عدم تجانس ر ترجیح و انتخاب در بیمه سلامت خصوصی: مورد استرالیا

A basic prediction of theoretical models of insurance is that if consumers have private information about their risk of suffering a loss there will be a positive correlation between risk and the level of insurance coverage. We test this prediction in the context of the market for private health insurance in Australia. Despite a universal public system that provides comprehensive coverage for inpatient and outpatient care, roughly half of the adult population also carries private health insurance, the main benefit of which is more timely access to elective hospital treatment. Like several studies on different types of insurance in other countries, we find no support for the positive correlation hypothesis. Because strict underwriting regulations create strong information asymmetries, this result suggests the importance of multi-dimensional private information. Additional analyses suggest that the advantageous selection observed in this market is driven by the effect of risk aversion, the ability to make complex financial decisions and income.

A basic prediction of theoretical models of insurance is that when consumers have private information about their risk of suffering a loss – or, equivalently, if insurers are prohibited from using observable information on risk in underwriting – insurance markets will be prone to adverse selection. Market equilibria with adverse selection are characterized by a positive correlation between risk and the level of insurance coverage (Chiappori et al., 2006 and Einav et al., 2010).1 A number of recent studies have tested this prediction using data from different types of insurance markets. While research on annuities finds evidence in support of the positive correlation hypothesis (Finkelstein and Poterba, 2002, Finkelstein and Poterba, 2004 and Finkelstein and Poterba, 2006), several studies on other types of insurance find either no correlation between risk and insurance coverage, or a negative correlation. Examples include studies of health insurance markets in the US2 (Hurd and McGarry, 1997, Cardon and Hendel, 2001, Asinski, 2005 and Fang et al., 2008), the UK (Propper, 1989), and Israel (Shmueli, 2001), long term care insurance (Finkelstein and McGarry, 2006), life insurance (Cawley and Philipson, 1999) and auto insurance (Chiappori and Salanie, 2000, Dionne et al., 2001 and Saito, 2006). Broadly, there are two possible explanations for this lack of evidence in supporting the positive correlation hypothesis. One is that the information asymmetries that are central to theoretical models of insurance markets are not empirically important. According to this argument, insurers are able to obtain enough information from consumers to adequately predict their losses and set premiums accordingly.3 The second possible explanation is that there is multidimensional private information. That is, in addition to private information about the risk of experiencing a loss, there are other factors that cannot be used in setting prices that increase the demand for insurance and are negatively correlated with the risk of suffering a loss. For example, if consumers who are more risk averse are also less likely to suffer a loss – perhaps because they are more inclined to undertake preventive efforts – the positive correlation between risk and insurance coverage due to adverse selection will be attenuated or perhaps even reversed. More generally, this situation can arise if consumers with low expected claims are more likely than high risks consumers to value certain features of an insurance product that are unrelated to the actuarial cost of coverage. An overall negative correlation between coverage and claims does not imply that there are no consumers purchasing insurance because they have a high expectation of claims. Rather, a negative correlation may be observed because such consumers are outnumbered by lower risk individuals whose purchase decisions are driven by other factors. In this paper, we investigate the issue of risk selection in the Australian market for private health insurance. Several features of this market make it an important case to study. First, much of the prior research on risk selection in health insurance has used data from the US, which is an outlier among industrialized countries in both the importance of private insurance in financing health care and the link between insurance coverage and employment. In contrast, Australia is typical of other developed countries in the way that private health insurance complements a universal public health care system and is purchased directly by individuals. Private insurance is used mainly to gain access to private hospitals, thereby avoiding having to wait for care through the public system. The factors that influence the demand for insurance and the institutional arrangements that facilitate pooling may be quite different in these systems as compared to the US. Another important feature of market for private health insurance in Australia is that the pricing of contracts is highly regulated. Premiums are required to be community rated, meaning that for a given contract the same price must be charged to all consumers regardless of age, gender, health status or any other individual characteristics. By prohibiting insurers from basing premiums on observable risk factors, community rating introduces a strong form of information asymmetry into the market, which simplifies the analysis. The most appropriate test for adverse selection is based on the correlation between risk and insurance coverage conditional on all variables that are used by insurers to set prices. The analysis of an unregulated market is complicated by a lack of such variables. Under community rating, we know that all consumers face the same prices. In addition to being a good case study for general issues related to risk selection in insurance markets, our analysis is directly relevant to health policy in Australia. From the mid-1980s to the 1990s, private insurance coverage was declining in Australia. Many observers attributed this trend to adverse selection caused by the community rating rules (AIC, 1997 and Barrett and Conlon, 2003). However, empirical research on this topic has produced mixed results. A better understanding of the factors that affect risk selection in this market is important for designing and evaluating future public policies related to private health insurance in Australia. Our analysis is based on two nationally representative surveys of Australian households. First, we use data from the Australian National Health Survey (NHS) to investigate the relationship among health care utilization, insurance coverage and various factors that are likely to affect the demand for insurance. In contrast to the textbook model of insurance markets, but like other recent studies, we find that the unadjusted relationship between health risk and insurance coverage is negative. Specifically, adults with private insurance for hospital care are in better self-reported health and have lower hospital utilization than adults without private coverage. The correlation between insurance and the number of nights spent in hospital reverses when we control for proxies for other types of private information. We consider proxies for preferences regarding risk and prevention, the ability to make complex financial decisions, as well as income and employment status.

