Endogeneity

An endogeneity problem is very common in econometrics because, compared to experimental sciences, it is not possible to control the data‐generating process. This chapter first summarizes the most important possible causes of endogeneity: simultaneity; covariate measured with error; and omitted variable. Concerning simultaneity, there is an additional problem as the model is not defined by one equation but by a system of equations. In this case, two strategies can be followed: estimating only the equation of interest; and estimating simultaneously all the equations. The latter approach leads to a more efficient estimator, as the correlation of the errors of all the equations is taken into account. In the chapter, the potential correlation between some covariates and the individual effects is treated drastically by using the within transformation, which wipes out the individual effects. Then, the chapter presents the error component instrumental variables estimator, and finally provides more empirical examples for the phenomenon ‘endogeneity’.