Fama-MacBeth Regression
Fama-MacBeth Two-Step Regression · Also known as: Two-step cross-sectional regression
The Fama-MacBeth procedure is a two-step regression methodology for analyzing cross-sectional relationships while controlling for time-series structure. Introduced by Fama and MacBeth (1973), it first estimates time-series parameters for each cross-sectional unit, then regresses outcomes on those parameters across the cross-section, averaging results over time. This approach elegantly separates within-unit dynamics from cross-sectional heterogeneity and provides standard errors robust to panel structure.
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 Fama-MacBeth when you have panel data with many units and time periods, and you wish to study how unit-level characteristics (betas, exposures) relate to outcomes in the cross-section. It is standard in asset pricing research and increasingly used in corporate finance and labor economics. The method is most reliable with balanced panels and many units.
Strengths & limitations
- Naturally separates time-series from cross-sectional relationships
- Provides standard errors robust to panel structure and heteroskedasticity
- Transparent two-step procedure easy to present and interpret
- Allows coefficients to vary across time, accommodating instability
- Step 1 estimates are noisy for units with short time series, inflating step 2 standard errors
- Requires many units for cross-sectional regression precision
- Ignores feedback from step 1 to step 2; doesn't account for parameter uncertainty from step 1
- Standard errors can be understated if step 2 residuals are highly correlated across time
Frequently asked
How do I compute correct standard errors in Fama-MacBeth?
The naive approach (ignoring step 1 estimation error) understates standard errors. Use Shanken (1992) adjustment or bootstrap methods. Most modern software implements corrected standard errors automatically; verify in the documentation.
What if I have missing observations or unbalanced panels?
Unbalanced panels require care. Estimate step 1 using available data for each unit, but interpret step-1 estimates cautiously for units with few observations. Consider robustness checks restricting to units with sufficient data.
Can Fama-MacBeth handle time-varying parameters?
Yes. Rolling-window Fama-MacBeth (re-estimating step 1 in rolling windows) captures parameter evolution. Alternatively, use dynamic versions that explicitly model time-variation in risk premia.
How do I choose the length of the time-series window in step 1?
Longer windows reduce noise but may miss structural breaks. A practical guide: for monthly data, use 3–5 years; for annual data, use 10–20 years. Conduct sensitivity analysis around your choice.
Sources
- Fama, E. F., & MacBeth, J. D. (1973). Risk, return, and equilibrium: Empirical tests. Journal of Political Economy, 81(3), 607-636. DOI: 10.1086/260061 ↗
- Shanken, J. (1992). On the estimation of beta-pricing models. Review of Financial Studies, 5(1), 1-33. DOI: 10.1093/rfs/5.1.1 ↗
How to cite this page
ScholarGate. (2026, June 3). Fama-MacBeth Two-Step Regression. ScholarGate. https://scholargate.app/en/econometrics/fama-macbeth-regression
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.
- Local ProjectionsEconometrics↔ compare
- Panel VARXEconometrics↔ compare
- TVP-FAVAREconometrics↔ compare