xtreg

xtreg is a Stata command used to estimate linear regression models when the data are observed across both entities (panels) and time. It supports several panel data estimators, including fixed effects, random effects, and between estimators. The data must be declared as panel data with a command such as xtset id time before running xtreg.

In the fixed effects option (xtreg ..., fe), unobserved, time-invariant characteristics are purged from the estimate by transforming the data within each panel. The resulting coefficients capture the effect of changes within a panel over time and do not rely on between-panel variation. A limitation is that time-invariant regressors cannot be identified under fixed effects.

The random effects option (xtreg ..., re) assumes that the unobserved panel-specific effects are uncorrelated with the regressors and employs a feasible generalized least squares approach to use both within- and between-panel variation. This estimator can be more efficient if the exogeneity assumption holds. The between estimator (xtreg ..., be) uses only cross-sectional averages and focuses on differences between panels rather than within-panel movements.

Postestimation, researchers commonly perform the Hausman test to choose between fixed and random effects, and LM-type tests (such as Breusch-Pagan variants) to assess random effects versus pooled models. Standard errors can be adjusted with vce(robust) or cluster-robust options to address heteroskedasticity or within-panel correlation.

Notes: xtreg assumes linear relationships and can be used with balanced or unbalanced panels. In short panels with dynamic specifications, Nickell bias may arise; alternatives like dynamic panel methods (for example, xtabond) are available.