Unbiasedness or Statistical Efficiency： Comparison between One-stage Tobit of MLE and Two-step Tobit of OLS

This paper tried to construct statistical and econometric models on the basis of economic theory in order to discuss the issue of statistical efficiency and unbiasedness including the sample selection bias correcting problem. Comparative analytical tool were one stage Tobit of Maximum Likelihood estimation and Heckman's two-step Tobit of Ordinary Least Squares. The results showed that the adequacy of model for the analysis on demand and choice, we believe that there is no big difference in explanatory variables between the first selection model and the second linear probability model. Since the Lambda, the self- selectivity correction factor, in the Type II Tobit is not statistically significant, there is no self-selectivity in the Type II Tobit model, indicating that Type I Tobit model would give us better explanation in the demand for and choice which is less complicated statistical method rather than type II model.