DOES INVESTOR SENTIMENT PLAY A ROLE IN HEDGE FUND RETURN?

Transcription

1 Indan Journal of Economcs & Busness, Vol. 10, No. 4, (2011) : DOES INVESTOR SENTIMENT PLAY A ROLE IN HEDGE FUND RETURN? NANDITA DAS * Abstract The current estmate of the hedge fund ndustry s over one trllon dollar and growth seems to contnue wth more than 8,000 funds. We study the exstence of nvestor sentment n the hedge fund ndustry usng MornngStar database. We calculate fund flows as a measure of nvestor sentment and analyze the fundamental characterstcs of hedge funds that are favored by nvestor sentment. We fnd an overwhelmng evdence of nvestor sentment n the hedge fund ndustry. We also fnd that nvestor sentment s not governed by the fundamental characterstcs of the hedge funds. I. INTRODUCTION Hedge funds have enjoyed healthy growth through the years and contnue to ncrease n popularty, especally among hgh net-worth ndvduals. Recently, an ncreasng number of nsttutons have allocated a small porton of ther assets to these alternatve nvestments owng to ther long-term success. The term hedge fund s used to descrbe a wde range of nvestment vehcles that can vary substantally n terms of sze, strategy, and organzatonal structures. The strateges nclude sellng stocks short on a bet they wll fall and usng borrowed money. Many of them hedge aganst declnes n the market, but the technques vary greatly. One commonalty surroundng hedge funds s the lmted amount of nformaton provded to potental nvestors. Typcally nformaton s lmted to perodc (monthly, quarterly, or annual) returns. Even the leadng hedge-fund databases provde ncomplete nformaton drawn from the fund-offerng documents such as contractual provsons (fee structure, mnmum nvestment sze, and wthdrawal provsons), descrptons of nvestments, styles of nvestment, and the perodc return. The current estmate of the ndustry sze s over one trllon dollar. The average hedge fund remans small, wth less than $200 mllon n assets, compared wth one bllon dollar for the average mutual fund. In 2004, the average fund lagged behnd the S& P ndex. Hedge fund performance was uneven n On an average hedge funds actually lost money n 4 out of 11 months. * Assocate Professor of Fnance, Delaware State Unversty, Delaware, DE

2 584 Nandta Das In ths paper we document the exstence of nvestor sentment n the hedge fund ndustry. We calculate fund flows as a measure of nvestor sentment and analyze the fundamental characterstcs of hedge funds that are favored by nvestor sentment. The paper proceeds as follows: Secton I gves a bref hstory of the hedge fund ndustry. Secton II provdes a revew of the lterature. Data, modelng, and results are outlned n Secton III. Secton IV summarzes our fndngs and contrbutons. II. A BRIEF HISTORY OF THE HEDGE FUND INDUSTRY In 1949, A.W. Jones ntroduced the concept of a hedge fund by combnng a leveraged long stock poston wth a portfolo of short stocks n an nvestment fund wth an ncentve fee structure. From ths smple concept, hedge fund nvestment practces and strateges contnue to evolve. Consequentally, many hedge fund characterstcs have changed sgnfcantly, but many of the fundamental features have remaned the same. Moreover, hedge funds are no longer unque to the U.S. markets, but have become a fxture n the global marketplace. In the Unted States, the funds normally offer ther shares n prvate placements and are lmted to 100 or fewer hgh net-worth nvestors n order to make use of regulatory exemptons provded under the Securtes Act of 1933, the Securtes Exchange Act of 1934, and the Investment Company Act of l940. Interest n hedge funds and ther performance has waxed and waned over tme, but recent publcty has lead to hedge funds enjoyng healthy growth. For nstance, the hgh net-worth nvestors created through the bull market of the late 1980s started to nvest n hedge funds as a means of enhancng ther returns. In 1990, there were about 600 hedge funds worldwde wth assets of approxmately $38 bllon. Accordng to ndustry publcatons, at the end of 1998, despte the publczed collapse of Long Term Captal Management (LTCM), there were some 3,300 hedge funds wth assets of approxmately $375 bllon. Addtonal nvestments at the turn of the century have pushed the hedge fund ndustry over the $1 trllon mark. Although hedge funds nvest n a varety of lqud assets smlar to mutual funds, they are qute dfferent. Under current federal law, hedge funds have no lmtatons on management, vrtually no lmts on the composton of the portfolos, and no mandatory dsclosure of nformaton about holdngs or performance. III. LITERATURE REVIEW The systematc study of hedge funds s a recent phenomenon, encouraged prmarly by the avalablty of data. Most of the lterature s less than a decade old, and focuses on performance attrbuton, performance evaluaton, characterstcs, and the mpact on the fnancal markets. When modelng hedge fund performance as a group, researchers typcally model hedge fund performance by treatng all the hedge funds n a database as a sngle group. Examples nclude Schneewes and Spurgn (1998), Ackermann et al. (1999). Researchers have also attempted to extract strateges from observed returns to reclassfy hedge funds based on observed return

