the effect of self-sufficiency on fruit and vegetable consumption of china's rural households

Agricultural & Applied Economics Association

The Effect of Self-Sufficiency on Fruit and Vegetable Consumption of China's RuralHouseholdsAuthor(s): Tong Han, Thomas I. Wahl and Ron C. MittelhammerSource: Review of Agricultural Economics, Vol. 23, No. 1 (Spring - Summer, 2001), pp. 176-184Published by: Oxford University Press on behalf of Agricultural & Applied Economics AssociationStable URL: http://www.jstor.org/stable/1349914 .

Accessed: 25/06/2014 07:20

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

Agricultural & Applied Economics Association and Oxford University Press are collaborating with JSTOR todigitize, preserve and extend access to Review of Agricultural Economics.

http://www.jstor.org

This content downloaded from 91.229.229.101 on Wed, 25 Jun 2014 07:20:32 AMAll use subject to JSTOR Terms and Conditions

http://www.jstor.org/action/showPublisher?publisherCode=oup

http://www.jstor.org/action/showPublisher?publisherCode=aaea

http://www.jstor.org/stable/1349914?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp


Review of Agricultural Economics-Volume 23, Number 1-Pages 176-184

The Effect of Self-Sufficiency on Fruit and Vegetable Consumption of China's Rural Households

Tong Han, Thomas I. Wahl, and Ron C. Mittelhammer

Multivariate analysis of variance and econometric analysis are used in conjunction to analyze China's rural household consumption behavior across households that differ in the level of self-sufficiency relative to the production/purchase of fruits and vegetables. There is a significant difference in consumption patterns for different households with different levels of self-sufficiency. Households that purchase but do not produce fruits and vegetables exhibit a notably more market-oriented price responsive behavior. The consumption decisions of households that produce fruits and vegetables, but do not purchase them, are influenced more heavily by implicit income effects than by implicit price effects.

China has consistently been an important market for the world's agricultural

exporters and especially for U.S. exporters, since the mid-1980s. Total U.S. exports of agricultural products to China have increased fourfold and exports of processed food products to China have increased eightfold since the early 1990s. U.S. exports of fruits and vegetables to China have increased even more rapidly, with average annual increases of more than 50% per year since the early 1990s. Given the importance of the Chinese market to the world market, and the United States in particular, projections of China's future food needs and changes in food consumption patterns have become increasingly important. Nearly all projections of China's future food demand predict an increase in food imports to meet the growing demand that is likely to result from an expanding population base and rapidly improving income levels.

0 Tong Han and Thomas Wahl are, respectively, research assistant and associate professor, Department of Agricultural Economics, Washington State University, Pullman. * Ron Mittelhammer is a professor of Agricultural Economics and of Statistics at Washington State University.



Self-Sufficiency and Fruit and Vegetable Consumption in China 177

China's rural economy was 62.3% self-sufficient in food consumption before 1978.1 By 1980, rural households produced less than 48.3% of the food they consumed. After China's rural economic reform, which began in the mid-1980s, self-sufficiency in rural households decreased further. By 1993, rural households purchased more than 63.7% of the food they consumed. Rural household self- sufficiency in fruits and vegetables decreased from 78.2% in 1980 to 51.3% in 1993 (Han).

A number of studies have examined food demand in China at the provincial, regional, and national levels (e.g., Kueh; Lewis and Andrews; Huang and Rozelle; Halbrendt et al.; Fan, Cramer, and Wailes; Fan, Wailes, and Cramer; Guan; Liu, Wahl, and Mittelhammer). In addition, two studies have used a two-stage budgeting LES-LA/AIDS system to estimate rural household food consumption across developed, developing, and undeveloped regions of China (Han) and across high, middle, and low income groups (Han and Wahl). The results of these studies indicate that different income groups have essentially the same consumption patterns and behavior. However, economic theory and intuition would suggest that households from different regions or income levels may have different demand functions. To the extent that previous studies have not fully addressed important factors affecting household behavior, further differentiating households by level of food self-sufficiency may contribute to a more accurate representation of rural household consumption patterns and result in better estimates of price and income effects.

