Panel Data Models: Fixed & Random Effects, Hausman Test

Panel Data Models: Key Concepts & Estimators

1. General Panel Data Model

The general panel data model is expressed as:

$y_{it} = \alpha_i + X_{it}’\beta + \epsilon_{it}$

  • $\alpha_i$: Represents the individual-specific, time-invariant effect.

  • Goal: Estimate $\beta$ despite unobserved heterogeneity.

2. Fixed Effects (FE) Estimator

Methods:

  • LSDV (Least Squares Dummy Variables): Not feasible for large $N$.

  • Within Estimator:

$y_{it} – \bar{y}_i = (X_{it} – \bar{X}_i)\beta + (\epsilon_{it} – \bar{\epsilon}_i)$

  • First-Difference (FD) Estimator:

$y_{i2} – y_{i1} = (X_{i2} – X_{i1})\beta + (\epsilon_{i2} – \epsilon_{i1})$

Note: Controls for correlation between regressors and $\alpha_i$.
Caution: Time-invariant regressors are dropped.

3. Random Effects (RE) Estimator

The random effects model is given by:

$y_{it} = \alpha + X_{it}’\beta + \nu_i + \epsilon_{it}$

Assumes: $\text{Cov}(X_{it}, \nu_i) = 0$

GLS transformation:

$y_{it} – \theta \bar{y}_i = (X_{it} – \theta \bar{X}_i)\beta + u_{it}$

with

$\theta = 1 – \frac{\sigma_\epsilon}{\sqrt{T\sigma^2_\nu + \sigma^2_\epsilon}}$


Key Panel Data Formulas

Panel Data Model

The fundamental panel data model is:

$y_{it} = \alpha_i + X_{it}’\beta + \epsilon_{it}$


Within Transformation

The within transformation for FE estimation is:

$y_{it} – \bar{y}_i = (X_{it} – \bar{X}_i)\beta + (\epsilon_{it} – \bar{\epsilon}_i)$


First Difference Transformation

The first difference transformation is:

$\Delta y_{it} = \Delta X_{it} \cdot \beta + \Delta \epsilon_{it}$


Random Effects Transformation

The transformed equation for RE estimation is:

$y_{it} – \theta \bar{y}_i = (X_{it} – \theta \bar{X}_i)\beta + u_{it}$

where

$\theta = 1 – \frac{\sigma_\epsilon}{\sqrt{T\sigma_\nu^2 + \sigma_\epsilon^2}}$


Hausman Test Statistic

The Hausman test statistic is used to choose between FE and RE models:

$W = (\hat{\beta}_{FE} – \hat{\beta}_{RE})’ [\text{Var}(\hat{\beta}_{FE}) – \text{Var}(\hat{\beta}_{RE})]^{-1} (\hat{\beta}_{FE} – \hat{\beta}_{RE})$