System GMM (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.
Key highlights
- 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.
Intuition
This section is available to Pro members. Upgrade to Pro
How it works
This section is available to Pro members. Upgrade to Pro
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.
Common pitfalls
This section is available to Pro members. Upgrade to Pro
Applications
This section is available to Pro members. Upgrade to Pro
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
- 1.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.
- 2.Blundell, R. & Bond, S. (1998). Initial Conditions and Moment Restrictions in Dynamic Panel Data Models. Journal of Econometrics, 87(1), 115-143.
- 3.Roodman, D. (2009). How to Do xtabond2: An Introduction to Difference and System GMM in Stata. Stata Journal, 9(1), 86-136.
You have read it. What now?
Cite this page
ScholarGate. (2026, June 1). System GMM. ScholarGate. https://scholargate.app/econometrics/system-gmm