinces, particularly in the 1990s. Based on the official statistics, however, the gap between the lowest and the highest income households in urban China does seem to have widened between 1985 and 1995. For example, in 1985, the highest-income-households surveyed accounting for 10 per cent of the total population had over three times more income than the lowest 10 per cent with a gap of 945 yuan, while they had over four times more income than that of the lowest with a gap of 5271 yuan in 1994 (SSB, 1986 and 1995) .

As these variations in average income per person are considerable, patterns of food consumption per person are also not uniform across individual households. Therefore, in estimating income and price effects on meat demand, it is imperative to segment households according to income levels. In this study, households are categorized as low, medium and high income based on urban households' income levels rather than the members of households in each group. At present, greater numbers of households fall into either the medium-income category or low-income class. However, in the future, more-and-more households may fall into the high-income class. It is assumed that in societies such as urban China, where visible changes have occurred in income distribution, meat demand analysis should be based on different income strata rather than on average estimate income variables. Thus, the hypothesis in the study is that consumers from different income strata respond differently to changes in prices and income when purchasing meats.

III. Price Changes and Price Subsidies

Apart from the fact that increases in disposable income tend to generate more consumption of meat, consumer prices are also expected to be another important factor affecting demand for ruminant-meat, pork and poultry-meat. The question is how significant and to what extent both own-prices and cross-prices influence urban household demand for these meat commodities. In response to the economic reforms, consumer prices have changed significantly in the last decade. Prior to the reform, the prices were determined administratively by the government and free market transactions were limited. The rationing of consumer goods especially grain, edible oil and housing generated much excess demand for non-staple food and consumer durables, which also distorted the price mechanism.

After 1985, the pricing system has been fundamentally transformed by abolishing the state monopoly in the purchase of agricultural commodities and developing free markets for meat and other consumer goods. These changes in the urban economy have made urban households almost entirely dependent on markets for their food needs. Over 90 per cent of retail prices and over 80 per cent of agricultural prices have been set by the market since the early 1990s (SSB, 1995) . Meat prices increased rapidly between 1985 and 1995 at an annual rate of 14.5 per cent, while food price as a whole increased by 8.1 per cent per year (SSB, 1996) . Free markets provide a more efficient mechanism for consumers to make rational decisions at the margin in their budget allocation because consumers may find it more satisfactory to purchase other food and consumer goods than grain or oil if the government distributes the amount of the price subsidy back to urban households, perhaps in the form of higher wages. Consequently, consumer utility should be maximized when rationing is removed, and at the same time the money for the price subsidy is distributed back to consumers to increase their income.

Despite the expansion of free markets under the economic reforms, coupons for grain and edible oils were still issued to urban households until 1993 as part of a government subsidy program (all food subsidies for urban households have been removed since early 1993). Within the Chinese food marketing system, pork and beef were occasionally and locally rationed to alleviate the impact of supply instability on the free market price of these products. However, since the reform, the government has gradually removed barriers to free trade and has relied more on markets to determine prices for most commodities. 

IV. Meat Consumption Patterns

The rapid growth in per capita disposable income over the last decade has generally changed the food consumption patterns of urban households. Low-income households are concerned first with consuming enough calories. They then add more variety and quality to their diets as incomes increase. The shift from staple food such as rice to relatively expensive animal products is gradual. In the wealthier groups, one would expect the direct consumption of cereals to fall, and meat consumption to rise rapidly. In China, consumer expenditures on meats have reflected these changing trends in consumption patterns. The proportion of meat expenditures to total food expenditures increased from 14 per cent in 1990 to 24 per cent in 1994. Pork has been the dominant source of animal protein in daily diets with consumption increasing from around 11 kg per person in 1980 to around 17 kg per person in 1990, an increase of 55 per cent. Since 1990, the rate of increase in pork consumption has slowed. Ruminant-meat and poultry-meat have been consumed in smaller quantities, but their rate of increase in consumption has been greater than that of pork.

