Do Factor Return Relationships Stay More Stable Than Stock Returns?

Summary

The document raises the claim, attributed to Grinold and Kahn, that relationships between factors and returns may be more stable than relationships between individual stocks and returns. It proposes an intuition: factors capture shared, systematic return variation across assets with similar exposures, whereas individual stock returns also contain idiosyncratic components that may change more readily.

The note is exploratory rather than conclusive. It asks whether empirical research supports the stability claim but provides no studies, data, tests, or findings to answer that question. Readers should treat the proposed explanation as a hypothesis to investigate, not as evidence that factor exposures or factor premia are reliably stable through time.

Key ideas

• The note cites a claim that factor return relationships may be more stable than stock return relationships.
• It suggests factors capture shared systematic variation among assets with similar exposures.
• Individual stock returns also include idiosyncratic variation, which may be less stable.
• The document offers an intuition but does not provide empirical evidence or resolve whether the claim holds.

Tags

# Are factor (return) relationships more stable than stock (return) relationships?


# Are factor (return) relationships more stable than stock (return) relationships?












In Grinold & Kahn (2000) (Expected Returns and the APT, Chapter 7), the authors state that - "One of the non-APT reasons to focus on factors is the knowledge that the factor relationship is more stable than the stock relationship".

I guess the idea behind it is that because these factors aim to describe the cross-sectional variation in returns of the assets/stocks themselves and therefore capture the systematic component of returns for a group that share an exposure to the same factor. Therefore, it is much harder for this systematic component to "change" as compared to idiosyncratic components.

How true is this? Are there papers that perform empirical studies to prove statistical findings on this?

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.