Socioeconomic Socioeconomic Determinants of Body Determinants of Body Mass Index of Adult Mass Index of Adult Chinese in the 1990s Chinese in the 1990s Zhehui Luo, Ph.D. MS Zhehui Luo, Ph.D. MS Department of Department of Epidemiology Epidemiology Michigan State University Michigan State University
41
Embed
Socioeconomic Determinants of Body Mass Index of Adult Chinese in the 1990s
Socioeconomic Determinants of Body Mass Index of Adult Chinese in the 1990s. Zhehui Luo, Ph.D. MS Department of Epidemiology Michigan State University. Outline. Simple Model Data and Econometric Issues Trends and Descriptives Results of Reduced Form Demand Results of Dynamic Demand - PowerPoint PPT Presentation
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Socioeconomic Socioeconomic Determinants of Body Determinants of Body Mass Index of Adult Mass Index of Adult Chinese in the 1990sChinese in the 1990s
Zhehui Luo, Ph.D. MSZhehui Luo, Ph.D. MS
Department of EpidemiologyDepartment of Epidemiology
Michigan State UniversityMichigan State University
OutlineOutline
Simple ModelSimple Model Data and Econometric IssuesData and Econometric Issues Trends and DescriptivesTrends and Descriptives Results of Reduced Form Results of Reduced Form
DemandDemand Results of Dynamic DemandResults of Dynamic Demand Results of Production FunctionResults of Production Function
The ModelThe Model
, , ; ,0,... 0{ , , } 1,...,
T tMax U h x lt t tt T tx l mit it it i n
, , , , ;1h f h m l zi it it it it ct itit
s. t.
wT l l hit it it it
(1 )1 1 1 1 1
n n nx m wA r A p x p m w l yt t it t it it it tt t i i i
The ModelThe Model
(1 ) 11
(1 )1
mU fr p wh h it itit t txptU f f rx m mtit t
(1 )1 1
r Uht tthusU tht
txpt
The ModelThe Model
If total time available for leisure and If total time available for leisure and work is not affected by health statuswork is not affected by health status
Labor is the same as the effective Labor is the same as the effective labor in Grossman’s senselabor in Grossman’s sense
Equilibrium condition is: Equilibrium condition is:
(1 ) 11
1
mU fr ph hit t txptU f fx m mtit t
Estimating EquationsEstimating Equations
, , , , , , , , , ;, , ,t t
x h l m g p p w yx m i i cx h l m hi
, , , , , ;1 1 1 1 1x m
h f h p p w yit itit t t it t
, , , , ;1h f h m l zi it it it it ct itit
The DataThe Data
Chinese Health and Nutrition Survey Chinese Health and Nutrition Survey (CHNS) 1989, 1991, 1993 and 1997(CHNS) 1989, 1991, 1993 and 1997
Carolina Population Center at the Carolina Population Center at the University of North Carolina, Institute University of North Carolina, Institute of Nutrition and Food Hygiene, Chinese of Nutrition and Food Hygiene, Chinese Academy of Preventive MedicineAcademy of Preventive Medicine
Sampling: multistage, random cluster Sampling: multistage, random cluster processprocess
The DataThe Data Adults aged 20 and aboveAdults aged 20 and above Excluding pregnant womenExcluding pregnant women Sample size: Sample size:
Years of Years of EducationEducation 7.327.32 6.016.01 6.046.04 6.586.58
% Employed% Employed 9696 8282 8080 7777
% HH water % HH water factoryfactory 3434 4141 4242 4949
Mean BMI, Mean BMI, WomenWomen 21.721.7 22.122.1 22.122.1 22.622.6
Mean BMI, MenMean BMI, Men 21.221.2 21.621.6 21.921.9 22.322.3
Trends in Food Trends in Food AvailabilityAvailability
Source: FAO 2002 Food Balance Sheet for China
Trends in overweight and Trends in overweight and obesityobesity
% Men and Women % Men and Women overweight (BMI overweight (BMI 25)25)
0
5
10
15
20
25
1989 1991 1993 1997
Men Women
% Men and Women% Men and Women
obese (BMI 30)obese (BMI 30)
0
0.5
1
1.5
2
2.5
3
1989 1991 1993 1997
Men Women
Trends in overweightTrends in overweight
% Men overweight % Men overweight by age groupsby age groups
% Women % Women overweight by age overweight by age groupsgroups
0
5
10
15
20
25
1989 1991 1993 1997
Age 20-39 Age 40-59 Age 60+
0
5
10
15
20
25
30
1989 1991 1993 1997
Age 20-39 Age 40-59 Age 60+
Trends in overweightTrends in overweight
% Men overweight % Men overweight by rural-urban by rural-urban residenceresidence
% Women % Women overweight by overweight by rural-urban rural-urban residenceresidence
0
5
10
15
20
25
30
1989 1991 1993 1997
Urban Rural
0
5
10
15
20
25
30
1989 1991 1993 1997
Urban Rural
Trends in overweightTrends in overweight
% Men overweight % Men overweight by education levelsby education levels
% Women % Women overweight by overweight by education levelseducation levels
0
5
10
15
20
25
30
