Of all the abundance of explanations out around the net, this one is the most intuitive. Simple as it is it tells much more than many more sophisticated and intricate ones, I’ve encountered so far. Thank you and keep it up!
Hi Nathan, thanks for the video. Maybe you can help me with this question: What if the conditions are right - i.e. steel price (Z) has no direct effect on car demand (Y), but a direct effect on car price (X) - but the predicted values of car price (X-hat) from the first step of the 2SLS now have no significant relation to car demand (Y) anymore? Although, the residuals do neither (augmente regression) and the Hausman test gives gives me results which allow me to not reject the null hypothesis of exogeneity. What does that mean?
Z is the instrumental variable which is the part of X that indirectly explains predicted car price (x itself) without having anything to do with "u", thus Z affects Y through X only. So Z is introduced by the author as a more error-free, purely exogenous version of X.