3 Does Investor Sentment Play a Role n Hedge Fund Return? 585 characterstcs. Examples nclude Fung and Hseh (1997), Brown and Goetzmann (2001). The second research focus, performance evaluaton, s essentally concerned wth comparng the return earned on a hedge fund wth the return earned on some other standard nvestment asset. Research n ths area can be dvded nto three groups: benchmarkng, performance persstence, and performance n a portfolo context. Key benchmarkng research supports the fact that hedge funds outperform mutual funds, even on a rsk adjusted bass. See, for nstance, Ackermann et al. (1999), Brown et al. (1999), Edwards and Lew (1999), Agarwal and Nak (2000), Edwards and Caglayan (2001), Kao (2002), Amn and Kat (2003) and Malkel and Saha (2005). The thrd research area focuses on hedge fund characterstcs. Ths area s the broadest focus group, startng wth general characterstcs and progressng to performance attrbutes, as n Brown et al. (2001). Characterstcs of the hedge fund ndustry, ncludng the fee structure, data condtonng bases, and the rsk/return characterstc of varous hedge fund strateges have been studed. For nstance, see Park and Staum (1998), Schneewes and Spurgn (1998), and Ackermann et al. (1999) for a thorough dscusson of hedge fund characterstcs. Returns are summarzed n Edwards (1999), Fung and Hseh (1999), and Lamm et al. (1999). Goetzmann et al. (1998) evaluate compensaton ssues. In the last area, researchers study the role of hedge funds n the fnancal market crss and the mplcatons for polcy. For nstance, the role of hedge funds n the Asan crss s documented n Yago et al. (1998, 1999), Echengreen and Matheson (1998), and Brown et al. (2000, 2001). The collapse of LTCM s referenced n Edwards (1999). A summary of the emprcal work on hedge funds leads to the followng conclusons: hedge fund returns are volatle the ncluson of hedge funds n dversfed portfolos rases effcency of portfolos hedge funds have a low correlaton wth tradtonal asset classes fund-of- hedge funds offer dversfcaton benefts to some extent hedge funds may have rsk-adjusted performance persstence dmnshng-return-to-scale may exst n the hedge fund ndustry hedge funds dd not have any drect role n precptatng rsk n the fnancal market ncentve fee structures do not lead hedge fund managers to take more rsk because of the possblty of non-survval hedge funds follow very dynamc strateges Many papers have documented the relatonshp between nvestor sentment and performance n mutual fund ndustry as well as the stock markets n general.

4 586 Nandta Das For example, Lamont and Frazzn (2005), Ippolto (1992), Chevaler and Ellson (1997), Srr and Tufano (1998) study ths relatonshp n the mutual fund ndustry. IV. DATA AND METHODOLOGY The databases popular among researchers and the nvestment communty nclude Center for Internatonal Securtes and Dervatves Markets (CISDM/Hedge) database (formerly, MAR/hedge), whch provdes a comprehensve coverage of global hedge funds; Hedge Fund Research (HFR) database, whch contans more equtybased hedge funds; and TASS, the nformaton and research subsdary of Credt Susse Frst Boston Tremont Advsers. The database provders all offer hedge fund classfcatons and ndces, unfortunately wthout much n common. Hedge fund categores lsted n a partcular database are based on the self-reported style classfcatons of the hedge fund managers. In addton, none of the databases provde nformaton on the complete hedge fund unverse. The databases also dffer on ther defnton of a hedge fund. For example, TASS s the only database that ncludes managed futures funds, whch lmt ther actvtes to futures market. Snce hedge fund managers employ a dverse array of nvestment strateges, the database provders must provde some sort of classfcaton scheme. Although all the major databases rely on the voluntary nformaton provded by the hedge fund managers, style defntons and the number of hedge fund categores dffer among the database provders. The data used for ths study s the monthly hedge fund return of the Center for Internatonal Securtes and Dervatves Markets/Hedge (CISDM/Hedge) database and MornngStar. CISDM/Hedge database was made avalable by Unversty of Massachusetts for ths research. The CISDM/Hedge database provdes monthly returns for all the funds. The study perod for the present research has been selected to be between January 1994 and December CISDM/Hedge data has 184,095 observatons of monthly returns for 2,930 funds. Some more funds had to be dropped from the study due to the unavalablty of some key data that could not be derved from the avalable nformaton. A study perod dataset from January 1994 to December 2004 s constructed from the avalable dataset. The avalable monthly return observatons that are used for the study are 167,009 for 2,930 funds and ther dstrbuton s shown n Table 1. The MornngStar has ndvdual hedge fund data for approxmately 1500 hedge funds domcled n US. Of course we wll have to consder the avalablty of data for ndvdual hedge funds and t s estmated that the data wll have an attrton rate of 50%. The researchers are analyzng the data avalable from 1995 to We wll also look at the 23 categores of HF data n MornngStar. Ths wll also gve an dea of the mpact that fnancal crss has had on hedge fund asset flow. Ths study consders after-fee returns and before-fee returns. In general, hedge funds charge two types of fees: an asset management fee and an ncentve fee. The asset management fee s based on amount of the assets n the fund, usually 1%, or 2% per year. The ncentve fee or the carred nterest s the hedge fund manager s

