Faster estimation of dynamic discrete choice models using index sufficiency

Many estimators of dynamic discrete choice models with permanent unobserved heterogeneity have desirable statistical properties but may be computationally intensive. In this paper we propose a method to quicken estimation for a broad class of dynamic discrete choice problems by exploiting index sufficiency. Index sufficiency implies a set of equality constraints which restrict the structural parameter of interest to belong in a subspace of the parameter space. We propose an estimator that uses the equality constraints, and show it is asymptotically equivalent to the unconstrained, computationally heavy estimator. Since the computational gains of our proposed estimator are due to the restriction of the parameter space to the subspace satisfying the equality constraints, we provide a series of results on the dimension of this subspace. Finally, we demonstrate the advantages of our approach by estimating a dynamic model of the U.K. fast food market.

Faster estimation of dynamic discrete choice models using index sufficiency Academic Article