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

This paper presents a method to estimate the effects of a counterfactual policy intervention in the context of dynamic structural models where all the structural functions (i.e., preferences, technology, transition probabilities, and the distribution of unobservable variables) are nonparametrically specified. We show that agents' behavior, before and after the policy intervention, and the change in agents' utility are nonparametrically identified. Based on this result we propose a nonparametric procedure to estimate the behavioral and welfare effects of a general class of counterfactual policy interventions. 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 blue-collar workers in Sweden