Risk-Taking and ManagerialIncentives: Seasonedversus New Funds of FundsYiNG LI AND JAMSHID MEHRAN

YiNG Llis assistant professorof Hiiance at IndianaUniversity South Bt-nd,South Bend. IN.yiagl@iusb.edu

J A M S H I D M E H R A N

is protessor oí finance JtIndiana University SouthBenii. South Bend. IN.jmehran@iusb.edu

ile the hedge funds industryhas grown at a fast pace duringthe past decade, lack of trans-parency, need for professionally

conducted due diligence, and the inaccessibilityto new investors of closed funds led to a signifi-cant growth in funds of funds (FOFs). Thesefunds invest in multiple individual hedge funds,hold shares in closed hedge funds, and provideprofessional oversight on individual hedgefunds. The percentage of hedge fund assets thatare managed by FOFs increased throughout thepast 18 years: Up from 5% in 1990 to 17% m2000, and 36% in 2006, it stood at 44% of the$1.7 trillion of total hedge fund assets at the endof the second quarter of 2007, according to areport by Hedge Fund J^esearch hic. FOFs arealso becoming a major venue for institutionalinvestors, such as pension funds, endowments,and wealthy individuals to access hedge funds.According to data collected in Pensions & Inpest-ments' first survey of hedge funds of funds, morethan half of the assets invested in FOFs are fromthese institutional investors. The industry is alsopredicting deeper involvement of FOFs in man-aging institutional money.

As new FOFs are started every year, severalquestions require an answer: Is there any dif-ference in risk-adjusted performance betweennewly started and existing FOFs? Do the riskprofiles of FOFs change over time? If so, howis this change related to managerial incentives?This article explores these issues by examining

performance, risk-taking behavior, and theunderlying managerial incentives of new andseasoned hedge fund managers. Throughout thearticle, "seasoned" and "existing" will be usedinterchangeably to refer to funds that are notnewly started ones.

Past studies on FOFs mainly exploretheir performance, risk characteristics, feesstructure, diversification benefits, and so on.Liang 12004] compares FOFs to individualhedge funds and CTAs and recognizes themas different asset classes. Fung, Hsieh, Naik,and Kamadorai |2()0H] examine the perfor-mance and risk profile, as well as the cap-ital formation of FOFs. Amo, Harasty, andHillion [2007] discuss the diversificationbenefits of FOFs. In this article, we comparethe performance and risk profile betweennew and seasoned FOF managers and findcollective evidence that younger managerstend to be more cautious in their risk-takingand some weak evidence that younger onesalso deliver better abnormal performances.

Professional managers are believed tochange in their risk-taking behavior as tlieyage. Chevalier and Ellison [1997, 1999| pointout that the (possible) early termination ofyounger (mutual fund) managers leads to theirimplicit incentives to avoid unsystematic risk.Avery and Chevalier [1999] show that youngermanagers are more likely to "herd" whilesenior managers do not to signal their ability.On the other hand, Ben-David, Graham,

1 0 0 RISK-TAKING AN!) MANAGERIAL INCENTIVES: SEASDNED VERSUS NEW FUNDS OF FUNDS WINTER 2009

and Harvey |2()()6] also present empirical evidence thatfinancial managers are usually overconfident.

Boyson |2005] conducts a study on the risk-takingbehavior for individual hedge funds and finds that moresenior risk-taking managers have a significantly higherprobability of failure than junior managers, whichsupports the hypothesis of less risk-taking senior man-agers. Our study on the risk-taking behavior of FOFsrelated to manager tenure, however, shows similarresults tor mutual funds studies and contrasting resultswith Boyson's findings on individual managers.

In this article, standard deviation is used as themeasure tor total risk-taking and overall ß exposure isused as the measure for systematic risk. We also use theHorfind.ihl measure to evaluate the concentration levelof an FOF with respect to different styles of individualhedge funds. Multitactor models are used to evaluateYOV performance. The factors are chosen to cover awide range, in correspondence to the wide range oftliiaruial instruments traded by hedge funds.

