I am trying to use a System Gmm for the first time and am really unsure how to use it properly. I have a categorial dependent variable y and a categorial independent variable x of a panel model. I try to account for reversed causality and want to model an interaction effect of x and y(t-1). For that, I created the code:
pgmm(y ~x + x:lag(y) | lag(y, 2:9), data=data,
effect="twoways", model="onestep", transformation="ld")
is this the correct coding of my model? Is it even possible to model categorial variables with a system Gmm?
best regards
