1+1=2? Evidence from Solo- and Team- Managed Mutual Funds

Transcription

1 1+1=2? Evidence from Solo- and Team- Managed Mutual Funds Yanfei Sun * and Jitka Hilliard Auburn University September 2016 * Department of Finance, Auburn University, 309 Lowder Business Building, Auburn, AL Tel: yzs0038@auburn.edu. Department of Finance, Auburn University, 313 Lowder Business Building, Auburn, AL Tel: jzh0023@auburn.edu.

2 Abstract In this paper, we use Morningstar Direct (MD) mutual fund data to examine the effect of manager structure on the mutual fund performance. By analyzing the actively managed and broadly diversified domestic equity funds from year 1990 to 2016, we find that team-managed funds have higher abnormal returns than the solo-managed funds. This superior performance comes mainly from the funds with three- and four-managers and funds with investment strategy categorized as Growth & Equity Income. Furthermore, we explore the reason of changing management structure. And we find managers have poor past performance are more likely to be fired. And the past performance only contributions the firing fund managers, and has no influence on hiring more managers. In addition, we examine the post-change effect, which turns out varies based on the fund past year return. For funds with negative past year return, the change in managers can improve its post-change performance, while for funds with positive past year return, the change has no influence on the post-change performance. 1

3 I. Introduction During recent three decades, mutual funds industry has successfully captured investors attention and undergone a rapid expansion. At the end of 2015, the U.S. mutual fund industry managed nearly $15.7 trillion assets. This is 15 times more assets under management than in 1990 (2016 Investment Company Fact Book 1 ). Although a lot of research on mutual fund performance has been done, the question how to choose a good fund remains largely unanswered. Investors use observable criteria, such as the past performance of the fund, the manager s working experience, and so on to help them in their choice of a mutual fund. In this paper, we examine the mutual fund management structure in context of the fund performance. More specifically, we test whether the number of managers matters and whether the change of the number of managers can affect the fund performance. Based on the management structure, there are two types of mutual funds: solo-managed funds and team-managed funds. Nowadays, more than three-quarters of US mutual funds (6,219 out of 8,072) have at least two managers. 2 Since managers make investment decisions based on their personal abilities and risk preferences, it is reasonable to assume that funds with different management structure may have different investment styles, risk levels and performance. Solo-managed fund is bounded by the manager s time and energy. Team-managed fund, on the other hand, does not face the same constraints. Composed from managers with different talents, experience and expertise, team-managed funds can access more resources and have advantage in 1 See 2016 Investment Company Fact Book, p.172, 2 See how solo star fund managers stack up against the team players, b387-64ab0a67014c.html#axzz4jpq3fgwk. 2

4 risk monitoring. In theory, when abilities of multiple managers are combined together to manage a specific fund, their aggregate expertise should be greater than the sum of the individual parts. At the same time, some researchers have criticized the performance of team-managed funds pointing toward the free rider and agency problems 3. In this paper, we examine the effect of management structure on the mutual fund performance. We ask two questions: (1) Does management structure influence fund performance? (2) Do changes in management structure have influence on future fund performance? We use Morningstar Direct mutual fund data from 1990 to We limit our analysis to actively managed and broadly diversified domestic equity funds. First, we compare the performance of solo- and team-managed funds. We find that team-managed funds have higher abnormal returns than the solo-managed funds. This superior performance comes mainly from the funds with threeand four-managers and funds with investment strategy categorized as Growth & Equity Income. This group of funds invests more in dividend paying stocks than other equity categories used in this study (Aggressive Growth and Growth) and relies less on soft information. This finding is consistent with Stein (2002) argument that the team structure impedes the transition of soft information. In the second part of paper, we explore the reason of changing management structure. We find that funds with poor past performance are more likely to decrease the number of fund managers. The past performance has no influence on hiring more managers. In addition, we examine the postchange effect and find that the effect varies based on the fund past year return. For funds with 3 See Can Multi-Manager Funds Outperform? 3

5 negative past year return, the change in managers can improve its post-change performance, while for funds with positive past year return, the change has no influence on the post-change performance. The rest of the paper is organized as follows. In Section II, we briefly review the related literature and develop our hypotheses. Section III introduces the data source and the methodology, Section IV reports the main results of our study and Section V concludes the paper. II. Literature Review The research on mutual fund management structure can be classified into two categories: (1) comparison of solo- and team-managed funds, and (2) explanation of changes in management structure. 1. Comparison of Solo- and Team-Managed Mutual Funds A comparison of performance between an individual person and a group is a hot topic in Psychology. Hill (1982) states that the group performance is generally qualitatively and quantitatively superior to the performance of the average individual. 4 He argues that this is due to the aggregation of member resources and better ability to assemble and integrate pieces of information to form a solution. Sah and Stiglitz (1986) also argue that the team decisions should reflect a compromise among the decisions of each member. Having different managers in a group can diversify opinions and eventually lead to less extreme decision (Baer, Kempf and Ruenzi, 2010). Furthermore, team members can correct other member s errors in the process of team 4 He also claims that Group performance, however, was often inferior to that of the best individual in a statistical aggregate and often inferior to the potential suggested in a statistical pooling model. 4

