Limited Dependent Variable Model

Classical linear regression model requires that the dependent variable, regressand, should vary between − ∞ and + ∞. But, most of the economic variables are restricted in a sense that they are nonnegative or even much more limited in their values. This chapter deals with econometric models of limited dependent variables capturing economic agent’s response in limited way. In such a model, the response variable is represented as 1 or 0, corresponding to responses of success or failure in a particular situation. A simple econometric model with binary dependent variable and a set of explanatory factors that we expect will influence the respondent’s decision is the linear probability model. But, linear probability model produces predicted probability of success that can take negative as well as the values exceeding unity. This limitation can be overcome by using a binary response model. Two commonly used binary response models are the binomial probit and binomial logit model.