Benchmark Fund Managers Carefully Before Comparing Alpha

Summary

The document discusses how to compare two mutual fund portfolios when their estimated alphas are statistically insignificant, even though one appears to outperform the other. Its answer cautions that an apparent performance difference may reflect different market exposures rather than manager skill, and that insignificant alpha estimates alone do not establish equal performance.

The suggested diagnostic is to check whether the benchmark captures relevant exposures. If important tilts, such as sector or company-size differences, are omitted, their effects may be absorbed into the regression error or misclassified as beta. The illustration contrasts technology and energy exposure and notes that performance differences within a sector can remain hidden when size is not modeled. The advice depends on having a sound alpha specification and is not a formal hypothesis-testing procedure; benchmark choice and omitted exposures limit the conclusion.

Key ideas

• Statistically insignificant alpha estimates do not by themselves prove that two managers perform identically.
• An apparent return difference may arise from different beta exposures or portfolio tilts.
• An incomplete benchmark can leave relevant sector or size effects unmodeled.
• Review benchmark specification before interpreting alpha estimates as evidence of manager skill.

Tags

# Compare fund managers with insignificant alphas?


# Compare fund managers with insignificant alphas?












For my thesis I am evaluating two mutual fund portfolios in order to check for differences in manager performance. My hypothesis is that there will be no differences in performance (in terms of alpha) between the two portfolios. Unfortunately even though one group seems to outperform the other all alphas are statistically insignificant. Do I reject the hypothesis (because one outperforms the other) or accept the hypothesis (because none of the alphas are signficant)? Thanks in advance

## Answer by jeff m (score 1)

https://quant.stackexchange.com/a/9817

Given that you're correctly measuring Alpha, the difference lies in the Beta exposures of the two managers. You may not be capturing certain tilts, which would show up in your error or incorrectly categorized as Beta. Consider the simple case where you have returns grouped into just technology vs. Energy for instance.

$R_p$ = $B_0$ + $B_t$$R_m$ + $B_e$$R_m$+ $e$

If there isn't a distinction for cap size, you could have large cap energy firms significantly outperform small cap energy, but wouldn't know it from the regression. Make sure you're correctly benchmarking returns before concluding insignificant alpha.

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.