Does Adverse Selection Matter? Evidence From a Natural Experiment

The empirical evidence of adverse selection in insurance markets is mixed. The problem in assessing the extent of adverse selection is that private information, on which agents act, is generally unobservable to the researcher, which makes it di¢ cult to distinguish adverse selection from moral hazard. Unique micro data, from a dental insurance natural experiment, is here used to provide a direct test of selection. The setup of the insurance makes it possible to observe a proxy for private information. Interestingly, results di¤er across the distribution of risk, with direct evidence of both adverse selection and advantageous selection. This dual selection can explain the limited empirical evidence for adverse selection in the literature: the two e¤ects may balance out on the aggregate level. The paper also presents a model of insurance choice that can harbor both adverse and advantageous selection, explaining the pattern in the data by di¤erences in the e¤ectiveness of prevention across high and low risks. Keywords: Asymmetric information, Adverse Selection, Advantageous Selection, Health Insurance. JEL codes: D81, D82, G22, I11. The paper has bene…ted from suggestions and comments by Magnus Johannesson, Douglas Lundin, Georges Dionne, Mark Pauly and seminar participants at Stockholm School of Economics, ENTER Jamboree 2003, Venice Summer Institute 2003 and EEA Congress 2003. I am grateful to the National Dental Service in Värmland, Sweden, especially to Jörgen Paulander, for providing data and odontological expertise, and to Ann-Britt Emilsson for generously answering questions on registration practices. Financial support from Apoteksbolagets Fond för Forskning och Studier i Hälsoekonomi och Socialfarmaci is acknowledged. Centre for Health Economics, Stockholm School of Economics, P.O. Box 6501, SE-113 83 Stockholm, Sweden, e-mail: [email protected]