Monte carlo evidences on finite sample performances of the simulated integrated conditional moment estimator for the binary choice model

In this paper, I propose a simulated integrated conditional moment (SICM) estimator for the binary choice model. The asymptotic property of the proposed SICM estimator is explored via Monte Carlo experiment since its asymptotic theory has not been fully developed. In particular, the SICM estimator is compared with method of simulated moment (MSM) and ML estimators by adopting a simple parametric distributional setup in the experiment. The experiment results show that the proposed SICM estimator is valid in the sense that it is consistent and its Monte Carlo variance decreases by 1/n times as the sample size increases. In particular, it is found that the variance of the SICM estimator is approximately twice that of the MSM estimator with one simulator.