HEFPA Paper Digest I Supon Limwattananon
WP2 WP1
Health shocks - Labor supply - Income (earned and unearned) - Medical spending Food consumption Non-food consumption - Health insurance - Coping methods: - Saving - Gift - Borrowing/loan - Asset - Social insurance/security - Informal solidarity Consumption insurance Financial risks
Paper 3
Which kind of insurance can protect welfare loss? For common minor illnesses vs. for unanticipated major illnesses First-dollar coverage with low capped benefits vs. catastrophic insurance with patient cost-sharing 1 1 Townsend (1995); Kochar (1995) found families in LICs were able to insure illness shocks fairly well.
Objective
Findings Section III: Section IV:
Panel data
Fixed-effects model
(1) ADLs vs. (2) Self-reported illness symptoms Symptom lasting > 1 mo. Any symptom mean (SD) % mean (SD) h ij
Problems on self-reported illnesses
L ij X ij Labor supply /wk This has to be imputed for informal sectors per wk (C/n) ij
h ij X ij L ij (1.1) per wk Section III
h ij L ij per wk Interpretation: Moving from being able to perform all ADLs to being able to perform none would result in - lowering hours of work by 30.9 hours per week (84% of baseline mean hours) % likelihood of becoming labor-force nonparticipant - a reduction of earnings by Rp.20,170 (~ baseline mean earnings) - an increase in medical spending of Rp.1,180
Section III
Section IV: Consumption insurance
ln(C/n) ij h ij X ij Moving from being able to perform all ADLs to being able to perform none would lower consumption by 19.5%
Section IV
Section V: The extent to which households are able to insure consumption Biased est.: For each Rp. that income falls, consumption falls by only 3% Unbiased est.: For each Rp. that income falls, consumption falls by 35% Households are able to insure only 65% of the consumption with respect to income loss due to a loss in ADL
Instrument Variable (IV) Y = 1 + 1 h + Z 1 + 1 C = 2 + 2 + X 2 + 2 Predicted Y as an instrument Step 1. Income Step 2. Consumption C = 0 + 0 Y + X 0 + 0 OLS method Y is endogenous: some unobserved variables affect both Y and C. Hence, Is biased (change in Y affecting change in C is spurious). Consumption: IV method
Paper 4
S ht-1
ln y ht S ht-1 XhtXht e.g., gift, remittance, pension, compensation e.g., wage/salary, agriculture, family business esp., direct, nonmedical costs of care
yhtyht
Rural Urban Effects on income per household An urban household’s earned income is more vulnerable to death shocks than an rural household’s. Statistically non-significant Regression coefficient and (t-statistics) Statistically non-significant An increase in unearned income offsets a decrease in earned income; whereby other-than death shocks in urban area are larger than in rural area.
Urban Rural Effects on income per capita Effect of a death on per capita income Is statistically non-significant and is less than on household income.
Effects on medical expenditure (1) Urban Rural
Effects on medical expenditure (2) Insured Uninsured
Effects on food consumption Rural Urban Households cannot smooth their food consumption in the face of some health shocks!
Effects on non-food/non-medical consumption Urban Rural The evidence is more mixed!
