Panel Arellano-Bond GMM Estimator
Panel Data Arellano-Bond Generalized Method of Moments Estimator · Also known as: Arellano-Bond GMM, AB-GMM, difference GMM estimator, dynamic panel GMM
The Arellano-Bond GMM estimator addresses the two core problems of dynamic panel models — individual fixed effects correlated with the regressors, and the endogeneity introduced by a lagged dependent variable — by first-differencing to remove fixed effects and then using lagged levels of the dependent variable as internal instruments.
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.
When to use it
Use the Arellano-Bond GMM estimator when you have panel data with a moderate number of time periods (T small to moderate, N large), a lagged dependent variable in the model, and fixed effects that are likely correlated with the regressors. It is well-suited to annual firm-level or country-level panels with T between 5 and 15 and N in the hundreds or thousands. Do not use it when T is very large relative to N (use time-series methods instead), when N is small (instrument proliferation causes the Sargan test to lose power), or when there is reason to believe errors are serially correlated at order 2 or higher — the instrument validity assumption is violated in that case.
Strengths & limitations
- Eliminates time-constant unobserved heterogeneity without requiring external instruments.
- Handles the Nickell bias from including a lagged dependent variable in a fixed-effects model.
- Exploits a growing set of internal instruments as T increases, improving efficiency.
- Two-step variant with Windmeijer-corrected standard errors is asymptotically efficient.
- Overidentification (Hansen) and serial-correlation (AR) tests provide built-in specification checks.
- Assumes errors are serially uncorrelated at order 2; if AR(2) is present, lagged levels are invalid instruments.
- Instrument count grows quadratically with T, causing the Hansen test to lose power in longer panels — instrument proliferation is a well-known pitfall.
- Performs poorly when N is small; the asymptotic approximation requires large N.
- The system GMM variant (Blundell-Bond) is often preferred when the series are near unit-root, because difference GMM loses efficiency in that case.
Frequently asked
What is the difference between Arellano-Bond and Blundell-Bond (system GMM)?
Arellano-Bond (difference GMM) instruments the first-differenced equation with lagged levels. Blundell-Bond (system GMM) adds a second set of moment conditions using lagged differences as instruments for the levels equation. System GMM is more efficient when the dependent variable is persistent (near a unit root), but it requires an additional stationarity assumption on initial conditions.
How many instruments is too many?
A common rule of thumb is to keep the instrument count below the number of groups N. Using the 'collapse' option in software like xtabond2 creates one instrument per variable and lag distance rather than one per period, substantially reducing proliferation.
What does the AR(2) test tell me?
It tests for second-order serial correlation in the first-differenced residuals. AR(1) correlation is expected by construction; AR(2) would imply that the level errors are correlated at lag 1, violating the moment conditions. A p-value above 0.05 for AR(2) supports the instrument validity assumption.
Can I use Arellano-Bond with a balanced panel requirement?
No — the estimator handles unbalanced panels naturally. Missing observations simply reduce the number of valid moment conditions available for that unit, but estimation proceeds consistently for the remaining observations.
Should I use one-step or two-step GMM?
Two-step GMM is asymptotically efficient, but its standard errors are severely downward-biased in finite samples. Always use Windmeijer's (2005) finite-sample correction to the two-step standard errors; most software packages (e.g., xtabond2 with the robust option) apply this automatically.
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 ↗
- 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 3). Panel Data Arellano-Bond Generalized Method of Moments Estimator. ScholarGate. https://scholargate.app/en/econometrics/panel-arellano-bond-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.
- Arellano-Bond GMM estimatorEconometrics↔ compare
- Dynamic Panel Data ModelEconometrics↔ compare
- Panel Fixed Effects ModelEconometrics↔ compare
- Panel Random Effects ModelEconometrics↔ compare
- Panel System GMMEconometrics↔ compare