Accounting for Unobserved Heterogeneity in Discrete-time, Discrete-choice Dynamic Microsimulation Models. An application to Labor Supply and Household Formation in Italy.

This paper analyzes the implications of unobserved heterogeneity in discrete-time, discrete-choice microsimulation models. We compare the predictions coming from simple pooled probit estimates with those obtained using random effect dynamic probit models, in a dynamic microsimulation of household formation and labor supply in Italy. We show that failing to account for unobserved heterogeneity has important quantitative consequences, which are often neglected in empirical microsimulation work.