QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Guo, X. Articles by Zhai, F. Search for Related Content PubMed PubMed Citation Articles by Guo, X. Articles by Zhai, F.

---

Articles

Food Price Policy Can Favorably Alter Macronutrient Intake in China¹

Xuguang Guo, Barry M. Popkin², Thomas A. Mroz^* and Fengying Zhai

Department of Nutrition, School of Public Health and Carolina Population Center, ^* Department of Economics and Carolina Population Center, University of North Carolina at Chapel Hill, NC 27516–3997 and Institute of Nutrition and Food Hygiene, Chinese Academy of Preventive Medicine, Beijing, China

²To whom correspondence and reprint requests should be addressed.

	ABSTRACT

TOP ABSTRACT INTRODUCTION SUBJECTS AND METHODS RESULTS DISCUSSION REFERENCES

 
The rapid change in diets, physical activity and body composition in low income countries has led to the coexistence of large pockets of undernutrition and overnutrition. Public health strategies for addressing this situation may be necessary, and price policy options are examined for China. Longitudinal dietary data collected in China in 1989–1993 on a sample of 5625 adults aged 20–45 y were examined. Three-day averages of food group consumption and nutrient intake were used in longitudinal statistical models to examine separately the effects of food prices on the decision to consume each food group and then the amount consumed. The effects of changes in six food prices on the consumption of each of six food groups, not just the food group whose price had changed, and on three macronutrients were estimated. The effects show large and significant price effects. If the joint effects of the nutrition transition are to be considered, then there are clear tradeoffs among which foods to tax and which to subsidize. Most important is the effect of prices in reducing fat intake of the rich but not adversely affecting protein intake for the poor. Increases in the prices of pork, eggs and edible oils are predicted to lower fat intake. Only increases in pork prices led to reduced protein intakes. This raises questions about earlier policy changes being implemented in China and provides insight into an important and controversial area for public health policy.

KEY WORDS: • the nutrition transition • food price policy • dietary intake • longitudinal analysis • China

	INTRODUCTION

TOP ABSTRACT INTRODUCTION SUBJECTS AND METHODS RESULTS DISCUSSION REFERENCES

 
The nutrition transition in many low income countries is associated with an increasing polarization of the population into those faced with obesity and other diseases linked with overnutrition and those faced with problems linked to dietary deficit (Popkin et al. 1995 ). China is a classic example of this situation, but many others exist (Popkin 1993 and 1998 ). Over the past two decades, China has achieved remarkable economic progress. From 1979 to 1987, income per capita quadrupled in rural areas and tripled in urban sites (Pinstrup-Andersen et al. 1990 ). Accompanying these changes was a rapid improvement of food supply and consumption. The emerging food consumption pattern represents a marked shift not only toward more food products perceived to provide a higher quality diet but also toward higher fat food and more processed food products. One result has been the marked reduction of undernutrition as a public health problem and the emergence of dietary excess and obesity as problems (Popkin et al. 1993 ). The shift in diet and the emerging gap between the public health needs of the rich and middle income people represent a new and complex issue. In China and most other lower income countries, excess fat intake and obesity are problems of the rich, whereas the poor face continued problems of food insecurity and undernutrition. As shown elsewhere, even the poor are beginning to be faced by these problems of dietary excess (Monteiro et al. unpublished results). Few studies have explored program or policy options for addressing the macronutrient component of this transition at the national level. This study explores the potential benefits of food price policy for favorably altering the macronutrient intake pattern in China.

The remarkable transition of the Chinese economy has led to specialized shifts in government price policies concerning grains, livestock and processed commodities such as edible oil. Since 1988, the Chinese government has initiated a series of price policies for gradual abolition of government grain procurement and urban rationing systems. Because price polices regarding income support and poverty alleviation have different effects across different income groups, this study investigates the effects of food prices on dietary intake patterns of the rich and the poor. This is done with longitudinal analysis. There is a strong literature demonstrating that behaviorally and statistically, the properties of longitudinal analysis of consumption relationships with income and price are more appropriate for policy analysis. With longitudinal data, one can exploit the fact that there could be unmeasured characteristics influencing individuals' consumption decisions through time (Kennedy 1992 , Maddala 1988 , Wonnacott and Wonnacott 1979 ).

