Hi guys,
I am running a diff-in-diff fixed effects model with 16 countries between 2002 to 2017 testing whether Spanish labour reform (treatment) had an impact on FDI post 2012 (treatment point)

In my first regression I had to bootstrap my standard errors as follows:

- bootstrap, reps(500) bca seed(1): xtreg netfdiinflow reformspain reform_others tradegdp inflation humancapital xrate investmentratio realgdppercapita corptax c.time_trend#i.country_n i.year, fe

I now want to undertake a system GMM as my refromspain independent variable is endogenous and so i want to take only the 1st and 2nd lag of the dependent variable as iv estimation, this is the regression i undertook:
- xtabond2 netfdiinflow L.netfdiinflow L2.netfdiinflow reformspain reform_others tradegdp inflation humancapital xrate investmentratio realgdppercapita corptax c.time_trend#i.country_n, gmm(L.netfdiinflow L2.netfdiinflow) iv(reformspain reform_others tradegdp inflation humancapital xrate investmentratio realgdppercapita corptax) twostep

My questions are:
1. Is taking the lag of the dependent variable enough given that its my reformspain dummy (the variable of interest) which is endogenous?
2. given i had to bootstrap my first eeuqation as my standard errors were clustered, do I have to bootstrap the system GMM as well or does the GMM deal with clustered standard errors itself? If I do have to bootstrap how would I immplement that into stata in terms of commands?
3. Could someone help me out with translating the system GMM regression I have done into a model as I don't understand the asscociated algebra too well.

Thank you!