This study examines the effects of levels of household food self-sufficiency on the response of households to market prices and income. Households are grouped into four categories based upon whether they purchased and/or produced fruits and vegetables. In 1993, the four groups (and the respective percentage of sample households in each group) were: group 1-households producing and purchasing fruits (85%) and vegetables (20%); group 2-households producing but not purchasing fruits (13%) and vegetables (11%); group 3-households not producing but purchasing fruits (2%) and vegetables (69%); and group 4-households not producing and not purchasing fruits and vegetables. The first three groups are relevant to this study because every household consumed at least some fruits and vegetables.

This study focuses on food consumption levels and, especially, the consumption patterns for fruits and vegetables of China's rural households across the aforementioned different self-sufficiency groups. The goal of this study is to analyze the extent to which these different groups of China's rural households have different consumption behavior and whether they do or do not share common demand behavior. In addition, the issue of whether expenditures on fruits and vegetables are responsive to increases in overall food expenditures is examined. Estimates of price and expenditure elasticities for fruits and vegetables are provided.

To analyze the degree to which consumption decisions respond to market prices, various models of the relationship between two responses, per capita consumption and expenditure share of fruits and vegetables, and a set of explanatory variables are examined. A two-stage budgeting model is used to estimate a complete demand system. The first stage includes demand for food, clothing, housing, durable goods, and other items. Clothing includes cotton, synthetic fiber, wool, silk and satin, cloth knitwear, clothing, and household fabrics such as bed



178 Review of Agricultural Economics

sheets and shoes. Durable goods include bicycles, sewing machines, clocks, wrist watches, electric fans, washing machines, refrigerators, sofas, wardrobes, writing desks, radios, television sets, VCRs, tape audio recorders, and cameras. Other items consist of the daily consumption of all goods and services (e.g., fuel and energy, education and entertainment, health care, etc.) other than food, clothing, housing, and durable goods.

In the second stage, the demands for grain, meat, leafy vegetables, root vegetables, other vegetables, dried vegetables, apples, grapes, other fruits and nuts, and other food is estimated. Leafy vegetables include Chinese cabbage, cabbage, spinach, rape, celery, fragrant flowered garlic, green Chinese onion, hol- low cabbage, asparagus, lettuce, and garlic bolt; root vegetables include radishes, carrots, ginger, and lotus root; other vegetables include cauliflower, cucumbers, wax gourd, onions, towel gourd, tomatoes, eggplant, green peppers, bean sprouts, fresh kidney beans, and others; dried vegetables include day-lily, edible fungus, Tremella, xianggu mushrooms, and others. Other fruits and nuts include pears, bananas, oranges, pineapples, peaches, persimmons, nuts, and kernel products.

The Models and Methods of Statistical Analysis Four multivariate statistical tests (Johnson and Wichern), including Roy's Maxi-

mum Root, Wilks' Lambda, Pillai's Trace, and Hotelling-Lawley Trace are used in the context of a multivariate analysis of variance (MANOVA) to test whether different self-sufficiency groups of households have different consumption patterns and, in addition, whether they are responsive to market prices and income. Two dependent variables, per capita consumption and expenditure share, the latter based upon market prices for fruits and vegetables, and a set of explanatory variables that includes logarithmic prices of grain, meat, other food, leafy, root, dried and other vegetables, apples, grapes, other fruit, plus income, self-sufficiency categories of the households, and geographic indicator variables, are used in the tests. The Wilks' Lambda test statistic, A, defined as:

(1) A Ee Ee + Hrl'

where Ee is the matrix of sum of square and cross-products (SSCP) not due to regressions (referring to estimated the residuals of the regressions), and Hr is the regression SSCP matrix. Wilks' Lambda acts as an index of how much vari- ability in the vector of dependent variables is due to regression, and assesses whether there is a significant association between the set of dependent variables and the explanatory variables. The other three tests are similar in objective to Wilks' Lambda but use different critical values in making test decisions. When the full contingent of explanatory variables is used, the overall null hypothesis amongst all of the tests is that there are no commodity price, household income, geographics, region, and sufficiency categories effects across different commodities and different self-sufficiency groups, a hypothesis which is a priori expected to be rejected. The effect of each explanatory variable is also assessed in a marginal contribution context and it is here where the marginal effect of the level of self- sufficiency is assessed.




A two-stage budgeting procedure is used to model the consumer's utility max- imization decision, which assumes the decision can be decomposed into two sep- arate stages. In the first stage, total expenditure is allocated across broad groups of expenditures using a linear expenditure system (LES). The advantage of the LES is that it is a parsimonious specification that provides an intuitive economic interpretation, albeit it incorporates relatively strong separability assumptions. However, the separability assumption is not overly restrictive for such broad commodity groups as food, housing, education, or clothing (Timmer and Aldermand). The general LES specification representing expenditure on group I is given by

(2) PIXI = PIRI

+ I3[Y

- E PIRI].

The LES functional form includes the parameter constraints 0 < P, < 1, LI [i = 1, and X1 > R1, where PIXI is expenditure on group I (PI and XI are aggregated price and quantity indices for commodities within group I), RI and P, are parameters and Y is total household expenditure.

In the second stage, group expenditures are allocated across individual commodities by using the Linear Approximate/Almost Ideal Demand System (LA/AIDS). Weak separability of the direct utility function across broad groups is both a necessary and sufficient condition for estimating the second stage of the two-stage budgeting procedure. The model applied to microlevel data. The share equation for the LA/AIDS model is

(3) Wi,I

= ai, I + E ij, ilog

pj, i+~+i, ilog( Y

,

where wi, I is the budget share of good i in commodity group I, Pi, I is the price of commodity j in commodity group I, YI is the Ith group's total expenditure, and PI is the Ith group's price index.

The two-stage budgeting model was estimated for the aforementioned three different self-sufficient groups and for fruits and vegetables, respectively. To obtain consistent and asymptotically efficient estimates in the presence of zero consumption, a two-step estimation procedure is employed following Heien and Wessells. In particular, an inverse Mill's ratio is computed for each household in the first step, which is then used as an instrument in the second-stage.

Equation (1) and the related multivariate statistical tests are used to test whether different self-sufficiency groups of households have different consumption patterns. Equations (2) and (3) are used to estimate elasticities of demand for fruits and vegetables across different groups. The SAS GLM procedure is used to compute the multivariate MANOVA statistics. A nonlinear seemingly unre- lated regression (SUR) estimator is used to estimate the LES model in the first stage and a linear SUR estimator is used to estimate the linear AIDS in the second stage. Adding up, homogeneity, and symmetry restrictions are imposed on the food group in the second stage. The same model is estimated for each of the three different self-sufficient groups of China's rural households. An aggregated (across self-sufficient groups) consumption analysis based on the entire national




subsample is also estimated in order to compare aggregated price and expenditure elasticities with the differentiated set of three self-sufficiency categories of households.

Data All major economic activities for 66,960 participating rural households were

recorded by the National Rural Household Survey, conducted by the General Organization for Rural Household Surveys of the State Statistical Bureau (SSB) in 1993.2 The sample contains information on the rural households' income, expenditure, production, and consumption, as well as their demographic characteris- tics. A 10% random subsample of the survey observations is used in this study. Five broad groups of goods-food, clothing, housing, durable goods, and other items-are used in the first stage of the two-stage analysis. In the second stage, food expenditures are allocated among the commodities used in the first stage analysis, including food, clothing, housing, durable goods, and all other commodities. The second stage analysis of the food group included the following food items: grains, meat, others (stimulants, sweets, and cooking oils), leafy vegetables, root vegetables, other vegetables, dried vegetables, apples, grapes, other fruits and nuts, and other food. Mean prices were used to proxy missing prices.