This study presents a policy analysis concerning how food price changes can affect dietary intake. It estimates the price effects on diet (presented as elasticities or the effect of a 1% change in price on the percentage change in dietary intake). The overall price elasticities are estimated for different income populations.

	SUBJECTS AND METHODS

TOP ABSTRACT INTRODUCTION SUBJECTS AND METHODS RESULTS DISCUSSION REFERENCES

 
Study sample.

Data for this application came from the household and individual components of the first three rounds of the China Health and Nutrition Survey (CHNS).³ The CHNS was designed as a longitudinal survey to allow a strict temporal ordering of presumptive causes and effects in statistical analysis. Formal details of the study and data have been given elsewhere (Popkin et al. 1993 , Zhai et al. 1996b ). Briefly, 3780 households, randomly selected from a sampling frame, consisting of 64 neighborhoods in urban areas and 126 villages in rural areas, were followed in 1989, 1991 and 1993. Of the 3780 households originally surveyed in 1989, 4.5% were lost to follow-up in 1991 and an additional 4.5% were lost in 1993, largely because of migration. A fourth survey was collected in 1997, but the data are not yet ready for use.

All adults aged 20–45 y with multiple-day, dietary data in 1989 were included in this study. There were 5625 adults with complete socioeconomic data and consecutive 3-d dietary recalls at the base year. Subjects in this age cohort were followed up in two successive surveys. Of them, 1055 and 853 adults were lost to follow-up in 1991 and 1993, respectively. Correspondingly, 823 and 491 subjects were recruited and/or returned to this cohort. As mentioned above, the sampling units in the CHNS were households rather than randomly selected individuals. Thus, the element of variance may be inflated regardless of whether these households were randomly or nonrandomly selected (Kish 1965 ). In a preliminary analysis, we used regression with Huber's correction of standard errors (Huber 1967 ) to examine this effect. The analysis showed that the effect was approximately equivalent to the study design (Paeratakul et al. 1998b ). Thus, this effect of nonrandomly selected individuals was ignored in the further analysis. The final sample for longitudinal analysis consisted of 6667 individuals, with 16,049 dietary measurements over the 4-y period. The sample frame for each cross-sectional survey and the longitudinal analysis is shown in Figure 1 .

View larger version (27K): [in this window] [in a new window]   Figure 1. Sample size changes from baseline to 4-y follow-up in Chinese men and women, aged 20–45 y according to the China Health and Nutrition Survey, 1989–1993.

All field work was completed by trained public health workers who were professionally engaged in the nutrition surveys at the provincial and subprovincial levels (for further details see Zhai et al. 1996a ). Detailed food consumption data were collected on household and individual levels for three consecutive days. Individual dietary data were obtained by 24-h dietary recall in combination with a weighing and measurement technique. A method was developed to obtain an accurate estimate of oil consumption on the basis of the proportion of animal products (including meat, fish, eggs and their products) and vegetables consumed by each individual in a household. Each individual's proportion of the total household meat and vegetables was used to allocate household cooking oil to each individual. The amount of oil allocated was added to the 24-h dietary recall to estimate individual dietary intakes.

Statistical analysis.

    Dependent variables. Six food groups and three macronutrients were selected as dependent variables for their importance in reflecting food behavior. The food groups were rice, wheat flour, coarse grain, pork, eggs and edible oils. The representative items in each food group were reported elsewhere (Guo et al. 1999 ). Macronutrients included total calories, protein, fat, the proportion of the Recommended Daily Allowances (RDA) for energy and protein, and the percentage of calories from fat. Measures of these nutrients were based on the average daily intake of 3 d of individual dietary data (Popkin et al. 1993 , Zhai et al. 1996a ). Table 1 presents the proportion of the population consuming each food group, along with the per capita consumption of each food group and related foodenergy content. Table 2 lists the average daily intake of macronutrients for this sample. Both tables also display the results stratified by income groups.

