Improving Smart Beta Attribution Analysis
Differentiating between factor exposures versus factor returns
December 2024. Reading Time: 10 Minutes. Author: Nicolas Rabener.
SUMMARY
- Return contribution analysis helps understand the different return drivers and measure alpha
- However, the contributions depend on the betas to factors and factor returns
- Managers can control betas, but not factor returns
INTRODUCTION
If you think value-focused funds have performed poorly over the last decade, then you better not look at small-cap funds.
The size factor in the U.S. stock market has shown a negative excess return since 1981 when Rolf W. Banz published his influential paper on small caps outperforming large caps. The underperformance is not restricted to the U.S. as only three out of 21 stock markets globally have shown positive excess returns for the size factor since 1990 (read The Illusion of the Small-Cap Premium).
Investors often lambast factor-focused mutual funds or smart beta ETFs when underperforming, but ignore that a fund has zero impact on the underlying factor performance and can only dial up or down the exposure to the factor.
In this article, we will explore ways to enhance contribution analysis for factor-focused funds.
FACTOR EXPOSURES
In this article, we use Alpha Architect`s U.S. Quantitative Value ETF (QVAL) as a case study. The ETF was launched in 2014, is actively managed, and holds a concentrated portfolio of 50 equal-weighted U.S. stocks sorted on quality and valuation metrics. The fund aims to provide high exposure to the value factor, albeit with a quality overlay to avoid value traps.
A returns-based factor exposure analysis highlights consistently positive exposure to the value factor as expected. One of the issues with measuring factor exposures is that it depends on the factor definitions, e.g. Alpha Architect uses EV/EBIT for selecting cheap stocks while we use a combination of price-to-earnings and price-to-book. Although the value factor shows the same large trends in performance regardless of the valuation metric, the factor indices are different, and this can occasionally lead to low or even negative betas (read