System GMM (Arellano-Bover / Blundell-Bond)
System Generalized Method of Moments Estimator (Arellano-Bover / Blundell-Bond) · Also known as: Arellano-Bover estimator, Blundell-Bond estimator, dynamic panel GMM, Sistem GMM (Arellano-Bover / Blundell-Bond)
System GMM is a generalized method of moments estimator for dynamic panel models that contain a lagged dependent variable. Introduced by Blundell and Bond (1998), building on Arellano and Bover, it augments the differenced equation of the earlier difference GMM (Arellano-Bond) with the equation in levels to deliver consistent estimates when N is large and T is small.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
+8 more
When to use it
Use System GMM for dynamic panels that include a lagged dependent variable and have a short time dimension with many units (typically N > 50 and T between 3 and 10). It is the preferred choice over difference GMM when the series are persistent and the lagged-level instruments would otherwise be weak. Key requirements are no second-order serial correlation in the residuals (Arellano-Bond AR(2) test), valid instruments (Hansen J / Sargan test), and keeping the instrument count well below N to avoid instrument proliferation.
Strengths & limitations
- Consistent estimation of dynamic panel models with a lagged dependent variable when N is large and T is small.
- Solves the weak-instrument problem of difference GMM for persistent series by adding the level equation to the moment conditions.
- Relies only on internal instruments (lags of the model variables), so no external instruments are required.
- Requires a reasonable number of units; with N < 50 the instruments lose validity and the Hansen J test becomes unreliable.
- Vulnerable to instrument proliferation: when the instrument count exceeds N the moment conditions overfit and the Sargan/Hansen test loses power.
- Specification fails if the AR(2) test is rejected, signalling that the dynamic structure is mis-specified and instrument validity breaks down.
Frequently asked
What is the difference between difference GMM and System GMM?
Difference GMM (Arellano-Bond) estimates only the first-differenced equation using lagged levels as instruments. System GMM (Blundell-Bond) adds a second equation in levels, instrumented by lagged differences. The extra moment conditions make the instruments much stronger for persistent series, where difference GMM alone would be weak.
Why must I check the AR(2) test?
The estimator relies on the idiosyncratic errors having no second-order serial correlation. First-order correlation in differences is expected, but if the Arellano-Bond AR(2) test is rejected (p < 0.05) the dynamic specification is wrong and the lagged instruments become invalid.
What is instrument proliferation and why does it matter?
GMM can generate a very large number of instruments from deeper lags. When the instrument count approaches or exceeds N, the moment conditions overfit the endogenous regressors and the Sargan/Hansen overidentification test loses power, often returning a p-value implausibly close to 1. Limiting the lag range or collapsing the instrument set keeps the count well below N.
Should I use one-step or two-step estimation?
The two-step estimator is asymptotically more efficient but its standard errors are downward biased in finite samples. The recommended practice is two-step estimation combined with Windmeijer's finite-sample correction so that inference is reliable.
Sources
- Arellano, M. & Bond, S. (1991). Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations. Review of Economic Studies, 58(2), 277-297. DOI: 10.2307/2297968 ↗
- Blundell, R. & Bond, S. (1998). Initial Conditions and Moment Restrictions in Dynamic Panel Data Models. Journal of Econometrics, 87(1), 115-143. DOI: 10.1016/S0304-4076(98)00009-8 ↗
- Roodman, D. (2009). How to Do xtabond2: An Introduction to Difference and System GMM in Stata. Stata Journal, 9(1), 86-136. DOI: 10.1177/1536867X0900900106 ↗
How to cite this page
ScholarGate. (2026, June 1). System Generalized Method of Moments Estimator (Arellano-Bover / Blundell-Bond). ScholarGate. https://scholargate.app/en/econometrics/system-gmm
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- OLS RegressionEconometrics↔ compare
- Panel Fixed EffectsEconometrics↔ compare
- Panel VAREconometrics↔ compare
- Random Effects ModelEconometrics↔ compare