The aim of this article is to estimate the type of selection that exists in the voluntary health insurance market in Colombia where the compulsory coverage is implemented through a managed care competition. We build a panel database that combines individuals’ information from the Ministry of Health and a database provided by two private health insurers. We perform the correlation test for health expenditure and coverage. Following Fang et al. (2008), we condition the estimation on health controls that are available to the econometrician but not to insurers. In both cases we obtain a positive correlation, suggesting that adverse selection predominates. In order to rule out some moral hazard effects, we estimate the correlation between previous health service consumption and insurance purchase. The positive correlation obtained is robust to the inclusion of controls for diagnosis, suggesting that despite some risk selection strategies, health insurers are not protected from adverse selection.
