Applying Fama–French Factors to European Equity Returns

Summary

The document raises implementation questions about using the Fama–French three-factor and five-factor models with monthly returns for large European companies. It asks whether a European risk-free benchmark, such as a German government bond yield, would be more suitable than the US Treasury bill rate distributed with the downloaded factor data. It also asks which market portfolio underlies the excess-market-return factor.

The author reports that regressions using the supplied factors yielded statistically insignificant betas for many factors, while a CAPM regression using an MSCI Europe market return produced significant betas. This comparison motivates checking that factor definitions, sample period, currency treatment, and market proxy align with the research question. The document does not provide answers or establish that either benchmark choice is correct; it also leaves unresolved how to handle returns across different quotation currencies.

Key ideas

• Risk-free rates and factor definitions should be matched to the market and currency studied.
• The document contrasts the supplied Fama–French market factor with an MSCI Europe return as a possible market proxy.
• Its reported regression significance differs across those market-return choices, but the comparison does not identify the cause.
• Currency conversion remains an open methodological question when combining European stocks quoted in different currencies.

Tags

# Fama&French models for asset returns applied to European Market


# Fama&French models for asset returns applied to European Market












I am trying to use the F&F 3-factor and 5-factor models for the European Market (monthly data frequency). I downloaded the dataset provided by Fama&French and tried to apply the regressions for the 300 biggest EU companies. However, my questions are:

- they use the yield on US T-bill. Wouldn't it be more appropriate to use the yield of German bund if considering the European Market?

- I don't manage to find which market portfolio they're using to estimate the factor "excess market returns". When using their dataset, the betas I get for most of the factors are statistically not significant at 5% significance level. I tried to make the same exercise with CAPM against the excess market return factor from FF dataset (vs. the excess market return factor considering the market portfolio as being the MSCI Europe index). The result was that the betas in the first case were not statistically significant (5% level), while with the market portfolio being the MSCI Europe index, all the betas were statically significant. Could someone tell me which market portfolio did they consider in computing the factor?

- Since I am using the log-returns for the stocks under analysis, would it be correct just to ignore (for simplicity) the currency in which the different stocks are quoted?

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.