GMM on static models

[Chiraz KARAMTI]

Hi, I was wondering if is it possible to use the GMM (xtabond2) on a static model. I'm aware that the GMM on dynamic models require GMM-style instruments to treat the problem with the lagged dependant variable in the model. However, suppose that we don't include y(t-1) in the model, do we still can use xtabond2 command? I'm aware also that GMM models can be used to treat variables mesasured withh error (using within transformation rather than differenciation).

[Sebastian Kripfganz]

Commands for estimating dynamic panel data models such as xtabond2 or xtdpdgmm can also be used for static panel data models, yes.

[Chiraz KARAMTI]

thank you. So I think it will be like an 2SLS model with robust standard errors (two-steps GMM).

[Sebastian Kripfganz]

2SLS is not the same as 2-step GMM. In fact, 2SLS is equivalent to 1-step GMM with a particular choice of the weighting matrix.

[Chiraz KARAMTI]

yes you're absolutely wright.

[Annur Wijayakusuma]

Hi Sebastian,

Please correct me if I am wrong about your comment #4. So if I run 2SLS, it will give me a different result to GMM? I run a two-step system GMM.
I am doing research about the effect of R&D on firm performance. I use System GMM based on Roodman (2009) to run my data analysis. What kind of condition that I can choose 2SLS for my analysis method? as I know when heteroskedasticity is presence, 2SLS is not a good option to choose.

Do you have a recommended article or book that I can read about this?. Appreciate your reply.

Regards,

Annur

[Sebastian Kripfganz]

As said above, the 2SLS estimator is equivalent to the 1-step GMM estimator with particular weighting matrix (the default in xtabond2 or xtdpdgmm). For a system GMM estimator with instruments for both the first-differenced and the level model, this 2SLS estimator is always inefficient (unless there are no unobserved unit-specific effects). However, it might be more robust to the "too-many-instruments" or "weak-instruments" problems.

More on GMM estimation of linear (dynamic) panel data models in Stata:

• Kripfganz, S. (2019). Generalized method of moments estimation of linear dynamic panel data models. Proceedings of the 2019 London Stata Conference.
• XTDPDGMM: new Stata command for GMM estimation of linear (dynamic) panel data models

[Sidra Iqbal]

Commands for estimating dynamic panel data models such as xtabond2 or xtdpdgmm can also be used for static panel data models, yes.

Dear Sebastian
is there any reference paper for this?

[Sidra Iqbal]

Dear Sebastian here is my result window. I dont know how to explain this. what type of model is this and what Sargan and Hasnen test is indicating.