Modeling Endogeneity Bias Using Observables

GALVAO, ANTONIO; MONTES ROJAS, GABRIEL VICTORIO; SONG, SUYONG

Posadas

Conferencia; XLIX Reunión Anual - Asociación Argentina de Economía Política; 2014

Asociación Argentina de Economía Política

This paper proposes an alternative solution to the problem of endogeneity by explicitly modeling the joint interaction of the endogenous variables and the unobserved causes of the dependent variable as a function of additional observable variables. We derive identification of the parameters and consistency and asymptotic normality of the proposed estimator. These additional variables are not standard instrumental variables because they are allowed to be correlated with the unobserved causes of the dependent variable. Moreover, our methodology diers from the proxy variables approach because it allows for additional unobserved common causes. We apply the proposed estimator to the investment equation model. Our empirical ndings support the idea that the proposed estimator is a useful alternative to instrumental variables economic applications with measurement errors.