6 deliberation (Shaw, 1932, Sharpe, 1981). Teams may also benefit from a broader resource of knowledge and capabilities, particularly when these capabilities are complementary. Based on these arguments, team-managed funds should outperform the solo-managed funds. There are also other theories that argue a very different point. For example, group shift hypothesis (Moscovici and Zavalloni, 1969; Hogg et al., 1990; Kerr, 1992) suggests that the opinion of team members shifts towards the opinion of the dominant person in a team, which makes the final decision no difference from the decision of the dominant individual. Massa, Reuter and Zitzewitz (2010) argue that team-managed funds may suffer from a free rider problem. The performance of an individual in a team-managed fund cannot be separately observed and therefore he/she may become a free-rider. This may lead to the underperformance of the team-managed funds. Another argument against the superior performance of team-managed funds is offered by Stein (2002). He argues that multiple managers structures may impede information transmission, especially for soft information. Managers may end up expending too much research effort on quantitative measures of a company to convince others to implement their ideas. Empirical evidence on the benefits (or handicaps) of team-managed funds is not conclusive and no unified conclusion has been reached. In following paragraphs, we present current empirical findings that compare the performance of solo- and team-managed mutual funds. Studies that suggest inferior performance of the team-managed mutual funds include studies of Chen, Hong, Huang, and Kubik (2004), Stein (2002), Goldman, Sun and Zhou (2016), Baer, Kempf and Ruenzi (2005), Massa, Reuter and Zitzewitz (2010), and other. Chen, Hong, Huang, and Kubik (2004) investigate the effect of asset scale on performance in the US equity mutual funds. They compare the performance of solo- and team-managed funds and find that team- 5

7 managed funds significantly underperform the solo-managed funds. They attribute this finding to the organizational diseconomies that make soft information hard to communicate among all the members (Stein, 2002). Goldman, Sun and Zhou (2016) hold the same view and claim that solomanaged funds have much more concentrated portfolios, tend to perform better, and have higher expense ratios than funds managed by multiple managers. Baer, Kempf and Ruenzi (2005) find that the team-managed funds have a slightly lower return than the solo-managed funds, but also have less risk and can attract higher inflows than solo-managed funds. Massa, Reuter and Zitzewitz (2010) find that solo-managed funds outperform team-managed funds. They argue that this is because solo-managed funds have named managers who receive more media attention than rather anonymous managers from team-managed funds. 5 On the other hand, other studies find that there is no difference between the solo- and teammanaged funds. For example, Prather and Middleton (2002) find no appreciable difference between the outcomes of team-managed and solo-manager funds. Bliss, Potter, and Schwarz (2008) find that in comparison to solo-managed funds, team-managed funds have similar returns, but less risk, lower turnover and lower costs. Therefore, they attract significantly greater flows than solomanaged funds. Karagiannidis (2010) also reports no difference in performance between the soloand team-managed funds using several performance measures. 6 Yet, other studies find that team-managed mutual funds generate better performance than their solo-managed counterparts. Han, Noe and Rebello (2012) claim that high ability managers 5 This issue has been eliminating after the SEC s 2004 release of Disclosure Regarding Portfolio Managers of Registered Management Investment Companies, 17 CFR Parts 239, 249, 270, and 274, Release No ; ; IC-26533; File No. S funds are obliged to disclose the identities of all team members. 6 He finds that mixed-team funds (multiple managers and multiple advisors) are worse when compare with the performance of pure teams, during in the bear market for growth-oriented funds. 6

