State dependence in self-assessed health in Spain

This paper studies for the first time the importance of the contribution of state dependence to the explanation of self-assessed health dynamics in Spain. With this objective in mind, we estimate a series of econometric models including a new proposal for a Heckman selection model with an initial conditions equation run as an ordered probit. Evidence suggests that state dependence and unobserved heterogeneity account for much of the probability of reporting a specific health status while the significance of observed heterogeneity vanishes when controlling for both. However, state dependence looses importance once the error structure of the estimations is improved.