Title: Assessing the Impact of Health Insurance on Demand for Medical Health Care: Preliminary Finding from
1Assessing the Impact of Health Insurance on
Demand for Medical Health Care Preliminary
Finding from IFLS 1997 Data
- Budi Hidayat1, Robert Nordyk2, Glenn Melnick2,
Cheryl Cashin1, Rainer Sauerborn1 - 1University of Heidelberg Germany
- 2USC/RAND Corporation - USA
2Introduction
- What is insurance?
- any kind of HI plan that covers financial
expenses or health services benefits to
beneficiaries - it ranges from traditional indemnity insurance
(reimburse medical expenses) to a HMO (provides
services as the benefits) (Thabrany, 1995)
3 Key Insurance Programs in Indonesia
4Objectives
- Evidence the pattern of health care use between
insured (ASKES, ASTEK, CARD PRIVATE) and
uninsured people - Asses the impact of insurance on demand for
medical care
5Theoretical framework
- Insurance increase the utilization by reducing
effective price (Feldstein, 1996)
Price
D (uninsured)
P
D1 (insured)
P1
Q
Q1
Quantity
0
6Theoretical framework (2)
Utility-based model of demand for health inputs.
Max UjU?Ui(Ci, Hi, C-i, H-i), (Ci, Hi, C-i,
H-i)?
- Health production function
Hi H(H0i, Mti, Mti-1, Xi, Zi, vt, ui, ?it )
- Household budget
PCCj PM(Ii)Mj Ydj, ?where Ydj Yj-PIIj
Demand functions for each individual
(reduced-form) Di D(Ii, I-i, PMi,
Pij, Ydj, PC, Ht-Ii (Mt-Ii,H0i),Xi, Zj, vt,ui,
?it)
7The Methods
Maximum Likelihood Estimation ? logistic
regression (probability odd-ratios)
Pui(visitgt0) Xß ?
1 Pui ------------ 1 e-Zui where
Zui Xß ?
1 1-Pui ---------- 1 eZui
1 eZui (Pui/(1-Pui)) ------------- eZui
? odd ratios 1 e-Zui
8The Data Source
- Indonesian Family Life Survey
- Panel Study fielded in 1993, 1997,(1999) and
2001. - Samples 7200 Households, gt 30,000 respondents,
13/27 provinces covering 83 of population. - Community/Facility Survey linked to HH Survey
- Issues related to health and nutrition, family
planning, expenditure and poverty, and education
- On this study ? IFLS 1997 (IFLS2)
- Insurance status Book 3
- Health care use (a) public hospitals (b)
private hospitals (c) public health centers (d)
private doctors (e) private paramedics (f)
private midwives (g) traditional healer and (h)
others - Health status self-reported morbidity,
self-reported of health status, and Activities of
Daily Living (ADL ) - Other covariates Roster
9Variables
- Dependent variables health care uses
- Two major categories
- Outpatient ? dichotomous variables (1 if at
least one visit at the last four weeks 0
otherwise) - pooled (all providers, public facilities
private facilities) - health centers, private doctors, public
hospitals, private hospitals - Inpatient ? dichotomous variables (1 if at
least one visit at the last twelve months 0
otherwise) - pooled all providers, pooled public facilities
and pooled private hospitals
10Variables (2)
- Independent variables
- Insurance status as dummy (1 if had insurance, 0
otherwise) e.g., ASKES, ASTEK, CARD, PRIVATE - Health status as dummy, e.g., having symptom,
having at least one limited daily living (ADL),
poor health status - Household covariates ? per capita expenditure
quintiles (adjusted with CPI in 1997) - Socio demographic ? age, gender, marital status,
education, provinces, urban/rural - Other covariates travel cost
11Distribution of HI (Fraction covered)
Source IFLS, 1997 Wave.
12HI coverage by expenditures quintiles
Total insured 14.4
Source IFLS, 1997 Wave.
13HI coverage by expenditures quintiles
Source IFLS, 1997 Wave.
14Rate outpatient care in all, public, and private
facilities
15Rate outpatient care in four types of providers
16Rate inpatient care in all, public, and private
facilities
17significant at 10 significant at 5
significant at 1
18significant at 10 significant at 5
significant at 1
19significant at 10 significant at 5
significant at 1
20Summary of insurance effects on the use of
outpatient care
1 omitted uninsured () in better access than
uninsured
significant at 10 significant at 5
significant at 1
21Summary of insurance effects on the use of
inpatient care
1 omitted uninsured () in better access than
uninsured
significant at 10 significant at 5
significant at 1
22Summary of findings and implications for further
research
- Evidence that insured people were more demanded
on medical care than uninsured counterparts. - The effects of insurance on demand for medical
care was higher on the outpatient than on the
inpatient care. - However, the study was not free from the
weakness, e.g. - Insurance enter into the demand models
EXOGENOUSLY, in fact, there is a possibility of
endogenous/adverse selection, leading bias
estimates. - Reassignment of insurance variable (ASKES
ASTEK) was done based on the employment status
found in IFLS Wave 1, leading underestimate of
coverage.
23Summary of findings and implications for further
research (2)
- Proposing further study ? Econometric modeling
(e.g. two stages procedures) for - (1) treating endogeneity of insurance
- (2) involving various dependent variables (count
data binary data)
- The results then worth full for
- Policy Analyses program evaluation (e.g.,
Effect insurance on health input and health
outcome) - Expanding HI
- Development insurance (e.g., through provide
basis data for actuarial).
24Thank you
25 The Poor Have Received Little Govt
Distribution of Public Subsidies, Rp/cap/mo
Source Van del Walle, The Johns Hopkins Univ.
Press, 1995.
26 Insurance pays little of that
27Rate outpatient in all providers by urban/rural
28Outpatient rate by insurance status urban/rural
29Outpatient rate by insurance status gender
30Outpatient rate overall by insurance status
gender