Different income groups in China have exhibited different meat consumption trends as shown in Table 2 ( SSB, 1986-1996 series; e.g. pp. 286, 1996 ). A wide gap in meat consumption exists among the three groups. The consumption of ruminant-meat by high-income consumers has nearly doubled that of the low-income group over the last decade, while the consumption of poultry-meat by high-income consumers has more than doubled that of the low-income group. High-income households also consume more pork than low-income households, but generally only 50 per cent more. Moreover, the relative proportion of ruminant-meat, pork and poultry-meat in per capita meat consumption has varied across income groups and over time. For example, pork consumption as a proportion of the consumption of all three meat types has varied from 75 per cent for the low income group to 73 per cent for the medium income group and 71 per cent for the high income group. Concurrently, the relative proportions of ruminant-meat and poultry-meat consumption were 9 per cent and 15 per cent for the low-income group, and 11 per cent and 17 per cent for the high-income households. Therefore, meat consumption patterns have varied markedly across different income groups.

Furthermore, it should be noted from Table 2 that where per capita consumption has fluctuated for all meats during the period of 1985 to 1995, that consumption of ruminant meat has varied the most. The recent declines in per capita meat consumption shown in Table 2 may partly be the result of the recent infrastructure reform in urban China. Recently, most households have needed to allocate a significantly greater proportion of their disposable income to housing, medication and education services that have previously been subsidized or provided chiefly by the government or employers. As a consequence, in recent years urban households would have had relatively less to spend on food consumption including meat.

Meanwhile, the Chinese government has made considerable efforts to encourage more consumption of beef and poultry-meat relative to pork as part of its grain conservation policy. Poultry-meat is presumed to have a higher feed conversion efficiency than pork (1kg of poultry-meat requires around 1.7 kg of grain compared with 2.3 kg of grain for 1 kg of pork), while cattle are presumed to use feed resources other than grain such as treated straw and stalk. For instance, the Chinese government's "Straw for Beef" program has seen major investments in cattle production with subsequent effects on beef production, consumption and prices. Furthermore, consumers in urban China have access to a much greater variety of meat products than previously as recent changes and emergence of diversified structures of the supply chain in the Chinese food industry and agro-industralization sector have emerged, including joint-venture enterprises and transnational fast-food stores. More-and-more urban residents now live in modern apartments that promote a wider usage of gas and electricity in household food storage and preparation equipment such as refrigerators, ovens and microwaves. All of these factors have influenced the changing food consumption patterns of urban consumers.

Thus, one would expect variations in meat consumption patterns across different income strata and over time. To explore these issues empirically, an analytical framework is applied to China's urban household survey data collected and complied by China's State Statistic Bureau.

Table 2 Urban Household Annual per Capita Consumption of Meat by Three Income Strata between 1985 and 1995

SOURCE: China Statistical Yearbooks, 1986 to1996 issues (e.g. pp617, 1987; pp286, 1996).

V: A Linear-Approximate Almost Ideal Demand System Model and Estimation - Methods

Selection of an appropriate demand model is crucial to any applied demand study. There are many criteria for demand model selection, but the basic criteria are consistency with economic theory, relative explanatory power, and simplicity of estimation (Wang, Halbrendt and Johnson, 1996) . To estimate meat demand elasticities in urban China, this study employs the almost ideal system (AIDS) developed by Deaton and Muellbauer (1980). This model satisfies the axioms of choice exactly, always-consistent aggregation of individual demands to market demands, and does not impose additive preferences. The system has been applied to both aggregate-level and micro-level data. Owing to these characteristics, this model is probably the most popular demand system applied by contemporary economists (see, for example, Alston, Foster and Green, 1994; Wu and Wu, 1994 ; Fan, Wailes and Cramer, 1995 and Wan, 1996 ).

The model expresses the budget share as a function of prices and real expenditure within the system:

(1)

where w_im is the budget share of the jth commodity{ ruminant-meat( r ), pork( g ) and poultry-meat( c )} for income group m (low-income, medium-income and high-income respectively);

Y_m stands for total expenditure within the system for different income groups;

p_j stands for the commodity prices;

P stands for price index.

The index P is assumed to be a function of the commodity prices:

(2)

This price index yields non-linearity in parameters of the model. To avoid the complication in estimation, Stone (1953) proposed the following that can be used to approximate this price index:

(3)

The resulting model is referred to as the linear approximate almost ideal demand system (LA/AIDS), which can be estimated using Zellner's Iterative Seemingly Unrelated Regression (ITSUR) procedure.

