Premium subsidies and demand for private health insurance: results from a regression discontinuity design

ABSTRACT

This article investigates the impact of a private health insurance (PHI) subsidy on the demand for PHI in the context of the Australian health care system. In particular, we focus on the subpopulation of elderly Australians and exploit discontinuous increases to the universal ‘PHI rebate’ that occur when people turn 65 and 70 years. Using a regression discontinuity design, we find the policy has little effect on take-up of PHI and is best interpreted as a wealth transfer to elderly Australians who already have insurance.

KEYWORDS:

• Private health insurance
• subsidy
• regression discontinuity design

JEL CLASSIFICATION:

• H51
• I11
• l13

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ Assuming providers cannot capture the benefit themselves by increasing premiums. Since the higher subsidy only applies to people aged 65 years and over, while regulation only allows providers to charge a single premium, this appears to be a reasonable assumption.

² Apart from hospital cover, Australians can purchase private health insurance (PHI) that covers a range of ancillary health services not covered by Medicare such as dental, optical and physiotherapy. Most people (around 85% of the insured) purchase a combined policy that covers both hospital and ancillary services.

³ This positive correlation between self-assessed health and PHI in Australia is well known and its causes have been explored by Doiron, Jones, and Savage (2008), Johar and Savage (2012) and Buchmueller et al (2013).

⁴ See Lee and Lemieux (2010) for an overview of the regression discontinuity methodology.

⁵ As recommended in the literature (e.g. Imbens and Lemieux (2008)), we test for discontinuity in the conditional mean around the age 70 threshold for several relevant socio-demographic characteristic (i.e. gender, marital status, education, income, retirement status and self-assessed health). Discontinuity in these variables may challenge validity of our findings suggesting the possibility that the effects are caused by one of these characteristics rather than by the policy. The results of this specification analysis are presented in the Appendix. There is only one variable for which we cannot reject a discontinuity at the age 70 at a reasonable statistical significance level (i.e. income non-respondent in the sample of males). These results lend confidence that any effects that we find are driven by the policy and not by discontinuous changes in any of these important characteristics.

In addition to this, we also test the robustness of our results by running models with additional controls (education, income and health) and re-estimating the model for a range of possible bandwidth. Our findings are robust to these specification tests. The results of these robustness checks are available from authors upon request.