Overview — Panel Data Econometrics

A bank, fund, or firm panel records repeated observations on distinct entities. Its main advantage is not a larger spreadsheet; it is the ability to separate within-entity changes from persistent entity differences. This topic asks whether those persistent differences can be ignored, absorbed as fixed effects, or treated as random and orthogonal to regressors. It ends at the boundary where a lagged dependent variable makes ordinary fixed-effects estimation inconsistent and dynamic-panel GMM becomes necessary.

By the end of this note you should be able to:

• formulate pooled OLS, fixed-effects, and random-effects panel models;
• identify the correlation assumption that separates FE from RE;
• use the Hausman test as a specification check rather than a model-selection ritual;
• explain why time-invariant regressors disappear under within transformation;
• recognise why dynamic short-$T$, large-$N$ panels call for methods such as Arellano--Bond.

1. Contents
2. The big picture