5 Does Investor Sentment Play a Role n Hedge Fund Return? 587 Table 1 Total Avalable Data and Study Perod Dataset Composton Total Database Study perod ( ) Category Funds Observatons Funds Observatons Number % Number % Number % Number % Event Drven , , Global Internatonal , , Global Regonal Establshed , , Global Regonal Emergng , , Global US , , Global Macro , , US Opportunstc , Long Only/Leveraged , , Market Neutral , , Sector , , Short Sellers , , Total 2, , , , share n a fund s proft. Usually ths s 20 percent and s pad annually n the Unted States. For offshore hedge funds, the ncentve fee s calculated monthly or quarterly. Two other mportant features of a hedge fund fee structure are the hurdle rate and the hgh water mark. The CISDM/Hedge database provdes nformaton on annual fee structure for each of the hedge funds. Subtractng 1/12th of the stated percent fee from the monthly return approxmates the admnstratve fee. Both the hurdle rate and the hgh water mark feature are consdered for computng the ncentve fee. For example, the ncentve fee was subtracted only f the fund n queston had a postve cumulatve return snce t last charged an ncentve fee and had crossed the hurdle rate. Ths takes care of the loss recovery requrement, the mnmum return requrement and assures that there s no double countng of fees. Our study uses an approach smlar to that of Lamont and Frazzn (2005). We study the ncremental flow of funds to dfferent categores and ndvdual hedge funds. We do not answer the queston as to why funds flow nto the hedge fund ndustry, whch s observed from the stellar ncrease n the sze of the ndustry from $38 bllon n 1990 to over $1 trllon n Rather, we try to analyze that, gven the fact that ndustry sze keeps growng what are the factors responsble for the dsproportonate growth of some hedge funds when other funds are experencng net outflow. We carry out our analyss at fund level and also at the category level. We use asset under management (AUM) to calculate the flow. The frst queston that we answer s about nvestor sentment n general. Our hypothess s stated below.

6 588 Nandta Das H 0 : There s an evdence of nvestor sentment as measured by net dsproportonate flow of funds n some hedge funds. H a : There s no evdence of nvestor sentment n the hedge fund ndustry. We measure nvestor sentment as the level of dfferental flow. Under normal crcumstances, every hedge fund should experence an nflow (outflow) of funds that s proportonal to the percentage of the hedge fund ndustry assets that the fund owns. Ths, we call the theoretcal flow for the hedge fund. In short, wth nothng changng, we expect the hedge fund s representaton n the ndustry, n terms of ts share to reman fxed. Dfferental flow s the extra flow of funds that a hedge fund receves over and above ts theoretcal flow (proportonal flow). The dfferental flow s calculated as a dfference of real flow and theoretcal flow. The model s expressed usng the followng equatons: RflMm = AUMm AUMm 1(1) + r (1) m RflY t 12 = RflMm (2) = 1 TflY t AUM t 1 Agg = FL Agg t (3) AUMt 1 AUM N Agg t 1 () AUMt 1 = 1 = (4) FL Agg t N = RflY (5) = 1 t where Dff () FL = RflY TflY (6) t t t RflM s the monthly real flow for hedge fund n month m, m AUM s the assets under management for hedge fund n month m, m AUM s the assets under management for hedge fund n month m-1, m 1 r s the monthly return for hedge fund n month m, m RflY s the annual real flow for hedge fund n year t, t TflY s the theoretcal annual flow for hedge fund n year t, t AUM s the year-end assets under management for hedge fund n year t-1, t 1

7 Does Investor Sentment Play a Role n Hedge Fund Return? 589 Agg AUMt s the year-end assets under management for all the funds n the 1 database or all funds n the category for year t-1. Agg FL s the flow of funds (new money) to the hedge fund ndustry (wth N t funds) or to the category (wth N funds) for year t, and Dff () FL s the dfferental flow of funds to hedge fund n year t, t Equatons 1 through 6 descrbe the model that we use to calculate nvestor sentment. When Dff () FL s postve there s a net nflow of dfferental funds t and when t s negatve the hedge fund experenced an outflow of funds. It s mportant to realze that we are not nterested n measurng nflow or outflow. We measure dfferental nflow or outflow experenced by that fund. Wth a growng ndustry, t s expected that there wll be net nflow of funds. Our varable of nterest s the dsproportonate flow of funds, represented by Dff (). FL t The asset under management fgure that s reported n the database conssts of two parts. There s some ncrease or decrease n assets under management that s solely due to the return of hedge fund. Then there s also the ncrease or decrease n assets under management that s due to nflow or outflow of funds. We calculate monthly real flow as return s updated on a monthly bass. Ths takes nto consderaton the mpact of return on real flow. Ths s more accurate than calculatng yearly flow usng average annual return as the annual return may not be a good representaton of the ups and downs n monthly return that hedge funds generally experence. The data on assets under management does not appear to have been updated on a monthly bass, though most of the funds n the database do have assets under management reported on a monthly bass. Careful observaton and analyss revealed that asset under management s updated only n the last month of the year, when the hedge fund managers have to calculate fees that they can charge the nvestors. Equaton 2 gves the annual real flow, obtaned by summng up all the monthly real flow for a partcular hedge fund. We calculate dfferental flow for three dfferent scenaros. In scenaro 1, we calculate dfferental flow wth hedge beng part of the unverse of hedge funds, whch n our case s the complete database. In scenaro 2, we calculate dfferental flow wth hedge fund beng a subset of the category t belongs to. In scenaro 3, we calculate category dfferental flow wth category beng a subset of the hedge fund database. Hypothetcally assume that the ndustry conssts of 10 funds wth assets under management of 10 mllon each. The ndustry sze s 100 mllon. Followng year, the ndustry experences an nflow of $10 mllon and a captal gan of 2 mllon. The ndustry sze s now 112 mllon. Under theoretcal flow measure, we would expect each hedge fund to receve $1 mllon new money (10 percent of 10 mllon).

