Measuring moral hazard and adverse selection by propensity scoring in the mixed health care economy of Hong Kong

OBJECTIVES: To evaluate the presence of moral hazard, adjusted for the propensity to have self-purchased insurance policies, employer-based medical benefits, and welfare-associated medical benefits in Hong Kong. METHODS: Based on 2005 population survey, we used logistic regression and zero-truncated negative binomial/Poisson regressions to assess the presence of moral hazard by comparing inpatient and outpatient utilization between insured and uninsured individuals. We fitted each enabling factor specific to the type of service covered, and adjusted for predisposing socioeconomic and demographic factors. We used a propensity score approach to account for potential adverse selection. RESULTS: Employment-based benefits coverage was associated with increased access and intensity of use for both inpatient and outpatient care, except for public hospital use. Similarly, welfare-based coverage had comparable effect sizes as employment-based schemes, except for the total number of public ambulatory episodes. Self-purchased insurance facilitated access but did not apparently induce greater demand of services among ever users. Nevertheless, there was no evidence of moral hazard in public hospital use. CONCLUSIONS: Our findings suggest that employment-based benefits coverage lead to the greatest degree of moral hazard in Hong Kong. Future studies should focus on confirming these observational findings using a randomized design.