Management team structure and mutual fund performance

Int. Fin. Markets, Inst. and Money 20 (2010) 197–211 Contents lists available at ScienceDirect Journal of International Financial Markets, Institutions & Money journal homepage: www.elsevier.com/locate/intfin Management team structure and mutual fund performance Iordanis Karagiannidis ∗ Eli Broad Graduate School of Management, Michigan State University, 315 Eppley Center, East Lansing, MI 48824, USA a r t i c l e i n f o Article history: Received 12 October 2009 Accepted 18 October 2009 Available online 27 October 2009 JEL classification: G23 L25 M12 Keywords: Mutual funds Team-manager Single-manager Performance a b s t r a c t We investigate the relationship between performance and portfolio management team structure of open-end mutual funds during 1997–2004. We first analyze differences in performance and risk taking between single-manager and multiple-manager mutual funds and find that the latter underperform the single-manager funds in terms of risk-adjusted returns during the 2001–2004 bear market. This underperformance is more evident among growthoriented funds. There are no differences observed in the 1997–2000 bull market. Not all multiple-manager funds, however, are managed by pure teams. When we compare the performance of single-manager and pure-team funds we do not find any differences in performance. The underperformance of multiple-manager funds documented in previous studies comes from multiple-manager funds that employ many investment advisors and, therefore, their exact management structure is unknown. We also document differences in management structure reporting between Morningstar and CRSP. © 2009 Elsevier B.V. All rights reserved. 1. Introduction The U.S. mutual fund industry has grown dramatically in the past decade. From $2.8 trillion in 1995, assets under management rose to a record-breaking $8.1 trillion in 2004 (ICI, 2005a). As the scale of the mutual fund industry has changed so have the funds themselves. For example, funds have introduced additional share classes to attract more investors and developed new channels to better reach the investment public. Moreover, the architecture of the portfolio management system has evolved with ∗ Tel.: +1 517 353 7568. E-mail address: [email protected]. 1042-4431/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.intfin.2009.10.003 198 I. Karagiannidis / Int. Fin. Markets, Inst. and Money 20 (2010) 197–211 teams of managers replacing single managers as the dominant decision-making unit.1 During this period and thereafter, mutual funds actively promoted the potential benefits of team management to its current and potential clients. By way of illustration, according to its Web site, “[the] Brazos Funds view the team-based approach as an important component in creating less risk for clients and increases their long-run returns.” Many studies, especially in the management and psychology literature, have examined the role and performance of teams. After reviewing more than 100 studies on team performance in a variety of circumstances, Stock (2004) reports that almost all of them find that teams behave differently than individuals but that these differences do not necessarily translate to superior performance no matter how measured. In the context of portfolio management, advantages of teams include being able to diversify style and judgment (Sharpe, 1981) and having a broader range of specialized skills and knowledge and abilities to process larger amounts of information (Hill, 1982; Herrenkohl, 2004). Disadvantages include the presence of free riders (Holmstrom, 1982; Rasmusen, 1987) and delays in decision making (Sah and Stiglitz, 1988). The role of risk is ambiguous. Barry and Starks (1984) suggest that overall teams may reduce risk taking, but others (Janis, 1982; Herrenkohl, 2004) contend that teams may actually increase risk because shared risk makes the risk borne by individuals seem less. Nevertheless, limited research has been done concerning the notion that the team is the superior fund management structure, and the empirical evidence that has been reported provides little or no support for this contention. For instance, Prather and Middleton (2002) find that there is no difference in the performance of team-manager and single-manager funds, while Chen et al. (2004) and Bär et al. (2005) document that team-manager funds do not perform as well as their single-manager counterparts. Massa et al. (2006) provide evidence that team underperformance is correlated with the anonymity of the managers, and Qiu (2003) shows that single-manager funds are more aggressive and tend to adjust their risk exposure more than team-managed funds. In a more recent paper, Han et al. (2008) find evidence suggesting a positive relation between fund performance and team management. All of the abovementioned studies characterize funds that list multiple portfolio managers as team-manager funds. This characterization, however, may be overly simplistic. Many times funds assign the management of their portfolio to more than one investment advisor and these advisors may be internal of external to the family. Carnahan (2004), for example, reports that Vanguard has contracts with 24 outside management companies for one-third of its funds. Thus, the team-manager category used in the previously cited studies contains (1) one investment advisor with multiple managers, (2) multiple investment advisors each with multiple managers, and (3) multiple investment advisors each with a single manager. Moreover, the single-manager designation is comprised of (1) one investment advisor with a single manager and (2) multiple investment advisors with a single manager. Even these distinctions are fuzzy representations of the decision-making structure. To illustrate, consider the following possibilities. First, co-managers who belong to the same advisory company may share the same pool of analysts and communicate with each other, and even though a consensus must be reached, individual members may be held accountable for specific recommendations. Thus, it is difficult to determine whether their investment decisions are independent of each other. Second whether the advisor is external or internal to the firm may make a difference because, according to Chen et al. (2004), externally advised funds are more likely to be closed down for poor performance than comparable internally run funds. The risk of closure may influence investment decisions. Finally, in the case of multiple advisors, their relative contribution to performance is unclear. This is because usually there is one major advisor who hires one or more sub-advisors. Kuhnen (2004) reports that about one-third of all mutual funds are managed by more than one advisory firm and that it is often the sub-advisor(s) that is responsible for the day-to-day management of the fund. This gives rise to 1 In one sense all funds are team managed. This is because many analysts and support staff work together in collecting and analyzing data. The distinction between single-manager and multiple-manager funds refers to whether a single individual or multiple individuals make the final transaction decisions. I. Karagiannidis / Int. Fin. Markets, Inst. and Money 20 (2010) 197–211 199 the multiple advisor and single manager set-up being functionally the same as a single advisor with a single manager. The purpose of our paper is to extend the research that investigates whether mutual management structure affects performance by more precisely defining alternative structures. Most studies on mutual fund performance use either the CRSP Survivorship Bias Free U.S. Mutual Fund database or Morningstar, Inc. We use Morningstar as our primary data source for two reasons. Morningstar lists the names of advisors and managers while CRSP only provides the names of managers. Second, Elton et al. (2001) suggest that Morningstar is more complete and accurate source of mutual fund information. However, to put our results in perspective, where possible we compare our Morningstar results to those obtained using CRSP. From Morningstar we hand-collect a comprehensive dataset of 2031 U.S. open-end, domesticequity mutual fund portfolios (7701 fund–year observations) covering December 1996 to December 2004, inclusively. Our sample period includes bull (1997–2000) and bear (2001–2004) markets. We separate the funds into three categories based on the structure of their portfolio management team: (1) single-manager funds (those with single and multiple advisors), (2) pure-team funds (those that list multiple managers but only one investment advisor) and (3) mixed-team funds (those that list multiple managers and many investment advisors). Our data does not allow us to determine the exact structure of the portfolio management team of mixed-team funds. We refer to funds that list many managers (pure-team and mixed-team funds) as multiple-manager funds. During this 8-year period, the assets of our sample of mutual funds, as we report in Table 1, increased from $1.15 trillion to $2.09 trillion, a change of 82%. In the beginning of the period, 32% of the funds in our sample were managed by multiple managers rather than a single individual. These multiple-manager funds accounted for 28% of the industry’s total assets. In December 2004, however, the number of multiple manager funds as well as their assets accounted for approximately 57% of the total. The average size of the single- and multiple-manager funds varies but there is no discernable trend. In some years the average size of single-manager funds is larger than multiple-manager funds and in others the reverse is true. The vast majority of multiple-manager funds are categorized as pure-team funds even though the percentage of multiple manager funds that use pure teams decreases over time. In 1997, almost 84% of multiple-manager funds are pure-team funds as compared to 69% in 2004. The portion of multiple-manager fund assets associated with pure teams, however, remains stable over time at approximately 86%. Our major results are as follows. First, we report that there is no difference in performance between single-manager and pure-team funds regardless of the performance measure used. Second, we find that the performance of mixed-team funds is significantly lower that the performance of pure teams only in the bear market for growth-oriented funds when the Sharpe ratio, the 1factor alpha and the 4-factor alpha are used as performance measures. For income-oriented funds, mixed-team funds also underperform in the bear market but only in the case of the 4-factor alpha. Our results survive a series of robustness checks. Finally, we document that there are noticeable differences in the categorization of single- and multiple-manager funds between the Morningstar and CRSP data sources, and we present results with both and show that there are significant differences. Our findings, taken together with the increased popularity of the team approach in portfolio management, present a puzzle. Namely, why do investors and fund families prefer team managed funds if they do not offer superior risk-adjusted returns? Unfortunately, our data do not provide an empirical answer to this question. It may be, however, that investors worry about the stability of management and do not want to risk losing the “star” stock picker of a single-manager fund. Thus, we conjecture that the typical mutual fund investor is not a financially sophisticated consumer (by nature or by choice) and may not even know whether a fund is managed by a single-manager or a team, let alone be aware of the names or expertise of the fund’s managers. These investors are likely to accept the mutual funds’ and possibly advisors’ assertions that the teams are superior management vehicles. These assertions are often part of a promotional package put together by the fund to enhance their reputation by making their name familiar to potential clients and reinforcing their brand with their current clients. Our results echo those of Gruber (1996), Huberman 200 I. Karagiannidis / Int. Fin. Markets, Inst. and Money 20 (2010) 197–211 Table 1 Data summary. Panel A: Portfolio by management structure Year 1997 1998 1999 2000 2001 2002 2003 2004 All years All funds Single-manager funds (% of all funds) Multiple-manager funds (% of all funds) Pure-team funds (% of multiple-manager funds) Mixed-team funds (% of multiple-manager funds) 702 475 67.66% 227 32.34% 190 83.70% 37 16.30% 741 472 63.70% 269 36.30% 223 82.90% 46 17.10% 818 487 59.54% 331 40.46% 266 80.36% 65 19.64% 857 1009 1054 1125 1395 7701 468 521 521 523 589 54.61% 51.64% 49.43% 46.49% 42.22% 53% 389 488 533 602 806 45.39% 48.36% 50.57% 53.51% 57.78% 47% 301 375 419 460 555 77.38% 76.84% 78.61% 76.41% 68.86% 77% 88 113 114 142 251 22.62% 23.16% 21.39% 23.59% 31.14% 23% Panel B: Total fund assets (in billions) Year 1997 1998 1999 2000 2001 2002 2003 2004 Total assets: all funds Single-manager funds Multiple-manager funds Pure-team funds Mixed-team funds 1154.92 831.32 323.60 280.35 43.25 1411.61 977.74 433.87 376.57 57.30 1726.88 1089.74 637.14 537.72 99.42 1791.21 1028.94 762.26 652.61 109.65 1564.67 771.88 792.79 691.06 101.73 1254.25 623.07 631.18 560.38 70.80 1685.05 774.84 910.22 820.25 89.97 2093.01 899.53 1193.48 1024.69 168.79 Panel C: Average portfolio size (in millions) Year 1997 1998 1999 2000 2001 2002 2003 2004 All years Average size: all funds Single-manager