Econometric Society European Meeting edition:59 location:Madrid date:20-24 August
This paper presents an estimator for the linear correlated random coefficient (CRC) model, which is an extension of Garens (1984) Selectivity Bias Method. This new estimator is robust in the sense that the functional forms of the dependencies between primary and secondary error terms and the correlated coefficient need not be known for these estimator to be consistent. Its asymptotic distribution is derived. In addition, conditions for consistency are given in the case of weakly dependent data. The CRC model is also extended to allow for a qualitative or censored endogenous variable and a censored response.