To be consistent with consumer theory, various restrictions are imposed. Since the conditional demand system is expressed as a budget-share, one equation has to be dropped from the system. Excluding one equation automatically implies the adding-up restriction. Thus, in this study, six equations are included consisting of ruminant-meat and pork for three income groups. The homogeneity and symmetry conditions are the only imposed restrictions. Parameters in the omitted poultry-meat equations can be recovered from the adding-up conditions. Thus, the system of 6 equations with 6 restrictions and 11 observations enables 42 degrees of freedoms for all the estimates. Given the above specifications, the income (expenditure) elasticities and uncompensated price elasticities in the AIDS model can be computed using the close approximation of the exact formulae developed by Green and Alston (1990) , which are given by Equations (4), (5) and (6) respectively.

(expenditure elasticity) (4)

(own-price elasticity) (5)

(cross-price elasticity) (6)

VI. Data Sources

The data used in this study are composed of 11 annual observations from 1985-1995 with three income groups. The data have been derived from urban household surveys conducted by the Urban Socio-Economic Survey Organisation within the State Statistical Bureau of China. The State Statistical Bureau classifies the lowest 20 per cent of total households according to their income as the low-income group, the next 60 per cent as medium- income, and the highest 20 per cent as the high-income group. The State Statistical Bureau does not usually publish cut-off points for the different income groups, and classify them in proportions of the total surveyed sample. However, according to the 1991 Yearbook, in 1990 the surveyed households having less than 3000 yuan annually were classified as the low income group, households with income from 3000 to 6000 yuan were classified as the medium income group, while households having over 6000 yuan were regarded as the high income group. These cut-off points would change in other years as incomes rise. The urban household annual per capita purchases of ruminant-meat, pork and poultry-meat were used as the consumption data. Only data during the period of 1985 and 1995 were employed as much of the pre-1985 data were incomplete. The amount of beef and mutton purchased were reported separately only in the last couple of years. Thus, average consumption of ruminant-meat was calculated to be consistent with the data set used. The base year was set as 1985, and retail price indices for ruminant-meat, pork and poultry-meat were calculated sequentially up to 1995.

Ideally, the generation of quality-adjusted prices is needed to ensure that certain qualities of meats consumed by the various groups are incorporated. However, there were no readily available means to adjust the prices for quality variables. As the principal goal of this study was to analyze the effects of income level on household meat consumption behavior, use of the best available price information was considered appropriate.

VII. Empirical Results

Parameter estimates for ruminant-meat and pork were obtained for each income group using the Deaton-Muellbauer iterative procedure. The parameter estimates for poultry-meat were recovered using the adding-up conditions. Most of the parameter estimates were significant at the 10 per cent level of significance (Table 3).

Table 3 Parameter Estimates of the LA/AIDS

Type of Meat Parameter Estimates

Low-Income a i g ir g ig g ic b i

1. *Denotes insignificance at 10 percent confidence.
2. Figures in parentheses are standard errors.
3. na denotes not available.

The principal goal of the study, however, was to analyze the effects of income level on household meat consumption behavior. Thus, the income (expenditure) elasticities and expenditure shares are reported in Table 4 . All income elasticities have positive signs as expected. However, the magnitudes are different among the three income groups of consumers with respect to each type of meat. The income elasticities for ruminant-meat are greater than one, implying that beef and mutton are luxury goods for all urban consumers irrespective of their income levels. However, medium-income households have the greatest income elasticity of 1.75 compared with an income elasticity of 1.1 for the low-income group. This suggests that demand for ruminant-meat would increase greatly but at a different pace for the three-income groups when their income rises. The magnitudes of income elasticities for pork are similar among the three groups, although they are relatively higher for the high-income households compared with the low-income and medium-income consumers.

