Hello, I am trying to use the Hausman Taylor IV panel regression model for impact evaluation for my master's thesis. I am having trouble trying to understand the following:
suppose, X'it= time varying exogenous variables.
and W'i= time invariant endogenous variables.
What I understand is that we use individual mean of X'i as instruments for W'i.
So for my model I have multiple time varying exogenous and time invariant endogenous variables.
Abd I am having trouble trying to understand which X'i's mean is being used as an instrument for which corresponding W'i? Or is it the case that a regression is formulated using the means of all Xi's and then that regression is used to estimate all the predicted values of W'i? As in, is the same regression used to calculate the predicted value of each individual time-invariant endogenous variable?
suppose, X'it= time varying exogenous variables.
and W'i= time invariant endogenous variables.
What I understand is that we use individual mean of X'i as instruments for W'i.
So for my model I have multiple time varying exogenous and time invariant endogenous variables.
Abd I am having trouble trying to understand which X'i's mean is being used as an instrument for which corresponding W'i? Or is it the case that a regression is formulated using the means of all Xi's and then that regression is used to estimate all the predicted values of W'i? As in, is the same regression used to calculate the predicted value of each individual time-invariant endogenous variable?