funds Multiple-manager funds Pure-team funds Mixed-team funds 1645.19 1750.15 1425.55 1475.52 1169.00 1905.01 2071.49 1612.9 1688.67 1245.58 2111.10 2237.66 1924.89 2021.50 1529.52 2090.09 2198.6 1959.54 2168.15 1246.04 1550.71 1481.53 1624.57 1842.84 900.22 1189.99 1195.91 1184.21 1337.43 621.04 1497.83 1481.53 1511.99 1783.15 633.58 1500.36 1527.22 1480.74 1846.28 672.46 1686.29 1743.01 1590.55 1770.44 1002.18 This table presents summary characteristics of the funds included in our dataset. The first row of panel A presents the number of distinct mutual fund portfolios each year after we account for multiple share classes. The rest of panel A shows the number and percentage of single-manager or multiple-manager funds as well as the number and percentage of multiple-manager funds that are pure-team and mixed-team funds. Panels B and C present the total assets managed and average portfolio size by each type of management team respectively. (2001) and Elton et al. (2004), who conclude that many investors make uninformed investment decisions. 2. Data, variables and method We begin our investigation by examining differences in performance between single-manager and multiple-manager funds to compare our results with existing literature. We first obtain management team structure at the beginning of each year t (end of year t − 1) for 1997–2004 and gather and calculate performance statistics for the same periods. We test for the difference in means of the single- and multiple-manager funds for all of our variables for the bull and bear markets combined and individually. Then employing ordinary least squares (OLS) regression, we use the management structure dummy variables and other fund characteristics at year t to explain performance during year t + 1 for the full sample period and the bull and bear market separately. Similar to Chevalier and Ellison (1999), we include instrumental variables in some of our regressions, using lagged observations as proxies for variables that are potentially endogenous. For all of the regressions we estimate clustered standard errors by fund and include as control variables the prospectus objectives and time dummy variables, even if we do not explicitly show them when we present our regression specifications and results. We separately report the performance results for growth-oriented (prospectus objective of growth- I. Karagiannidis / Int. Fin. Markets, Inst. and Money 20 (2010) 197–211 201 and aggressive-growth) and income-oriented funds (prospectus objective of growth-income and equity-income).2 2.1. Mutual fund data Our Morningstar data are from its CD-based Principia Mutual Funds Advanced database.3 We identify all the funds in existence in December 1996 and follow them through December 2004 or until they disappear from the database. Our working sample includes all actively managed domestic equity funds with a self-declared investment objective of growth, aggressive growth, growth-income, or equity-income. Excluded are index funds, balanced funds, funds of funds, as well as other types of funds with investment restrictions such as socially conscious funds, life cycle funds, target retirement funds and tax managed funds. Because some of our variables are lagged one year, we also exclude funds that do not have at least two consecutive years of data. To make ensure that this restriction does not bias our sample towards one type of management structure over the other, we examine the management structure of the funds that drop out. We do not observe any evidence that these funds belong to any particular management team structure. We also exclude fund–years that have obvious reporting errors that cannot be corrected as well as those fund–years that have less than five stocks under management. For each fund we obtain annual and monthly returns, annual expense ratios and loads, net assets, fund inception dates, fund family names (if applicable), as well as portfolio characteristics such as turnover, number of holdings, percentage of assets invested in the top 10 holdings, stock, cash, and bond holdings, and manager names. For most funds a single value for the management fee is provided. For others a minimum and maximum management fee range is given; in these cases we use the midpoint to represent the fund’s management fee. In the “manager name” field, Morningstar lists the name(s) of the manager(s) or the term “Management Team” when there multiple managers and their names are not available.4 We also collect the data on whether the fund is managed by one or more investment advisory finds from the advanced analytics view of the database. The reporting unit used by Principia is the fund share class. However, although the various share classes offer investors freedom to choose how to pay for broker fees, the underlying portfolio and consequently the before-fee performance (gross return) is exactly the same. Thus, because our unit of observation is the fund, we aggregate multiple share classes into one fund observation. To identify different share classes of the same fund, we match them by turnover, number of holdings, percentage invested in stock, and percentage invested in the top 10 holdings. We use fund names to verify our matching.5 We also match our Morningstar sample to the CRSP Survivorship Bias Free U.S. Mutual Fund Database. If a fund is managed by an individual, CRSP reports the manager’s name. If a fund is managed by more than one individual, the database reports manager names, or the terms “team”, “management team”, “committee” or the name of the lead manager while noting that the fund is managed by a team. Of the total 7701 fund–year observations of our Morningstar sample we were able to match 7360 fund–years to CSRP. There were 573 instances where a fund listed as multiple-manager fund in Morningstar shows up as a single-manager fund in CRSP and 729 instances where a fund listed as single-manager fund in Morningstar it appears as a multiple-manager fund in CRSP. The categorization is the same for only 6058 (about 82%) of 7360 fund–years. 2 Ding and Wermers (2005) examine the effect of manager characteristics and report that their findings are significant only for growth fund managers. They suggest that their results might reflect the need for more experience or specialized skills to accurately forecast the earnings of growth stocks. 3 Morningstar, Inc. established the Principia database in January 1996. Throughout our sample period, this database has been named Principia Mutual Funds Plus, Principia Mutual Funds Pro Plus, or Principia Mutual Funds Advanced. 4 The exact description of “Management Team” is: “This is used when there are more than two persons involved in fund management, and they manage together, or when the fund strongly promotes its team-managed aspect”. 