I am estimating an economic growth model that has 44 countries and 30 years, and in most of my estimation, I am using 3-year averages. I am getting confused when it comes to using xtabond2 in Stata. Almost all the YouTube tutorials suggest putting all control variables in iv() as exogenous, while some of the sources online, like Stata Forum and even AI, suggest that variables should be included in gmm() as endogenous. I don't know which I should follow. I even read the Roodman 2009 guide, and it seems to be unclear since he uses the arlleno bond example and they treated 2 of the variables as endogenous and the rest as exogenous. The interesting part is that whether I put all variables in iv() or all in gmm(), my main conclusion does not change; that is, my variables' coefficients still have the same sign, and most of them are significant. Of course, AR 1, AR 2 and Hansen tests all pass in both cases, but Hansen seems to hit the sweet spot of 0.25 more often in the iv() case.…