Quantities consumed and commodity expenditures are used to infer implicit values for individual commodities. In particular, prices paid by individual households were derived by dividing expenditures by quantities. For households in self-sufficient group 2 that did not purchase fruits and vegetables, average prices at the village or county in which the observation occurred are used. The aggregated prices for the goods in both the first and second stages of the analysis were computed using Stone's price index with expenditure shares used to weight the commodities contained within each group category.

Estimation Results The results of the Wilks' Lambda test, applied marginally to the various

explanatory variables, are presented in table 1. Roy's Maximum Root, Pillai's Trace, and Hotelling-Lawley Trace results are not presented, but all support pre- cisely the same final conclusions as the Wilks' Lambda test. The null hypothesis of no commodity price, income, geographic, or self-sufficiency effect is for the most part rejected at conventional levels of significance for each of the commodities. In particular, significant self-sufficiency effects suggest that there are significant differences in consumption patterns between China's rural households which are producing and purchasing fruits and vegetables (group 1); households producing and not purchasing fruit and vegetables (group 2); and households not producing but purchasing fruits and vegetables (group 3).

Unconditional price and expenditure elasticities for major food items among the three household self-sufficient groups are presented in table 2. Demand for all fruits and vegetables for household self-sufficient group 3, which is purchasing but not producing, is more elastic with respect to price than for group 1, which is both purchasing and producing. Expenditure elasticities for leafy, root, other fresh vegetables, and other fruit in group 3 are all higher than for group 1, suggesting that group 3 exhibits more income-responsive market behavior than does group 1.



Table 1. MANOVA test criteria and F approximations for the hypothesis of no independent variable effect for fruits and vegetables

Other Leafy Veg. Root Veg. Other Veg. Dried Veg. Plains Northern Household Grain Price Meat Price Food Price Price Price Price Price Income Area Region Category

Leafy 0.992 0.981 0.983 0.902 0.998 0.988 0.163 0.747 0.969 0.980 0.957 vegetables (0.0001) (0.0001) (0.0001) (0.0001) (0.0001) (0.0001) (0.0001) (0.0001) (0.0001) (0.0001) (0.0001) Root 0.999 0.996 0.993 0.999 0.930 0.999 0.168 0.888 0.969 0.996 0.998 vegetables (0.0244) (0.0001) (0.0001) (0.1256) (0.0001) (0.137) (0.0001) (0.0001) (0.0001) (0.0001) (0.0523) Other 0.996 0.984 0.997 0.998 0.997 0.861 0.168 0.771 0.987 0.935 0.992 vegetables (0.0002) (0.0001) (0.0008) (0.0020) (0.0002) (0.0001) (0.0001) (0.0001) (0.0001) (0.0001) (0.0001) Dried 0.998 0.999 0.999 0.999 0.999 0.999 0.999 0.948 0.998 0.992 0.998 vegetables (0.0159) (0.4549) (0.591) (0.6058) (0.6215) (0.4244) (0.2381) (0.0001) (0.0302) (0.0001) (0.099) Apples 0.999 0.998 0.999 0.956 0.999 0.997 0.959 0.997 0.963 0.993 0.993

(0.3108) (0.0259) (0.0837) (0.0001) (0.3116) (0.0024) (0.0001) (0.0012) (0.0001) (0.0001) (0.0001) Grapes 0.998 0.999 0.999 0.993 0.999 0.999 0.996 0.997 0.999 0.999 0.999

(0.0266) (0.8553) (0.2742) (0.0001) (0.9069) (0.6946) (0.0001) (0.0003) (0.4337) (0.0001) (0.0001) Other fruit 0.997 0.996 0.999 0.927 0.993 0.998 0.929 0.992 0.999 0.997 0.997

(0.0005) (0.0001) (0.8270) (0.0001) (0.0001) (0.0292) (0.0001) (0.0001) (0.3620) (0.0019) (0.0019)

Note: Numbers in parentheses are p values.