5 Multiple share classes of the same fund have basically the same name. Their names differ only by the name of the share class, e.g., Vanguard Growth A, Vanguard Growth B, and so forth. Nanda et al. (2005) document that at the end of 2002 more than 50% of mutual funds offered more than one share class. 202 I. Karagiannidis / Int. Fin. Markets, Inst. and Money 20 (2010) 197–211 2.2. Return and risk variables We measure a fund’s yearly performance using raw annual fund returns, the Sharpe ratio, the 1factor alpha from the market model, and the 4-factor alpha from Carhart’s (1997) expanded market model. We obtain the 1-factor and 4-factor alphas, respectively, by estimating: Rik − Rfk = ˛i + ˇi1 EMRk + εik , (1) Rik − Rfk = ˛i + ˇi1 EMRk + ˇi2 SMBk + ˇi3 HMLk + ˇi4 UMDk + εik , (2) where Rik − Rfk is the month k excess return for fund i with Rfk representing the risk-free rate, EMRk is the excess market return (market return less the risk-free rate), SMBk is the difference in returns between small and big stock portfolios, HMLk is the difference in returns between high and low bookto-market portfolios, and UMDk is the return on a momentum portfolio. To obtain Rik , we use monthly gross fund returns, which we calculate by adding one-twelfth of the annual expense ratio to the monthly net returns. We use gross returns, which unlike net returns are the same for all classes of the same fund, because we want to measure the performance of various management team configurations. If management configurations receive rents through higher expenses, the performance superiority of one configuration over another might not show up in net returns. We use the value-weighted NYSE/AMEX/Nasdaq composite index from Wharton Research Data Services (WRDS) and the one-month T-bill rate from Ibbotson Associates as our risk-free rate to calculate excess market returns. Returns for SMB, HML, and UMD are from Kenneth French’s website (http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/). Our measures of portfolio risk are the betas (ˇi1 ) from Eqs. (1) and (2), which measure 1-factor and 4-factor systematic risk, respectively, and the standard deviation of monthly gross returns realized during the year. 2.3. Typical fund characteristics As we report in Table 2, although there are differences between single- and multiple-manager funds and in our bull and bear markets, the typical mutual fund has in the neighborhood of $1.6 billion under management, involving approximately 100 securities, with the largest 10 holdings accounting for 30–35% total assets. It turns its portfolio almost every year and charges a management fee of 70–75 basis points. The fund is 13–14 years old and there is about a 75% chance that it uses an internal advisor. Single-manager funds on average have better raw returns than both pure-team and mixed-team funds in the bull market period but lower raw returns during the bear market period. The situation is mixed when we focus on other performance measures. Regardless of market conditions and sample choice single-manager funds trade more, hold fewer securities in their portfolio and have more concentrated holdings. 3. Models and results 3.1. Single-manager versus multiple-manager funds We begin our analysis by focusing on performance differences between single-manager and multiple-manager funds to compare our results to prior research. Morningstar and CRSP data are used. We explore differences in performance using difference in means tests and multiple regressions. The latter permits us to control for the influences of other variables. Our performance measures are raw annual returns, Sharpe ratios, 1-factor alphas and 4-factor alphas. The latter two metrics adjust returns for various types of systematic risk while the Sharpe ratio adjusts for total risk. When we present results with the Morningstar single- versus multiple-manager categorization we report results using the full sample of 7701 fund–years and not only for the 7360 fund–years that could be matched to CRSP. However, we do compute all of the statistical tests with the reduced sample and the results are virtually identical. Table 3 presents the tests for differences in the means between single-manager and multiplemanager funds for the full sample and the bull and bear markets, respectively. Table 4 presents the 203 I. Karagiannidis / Int. Fin. Markets, Inst. and Money 20 (2010) 197–211 Table 2 Summary statistics. Panel A: Full sample (1997–2004) Single-manager funds Pure-team funds Mixed-team funds Mean S.D. Mean S.D. Mean S.D. Raw annual returns S.D. of returns Sharpe ratio 1-Factor alpha 4-Factor alpha Management fee (%) Total assets Fund age Portfolio turnover Securities in portfolio Portfolio concentration 9.7528 18.3650 0.5659 0.4541 0.0658 0.7301 1582.1500 13.6777 99.3459 91.8755 33.1523 21.8690 8.6354 1.2377 12.2340 10.5380 0.2437 5208.3600 14.6542 114.1455 80.8200 12.4011 8.3467 17.3670 0.5812 0.2197 -0.1460 0.7436 1625.0900 14.6730 87.8684 106.6995 31.3460 21.6560 8.0921 1.2754 10.9360 9.0798 0.2280 5006.6900 15.2811 83.2187 141.7611 12.0758 8.1550 16.3150 0.6231 -0.2780 -0.0570 0.7389 831.7103 9.7118 89.7500 117.1951 29.5099 20.4650 7.3796 1.2596 9.3368 7.8396 0.2110 2229.2200 10.9803 65.0354 151.1540 10.8169 N (fund–years) 4056 2789 856 Single-manager funds Pure-team funds Mixed-team funds Mean Mean Panel B: Bull market (1997–2000) Mean S.D. Raw annual returns S.D. of returns Sharpe ratio 1-Factor alpha 4-Factor alpha Management fee (%) Total assets Fund age Portfolio turnover Securities in portfolio Portfolio concentration 17.8700 20.3840 0.7095 2.0694 1.9137 0.7125 1746.0200 13.9783 92.4863 90.7603 34.3575 17.9200 8.6106 0.8692 14.4600 12.2660 0.2342 5712.6700 15.0826 97.1939 78.0187 12.4536 N (fund–years) 1902 16.0600 20.2700 0.6212 2.1667 2.4605 0.7221 1627.9100 15.5622 80.8837 95.3643 33.4496 S.D. 17.7940 8.1456 0.8557 13.6950 11.0930 0.2162 4579.1000 15.8735 71.6745 95.5212 12.4357 980 14.4690 20.2370 0.5282 2.7968 2.8615 0.6928 1163.2400 11.7511 90.8263 109.7034 30.6188 S.D. 18.1670 7.7673 0.8616 12.5270 9.8760 0.1999 2594.6200 13.3239 59.6678 110.9248 9.9113 236 Panel C: Bear market (2001–2004) Single-manager funds Pure-team funds Mixed-team funds Mean S.D. Mean S.D. Mean S.D. Raw annual returns S.D. of returns Sharpe ratio 1-Factor alpha 4-Factor alpha Management fee (%) Total sssets Fund age Portfolio turnover Securities in portfolio Portfolio concentration 2.5855 16.5830 0.4391 −0.9720 −1.5660 0.7457 1437.4500 13.4123 105.4030 92.8603 32.0881 22.5310 8.2587 1.4778 9.6398 8.4036 0.2508 4715.0900 14.2635 126.9653 83.2209 12.2594 4.1680 15.7940 0.5596 −0.8350 −1.5590 0.7552 1623.5600 14.1913 91.6523 112.8402 30.2065 22.4060 7.6160 1.4527 8.9280 7.4042 0.2334 5224.9700 14.9328 88.6378 161.0622 11.7231 5.7515 14.8210 0.7680 −0.8590 −1.8940 0.7565 705.5152 8.9356 89.3403 120.0468 29.0878 20.7900 6.6496 1.3802 7.4739 6.4513 0.2126 2061.6500 9.8463 67.0071 163.8828 11.1209 N (fund–years) 2154 1809 620 This table presents summary statistics by management team structure for all 7701 fund–year observations in the dataset. Variables are as defined in Section 2. Panel A presents summary statistics for the full sample period while panels B and C present the same statistics doe the 1997–2000 and 2001–2004 periods respectively. 