Interpreting Factor Betas When Value and Growth Correlations Shift

Summary

The document raises a factor attribution question using a fund’s estimated value and growth coefficients around the Covid market shock. As the correlation between the factors moved from negative to positive during the selloff, the fund’s estimated value exposure became less negative, while its growth coefficient also changed. The question is whether that pattern represents a deliberate rotation by the manager or a statistical effect of factors moving together in a volatile market.

No answer or analytical method is included, so the observations do not establish that the manager changed the portfolio. The key interpretive issue is that factor coefficients are conditional on the model and the relationships among its explanatory factors; when those relationships shift, estimates may move even without a corresponding decision. Distinguishing active repositioning from market-driven estimation changes would require additional evidence, such as portfolio holdings and a stable attribution framework. The document offers an empirical question and illustrative coefficients, not a resolved finding.

Key ideas

• The observed value and growth coefficients changed as their correlation shifted during a market shock.
• Changing correlations can complicate interpretation of estimated factor exposures.
• Coefficient movements alone do not show whether a manager deliberately rotated the portfolio.
• Holdings data and a consistent attribution method would help separate portfolio decisions from estimation effects.
• The document poses the question but provides no resolution or supporting analysis.

Tags

# Interpreting factor coefficients when correlation flips


# Interpreting factor coefficients when correlation flips












I am looking at mainly value and growth factor coefficients of a fund during the recent Covid market “crisis”.

I have found that said fund had a negative coefficient to value at the start of 2020 (let’s say around -0.4) and a positive growth coefficient (around 0.3).

The correlation at this time between the two factors was also negative at around -0.5.

In March 2020 when the markets nosedived the correlations of the two factors suddenly jumped to positive 0.6 as everything started moving together. At this same time the value coefficient of the same fund jumped to around -0.05, and growth fell a bit.

More recently at the end of Jan 2021 the correlation between the two factors is still high at around 0.36 (relatively high for these two factors) but it has dropped a bit and the fund’s value coefficient has basically followed it down slightly.

My question therefore is how to interpret this, it looks as though the fund has only increased exposure (reduced negative exposure) to the value factor because the correlation to other factors has likely increased ie the market is all behaving similarly.

I’m trying to ascertain if there has been a factor rotation from growth to value and whilst I see increased value exposure and reduced growth exposure how are you supposed to tell whether that was as a result of the manager’s decisions or purely the market doing wacky things in a very volatile environment?

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.