This is reasonable given the position of pork as the dominant and traditional meat in the diet for most Chinese. For poultry-meat, the low-income group has much higher income elasticities compared with that of the middle income and high-income groups. This suggests that given the relatively lower prices for poultry-meat, low-income households prefer to buy more of the cheaper meat as their income rise. [The retail price for poultry-meat was about 6 yuan per kilo in 1992, compared with over 8 yuan for beef with the pork price being between that for the other two meats (MOA, 1992) .] High-income households have the highest expenditure share for ruminant-meat, while low-income consumers have the highest expenditure share for pork and poultry-meat. This is consistent with the notion that beef and mutton are luxury goods, while pork and poultry-meat are normal goods for urban Chinese households.

Table 4 Expenditure Elasticities and Expenditure Shares by Income Groups

Except for some cross-price elsticities, most of the price elasticities have the expected signs and magnitudes. Uncompensated own-price elasticities are presented in Table 5. All have negative signs in line with economic theory. However, the magnitudes of own-price elasticities of demand vary among the three income groups for different types of meat. Own-price elasticities for ruminant-meat are much higher than those for other meats and greater than one for both low- and medium-income households. This indicates that demand for beef and mutton in urban China is very price elastic.

On the other hand, in terms of pork and poultry-meat consumption, the magnitudes of own-price elasticities are between -0.25 and -0.43, except for pork for the medium-income group and poultry meat for the low-income households. The latter elasticities imply that the demand for pork and poultry-meat by urban households in China is price inelastic. Some of the cross-price elasticities have negative signs, but the magnitudes are very small. In general, the results suggest that own-prices as well as incomes are the predominant factors determining consumer choice and meat consumption patterns in China rather than relative prices.

Table 5 Estimated Own-Price Elasticities for Three Income-Group Households

The results of this study broadly coincide with previous studies where income elasticities ranged from 0.54 to 1.75, and own price elasticities from 0.25 to 1.11 (see Table 1) . The results also fall into the range of income elasticites (0.57-1.0) and price elasticities (0.34-1.04) in South Korea and Japan from previous studies such as Hayes, et al, (1990) and Hayes, et. al, (1991). The Hayes et al . studies were based on 1961-1987 and 1947-1978 average data in South Korea and Japan respectively and also employed an LA-AIDS model. Thus it appears that meat demand and consumption in urban China in the past decade may, in part, be comparable to that in South Korea and Japan during 1960s and 1970s. 

VIII. Implications

China is not only one of the biggest meat producing countries but also one of the largest meat consumers in the world. China's importance in world food markets will continue to grow. This is especially the case in Australian and Asian meat markets of major interest to member countries of APEC. The empirical results of this study suggest several points of interest for researchers, policy makers, planners and traders in these countries. First, income elasticities for ruminant-meat are highly elastic for all income strata suggesting that Chinese urban households will consume more beef and mutton as incomes increase. In terms of poultry-meat, the income elasticity is also highly elastic for the low-income group, implying that Chinese consumers with low incomes will increase their consumption of poultry-meat as their incomes rise. Second, own-price elasticities of all meat items for mainly low-income and medium-income households are fairly elastic. This suggests that any changes in meat prices could bring about a significant shift in meat consumption patterns. Third, given the emergence of large urban unemployment, a major challenge confronting the government is how to design appropriate policies for the relative enhancement of low-income groups. Identifying elasticities for different income groups enables Chinese decision-makers to gauge precisely the impact of their policies on various income groups, and so better design policies targeted at low-income groups.

The strength of study relative to previous meat demand studies in China is the use of observations pertaining to different income groups rather than average estimates for the population as a whole. A further partitioning of income groups with longer time series data and incorporating socio-demographic variables would further enhance this study. Caution should be taken when interpreting those empirical results since the statistical information on consumption data in China is rather scarce, incomplete and controversial. These data problems limit strong interpretation of empirical findings. Nevertheless, this study opens up discussion on the important issue of consumption patterns for different income strata in urban China. Further studies will enhance the potency of these preliminary findings. Only through more disaggregated analysis can the information needed to design appropriate policies for different socio-income and demographic groups be effected.

REFERENCES

Brown, L.R . Who Will Feed China? Wake-up Call for a small Planet . W.W. Norton and Company. New York, 1995

Chern, W. S . and Wang, G. "The Engel Function and Complete Food Demand System for Chinese Urban Households". China Economic Review 1994, 4 , 35-57.

Deaton, A . and Muellbauer, J. "An Almost Ideal Demand System". The American Economic Review 1980, 70 , 312-326.