204 Table 3 Performance t-tests (CRSP versus Morningstar). Full sample (1997–2004) Bull market (1997–2000) Bear market (2001–2004) MSTAR difference in means (single-multiple) CRSP difference in means (single-multiple) MSTAR difference in means (single-multiple) CRSP difference in means (single-multiple) MSTAR difference in means (single-multiple) 1.1628** (0.0221) −0.0880* (0.0029) 0.2143 (0.4265) 0.0391 (0.8642) 1.4511* (0.0033) −0.0250 (0.3802) 0.3514 (0.1767) 0.3147 (0.1537) 1.4281** (0.0334) 0.0635*** (0.0505) −0.7920 (0.1313) −1.3060* (0.0027) 2.1181* (0.0013) 0.1063* (0.0008) −0.2200 (0.6662) −0.6250 (0.1368) −3.1100* (0.0001) −0.2440* (0.0001) −0.0740 (0.7961) −0.2440 (0.3200) −1.9870* (0.0026) −0.1460* (0.0007) 0.0196 (0.9424) 0.0783 (0.7362) 7360 7701 3001 3118 4359 4583 Panel B: Growth-oriented funds Raw annual returns 1.3106** (0.0457) Sharpe ratio −0.0820** (0.0229) 1-Factor alpha 0.6298*** (0.0747) 4-Factor alpha 0.2180 (0.4684) 1.5773** (0.0131) −0.0230 (0.5105) 0.6056*** (0.0749) 0.4987*** (0.0841) 1.0434 (0.2642) 0.0369 (0.3751) −1.2590*** (0.0843) −1.5890* (0.0096) 2.0488** (0.0256) 0.0889** (0.0289) −0.3390 (0.6316) −0.5420 (0.3554) −2.7820* (0.0010) −0.2090* (0.0001) 0.4045 (0.2634) −0.0280 (0.9287) −1.7770** (0.0283) −0.1240** (0.0153) 0.1729 (0.6117) 0.1727 (0.5562) N (fund–years) 5415 1962 2039 3199 3376 Panel C: Income-oriented funds Raw annual returns 0.8083 (0.2634) Sharpe ratio −0.1000** (0.0431) 1-Factor alpha −0.7640** (0.0304) 4-Factor alpha −0.3570 (0.2250) 1.1429 (0.1050) −0.0300 (0.5354) −0.2560 (0.4583) −0.0810 (0.7795) 1.8303** (0.0228) 0.1086** (0.0355) −0.2000 (0.7501) −0.8630*** (0.0888) 1.9592** (0.0123) 0.1348* (0.0072) −0.2500 (0.6823) −0.8410*** (0.0936) −4.0230* (0.0003) −0.3420* (0.0001) −1.2810* (0.0009) −0.7410** (0.0214) −2.5720** (0.0160) −0.2090* (0.0098) −0.2460 (0.5137) −0.0560 (0.8561) N (fund–years) 2286 1039 1079 1160 1207 Panel A: All funds Raw annual returns Sharpe ratio 1-Factor alpha 4-Factor alpha N (fund–years) 5161 2199 This table presents tests for differences in the means between single-manager and multiple-manager funds using the Morningstar and CRSP classification. Panel A presents results for all funds while panels B and C present results for growth-oriented and income-oriented funds respectively. p-Values appear in parentheses below the coefficients. * Indicate significant at the 1% respectively. ** *** Indicate significant at the 5% respectively. Indicate significant at the 10% respectively. I. Karagiannidis / Int. Fin. Markets, Inst. and Money 20 (2010) 197–211 CRSP difference in means (single-multiple) Table 4 Performance regressions (CRSP versus Morningstar). Full sample (1997–2004) Bull market (1997–2000) Bear market (2001–2004) MSTAR dummy coefficient (multiple = 1) CRSP dummy coefficient (multiple = 1) MSTAR dummy coefficient (multiple = 1) CRSP dummy coefficient (multiple = 1) MSTAR dummy coefficient (multiple = 1) 0.2864 (0.3447) 0.0242 (0.1075) 0.0390 (0.8859) 0.1324 (0.5621) −0.0234 (0.9381) −0.0051 (0.7265) −0.4728*** (0.0707) −0.3979*** (0.0791) 0.4026 (0.5401) 0.0382 (0.1598) 0.4877 (0.3733) 0.8241*** (0.0844) −0.4458 (0.4923) −0.0060 (0.8247) −0.6334 (0.2416) −0.1214 (0.7963) 0.0314 (0.9108) 0.0096 (0.5511) −0.2671 (0.3572) −0.2529 (0.3106) 0.0639 (0.8135) −0.0124 (0.4313) −0.4685*** (0.0913) −0.5783** (0.0131) 5618 5736 2014 2060 3604 3676 Panel B: Growth-oriented funds Raw annual returns 0.0846 (0.8310) Sharpe ratio 0.0085 (0.6447) 1-Factor alpha −0.3118 (0.3884) 4-Factor alpha −0.0621 (0.8384) −0.1126 (0.7738) −0.0104 (0.5548) −0.6819** (0.0470) −0.5794*** (0.0519) 0.9781 (0.2872) 0.0609*** (0.0814) 0.7092 (0.3755) 1.1011 (0.1115) −0.3452 (0.7024) −0.0058 (0.8682) −0.7159 (0.3558) −0.0633 (0.9225) −0.4547 (0.2078) −0.0243 (0.2126) −0.8975** (0.0152) −0.5857*** (0.0691) −0.0498 (0.8873) −0.0189 (0.3172) −0.6859*** (0.0511) −0.7711* (0.0100) N (fund–years) 3965 1303 1329 2581 2636 Panel C: Income-oriented funds Raw annual returns 0.7412** (0.0453) Sharpe ratio 0.0620* (0.0069) 1-Factor alpha 0.7770** (0.0202) 4-Factor alpha 0.6002** (0.0436) 0.1881 (0.6026) 0.0119 (0.5959) 0.0658 (0.8369) 0.0102 (0.9724) −0.0012 (0.9986) 0.0120 (0.7733) 0.3379 (0.5803) 0.6596 (0.2464) −0.0656 (0.9285) 0.0128 (0.7639) −0.0506 (0.9363) 0.0869 (0.8846) 0.8304** (0.0153) 0.0782* (0.0020) 1.0041* (0.0052) 0.5502*** (0.0723) 0.2277 (0.4877) 0.0024 (0.9212) 0.1092 (0.7502) −0.0761 (0.7973) N (fund–years) 1771 711 731 1023 1040 Panel A: All funds Raw annual returns Sharpe ratio 1-Factor alpha 4-Factor alpha N (fund–years) 3884 1734 *** Indicate significant at the 10% respectively. 205 This table presents tests results from regression (3) using the Morningstar and CRSP classification. Only the coefficient of the multiple-manager dummy is reported and p-values appear in parentheses below the coefficients. Panel A presents results for all funds while panels B and C present results for growth-oriented and income-oriented funds respectively. * Indicate significant at the 1% respectively. ** Indicate significant at the 5% respectively. I. Karagiannidis / Int. Fin. Markets, Inst. and Money 20 (2010) 197–211 CRSP dummy coefficient (multiple = 1) 206 I. Karagiannidis / Int. Fin. Markets, Inst. and Money 20 (2010) 197–211 regression results. Based on Ding and Wermers (2005) observation that growth fund managers behave differently than other fund managers we split our sample of funds into (a) aggressive growth and growth (AG&G) and (b) growth and income and income (GI&I). We focus our discussion on the AG&G and GI&I funds and the bull and bear markets. Turning first to the Morningstar results in Table 3 if raw returns or the Sharpe ratio are our performance metric single-manager AG&G and GI&I funds outperform their multiple-manager counterparts for the bull market. However, in the bear market the opposite occurs with the multiple-manager funds outperforming the single-manager funds. If we use the 1-factor alpha or the 4-factor alpha as our metric, neither management structure is superior for either market and for either fund style. Results are qualitatively the same between Morningstar and CRSP