8 590 Nandta Das Instead hedge fund 1 receves 1.5 mllon of the new money, whereas hedge fund 2 receves only 0.5 mllon. In ths case hedge fund 1 has experenced a dfferental flow of +0.5 mllon whereas hedge fund 2 has a dfferental flow of -0.5 mllon. So nvestor sentment favored hedge fund 1. To test ths hypothess of nvestor sentment, we sort all the hedge funds on the bass of dfferental flow, the varable of nterest. We report results of three years from our study perod of 11 years. Table 2 shows 10 BIG OUTFLOW hedge funds for each of the three years (1994, 1999, and 2004) out of the study perod (1994 to 2004). The dfferental flow of hedge funds s calculated wth respect to total hedge funds, whch n our case s the total database. These are the hedge funds that Table 2 Dfferental Outflow of Funds wth respect to Total Funds Fund ID Real Flow Theoretcal Dfferental Category (US$M) Flow (US$M) Flow (US$M) Panel A Global Macro Global Macro Global Regonal Establshed Global Macro Global US Market Neutral Global US Global Internatonal Event Drven Global Macro Panel B Market Neutral Global Macro Market Neutral Global Macro Market Neutral Global Regonal Establshed Global Regonal Emergng Global Regonal Establshed Event Drven Market Neutral Panel C Market Neutral Market Neutral Market Neutral Global Regonal Establshed Market Neutral Market Neutral Global Regonal Emergng Market Neutral Global Macro Event Drven

9 Does Investor Sentment Play a Role n Hedge Fund Return? 591 experenced a net dfferental outflow for the representatve year. Hedge fund 4607 has a negatve real flow (outflow) of $150 mllon. If hedge fund 4607 had receved flows (new money) n proporton to ts prevous (1993) year-end sze, t would have had an nflow of $2755 mllon. Havng found an evdence of nvestor sentment n the hedge fund ndustry, the next step n our analyss s to see what causes ths nvestor sentment. Is there a partcular type of hedge fund that nvestors prefer? To answer ths queston, we come up wth a set of hypothess descrbed below. The hypotheses related to the nvestor sentment hypothess stated above are: H 0 : Category membershp has an effect on net dfferental flow of funds experenced by the hedge fund. H a : Category membershp has no effect on net dfferental flow of funds experenced by the hedge fund. H 0 : Amount of leverage has an effect on net dfferental flow of funds experenced by the hedge fund. H a : Amount of leverage has no effect on net dfferental flow of funds experenced by the hedge fund. H 0 : Sze has an effect on net dfferental flow of funds experenced by the hedge fund. H a : Sze has no effect on net dfferental flow of funds experenced by the hedge fund. H 0 : Portfolo allocaton has an effect on net dfferental flow of funds experenced by the hedge fund. H a : Portfolo allocaton has no effect on net dfferental flow of funds experenced by the hedge fund. The last column n Table 2 provdes the category membershp of the hedge fund n queston. For year 1994 (panel A) t appears that the funds that experenced dfferental outflow n general belonged to the category Global Macro. Smlar analyss for year 1999 and 2004 reveals that hedge funds that have been experencng a net dfferental outflow n general belong to the category Market Neutral. In general, we fnd that the funds that have experenced dfferental net outflow belong to the category Global Macro for 1994 to 1998, and Market Neutral for 1999 to 2004 wth the excepton of the year Table 3 shows top 10 BIG INFLOW (wth respect to total funds) hedge funds. For year 1994 (panel A) t appears that the funds that experenced dfferental nflow belong to the category Global Macro (5 out of 10 hedge funds). Smlar analyss for year 1999 and 2004 reveals that hedge funds that have been experencng a net dfferental nflow belong to the category Regonal Establshed and Event Drven respectvely. In general, we fnd that the funds that have experenced dfferental net nflow belong to the category Market Neutral for 1998 to 2003 except year 1999 (4 out of 11 years of study) and varous categores for the other years of study.

