Abstract:
|
In Tobit models, the response variable is censored with lower or upper limits. We propose a model checking procedure for the errors-in-variable Tobit regression based on Zheng's nonparametric test statistic. In Errors-in-variable Tobit model, we first adopt least square estimators using nonparametric kernel mean estimators with validation data. With the estimators, the empirical residuals are used to construct a Zheng's nonparametric test statistic. It is shown that the proposed test statistic is asymptotically normally distributed.
|