Generalized Method of Moments (GMM) Estimation
Generalized Method of Moments Estimation · Also known as: generalized method of moments, GMM, Arellano-Bond estimator, Genelleştirilmiş Momentler Yöntemi (GMM)
The Generalized Method of Moments is a general-purpose econometric estimator that recovers parameters from population moment conditions, introduced by Lars Peter Hansen in 1982. It is widely used for instrumental-variable estimation, dynamic panel-data models (the Arellano-Bond estimator), and time-series applications.
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
GMM is appropriate when you can specify valid moment conditions and have at least about 100 observations, typically with panel or time-series data. It shines for dynamic panels where lagged outcomes appear as regressors (Arellano-Bond), and for instrumental-variable settings where regressors are endogenous. The number of instruments should be at least the number of parameters, and over-identification should be checked with the Hansen J test. For panel GMM a short time dimension (small T) with many units (large N) is preferred.
Strengths & limitations
- Requires only moment conditions, not a full distributional assumption, making it robust and broadly applicable.
- Handles endogenous regressors and dynamic panel models that ordinary least squares cannot estimate consistently.
- The efficient weighting matrix yields asymptotically efficient estimates, and the Hansen J statistic provides a built-in specification test.
- Relies on large-sample (asymptotic) theory; with fewer than 100 observations finite-sample bias can be severe.
- Instrument proliferation — too many instruments relative to the number of units — weakens the Hansen J test and invites overfitting.
- Results hinge on the validity of the moment conditions; invalid or weak instruments bias the estimates.
Frequently asked
How does GMM differ from OLS?
OLS minimises squared residuals and assumes exogenous regressors. GMM instead works from moment conditions — typically that instruments are uncorrelated with the error — so it can consistently estimate models with endogenous regressors or lagged dependent variables that OLS cannot handle.
What is the Hansen J test?
When there are more moment conditions (instruments) than parameters, the model is over-identified and the conditions cannot all hold exactly. The Hansen J statistic tests whether the remaining moment slack is small enough to be consistent with valid instruments; a rejection signals invalid moments or mis-specification.
What is instrument proliferation and why is it a problem?
Using too many instruments — more than the number of cross-sectional units — weakens the Hansen J test and lets the model overfit the endogenous regressors. If this happens, prefer a parsimonious instrument set or fall back to two-stage least squares.
How large a sample does GMM need?
GMM relies on asymptotic theory, so at least about 100 observations are recommended. With smaller samples finite-sample bias can be severe, and a simpler estimator such as OLS may be more reliable.
Sources
- Hansen, L. P. (1982). Large Sample Properties of Generalized Method of Moments Estimators. Econometrica, 50(4), 1029-1054. DOI: 10.2307/1912775 ↗
- 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. DOI: 10.2307/2297968 ↗
How to cite this page
ScholarGate. (2026, June 1). Generalized Method of Moments Estimation. ScholarGate. https://scholargate.app/en/econometrics/gmm-estimation
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.
- 2SLS RegressionEconometrics↔ compare
- Instrumental Variables in Health ResearchHealth Economics↔ compare
- OLS RegressionEconometrics↔ compare
- Panel Fixed EffectsEconometrics↔ compare
- Tobit ModelEconometrics↔ compare