Xtabond2 with country-time interaction

[Simone Rossi]

Hi everybody
I'm trying to solve a problem with GMM estimations.
I have a dataset that comprises 100 banks from 8 countries observed over a period of 7 years.
In my first draft I included in GMM estimation several macroeconomic controls (different for each country-year combination).
Now a referee is asking me to test the same regression using interacted country-time dummies instead of macro controls... While I can do it easily in dynamic estimations just including i.year##i.country in the command line, xtabond2 fails to provide an output, just saying that "interactions are not allowed".
Does this mean that I need to build all the interactions "by hand"? In this case, can you suggest me a convenient way to do this task easily?
Thanks a lot
Simone

[Sebastian Kripfganz]

See help xi, but please notice that the degrees of freedom of the Sargan/Hansen test (and therefore also the corresponding p-values) are incorrectly computed by xtabond2 when there are any omitted coefficients in the regression output (which is likely to happen when dummies are specified).

[Simone Rossi]

Thanks a lot Sebastian...
I have solved the problem!