Dietary Diversity in Urban and Rural China: An Endogenous Variety Approach Jing Liu September 6, 2011
Road Map What is endogenous variety? Why is it? A structural framework illustrating this idea An application to an analysis of dietary diversity in China
The standard consumer demand problem follows an endogenous quantity approach.
The standard consumer demand problem follows an endogenous quantity approach. Endogenous Quantity Max X s.t. X (P, Y ) U(x i ), i = 1,..., N N p i x i Y i=1
The standard consumer demand problem follows an endogenous quantity approach. Endogenous Quantity Max X s.t. U(x i ), i = 1,..., N N p i x i Y i=1 Consumers: access all varieties with no cost adjust quantities to achieve max utility X (P, Y )
Utility and expenditure are aggregation of variety as well as quantity. Utility: more is better vs. the love of variety Budget: a constraint of consumption diversification
Utility and expenditure are aggregation of variety as well as quantity. Utility: more is better vs. the love of variety Budget: a constraint of consumption diversification Accessing to an additional variety is rarely free due to e.g. shopping time, storage and transportation cost. upward sloped MC downward sloped MB
Utility and expenditure are aggregation of variety as well as quantity. Utility: more is better vs. the love of variety Budget: a constraint of consumption diversification Accessing to an additional variety is rarely free due to e.g. shopping time, storage and transportation cost. upward sloped MC downward sloped MB The consumer will not take the additional variety unless it brings in non-negative net marginal benefit. Variety is endogenous when access to it is not free.
New approach Variety is accessed with a cost. Consumers adjust n to achieve max utility. The optimal choice ends up with some but not all varieties. Endogenous variety s.t. Max n U(x i ), i = 1,..., n n p i x i + AC(n) Y i=1
New approach Variety is accessed with a cost. Consumers adjust n to achieve max utility. The optimal choice ends up with some but not all varieties. Endogenous variety s.t. Max n U(x i ), i = 1,..., n n p i x i + AC(n) Y i=1 Access Cost (AC): contains fixed and variable cost; may vary widely due to either different household production technology or infrastructure.
Structural Model s.t. Max n U(x i ), i = 1,..., n n p i x i + AC(n) Y i=1 AC(n) = F (Z)n ɛ
Structural Model s.t. Max n U(x i ), i = 1,..., n n p i x i + AC(n) Y i=1 AC(n) = F (Z)n ɛ [ n = λ Y F (Z) ] 1/ɛ AC, Variety access cost Y, total expenditure Z, a vector of household and community characteristics λ, parameters associated with utility function and price index
Empirical Model Hypothesis of interest: Rich people consume more varieties. High access cost prevents consumers from reaching more varieties.
Empirical Model Hypothesis of interest: Rich people consume more varieties. High access cost prevents consumers from reaching more varieties. Empirical model: n = γ 0 + γ 1 INC + γ 2 INC 2 + γ 3 AC + B X + u 1 AC is determined by household and community characteristics.
