Limit predictions in the Tobit model: an alternative derivation approach

This method considers the problem of predicting already censored values of the dependent variable in the Tobit model. Assuming normality, it suggests a prediction formula that is equivalent to the classical correction using the inverse Mills ratio (IMR). The main contribution is providing an alternative derivation linking Ordinary Least Squares (OLS) and Maximum Likelihood Estimation (MLE) approaches, providing additional practical implementation insights. An extensive experimental investigation and robustness checks demonstrate the method's performance.