In this paper we analyze the nature of risk selection into private health insurance in Australia. In this market, underwriting regulations prohibit insurers from using any information on consumer risk in setting premiums. Under these circumstances, theoretical models with one-dimensional private information would predict a positive correlation between insurance coverage and expected claims. We test this prediction using data from a nationally representative household survey that provides information on both private insurance for hospital care and hospital utilization. As in other recent studies using data on other types of insurance in other countries, our results do not support the positive correlation hypothesis. We find that adults covered by private hospital insurance spend fewer nights in hospital over the course of a year than individuals without private coverage. Privately insured individuals are in better self-reported health and are less likely to report a recent physician's visit. Since private insurance in Australia provides no reimbursement for outpatient care, this latter result can be interpreted as evidence that the privately insured group is healthier than the group without private coverage. Thus, our results are consistent with advantageous, rather than adverse selection. Given the information asymmetries induced by the underwriting regulations, this finding suggests that there must be other factors that positively affect the demand for private health insurance but are negatively related to expected claims. Previous studies suggest three possible factors: preferences regarding risk and prevention, cognitive ability and income. Several pieces of evidence suggest that all three are important determinants of private insurance coverage in Australia. First, when we condition on a limited set of covariates the correlation between health insurance and utilization becomes positive. Questions on the reasons for purchasing health insurance show that, consistent with textbook theory, some consumers do buy insurance because they anticipate the need for medical care. As would be expected, these individuals are in poorer than average health and have higher utilization. However, these consumers are outnumbered by healthier consumers who purchase insurance for other reasons. Consumers who purchase private insurance for reasons related to risk aversion or because they prefer the convenience and amenities provided by private hospitals are in better health than individuals without private coverage. Our analysis of household expenditure data indicates that consumers who purchase private health insurance are also significantly more likely to purchase other types of insurance. Our results provide insights into the factors driving empirical findings of advantageous selection in a setting where private health insurance complements a universal public health care system. In Australia as in other countries, patients who receive care through the public system face relatively low direct charges but may face very high implicit costs as a result of long waiting lists for certain procedures. Consumers with a high opportunity cost of time will face high total costs (explicit plus implicit) even if they are relatively unlikely to need care. By making it possible to “jump the queue” private coverage essentially allows patients to insure against these time costs. As a result, high-income patients who are low risk from the perspective of insurance companies will have a higher than average demand for private insurance. To the extent that high-income patients have a relatively strong preference for amenities that are available to private patients, such as the ability to choose one's own doctor and having a private room, this will also contribute to advantageous selection. The results have interesting policy implications. As noted in the introduction, for over a decade the Australian government has used subsidies to encourage the purchase of private insurance. In the presence of advantageous selection, de Meza and Webb (2001) show that a tax on insurance is welfare improving. This would suggest that subsidies, such as those in Australia, are likely to be welfare reducing. On the other hand, by allowing patients with the greatest opportunity cost of time to avoid long waits for elective procedures, private insurance may lower the welfare costs associated with non-price rationing.