Published online by Cambridge University Press: 06 April 2009
This paper addresses the bias associated withparametric measurement of timing skill based onmonthly timer returns when timers can make dailytiming decisions. Simulations suggest that theclassic Henriksson-Merton parametric measure oftiming skill is weak and biased downward whenapplied to the monthly returns of a daily timer. Thepaper proposes an adjustment that mitigates thisproblem without the need to collect daily timerreturns. Four tests of timing skill, carried out ona sample of 558 mutual funds, show that very fewfunds exhibit statistically significant timingskill. More encompassing, the adjusted-FF3 test(based on the specification that incorporates boththe proposed adjustment and the Fama-Frenchthree-factor model) is the least biased measure oftiming skill among the four—it provides for asharper inference regarding timing skill and helpsmitigate biases associated with the choice ofinvestment style.
All authors, Yale School of Management, 135Prospect Street, Yale University, New Haven, CT06510. We thank Don Chance (the referee) forseveral constructive suggestions. We also thankSteve Shellans and Ed Owens for usefuldiscussions. We are grateful to Ken French, whograciously provided SMB and HML factor returns.Finally, we thank the 1999 WFA conferenceparticipants, especially the discussant MarkGrinblatt, for valuable comments.