Another Look at the Identification of Dynamic Discrete Decision Processes: With an Application to Retirement Behavior

This paper deals with the estimation of the behavioral and welfare effects of counterfactual policy interventions in dynamic structural models where all the primitive functions are nonparametrically specified (i.e., preferences, technology, transition rules, and distribution of unobserved variables). It proves the nonparametric identification of agents' decision rules, before and after the policy intervention, and of the change in agents' welfare. Based on these results we propose a nonparametric procedure to estimate the behavioral and welfare effects of a general class of counterfactual policy interventions. The nonparametric estimator can be used to construct a test of the validity of a particular parametric specification. We apply this method to evaluate hypothetical reforms in the rules of a public pension system using a model of retirement behavior and a sample of workers in Sweden.