Between Estimator for Panel Data
Also known as: Between-Groups Estimator, Cross-Sectional Averages Estimator, Panel Between Estimator, Gruplar-Arası Tahmin Edici
The Between Estimator is a panel data regression technique that identifies regression coefficients exclusively from cross-sectional variation across individuals, by collapsing the panel to individual-specific time-averaged observations and applying ordinary least squares to those group means. It is used in economics, sociology, and political science when researchers are interested in long-run or structural differences between units rather than short-run within-unit dynamics.
Key highlights
- Simple to compute: reduces to OLS on individual means, reproducible in any statistical software
- Directly targets cross-sectional variation, which is often the substantively relevant variation in long-run studies
- Provides variance component estimates (sigma^2_mu, sigma^2_epsilon) required for constructing the random effects GLS estimator
- Remains consistent under the random effects assumption, complementing within (fixed effects) and pooled OLS estimates
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 the Between Estimator when the research question concerns persistent, structural differences across units—for example, whether countries with higher average education levels also have higher average incomes. It is appropriate under the random effects assumption (individual effects uncorrelated with regressors), though it remains a useful diagnostic even when that assumption fails. The estimator is less suitable when within-unit dynamics are the focus, when T is small and time-averaging discards important information, or when unobserved heterogeneity is correlated with regressors—a situation where fixed effects is preferred.
Strengths & limitations
- Simple to compute: reduces to OLS on individual means, reproducible in any statistical software
- Directly targets cross-sectional variation, which is often the substantively relevant variation in long-run studies
- Provides variance component estimates (sigma^2_mu, sigma^2_epsilon) required for constructing the random effects GLS estimator
- Remains consistent under the random effects assumption, complementing within (fixed effects) and pooled OLS estimates
- Discards all within-individual variation, so it is inefficient when both within and between variation are informative
- Inconsistent when individual effects are correlated with regressors (endogeneity), unlike the fixed effects estimator
- Based on only N observations (one per unit), so precision degrades rapidly with small cross-sections
- Cannot identify time-invariant regressors differently from the fixed effects model, but unlike FE it does not absorb them
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
How does the Between Estimator differ from pooled OLS?
Pooled OLS uses every observation and gives equal weight to all T_i time points per individual. The Between Estimator collapses each individual to one observation (the time mean) and runs OLS on those N means. They coincide only in special balanced cases; generally, pooled OLS mixes within and between variation while the Between Estimator isolates between variation exclusively.
When is the Between Estimator consistent?
Consistency requires that the unobserved individual effect mu_i is uncorrelated with the regressors—the random effects assumption. When this holds, both the Between and within estimators are consistent, but GLS random effects is more efficient. If correlation exists, the Between Estimator is inconsistent and the within (fixed effects) estimator should be preferred, as confirmed by a Hausman specification test.
Can the Between Estimator handle time-varying regressors?
Yes, but it uses only the time-mean of those regressors, discarding all within-period fluctuation. Variables that change over time for a given unit are effectively treated as fixed at their individual average. This makes the estimator unsuitable for studying how changes in X affect changes in Y over time, which is the province of the within or first-differences estimator.
Sources
- 1.Baltagi, B. H. (2008). Econometric Analysis of Panel Data (4th ed.). John Wiley & Sons.ISBN 978-0-470-51886-1
You have read it. What now?
Cite this page
ScholarGate. (2026, June 2). Between Estimator. ScholarGate. https://scholargate.app/econometrics/between-estimator