Dixit, P . M.and Webb, S. E. H. "Changes in China's Meat Consumption Patterns: Implications for International Grain Trade". In Issues in Agricultural Development ; M. Bellamy and B. Greenshelds, eds. International Association of Agricultural Economists, 1992; pp 159-167.

Fan, S ., Wailes, E. J. and Cramer, G. L. "Household Demand in Rural China: A Two-Stage LES-AIDS Model". American Journal of Agicultural Economics 1995, 77 , 54-62.

Fan, S., Cramer, G.L. and Wailes, E.J. "Food Demand in Rural China: Evidence from Rural Household Survey". Agricultural Economics 1994, 11, 61-69.

Gao, X. and Cramer, G. L. "Partial Rationing and Chinese Urban Household Food Demand Analysis". Journal of Comparative Economics 1996, 22 , 43-62.

Gao, X., Wailes, E. and Cramer, G. "A Two-Stage Rural Household Demand Analysis: Microdata vidence from Jiangsu Province, China". American Journal of Agricultural Economics 1996 , 78, 604-613.

Green, R .; Alston, J. M. "Elasticities in AIDS Models". American Journal of Agricultural Economics 1990, 72 , 442-445.

Halbrendt , C., Tuan, F., Gempesaw, C. and Dolk-Etz, D. "Rural Chinese Food Consumption: the Case of   Guangdong". American Journal of Agricultural Economics 1994, 76 , 794-799.

Hayes, D.J ., Ahn, H. and Baumel, C.P. "Meat Demand in South Korea: A Systems Estimate and Policy Projections". Agribusiness 1991, 7 , 433-446.

Hayes, D.J., Wahl, T.I. and Williams, G.W. "Testing Restrictions on a Model of Japanese Meat Demand". American Journal of Agricultural Economics 1990, 73 , 556-565.

Kuznets , S. "Economic Growth and Income Inequality". American Economic Review 1955, 45, 1-28.

Lewis, P . and Andrews, N. "Household Demand in China". Applied Economics 1989, 21, 793-807.

Lin, J ., Huang, J.K. and Rozelle, S. "China's Food Economy: Past Performance and Future's Projection". Working Paper, Beijing, China Center for Economic Research, 1996, 1-29

Meng, X . and Wu, H. X. "Household Income Determination and Regional Income Differential in Rural China," Chinese Economy Research Unit, The University of Adelade,, 1995.

MOA , "Market Price of Animal Husbandry Products", The Information Center, 1992.

Paarlberg, R.L . "Feeding China: A Confident View". Food Policy 1997, 22 (3), 269-279.

Park, J. L ., Holcomb, R. B., Paper, K. C. and Capps, O. J. "A Demand Systems Analysis of Food Commodities by U.S. Households Segmented by Income". American Journal of Agricultural Economics 1996, 78 , 290-300.

Qi, J.C . "Reform and Urban Bias in China". Journal of Development Studies 1993, 29, 129-148

SSB , China Statistical Yearbooks . China Statistical Information and Consultancy Service Centre: Beijing, 1981-1996.

Stone, J.R . The Measurement of Consumers' Expenditure and Behavior in the United Kindom, 1920-1938 . bridge, 1953, Vol. 1. The World Bank. China: Options for Reform in the Grain Sector. Washington DC, the World Bank. 1987.

Wan, G. H . "Income Elasticities of Household Demand in Rural China: Estimates from Cross-sectional Survey Data". Journal of Economic Studies 1996, 23 , 18-33.

Wang, Q. B ., Halbrendt, C. and Johnson, S. R. "A non-tested test of the AIDS vs. the translog demand system". Economics Letters 1996, 51 , 139-143.

Wu, H.X . and Findley, C. "China's Grain Demand and Supply: Trade Implications". Chinese Economy Research Unit, The University of Adelaide, 1997.

Wu, Y. R . and Wu, X. H. "Household Grain Consumption in China: Effects of Income, Price and Urbanisation". Chinese Economy Research Unit, The University of Adelaide, 1994.

Yang, D . and Wei, H. "Rural Enterprise Development and Regional Policy in China". Asian Perspective 1996, 35, 1.