|
| |
| Kazumitsu Nawata |
| |
| ''A monte carlo analysis of the type II tobit maximum likelihood estimator when the true model is the type I tobit model'' |
| ( 2007, Vol. 3 No.54 ) |
| |
| |
| Type I (censored regression) and Type II Tobit (sample selection) models are widely used in the various fields of economics. The Type I Tobit model is a special case of the Type II Tobit model. However, the dimension of the error terms decreases and the distribution of the error terms degenerates in the Type I Tobit Model. Therefore, we cannot use the standard asymptotic theorems for the Type II Tobit Maximum Likelihood Estimator (MLE) when the sample is obtained from the Type I Tobit model. Results of Monte Carlo experiments show strange behavior that has never been reported before for the Type II MLE. |
| |
| |
| Keywords: |
JEL: C2 - Single Equation Models; Single Variables: General
|
| |
| Manuscript Received : Nov 03 2007 | | Manuscript Accepted : Nov 03 2007 |
|