    Food prices. The CHNS collected food prices from each sample community. Prices in this analysis came from the following three sources: the state store (SS),⁴ the free market (FM) and authority price records published by the State Statistical Bureau (SSB) of China, which provides the provincial average. The SS prices were no longer used after the 1991–1992 price reform in China. The FM price had incomplete data for pork and coarse grains in some of the sampling points in the southern provinces. This related in part to the lack of availability of prices for the pork products we were studying and the fact that the components of coarse grains (corn and sorghum) were less likely to be sold in the free markets. In almost all situations, we selected the free market prices as the basis. Only when the goods we were studying were not sold in the free market, did we use the prices from the ration system. Unreported analysis compared price data collected from each community with two series of government price bureau food price data collected for each province each month, and rural and urban food prices collected for each province each month by the Ministry of Agriculture. Community time-varying price data were found to be more precise and to affect consumption decisions more readily.

    Income. The income variable used in this study represents household per capita income. It included all cash and noncash income components, except food subsidies.⁵ To reduce the potential for biases due to measurement error in the income measure, a standard economics procedure was used to create a predicted income measure (Maddala 1988 ). We also examined differences in behavior between the rich and the poor, and we conducted separate analyses on subsamples defined by their measured income level. We defined household income categories as low (poor), middle and high (rich), based on income tertiles of household per capita income in 1989.

Price and income variables were deflated by the consumer price index (CPI) (SSB 1990 , 1992 and 1994 ) for the particular time period in which the surveys were conducted and for the particular region in which the samples were located. The use of real deflated values in the regressions was designed to remove the effect of inflation and allow the analysis to focus on the effect of the increase in real price and real income. The CPI with the index of 1980 (index = 100%) was used as the baseline to deflate the nominal values for urban and rural consumers. Deflated community-level food prices were assigned to each individual corresponding to the time of their interviews (and hence the time represented for the collection of income data).

Seven household demographic variables were used in this analysis. Age, household size and education were continuous variables. Generally, it is difficult to obtain an accurate age in China because the Western and Chinese calendars are used interchangeably. The age difference can be as much as 1.5–2.0 mo between the Chinese lunar calendar and the Western one. We converted all Chinese lunar calendar dates to Western dates and used them to calculate age. Four dichotomous variables were included in the model building to indicate gender, urban residence and region of residence (two dummies variables for the three regions). The measurement of region was developed by the World Bank in collaboration with the SSB (World Bank 1995 ). It reflected contiguous groupings with comparable income levels. With respect to agricultural economics and food behavior, we regrouped all samples into three regions, i.e., the South Hinterland (Guezhou, Guangxi, Hunan), the Central Core (Henan, Hubei, Jingsu) and the North (Liaoning, Shangdong).

Model specification.

There are important behavioral and statistical reasons to study food consumption decisions as a two-step process. There are really two consumption decisions, i.e., whether to consume the specific food and then based on consuming the food, the quantity consumed. We utilized a procedure developed to handle this issue (Haines et al. 1988 ). In addition, we considered income, price and consumption relationships to be nonlinear and used a logarithmic transformation of our data.