Major findings are as follows. First, we find weakevidence for risk-adjusted performance differentials.Second, we find that newly started FOFs take less totalrisk and less systematic risk compared to the existingones. Calculated based on 12-month, 24-month, and36-month rolling windows, the volatility of existingFOFs stays above that of newly started FOF managersand the difference persists until after about two to threeyears of the establishment of a new FOF. The grossleverage measure, approximated by aggregating absolutevalue of risk exposures (j9), is also consistently higherfor existing FOFs than for new FOFs. Third, the Her-flndahl measure for existing FOFs is consistently higherthan th.1t of new FOFs. showing evidence for higherinvestment concentration. Fourth, we attribute the risk-taking differential to "herding" theory as documentedby Chevalier and Ellison [1997. 1999] for mutual funds:newly started FOFs are more cautious in risk-taking and"herd" more than the existing ones. Finally we note thatthe above differences in risk-taking behavior betweennewly started and existing FOFs are found to disappearafter two to three years.

DATA AND METHODOLOGY

We use the electronic copy ofthe 2005 fund offunds directory published by Alternative Investment Cen-ter.' As claimed by the data vendor, they focus exclusively

on proven stn*ainlined procedures for data verification,contirmation, and authentication, and this database is themost complete database specific to global funds of hedgefunds and th(ir managers." An advantage of studyingFOF returns instead of individual hedge funds comesfrom reduced back-fill bias, which can be quite seriousin hedge fund studies.-* Our data include 1,120 FOFs,with information on the start date, net-ot-all-fees per-formance dat.i, assets under management (AUM). andother fund characteristics. We take the beginning of eachyear as the observation point for a rolling window anddefine newly started FOFs to be those with less thanone year's return history. 1 he ones with more than oneyear's return history at observation point are defined tobe existing FOFs. Exhibit I is a graph ofthe number ofnewly started FOFs every year between before 19K7 and2005. Exhibit 2 reports summary statistics on numberof funds and AUM of our FOF data. We have severalobservations: the FOF industry has a boom after 1998.For example, in 201)1 alone, 230 new FOFs were started,compared with fewer than 100 on average before 2000.Besides the increase in number of FOFs, the new moneyattracted to FOFs reached an all-tiine high in 2001. in theamount of $42 million for the year, almost double that for2000. After 2002, the growth in the FOF industry slowsdown quite a bit. The mean and median AUM of newlystarted FOFs are usually lower than the seasoned ones andhave a decreaMng trend over the years.

Exhibit i lists the abnormal performance of newlystarted and existing FOF portfolios based on multifactormodel (1). On the one hand, the alphas are positivefor both newly started and existing managers over thesample years, ranging from 8 to 69 basis points a year.The differen--e in alpha between new and seasonedFOFs is not significant acct)rding to a two-sample r-test"*in most ofthe sample years, and in only three out ofthe seven sample years do newly started FOFs seem tooutperform the existing FOFs. There is also a trend ofdecreasing alp las after 2000, consistent with the findingsof Fung et al. [2008|. On the other hand, the abnormalperformance seems to decrease in magnitude after 2002,suggesting intense competition within the industry aswell as decreasing economic profit due to that. Thismight explain why less new capital flows into the FOFindustrv after 2002.

(1)

WINTER 2009 THE JOURNAL of ALTERNATIVE iNvtsTMKNTs 1 0 1

ÍBEll

E X H I B I T 1

Number of Newly Started FOFs, 1987-2004

Funds Started Each Year250

Thirteen factors are used in the model, coveringa wide range of traded instruments. They include fourequity factors (Fama-French three factors and Car-hart [1997] momentum factor), three hond factors(high-yield, term premium, and convertible factors), fivelook-hack straddle factors as described in Fuug and Hsieh[2001] (PTFSBD. PTFSFX, PTFSCOM, PTFSIR, and

PTFSSTK) and an emerging market factor. Summarystatistics on these factors are reported in Exhibit 4. Wealso check the correlation matrix of these factors and donot find high correlation between them.

Our data do not provide direct information onhow much experience a fund manager has had ni theFOF industry. This might lead to some potential noisesfor our comparison between the managers in newlystarted funds versus existing ones. First, a manager istreated as "new" if she/he is with a newly started FOF,despite his/her previous experience. Second, if a FOFchanges manager, we treat the new manager as someonewith an existing fund. Both noises are likely to lead toweaker results in our proposed comparison. What i.smore. FOFs are usually organized as limited partner-ships and change ot manager (general partner) leads todissolution of the tund. So the latter noise source is lesslikely to come into play/'

We define newly started FOF managers in a cer-tain sample year to be those who have been in businessfor less than 12 months at the observation point. Theexisting managers are those with more than one year's

E X H I B I T 2Summary Statistics

Before198719881989199019911992199319941995199619971998199920002001200220032004

of FOFs

TotalFunds

7111426406080113166218269349452573803101110901120

NewFunds

7431214202033535251801031212302087930

ExistingFunds

0711142640608011316621826934945257380310111090

Mean

NewFunds

N/A442985102072921162201892691193093071791841377332

AUM ($M)

ExistingFunds

N/A790651535524413373423274325312277284290266243222211

Median

NewFunds

N/A1417137137502046761103668455647504227

AUM ($M)

ExistingFunds

N/A550215131103109857268687591706562565553

NewMoney inFOFs($M)

N/A1768294612028985840232072601001713988606924720316212165942320284965767960

102 R[SK-TAKIN(; AND MANAGKKIAI INCENTIVES; SEASONED VERSUS NEW FUNDS OF FUNDS WINTER 2009

E X H I B I T 3

Abnormal Performance Based on the 13-Faclor Model

1995-1997

1996-1998

1997-1999

1998-2000

1999-2001

2000-2002

2001-2003

Mean AiphaNewFOFs

0.32%

0.25%

0.59%

0.35%

0.65%

0.08%

0.09%

ExistingFOFs

0.29%

0.22%

0.39%

0.26%

0.69%

0.08%

0.15%

***indicates significance at 1% level; **itidicares significance at

E X H I B I T 4Summary Statistics of Factors

A. PTFS FactorsPTFSBDPTFSFXPTFSCOMPTFSiRPTFSSTK

B. TraditionalExmi<tSMBHMLUML

Factors

Emerging MarketConvertibleTerm PremiumHigh Yield

Mean

0.796-0.123-0.514-0.242-5.696

0.6310.2920.2010.8230.4100.8180.6530.596

Mean t-statisticsNewFOFs

1.689

1.407

2.747

1.592

3.418

0.313

0.865

Adj. R-squareExisting NewFOFs

1.148

0.943

1.934

1.228

3.554

0.334

1.710

FOFs

0.725

0.863

0.728

0.701

0.716

0.478

0.644

5% level; *indica¡es significance ai 10% level

Stdev

15.99419.23112.48418.85213.287

4.5343.7194.4725.5286.7153.6282.5382.066

Min

-24.215-29.f)75-22.935-24.(i45-30.192

-15.990-11.fiOO-20.'90-25.000-28.914-12.ÍÍ42-8.H75-7.;i72

ExistingFOFs

0.730

0.825

0.764

0.745

0.828

0.544

0.788

Max

66.22490.26764.99698.69846.149

8.16014.62014.92018.38013.77012.6867.7327.492

T-value ofTwo-sampieComparison

2.120**

-0.950

0.760

-0.830

2.487*"

1.981**

-0.334

pi

0.1430.038

-0.1230.0400.200

0.045-0.030

0.111-0.0710.1660.0910.0990.147

WiNTCR 2(11)9 THEJOURNAI OF ALTERNATIVE INVESTMENTS 103

history since their initial date. We choose the observa-tion point to be the beginning of each sample year. So forsample year 1994 the observation point is January 1994,newly started FOFs are those that were started betweenJanuary 1993 and December 1993, and existing fundsare those that were started in or before December 1992.We equally weight all FOFs and group them into newlystarted and existing FOFs portfolio for each sample yearbased on this rule.

We measure the total risk ot FOFs using returnvolatility. We calculate and compare the 12-, 24-, and36-month return volatility for newly started and existingfunds in sample years 1995 to 2001. Interestingly, thereturn volatility for these two portfolios presents a verysimilar pattern over the seven sample years, as shownin Exhibit 5, which plots the return volatility over a36-month window: The newly started FOFs have alow er return volatility than the more seasoned FOFsin the initial stage. This difference continues for about2—3 years, when the return volatility for the two gradu-ally converges.

With multifactor models, we are able to decomposethe return of a fund into return from various risk factors.This is especially true for mutual tunds, as illustrated inSharpe [1992]. We follow a similar route to extract thefactor loadings (ß) for FOFs and sum up their absolutevalues to reilect the fact that even though hedge tundsmay take either direction in a trade. FOFs usually are longin individual hedge funds. Thirteen risk factors are usedto capture the broad trading universe for hedge funds,namely four equity factors, three bond factors, tlve straddle

factors, and oue emerging market tactor. We name theresulting measure "GLM" {gross leverage measure) andtreat it as a proxy for leverage usage in FOF portfolios."Christie [2007] uses another measure for FOF leverage bydet'ming it to be the ratio between the aggregated leveredand unlevered returns. His proxy, similarly, requires aseries of assumptions, as leverage of hedge funds and FOFsis not public information and can only be inferred.

(2)

EMPIRICAL RESULTS

We see from Exhibit 6 that based on the 13 risk fac-tors, the GLM is slightly over 1 tor existing FOFs in theinitial sample years and it decreases over the later years,lu contrast, the new FOF portfolio never had a GLMvalue more than 1, showing that newly started FOFsemploy less leverage than their more experienced coun-terparties. After 1998, both new and seasoned FOFs haveGLMs less than 1, showing that both are more cautiousin terms ot using leverage. The fall ot 1998 is known tobe the most turbulent period to the hedge funds industry,partly due to the failure of LTCM with extreme lever-ages. Our findings may be due to voluntary and forcedreduction in risk exposures: on one hand, managers aremore cautious; on the other hand, big brother's failurein the period may lead to credit rationing for all FOFsand especially newly started FOFs.

E X H I B I T 5The 36-Monlh Rolling Window Return Volatility for Newly Started and Existing FOFs

0.03

0.025

1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67 70

Stdev Existing Stdev Newly Started

104 AND MANAGERIAL INCENTIVES: SEASONEU VERSUS NEW FUNDS OF FUNDS WINTER 2009

E X H I B I T 6

GLM for Newly Started and Existing FOFs (Thirteen Factors)

SampleYear

1995-19971996-19981997-19991998-20001999-20012000-20022001-2003

Obs

84726048362412

Mean

Seasoned

1.0741.0170.7990.7680.5610.6060.486

New0.9210.7690.7050.5630.3560.3620.453

Comparison of Mean

P-vaiue

o.o3e0.001o.O5e

<o.ooc<o.ooc<o.ooc

o.ooe

95% Ct Low

-0.0130.097

-0.0220.1550.1580.1970.008

(Old > New)

95% CI High

0.3190.3990.2100.2540.2120.2910.058

What is more, newly started FOFs consistentlyhave ;t lower GLM throu^ihout the seven sample years,and a two-sample f-test shows that the difference issignificant throughout.

To reconfirm that our estimates of risk exposuresare reasonable, we utilize the reported results In Fung.xud Hsieh [2008] ' and find that their magnitude of GLMmi asures is comparable to ours. This suggests that ourfinding is not sample-dependent.

It the newly started FOFs have incentives to bemore cautious, we expect them to have lower unsys-tematic risk besides having lower total risk. Withevidence that newly started FOFs arc cautious to leverup systematic risk, we continue to examine whethernewly started FOFs are better diversified to minimizeunsystematic risk.

Herfindahl index, calculated as in Equation (3). iswidely used to measure the concentration level of a busi-ness organization. Getmansky [2004] uses it to measureconcentration level within styles of hedge fund industry.

(3)7=1

where I* represent the percentages of investment in oneasset class by a FOF. The Herfindahl measure (here-after H-measure) is calculated based on the 13-factormodel for both newly started and existing FOFs. if aportfolio concentrates on a few risk factors, we shoulddetect a higher H-measure and vice versa. We find thatfor six out of seven sample three-year windows, theportfolio of existing FOFs has a higher H-measure,

showing a higher concentration level for investmentinto asset classes. A two-sample r-iest also shows thatthis difference is statistically significant (p-value <0.001) in the six sample windows., while the excep-tion is not significant (/»-value = 0.791).

We are very curious about this exception: a coin-cidence of ttiis finding is that durmg that estimationperiod, many more new FOFs were started as theindustry underwent a boom. Our suspicion is that thelarge number of newly started FOFs have less new capitalto chase on average and this leads to capital rationmg,so that newl)/ started FOFs may not be able to diversifyas they wish. More funds chasing limited capital and thatit leads to wt akcr alpha durmg the time period is alsodocumented in Fung et al. [2008].

In summary, newly started FOFs tend to be morediversified with less risk exposures and exhibit lowertotal risk in tl e initial years than i he existing ones as evi-denced by lower GLM, H-nieasure, and return volatility.However, the risk-adjusted performance ofthe two is notso different. Ht-nce, we infer that newly started FOFsshould have Ligher systematic risks or they are better atchoosing systematic risks that deliver good returns. Funget al. [2(H)8| divide FOFs into "have alphas*' and "havebetas" and fii d the "have alphas" are better at deliveringabnormal performance than "have betas," which usuallyhave higher lisk exposures instead of delivering alphasconsistently. An interesting study would be to combineFung et al. [.Î(H)8[ and this article to see whether theyounger FOFs tend to belong to "have alpha" groupswhile the more seasoned FOFs tend to belong to "havebeta" groups.

WlNTF.U 2009 THEJOURNAI OF ALTERNATIVE [NVKSTMENTS 105

all I III! 1.1. niBÉHik I H H

E X H I B I T 7Herfindahl Measure for Newly Started and Existing FOFs (13 Factors)

Sample Year

1995-19971996-19981997-19991998-20001999-20012000-20022001-2003

Obs

84726048362412

Mean

Existing

0.1700.1600.1610.1420.1590.1410.195

New

0.1230.0970.1320.0930.0980.0610.207

Comparison of Mean (Old >

P-value

<0.001<0.001<0.001<0.001<0.001<0.001

0.791

95% Ci Low 95

0.0340.0510.0150.0410.0390.073

-0.040

• New)

% Ci Higli

0.0590.0750.0420.0570.0830.0870.017

ROBUSTNESS

For robustness testing, we use a different set ofrisk factors, the pure strategy indices, to explain FOFreturns. Since FOFs invest in individual hedge funds andindividual hedge funds belong to various styles based onmajor financial instruments they use and trading strate-gies they employ, FOFs can be considered as investedin various hedge fund strategies to achieve diversifica-tion. Pure strategy indices are constructed from hedgefund style indices provided by various data vendors.'"We regress the FOFs' return on pure strategy indices toevaluate their exposures to the various strategies, andcalculate the GLM and H-measure to compare betweennewly started and existing FOFs.

We then conduct the same exercises as in theprevious section for newly started and existing FOFs.The results are reported in Exhibit 8. The pure strategyindices do a reasonably good Job explaining FOF returnvariations, with mean-adjusted R-s around 80% ingeneral." The results from using pure strategy indicesas risk factors are consistent with our previous findingsfrom the 13 factors. The GLM and H-measure are againfound to be higher for existing FOFs, showing higherconcentration level on strategy investments and riskexposures. We also notice that the magnitude of GLMmeasure is higher with this new set of risk factors.

CONCLUSION

This article examines the differences in perfor-mance, risk-taking behavior, and underlying mana-gerial incentives between newly started and existingFOFs. We investigate the difference in total risk, as

well as level of diversification and risk exposure forFOFs. We find that consistent with Chevalier andEllison [1997, 1999), newly started FOFs tend to bemore cautious in taking risk and are more diversi-fied with lower risk exposure. Newly started FOFstend to have lower aggregate risk exposure (lowerleverage usage), be more diversified, and have lowertotal return volatility than their more seasoned coun-terparties. This is true despite the fact that being moreestablished in the industry with access to more hedgefunds that might have closed to new investment,existing FOFs are able to be more fully invested withmore diversified portfolios. It is also surprising tofind that even though FOFs can have double layersof leverage, this does not seem to be the case judgingfrom the magnitude of GLM. It raises concern thatmany FOFs are having investors" capital sitting withthem idling and cannot find real talents to invest.Fung et al. [2008] also point out a worrying picture:The magnitude of alpha is decreasing in the recentyears while more capital is flowing into the industrychasing performance. We ask this question: Are themore seasoned FOF managers more "entrenched"?This is a future research topic well worth pursuing.

ENDNOTES

'This database from Alternative hivestment Center liassince been acquired by Barclays and now is a part of Barclay-Hedge Datahase.

-See http://www.harclayhedge.com/products/fund-of-fund-datafeeder.html.

•*Accordingto Fung and Hsieh [20001, back-fill bias inindividua! fund.s can be as high as 2% per year, while that for

1 0 6 RISK-TAKINC; AND MANAGERIAL ¡NCFNTivriS: SEASONED VERSUS NEW FUNDS OF FUNDS WINTER 2009

E X H I B I T 8

Newly Started and Existing FOFs (Pure Strategy Index Factors)

A. GLM

SampleYear

1995-19971996-19981997-19991998-20001999-20012000-20022001-2003

B . H-Measure

SampleYear

1995-19971996-19981997-19991998-20001999-20012000-20022001-2003

Obs

84726048362412

Obs

84726048362412

Mean

Seasoned

1.4871.4461.3841.1521.0281.0351.Û98

Mean

Seasoned

2200210021002190249021002240

New

1.4941.0361.1690.9350.8940.9921.224

New

2100116017601450189020702240

CoTparison of Mean (Old

P-value

-0.165<0.00()1<0.00()1<0.00()1

O.OIH0.1 Íi30.997

95% CI Low

-0.0990.3170.1360.1160.052

-0.040-0.212

Comparison of Mean (Old

P-value

0.3Í.8<0.00(i1

0.0(2

<o.oo(n0.00010.3'30.505

95% CI Low

-180850100520270

-170-270

>New)

95% CI High

0.0830.5020.2960.3180.2160.125

0.04

> New)

95% CI High

0.0241070520970930290260

FOFs is estimated to be about 0.7%. Malkiel and Saha [20051h:ivif similar estimates.

"The r-statistics are not reported to save space.''Factor returns for Fama-French and Carhart ¡1997]

.ire taken from French's website. Factor returns for Fung andHsieh [2001] are taken from Hsieh's website.

''Some FOFs are managed by a fund family and canchange managers as a mutual fund does. We conduct a searchby visiting the FOF website to check whether the managerremains the same. There are some websites that are not acces-sible, and some do not provide inforjnatiou on the manager.However, out of the ones that do. [nore than 95% of managersare the same as provided in our data.

^The four equity factors are three Fama-French factorsand Carhart [1997] momentum factor; the three bond factorsare return on long-term, high yield, and convertible bonds;the five straddle factors are return on constructed straddlesbased on tutures contracts, including currency futures, bondfutures, commodity futures, interest rate futures, and stockindex futures. The emerging market factor is return on MSCI

emerging markets. All returns are in excess of the risk-freerate.

"The adjusted R~ for the regression based on whichGLM is calculated shows reasonable explanatory power forFOFs. ranging from 50-80%.

''They run a seven-factor model to capture FOFs'alphas and risk exposures usmg a combined FOF sample fromC!SDM, HFR and TASS databases.

"Various data vendors provide different index valuesfor similar/same styles since the managers reporting to themare not all the -ame. However, the style indices are highlycorrelated as they describe similar/same strategies. The purestrategy indices are a weighted average of indices for thesame style, and the weight is based on the contribution ofan index to th.: first principal component extracted fromthe style indice* of various sources (data vendors). The datavendors we usr in constructing our pure strategy indicesinclude CiSDM, HFR, and TASS. There are seven purestrategy indices constructed: convertible arbitrage, eventdriven, equity hedge, merger arbitrage, equity marketneutral, global macro, and managed futures. These seven

WtNTFR 2 T H E J O U R N A I I>F ALTERNATIVE INVESTMENTS 1 0 7

111 * . U , I.

pure strategy indices serve as risk factors m Regression (1).More detailed discussion on pure strategy indices can befound in Lhabitant [2004].

