Interpreting Fama–French Factor Regressions with Insignificant Coefficients

Summary

The document asks how to interpret Fama–French three-factor regressions on portfolios sorted by book-to-market. The reported pattern is that the low book-to-market portfolios have insignificant intercepts and factor loadings, the higher book-to-market portfolios have significant HML loadings, and a high-minus-low portfolio has no significant coefficients. The question is whether these findings imply poor explanatory power or no return spread attributable to the model.

The included responses explain that coefficient significance concerns evidence against a zero loading or intercept, not by itself whether the model explains returns. They also discuss how portfolio construction and statistical uncertainty affect interpretation. However, the replies focus mainly on filtration and Brownian motion concepts unrelated to the factor-model question, so they do not resolve the empirical interpretation. The document offers no regression tables, sample details, standard errors, or diagnostics; conclusions about explanatory power would require those details and appropriate tests of the portfolio spread.

Key ideas

• An insignificant factor loading means the regression does not establish that the portfolio has nonzero exposure to that factor.
• An insignificant intercept does not alone prove that a factor model fully explains portfolio returns.
• A high-minus-low portfolio requires its own regression and uncertainty assessment to test whether the return spread is explained.
• The included answers mainly address unrelated stochastic-calculus concepts and do not settle the portfolio results.

Tags

# How should I interpret the (insignificant) coefficients of Fama-French 3-factor model?


# How should I interpret the (insignificant) coefficients of Fama-French 3-factor model?












I am writing a mid-term thesis on the Fama-French factor model.

I have built 5 portfolios sorted by the Book-to-Market ratio. The first portfolio is the lowest-BM group and the last portfolio is the highest-BM group. The interception and the coefficients are all insignificant for the first two portfolios (low-BM). The beta_HML is significant for the last three portfolios(high-BM). I also set a group which is the difference between the last and the first portfolio, the coefficients are all insignificant.

How can I interpret such a result?

Does the model fail to explain excess returns of the first two groups?

Does the result mean that the 3-factor model couldn't explain the returns difference between the last group and the first group?

Thank you!

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.