8 rationally self-select into solo-managed funds. Controlling for the managerial self-selection bias, he finds that the team-managed funds perform better, allocate funds more conservatively, and trade less aggressively than solo-managed funds. Adams, Nishikawa, and Rao (2015) find that teammanaged funds with highly independent boards perform better. Patel and Sarkissian (2016) explain that the previous studies underestimate the performance of team-managed funds by using the CRSP and Morningstar Principia database, which provide a low accuracy rate in reporting manager information. After using a relatively new and better data source, Morningstar Direct, they find that team-managed funds, especially three-managers funds, have better returns without taking extra risks. 2. Changes in management structure. Last two decades brought a big increase in the proportion of team-manger funds. Bliss, Potter and Schwarz (2008) find that the number of team-manger funds has grown at seven times the rate of funds managed by individuals. Besides the performance difference, there are several other explanations. The most common accepted reason is increasing fund size. Large funds need more accurate analysis to select stocks and closer monitoring of the portfolios, which cannot be fully done by an individual manager. Therefore, multiple managers are better fit for large size funds. Besides the voluntary turnover, like retire, some changes may due to bad past performance. For example, poor performance managers are more likely to be fired (Khorana, 1996; Chevalier and Ellison 1999; Porter and Trifts, 2014) and new managers are more likely to be hired after large positive excess returns (Goyal and Wahal, 2008). 7

9 Khorana (2001) documents the effect of changes in managers. For the underperforming funds, the change can significantly improve the performance relative to the past performance. However, for the outperforming funds, the change deteriorates in post-replacement performance. Based on these studies, we want to examine the relation between fund management structure and fund performance and investigate the effect of changes in management structure on future performance of the fund. Our null hypotheses are as follows: Hypothesis 1: The fund management structure and the number of fund managers have no effect on fund performance. Hypothesis 2: The change of fund management structure does not influence post-change fund performance. III. Data and Methodology 1. Sample and data collection The data are obtained from Morningstar Direct mutual fund database, which is a survivorship-bias free dataset. Following the existing literature, we focus on actively managed domestic equity mutual funds, whose prospectus objective belong to Aggressive Growth, Growth, Growth & Income, or Equity Growth. 7 We exclude the index funds, balanced funds and funds of funds. 7 Aggressive Growth: Fund that seek rapid growth of capital and that may invest in emerging market growth companies without specifying a market capitalization range. They often invest in a small or emerging growth companies and are more likely than other funds to invest in IPO s or in companies with high price/earnings and price/book ratios. Growth: Funds that pursue capital appreciation by investing primarily in equity securities. Current income, if considered at all, is a secondary concern. Growth and Income: Growth of capital and current income are near-equal objectives for these funds. Investments are typically selected for both appreciation potential and dividendpaying ability. Equity Income: Funds that are expected to pursue current income by investing at least 65% of assets in dividend-paying equity securities. 8

10 Managers of these passively managed funds follow the market and do not make substantial investment decisions. Therefore, their judgement is not so crucial to the fund performance. Managers of the actively managed funds, on the other hand, use their information and judgement to make investment decisions. The managerial structure of these funds may affect the decisionmaking process and be reflected in the fund performance. We obtain the monthly mutual funds data from January 1990 to June In total, we have 12,375 funds, and 728,997 fund-month observations. Morningstar Direct reports all the data at the fund share-class level. The different share-classes of a fund hold the same underlying investment portfolio but differ in realized returns on account of fees, expenses, and sales charges. To avoid multiple counting, we aggregate all the classes to the fund level. More specifically, for each fund, we calculate fund age using the inception date of the oldest share class in the fund, and take average of the expense ratio and turnover ratio across different share classes to form the fund level data. We winsorize the gross return, expense ratio and turnover ratio at the bottom and top 1% level to avoid the potential misreporting. We also collect the overall rating 8 and Sharpe ratio from Morningstar. Morningstar provides the fund level assets data. We aggregate all fund assets together to get family assets size. As to fund net flow, we follow Sirri and Tufano (1998) approach to calculate: Flow TNA = (1 + R ), (1) TNA it, it, it, it, 1 8 The Morningstar Rating brings load-adjustments, performance (returns) and risk together into one evaluation. To determine a fund's star rating for a given time period (three, five, or 10 years), the fund's risk-adjusted return is plotted on a bell curve: If the fund scores in the top 10% of its category, it receives 5 stars (Highest); if it falls in the next 22.5% it receives 4 stars (Above Average); a place in the middle 35% earns 3 stars (Average); those lower still, in the next 22.5%, receive 2 stars (Below Average); and the bottom 10% get only 1 star (Lowest). 9