Often the degree of sensitivity of a dependent variable to independent variables is represented by a measure called an "elasticity," the percentage change in the dependent variable resulting from a 1% change in the explanatory variable (Maddala 1988 ). In economics, often the best way to understand the effect of a change in a food price on consumption is to express the results in terms of what is called a "price elasticity" (Popkin and Haines 1981 ). A price elasticity measures the percentage change in the quantity of a food item consumed resulting from an increase of 1% in the item's food price. The own-price elasticity is defined as the direct effect of a food price on the consumption of the same food item (e.g., the effect of a change in food price of pork on pork consumption). The cross-price elasticity is the effect of the price change of a given food on the consumption of other food items (e.g., the effect of a change in the price of flour products on the consumption of rice products). As the price of a particular food changes, not only will consumption of that food change but consumption of "complements" and "substitutes" will also change. A complement is a food directly linked in consumption with the food studied (e.g., the amount of ready-to-eat cereal consumed might decrease with an increase in the price of milk because they are consumed jointly); a substitute is a replacement (e.g., millet, and other coarse grain products, and flour products compete as the staple food in Northern China; increased millet consumption might replace some rice consumption when the price of rice increases). As a first step, this analysis presents the way a change in the price of a single food can affect consumption of the food consumed as well as other related foods. Then, the overall effect of a change in the price of a specific food can be studied by combining both effects (the direct own-price elasticity and the indirect cross-price elasticity).

Constant elasticity models in which the dependent variable and the key independent variable are both transformed into logarithms are frequently used in food demand studies to estimate the price and income elasticities. In this model (often termed a log-log model), each estimated slope coefficient directly measures the expected percentage change in the dependent variable due to a 1% increase in each of the explanatory variables, holding constant all of the other explanatory variables. If an explanatory variable is not expressed as a logarithm, then the coefficient on it measures the expected percentage change in the dependent variable due to a unit change in the explanatory variable, holding constant the other explanatory variables in the model. Other important statistical issues are handled with the longitudinal modeling used in this study.

The SAS program (version 6.12) was used for data management (SAS Institute 1997 ). All analyses were done with the STATA statistical package, release 5.0 (STATA 1997 ).

	RESULTS

TOP ABSTRACT INTRODUCTION SUBJECTS AND METHODS RESULTS DISCUSSION REFERENCES

 
Price elasticities for the likelihood of consuming specific foods.

Multivariate logistic regression was used first to assess the effect of food prices on the likelihood of any consumption within each food group, controlling for income and other sociodemographic variables. Table 3 indicates that all elasticities with respect to the effect of a specific food group's price on the likelihood of consumption of that food group were negative and significantly different from zero. An increase in the price of each food group led to significant reductions in the probability of consuming any food within the food group. These own-price elasticities were estimated to be about -1.6 and -2.0 for the probabilities of consuming edible oils and rice, respectively. The cross-price elasticities with respect to the price of rice were significantly positive for the likelihood of consuming coarse grains (1.63) and wheat flour (0.77). An increase in the price of rice led to increased likelihood of consumption within these substitute food groups. The elasticity of the likelihood of pork consumption was large (-1.3), as were the pork price elasticities with respect to the other food groups. Increases in the price of pork led to large and significant reductions in the likelihood of consuming rice and eggs, and large and significant increases in the likelihood of consuming wheat flour, course grains and oils.

After controlling for the same sociodemographic factors as those in the analysis of the consumption participation decision, own-price elasticities for the quantity consumed had the same signs as they did for the likelihood of consuming each food group in Table 4. However, the magnitude of coefficients determining the amount of food consumed were smaller. The own-price elasticity was reduced to -0.12 for rice, -0.16 for wheat flour, -0.04 for course grains, -0.38 for pork, -0.16 for eggs and -0.30 for edible oils. Although the elasticities were smaller, all but the coarse grain own-price elasticity remained significant. Also, of the 36 own- and cross-price coefficients (six for each of six food groups) estimated for the longitudinal sample for the determinants of the quantity of food consumed (among those who consumed food in the group), 25 were significant at the 5% level.

Based on the parameter estimations and the probability of consumption of each food group, the overall price elasticities were calculated and reported in Table 5. These measure the expected percentage change in the quantity consumed within each group, accounting for both the probability of consuming a positive amount and the change in the amount consumed conditional on having consumed a positive quantity. The overall own-price elasticity was -0.4 for rice and wheat flour, -0.5 for pork, -0.1 for eggs and -0.3 for edible oils. Consider the own- and cross-price elasticities for pork. A 10% increase in the price of pork would result, overall, in a 5% decline in pork consumption, a 9% decline in rice consumption, a 2% increase in wheat flour consumption, a 4% increase in course grain consumption, a 3% decrease in egg consumption and a 3% increase in oil consumption. The own- and the cross-price effects indicated substantial responses to changes in the price of pork for each of the food groups.

