dung phd conference 2012
Post on 14-May-2015
406 Views
Preview:
TRANSCRIPT
Preliminary results
Dung Doan
Crawford School of Public Policy 27 November 2012
Income and diet quality in China2004-2009
Research context Structural shift in dietary consumptions in developing
countries Characterized not by calories deficiency, but by unbalanced diet Significant impacts on dietary welfare, health, and labor quality
Warnings of deteriorating diet quality as income rises in China Shift away from Chinese tradition diet Higher consumption of animal foods and edible oils Lower grain and vegetable consumption Calories intake decreases Increasing income hardly reduces vitamin A & D deficiencies
2
Literature review Strong focus on diet adequacy
Levels of food and nutrient consumption as dependent variables E.g. Berhman & Deolalikar (1987), Ravallion (1990), Skoufias (2002,
2009), Mangyo (2008)
Studies on consumption quantity are insufficient to inform about how diet quality changes as income grows Estimated income elasticity varies widely across studies Consumption quantity reveals limited info about changes in diet structure
and quality
Other aspects of diet quality are under-studied, despite being well-grounded in nutrition literature
3
Research questions Does diet quality necessarily deteriorate as income rises in
China? Are there income effects and what form do they take? Do income effects change over time?
Are there education effects?
Contribution: Directly examine income effects on an essential aspect of diet quality
that no economists has paid attention to: diet variety Most recent time period
4
Data
5
China Health and Nutrition Survey 2004, 2006, 2009
Multi-stage, randomized cluster sampling design
9 provinces, approx. 3,800 households/year
Pool sample: 20,307 adults (18-60 years old) from 4,506 households
Methodology
Dependent variable: Diet variety, measured by number of major food groups consumed
Repeated cross-sectional regression models OLS :
Poisson quasi-maximum likelihood:
where
6
Income effects
7
OLS Poisson
Model Variable 2004 2006 2009 2004 2006 2009
1 Income 0.073*** 0.016 0.0092 0.064*** 0.016** 0.0094*
2 Income 0.114*** 0.088*** 0.035*** 0.114*** 0.087*** 0.036***
Income squared -‐0.0082* -‐0.0046*** -‐0.0014*** -‐0.0094*** -‐0.0051*** -‐0.0016***
3
QuinJle 2 0.0074 0.056** -‐0.038 0.014 0.076** -‐0.038
QuinJle 3 -‐0.034 0.015 -‐0.043 -‐0.024 0.039 -‐0.041
QuinJle 4 0.108*** 0.117*** -‐0.022 0.119*** 0.137*** -‐0.018
QuinJle 5 0.138*** 0.149*** 0.101*** 0.143*** 0.159*** 0.096***
* p < 0.10, ** p < 0.05, *** p < 0.01
Upward concave relationship between income and diet variety Income effects vary across income groups Income effects weaken over time
Income effects (cont’)
8
Marginal income effect is positive, stronger among low income groups
Income effects weaken over time
0.00
0.02
0.04
0.06
0.08
0.10
0.12
Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5
Marginal effects at mean income
2004 2006 2009
Education effects
Higher education attainment, higher diet diversity Difference in impacts of vocational training and university education are
not significant at 10% level
9
OLS Poisson
Educa=on level 2004 2006 2009 2004 2006 2009
Primary 0.128*** 0.032 0.042 0.133 0.036 0.049
Secondary 0.154*** 0.106*** 0.088*** 0.159*** 0.107*** 0.095***
High school 0.217*** 0.161*** 0.150*** 0.217*** 0.164*** 0.158***
VocaJonal training 0.322*** 0.340*** 0.296*** 0.306*** 0.321*** 0.287***
University & above 0.338*** 0.279*** 0.376*** 0.301*** 0.243*** 0.341***
* p < 0.10, ** p < 0.05, *** p < 0.01 Base educaJon category is “No educaJon”
Preliminary conclusion Household income has positive impact on diet variety.
Income effects are stronger among low income groups, but diminish over time.
Education matters.
10
Issues and Steps forward Refine model specification and functional form
Address issue of endogeneity between household income and diet variety
Explore a diet quality index as a more comprehensive measure of diet quality
11
Thank you
12
Rationale for diet variety Diet variety as dependent variable
Unambiguous and well-grounded in nutrition literature Widely recommended by government’s dietary guidelines Incorporated in well-known diet quality indexes
Construction of variety measure Diet variety is most commonly measured by no. of food groups
consumed (Ruel 2005) Food group classification is based Chinese Food Composition Table
(Yang et al. 2004, 2009), Chinese dietary guidelines 2007, and Diet quality index – International (Kim et al. 2003)
13
Income effects
Upward concave relationship between income and diet variety Marginal income effect is smaller among high income groups
Income effects weaken over time – flatter curves Note: The curves represent the predicted quadratic relationship between income and diet variety, while holding all other
regressors constant. Thus, values on the vertical axis do not represent nominal value of diet variety.
14
- 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45
0 3 6 9 12 15 18
Pred
icte
d di
et v
arie
ty
Household income per capita (10,000 Yuan)
Predicted diet variety based on Model 2 and mean income of percentiles
2004 2006 2009
Income marginal effects
Marginal income effect is smaller among high income groups Need to refer to nutrition literature to assess the benefit of
having 1 additional food group in diet
15
OLS Poisson
Income Marginal effect 2004 2006 2009 2004 2006 2009
Sample mean 0.099 0.078 0.031 0.097 0.076 0.031
Mean of quinJle 1 0.112 0.086 0.034 0.111 0.086 0.035
Mean of quinJle 2 0.108 0.084 0.033 0.107 0.083 0.034
Mean of quinJle 3 0.104 0.081 0.032 0.102 0.080 0.032
Mean of quinJle 4 0.097 0.077 0.030 0.094 0.075 0.030
Mean of quinJle 5 0.076 0.061 0.024 0.070 0.058 0.023
Compare Model 2 and Model 3
16
0.00
0.05
0.10
0.15
0.20
0.25
Quintile 2 Quintile 3 Quintile 4 Quintile 5
Income effects, as compared to Quintile 1
2004 - model 2 2006 - model 2 2009 - model 2 2004 - model 3 2006 - model 3 2009 - model 3
Community-fixed effects OLS Poisson
Province 2004 2006 2009 2004 2006 2009 Heilongjiang -‐0.011 -‐0.049 -‐0.088** -‐0.0038 -‐0.0392 -‐0.076*
Jiangsu -‐0.3889** -‐0.706*** -‐0.797*** -‐0.347*** -‐0.617*** -‐0.718***
Shandong -‐0.289*** -‐0.690*** -‐0.487*** -‐0.247*** -‐0.641*** -‐0.419***
Henan -‐0.359*** -‐0.642*** -‐0.625*** -‐0.316*** -‐0.568*** -‐0.564***
Hubei -‐0.683*** -‐0.917*** -‐0.894*** -‐0.670*** -‐0.879*** -‐0.849***
Hunan -‐0.453*** -‐0.800*** -‐0.881*** -‐0.434*** -‐0.722*** -‐0.829***
Guangxi -‐0.838*** -‐1.216*** -‐1.175*** -‐0.991*** -‐1.369*** -‐1.088***
Guizhou -‐0.435*** -‐0.6623*** -‐0.600*** -‐0.395*** -‐0.587*** -‐0.492***
• p < 0.10, ** p < 0.05, *** p < 0.01 • Base province is Liaoning
17
Provincial factors play a significant role in determining diet variety
Considerable differences across some provinces
Checking multicollinearity
18
Model 2 2004 2006 2009 Age squared 56.37 56.76 55.24 age 56.03 56.45 54.96 HH income pc 5.26 4.18 3.74 HH income pc squared 4.53 3.83 3.53 EducaJon level -‐ 2 3.21 3.18 3.02 Province_43 2.46 2.74 2.7 EducaJon level -‐ 1 2.44 2.58 2.59 EducaJon level -‐ 3 2.43 2.51 2.26 Province_32 2.29 2.45 2.13 Province_42 2.25 2.44 2.11 Province_23 2.23 2.41 2.08 Province_37 2.06 2.28 1.92 Province_41 2.04 2.06 1.87 EducaJon level -‐ 4 1.84 2.03 1.85 Province_52 1.75 1.94 1.85 Price -‐ soy oil (log) 1.65 1.91 1.79 EducaJon level -‐ 5 1.64 1.88 1.76 Province_45 1.59 1.67 1.54 Price -‐ tofu (log) 1.49 1.62 1.51 Price -‐ cabbage (log) 1.35 1.55 1.42 Rural 1.33 1.41 1.38 Price -‐ rice (log) 1.23 1.38 1.32 Price -‐ tofu (log) 1.21 1.33 1.24 HH size 1.2 1.27 1.23 Price -‐ pork (log) 1.13 1.24 1.22 Gender 1.06 1.06 1.04 Mean VIF 6.23 6.31 6.05
Descriptive Statistics
19
No. of food groups consumed 2004 2006 2009
Mean 1.72 1.77 1.89 Median 2 2 2 Standard deviaJon 0.77 0.80 0.83 Percentage
0 0.24% 0.31% 0.10% 1 43.38% 41.76% 36.25% 2 43.30% 41.25% 42.18% 3 10.43% 13.85% 17.94% 4 2.36% 2.56% 3.15% 5 0.29% 0.28% 0.39%
HH income pc (Yuan), adjusted to adult equivalence scale 2004 2006 2009
Sample mean 9,111 10,828 14,951
Mean of quinJle 1 1,352 1,505 1,622
Mean of quinJle 2 3,899 4,123 6,481
Mean of quinJle 3 6,439 7,219 10,442
Mean of quinJle 4 10,455 11,649 16,682
Mean of quinJle 5 23,256 28,671 38,782
Dietary guidelines for Chinese residents
20
1. Eat a variety of foods, mainly cereals including appropriate amount of coarse grains
2. Consume plenty of vegetables, fruits and tubers
3. Consume milk, soybean or dairy-‐ or soybean-‐products everyday
4. Consume appropriate amounts of fish, poultry, eggs and lean meat
5. Use less cooking oil; choose a light diet which is also low in salt
6. Do not over eat, exercise every day, and maintain a healthy body weight
7. RaJonally distribute the daily food intake among the three meals, correctly choose snacks
8. Drink sufficient amount of water every day, raJonally select beverages
9. If you drink alcoholic beverages, do so in limited amounts
10. Choose fresh and sanitary foods
Source: Ke (2011)
Food classification
21
Food groups Foods sub-‐groups
Grain Wheat, rice, corn, barley, millet, tubers
Vegetable Root, stem, leafy, and flowering vegetables
Fruit Fruits
Meat Pig, cakle, mukon, poultry , fish, egg, beans, and products
Dairy Milk and dairy products
Others Fungi, nuts & seeds, infant food, ethnic foods & cakes, fast foods, beverages, alcoholic beverages, sugars & preserves & honey, fats & vegetable oils, condiments
Source: Chinese Food Composition Tables (Yang et al. 2004, 2009)
Food classification
22
Source: Ke (2011)
top related