(I-i

ri~h

0

00



Table 2. Comparison of unconditional own-price and expenditure elasticities across different self-sufficient groups in China's rural households, 1993

Own-Price Elasticities Expenditure Elasticities

Group Ia Group 2 Group 3 National Group 1 Group 2 Group 3 National

Grain -0.759 -0.943 -0.586 -0.784 1.002 1.065 1.138 1.092 Meat -0.146 -0.763 -0.338 -0.219 0.529 0.779 0.300 0.462 Otherb -0.614 -0.819 -0.284 -0.440 0.772 0.856 0.611 0.692 Leafy vegetables -0.272 -1.511 -0.917 -0.182 1.403 0.653 1.695 1.429 Root vegetables -0.578 -0.302 -0.878 -0.623 1.160 1.287 1.437 1.318 Other fresh vegetables -0.421 -0.968 -0.775 -0.863 0.011 0.742 0.810 0.621 Dried vegetables -0.199 -0.114 -0.995 -0.389 0.955 0.867 0.598 0.748 Apples -0.627 -0.928 -0.633 -0.693 0.683 0.945 0.658 0.667 Grapes -1.148 -1.017 -1.356 -1.042 0.942 0.884 0.931 0.903 Other fruitc -0.899 -1.759 -1.271 -1.011 0.129 0.922 0.387 0.206

aGroup 1-producing and purchasing, Group 2--producing and not purchasing, and Group 3-not producing and purchasing. bOther includes sweets, stimulants, and cooking oil. cOther fruit includes bananas, oranges, pineapples, peaches, persimmons, watermelons, nuts, kernel products, and others.

too

c,-

trl 0

0r




Demand for a number of fruits and vegetables by group 1 is considerably less elastic with respect to price than is indicated for the overall aggregate (national) group. Expenditure elasticities for group 1 and the overall aggregate group (national) are in most cases similar. This result suggests that there is less market price responsive behavior in group 1 in a number of instances. Group 2, which produces but does not purchase, has the highest own implicit price elasticities for leafy (-1.51), other fresh vegetables (-0.97), apples (-0.93), and other fruits (-1.76) among the three groups. Demand for root (-0.30) and dried (-0.11) vegetables in group 2 is less elastic with respect to price than in group 1 or 3. Group 2 consumes more fruits and vegetables at lower market prices, and supplies more fruits and vegetables at higher prices. Group 2 has relatively high own-price elasticities for perishable fruits and vegetables (i.e., apples, other fruit, and leafy and other fresh vegetables). This result may be reflecting large responses to market prices during harvest, whereby households in group 2, who produce but do not purchase, are influenced more heavily by implicit income consequences of consumption versus selling.

We note that region is also a major contributing factor for explaining differences in food consumption behavior across self-sufficiency categories. This result was not surprising because it was known a priori that households in the plains area are more likely to purchase fruits and vegetables than produce them.

Conclusions The empirical results provided via multivariate statistical analysis and econo-

metric analysis appear to provide consistent conclusions regarding the heterogeneity of China's rural household fruit and vegetable consumption behavior across different self-sufficiency categories of China's rural households relative to production/purchase of fruits and vegetables. China's rural households that produce and purchase fruits and vegetables (group 1), produce but do not purchase fruits and vegetables (group 2), and do not produce but purchase fruits and vegetables (group 3) exhibit notably different consumption behaviors. Moreover, the consumption decisions of all three household self-sufficiency groups appear to be influenced by both income and prices.

