Abstract
Would you go to the dentist more often if it were free? Many people feel discomfort, maybe even anxiety, when going to the dentist. Only a limited extent of moral hazard would therefore be expected in a private dental insurance, especially if the access to services is restricted by a managed care feature. The challenge of assessing the bite of moral hazard is to separate it from adverse selection, since agents act on private and generally unobservable information. Micro data is here used to analyze the impact of a private dental insurance natural experiment launched in 1999 on the utilization of dental care. Data is available both for patients selecting into the insurance and patients choosing not to opt in. Dental consumption can be observed for each patient both before and after the insurance was launched. The set-up of the insurance scheme makes it possible to observe a proxy for private information. With the available data, the treatment effect of the insurance can be separated from self-selection, using propensity score matching, OLS and IV estimators. The effect is assessed using both difference-in-differences and cross-sectional estimations. The estimated moral hazard effect on total dental costs is large; 51-84 percent depending on the estimator.