Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

25
Wealth, Education and Demand for Medical Care ___Evidence from Rural China Feng Jin Qin Bei Yu Yangyang

description

Wealth, Education and Demand for Medical Care ___ Evidence from Rural China. Feng Jin Qin Bei Yu Yangyang. Background. In many developing areas, health is much more important, since a person with poor health is more likely to be burdened with the tremendous medical expenditure - PowerPoint PPT Presentation

Transcript of Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

Page 1: Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

Wealth, Education and Demand for Medical Care

___Evidence from Rural China

Feng JinQin Bei

Yu Yangyang

Page 2: Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

Background

• In many developing areas, health is much more important, since a person with poor health is more likely to be burdened with the tremendous medical expenditure

• Due to the collapse of health care system and the increasing of medical price, the sick people suffer heavy financial burden

• In rural China, disease has been cited as one of the top two reasons accounting for impoverishment

Page 3: Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

Wealth education and health(raw data)

22.

53

3.5

Hea

lth (S

RH

)

0 20 40 60 80 100age

Fitted values Fitted values

22.

53

3.5

Hea

lth (S

RH

)

0 20 40 60 80 100age

Fitted values Fitted values

Age-wealth-health Graph A Age-education-health Graph B

Page 4: Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

Research Objective

• To check the expected medical burden of people with certain wealth and education stocks

Who has the higher probability to be sickWho has more expenditure after sickWho has more heavy medical burden

• To test the hypothesis of use-related deprecation rate on health

Page 5: Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

Literature

• Grossman (1972): health capital and demand for health

• Muurinen (1982): three separated stocks (education, wealth and health) are substitutable

• Empirical tests :Muurinen and Le Grand (1985) , Van Doorslaer (1987) , Case and Deaton (2004)

Page 6: Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

Model

• Demand for medical care• Two part model of health expenditure• Measurement issues

Page 7: Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

Demand for medical expenditure

))(),(),(),(),(),(()( tXtEtagetptwtHftM

( ) ( ( ), ( ), ( ), ( ), ( ))H t g w t p t age t E t X t

( ) ( ( ), ( ), ( ), ( ), ( ))M t m w t p t age t E t X t

Page 8: Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

Two-part models

• The first part: decision for participation

• The second part: decision for expenditure

iccicic XI 1

iccicicic uXIMED 2)0|log(

)exp()()|( 1 icicicic XXXMEDE

Page 9: Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

Marginal effect( ) (Pr( 0) ( | 0))

( | 0) Pr( 0)Pr( 0) ( | 0)

k k

k k

E y y E y yx x

E y y yy E y yx x

( | 1) ( | 0)

Pr( 0 | 1) Pr( 0 | 0) ( | 0 | 1)

Pr( 0 | 0) ( | 0 | 1) ( | 0 | 0)

k k

k k k

k k k

E y x E y x

y x y x E y y x

y x E y y x E y y x

Page 10: Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

Measurement issues

• Wealth: household income per capita

• Medical price: possible problem of self-selection. Using survey data of health providers. We use average price paying for a treatment of cold or flu in the community

• Insurance: public insurance, worker insurance, cooperative medical insurance and all kinds of insurance. whether the individual has medical insurance

Page 11: Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

Data

• CHNS (China Health and Nutrition Survey) 1991,1997 data of rural China

• collected by Carolina Population Center (CPC) at the University of North Carolina at Chapel Hill, the Institute of Nutrition on Food Hygiene, and the Chinese Academy of Preventive Medicine

Page 12: Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

Sample size

province community Household individual Illness samples

1991 8 128 2550 10374 884 (8.59%)

1997 8 128 2590 9583 504 (5.18%)

Page 13: Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

Medical expenditure and household income

Percentage of samples who have medical expenditure when ill (%)

Medical expenditure (mean) (yuan)

Household adjusted income (total sample)(mean) (yuan)

Household adjusted income (illness sample)(mean) (yuan)

1991 77.26 172.69 1494.27 1470.95

1997 84.13 312.96 1302.48 1298.22

Page 14: Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

Full samples Illness samples

Variable definition 1991 1997 1991 1997

edu Education (year) 5.78 5.04 4.73 4.02

age1 =1, 15 to 35 years old

0.46 0.44 0.25 0.17

age2 =1, 36 to 65 years old

0.42 0.43 0.58 0.59

age3 =1, older than 65

0.09 0.10 0.15 0.23

Education and age

Page 15: Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

Price and Insurance

Province Medical Price Health insurance coverage (%)

