Abstract
The Australian government introduced three major private health insurance policy initiatives in recent years. These are, in chronological order, (i) the Private Health Insurance Incentives Scheme (PHIIS), which imposes a tax levy on high-income earners who do not have private health insurance and provides a means-tested subsidy schedule for low-income earners who purchase PHI; (ii) a 30% premium rebate for all private health insurance policies to replace the means-tested component under PHIIS; and (iii) lifetime health cover, which permits a limited form of age-related risk rating by insurance funds. Together, these policy changes have been effective in encouraging the uptake of PHI; the percentage of the population covered by PHI rose from 31% in 1999 to 45% at the end of 2001. The difficult issue, however, is in disentangling the effects of the three policy changes, given that they were introduced in quick succession. This article attempts to evaluate the effect of lifetime health cover using a regression discontinuity design, an approach that makes use of cross-section data that allows the effect of lifetime health cover to be isolated via local regression. The results suggest that the importance of lifetime health cover appears to be grossly over-rated in previous studies. Our estimates indicate that it accounts for roughly 22–32% of the combined effects of all the policy initiatives introduced in the late 1990s. While these figures suggest that its effect is clearly significant, it is nonetheless nowhere near the effect often associated with lifetime health cover.
Notes
1 Note that a ceiling of 70% applies, and the premium is determined at entry and once determined, would not rise with the member's age. Also, people over the age of 65 years in July 2000 are exempt from LHC.
2 However, using a modified Rothschild--Stiglitz insurance model, Brown and Connelly (Citation2005) arrives at a contrary view that LHC is unlikely to be an effective policy, especially in attracting individuals with low health risks.
3 Another reason that those above 64 years of age are deleted from the sample is because LHC only applies to individuals below 68 years old as at July 2000. One could, in principle, apply the same RD design to those just above and below 68 years of age, but there are not enough observations to give reliable estimates.
4 There are a number of individuals with ancillary but not hospital cover. These individuals are regarded as without PHI in the empirical estimation below.
5 The computation is performed in STATA using the command LOWESS, with the default bandwidth of 0.8.
6 An alternative, albeit somewhat dubious, estimate is to take the difference between the sample proportion of single individuals with PHI in 2001 and 1995. In our case, the overall sample proportion of individuals with PHI is 0.365 for 2001 and 0.25 for 1995, the difference of 0.115 is thus a rough estimate of the total combined effects. This is, however, likely to underestimate the combined effects, since PHI membership is, from , clearly on a declining trend. This means the proportion of individuals with PHI would have been even lower by 2001 had there been no policy initiatives.
7 See http://www.phiac.gov.au/statistics/trends/index.htm, accessed on 24 October 2005.
8 See Appendix B for an estimate of the effects of the Medicare Levy Surcharge using RD.