Essays on private information and choice

Abstract

This paper provides new evidence on the existence and nature of imperfect information in health insurance markets that has important implications for assessing market-based health care reform proposals. Empirical research has traditionally concentrated on testing for positive correlation between coverage and risk predicted by standard moral hazard and adverse selection models. Recent theoretical advances demonstrate that the lack of such correlation does not signal the absence of informational asymmetries and is consistent with more complex models with unobserved risk-aversion. I extend the empirical literature by testing for selection on more than one type of latent information risk type and risk-aversion type. In order to separate an incentive (moral hazard) effect from effects of selection on multiple types of unobservables, I specify a hybrid endogenous treatment model with explicit modeling of indicators of latent attitudes toward risk-taking behavior. The model is estimated using Markov Chain Monte Carlo (MCMC) methods. I use data from the 2000 Medical Expenditure Panel Survey (MEPS), which contains a set of attitudinal variables necessary to estimate the model. Although the overall selection effect appears to be insignificant, the results indicate that individuals do, in fact, possess private information which increases their propensity to be insured and to utilize health care. This suggests that lack of conditional correlation between insurance coverage and health care utilization results from informational asymmetries of multiple types inducing selection in opposite directions. In addition, I find strong evidence of moral hazard in outpatient