Interaction term in GMM

[Deniz Guvercin]

The code I am running GMM for is the following:

xtabond2 Y L.Y A B C D A2 A*B C2, gmm(L.Y A A2, collapse lag (2 2)) iv(C B C2) small nocons,

Is it ok to exclude the interaction term from both list of endogenous variables and exogenous variables?

Thanks for your help!

[Phil Bromiley]

You'll increase your chances of a useful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex.

I don't know explicitly about GMM, but for the most part, you must include interactions that include endogenous variables among the endogenous variables. If one of the components is endogenous, the combination is also.

[Sebastian Kripfganz]

Duplicate post here: https://www.statalist.org/forums/for...teraction-term

There is no general answer to the question of how to treat the interaction term. You possibly could treat it as an endogenous regressor and choose appropriate lags as instruments. Alternatively, lags of the exogenous regressor B might be valid instruments if they are sufficiently correlated with the interaction term.

In your case, I generally doubt that your instruments are sufficiently correlated with the endogenous variables D and A*B.