International Economic Review
Using data on the schooling decisions of a sample of white males from the National Longitudinal Survey of Youth of 1979, the authors of this paper compare the performance of simulated method of moments (SMM) and maximum likelihood (ML) estimation in dynamic discrete choice models.
They report that the ML estimates were close to the true structural objects of interest, but their version of SMM failed to recover the true net returns to education and policy effects. The SMM was unable to distinguish between systematic and unsystematic cost components driving educational choices.
Based on their analysis, the authors recommend that more exacting model specification tests and Monte Carlo evidence be provided to verify the performance of SMM in any application.