Barber, Huang, and Odean (2016) investigate which risk factors mutual fund investors actually attend to when evaluating fund performance. They analyze flows into over 3,900 U.S. actively managed equity mutual funds from 1996 to 2011.
Main finding: CAPM alpha dominates
In a horse race among six competing performance models (market-adjusted returns, CAPM, FF3, Carhart 4-factor, a 7-factor model adding industry, and a 9-factor model adding profitability and investment), CAPM alpha is the best predictor of fund flows. Its partial effect on flows is roughly double that of its nearest competitor (market-adjusted returns).
When fund returns are decomposed into eight components (alpha plus seven factor-related returns from a 7-factor model), the most surprising result is that flows do not respond as strongly to returns related to a fund’s market risk (beta). Investors largely treat returns from size, value, momentum, and industry tilts as alpha rather than as factor exposures.
Sophistication gradient
Using three proxies for investor sophistication, the paper finds that more sophisticated investors use more sophisticated benchmarks:
-
Distribution channel. Investors in direct-sold funds (more sophisticated) respond less to factor-related returns than those in broker-sold funds (less sophisticated). Broker-sold fund investors are more likely to reward managers for returns attributable to style tilts.
-
Investor sentiment. During periods of high mutual fund inflows (high investor sentiment), flows are more responsive to factor-related returns, consistent with less sophisticated investors trading more during these periods.
-
Wealth. Using data from a large discount brokerage (1991-1996), wealthier investors, who tend to be more financially literate, are less responsive to factor-related returns when assessing fund performance.
Implications for factor pricing
The finding that most investors evaluate funds using the CAPM rather than multi-factor models has implications for the factor investing debate:
- It suggests that returns from size, value, and momentum tilts are largely unrecognized by average mutual fund investors, who reward managers for these returns as if they were skill
- This does not resolve whether factors represent risk or mispricing: investors who attend to market beta may do so because they associate it with risk, but the same cannot be inferred about factors they ignore
- The paper contradicts Berk and Van Binsbergen’s (2016) interpretation that the CAPM victory in the flow horse race validates the CAPM as the true asset pricing model. Instead, it reflects investor sophistication constraints.
Mechanism
How investors attend to market beta remains somewhat mysterious. It is not explained by:
- Morningstar style-box chasing (average beta varies little across style boxes)
- Morningstar star ratings (inclusion of ratings dampens but does not change relative importance of return components)
- Direct estimation of factor models (implausible for most retail investors)
Morningstar category assignments do allow investors to partially attend to size and value tilts, explaining weak flow responses to these factors.