Hi all,
I am running xtabond2 on the model , and the command is:

xtabond2 lVal l.lVal l(0/5).Init_a l(0/5).Init_b l(0/5).Init_c l(0/5).Init_d log_Gdp i.Year,

gmm(l.lVal l(0/5).Init_a l(0/5).Init_b l(0/5).Init_c l(0/5).Init_d, collapse) iv(log_Gdp i.Year) twostep robust small orthogonal

My data include 39 countries, 17 products and for the period from 2002-2019
i.Year is dummy variable for year (from 2002-2019)
Init_a, Init_b, Init_c, Init_d are 4 dummy variable account for separate factor
lVal is the log of the trade Value. Therefore this is a log-linear model
Here is the result: Hansen= 0.068, p (AR2)=0.2
However, the coefficient of dummy variable Init_a doesn't make sense, as it is too different from other paper
I.e. I got coefficient value of very large negative value (for example -70), when interpreting the result, switching from 0 to 1 in Init_a will decrease the trade value by 100* (e^(-70)-1)=100%
Could you please give me any advice to fix this problem?

I am thinking about adding the country factor dummy variable but I don't know how to include all 39 countries into the regression
I created the factor dummy variables for the countries already, similar to Year dummy, which is i.Country
But when I include them in the model, I got this warning from Stata "Country: string variables may not be used as factor variables"

Any suggestions would be very helpful!

Thanks a lot