10 592 Nandta Das Table 3 Dfferental Inflow of Funds wth respect to Total Funds Fund ID Real Flow Theoretcal Dfferental Category (US$M) Flow (US$M) Flow (US$M) Panel A Global Macro Market Neutral Global Regonal Establshed Global Regonal Establshed Global Macro Global Macro Global Macro Global Macro Global Regonal Emergng Global Regonal Establshed Panel B Global Regonal Establshed Market Neutral Global Regonal Establshed Market Neutral Global Internatonal Global Regonal Establshed Market Neutral Global Regonal Establshed Global Internatonal Global Macro Panel C Market Neutral Event Drven Global Internatonal Event Drven Event Drven Market Neutral Global Macro Event Drven Global Regonal Establshed Global Regonal Establshed

11 Does Investor Sentment Play a Role n Hedge Fund Return? 593 Table 4 Dfferental Outflow of Funds wth respect to Category Funds Fund ID Real Flow Theoretcal Dfferental Category (US$M) Flow (US$M) Flow (US$M) Panel A Global Macro Global Macro Global Internatonal Global Internatonal Global Regonal Establshed Global Regonal Emergng Global US Global Macro Global Internatonal Market Neutral Panel B Market Neutral Global Regonal Establshed Global Regonal Establshed Global Macro Global Regonal Establshed Global Regonal Establshed Global Regonal Establshed Global Regonal Establshed Global Regonal Establshed Global Macro Panel C Market Neutral Global Regonal Establshed Market Neutral Market Neutral Event Drven Event Drven Market Neutral Global Regonal Emergng Event Drven Event Drven

12 594 Nandta Das Comparng the results of Table 2 and Table 3, t s not clear f category membershp has had anythng to do wth the dfferental outflow or dfferental nflow of funds. For the year 1994, the hedge funds that experenced dfferental outflow belonged to the category Global Macro but so dd the funds that experenced dfferental nflow. Even for the results across the years, category membershp does not seem to be a domnant feature that would explan the dfferental flow of these hedge funds. Dfferental nflow and outflow results for hedge funds wth respect to category are shown n Table 5 and Table 6 respectvely. Note that the category US Opportunstc dsappeared from the database n For year 1994 (panel A) t appears that the funds that experenced dfferental outflow belonged to dfferent categores. Smlar analyss for year 1999 and 2004 reveals that hedge funds that have been experencng a net dfferental outflow belong to the category Global Regonal Establshed and Market Neutral respectvely. For dfferental nflow wth respect to category, t appears that the funds n general belong to the category Global Macro for 1994, and Market Neutral for 1999 and Analyss of dfferental flow (nflow and outflow) wth respect to category funds for the complete study perod reveals that most of these funds belong to category Market Neutral (8 out of 11 years of study for outflow, and 7 out of 11 years of study for nflow). We can conclude that category membershp does not seem to have an mpact on the dfferental flow (nflow and outflow) of funds experenced by some hedge funds. Specfcally, when we look at the dfferental nflow and outflow of funds t appears that there has been some redstrbuton of funds wthn the category Market Neutral resultng n both maxmum dfferental nflow and maxmum dfferental outflow occurrng n funds that belong to Market Neutral. Snce category membershp does not gve much detal, we further analyzed these BIG INFLOW HFs (10 hedge funds wth maxmum dfferental nflow) and BIG OUTFLOW HFs (10 hedge funds wth maxmum dfferental outflow) to help us determne what are the characterstcs of the IN FASHION HFs, f any. But before we proceed wth ths analyss, t s mportant to probe nto the dfferental flow results (Table 2 to Table 5) a lttle more. When we compare the results of dfferental nflow of funds wth respect to total hedge funds (Table 3 and Table 4) wth dfferental flow of funds wth respect to category funds (Table 5 and Table 6), we observe that the Fund ID does not appear to change much. In general, the hedge funds that experenced a net dfferental nflow (outflow) wth respect to total funds also had dfferental nflow (outflow) wth respect to category funds. For example, t appears that for the year 1994 the hedge funds that experenced outflow of funds wth respect to total funds and also wth respect to category funds are 4607, 4608, 3385, 4548, 3341, 3622, and 1239 (7 out of 10 funds). These funds had a dfferental outflow of funds when n general the total hedge fund and the respectve categores were experencng an nflow (or less outflow) of funds. Ths renforces our analyss that category

13 Does Investor Sentment Play a Role n Hedge Fund Return? 595 Table 5 Dfferental Inflow of Funds wth Respect to Category Funds Fund ID Real Flow Theoretcal Dfferental Category (US$M) Flow (US$M) Flow (US$M) Panel A Global Regonal Establshed Global Macro Global Regonal Establshed Market Neutral Global Macro Global Macro Global Regonal Emergng Global Macro Global Macro Global Regonal Establshed Panel B Market Neutral Global Regonal Establshed Market Neutral Market Neutral Global Internatonal Market Neutral Global Regonal Establshed Global Internatonal Market Neutral Global Macro Panel C Market Neutral Global Internatonal Market Neutral Event Drven Market Neutral Global Macro Market Neutral Event Drven Global Regonal Establshed Global Regonal Establshed

14 596 Nandta Das Table 6 Characterstcs of BIG OUTFLOW HFs wth Respect to Total Funds Fund ID Leverage AUM Portfolo Category (US$M) S B C W O F Panel A Y N N N Y N Global Macro Y Y Y Y Y Y Global Macro 3385 X N N N N Y Y Global Regonal Establshed 4548 X Y N N N Y N Global Macro Y N Y N N Y Global US 3622 X Y N N N N N Market Neutral 3172 X N N Y N N Y Global US Y N N N N N Global Internatonal 1603 X X X X X X X Event Drven Y Y N Y Y Y Global Macro Panel B Y N N Y Y N Market Neutral 1601 X X X X X X X Global Macro 3786 X X X X X X X Market Neutral Y N N N Y N Global Macro Y N N Y Y N Market Neutral Y N N N N N Global Regonal Establshed Y N N N Y N Global Regonal Emergng Y Y Y Y Y N Global Regonal Establshed Y N N N Y N Event Drven N Y N N Y Y Market Neutral Panel C Y N N Y Y N Market Neutral Y N N N N N Market Neutral Y Y N Y Y Y Market Neutral 1795 X X X X X X X Global Regonal Establshed Y Y Y N Y Y Market Neutral 526 X X X X X X X Market Neutral Y N N N N N Global Regonal Emergng Y N N N Y Y Market Neutral Y Y N Y Y Y Global Macro Y N N N Y Y Event Drven Note: S=Stocks, B=Bonds, C=Currency, W=Warrants, O=Optons and F=Futures

