INDIVIDUAL DEMAND FOR HEALTH INSURANCE WHEN HEALTH CARE IS WIDELY ACCESSIBLE 1 Arpah Abu-Bakar,...
-
Upload
martha-lee -
Category
Documents
-
view
215 -
download
0
Transcript of INDIVIDUAL DEMAND FOR HEALTH INSURANCE WHEN HEALTH CARE IS WIDELY ACCESSIBLE 1 Arpah Abu-Bakar,...
INDIVIDUAL DEMAND FOR HEALTH INSURANCE WHEN HEALTH CARE IS
WIDELY ACCESSIBLE 1
Arpah Abu-Bakar, Angappan Regupathi, Syed M Aljunid
2013 China International Conference on Insurance and Risk Management 17-20 July Kunming, China
OUTLINE
• Motivation of the Study• Research Objectives• Hypotheses• Research Methodology• Findings• Conclusion and Recommendations
2
• Why Individual Demand?– Most previous studies focus on household
demand
• Why Malaysia?– Different Health Care System
• Public Health Care is Widely Accessible but• High private spending on health care [OOP cost is
the highest source of financing]– Multi-ethnicity & Multi-religion Society
3
Motivation
• Why Current Empirical Evidence?– The Government is looking for a mechanism
to reduce its financing burden• “…health financing scheme to meet health care cost.” (7th
Malaysia Plan)• “provide consumer with a wider choice in the purchase of
health services from both the public and private sector.” (8th Malaysia Plan)
• Towards achieving better health through consolidation of services…between the public and private sectors (9th Malaysia Plan)
• Promoting private health care. (10th Malaysia Plan)
4
Motivation
Objectives
1. To determine the factors that affect the individual demand for private health insurance
2. To predict the likelihood of a person buying health insurance, given the person’s characteristics
5
6
Variables Effect on Probability of Purchase
X1 = income +
X2 = age +
X3 = gender Female +
X4= race-religion Non-Muslim +
X5 = education level +
X6 = marital status Married +
X7= household size -
X8 = type of occupation Service -
X9= job sector Public -
X10= urban vs rural Urban +
X11= distance to the private hospital -
X12= frequency of visit to inpatient and outpatient +
X13= out-of-pocket cost +
X14 = health status NS
X15= attitude towards risk Risk Averse +
HYPOTHESIS
DATA
• National Health & Morbidity Survey III; Cross sectional data (2006)
• Data Screening– 34,539 respondents answered Module B, 18.8% owned
some type of MHI• Module B: Health Expenditure, Hospitalization, Private Health
Insurance
– 14,233 cases with no missing values; Split to two data set. Further split into Salaried & Non-salaried individuals
– N = 4997 to fit the model and N = 5119 to test the model
7
METHODS
• Unit of Analysis– An individual who can purchase health
insurance for him/herself• Logistic Regression◦ DV – Either own or do not own HI◦ Base Category - Malay male individuals, not
married, have good health, live in urban area. Have tertiary education and work in the public sector.
Race-Religion
MalaysNon-Malay MuslimsNon-Muslims
9
RaceMalaysChineseIndiansOther BumiputrasOthers
ReligionIslamChristianityBuddhismHinduismOthers
Original Categories for Race and Religion
Risk Attitude
Wearing Helmet
Wearing Seatbelt
Risk Attitude
1 1 1 – Risk Averse
1 2 2
2 1 2 – Moderate Risk Averse
1 3 3
3 1 3
2 2 3 – Risk Neutral
2 3 4
3 2 4 – Moderate Risk Taker
3 3 5 – Risk Taker
• Smokers vs Non Smokers
• Safety Behaviors
10
Risk Attitude Scales for Safety Behaviors
11
SalariedVariables Coefficient Odd Ratio
Income 1.193** 3.298Age 0.143** 1.154
Age Square -0.002** 0.998
Female 0.203* 1.225Non Malay MuslimNon Muslim 1.171** 3.227Secondary Education -0.256* 0.774Primary Education -0.871** 0.419No Education -0.914** 0.401Married
Divorcee
Widow/widower
Household Size
Non Service SectorPrivate Sector Employee -0.734** 0.48Self-employed -0.691** 0.501RuralDistance to Private HospitalNumber of In/Outpatient VisitsOOP costBad Health StatusSafety Behaviour -0.122* 0.885Constant -11.12** 0.000
Results
Performance Criteria
• Classification table– 64.6% of individuals correctly classified
as owners– 78.9% of individuals correctly classified
as non-owners
• Hosmer and Lemeshow Test– Chi-Square is not significant
• Cox & Snell’s R sq = .217; Nagelkerke R sq = .316
13
Variables Hypothesis Actual Results
X1 = income + +
X2 = age + Not linear
X3 = gender Female + Female +
X4= race-religion Non-Muslim + Non-Muslim +
X5 = education level + +
X6 = marital status Married + NS
X7= household size - NS
X8 = type of occupation Service - NS
X9= job sector Public - Public +
X10= urban vs rural Urban + NS
X11= distance to the private
hospital- NS
X12= frequency of visit to inpatient
and outpatient+ NS
X13= out-of-pocket cost + NS
X14 = health status NS NS
X15= attitude towards risk Risk Averse + Risk Averse +
Findings: Hypothesis vs Results
ResultsSignificant variables Previous EvidenceThe higher the income the more likely to purchase HI
Propper (1989); Kronik & Gilmer (1999); Gruber & Poterba (1994); Liu & Chen (2002)
Female was more likely to purchase HI Auerbach & Ohri (2006); Liu & Chen (2002); Long & Marquis (2002); Marquis & Long (1995)
Non Muslims were 3 times more likely to purchase HI
Gruber & Poterba (1994)
More educated individuals were more likely to purchase HI
Auerbach & Ohri (2006), Gruber & Poterba(1994), Marquis et al.(2004), Besley et al. (1999) & Dewar (1998)
Self-employed & private sector employees were less likely to purchase HI compared to civil servants
Not consistent with Besley et al. (1999)
As age increase the odd of buying HI increases
Consistent with Auerbach & Ohri (2006); Marquis et al (2006)
Risk Averse individual is more likely to purchase HI
Pauly & Herring (2007)
14
Prediction Model
• Reduced-Form Model• Measure the Predictive Power of the Model
using the 2nd half of the data set• Chi-square test shows that there is a
significant association between the predicted purchased of HI and the actual purchase of HI
• This model is useful for predicting potential HI buyers
15
CONCLUSION
• Those who are likely to buy are high income earners, older and more educated individuals, female, non-Muslims, public servants and risk averse individuals
• Efforts to increase health insurance ownership– Awareness program– Programs to increase individual income level
• Using prediction model to predict health insurance potential buyers
16
Limitations
• Cross-sectional data– Out-of-pocket cost– Number of in-patient/out-patient visits
• Potential duplicate HI ownership
17
Thank you!
• Correspondence:Arpah Abu-BakarBanking and Risk Management DepartmentCollege of Business, Universiti Utara [email protected] / [email protected]
18
Main references• Propper, C. (1989). An Econometric Analysis of the Demand for Private
Health Insurance in England and Wales. Applied Economics, 21(6), 777-792.• Manning, W. G., Newhouse, J. P., Duan, N., Keeler, E. B., & Leibowitz, A.
(1987). Health Insurance and the Demand for Medical Care: Evidence from a Randomized Experiment. The American Economic Review, 77(3), 251-277.
• Liu, T. C., & Chen, C. S. (2002). An Analysis of Private Health Insurance Purchasing Decisions with National Health Insurance in Taiwan. Social Science and Medicine, 55, 755-774.