Pr (y>0) E[ y| y>0] (Housing)
Health shocks Year (wave 1: N = 5,673) Year (wave 2: N = 5,495) % Dead: Any HH members0% 2.3% % Ill: Any HH members57.0%53.8% % Ill: HH head37.8%35.1% % Ill: Other than head19.1%18.7% % Ill: Working member43.4%40.7% % Hospitalized: Any HH members21.3%NA % Hospitalized: HH head10.3%NA % Hospitalized: Other than head11.0%NA % Hospitalized: Working member14.7%NA
Economic consequences Year (wave 2) Year (wave 3) Mean Unearned income (Baht) Mean Earned income (Baht) 4,357 4,664 Mean Total income (Baht) 5,269 5,656 % Catastrophic health exp. (>10% total exp.)3.3% % Positive health exp.57.3%51.0% Mean Health exp. (Baht), given positive Mean Food expenditure (Baht) 1,173 1,260 Mean Nonfood/non-health exp. (Baht) 1,867 2,186 % Positive education exp.56.2%55.8% Mean Education exp. (Baht), given positive
Covariates for adjustment Year 2006Year 2007 % Urban 29.1%31.7% % Rural 70.9%68.3% % Bangkok 15.8%17.1% % Central 16.9%17.4% % North 19.1%19.4% % Northeast 35.2%35.0% % South 12.9%11.1% Mean Age (years) % Male 70.1% % Couple 73.9%75.2% % Primary educated 73.9%71.2% % Secondary educated 14.2%15.2% % Higher educated11.8%13.6% Household head
Effect on per capita income Thailand –whole country Health shock UnearnedEarnedTotal Dead: Any HH members Ill: Any HH members Ill: HH head Ill: Other than head Ill: Working member Hospitalized: Any HH members Hospitalized: HH head Hospitalized: Other than head Hospitalized: Working member
Effect on per capita income Thailand –urban vs. rural Health shock UnearnedEarnedTotal Dead: Any HH members Hospitalized: Any HH members Hospitalized: HH head Hospitalized: Other than head Hospitalized: Working member Dead: Any HH members Hospitalized: Any HH members Hospitalized: HH head Hospitalized: Other than head Hospitalized: Working member Rural Urban
Effect on per capita health spending (W2) Thailand –whole country Health shock (W1) Catast. exp. Positive exp. Exp. if positive Overall exp. Ill: Any HH members Ill: HH head Ill: Other than head Ill: Working member Hospitalized: Any HH members Hospitalized: HH head Hospitalized: Other than head Hospitalized: Working member
Effect on per capita health spending (W2) Thailand –urban vs. rural Health shock (W1) Catast. exp. Positive exp. Exp. if positive Overall exp. Hospitalized: Any HH members Hospitalized: HH head Hospitalized: Other than head Hospitalized: Working member Hospitalized: Any HH members Hospitalized: HH head Hospitalized: Other than head Hospitalized: Working member Rural Urban
Effect on per capita health spending (W3) Thailand –CS vs. SS vs. UC Health shock (W2) Catast. exp. Positive exp. Exp. if positive Overall exp. Ill: Any HH members Ill: HH head Ill: Any HH members Ill: HH head Ill: Any HH members Ill: HH head UC SS CS
Effect on per capita non-health spending Thailand –whole country Health shock FoodNonfood Dead: Any HH members Ill: Any HH members Ill: HH head Ill: Other than head Ill: Working member Hospitalized: Any HH members Hospitalized: HH head Hospitalized: Other than head Hospitalized: Working member
Effect on per capita non-health spending Thailand –urban vs. rural Health shock FoodNonfood Dead: Any HH members Hospitalized: Any HH members Hospitalized: HH head Hospitalized: Other than head Hospitalized: Working member Dead: Any HH members Hospitalized: Any HH members Hospitalized: HH head Hospitalized: Other than head Hospitalized: Working member Rural Urban
Effect on per capita education spending Thailand –whole country Health shock Positive exp. Exp. if positive Overall exp. Dead: Any HH members Ill: Any HH members Ill: HH head Ill: Other than head Ill: Working member Hospitalized: Any HH members Hospitalized: HH head Hospitalized: Other than head Hospitalized: Working member
Effect on per capita education spending Thailand –urban vs. rural Health shock Positive exp. Exp. if positive Overall exp. Dead: Any HH members Hospitalized: Any HH members Hospitalized: HH head Hospitalized: Other than head Hospitalized: Working member Dead: Any HH members Hospitalized: Any HH members Hospitalized: HH head Hospitalized: Other than head Hospitalized: Working member Rural Urban
VHLSS (Vietnam Household Living Standards Survey) Coping Strategies Source: VHLSS 2006 Report
Paper 5
(coping strategies) C = 2,580/2,760 = 0.93 Mean of proportion of exp. financed by coping Coping-adjusted health expenditure ratio (P) (High spending households) P = unadjusted C
Cumulative distributions of health expenditure ratios w = Unadjusted; P = Coping-adjusted (w) (P)