for raw returns and the Sharpe ratio. However, for CRSP we find significant underperformance of multiple-manager funds in the bull market for AG&G funds equal to 158 basis points in terms of 4-factor alphas. For GI&I funds we find that multiple-manager funds underperform their single-manager counterparts in terms of 1-factor and 4-factor alphas in the bear market. To delve deeper into the differences in performance (Perf) between single- and team-managed funds, we estimate the following model for each of the four performance metrics: Perfi,t = a + b1 MMi,t−1 + b2 Perfi,t−1 + b3 Turnoveri,t−1 + b4 MgtFeei,t−1 + b5 LogAssetst + b6 FundAgei,t−1 + εt (3) where MM is a dummy variable that takes the value of one if the fund is team-managed and zero otherwise at the end of year t − 1, MgtFee is the management fee in percent charged by the management company at the end of year t − 1, LogAssets is the logarithm of the fund’s total assets at the end of year t − 1, FundAge is the fund’s age in years at time t − 1. We add lagged performance (Perf) and portfolio turnover (Turnover) to the list of independent variables to capture the possibilities of earnings persistence and portfolio churning, respectively. For the total sample, the prospectus objective and time dummy variables are included in the regressions. For the AD&G and GI&I funds only the time dummy variables are included but as before their coefficients are not reported. We only report the multiple managers dummy coefficients from all regressions to make results easily comparable. As we show in Table 4, looking at the Morningstar results, the form of management structure is not significantly related to performance no matter how measured in the bull market period for AG&G and GI&I funds. This lack of relevance also holds for the GI&I funds in the bear market. In contrast, in the case of AG&G funds the underperformance of multiple-manager funds amounts to 68.6 basis points annually for the 1-factor alpha and 77.1 basis points for the 4-factor alpha. When we look at the same regression results using the CRSP categorization we find significant results for GI&I funds in the bear market. Multiple-manager funds over perform by 83 basis points in terms of raw returns and approximately 100 basis points in terms of the 1-factor alpha when compared to their single-manager counterparts. The only significant results for AG&G funds when using the CRSP categorization is that multiple-manager funds underperform by 89.8 basis points in the bear market period. In sum, the results from Tables 3 and 4 are mixed. There is some evidence that single-manager funds perform at least as well or better than multiple-manager funds. That they provide superior performance depends on the performance measure, the fund’s style and the overall market environment. The results also depend on whether the management structure is derived from information provided by Morningstar or CRSP. 3.2. Single-manager versus pure-team funds In this section we ignore mixed-team funds and consider only single-manager and pure-team funds. This reduces our sample to 5150 fund–year observations. We estimate the following regressions for each of our four performance metrics: Perfi,t = a + b1 PTi,t−1 + b2 Perfi,t−1 + b3 Turnoveri,t−1 + b4 MgtFeei,t−1 + b5 LogAssetst + b6 FundAgei,t−1 + εt (4) I. Karagiannidis / Int. Fin. Markets, Inst. and Money 20 (2010) 197–211 207 Table 5 Performance single-manager versus pure-team funds. Full sample (1997–2004) Dummy coefficient (team = 1) Bull market (1997–2000) Dummy Coefficient (team = 1) Bear market (2001–2004) Dummy coefficient (team = 1) 0.0610 (0.8518) 0.0034 (0.8252) −0.3720 (0.1858) −0.2447 (0.3116) −0.4547 (0.5173) −0.0044 (0.8791) −0.6455 (0.2707) −0.0512 (0.9199) 0.1985 (0.4974) 0.0009 (0.9553) −0.2499 (0.3964) −0.3471 (0.1571) 5150 1903 3247 Panel B: Growth-oriented funds Raw annual returns −0.0706 (0.8668) Sharpe ratio −0.0034 (0.8575) 1-Factor alpha −0.5811 (0.111) 4-Factor alpha −0.4191 (0.1813) −0.4576 (0.6401) −0.0115 (0.7544) −0.8091 (0.3358) 0.0249 (0.9716) 0.0498 (0.8936) −0.0057 (0.7745) −0.4590 (0.2123) −0.5495*** (0.0776) N (fund–years) 1241 2352 Panel C: Income-oriented funds Raw annual returns 0.5010 (0.1962) Sharpe ratio 0.0310 (0.2162) 1-Factor alpha 0.2716 (0.4436) 4-Factor alpha 0.1683 (0.6045) 0.2351 (0.7538) 0.0353 (0.4365) 0.1299 (0.8452) 0.1009 (0.8770) 0.4940 (0.1978) 0.0178 (0.5337) 0.3278 (0.4084) 0.1740 (0.5983) N (fund–years) 662 895 Panel A: All funds Raw annual returns Sharpe ratio 1-Factor alpha 4-Factor alpha N (fund–years) 3593 1557 This table presents tests results from regression (4) using the Morningstar classification. Only the coefficient of the pure-team dummy is reported and p-values appear in parentheses below the coefficients. Panel A presents results for all funds while panels B and C present results for growth-oriented and income-oriented funds respectively. *** Indicate significant at the 10% respectively. where PT is a dummy variable that takes the value of one if the fund is managed by a pure team and zero otherwise at the end of year t − 1, MgtFee is the management fee in percent charged by the management company at the end of year t − 1, LogAssets is the logarithm of the fund’s total assets at the end of year t − 1, FundAge is the fund’s age in years at time t − 1. We add lagged performance (Perf) and portfolio turnover (Turnover) to the list of explanatory variables to capture the possibilities of earnings persistence and portfolio churning, respectively. Prospectus objective and time dummy variables are included in the regressions but their coefficients are not reported. We report our regression results in Table 5. We do not find any evidence of underperformance or over performance of any one management structure in any of the three periods or for either of the two fund styles. These results suggest that the underperformance of multiple-manager funds is due to the presence of more than one investment advisors (mixed teams) and not due to the type of management structure (single managers versus teams). 