1: \begin{abstract}

2: This paper develops and implements a practical simulation-based method

3: for estimating dynamic discrete choice models. The method, which can

4: accommodate lagged dependent variables, serially correlated errors,

5: unobserved variables, and many alternatives, builds on the ideas of

6: indirect inference. The main difficulty in implementing indirect inference

7: in discrete choice models is that the objective surface is a step

8: function, rendering gradient-based optimization methods useless. To

9: overcome this obstacle, this paper shows how to smooth the objective

10: surface. The key idea is to use a smoothed function of the latent

11: utilities as the dependent variable in the auxiliary model. As the

12: smoothing parameter goes to zero, this function delivers the discrete

13: choice implied by the latent utilities, thereby guaranteeing consistency.

14: We establish conditions on the smoothing such that our estimator enjoys

15: the same limiting distribution as the indirect inference estimator,

16: while at the same time ensuring that the smoothing facilitates the

17: convergence of gradient-based optimization methods. A set of Monte

18: Carlo experiments shows that the method is fast, robust, and nearly

19: as efficient as maximum likelihood when the auxiliary model is sufficiently

20: rich. 

21: \end{abstract}

22: