Do Model and Benchmark Specification Error Affect Inference in Measuring Mutual Fund Performance?

71 Pages Posted: 25 Mar 2005

See all articles by Naveen D. Daniel

Naveen D. Daniel

Drexel University - Department of Finance

Jeffrey L. Coles

University of Utah - Department of Finance; Arizona State University (ASU) - Finance Department

Federico Nardari

University of Melbourne - Department of Finance

Date Written: March 2005

Abstract

This paper examines the implications for mutual fund performance measurement of two likely sources of specification error. We compare three well-known models, those of Jensen (1968), Treynor and Mazuy (1966), and Henriksson and Merton (1981), and two commonly-used timing benchmarks, the S&P 500 index and CRSP value-weighted index. The practical question we address is whether investors and researchers are likely to make invalid inferences about fund manager performance when using the wrong model or benchmark. Although prior studies recognize at the conceptual level the potential impact of model and benchmark misspecification, the existing literature does not explore empirically the magnitude and significance of the inferential errors when analyzing actual mutual fund data. Based on Monte Carlo simulations calibrated to real data, we find that: (1) model misspecification results in severely biased measures of both selectivity and timing ability, especially for extreme (good and bad) performers, but biases in measures of overall performance are economically insignificant; (2) benchmark misspecification results in qualitatively similar inferences, although statistical significance is not as strong; and (3) the power of detecting ability for an individual fund and for distinguishing a good fund from a bad fund is typically quite low and such power is not appreciably altered by model and benchmark misspecification. All of these results are robust to alternative asset pricing specifications (CAPM versus Carhart 4-factor) and the periodicity of the simulation (calibrated to daily versus monthly data). The use of daily fund returns actually amplifies our conclusions about the biases induced by model misspecifications. The biases we identify appear to be difficult to correct by using standard model selection criteria and misspecification tests.

Keywords: Mutual Fund Performance, Timing Models, Simulation, Power

JEL Classification: G11

Suggested Citation

Daniel, Naveen D. and Coles, Jeffrey L. and Nardari, Federico, Do Model and Benchmark Specification Error Affect Inference in Measuring Mutual Fund Performance? (March 2005). Available at SSRN: https://ssrn.com/abstract=687435 or http://dx.doi.org/10.2139/ssrn.687435

Naveen D. Daniel (Contact Author)

Drexel University - Department of Finance ( email )

LeBow College of Business
Philadelphia, PA 19104
United States
215-895-5858 (Phone)
215-895-2955 (Fax)

HOME PAGE: http://lebow.drexel.edu/Faculty/DanielN

Jeffrey L. Coles

University of Utah - Department of Finance ( email )

David Eccles School of Business
Salt Lake City, UT 84112
United States
801-587-9093 (Phone)

Arizona State University (ASU) - Finance Department ( email )

W. P. Carey School of Business
P.O. Box 873906
Tempe, AZ 85287-3906
United States
480-965-4475 (Phone)
480-965-8539 (Fax)

Federico Nardari

University of Melbourne - Department of Finance ( email )

Faculty of Economics and Commerce
Parkville, Victoria 3010 3010
Australia

Register to save articles to
your library

Register

Paper statistics

Downloads
16
Abstract Views
330
PlumX Metrics
!

Under construction: SSRN citations will be offline until July when we will launch a brand new and improved citations service, check here for more details.

For more information