"These are pseudo-R's since we impose the positivecoefficients constraint to reflect the fact that FOFs can onlybe considered long certain hedge funds strategies and that anR- from a constrained regression is not the same as what weusually mean by R-. However, high pseudo-R- still reflectsbetter explanatory power of the model.

REFERENCES

Amo. Aune-Valer, Helene Harasty. and Pierre Hillion. "Diver-sification Benefits of Funds of Hedge Funds: Identifying theOptimal Number of Hedge Funds." The Journal of Altematiue¡iivcsimnns. Vol. H). No. 2 (Fall 2007). pp. 10-22.

Avery, Christopher N., and Judith A. Chevalier. "Herdingover the Career." Economics Leiters, 63 (1999), pp. 327-333.

lien-David, Itzhak., John R. Graham, and Campbell R.Harvey. "Managerial Overconfidence and CorporatePolicies." Working Paper, Duke University, April 2006.

Boyson, N.M., "Another Look at Career Concerns: A Studyof Hedge Fund Managers." Working Paper, NortheasternUniversity, November 2(X)5.

Carhart, Mark M. "On Persistence in Mutual Fund Perfor-mance." JcHnííi/c/'F/íJfiíife, Vol. 52, No. 1 (1997), pp. 57-82.

Chevalier, Judith, and Glenn Ellison. "Risk-Taking hyMutual Funds as a Response to Incentives." TheJournai ojPolitical Economy. Vol. 105. No. 6 (1997), pp. 1167-1200.

-. "Career Concerns of Mutual Fund Managers." QuarterlyJournal of Economics, Vol. 114, No. 2 (1999), pp. 389-432.

Christie, Steven. "Downside Leverage and Event Riskin Fund-of-Funds Portfolios." TheJournai of Altcrnútivehweítments. Vol. 10, No. 2 (Fall 2007). pp. 1-8.

Fama, E., and K. French. "ComnK)n l<.isk Factors in theReturns on Stocks and ^ox\á^." Journal of Financial Economics,

33 (1993), pp. 3-56.

Fung, William, and David A. Hsieh. "PerfornKinceCharacteristics of Hedge Funds and C TA Funds: Naturalversus Spurious Biases." JoHnju/ of Quantitative ami FinancialAnalysis, 35 (2000), pp. 291-307.

Fung, William, David A, Hsieh, Narayan Y. Naik, and RarunRamadorai, "Hedge Funds: Performance, Risk and CapitalFormation." /(íí/nid/ of Finance. Vol. (>3, No. 4 (2008), pp.1777-1803.

. "The Risk in Hedge Fund Strategies: Theory andEvidence from Trend-Followers." Review of Financial Studies,14(2001), pp. 313-341.

Getmansky. G. "Life Cycle of Hedge Funds."' Working Paper,Massachusetts Institute of Technology, 2004.

Lhabitant. Francois-Serge. "Intelligent Hedge Fund Investing,Successful Avoiding Pitfalls through Better Risk Evaluation."In Evaluating Hedge Fund lm'estment.<: The Role of Pure StyleIndices. Risk Books, 2004.

Liang, Bing. "Alternative Investments: CTAs, Hedge Funds,and Funds-of-Funds." The Journal of Investment Manajiemcni,Vol. 2, No. 4 (4th Quarter 2004), pp. 76-93.

Malkiel, B.C.. and A. Saha. "Hedge Funds: Risk and Return."Financial Analysis Journal, Vol. 61, No. 6 (2005). pp. HI-88.

Sharpe. W.F. "Asset Allocation: Management Style andPerformance Measurement." The Journal oj Portjolio Maiiage-nu-iit (Winter 1992), pp. 7-19.

To onlcr reprints of this article, please contact Dewey Palmieriat dpaUuieriífüJijournalí.com or 212-224-3675.

1 0 8 RisK-T.\KiN(; AND MANA(;ERIAL INCENTIVES; SEASONEI) VERSUS NEW FUNDS OF FUNDS WINTER 2009