RDA are used to interpret the nutritional quality of the diet. In this study, the effects of food prices on energy and protein intakes are expressed as a percentage of the Chinese RDA. It was not surprising that their patterns were similar to those for absolute intakes. In addition, dietary fat is expressed as the percentage of energy from fat. The effect of the price of pork on the percentage of energy from fat was significantly negative, with a 5.5% decrease in response to a 10% increase in the price.

Because of the high budget share spent on food among the poor, the negative effect of food price increases was expected to be greater among the poor than among the rich. Table 6 presents the estimates of own-price elasticities stratified by income level. This table is based upon the same type of analysis used to construct Table 5 , but the analyses were conducted separately for the poor and the rich. These larger elasticities show that poor consumers were more responsive to price changes than rich, except for coarse grains and edible oils. For instance, the price elasticity for pork was -0.96 among the poor and -0.33 among the rich. Pork consumption declined significantly after a pork price increase. An increased cost of food caused larger quantity adjustments among the poor as expected. With respect to the price of rice, the elasticity shifted from -0.25 among the rich to -0.54 among the poor. Food-group consumption clearly had substantially different price responses for the rich and the poor, especially for rice and pork. Figure 2 illustrates the difference of own-price elasticity for these food groups by income subpopulations.

View larger version (15K): [in this window] [in a new window]   Figure 2. Distribution of price elasticity of various food groups in income subpopulations of Chinese men and women, aged 20–45 y.

View larger version (13K): [in this window] [in a new window]   Figure 3. Effect of food price of various food groups on fat intake by poor and rich Chinese men and women, aged 20–45 y.

View larger version (10K): [in this window] [in a new window]   Figure 4. Effect of food price of various food groups on energy intake by poor and rich Chinese men and women, aged 20–45 y.

	DISCUSSION

TOP ABSTRACT INTRODUCTION SUBJECTS AND METHODS RESULTS DISCUSSION REFERENCES

Overall, there was evidence of substantial responses to price in food consumption. These elasticities indicated that the increase of food own-price led to a reduced probability of consumption for each food group. The own-price elasticities with respect to the probabilities of consuming edible oils and rice were -1.6 and -2.0. In other words, the likelihood of consuming edible oils and rice would decrease by ~16–20% for each 10% increase in real price. The effect of an increase in the price of rice was to increase the likelihood of consuming coarse grains and wheat flour, that is, an increase in the cost of rice resulted in lower rice consumption but higher consumption of the substitutes, i.e., coarse grains and wheat flour. The cross-price elasticities of pork were very large. For a 1% increase for the price of pork, the probability of consuming rice and eggs decreased by 4.9 and 1.4%, respectively. The probability of consuming coarse grains and edible oil increased by 2.9 and 1.9%, respectively, after a 1% increase in the price of pork.

Clearly, different foods have quite different price-consumption relationships. In particular, cross-price elasticities were different for different foods and prices. We focus on price changes and do not address the equally important effects of income, which is presented elsewhere (Guo 1998 ).

Our descriptive study indicated that along with a reduction in food quantities, food consumption patterns shifted toward greater use of staple foods and other foods of vegetable origin (Guo et al. 1999 ). This shift was induced by changes in price relations among the food groups. The removal of food subsidies favoring rice and wheat caused a rapid increase in the retail prices of these foods relative to price increases for other food groups. As a result, consumer demand shifted toward less expensive substitutes, and rice and wheat consumption declined. When the price of rice increased, there was a large substitution of coarse grains and, to a lesser extent, wheat flour for rice. Consequently, there were shifts in food consumption to lower priced sources of energy and protein. Not surprisingly, the price elasticities of grains (rice, flour and coarse grains) were small for energy, protein and fat intake. It was worthwhile noting that increases in the prices of rice and edible oils did not adversely affect energy and protein intakes, but that they were inversely associated with fat intake. It seems that the changes in food prices, especially changes in the price of pork and edible oils, have practical implications that may affect the issues of dietary excess and obesity during the period of nutrition transition in China.