11 where TNAt is the total fund assets at the end of month t, and Rt is the fund gross return in the same month. Management information, our variable of interest, is derived from Morningstar manager history 9 data. We define a fund as a solo- or a team-managed fund based on the number of fund managers at the end of each month. Moreover, we also classify the funds into five categories based on the number of managers in the fund: one, two, three, four and five plus indicating one manager, two managers, three managers, four managers and five or more managers, respectively. Manager tenure measures the number of years the fund managers have stayed in that fund. In case of the team-managed funds, we use the average tenure of all the managers to represent the overall tenure. Figures 1A and 1B show the number of funds and the net assets in each manager category. Overall, the number and the size of mutual funds are growing over the sample period. Since 2003, the proportion of solo-managed funds is decreasing in favor of team-managed funds (Figure 1A). Among the team-managed funds, the majority of funds have two or three managers. Net assets of mutual funds dropped largely during the two recession periods (Figure 1B). Similarly, as for the number of funds, team-managed funds are increasing their assets under management both absolutely and relatively to solo-managed funds. In recent time period (second quarter of 2016), two-manager funds hold the largest amount of net assets followed by funds with 5 or more managers. [Insert Figure 1 Here] 9 The Morningstar Direct reports the whole history of the funds, for example, the 1290 GAMCO fund, the manager history shows [ ] Mario J. Gabelli;[ ] Kenneth T. Kozlowski;[ ] Alwi Chan. 10

12 2. Performance and Risk Measurements To measure fund performance, we rely on fund monthly gross returns. We use CAPM, Fama French (1993) three-factor model, Carhart (1997) four-factor model and Pastor-Stambaugh (2003) liquidity model to estimate each fund s alpha. The models are as follows, r r = α + β ( r r ) + ε ; (2) it, f, t it, 1 mt, f, t it, r r = α + β ( r r ) + β SMB + β HML + ε ; (3) it, f, t it, 1 mt, f, t 2 t 3 t it, r r = α + β ( r r ) + β SMB + β HML + β MOM + ε ; and (4) it, f, t it, 1 mt, f, t 2 t 3 t 4 t it, r r = α + β ( r r ) + β SMB + β HML + β MOM + β LIQ + ε, (5) it, f, t it, 1 mt, f, t 2 t 3 t 4 t 5 t it, where ri,t is the monthly gross fund return, and rf,t is the one-month U.S. Treasury bill rate, rm,t is the monthly return on the CRSP value-weighted index, SMBt, HMLt, and MOMt, and LIQt are returns on the size, book-to-market, momentum, and liquidity factors, respectively. We adopt the rolling window approach and use 12-month horizon gross return to run these models. To measure the risk of the fund, we follow Bar, Kempf, Ruenzi (2011) methods: the total risk is the standard deviation of the past 12 months returns, the systematic risk is the beta from CAPM model, and idiosyncratic risk is the standard deviation of fund i s residual fund returns from the same model. Table 1 reports the summary statistics of fund characteristic by different manager group. On average, mutual funds with five or more managers have largest fund size, but three-managers group has a higher percentage net flows (Panel A). Solo-managed funds have higher expense ratio and 11

13 relatively higher overall rating. Five-or-more manager funds have lower turnover ratio indicating lower trading. Since the team-manager funds emerged mainly in the latter part of our sample period, it is not surprising that they have shorter average manager tenure compared to solo-managed funds. Panel B of Table 1 compares the performance and risks among different groups. On one side, the solo-managed funds have best performance with highest gross and net return and abnormal return. On the other side, the solo-managed funds also have highest risks, in terms of total risk, systematic risk and unsystematic risk. [Insert Table 1 Here] In the data summary, we combine all data together and take the average for fund characteristics variables and ignore the years trending effect: the solo-funds are popular in earlier years, in which the mutual funds earned higher return. In this case, the data summaries are not providing an accurate picture of the fund properties and the good performance by solo-managed funds may be overestimated. 3. Methodology To evaluate the effect of managerial structure on fund performance, we use the following models: α = β + β Manager _ structure + β Fund _ characteristics it, 0 1 it, 6 2 it, 6 + β Manager _ tenure + Year _ dummy + ε ; 3 it, 6 it, (6) Risk = β + β Manager _ structure + β Fund _ characteristics it, 0 1 it, 6 2 it, 6 + β Manager _ tenure + Year _ dummy + ε, 3 it, 6 it, (7) where the performancei,t and riski,t represent fund abnormal return or the fund risk as discussed in the previous subsection. For manager_structure variable, we construct two sets of measures: (1) 12

