System GMM and fixed effects

[alessandro gst]

Dear all,
I am trying to estimate a model for panel data and I was thinking about using System GMM. I have been reading Roodman (2009) and other papers, but I do not fully understand whether System GMM is able to control for unobserved heterogeneity or not. In other words, if I am sure I need to control for unit heterogeneity, would difference GMM be more robust than system GMM?

thank you in advance for your help

Best regards

[Sebastian Kripfganz]

Difference GMM means that you are effectively removing the time-invariant unobserved heterogeneity from the model. System GMM uses additional instruments for the untransformed model. To be valid, those instruments require the additional assumption that they are uncorrelated with the unobserved heterogeneity. In that sense, the difference GMM estimator is more robust by not relying on this additional assumption.

Further information that might be useful:

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