1991年 1997年 1991年 1997年

Heilongjiang - 22.35 - 9.2

Liaoning 4.31 - 18.0 -

Jiangsu 5.58 21.98 45.7 36.8

Shandong 2.75 18.63 18.1 39.3

Henan 2.61 8.85 14.0 18.1

Hubei 5.24 17.53 20.4 9.7

Guanxi 2.20 9.64 10.3 12.4

Guizhou 3.72 12.17 7.8 5.5

Hunan 3.46 21.21 9.1 3.8

Page 16: Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

Results

• Possibility to be sick (Xtprobit)• Possibility to have expenditure (Xtprobit)• Possibility to have expenditure after sick (b

udget constrain) (Xtprobit)• Medical expenditure (random effect)• Marginal effect on medical expenditure of t

wo part model

Page 17: Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

Marginal effect on probability of illness and having medical expenditure

Dependent variable

ill(full sample)

haveexp(full sample)

haveexp(sick sample)

Ln(inc) -0.004( 0.03) 0.000( 0.03) 0.069***( 0.021)Ln(price) 0.02( 0.04) 0.003( 0.04) 0.027( 0.025)

Ln(cominc) 0.011( 0.008) 0.006( 0.007) -0.086*( 0.053)Age1 -0.058***( 0.001) -0.040***( 0.006) 0.058*( 0.039)Age2 -0.017***( 0.006) -0.008( 0.005) -0.060( 0.035)Edu -0.002***( 0.000) -0.001**( 0.001) 0.011***( 0.004)Job -0.021***( 0.006) -0.0141***( 0.005) 0.083**( 0.037)

Insurance 0.017***( 0.1161) 0.009*( 0.005) -0.026( 0.037)Year1997 -0.038*** (0.007) -0.022*** (0.006) 0.107***( 0.355)Sigma_u 0.389 (0.033) 0.409 (0.036) 0.450 (0.083)

LR test of rho=0

Prob>=chaibar2=0 Prob>=chaibar2=0 Prob>=chaibar2=0

Observation 14863 14881 1102

Page 18: Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

Medical Expenditure

Dependent variable: Ln(medical expenditure) Random effect model

Ln(inc) 0.104( 0.110)Ln(price) 0.223*( 0.133)

Ln(cominc) 0.072( 0.264)Age1 -0.078( 0.224)Age2 -0.088( 0.192)Edu -0.036**( 0.019)Job 0.010( 0.172)

Insurance 0.085( 0.178)Severe2 1.131***( 0.134)Severe3 1.957***( 0.190)

Year1997 0.217( 0.203)R-sq 0.150

Observation 884

Breusch and Pagan Lagrangian multiplier test for random effects:

Test: Var(u) = 0

chi2(1) = 11.58 Prob>chi2= 0.0007

Page 19: Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

Marginal effect of two-part model

variableMarginal effect (unconditional)

Marginal effect (conditional on illness)

income 0.006 0.331price 0.020 0.280Education -0.039 0.009Age2 -0.146 0.114Age3 -0.024 0.270job -0.047 0.242insurance 0.036 -0.038

Page 20: Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

Test the endogineity of medical insurance

• people with medical insurance might have some unobserved characteristics which influence their medical expenditure

• We use “if the village enterprises subsidize the insurance” as an Instrument Variable

Page 21: Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

*1 2i i i iy x z

* 0iy 1iy indicate the individual has medical insurance

The first stage regression of ivprobit (instrumented: insurance)

Dependent variable: having insurance

Participating equation (full sample)

Participating equation(illness sample)

Have subsidy or not 0.153*** (0.009) 0.082** (0.034)

Walt test for exogeneity Prob>Chi(2) =0.292 Prob>Chi(2) =0.783observations 14881 1102

Page 22: Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

Medical burden by income group and education group

inc1 inc2 inc3 inc4 inc5 edu1 edu2

Probability of being ill

1991 0.089 0.088 0.089 0.097 0.102 0.133 0.050

1997 0.052 0.051 0.053 0.054 0.057 0.082 0.028

Medical burden (%)

full smaple

1991 1.03 0.53 0.41 0.34 0.24 0.86 0.21

1997 1.19 0.55 0.4 0.3 0.21 0.97 0.22

illness sample

1991 13.53 7.09 5.99 3.86 2.81 8.1 5.5

1997 37.63 12.01 9.64 7.68 5.01 18.61 10.6

Page 23: Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

conclusions

• the less educated people have higher probability to be sick and expend more on medical care after sick.

• the income elasticity of demand for medical care is low, so the lower income people have heavier medical burden .

• due to the lack of price elasticity, the medical burden of lower income people is growing more fast

Page 24: Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

Policy implications• the inequality is much larger if taking account of

the health inequality and the heavier medical burden imposed in poor people

• it is particularly emergency to establish appropriate and widely covered public health insurance to share the risk of illness and medical expenditure

• A proper insurance scheme will play a redistributive role, since the poor and the low educated people face higher risk

Page 25: Wealth, Education and Demand for Medical Care ___ Evidence from Rural China

Thank You!