Mundlak-Chamberlain Correlated Random Effects
Also known as: Correlated Random Effects, CRE Estimator, Mundlak Device, Korelasyonlu Rassal Etkiler
The Mundlak-Chamberlain correlated random effects (CRE) estimator, introduced by Mundlak (1978) and extended by Chamberlain (1982), is a panel data technique that reconciles the fixed effects and random effects approaches by explicitly modelling the correlation between unobserved individual heterogeneity and the observed regressors. By including within-group means of time-varying covariates as additional regressors in a random effects framework, CRE yields estimates numerically equivalent to the within (fixed effects) estimator while permitting identification of time-invariant variables.
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 Mundlak-Chamberlain estimator when you have longitudinal or panel data and suspect that unobserved individual heterogeneity correlates with at least some regressors, yet you also need to estimate the effect of time-invariant variables (e.g., gender, country, sector) that would be swept away by fixed effects. The method is valid under a linear projection of the individual effect onto group means. It is particularly valuable when the panel is short (small T, large N), when software for true fixed effects with time-invariant regressors is unavailable, or when you want a formal endogeneity test embedded in the estimation. It is not suited to nonlinear models without modification, and it assumes the linear CRE projection is correctly specified.
Strengths & limitations
- Identifies coefficients on time-invariant regressors that fixed effects cannot estimate
- Provides a built-in Hausman-type endogeneity test via the joint significance of group-mean coefficients
- Numerically equivalent to the within estimator for time-varying regressors under standard assumptions
- Easy to implement in any GLS-capable software without dedicated panel routines
- Requires correct specification of the linear projection of the individual effect onto group means
- Does not handle time-varying unobserved heterogeneity or interactive fixed effects
- Extension to nonlinear models (logit, probit, count) requires additional assumptions and is not always tractable
- Efficiency gains over fixed effects are modest when group means are highly collinear with the original regressors
Frequently asked
Is the Mundlak estimator exactly identical to fixed effects?
For time-varying regressors, the CRE slope estimates are numerically identical to within (fixed effects) estimates under balanced panels. The advantage of CRE is that it additionally identifies coefficients on time-invariant regressors and provides a formal test of the random effects assumption, neither of which the standard within estimator can do.
How does the CRE approach differ from simply running a Hausman test before choosing an estimator?
The Hausman test guides model selection but does not estimate time-invariant effects. The Mundlak device integrates the test into the estimation itself: the gamma coefficients simultaneously control for endogeneity and serve as the test statistic, yielding a single coherent model rather than a pre-test followed by a separate estimator.
Can the Mundlak-Chamberlain approach be applied to probit or logit panel models?
Chamberlain (1984) extended the CRE idea to binary outcome models by projecting the latent individual effect onto covariate means and integrating out the remaining error. Wooldridge (2010) provides a practical implementation for probit; however, the approach requires distributional assumptions on the individual effect and is more demanding than the linear case.
Sources
- Mundlak, Y. (1978). On the pooling of time series and cross section data. Econometrica, 46(1), 69–85. DOI: 10.2307/1913646 ↗
How to cite this page
ScholarGate. (2026, June 2). Mundlak-Chamberlain Correlated Random Effects. ScholarGate. https://scholargate.app/en/econometrics/mundlak-chamberlain
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.
- Hausman TestEconometrics↔ compare
- Panel Fixed EffectsEconometrics↔ compare