Moral hazard and health insurance: an evaluation based on Rubin's causal framework#(in French)

The aim of this paper is to estimate moral hazard effects on health expenditures based on French data. Recent studies have underlined that selection on the market of insurance is partly due to unobservable characteristics and not only to observable ones (such as health condition). Methods using instrumental variables have therefore gained strong interest. At the same time, papers on evaluation have recently focused on conditions that are requested for a correct application of such methods. The estimation method that is proposed in this paper is based on the assumption that having a supplemental insurance scheme is exogenous when purchasing is mandatory, as it is sometimes the case in France. The idea of using this information in the French context is not new and has been used in several studies before. The contribution of this paper doesn't lie in this assumption but rather in the particular attention given to the scope of effects it allows to estimate. For example, we argue that assignment is not ignorable among the entire population, but only among employees. The second contribution of this paper is to set up the analysis in the Rubin Causal Model. The idea of potential outcomes is particularly well adapted to the problem of moral hazard on healthcare demand. This framework sets five assumptions under which identification is possible. We find substantial moral hazard effects even if the results, as always with instrumental methods, are not very precise.