Reporting Arellano-Bond GMM estimation results

[Abhinaya Kanakasabai]

Hi,

I ran the following to perform the Arellano-Bond GMM estimation as I am concerned about the potential violation of strict exogeneity assumption in my dynamic panel data model previously constructed:

quietly: xtabond2 lnGDPper l(1/4).lnGDPper high_frac_dem low_frac_dem i.year, gmmstyle(lnGDPper, laglimits(2 .)) gmmstyle(high_frac_dem, laglimits(1 .)) gmmstyle(low_frac_dem, laglimits(1 .)) ivstyle(i.year, p) noleveleq robust nodiffsargan
local adjR2arellano=e(r2_a)
ttest low_frac_dem == high_frac_dem
return list
local pvalarellano = r(p)
outreg2 using "`path'/Arellano.xls", replace bdec(3) rdec(3) aster nocons label ctitle("Log of GDP per capita") keep(low_frac_dem high_frac_dem lag1_lnGDPper lag2_lnGDPper lag3_lnGDPper lag4_lnGDPper lnGDP_trans)

The final table I obtained contains only high_frac_dem and low_frac_dem variables. I'm not sure where my command is going wrong in not incorporating the lagged variables in my table. Also, if I add:

addstat("p-value, H0:bL=bH", `pvalarellano', "Adjusted R-sq", `adjR2arellano')

at the end of the above xtabond2 command, I'm getting invalid syntax error r(198); Could someone please clarify these two?

Thank you!

[Abhinaya Kanakasabai]

George Ford could you please help?

[George Ford]

Drop the ivstyle(i.year, p) and see if you get the r(198) error.

If you still do, then drop some of the other options to see if that's the source of the problem, and then figure out why.

Your outreg2 line does not include year, so there are only two variables to report (and the lags).

[Abhinaya Kanakasabai]

George Ford Thank you for your reply! Dropping add stat and ivstyle made the error disappear, but still not producing the estimates of the lagged variables in my table

[George Ford]

the outreg2 table?

look at the e(b) matrix after estimating it and make sure you are using the correct names.

I'm guessing it will look something like this:

keep(low_frac_dem high_frac_dem L1.lnGDPper L2.lnGDPper L3.lnGDPper L4.lnGDPper lnGDP_trans)