Choosing Additional Factors for the Fama–French–Carhart Model
Summary
The document considers whether to augment the Fama–French–Carhart equity regression with Quality Minus Junk and Betting Against Beta. The responses emphasize that factor selection depends on the dataset and on whether the added variables provide distinct explanatory information. Adding regressors may raise in-sample R-squared while weakening interpretability, reducing the reliability of existing coefficients, or duplicating information already represented by other factors.
Suggested checks include examining coefficient significance and conducting robustness analysis rather than relying on fit alone. One response names the Pastor–Stambaugh liquidity factor as a commonly used extension, while another notes potential overlap between BAB and the safety component of QMJ. The exchange offers general modeling advice rather than empirical comparisons or a universal factor count: useful choices depend on research goals, sample, and factor construction, and extra parameters require careful assessment.
Key ideas
- Adding QMJ or BAB depends on the dataset and whether each factor contributes distinct explanatory information.
- A higher R-squared alone does not establish that an expanded regression is better.
- Researchers should inspect coefficient significance and test robustness when adding factors.
- QMJ and BAB may overlap, while liquidity is another possible extension to consider.
Tags
Full text
# Adding more factors to Fama French Carhart 4 factor model
# Adding more factors to Fama French Carhart 4 factor model
Does it make sense to add more factors such as Quality Minus Junk (QMJ) and Betting Against Beta (BAB) in the Fama-French-Carhart model? Also, if anyone can point me to an article it would be appreciated.
$$r=α+R_f+β_m(R_m−R_f)+β_s⋅SMB+β_v⋅HML+β_{umd}⋅UMD$$
Would add QMJ and BAB to the regression.
## Answer by br0323 (score 1)
https://quant.stackexchange.com/a/55211
This depends on your data and whether it would isolate the new factors completely.
Adding more factors is sometimes difficult as it can decrease the strength of your model and muddle up the previously "good" model, such as Carhart.
QMJ is used to check for quality, but since you are adding other factor in addition to it, you could do other robustness checks instead of QMJ.
## Answer by Prabhnoor Duggal (score 1)
https://quant.stackexchange.com/a/55527
You can add the factors and perform the regression but be careful while assessing the effect of adding more parameters to your model, event though the basic model power may seem to increase(R-square) but checking the parameters in depth(p-value, t-stat) is always useful. In general, adding more than 5 factors to your model counters your goal. So be careful on which factors to use.
## Answer by shoonya (score 1)
https://quant.stackexchange.com/a/64347
The most often used additional factor is Pastor-Stambaugh.
The Fama-French model is augmented with a proxy for the Pastor-Stambaugh liquidity factor. r = RF + βmkt (RM - RF) + βS x SMB + βV x HML + βL x LIQ
You could check the replication issue at Critical Finance
https://www.nowpublishers.com/article/Details/CFR-0074 https://www.nowpublishers.com/article/Details/CFR-0075
## Answer by Lsvob (score 0)
https://quant.stackexchange.com/a/71722
I do not believe adding both of these factors would be the best choice given that the BAB factor is already taken into account in the safety subcomponent of the QMJ factor. I would highly recommend looking into Taming the Factor Zoo by Feng, Giglio and Xiu (2020), summarized here by AQR as it aims to help with answering precisely your question.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.