In addition, given common economic assumptions about price responses, an increase in the price of a food tends to drive consumption away from that good (and its complements) and toward its substitutes. It was observed that an increase in the price of rice negatively affected the consumption of pork, eggs and oil, and positively affected the consumption of wheat flour and coarse grains. Because the cross-price elasticity with respect to the price of rice was the highest (0.37) for coarse grains, the increased rice cost resulted in a substantially higher consumption of sorghum, corn and millet.

In this study, food consumption of the poor was affected significantly by price changes, but substitutions between major staples buffered the effects on overall nutrient intakes. There was an important effect of changes in food prices. When the price of pork rose, the dietary pattern of the poor shifted to relatively cheaper foods, such as oil, wheat flour and coarse grains. From the comparison of cross-price elasticities, it is clear that there are systematic effects of price changes on diet that are logical and fit our understanding of Chinese dietary behavior. Note, also, that grains fed to livestock can buffer grain price increases (Behrman et al. 1988 ). As grain prices go up, pork prices will also rise. Pork consumption will then decline, freeing up some grain for direct consumption. As the gap between rich and poor widens, however, rich consumers may not greatly reduce their consumption of meat, even at higher prices, and the burden of reducing grain demand to the level of supply may fall mostly on the poor. This may explain why many of the poor do not derive a large share of their incomes from either wage labor in food production or the sale of food. A large proportion are net purchasers of other foods. Undoubtedly, increases in food prices have much less favorable effects on the poor.

The main implication of this study is that one should consider price changes because of their affect differential effect on the rich and the poor. In addition, the range of nutrients and foods that are affected by change should be considered to provide some understanding of the dynamic situation. Combining this work with an understanding of the needs of the rich and the poor in terms of health and nutritional status allows us to prepare sound price policy recommendations.

As was shown in this analysis, price changes for animal protein foods had a large effect on reducing fat intake. Given the rapid increases in obesity in China and the role that fat plays in this change (Paeratakul et al. 1998a , Popkin and Doak 1998 ), reducing fat intake is important. At the same time, one must worry about the protein intake of the poor. At this stage in China's nutrition transition, dietary excess is mainly a feature of China's middle and upper income groups. The poor face dietary deficit, even increases in undernutrition in some areas of the country (Popkin et al. 1995 ). One goal of price policy would be to reduce the fat intake of the rich but not adversely affect protein intake of the poor.

Two alternate price policies are to increase the price of pork and edible oils. As was shown in this analysis, the effects are quite different. Pork price increases would reduce fat intake more but also reduce protein intake for the poor, whereas oil price increases would not adversely affect protein intake (in fact, it would increase protein intake slightly). This runs counter to what was proposed at a meeting on food and nutrition planning a number of years ago in China. On the basis of limited data, it seemed logical to increase pork prices, and work has been underway for half a decade to do that. These results indicate this may not be an appropriate policy option.

This study has focused on only a limited set of foods and macronutrients. The foods selected represent ~75% of the intake of fat in the Chinese diet. Food prices can affect the intake of many important vitamins and minerals by affecting both the foods studied as well as other key food groups such as fruit and vegetable intake. Much of the focus of China's price policy has been on ways to enhance energy and protein intake as it relates to the reduction of undernutrition and hunger. Our research has attempted to expand that focus to fat intake and the issue of excessive fat intake. Iron deficiency anemia, iodine deficiency disorder, low calcium intake and subclinical vitamin A deficiency in some regions are important nutrition deficiencies that are not studied in this focus on the effects of food price changes on macronutrient intake in China.

