Holdings-Based Fund Performance Measures: Estimation and Inference
105 Pages Posted: 16 Jun 2018 Last revised: 19 Dec 2018
Date Written: December 18, 2018
Abstract
This paper introduces a panel regression framework for holdings-based investment performance measures. Fixed effects decompose performance into time series and cross-sectional predictive ability. Time-series predictive ability is the traditional focus, but cross-sectional ability strongly influences classical measures. A large component of the cross-sectional ability is “buy-and-hold drift,” where buy-and-hold weights drift towards higher alpha stocks. Controlling for this, we find that the time-series predictive ability of the typical equity mutual fund is insignificant or even negative before costs, and the cross-sectional predictive ability is typically no better than a buy-and-hold strategy.
Suggested Citation: Suggested Citation