Working Paper: NBER ID: w12461
Authors: Louis K.C. Chan; Stephen G. Dimmock; Josef Lakonishok
Abstract: Academic and practitioner research yields a proliferation of methods using size and value/growth attributes or factors to evaluate portfolio performance. We assess the relative merits of several of the most widely-used procedures, including variants of matched-characteristic benchmark portfolios and time-series return regressions, by applying them to a sample of active money managers and passive indexes. Estimated abnormal returns display large variation across approaches. The benchmarks most widely used in academic research --- attribute-matched portfolios from independent sorts, the conventional three-factor time series model, and cross-sectional regressions of returns on stock characteristics --- have poor ability to track returns. Simple alterations are provided that improve the performance of the methods.
Keywords: money manager performance; benchmarking; portfolio performance; abnormal returns
JEL Codes: G11; G12; G14; G23
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Choice of benchmarking procedure (C52) | Estimated abnormal returns of active money managers (G11) |
| Independent sorts on size and book-to-market ratios (G14) | Tracking error volatilities (C58) |
| Independent sorts on size and book-to-market ratios (G14) | Covariation with managed portfolios (G11) |
| Independent sort method (C69) | Divergence in abnormal return estimates (C22) |
| Regression-based benchmarks incorporating a composite value measure (C51) | Accurate assessments of performance (C52) |
| Choice of benchmarking method (C52) | Materially different conclusions about a manager's skill (D29) |