Large coefficient in GMM (xtabond2)

[Allen Smith]

Hi all,

I am using the following code for GMM:

 xtabond2 $Dep_variable l.$Dep_variable $Ind_variable i.year i.Industry i.Country, gmm( $Dep_variable , lag(3 6) collapse eq(l) ) gmm($Ind_variable) iv(i.year i.Industry i.Country , eq(l) ) ortho robust small nocons artest(5) cluster(FirmID)

The coefficient for the key independent variables increase from 2.34 to 33.67. Furthermore, as the number of lags increases, so does the coefficient. If having a large coefficient is considered acceptable in GMM, could someone kindly direct me to a reference that explains why? Thank you very much!