In summary, the time is here for China to develop policies for nutrition education and intervention that would avert some of the adverse health effects of the nutrition transition. The CHNS study provides a series of useful references to steer the Chinese population toward a more healthy diet. There is potential, as this study shows, for a meaningful price policy to be instituted. Fat intake can be reduced. Moreover, this approach could certainly be expanded to consider the same policy options in other countries. In the case of Scandinavian countries, aggressive state policies related to taxation and import tariffs, as well as consumer education, are believed to have had an effect on dietary choices and public health (Milio 1990 and 1991 ). As we show, price policy, albeit an approach that is complex in itself as well as politically, offers an important element in the public policy arsenal that nutritionists should embrace.

View larger version (10K): [in this window] [in a new window]   Figure 5. Effect of food price of various food groups on protein intake by poor and rich Chinese men and women, aged 20–45 y.

	ACKNOWLEDGMENTS

	FOOTNOTES

 
¹ Funding for parts of the project design, data collection and computerization was provided by the Chinese Academy of Preventive Medicine (CAPM), the Carolina Population Center (CPC) of the University of North Carolina at Chapel Hill (UNC-CH) and the National Institutes of Health (NIH) (P01HD28076–01). Collaborative training and development work was funded by the National Science Foundation (grant #37486) and the Ford Foundation. Funds for the research reported in this article were provided by NIH.

³ Abbreviations used: CHNS, China Health and Nutrition Survey; CPI, consumer price index; FM, Free Market; RDA, Recommended Daily Allowances; SS, State Store; SSB, State Statistical Bureau of China.

⁴ To assess the effect of food prices in a free-living population, we took the income without food subsidies to examine the price elasticity for urban and rural residents.

⁵ As a part of national adjustment of economic structure, all food subsidies were abolished in1993.

Manuscript received October 21, 1998. Initial review completed November 18, 1998. Revision accepted January 29, 1999.

	REFERENCES

TOP ABSTRACT INTRODUCTION SUBJECTS AND METHODS RESULTS DISCUSSION REFERENCES

 

1. Behrman J. R., Deolalikar A. B., Wolfe B. L. Nutrients: impacts and determinants. World Bank Econ. Rev. 1988;2:299-320[Abstract/Free Full Text]

2. Guo, X. (1998) Impact of Income and Food Prices on Food Consumption and Dietary Fat Intake in China, 1989–1993. UMI Dissertation Services: A Bell & Howell Company, Ann Arbor, MI.

3. Guo, X., Popkin, B. M. & Zhai, F. (1999) Patterns of change in food consumption and dietary fat intake in Chinese adults, 1989–1993. Food Nutr. Bull. (in press).

4. Haines P. S., Guilkey D. K., Popkin B. M. Modeling food consumption decisions as a two-step process. Am. J. Agric. Econ. 1988;70:543-552

5. Huber, P. J. (1967) The behavior of maximum likelihood estimates under non-standard conditions. Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability 1: 221–233.

6. Kennedy P. A Guide to Econometrics 1992 The MIT Press Cambridge, MA.

7. Kish L. Survey Sampling 1965 Wiley New York, NY.

8. Maddala G. S. Introduction to Econometrics 1988 Macmillan New York, NY.

9. Millio N. Nutrition Policy for Food-Rich Countries: A Strategic Analysis 1990 Johns Hopkins University Press Baltimore, MD.

10. Millio N. Toward health lessons in food and nutrition policy development from Finland and Norway longevity. Scand. J. Soc. Med. 1991;19:209-217[Medline]

11. Paeratakul S., Popkin B. M., Ge K., Adair L. S., Stevens J. Changes in diet and physical activity affect the body mass index of Chinese adults. Int. J. Obes. 1998;22:424-431

12. Paeratakul S., Popkin B. M., Kohlmeier L., Hertz-Picciotto I., Guo X., Edwards L. Measurement error in dietary data: implications for the epidemiologic study of the diet-disease relationship. Eur. J. Clin. Nutr. 1998;52:722-727[Medline]

13. Pinstrup-Andersen, P., Yang, D., Xian, Z. & Yang, Y. (1990) Changes in incomes, expenditures, and food consumption among rural and urban households in China during the period 1978–88. In: Proceedings of the International Symposium on Food: Nutrition and Social Economic Development. The Chinese Academy of Preventive Medicine, Beijing, China.