15 Does Investor Sentment Play a Role n Hedge Fund Return? 597 membershp does not appear to have an mpact on the dfferental flow experenced by hedge funds. Also, t s mportant to realze that hedge fund classfcaton vares among dfferent databases. The classfcaton schemes provded by dfferent database provders are overlappng and many-a-tmes confusng from the perspectve of the nvestor. Table 6 and Table 7 provde the general characterstcs of these BIG OUTFLOW HFs and BIG INFLOW HFs wth respect to total funds respectvely for the three representatve years (1994, 1999, and 2004). The characterstcs that we looked nto are leverage, sze, and portfolo allocaton. Leverage does not seem to have any mpact on the net dfferental flow (nflow and outflow) of funds experenced by hedge funds. Both the BIG OUTFLOW HFs and BIG INFLOW HFs have n general a low leverage except for a few funds. As far as the sze of hedge fund s concerned we were nterested n seeng f there s any dseconomes-of-scale for nvestor sentment that can be nferred from the dfferental nflow or outflow of funds. It appears that nvestors do not prefer any specfc sze. For example, hedge fund 93 (Table 7, panel C) has a sze of $15.7 bllon and s one of the BIG INFLOW HFs for the year Ths hedge fund entered the database n 1998 and s fve years old. There does not appear to be any threshold sze for hedge funds to be favored by nvestors. The last hypothess that we test s to see f portfolo allocaton has any effect on net dfferental flow of funds experenced by the hedge fund. It s observed from Table 6 that the BIG OUTFLOW HFs nvest n stocks and optons for all the three years. For the year 1994, the BIG OUTFLOW HFs nvested n futures as well, whereas for the year 2004, they nvested n warrants and futures as well. Analyzng the portfolo allocaton of BIG OUTFLOW HFs for all the eleven years of study we fnd that these funds n general do not nvest n currency, warrants and futures. Analyzng the BIG INFLOW HFs gves us some nterestng results. For the years 1994 to 1998, the BIG INFLOW HFs nvested n stocks, bonds and optons. Then for the years 1999 and 2000, these funds as a group stopped nvestng n optons. For year 2001 and 2002 there was a trend to nvest n all the nstruments (stocks, bonds, currency, warrants, optons and futures). A conservatve trend s observed for the year 2003 and 2004 where these funds are nvestng only n stocks and bonds. Smlar results are found for MornngStar database but when the study perod s broken nto pre and post crss, the results are dfferent. Table 8 and Table 9 provde the general characterstcs of these BIG OUTFLOW HFs and BIG INFLOW HFs wth respect to category funds respectvely for the three representatve years (1994, 1999, and 2004). The results are smlar to the BIG OUTFLOW HFs and BIG INFLOW HFs wth respect to total funds. We can thus conclude that portfolo allocaton has no effect on net dfferental flow of funds experenced by the hedge fund. Table 10 provdes the results of dfferental flow of category for the three representatve years of study. For the year 1994, the category Global Macro had

16 598 Nandta Das Table 7 Characterstcs of BIG INFLOW HFs wth Respect to Total Funds Fund ID Leverage AUM Portfolo Category (US$M) S B C W O F Panel A Y N N N N N Global Macro N Y N N N N Market Neutral Y Y Y Y Y N Global Regonal Establshed Y N N N N N Global Regonal Establshed N Y N N N N Global Macro 1601 X X X X X X X Global Macro Y Y N Y Y Y Global Macro Y Y N N N N Global Macro Y N N N Y N Global Regonal Emergng Y Y N N Y Y Global Regonal Establshed Panel B X X X X X X X Global Regonal Establshed 304 X X X X X X X Market Neutral N Y N N Y Y Global Regonal Establshed 3073 X X X X X X X Market Neutral Y N N N N Y Global Internatonal 1798 X X X X X X X Global Regonal Establshed Y Y N N N N Market Neutral Y Y N N N N Global Regonal Establshed 1602 X Y Y Y Y Y Y Global Internatonal N Y N N N N Global Macro Panel C N Y N N N N Market Neutral Y N N N N N Event Drven Y N N N Y Y Global Internatonal N Y N N N N Event Drven N Y N N N N Event Drven 1633 X N X X X X X Market Neutral Y N Y N N N Global Macro Y N N Y Y N Event Drven Y N Y N N Y Global Regonal Establshed Y Y Y N N N Global Regonal Establshed