3.3. Pure-team versus mixed-team funds To supplement our findings from Sections 3.1 and 3.2 we estimate another set of performance regressions. This time we focus only on multiple-manager funds 5736 fund–year observations, and 208 I. Karagiannidis / Int. Fin. Markets, Inst. and Money 20 (2010) 197–211 Table 6 Performance pure-team versus mixed-team funds. Full sample (1997–2004) Dummy coefficient (multiple = 1) Bull market (1997–2000) Dummy coefficient (multiple = 1) Bear market (2001–2004) Dummy coefficient (multiple = 1) −0.3984 (0.3523) −0.0400*** (0.0624) −0.7198** (0.041) −0.9019* (0.0026) −0.3610 (0.7326) −0.0149 (0.7476) −0.4440 (0.5955) −0.4516 (0.5409) −0.4813 (0.1955) −0.0580** (0.0105) −1.0925* (0.0035) −1.2309* (0.0002) 5736 2060 3676 Panel B: Growth-oriented funds Raw annual returns −0.2751 (0.6421) Sharpe ratio −0.0408 (0.1533) 1-Factor alpha −0.8984*** (0.0684) 4-Factor alpha −1.0836* (0.0098) 0.2840 (0.8604) 0.0282 (0.6744) −0.0282 (0.9829) −0.4526 (0.6751) −0.4107 (0.4134) −0.0663** (0.0249) −1.3178* (0.0084) −1.3716* (0.0023) N (fund–years) 1329 2636 Panel C: Income-oriented funds Raw annual returns −0.9866** (0.0461) Sharpe ratio −0.0617** (0.0188) 1-Factor alpha −0.7006*** (0.0640) 4-Factor alpha −0.6219*** (0.0901) −1.3553 (0.2396) −0.0853 (0.1708) −0.8364 (0.3904) −0.1434 (0.8777) −0.6671*** (0.0955) −0.0499*** (0.0802) −0.6432*** (0.0994) −0.9182* (0.0097) N (fund–years) 731 1040 Panel A: All funds Raw annual returns Sharpe ratio 1-Factor alpha 4-Factor alpha N (fund–years) 3965 1771 This table presents tests results from regression (5) using the Morningstar classification. Only the coefficient of the pure-team dummy is reported and p-values appear in parentheses below the coefficients. Panel A presents results for all funds while panels B and C present results for growth-oriented and income-oriented funds respectively. * Indicate significant at the 1% respectively. ** Indicate significant at the 5% respectively. *** Indicate significant at the 10% respectively. use the following model: Perfi,t = a + b1 MTi,t−1 + b2 Perfi,t−1 + b3 Turnoveri,t−1 + b4 MgtFeei,t−1 + b5 LogAssetst +b6 FundAgei,t−1 + εt (5) where MT is a dummy variable that takes the value of one if the fund is managed by a mixed team with an unknown structure and zero if the fund is managed by a pure team at the end of year t − 1, MgtFee is the management fee in percent charged by the management company at the end of year t − 1, LogAssets is the logarithm of the fund’s total assets at the and end of year t − 1, FundAge is the fund’s age in years at time t − 1. We add lagged performance (Perf) and portfolio turnover (Turnover) to the list of independent variables to capture the possibilities of earnings persistence and portfolio churning, respectively. Prospectus objective and time dummy variables are included in the regressions but once again their coefficients are not reported. I. Karagiannidis / Int. Fin. Markets, Inst. and Money 20 (2010) 197–211 209 We report in Table 6 that there are highly significant differences between pure and mixed teams. Mixed teams underperform pure teams by 72 1-factor alpha basis points or 90 4-factor alpha basis points in the full sample period. Most of the underperformance is present among growth funds (108 basis points in terms of 4-factor alphas). There are no significance performance differences during the bull market period. In the bear market period, however, we find that both growth funds and income funds managed by mixed teams underperform by 137 and 92 basis points, respectively, in terms of 4-factor alphas. 3.4. Robustness checks We perform several robustness checks. First, we repeat all analysis of the using only the funds that have the same management structure in Morningstar and CRSP. Our results are unchanged. Second, there are some single-manager funds that list multiple advisors. We compare these funds to single-manager funds with only one advisor. We find that single-manager funds with multiple advisors do not do worse than single-manager funds with one advisor. This means that mixed teams are the worse performance category. Third, we compare single-manager funds with one advisor and multiple-manager funds with one advisor (pure teams) and do not find any difference in performance. Fourth, we run regressions (3) and (4) by including two additional variables to account for family effects. More specifically, we include fund family size (log of total assets managed by the fund family) and fund family management policy (the percentage of funds in the family managed by teams). Our major results in Tables 4 and 5 remain unaffected. Finally, similar to Han et al. (2008), we use Heckman’s method to control for possible endogeneity in our performance regressions. We regress the multiple-manager (or pure-team) dummy variable, which takes the value of one if the fund is a multiple-manager (or pure-team) fund and zero otherwise, on several mutual fund characteristics. From this probit regression we obtain the inverse Mill’s ratio for each fund and use this ratio as an additional explanatory variable in our performance regressions. In most instances the inverse Mill’s ratio is insignificant indicating that the coefficients in our original regressions are unbiased. In the few times where the inverse ratio is significant, the multiplemanager (or pure-team) dummy coefficient remains significant with qualitatively the same sign and size. 4. Concluding remarks We examine the effect of the portfolio management team structure on mutual fund portfolio performance. Similar to several previous studies we document that multiple-manager funds underperform their single-manager counterparts in terms risk-adjusted returns in the 2001–2004 bear market but not in the 1997–2000 bull market for growth-oriented but not for income-oriented funds. We conjecture, however, that this does not mean that team management structure is inferior to the single-manager approach. We investigate our conjecture by distinguishing between pure-team funds (single advisor with multiple managers) and mixed-team funds (multiple managers and multiple advisors) whose structure detail is unknown. We find that the underperformance comes from mixed-team funds suggesting that there are no differences in performance between single managers and teams. In addition, we document that there are differences in management team structure reporting between the CRSP Mutual Fund Database and Morningstar and these differences may yield contradicting results. Our results lead to a question: Why have team-manager funds thrived and grown in number at the market share expense of single-manager funds, when single-manager funds have performed better or at least not worse? A plausible explanation involves uninformed investors and industry selfpromotion. As Gruber (1996) points out, sophisticated mutual fund investors make decisions based on expected performance while their unsophisticated counterparts make decisions based on advertising and various types of advice. Huberman (2001) concludes that investors are not objective with respect to risk-return trade-offs, are disposed to invest in a company that they know (or think that 210 I. Karagiannidis / Int. Fin. Markets, Inst. and Money 20 (2010) 197–211 they know), and tend to favor stocks that receive positive coverage by the media.6 Moreover, empirical evidence suggests that a large portion of