Two-Step System GMM

[Ravi Patel]

Hi All,

I am estimating a dynamic panel data model with the two-step system GMM estimator using the command -xtabond2-, I wanted to understand whether it is acceptable to lag independent variables and use them as instrumental variables when performing this analysis.

Any help/guidance would be much appreciated.

Thanks in advance,
Ravi

[George Ford]

xtbond2 automatically does so for the difference equation.

webuse abdata, clear
xtabond n l(0/1).w l(0/2).(k ys) yr1980-yr1984, lags(2) twostep

[Ravi Patel]

George Ford I see thanks, as I am looking at the level equation will that mean that my code below is valid even with the lags as it passes the Arellano-Bond for serial correlation and the Hansen test?

(CA1 = capital adequacy lagged by one period - this follows for all)

Code:
xtabond2 ROE L.ROE total_assets capital_adequacy asset_quality liquidity_ratio efficiency_ratio GDP_growth inflation year*, iv(size1 CA1 AQ1 ER1 LR1 GDP_growth1 inflation1 year*, equation(level)) gmm(L.ROE) twostep robust noconstant small orthogonal nodiffsargan

[George Ford]

https://journals.sagepub.com/doi/pdf/10.1177/1536867X0900900106

The paper discusses implementation with l.Y. There may be some options you have to include.