14 solo- versus team-managed, and (2) the number of fund managers. The fund characteristics variables include fund age, size, expense ratio, turnover ratio, family size, objective dummy. We take natural logarithm of the fund size, family size, the fund age and manager tenure variables. In the regression, we also include the year fixed effect. Moreover, to avoid the contemporaneous effect, we take 6-month lag for all independent variables. Then we use the probit model to capture the reasons of management structure change. The model is shows as follows: Pr( change = 1) = θ + θ Fund _ characteristics + θ Manager _ tenure + Year _ dummy + ε, (8) it, 0 1 it, 6 2 it, 6 it, where changei,t indicates the change in manager number, for example, changei,t equals to one if fund i increases its manager number in month t, or decrease its manager number in month t. The fund characteristics include fund age, size, expense ratio, turnover ratio, family size, objective dummy as well as fund past performance and risk. IV. Empirical Results Table 2 reports the relationship between fund abnormal return and management structure. The dependent variable, alpha, are getting from four different models, CAPM, Fama French model, Carhart model, and Pastor-Stambaugh (2003) liquidity model. We find that the team-managed funds earn a higher abnormal return than the solo-managed funds (column 2 and 4). Then, when we divide the funds into more detailed subgroups based on the number of managers (results are presented in column 3, 5, 7, 9), we find that the three-manager and four-managers groups are the main contributors to the superior performance of team-managed funds. Consistently with Chen et al. (2004), we find that larger fund size erodes the fund performance. Family size, on the other 13

15 hand, improves performance suggesting that stronger mutual fund families (fund firms have more assets under management) have better resources, and can achieve better performance than their counterparts (fund firms have less assets under management). Surprisingly, we find that longer manager tenure leads to relatively lower abnormal return. [Insert Table 2 Here] Next, we follow Patel and Sarkissian (2016) and divide our data into three subgroups based on their objectives. Then we examine which of these subgroup contributes to performance differences found in Table 2. Our results are presented in Table 3. We find significant and positive coefficient of team-managed dummy variable for Growth and Equity Income subgroup. These funds are funds with the least aggressive investment strategy in equity markets, investing largely into dividend paying stocks. Therefore, these funds rely less on soft information than funds with aggressive growth or just growth investment strategies. Our finding is consistent with Stein (2002) argument that the team structure hampers the usage of soft information. Compared with other funds, Growth and Equity Income funds rely less on soft information, and therefore exhibit their performance advantage. [Insert Table 3 Here] Table 4 presents analysis of fund risk in context of management structure. Our risk measurements include the fund total risks, which is the standard deviation of the past 12 months return, the market beta, the coefficient of size, book-to-market, and momentum, as well as the unsystematic risk using the standard deviation of fund i s residual fund returns from Carhart model. We find that team-managed funds have significantly lower total risk as well as unsystematic risk (Table 4). 14

16 Meanwhile, these funds have also lower loading on the market risk premium and the size factor, but higher loading on the value and momentum factors. In other words, compared to solo-managed fund, the team-managed funds investments are slightly less correlated with the market and are biased towards larger-capitalization and value companies and past winners. [Insert Table 4 Here] Table 5 reports the results of probit regression capturing the reasons of changes in management structure (eq 8). We use four measures to capture the fund performance: past year return, abnormal return from Carhart model (α), the cumulative abnormal return from month t-6 to t-1 (CAR), and the underperformance indicator (underperform). The underperformance indicator is based on the Morningstar overall rating: if the rating of fund i is equal or less than 2, we treat it as underperforming fund. Not surprisingly, we find that the past performance has significant effect on manager turnover. This effect is especially strong for underperforming funds. This is not surprising since Morningstar rating used for constructing the underperformance indicator compares the fund to its peers. In addition, the fund flow plays a crucial role in deciding management structure. This is reasonable, because the fund flow reflects market evaluation of the fund. If a fund is on longer able to generate satisfactory net flow, the fund managers, of course, should be punished. Funds with low fund flow are more likely to undergo changes in their management. We also find that lager funds and funds from larger fund family have higher probability of changes in management structure. This may be due to higher standard of large fund and large fund families for their fund managers. These funds have more attention and are monitored more closely. Therefore, once the fund manager does not perform well, he is more likely to be replaced. 15

17 Last two columns of Table 5 show results for changes representing the increase and decrease in number of managers. We find that decrease in number managers (firing) is more likely after bad performance of the fund. This is consistent with Khorana (1996), Chevalier and Ellison (1999) and Porter and Trifts (2014). We did not find any evidence that the increase in number of managers depends on the past performance. [Insert Table 5 Here] Khorana (2001) documents significant improvements in post-replacement performance relative to the past performance of the fund. In this paper, we also investigate the post-change performance. We divide funds that had a change in number of managers into two groups based on their past year performance: positive and negative past year return. Using event study methodology, we calculate cumulative abnormal returns. For example, if a fund i has a change in the number of managers at month t, we treat the change as an event and calculate the cumulative abnormal return from month t-15 to t+15. To avoid the influences on multiple changes, we limit our sample to the funds that had only one change from month t-30 and t+30. We aggregate all the fund change information to get the average change effects. The results are presented in Figure 2. The Panel A presents the cumulative abnormal return calculated by Fama French 3-factor model, and Panel B presents the results using Carhart model. Similar to Khorana (2001), we find that for the negative past year return group, change in the number of managers can dramatically improve fund return. In contrast, funds in the positive past year return group don t experience the same benefit from changes in number of managers. [Insert Figure 2 Here] 16

18 V. Conclusion In the past two decades, the number of team-managed funds has grown. Nowadays, the majority of mutual funds are run by two or more managers. With increasing number of team-managed funds, it seems that market and fund families prefer team-managed funds to solo-manger funds. There is, however, no unified conclusion whether the solo- and team-managed funds differ in their performance. In this paper, we investigate this question and compare the performance of solo- and teammanaged mutual funds. Our study focuses on actively managed domestic mutual funds over past 25 years. Using Morningstar Direct data, we are able to define the management structure every month and compare returns and risks of the funds with different management structures. We find that some evidences suggest that team-managed funds have higher abnormal return than solomanaged funds. After separating team-managed funds into subgroups based on number of managers, we find that the three- and four-manager fund groups have significantly higher returns. We find that the superior performance of team-managed funds is due mainly to Growth & Equity- Income funds. We further analyze the reasons for the change in number of managers. We find that managers who had bad past performance are more likely to be fired but hiring new managers does not seem to directly depend on past performance. In addition, we also examine the post-change effect. For fund with negative past year return, the change in number of managers improves its future performance. We cannot observe this pattern in positive past year return group. 17

19 References Adams, John, Takeshi Nishikawa, Ramesh Rao, 2015, Mutual fund performance, management teams, and boards, working paper. Baer, Michaela, Alexander Kempf, and Stefan Ruenzi, 2005, Team Management and Mutual Funds, CFR Working Papers 05-10, University of Cologne, Centre for Financial Research. Bar, M., A. Kempf, and S. Ruenzi, 2011, Is a Team Different from the Sum of its Parts? Evidence from Mutual Fund Managers, Review of Finance, 15, pp Bliss, Richard T, Mark E Potter, and Christopher Schwarz, 2008, Performance Characteristics of Individually-Managed versus Team-Managed Mutual Funds, The Journal of Portfolio Management, 34, pp Carhart, Mark, 1997, On Persistence in Mutual Fund Performance, Journal of Finance, 52, pp Chen, Joseph, Harrison Hong, Ming Huang, and Jeffrey D Kubik, 2004, Does Fund Size Erode Mutual Fund Performance? The Role of Liquidity and Organization, American Economic Review, 94, pp Chevalier Judith and Glenn Ellison, 1999, Career Concerns of Mutual Fund Managers, Quarterly Journal of Economics, 14(2), pp Daniel, Kent, Mark Grinblatt, Sheridan Titman, and Russ Wermers, 1997, Measuring Mutual Fund Performance with Characteristic-Based Benchmarks, The Journal of Finance, 52, pp Fama, Eugene and Kenneth French, 1993, Common Risk Factors in the Returns on Stocks and Bonds, Journal of Financial Economics, 22, pp Goldman Eitan, Zhenzhen Sun, and Xiyu (Thomas) Zhou, 2016, The Effect of Management Design on the Portfolio Concentration and Performance of Mutual Funds, Financial Analysts Journal, 72(4), pp Goyal, Amit and Sunil Wahal, 2008, The Selection and Termination of Investment Management Firms by Plan Sponsors, Journal of Finance, 63(4), pp Han, Yufeng, Michael J. Rebello, and Thomas H. Noe, 2012, Horses for Courses: Fund Managers and Organizational Structures, SSRN Electronic Journal. Hill Gayle, 1982, Group versus individual performance: Are N+ 1 Heads Better Than One? Psychological bulletin, 91(3), pp

