system GMM #2
Description
From Prof Arkangel Cordero (via email)
“system gmm” estimator (Arellano & Bond, 1991; Arellano and Bover 1995, Blundell & Bond, 1998; Roodman, 2009) for a “dynamic” (meaning including a lagged dependent variable) panel model. The closest I found in Python is the pydynpd (GitHub - dazhwu/pydynpd: This python package estimates dynamic panel data model using difference GMM and system GMM.). But this package only runs on CPUs.
I took a look at your “gmm” package, and I can envision using it to run what econometricians call a “difference-gmm” model by manually differencing variables and generating the lagged instrumental variables. My questions are the following:
- Would you be able to provide some guidance on how to estimate what econometricians call “system-gmm” model with your gmm package?
2 Relatedly, would you be able to provide some guidance on how to conduct the corresponding test for :
a. over-identification
b. under-identification
(something akin to this: UNDERID: Stata module producing postestimation tests of under- and over-identification after linear IV estimation (repec.org)),
c. Possibly, a test for weak-instruments.
References:
Arellano, M., & Bond, S. (1991). Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations. The Review of Economic Studies, 58(2), 277-297.
Arellano, M., & Bover, O. (1995). Another look at the instrumental variable estimation of error-components models. Journal of Econometrics, 68(1), 29-51.
Blundell, R., & Bond, S. (1998). Initial conditions and moment restrictions in dynamic panel data models. Journal of Econometrics, 87(1), 115-143.
Roodman, D. (2009). How to do xtabond2: An introduction to difference and system GMM in Stata. The Stata Journal, 9(1), 86-136. https://journals.sagepub.com/doi/pdf/10.1177/1536867X0900900106