On Efficiency of Methods of Simulated Moments and Maximum Simulated Likelihood Estimation of Discrete Response Models

This article has considered methods of simulated moments for estimation of discrete response models. We have introduced a modified method of simulated moments of McFadden [1989]. Using the same number of Monte Carlo draws as in McFadden's method of simulated moments, our estimator is asymptotically efficient relative to McFadden's estimator. In addition to the method of simulated moments, we have considered also maximum simulated likelihood estimation methods. The estimators are shown to be consistent and asymptotically normal without excessive number of Monte Carlo draws