1989 1991 1993 19970
5
10
15
20
25
1989 1991 1993 1997
No EduSome PrimaryPrimaryMiddle SchoolHigh SchoolCollege
BMI Distribution: MenBMI Distribution: Men
Russel Davidson and Jean-Yves Duclos (2000), performed with DAD 4.3
BMI Distribution: BMI Distribution: WomenWomen
Russel Davidson and Jean-Yves Duclos (2000), performed with DAD 4.3
Estimating EquationEstimating Equation
The linear representation:The linear representation:
Baltagi and Wu (1999)Baltagi and Wu (1999)
2~ (0, )
2 2~ (0, ) , ~ (0, )1
1,...,
1,...,
,
h X uit it it
where u IID uit i it i
IID or IIDe vit it it itit
i n
t Ti
Estimating EquationEstimating Equation
Econometric IssuesEconometric Issues Measurement ErrorsMeasurement Errors Functional form: nonlinearity of effectsFunctional form: nonlinearity of effects Age, cohort and period effectsAge, cohort and period effects
Age, Period and Cohort Age, Period and Cohort AnalysisAnalysis
To identify: To identify: Origin-related process – cohort effect Origin-related process – cohort effect
dominantdominant Aging processAging process Instantaneous process – function of timeInstantaneous process – function of time
Linear dependency: Linear dependency: Deaton (1997), Holford (1983)Deaton (1997), Holford (1983)
, , , , , ,y ui c a t c a t i c a t
0, 0, 01 1 1 1 1 1
P A Tor c a tc a t
(3.1)
Age, Period and Cohort Age, Period and Cohort AnalysisAnalysis
Spacing of the data determines the Spacing of the data determines the number of extra constraints (n) number of extra constraints (n) neededneeded
CHNS 1989, 1991, 1993, 1997: n=2CHNS 1989, 1991, 1993, 1997: n=2 If arbitrarily impose two constraints:If arbitrarily impose two constraints:With Perfect Coliearity
age
Male age effect Female age effect
21 75-3.31
3.07
With Perfect Colinearity
cohort
Male cohort effect Female cohort effect
1906 1960-5.71
8.44
Age, Period and Cohort Age, Period and Cohort AnalysisAnalysis
Meaningful constraints:Meaningful constraints:
Source: State Statistical Bureau (1992): 1990 Population Census of China.
Age, Period and Cohort Age, Period and Cohort AnalysisAnalysis
Results from age dummies, five-year Results from age dummies, five-year cohort dummies and year dummies:cohort dummies and year dummies:
Ma
le/F
em
ale
Ag
e E
ffect
sId
en
tifie
d w
ith F
ive
-ye
ar
coh
ort
s
age
Male Age Effect Fmale Age Effect
20 25 30 35 40 45 50 55 60 65 70 75
-1.60
-1.20
-0.80
-0.40
0.00
0.40
0.80
1.20
1.60
2.00
2.40
Reduced Form Demand Reduced Form Demand (I)(I)
Male Female
BMI Calories % Fat % Protein
BMI Calories % Fat % Protein
Some primary 0.02 2.07 0.50 0.14 0.22 14.19 0.33 0.17
Primary 0.04 -1.25 0.89 0.19 0.47 1.80 1.22 0.45
Middle School 0.05 -42.83 1.78 0.37 0.01 -52.28 1.93 0.58
High School 0.21 -66.05 2.07 0.48 -0.32 -76.89 2.70 0.79
College 0.10 -57.37 2.36 0.79 -0.63 -42.83 3.02 1.23
Results from the basic model:Results from the basic model: Education effect on female BMI inversely U-shapedEducation effect on female BMI inversely U-shaped % Calories from fat and protein increase with % Calories from fat and protein increase with
educationeducation Male BMI increases with productive assets, female Male BMI increases with productive assets, female
inversely U-shapedinversely U-shaped % Calories from fat and protein increase with % Calories from fat and protein increase with
assets, except for urban men – U-shapedassets, except for urban men – U-shaped
Results from the augmented model:Results from the augmented model: Prices significant without controlling Prices significant without controlling
community dummies in BMI regressionscommunity dummies in BMI regressions Negative price elasticity of rice, eggs for Negative price elasticity of rice, eggs for
calories, fat and protein; positive price calories, fat and protein; positive price elasticity of porkelasticity of pork
Better water source – fat, protein intakes and Better water source – fat, protein intakes and male BMI: positivemale BMI: positive
More excreta – calories and protein: negativeMore excreta – calories and protein: negative
Dynamic BMI DemandDynamic BMI Demand
Estimating Equation:Estimating Equation:
Instruments: previous years HH Instruments: previous years HH resources, community characteristicsresources, community characteristics
Results:Results: In short terms lagged BMI is good In short terms lagged BMI is good
summary measure of health statussummary measure of health status Age effect for women still strong, aging Age effect for women still strong, aging