Simulation evaluation of emerging estimation techniques for multinomial probit models

摘要: Abstract A simulation evaluation is presented to compare alternative estimation techniques for a five-alternative multinomial probit (MNP) model with random parameters, including cross-sectional and panel datasets scenarios without correlation among parameters. The different assessed are: (1) maximum approximate composite marginal likelihood (MACML) approach; (2) Geweke-Hajivassiliou-Keane (GHK) simulator Halton sequences, implemented in conjunction the (CML) (3) GHK approach sparse grid nodes weights, (4) Bayesian Markov Chain Monte Carlo (MCMC) approach. In addition, comparison purposes, sequences was traditional, full information as well. results indicate that MACML provided best performance terms of accuracy precision parameter recovery time all data generation settings considered this study. For settings, when combined CML approach, better than albeit not did perform well recovering parameters dimension integration increased, particularly so datasets. MCMC performed correlations but exhibited limitations correlated