Serial Correlation in Fama–MacBeth Regressions with Momentum

Summary

The document raises an inference question about applying the Fama–MacBeth procedure to panel data on expected REIT returns. The proposed cross-sectional model includes past returns as a momentum determinant, and the author asks whether this calls for correcting serial correlation and heteroskedasticity in the time series of estimated coefficients, for example with a Newey–West adjustment. The setup refers to prior REIT return research but gives no equation details or data analysis.

No answer or resolution is included, so the text does not establish whether adding lagged returns itself creates a need for a particular correction. The useful methodological distinction is that a regressor’s time-series content and the dependence structure of the resulting estimation errors are separate considerations. Researchers applying the procedure need to assess the time-series behavior of the period-by-period coefficient estimates and choose inference accordingly. The document offers no empirical evidence, correction settings, or comparison of standard errors, so it serves as a research question rather than a complete recommendation.

Key ideas

• The author studies cross-sectional determinants of expected REIT returns using Fama–MacBeth regressions.
• Past returns are included as a momentum-related explanatory variable.
• The question is whether inference should account for serial correlation and heteroskedasticity in the time series of estimates.
• The document poses the issue but provides no answer, data analysis, or recommended correction settings.

Tags

# serial correlation, Fama MacBeth (1973) procedure incorporating momentum


# serial correlation, Fama MacBeth (1973) procedure incorporating momentum












I have a question regarding the use of the Fama-MacBeth (1973) procedure on panel data.

I am investigating the cross sectional determinants of expected REIT return following the procedure from: Chui, A. C., Titman, S., & Wei, K. C. (2003a). The cross section of expected REIT returns. Real Estate Economics, 31, 451-479.

The equation I am investigating is as follows:

I am aware that the Fama MacBeth procedure accounts for cross correlation between variables but not for time series/ auto correlation.

However, in my equation I introduce past returns (momentum) as one of the cross sectional determinants:

Now my question is as follows:

By introducing past returns, should I also account for time series correlation for instance by performing the Newey and West (1987) correction for serial correlation and heteroskedastcity?

Thank you in advance for your help and if anything is unclear I will gladly explain.

Kind regards,

Spike Gontscharoff

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.