Hi all!
I am just getting into GMM (dynamic panel) for the first time, and I'm having much difficulty. I'm somewhat familiar with first differencing and fixed effect modeling, but this is definitely a step (or two) up.
So, I have a crime rate as outcome variable (DV). There are 60 counties and 7 consecutive years from 2011 to 2017. About half of those 60 counties initiated crime control policy (policy1) at varying years (some in 2014, some in 2015, and others in 2017 (yes, those others initiated the policy in the very last year). So I coded the policy1 as a dummy for a specific county in a particular year where the policy was in full force and effect. For example, a county that initiated the policy in 2015 would have zeros until 2014 and 1 for 2015, 2016, and 2017.
While I'm interested in the effect of this policy on crime rate, I'd like to control for another policy (policy2) and unemployment rate (unemp). These are all time-varying, and policy2 is dummy-coded just like policy1.

Code:
xtabond2 DV L.DV policy1 policy2 unemp year yr11 - yr17, ///
gmm(L.DV policy1 policy2 unemp, collapse) ivstyle(year yr11 - yr17, eq(level)) ///
ivstyle(year yr11 - yr17, eq(diff)) ///
robust two small
Am I on the right track here? Or, does this model explain what I want to investigate?
Any advice would be appreciated.