17 Does Investor Sentment Play a Role n Hedge Fund Return? 599 Table 8 Characterstcs of BIG OUTFLOW HFs wth respect to Category Funds Fund ID Leverage AUM Portfolo Category (US$M) S B C W O F Panel A Y N N N Y N Global Macro Y Y Y Y Y Y Global Macro Y N N N N N Global Internatonal Y N N N Y Y Global Internatonal 3385 X N N N N Y Y Global Regonal Establshed Y Y Y Y Y N Global Regonal Emergng Y N Y N N Y Global US 4548 X Y N N N Y N Global Macro Y N N N N N Global Internatonal 3622 X Y N N N N N Market Neutral Panel B Y N N Y Y N Market Neutral Y Y Y Y Y N Global Regonal Establshed Y N N N N N Global Regonal Establshed 1601 X X X X X X X Global Macro Y Y N N Y Y Global Regonal Establshed Y Y Y N Y Y Global Regonal Establshed N Y N Y N N Global Regonal Establshed 3148 X Y N N N Y N Global Regonal Establshed N Y N Y N N Global Regonal Establshed Y N N N Y N Global Macro Panel C Y N N Y Y N Market Neutral 1795 X X X X X X X Global Regonal Establshed Y N N N N N Market Neutral Y Y N Y Y Y Market Neutral Y N N N Y Y Event Drven Y N N N N N Event Drven Y Y Y N Y Y Market Neutral Y N N N N N Global Regonal Emergng N Y Y Y Y Y Event Drven Y N N N N N Event Drven

18 600 Nandta Das Table 9 Characterstcs of BIG INFLOW HFs wth Respect to Category Funds Fund ID Leverage AUM Portfolo Category (US$M) S B C W O F Panel A Y Y Y Y Y N Global Regonal Establshed Y N N N N N Global Macro Y N N N N N Global Regonal Establshed N Y N N N N Market Neutral Y Y N Y Y Y Global Macro N Y N N N N Global Macro Y N N N Y N Global Regonal Emergng Y Y N N N N Global Macro 1601 X X X X X X X Global Macro Y Y N N Y Y Global Regonal Establshed Panel B Y N N N N N Market Neutral 1798 X X X X X X X Global Regonal Establshed Y Y N N Y N Market Neutral 304 X X X X X X X Market Neutral Y N N N N Y Global Internatonal 3073 X X X X X X X Market Neutral Y Y N N N N Global Regonal Establshed 1602 X Y Y Y Y Y Y Global Internatonal Y Y N N N N Market Neutral N Y N N N N Global Macro Panel C Y N N N Y N Market Neutral Y N N N Y Y Global Internatonal N Y N N N N Market Neutral N Y N N N N Event Drven N Y N N N N Market Neutral Y N Y N N N Global Macro 1633 X N X X X X X Market Neutral Y N N Y Y N Event Drven Y N Y N N Y Global Regonal Establshed Y Y Y N N N Global Regonal Establshed

19 Does Investor Sentment Play a Role n Hedge Fund Return? 601 Table 10 Dfferental Flow of Category wth Respect to Total Funds Category Real Flow Theoretcal Dfferental Flow (%) (US$B) Flow (US$B) Flow (US$B) Panel A Global Macro Event Drven Global US Market Neutral US Opportunstc Long Only/Leveraged Sector Short Sellers Global Regonal Establshed Global Internatonal Global Regonal Emergng Panel B Market Neutral Global Macro Event Drven Global Regonal Emergng Long Only/Leveraged Short Sellers Global Internatonal Sector Global Regonal Establshed Panel C Market Neutral Sector Global Regonal Emergng Long Only/Leveraged Short Sellers Global US Global Macro Global Internatonal Event Drven Global Regonal Establshed

20 602 Nandta Das the maxmum dfferental outflow whereas category Global Regonal Emergng has maxmum dfferental nflow of funds. Panel B and panel C report the results for the year 1999 and 2004 respectvely. Table 11 provdes the organzaton of categores sorted on the bass of the results of dfferental flow for each year of study. For the years 1994 to 1998 the category Global Macro had maxmum outflow, whereas category Market Neutral had maxmum outflow for the years 1999 and Further analyss reveals that nvestors n aggregate were pullng out of the category Global Macro and redrectng ther sentment to the category Market Neutral. Market Neutral was the n-fashon category from 1995 to 2001 except for the year The year 1999 appears to have experenced a fad n nvestor sentment, wth nvestors favorng the category Global Regonal Establshed. Ths category has emerged agan as n-fashon category n the year In fact for Fgure 1: CISDM/Hedge Classfcaton of Hedge Funds