20 Hogg, Michael, John Turner, and Barbara Davidson, 1990, Polarized Norms and Social Frames of Reference: A Test of the Self-Categorization Theory of Group Polarization. Basic and Applied Social Psychology, 11, pp Investment Company Institute, 2016 Investment Company Fact Book, May 2015, Karagiannidis, Iordanis, 2010, Management Team Structure and Mutual Fund Performance, Journal of International Financial Markets, Institutions and Money, 20, pp Kerr, Norbert L., 1992, Group Decision Making at a Multialternative Task: Extremity, Interfaction Distance, Pluralities, and Issue Importance, Organizational Behavior and Human Decision Processes, 52, pp Khorana, Ajay, 1996, Top Management Turnover: An Empirical Investigation of Mutual Fund Managers, Journal of Financial Economics, 40(3), pp Khorana, Ajay, 2001, Performance Changes Following Top Management Turnover: Evidence from Open-End Mutual Funds, Journal of Financial and Quantitative Analysis, 36, pp Massa, Massimo, Jonathan Reuter, and Eric Zitzewitz, 2010, When Should Firms Share Credit with Employees? Evidence from Anonymously Managed Mutual Funds, Journal of Financial Economics, 95, pp Moscovici, Serge; Zavalloni, Marisa, 1969, The Group as A Polarizer of Attitudes, Journal of Personality and Social Psychology, 12(2), pp Pastor, Lubos, and Stambaugh Robert, 2003, Liquidity Risk and Expected Stock Returns, Journal of Political Economy, 111, pp Patel, S., and S. Sarkissian. 2016, To Group or Not To Group? Evidence from Mutual Funds Databases, The Journal of Financial and Quantitative Analysis, forthcoming. Pizzani, Lori, 2004, The Mutual Fund Scandal: Why Did Unethical Behavior Go Unchecked? CFA Magazine,15, pp Porter G E, Trifts J W, 2014, The Career Paths of Mutual Fund Managers: The Role of Merit, Financial Analysts Journal, 70(4), pp Prather, Larry J., and Karen L. Middleton, 2002, Are N+1 heads better than one? Journal of Economic Behavior & Organization, 47, pp

21 Sah Rajj Kumar, Joseph Stiglitz. 1986, The architecture of economic systems: Hierarchies and polyarchies, Journal of American Economic Review, 76(4), pp Shapre W, 1981, Decentralized Investment Management. The Journal of Finance, 36(2), pp Shaw, Marjorie, 1986, A Comparison of Individuals and Small Groups in the Rational Solution of Complex Problems, The American Journal of Psychology, 44(3), pp Sirri, Erik, and Peter Tufano, 1998, Costly Search and Mutual Fund Flows, Journal of Finance, 53(5), pp Stein, Jeremy, 2002, Information Production and Capital Allocation: Decentralized versus Hierarchical Firms, Journal of Finance, 57(5), pp

22 Figure 1A. Number of Funds at The End of Year Over the Sample Period 4,000 3,500 3,000 2,500 2,000 1,500 1, Q Figure 1B. Net Assets Value of Funds at The End of Year Over the Sample Period Billion of USD $7,000 $6,000 $5,000 $4,000 $3,000 $2,000 $1,000 $ Note: This figure describes the number of funds and the net assets in each manager category. We classify the funds into five categories based on the number of managers in the fund: one, two, three, four and five plus, which indicate one manager, two managers, three managers, four managers and five or more managers, respectively. 21

23 Figure 2. Change in Number of Managers and Post-Change Performance Panel A. Fama French Three-Factor Model to Get Cumulative Abnormal Return Month Positive past year return Negative past year return Panel B. Carhart Four-Factor Model to Get Cumulative Abnormal Return Month Postive past year return Negative past year return 22

24 Table 1. Data Summary Panel A. Team Size and Fund Characteristics Fund size ($ millions) 2,250 1,350 1,450 1,400 3,750 Net flow (%) Expenses ratio (%) Turnover ratio (%) Morningstar rating Sharpe ratio (%) Managers tenure Panel B. Team Size, Fund Return and Risk Gross return Net return Excess return Alpha (CAPM) Alpha (Three-factor model) Alpha (Carhart model) Alpha (Five-factor model) Total risk Systematic risk Unsystematic risk Note: We classify the funds into five categories based on the number of managers in the fund: one, two, three, four and five plus, which indicate one manager, two managers, three managers, four managers and five or more managers, respectively. The alphas are calculated from CAPM, Fama French (1993) three-factor model, Carhart (1997) four-factor model and Pastor-Stambaugh (2003) liquidity, respectively. The total risk is the standard deviation of the past 12 months return, the systematic risk is the beta from CAPM model, and idiosyncratic the standard deviation of fund i s residual fund returns from the same model. All the alphas are based on monthly gross return. 23