China Health and Nutrition Survey (CHNS), 2006
The darker shaded regions are the provinces in which the survey has been conducted. China Health and Nutrition Survey (CHNS), 2006
Variable Specification Dependent variable: Number of items a respondent consumed per day;
Variable Specification Dependent variable: Number of items a respondent consumed per day; Explanatory variables: Demographic information Actual annual household net income Food access cost related household and community characteristics
Variable Specification Dependent variable: Number of items a respondent consumed per day; Explanatory variables: Demographic information Actual annual household net income Food access cost related household and community characteristics Household Own refrigerator Own transportation Community Distance to the nearest Chinese supermarket Distance to the nearest farmer s market # of supermarket within 5 km # of farmer s market within 5 km # of indoor restaurant Bus stop in the community
Sample Description Rural Urban Variable Mean Std Mean Std # of items consumed per day 9.12 2.44 10.54 2.67 Age 42.95 10.68 42.53 11.34 Gender (1=female) 0.52 0.50 0.53 0.50 Education (1=high school finished) 0.20 0.40 0.42 0.49 Perception (1=agree diverse diet is good for health) 0.81 0.39 0.86 0.35 Married (1=married) 0.89 0.31 0.83 0.38 Household size 2.98 1.32 3.23 1.22 Household income (1,000 Yuan) 25.75 30.47 35.24 35.92 # of indoor restaurant 13.56 17.83 16.66 20.69 Own transportation (1=yes) 0.85 0.36 0.74 0.44 Own refrigerator (1=yes) 0.44 0.50 0.73 0.45 Dist. to nearest Chinese supermarket (km) 9.91 12.09 4.30 5.74 Dist. to nearest farmer s market (km) 1.60 2.91 0.85 1.31 # of supermarket within 5km 1.73 2.20 5.86 10.22 # of farmer s market within 5km 2.76 3.54 4.72 5.31 Community area (km 2 ) 16.75 66.31 10.01 45.81 Bus stop (1=yes) 0.67 0.47 0.81 0.40
Dietary Diversity Comparison Questions: Is rural diet different from urban diet? How? Could the difference be driven by income or food accessibility?
Dietary Diversity Comparison Questions: Is rural diet different from urban diet? How? Could the difference be driven by income or food accessibility? First glance: Category Percentage Total # of items Rural 66.16 903 Urban 33.84 1,008
Dietary Diversity: Number of Items Top 30 urban popular food
Dietary Diversity: Number of Items Top 30 urban popular food Top 30 rural popular food
Dietary Diversity: Number of Items Top 30 urban popular food Top 30 rural popular food The percentage of population consuming a specified item is generally higher in urban. 24 of the 30 items overlap.
Rank Urban Rural 1 Rice Rice 2 Pork (lean and fat) Pork (lean and fat) 3 Egg Egg 4 Potato Potato 5 Bok Choy Bok Choy 6 Noodle Hot pepper 7 Tomato Fresh noodle 8 Hot pepper Wheat flour 9 Steamed bun Baby bok choy 10 Soybean Soybean 11 Green bean Steamed bun 12 Baby bok choy Green bean 13 Wheat flour Eggplant 14 Steamed rice Tomato 15 Cucumber Pumpkin 16 Eggplant Ash gourd 17 Apple Cucumber 18 Deep fried dough stick Dried noodle 19 Soy milk Loofah 20 Beef Pork (lean) 21 Pumpkin Chinese leek 22 Chicken White radish 23 Ash gourd Long purple eggplant 24 Chinese leek Deep fried dough stick 25 Dried noodle Corn flour 26 Milk Pickled bok choy 27 Soybean sprout Steamed rice 28 White radish Soybean sprout 29 Cauliflower Firm tofu 30 Goldfish carp Beef
Rank Urban Rural 1 Rice Rice 2 Pork (lean and fat) Pork (lean and fat) 3 Egg Egg 4 Potato Potato 5 Bok Choy Bok Choy 6 Noodle Hot pepper 7 Tomato Fresh noodle 8 Hot pepper Wheat flour 9 Steamed bun Baby bok choy 10 Soybean Soybean 11 Green bean Steamed bun 12 Baby bok choy Green bean 13 Wheat flour Eggplant 14 Steamed rice Tomato 15 Cucumber Pumpkin 16 Eggplant Ash gourd 17 Apple Cucumber 18 Deep fried dough stick Dried noodle 19 Soy milk Loofah 20 Beef Pork (lean) 21 Pumpkin Chinese leek 22 Chicken White radish 23 Ash gourd Long purple eggplant 24 Chinese leek Deep fried dough stick 25 Dried noodle Corn flour 26 Milk Pickled bok choy 27 Soybean sprout Steamed rice 28 White radish Soybean sprout 29 Cauliflower Firm tofu 30 Goldfish carp Beef
Ten most differently consumed food items
Dietary Diversity: Composition Summarize top 30 popular items based on USDA food pyramid: Food Group Subgroup Urban Rural Grains 25 30 Vegetables 41 56 Fruits 3 0 Protein Meat, poultry, seafood, egg 22 11 Beans 3 4 Dairy 6 0 Total 100 100
Dietary Diversity: Composition Summarize top 30 popular items based on USDA food pyramid: Food Group Subgroup Urban Rural Grains 25 30 Vegetables 41 56 Fruits 3 0 Protein Meat, poultry, seafood, egg 22 11 Beans 3 4 Dairy 6 0 Total 100 100 Rural diet is more grain and plant based, and not well balanced by the intake of fruit, meat and diary product. These products are featured by higher price or higher access cost, or both.
