Using Fama–MacBeth Regressions with Static Country Characteristics

Summary

The document raises a practical question about applying the two-step Fama–MacBeth procedure to cross-country momentum returns. The proposed study has monthly momentum premia for each country as the dependent variable and includes both characteristics that change over time and fixed country attributes, such as an individualism index. The researcher is unsure how to estimate the regression when a characteristic has no time variation and whether constructing the momentum return series itself counts as the first step.

The post provides context from an empirical study of momentum across more than forty countries, but it does not include an answer, regression specification, or estimation results. It therefore serves mainly as a statement of a methodological issue: how to handle cross-sectional explanatory variables alongside time-varying returns in a Fama–MacBeth analysis. Readers would need additional methodological guidance to resolve the question or assess the study’s implementation.

Key ideas

• The question concerns Fama–MacBeth estimation of country-level monthly momentum returns.
• The proposed regressors mix time-varying variables with fixed country characteristics.
• Forming momentum portfolio returns does not by itself explain how to perform the two regression steps.
• The document presents the estimation question but supplies no answer or empirical result.

Tags

# Fama-Macbeth Two Step Confusion


# Fama-Macbeth Two Step Confusion












I am attempting to replicate the work from "Individualism and Momentum around the World" (Chui, Titman, Wei, 2010, `https://onlinelibrary.wiley.com/doi/pdf/10.1111/j.1540-6261.2009.01532.x`). In Section VI (Other Determinants of Cross-Country Momentum: Regression Analysis), the author performs a Fama Macbeth regression where the dependent variable is country specific monthly momentum returns and the dependent variables are a combination of static and those that vary over time.

The two step Fama Macbeth procedure is confusing me a bit. I already have the monthly momentum returns per country (i.e. stocks returns have already been sorted in winner and loser portfolios etc. and I have the momentum premia now) but I am unsure how to perform the Fama Macbeth regression now given that some variables do not vary over time (an example is the individualism index). Does the fact that I have the monthly momentum portfolio return count as Step 1 in the FM regression?

This research does cover 40+ countries.

Below is a sample of data for the US: I am just not clear on how to even run regression (4) since there is no time variation

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.