Dear statalist users,
I am using xtabond2 in Stata 13.1 to estimate the effects of police inspections on the crime rate, considered as endogenous. The other regressors are considered predetermined. However, I have a doubt about how to properly specify L(0/1).inspections in the instrument set.
Here is the code:
xtabond2 crime L.crime L(0/1).(inspections) assoc_rate Education lgdp unem den y2009-y2016, ///
gmmstyle(crime inspections, collapse laglimits(2 4) eq(diff)) ///
gmmstyle(assoc_rate Education lgdp unem den, collapse lag(1 3) eq(diff)) ///
ivstyle(LD.(crime inspections), eq(level) mz) ///
ivstyle(D.assoc_rate D.Education D.lgdp D.unem D.den, eq(level) mz) ///
ivstyle(y2009-y2016, eq(level) mz) ///
robust small artests(3)
Is it the correct specification for the endogenous regressors or should I specify it as:
gmmstyle(crime inspections L.inspections, collapse laglimits(2 4) eq(diff)) ///
ivstyle(LD.(crime inspections L.inspections), eq(level) mz) ///
?
I am using xtabond2 in Stata 13.1 to estimate the effects of police inspections on the crime rate, considered as endogenous. The other regressors are considered predetermined. However, I have a doubt about how to properly specify L(0/1).inspections in the instrument set.
Here is the code:
xtabond2 crime L.crime L(0/1).(inspections) assoc_rate Education lgdp unem den y2009-y2016, ///
gmmstyle(crime inspections, collapse laglimits(2 4) eq(diff)) ///
gmmstyle(assoc_rate Education lgdp unem den, collapse lag(1 3) eq(diff)) ///
ivstyle(LD.(crime inspections), eq(level) mz) ///
ivstyle(D.assoc_rate D.Education D.lgdp D.unem D.den, eq(level) mz) ///
ivstyle(y2009-y2016, eq(level) mz) ///
robust small artests(3)
Is it the correct specification for the endogenous regressors or should I specify it as:
gmmstyle(crime inspections L.inspections, collapse laglimits(2 4) eq(diff)) ///
ivstyle(LD.(crime inspections L.inspections), eq(level) mz) ///
?
