GMM or Fixed Effect Estimation

[Sadam Khan]

Hi. I need help regarding employing FE or GMM in my regression.
although I estimated the model through fixed effect but my mentor asked me to introduce the lag of dependent model as a regressor which changed the model nature from static to dynamic panel model. Now I am using two step difference GMM but the issue is that the Hansen test probability is always close to zero and I am directed that: for validity of the instruments it should be close to 1 or above 1. I estimated the model using different hit and tries by changing the lags, IVs and other options but it didn't change from zero. Now I am getting confused. should I go back for FE or continue with GMM. Note that al the regressors and the dependent variables are in difference form (returns).
The FE regression command is as given

xtreg Rs Rf Rm rExr zmv rBrent vBrentttt, fe vce(cluster compid)

Rs= stock returns, Rf= Risk free rate, Rm= Market return, rExr= Exchange rate returns, zmv= Size of company, rBrent= brent price returns, vBrenttt= Brent price volatility
GMM command:

xtabond2 Rs l.Rs Rf Rm rExr zmv rBrent volBrent, twostep small iv(Rf Rm rBrent volBrent rExr d*) gmm(Rs Rm zmv,lag(2 6) collapse)nodiffsargan robust orthogonal

Moreover if I introduce time dummies

d*

with iv( ) style and on dependent side like:

xtabond2 Rs l.Rs Rf Rm rExr zmv rBrent volBrent d*, twostep small iv(Rf Rm rBrent volBrent rExr d*) gmm(Rs Rm zmv,lag(2 6) collapse)nodiffsargan robust orthogonal

it says collinearity issue and omit many of the independent regressors
I am unable to post the data sample as stata says

 dataex
input statement exceeds linesize limit. Try specifying fewer variables
r(1000);