21 Does Investor Sentment Play a Role n Hedge Fund Return? 603 each of the years 1999, 2002 and 2003 a dfferent category was n-fashon. The category Event Drven made ts entry as an n-fashon category n the year Ths category contnued to attract nvestors as can be seen from Table 11. V. CONCLUSION In ths paper we document the exstence of nvestor sentment n the hedge fund ndustry. To test the hypothess of nvestor sentment, we sort all the hedge funds on the bass of dfferental flow, the dfferental flow (nflow and outflow) of funds experenced by ndvdual hedge funds and dfferent categores. We fnd an overwhelmng evdence of nvestor sentment n the hedge fund ndustry for both ndvdual hedge funds and categores. For ndvdual hedge funds we analyze to see f category membershp has any mpact on the dfferental flow of funds. We fnd that the funds that have experenced dfferental net outflow belong to the category Global Macro for 1994 to 1998, and Market Neutral for 1999 to 2004 wth the excepton of the year In aggregate t appears that category membershp does not mpact the flow of funds. Even for the results across the years, category membershp does not seem to be a domnant feature that would explan the dfferental flow of funds. There appears to be some redstrbuton of funds wthn the same category resultng n both maxmum dfferental nflow and maxmum dfferental outflow occurrng n hedge funds that belong to the same category. We analyze further to see f the nvestor sentment s drven by the fundamental characterstcs of hedge funds, namely sze, leverage, and portfolo allocaton. Leverage and sze do not appear to be the dscernng factor for nvestor sentment. There does not appear to be any threshold sze for hedge funds that s favored by nvestors. We also fnd that portfolo allocaton has no effect on net dfferental flow of funds. Investor sentment s not governed by the fundamental characterstcs of the hedge funds. We calculate the dfferental flow of funds for all the categores for each year of study. It appears that nvestors n aggregate were pullng out of the category Global Macro and redrectng ther sentment to the category Market Neutral. Market Neutral was the n-fashon category from 1995 to The category Event Drven made ts entry as an n-fashon category n the year Ths category contnued to attract nvestors tll the end of our study perod. The results are qute nterestng for pre and post fnancal crss for hedge fundsusng Mornngstar database. It appears that hedge funds have lost ther charm n attractng new money. It could be because of regulatons or because of lack of confdence n the global economy. References Ackerman, Carl, Rchard McEnally and Davd Ravenscraft (1999), The Performance of Hedge Funds: Rsk, Return, and Incentves, The Journal of Fnance, Agarwal, Vkas, Naveen D. Danel and Narayan Y. Nak (2004), Flows, Performance, and Manageral Incentves n Hedge Funds, EFA 2003 Annual Conference Paper No. 501.

22 604 Nandta Das Brown, Stephen J., Wllam N. Goetzmann and James Park (2001), Careers and Survval: Competton and Rsk n the Hedge Fund and CTA Industry, The Journal of Fnance, (56:5), Brown, Stephen J. and Wllam N. Goetzmann (1995), Performance Persstence, The Journal of Fnance 50: 2, Brown, Stephen J., Wllam N. Goetzmann and James Park (2000), Hedge funds and the Asan Currency Crss of 1997, The Journal of Portfolo Management, Brown, Stephen J., Wllam N. Goetzmann and Roger G. Ibbotson (1999), Offshore Hedge Funds: Survval and Performance , Journal of Busness 72: 1, Chen, Na-Fu, Rchard Roll and Stephen A. Ross (1986), Economc Forces and the Stock Market, The Journal of Busness 59: 3, Connor, Gregory (1995), The Three Types of Factor Models: A Comparson of Ther Explanatory Power, Fnancal Analysts Journal 51: 3, Fama, Eugene F. and James D. MacBeth (1973), Rsk, Return, and Equlbrum: Emprcal Tests, The Journal of Poltcal Economy 81: 3, Frankln, R. Edwards and Jmmy Lew (1999), Hedge Funds Versus Managed Futures as Asset Classes, Journal of Dervatves, Frankln, R. Edwards (1999), Hedge Funds and the Collapse of Long Term Captal Management, Journal of Economc Perspectves 13: 2, Fung, Wllam and Davd A. Hseh (2002), Benchmarks of hedge-fund performance: Informaton content and measurement bas, Fnancal Analysts Journal, (2001a), The Rsk n Hedge Fund Strateges: Theory and Evdence from Trend Followers, The Revew of Fnancal Studes 14:2, (2001b), Performance Characterstcs of Hedge Funds and Commodty Funds: Natural versus Spurous Bases, Journal of Fnancal and Quanttatve Analyss 35: 3, (1997), Survvorshp Bas and Investment Style n the Returns of CTAs: The Informaton Content of Performance Track-Records, Journal of Portfolo Management 24: 1, Goetzmann, Wllam N., Jonathan Ingersoll Jr. and Stephen A. Ross (1998), Hgh Water Marks, NBER Workng Paper, February. Lamont, Owen and Andrea Frazzn (2005), Mutual Fund Flows and the Cross-Secton of Stock Returns, Yale Internatonal Center for Fnance Workng Paper No , May. McFall, Lamm Jr. (1999), Portfolo of Alternatve Assets: Why not 100% Hedge Funds, The Journal of Investng, Park, James M. and Jeremy C. Staum (1998), Performance Persstence n Alternatve Investment Industry, Workng Paper Columba Unversty. Schneewes, Thomas (1998), Managed Futures, Hedge Fund and Mutual Fund Performance: An Equty Class Analyss, The Journal of Alternatve Investments, Schneewes, Thomas and Rchard Spurgn (1998), Estmaton: A Mult-factor Analyss of Hedge Fund, Managed Futures, and Mutual Fund Return and Rsk Characterstcs, Journal of Alternatve Investments, Schneewes, Thomas (1998), Dealng wth the myths of Hedge Fund Investment, The Journal of Alternatve Investments,