trouble replicating xtabond2 with ivreg2

Hi,

I am trying to run the equivalent regression of my gmm in difference regression:
xtabond2 donationamount2 movingavwindowdonations, gmm (movingavwindowdonations, lag (2 3)) twostep nolevel

using ivreg2, the reason for this is because I want to test the ivreg2 regression for homoskedasticity and it is not possible to do this after xtabond2, i know there will be less observations in the ivreg2 regression but it is still a similar regression.

currently i am using: ivreg2 d.donationamount2 (d.movingavwindowdonations = f2d.donationamount2 f3d.donationamount2), gmm2s,

but this gives me completely different estimates of movingavdonations.

What am I doing wrong?

Thanks for any advice.