The results demonstrate that there is a substantial heterogeneity in consumption behavior for different households with different levels of self-sufficiency. The estimated national level price and income elasticities are, except for one case (price elasticity for leafy vegetables), within the range of the household category estimates. The price elasticities among the three groups range from -0.114 to -1.511 for vegetables and from -0.627 to -1.759 for fruit. The expenditure elasticities among the three groups range from 0.011 to 1.695 for vegetables and from 0.129 to 0.945 for fruit. The range of these estimates is much larger than studies that have focused on different income groups or regions alone. It therefore appears that capturing the effects of the household self-sufficiency categories should be an important consideration in any future modeling of the food consumption behavior of rural Chinese households.

Endnotes SSelf-sufficiency in food consumption is measured as household food production relative to total

household food consumption, which includes both produced and purchased food.




2 China's official aggregate data reports have historically suffered from data discrepancies, especially reported total acreage. See Rozelle; Han, Wailes and Cramer; or Renan for a discussion of Chinese data collection procedures and problems.

References Fan, S., G.L. Cramer, and E. Wailes. "Food Demand in Rural China: Evidence from Rural Household

Survey." Agr. Econ. 11(1994):61-9. Fan, S., E. Wailes, and G.L. Cramer. "Household Demand in Rural China: A Two-Stage LES-AIDS

Model." Am. J. Agr. Econ. 77(1995):847-58. Guan, X. "Food Consumption in Urban Coastal China: A Cross Sectional Analysis." Unpublished MA

thesis. Washington State University, 1996. Halbrendt, C., F. Tuan, C. Gempesaw, and D. Dolk-Etz. "Rural Chinese Food Consumption: The Case

of Guangdong." Am. J. Agr. Econ. 76(1994):794-808. Han, T. "Rural Household Food Consumption in China: Evidence from the Rural Household Survey."

Unpublished MA thesis. University of Arkansas, 1996. Han, T., and T.I. Wahl. "China's Rural Household Demand for Fruit and Vegetables." J. Agr. Appl.

Econ. 30(July 1998):141-50. Han, T., E. Wailes, and G. Cramer. "Rural and Urban Data Collection in the People's Republic of

China." In: The China Market: Data and Information Systems, Proceedings of WRCC-101, IMPACT Center, Washington State University, 1995.

Heien, D., and C.R. Wessells. "Demand System Estimation with Microdata: A Censored Regression Approach." J. Bus. Econ. Stat. 8(1990):365-71.

Huang, J., and S. Rozelle. "Rural Consumption after Liberalization: Income, Quality and the Demand for Food in China." Research Paper, University of California at Davis, 1990.

Johnson, R., and D.W. Wichern. "Applied Multivariate Statistical Analysis." 2nd ed., Englewood Cliffs: Prentice Hall, 1992.

Kueh, Y.Y. "Food Consumption and Peasant Incomes in Post-Mao China." China Quality 100(March, 1988):634-70.

Lewis, P., and N. Andrews. "Household Demand in China." J. Appl. Econometrics 21(1989):793-807. Liu, H., T.I. Wahl, and R.C. Mittelhammer. "A Fruit Demand Model for Urban Chinese Consumers."

Unpublished manuscript. Department of Agricultural Economics, Washington State University, 1996.

Renan, Z. "The Information System of the Ministry of Agriculture of the People's Republic of China." In: The China Market: Data and Information Systems, Proceedings of WRCC-101, IMPACT Center, Washington State University, 1995.

Rozelle, S. "Rural Household Data Collection in Reform China." In: The China Market: Data and Infor- mation Systems, Proceedings of WRCC-101, IMPACT Center, Washington State Unversity, 1995.

Timmer, P., and H. Aldermand. "Estimating Consumption Parameters for Food Policy Analysis." Am. J. Agr. Econ. 61(1979):982-7.

United Nations, Food and Agricultural Organization (FAO). "Demand Prospects for Rice and Other Food Grains in Selected Asian Countries." FAO Economic and Social Development Paper, No. 97 (1991).



the effect of self-sufficiency on fruit and vegetable consumption of china's rural households

Documents