Advantageous selection in private health insurance: The case of Australia

When consumers have private information about risk of suffering a loss, or equivalently, if insurers are prohibited from using observable information on risk in underwriting, theoretical models of insurance predict adverse selection. Yet the most common finding in empirical studies is that of no positive correlation between risk and insurance coverage. This is found for different types of insurance (e.g. car, health, life) and in different countries (e.g. France, US, UK, Israel) suggesting a fundamental relationship involving private information and consumer preferences. In this paper, we investigate the nature of risk selection in the Australian market for private health insurance in which community rated private health insurance complements a universal public health care system. We use National Health Survey data on hospital utilisation and individual characteristics to construct an empirical analogue for the risk variable in the Rothschild and Stiglitz model. Estimating the relationship between insurance and risk semi-parametrically, we find robust evidence of favourable selection. To explore the extent to which underlying risk preferences rather than risk drives the decision to purchase health insurance, we use Household Expenditure Survey data to model decisions to purchase a range of insurance products (health, life, accident, home, car) and to engage in risky behaviours (smoking and various forms of gambling). Correlations between residuals in the model suggest that advantageous selection is driven by risk aversion, which theoretical models do not typically capture.