14. Popkin B. M. Nutritional patterns and transitions. Popul. Dev. Rev. 1993;19:138-157

15. Popkin B. M. The nutrition transition and its health implications in lower income countries. Public Health Nutr 1998;1:5-21[Medline]

16. Popkin B. M., Doak C. The obesity epidemic is a worldwide phenomenon. Nutr. Rev. 1998;56:106-114[Medline]

17. Popkin B. M., Ge K., Zhai F., Guo X., Ma H., Zohoori N. The nutrition transition in China: a cross-sectional analysis. Eur. J. Clin. Nutr. 1993;47:333-346[Medline]

18. Popkin B. M., Haines P. S. Factors affecting food selection—the role of economics. J. Am. Diet. Assoc. 1981;79:419-425[Medline]

19. Popkin B. M., Paeratakul S., Zhai F., Ge K. Body weight patterns among the Chinese: results from the 1989 and 1991 China Health and Nutrition Surveys. Am. J. Public Health 1995;85:690-694[Abstract/Free Full Text]

20. SAS Institute Inc SAS Language Reference 1997 SAS Institute Cary, NC.

21. STATA (1997) Stata User's Guide, Release 5. Stata Press, College Station, TX.

22. State Statistical Bureau of China (1990) The 1989 Price Statistical Yearbook for China [in Chinese]. China Statistical Publication House, Beijing, China.

23. State Statistical Bureau of China (1992) The 1991 Yearbook of Statistical Index for Price in China [in Chinese]. China Statistical Publication House, Beijing, China.

24. State Statistical Bureau of China (1994) The 1993 Yearbook of Statistical Index for Price in China [in Chinese]. China Statistical Publication House, Beijing, China.

25. Wonnacott R. J., Wonnacott T. H. Econometrics 1979 Wiley New York, NY.

26. World Bank (1995) China regional disparities. Report no. 14496-CHA. The World Bank, Washington, DC.

27. Zeger S. L., Liang K. Y. An overview of methods for the analysis of longitudinal data. Stat. Med. 1992;11:1825-1839[Medline]

28. Zhai F., Guo X., Popkin B. M., Ma L., Yu W., Jin S., Ge K. The evaluation of the 24-hour individual recall method in China. Food Nutr. Bull. 1996;17:154-161

29. Zhai F., Jin S., Ge K. Summary report of China Health and Nutrition Survey—an eight-province case study. China J. Hygiene Res. 1996;25(suppl.):16-25

30. Zohoori N., Savitz D. A. Econometric approaches to epidemiologic data: relating endogeneity and unobserved heterogeneity to confounding. Ann. Epidemiol. 1997;7:251-257[Medline]

This article has been cited by other articles:

				D. J. Hoffman Upper Limits in Developing Countries: Warning Against Too Much in Lands of Too Little J. Am. Coll. Nutr., December 1, 2004; 23(suppl_6): 610S - 615S. [Abstract] [Full Text] [PDF]

				J. D. Stookey, L. Adair, J. Stevens, and B. M. Popkin Patterns of Long-Term Change in Body Composition Are Associated with Diet, Activity, Income and Urban Residence among Older Adults in China J. Nutr., September 1, 2001; 131(9): 2433S - 2440. [Abstract] [Full Text] [PDF]

				B. M. Popkin The Nutrition Transition and Obesity in the Developing World J. Nutr., March 1, 2001; 131(3): 871S - 873. [Abstract] [Full Text]

				Y. Wang, K. Ge, and B. M Popkin Tracking of body mass index from childhood to adolescence: a 6-y follow-up study in China Am. J. Clinical Nutrition, October 1, 2000; 72(4): 1018 - 1024. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Guo, X. Articles by Zhai, F. Search for Related Content PubMed PubMed Citation Articles by Guo, X. Articles by Zhai, F.