Predicting hedge fund performance when fund returns are skewed

• Author: Andrea J. Heuson,Mark C. Hutchinson,Alok Kumar
• Date: 01 December 2020
• Published date: 01 December 2020
• DOI: http://doi.org/10.1111/fima.12304

DOI: 10.1111/fima.12304

ORIGINAL ARTICLE

Predicting hedge fund performance when fund

returns are skewed

Andrea J. Heuson1Mark C. Hutchinson2Alok Kumar1

1Miami Herbert Business School, University of

Miami, CoralGables, Florida

2Cork University Business School, University

College Cork, Cork, Ireland

Correspondence

AndreaJ. Heuson, Miami Herbert Business

School,University of Miami, 5250 University

Drive,Coral Gables, FL, 33146.

Email:aheuson@miami.edu

Fundinginformation

ScienceFoundation Ireland, Grant/AwardNum-

ber:18/SPP/3459

Abstract

We show that fund-specific return skewnessis associated with man-

agerial skill and future hedge fund performance. Specifically, skew-

ness in fund returns reflects managerial skill in avoiding large draw-

downs. Using a new measure of investment skill that accounts for

this managerial ability,we demonstrate that traditional performance

measures underestimate (overestimate) managerial performance

when returns exhibitpositive (negative) fund-specific skewness. Our

new measure is particularly valuable during periods of economic

crisis, when the annual risk-adjusted outperformance is 5.5%.

KEYWORDS

fund-specific skewness, hedge funds, investment skill, performance

measurement, performance persistence

JEL CLASSIFICATION

G10, G19, G20

1INTRODUCTION

The returns of more than 90% of the hedge funds in the Lipper/TASS database exhibitconsiderable skewness. Skew-

ness in fund returns could reflect managerial attempts to cater to the skewness preferences of fund investors. Alter-

natively,it could arise from manager-specific skills such as dynamic trading or superior risk management to minimize

large losses. Fund return skewnessis more likely to appear during periods of financial crisis because skilled fund man-

agers tend to employ specialized investment and risk management strategieswhen there is greater potential for large

losses.

If fund return skewness is more likely to reflect managerial skill, higher fund skewness should be associated with

higher future performance. In contrast, if high fund skewness reflects managerial desire to cater to known skewness

preferences of hedge fund investors, higher fund skewness is likely to be associated with lower future performance

because idiosyncratic skewness is known to be associated with lower average returns (e.g.,Boyer, Mitton, &Vorkink,

2009; Conrad, Dittmar,& Ghysels, 2013). Although investor preference for skewness is well documented, hedge fund

c

2019 Financial Management Association International

Financial Management. 2020;49:877–896. wileyonlinelibrary.com/journal/fima 877

878 HEUSON ET AL.

studies have not identified a clear relation between historical skewness and future fund returns (Agarwal, Bakshi, &

Huij, 2009; Bali, Brown, & Caglayan, 2012). Given this inconclusive evidence, fund-level skewness is likelyto reflect

both managerial skill and skewness preferences of investors.

In this article, we propose an improved performance measurement framework to more accuratelycharacterize the

relation between fund-specific skewness and future hedge fund returns. Specifically, we develop a fund skewness-

adjusted alpha (ADJ alpha or 𝛼ADJ) measure to predict managerial performance. Our new performance measure is an

improvement over traditional factor model alphas, which do not explicitlyaccount for skewness in observed returns.

Consequently, these factor models cannot accurately evaluate positive fund return skewness strategies that limit

downside risk.

Our ADJ alpha measure is motivated by Leland (1999) and Glode (2011).1Specifically, Leland (1999) posits that

investors should evaluate a fund’s alpha relative to its factor model risk exposure and information on fund-specific

skewness because fund investment strategiesthat generate positive skewness limit downside risk. Comparing a fund

manager who can generate a certain alpha to another who can generate the same alpha with more positive fund-

specific skewness reveals that the latter is likelyto be preferred by a fund investor because the manager may protect

the portfolio from extreme negative returns. In fact, Glode (2011) shows that mutual fund managers generate incre-

mental utility for investors through their skill in reducing losses in bad market states, despite appearing to underper-

form when assessed by unconditional alpha.2

Our new performance measure captures this economic intuition and reflects both a fund’s traditional alpha rela-

tive to its factor model exposure and fund-specific skewness. Specifically,it assigns a higher performance ranking to

fund managers with both high alpha and positive fund-specific skewness. We validate this new performance measure

(ADJ alpha) using both simulated and actual hedge fund returns data, and then show that ADJ alphas are better than

traditional alphas at predicting future hedge fund returns.

In particular, using a large sample of hedge fund returns from the Lipper/TASS database, we decompose total

return skewness into systematic skewness and fund-specific skewness.We find that fund-specific skewness is persis-

tent and positively associated with future hedge fund returns. This evidence is consistent with Leland’s (1999) con-

jecture that positive fund-specific skewness likely reflects managerial skill in avoiding large losses. In addition, we

demonstrate that funds that have both high alpha and greater fund-specific skewness exhibit superior future per-

formance. This result supports the view that traditional performance measures systematically underrate (overrate)

performance when returns have positive (negative)skewness. Consequently, accounting for fund-specific skewness in

returns improves performance accuracy and allows us to better identify skilled hedge fund managers on an ex ante

basis.

Tofurther demonstrate the superiority of our new performance measure, we use the Fung and Hsieh (2004) factor

model specification as the benchmark and compare our ADJ and standard alpha estimates. We find economically sig-

nificant differences. Repeating this analysis on subsamples of returns drawnfrom crisis and noncrisis periods, we show

that the ability of our measure to identify superior managerial performance exante is particularly valuable during peri-

ods of economic crisis when hedge fund returns are more likely to exhibit skewness.During these periods, portfolios

formed using traditional performance measures underestimate fund alpha byup to 5.50%. This evidence suggests that

adjusting for fund-specific skewness is valuable for assessing hedge fund performance during crisis periods.

These findings contribute to the growing hedge fund literature that attempts to identify skilled hedge fund man-

agers’ ex ante and cross-sectional determinants of hedge fund performance. Specifically,Aragon (2007) and Agarwal,

Daniel, and Naik (2009) show that greater incentives, more discretion, and more stringent lockups are associated with

higher future returns. Empirical evidence also suggests that greater exposureto macroeconomic factors and increased

1Thefindings by Goetzmann, Ingersoll, Spiegel, and Welch (2007) provide another motivation for our article. They show that performance measures estimated

using standard statistical techniques inappropriately are at risk of manipulation bymanagers, and they cite hedge funds as a specific example of investments

whosereturns “can deviate substantially from normality” (p. 1505).

2Similarly,more recently, Barroso and Santa-Clara (2015) show that higher skewness at the portfolio level, through hedging or dynamic risk management, is

associatedwith reduced drawdowns and enhanced future performance.