Cinzia Di Novi

We use the 2003/2004 Medical Expenditure Panel Survey in conjunction with the 2002 National Health Interview Survey to test for adverse selection in the private U.S. health insurance markets. The key idea is to test whether individuals who are more exposed to health risks also buy insurance contracts with more coverage or higher expected payments. The critical statistical problem is that the extension of insurance is only measured for those who are insured and face positive health care expenditure, so there is a possible sample selection bias effect. The procedure used is based on a method suggested by Wooldridge (1995). The method also accounts for heterogeneity across individuals. The simultaneous account taken of both possible sources of bias is new for this kind of application. Key words: adverse selection, health insurance, risk profile, two-step estimation, sample-selection bias. JEL Classification: I11, I18, D82.