Question

Econometrics Question

Discuss the following statement: “If X is not correlated with u; the best choice of instrument is using the regressor itself”.

Answer 1

Ans: As per the explanation significant correlation is not a pre-requisite for running a regression. we can find some of the cases where the relation looks very stronger but the regression result is poor and vice-versa. Correlation is a causal relation where regression will gives idea about how an independent variable is numerically related to the dependent variable. We always not looking for a causal relation sometimes it is more than that which is described by the regression. When we try to build an econometric model or try to predict a key response, better we use regression. for a quick relationship in any direction (+ve or -ve) and strength of a relationship, correlation is your best bet.