Hi there!

I am studiyng whether financial literacy causes over-indebtedness among households. So, my response variable is dummy (over-indebted or not) and my variable of interes is categorical (5 categories corresponding to the correct answers on financial literacy questions). I have a panel data, so I am able to use xtlogi .., fe to account for ommited variable bias (even in the presense of the list of control variables in the model). But I think that finanncial literacy variable is endogeous and Instrument variable inclusion is necessary but I am not sure that this option possible as an alterantive to xtlogit...fe.

I have seen several options:
xtivreg ..., fe - but it is not sutable for binary oucome
ivprobit... , vce(cluster x) - but no fe
ivprobit ... , twostep - but no fe

Could you please guide me on what are the possible options I can use in order to build an instrument variable model in my case?

Thanks for any feedback in advance!

Best wishes,
Artem Abduramanov