Why Factor Models May Need Sector-Specific Factors

Summary

The document asks whether standard equity factor models, such as Fama–French and Carhart, should be adapted for individual sectors. It highlights that a characteristic like leverage may have different implications for financial firms than for other companies, while suggesting that sector-specific factors could still help explain returns within that industry.

The author points to a paper arguing that factor models may not work uniformly across stock types and describes an open-source climate risk model as motivation. In that example, an energy-specific carbon risk factor appears more useful for explaining energy stock returns than a broad market carbon factor, but less useful outside the sector. The document presents this as an observation and research question, not a validated general result; it supplies no detailed methodology, dataset, or evidence with which to assess the comparison.

Key ideas

• Factor meanings and usefulness may differ across sectors.
• Leverage can have different economic interpretations for financial and nonfinancial firms.
• Sector-specific factors may explain returns within an industry better than broad market factors.
• The climate risk example motivates testing energy-specific and broad carbon factors.

Tags

# are there any good papers on sector-specific versions of factor models?


# are there any good papers on sector-specific versions of factor models?












Does anyone know of any good papers that build sector-specific (utilities, financials, energy, etc.) versions of factor models like the Fama French 3-factor or Carhart 4-factor models?

For example, Fama and French (1992) said "We exclude financial firms because the high leverage that is normal for these firms probably does not have the same meaning as for nonfinancial firms, where high leverage more likely indicates distress."

But just because leverage in financial firms doesn't have the same meaning as other industries doesn't mean that it couldn't be used as a factor. Perhaps some combination of factors could be built for the financial sector?

I found a paper called "Designing Factor Models for Different Types of Stock: What's Good for the Goose Ain't Always Good for the Gander" It has some good points but is short.

(A little more background in case it helps: I'm building an open source climate risk model with a carbon risk factor, and I found that an energy specific carbon risk factor is very good for explaining energy stock returns, but weaker in other sectors, than a broad market carbon risk factor.)

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.