Variety Engel Curve: how the number of items consumed varies with household income.
Variety Engel Curve: how the number of items consumed varies with household income.
Regression Diagnostics
Regression Diagnostics Heterogeneity split sample Chow-test: F 27,3381 = 8.31, P-value < 0.0001 add province fixed-effect
Regression Diagnostics Heterogeneity split sample Chow-test: F 27,3381 = 8.31, P-value < 0.0001 add province fixed-effect Heteroskedasticity Breusch-Pagan test: Category F-value P-value Rural 4.81 < 0.0001 Urban 2.09 0.0011 Distance and # of supermarket are negatively correlated with the variance. Calculated the heteroskedasticity consistent-covariance matrix estimator(hccme).
Urban Rural Variable Coefficient Std.Err. Coefficient Std.Err. Intercept 7.03 0.62 7.44 0.46 Age 0.004 0.01-0.001 0.00 Female -0.12 0.12-0.41 0.09 High School 0.26 0.15 0.23 0.13 Perception -0.25 0.19 0.33 0.11 Married 0.34 0.19-0.003 0.17 HHINC 0.02 0.01 0.004 0.00 HHINC 2-0.00007 0.00-0.00002 0.00 HHSIZE 0.09 0.06 0.31 0.04 Neighborhood Area -0.005 0.00-0.002 0.00 # of Restaurant 0.01 0.00 0.02 0.00 Refrigerator 0.70 0.15 0.55 0.11 Transportation 0.48 0.16 0.25 0.14 Bus Stop -0.12 0.22 0.55 0.13 Dist SM -0.06 0.01-0.03 0.01 Dist FM -0.23 0.09-0.06 0.03 Dist SM Dist FM 0.02 0.01 0.004 0.00 # of SM within 5 km -0.01 0.01-0.27 0.04 # of FM within 5 km 0.03 0.02 0.002 0.01 Home farming (1= yes) -0.51 0.16 Value consumed 0.06 0.01 Sample Size 1464 1971 Adjusted R 2 0.207 0.322 Province fixed effect Yes Yes
Concluding Marks When access cost is not zero, consumer utility can be endogenously determined by variety in addition to by quantity. High access cost prevents consumers from reaching more varieties.
Concluding Marks When access cost is not zero, consumer utility can be endogenously determined by variety in addition to by quantity. High access cost prevents consumers from reaching more varieties. Policy implication Enhance infrastructure investment
Concluding Marks When access cost is not zero, consumer utility can be endogenously determined by variety in addition to by quantity. High access cost prevents consumers from reaching more varieties. Policy implication Enhance infrastructure investment Contribution Investigated the underlying causes of dietary diversity, especially the role of economic and environmental factors. Systematically compared dietary diversity in urban and rural China.
Concluding Marks When access cost is not zero, consumer utility can be endogenously determined by variety in addition to by quantity. High access cost prevents consumers from reaching more varieties. Policy implication Enhance infrastructure investment Contribution Investigated the underlying causes of dietary diversity, especially the role of economic and environmental factors. Systematically compared dietary diversity in urban and rural China. Shortcoming Absence of price Measurement of dietary diversity