Running Fama–French Factor Regressions on Portfolio Excess Returns
Summary
The document explains how to set up a time-series factor regression for a portfolio return series. First, subtract the period’s risk-free rate from the portfolio return to form its excess return. Then regress that series on the market excess return, SMB, and HML for the three-factor specification. The same structure can be extended with additional factors for other models, although those extensions are not detailed in the answer.
The regression intercept is portfolio alpha. The response says that if the factor model is appropriate, estimated alphas may be statistically indistinguishable from zero. It also describes alpha as an error term in the cross-sectional relationship between portfolio betas and average returns. This addresses the regression setup, but not the question about reading monthly observations from the source data. The answer gives no EViews-specific steps, factor-model diagnostics, or guidance on statistical inference, so implementation details and model suitability require further work.
Key ideas
- Convert each portfolio return into an excess return by subtracting the matching period’s risk-free rate.
- Regress portfolio excess returns on the market excess return, SMB, and HML for the three-factor model.
- The intercept is alpha, which may be statistically indistinguishable from zero if the model is appropriate.
- The answer does not provide EViews procedures or resolve how to select the monthly return observation.
Tags
Full text
# fama French regression in Eviews
# fama French regression in Eviews
I'm trying to figure out how to perform CAPM, the fama french 3 Factors and 5 Factors and the Carhart 4 factors regressions in Eviews.
I downloaded all the data from French's website. The 3 Factors data, 5 factors data and the monthly return on 25 portfolios sorted on size and Book-to-Market-Value.
Question 1:
The Picture below is a screenshot of the monthly returns for the 25 portfolios sorted on size and Book-to-market value obtained from French's website.
What value should I use for the monthly returns of let's say year 1926?
Question 2:
For the Regression in Eviews, should I input the Fama French 3 Factors ($SMB$, $HML$, $R_{m}$) together with returns in question 1 in this equation:
$$R_{i,t} - R_{f} = \alpha_i + \beta_i (R_{m} - Rf) + \gamma_i SMB + \delta_i HML + \varepsilon_{i,t}$$
## Answer by Matthew Gunn (score 1)
https://quant.stackexchange.com/a/35268
For each return series $i$ you want to form an excess return over the risk free rate.
$$ R^x_{it} = R_{it} - R^f_t$$
Then for each return series, run the regression:
$$ R^x_{it} = \alpha_i + \beta_{i1} \mathit{RMRF}_t + \beta_{i2} \mathit{SMB}_t + \beta_{i3} \mathit{HML}_t + \epsilon_{it}$$
If the factor model is correct, the estimated alphas probably will be statistically indistinguishable from zero.
This answer the interpretation of alpha. The intercept alpha from a time-series regression is an error term in the cross-sectional linear relationship between portfolio betas average returns.
If confused, I'd recommend material by Eugene Fama and/or John Cochrane as they are clear and careful writers.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.