mutual fund investors are unsophisticated and most likely uninformed.7 Further, anecdotal evidence provided by investment professionals indicates that the fear of losing investors following the departure of a successful manager is an important reason for a fund’s board of directors adopting a team-manager approach.8 Thus, many mutual funds have actively promoted the potential benefits of the teams, especially consistency in performance brought about by the stability of management. It may well be that the emergence of team-manager mutual funds is simply the result of effective advertising and public relations efforts by the industry. As Elton et al. (2004) assert in the context of index funds, poorly performing funds survive because of the presence of uninformed investors and the willingness of distributors to sell these funds. Acknowledgements We thank Charles J. Hadlock, Jun-Koo Kang and Stephen Dimmock for helpful comments and suggestions, and William Alsover, Paul Cook, Jeffrey deGraaf, Jay Keranen, John Koczara, Patrick Lynch and Patrick Sahm for their institutional insights. All errors remain our own. References Bär, M., Alexander, K., Ruenzi, S., 2005. Team Management and Mutual Funds, Working Paper. Centre for Financial Research (CFR), Cologne, Germany. Barry, C., Starks, L., 1984. Investment management and risk sharing with multiple managers. Journal of Finance 39, 477–491. Carhart, M., 1997. On the persistence of mutual fund performance. Journal of Finance 52, 57–82. Carnahan, I., 2004. The other side of Vanguard: the fund firm synonymous with indexed funds turns out to be a nifty active money manager. Who knew? Forbes (December), 190–192. Chen, J., Harrison, H., Ming, H., Kubik, J., 2004. Does fund size erode mutual fund performance? The role of liquidity and organization. American Economic Review 94, 1276–1302. Chevalier, J., Ellison, G., 1999. Are some mutual fund managers better than others? Cross-sectional patterns in behavior and performance. Journal of Finance 54, 875–899. Ding, B., Wermers, R., 2005. Mutual fund performance and governance structure: the role of portfolio managers and boards of directors. In: American Finance Association 2006 Annual Meeting, Boston, MA. Elton, E.J., Gruber, M.J., Blake, C.R., 2001. A first look at the accuracy of the CRSP mutual fund database and a comparison of the CRSP and the Morningstar mutual fund database. Journal of Finance 56, 2415–2430. Elton, E.J., Gruber, M.J., Busse, J.A., 2004. Are investors rational? Choices among index funds. Journal of Finance 59, 261–288. Grinblatt, M., Keloharju, M., 2001. How distance, language and culture affect stockholdings and trades. Journal of Finance 56, 1053–1073. Gruber, M.J, 1996. Another puzzle: the growth of actively managed mutual funds. Journal of Finance 51, 783–810. Han, Y., Noe, T., Rebello, M., 2008. Horses for courses: fund managers and organizational structures. Working Paper, Available at SSRN http://ssrn.com/abstract=1100676. Hill, G., 1982. Group vs. individual performance: are N + 1 heads better than one? Psychological Bulletin 91, 517–539. Herrenkohl, R., 2004. Becoming a Team. Cincinnati, OH, South-Western. Holmstrom, B., 1982. Moral hazard in teams. The Bell Journal of Economics 13, 324–340. Huberman, G., 2001. Familiarity breeds investment. Review of Financial Studies 14, 659–680. Investment Company Institute (ICI), 2005a. 2005 Investment Company Fact Book. Investment Company Institute, Washington, DC. Investment Company Institute (ICI), 2005b. Ownership of mutual funds through professional advisors. Fundamentals 14 (April (3)), 1–4. Investment Company Institute (ICI), 2006. Understanding Investor Preferences for Mutual Fund Information. Investment Company Institute, Washington, DC. 6 The importance of familiarity in investment decisions has been well documented. See, for example Grinblatt and Keloharju (2001), Ivković and Weisbenner (2005, 2007). 7 For example, ICI (2007) reports that only 13% of mutual fund assets are owned by institutional investors and 53% of these are held in money market accounts with the remainder residing in stock, bond, and hybrid accounts. Furthermore, only 14% of investors invest their wealth in non-retirement accounts are considered to be self-sufficient (ICI, 2005b), and, most importantly, only about 25% of mutual fund investors seek information about a fund’s portfolio manager before making an investment decision (ICI, 2006). 8 A case in point is Elizabeth Branwell’s departure from Mario Gabelli’s GAMCO. Under Ms. Branwell’s leadership the Gambelli Growth Fund became one of the top performers in its style class. In 1994 she left the fund, founded her own company, and convinced the Securities and Exchange Commission to let her use her track record with Gambelli to help her get started. Many of Ms. Branwell’s former clients followed her, and Gabelli took her to arbitration but lost. I. Karagiannidis / Int. Fin. Markets, Inst. and Money 20 (2010) 197–211 211 Investment Company Institute (ICI), 2007. 2007 Investment Company Fact Book. Investment Company Institute, Washington, DC. Ivković, Z., Weisbenner, S., 2005. Local is as local does: information content of the geography of individual investors’ common stock investments. Journal of Finance 60, 267–306. Ivković, Z., Weisbenner, S., 2007. Information diffusion effects in individual investors’ common stock purchases: covet thy neighbors’ investment choices. Review of Financial Studies 20, 1327–1357. Janis, I., 1982. Groupthink: A Psychological Study of Policy Decisions and Fiascoes. Houghton Mifflin Company, Boston, MA. Kuhnen, C., 2004. Dynamic Contracting in the Mutual Fund Industry, Working Paper. Stanford Graduate School of Business, Palo Alto, CA. Massa, M., Reuter, J., Zitzewitz, E., 2006. The Rise of Anonymous Teams in Fund Management, Working Paper. INSEAD, Fountainbleu, France. Nanda, V.K., Wang, J.Z., Zheng, L., 2005. The ABC’s of mutual funds: a natural experiment on fund flows and performance. In: American Finance Association 2005 Annual Meeting, Philadelphia, PA. Prather, L., Middleton, K., 2002. Are N + 1 heads better than one? The case of mutual fund managers. Journal of Economic Behavior & Organization 47, 103–120. Qiu, J., 2003. Termination risk, multiple managers and mutual fund tournaments. European Finance Review 7, 161–190. Rasmusen, E, 1987. Moral hazard in risk-averse teams. The RAND Journal of Economics 18, 428–435. Sah, R., Stiglitz, J., 1988. Committees, hierarchies and polyarchies. The Economic Journal 98, 451–470. Sharpe, W, 1981. Decentralized investment management. Journal of Finance 36, 217–234. Stock, R., 2004. Drivers of team performance: what do we know and what have we still to learn? Schmalenbach Business Review 56, 274–306.