This paper is concerned with the problem of estimating the demand for health care with panel data. A random effects model is specifed in a semiparametric Bayesian fashion using a Dirichlet process prior. This results in a very exible mixture distribution with an in nite number of components for...