PT - JOURNAL ARTICLE AU - Jonathan B. Berk AU - Jules H. van Binsbergen AU - Max Miller TI - Mutual Funds: <em>Skill and Performance</em> AID - 10.3905/jpm.2020.1.143 DP - 2020 Mar 31 TA - The Journal of Portfolio Management PG - 17--31 VI - 46 IP - 5 4099 - https://pm-research.com/content/46/5/17.short 4100 - https://pm-research.com/content/46/5/17.full AB - The authors summarize the recent literature on mutual fund manager skill and performance. They discuss the latest contributions in the field and reinterpret them through the lens of the rational expectations framework (efficient market hypothesis). They further discuss the importance of (1) the choice of benchmark model and (2) the time-series and cross-sectional sample selected in performance studies. The article has three main conclusions. First, although net alpha is a measure of the abnormal return of an extra dollar invested in a particular fund (i.e., performance), it does not measure mutual fund manager skill. To measure the latter, the product of gross alpha and the size of the fund—value added—is needed. Second, the set of real-time available index funds is the relevant counterfactual to use when assessing the skill and performance of investment managers. Nontradable factors that are constructed with the benefit of hindsight are not a realistic benchmark. Third, the authors can think of no good reason to exclude high-quality mutual fund data either in the cross section or time series when making inferences regarding skill and performance.TOPICS: Manager selection, mutual fund performance, performance measurementKey Findings• Net alpha is not an appropriate measure of mutual fund manager skill; value added, the product of assets under management and gross alpha, is a measure of mutual fund manager skill.• The set of real-time available index funds is the relevant counterfactual investment opportunity set for assessing skill and performance of investment managers.• Studies of skill and performance should not exclude high-quality data on mutual funds from either the cross section or the time series.