Working Paper: NBER ID: w7779
Authors: Lubos Pastor; Robert F. Stambaugh
Abstract: Our framework for evaluating and investing in mutual funds combines observed returns on funds and passive assets with prior beliefs that distinguish pricing-model inaccuracy from managerial skill. A fund's alpha' is defined using passive benchmarks. We show that returns on non-benchmark passive assets help estimate that alpha more precisely for most funds. The resulting estimates generally vary less than standard estimates across alternative benchmark specifications. Optimal portfolios constructed from a large universe of equity funds can include actively managed funds even when managerial skill is precluded. The fund universe offers no close substitutes for the Fama-French and momentum benchmarks.
Keywords: mutual funds; performance evaluation; alpha; benchmark; managerial skill
JEL Codes: G11; G12; C11
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| inclusion of nonbenchmark passive assets (G23) | more precise estimates of alpha (C51) |
| inclusion of nonbenchmark passive assets (G23) | estimates of alpha vary less across different benchmark specifications (C51) |
| nonbenchmark assets (G19) | valuable information for estimating alpha (C51) |
| benchmark choice (D43) | alpha estimation (C51) |
| nonbenchmark assets (G19) | consistency of alpha estimation across different benchmark definitions (C51) |
| estimates typically indicate negative alphas (C13) | lack of managerial skill (M54) |