Adverse Selection and Advantageous Selection in Insurance Markets

This dissertation consists of three essays about adverse selection and advantageous selection in life insurance and health insurance markets.

Firstly, I confirm the advantageous selection in voluntary private health insurance markets in Europe and detect the sources of such advantageous selection by using data from Survey of Health, Ageing and Retirement in Europe (SHARE). Specifically, I find, on the extensive margin, individuals with symptom are less likely to own VPHI than those without any symptom; on the intensive margin, the more the number of symptoms the individual has, the less likely she has VPHI. Same conclusion can be obtained when using a subjective measure of health. The sources of this advantageous selection include asset, education, longevity expectations, as well as cognitive ability. Conditional on these factors, individuals whose health is worse are more likely to purchase VPHI.

Secondly, I identify the adverse selection problem in life insurance markets in the presence of both adverse and advantageous private information. Conventional theory for private information of adverse selection predicts a positive correlation between insurance coverage and ex post risk. However, Cawley and Philipson (1999) reported a neutral or even negative correlation between mortality risk and insurance coverage in the life insurance market. A recent growing literature has shown that such puzzle could be attributed to the multiple dimensions of private information coexisting in the market. Specifically, I provide evidence of the existence of private information both on mortality risk and on life insurance preferences. I show that these two dimensions of private information have an offsetting effect on the relationship between subsequent mortality and life insurance purchases, which makes the identification of the private information on mortality risk difficult under the traditional setting. Instead, I apply the mixture density model and successfully detect a positive correlation between individual mortality and insurance coverage.

Moreover, I examine the mortality risk related to each of the two main types of life insurance contracts ? term and whole life insurance. Our two-period model shows that, given an individual, the relative income, rather than the risk, dominates the choice between whole and term life insurance policies, indicating that a systematic risk difference between these two pools should not be observed. Moreover, when the income of these two periods are the same, whole life insurance policies, the one with more capability of avoiding reclassification risk, would be always favored if the individual is risk averse. Empirical results support the conclusions made in the theoretical model. This paper also, empirically confirms the partial lock-in of consumers embodied in the more front-loading contract as proposed by Hendel and Lizzeri (2003). Specifically, I find as a more front-loaded contract, whole life insurance policy is associated with a lower lapsation rate and thus retains a healthier pool after 65 years old.