25 Table 2. Team Size and Fund Abnormal Return α = β + β Manager _ structure + β Fund _ characteristics it, 0 1 it, 6 2 it, 6 + β Manager _ tenure + Year _ dummy + ε, 3 it, 6 it, CAPM CAPM 3-factor 3-factor 4-factor 4-foract 5-factor 5-factor Team-managed i,t ** * (0.047) (0.050) (0.690) (0.553) Two-mangeri,t (0.270) (0.121) (0.897) (0.507) Three-manger i,t ** (0.024) (0.131) (0.831) (0.348) Four-manager i,t * (0.082) (0.150) (0.224) (0.348) Five-or-more-manager i,t (0.341) (0.250) (0.612) (0.631) Fund Size i,t *** *** *** *** *** *** *** *** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Family Size i,t ** ** *** *** ** ** ** ** (0.009) (0.010) (0.001) (0.001) (0.044) (0.046) (0.022) (0.023) Fund Flow i,t *** *** *** *** *** *** *** *** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Fund Age i,t * * *** *** *** *** ** ** (0.065) (0.062) (0.000) (0.001) (0.000) (0.000) (0.018) (0.017) Manger Tenure i,t ** ** ** ** ** ** *** *** (0.011) (0.011) (0.026) (0.027) (0.002) (0.002) (0.001) (0.001) Total Risk i,t * * *** *** *** *** (0.169) (0.170) (0.083) (0.083) (0.000) (0.000) (0.000) (0.000) Expense Ratio i,t (0.213) (0.220) (0.200) (0.198) (0.953) (0.957) (0.708) (0.717) Turnover ratio i,t *** *** *** *** ** ** *** *** (0.000) (0.000) (0.000) (0.000) (0.005) (0.005) (0.000) (0.000) Constant ** ** *** *** ** ** * * (0.010) (0.010) (0.001) (0.001) (0.005) (0.005) (0.091) (0.091) Objective FE Yes Yes Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Yes Yes Number of Observations 389, , , , , ,951 37, ,651 Adjusted R square Note: Team-manged indicates that fund has more than one manager. We also classify the funds into five categories based on the number of managers in the fund: one, two, three, four and five plus, which indicate one manager, two managers, three managers, four managers and five or more managers, respectively. The alphas are calculated from CAPM, Fama French (1993) three-factor model, Carhart (1997) four-factor model and Pastor-Stambaugh (2003) liquidity, respectively. Total risk is the standard deviation of the past 12 months return. Manager tenure measures the year the fund managers have stayed in that fund. In case of the teammanaged funds, we use the average tenure of all the managers to represent the overall tenure. p-values in parentheses, and are calculated by robustness standard errors. * indicates p < 0.10, ** indicates p < 0.05, and *** indicates p <

26 Table 3. Team Size, Fund Objective and Fund Abnormal Return α = β + β Manager _ structure + β Fund _ characteristics it, 0 1 it, 6 2 it, 6 + β Manager _ tenure + Year _ dummy + ε, 3 it, 6 it, Whole Sample Aggressive Growth CAPM Growth Growth & Equity- Income Whole Sample Fama French 3-factor Model Aggressive Growth Growth Growth & Equity- Income Team-managed i,t ** * * * (0.047) (0.834) (0.234) (0.066) (0.050) (0.614) (0.874) (0.097) Fund Size i,t *** *** *** * *** (0.000) (0.312) (0.233) (0.000) (0.000) (0.507) (0.093) (0.000) Family Size i,t ** ** *** ** (0.009) (0.324) (0.893) (0.048) (0.001) (0.797) (0.514) (0.021) Fund Flow i,t *** ** * *** *** ** ** *** (0.000) (0.001) (0.055) (0.000) (0.000) (0.003) (0.003) (0.000) Fund Age i,t * * *** *** (0.065) (0.983) (0.358) (0.093) (0.000) (0.863) (0.378) (0.001) Manger Tenure i,t ** ** (0.011) (0.904) (0.808) (0.110) (0.026) (0.828) (0.975) (0.111) Total Risk i,t * * (0.169) (0.435) (0.159) (0.412) (0.083) (0.626) (0.119) (0.058) Expense Ratio i,t (0.213) (0.884) (0.683) (0.119) (0.200) (0.705) (0.425) (0.354) Turnover ratio i,t *** *** *** *** (0.000) (0.368) (0.465) (0.000) (0.000) (0.102) (0.288) (0.000) Constant ** *** *** *** * (0.010) (0.271) (0.000) (0.270) (0.001) (0.253) (0.000) (0.056) Objective FE Yes No No No Yes No No No Year FE Yes Yes Yes Yes Yes Yes Yes Yes Number of Observations 389,349 10,548 23, , ,263 10,546 23, ,765 Adjusted R square Note: Team-managed indicates a fund has more than one manager. The alphas are calculated from CAPM and Fama French (1993) three-factor model, Total risk is the standard deviation of the past 12 months return. Manager tenure measures the year the fund managers have stayed in that fund. In case of the team-managed funds, we use the average tenure of all the managers to represent the overall tenure. p-values in parentheses, and are calculated by robustness standard errors. * indicates p < 0.10, ** indicates p < 0.05, and *** indicates p <

27 Table 4. Management Structure and Fund Risk Risk = β + β Manager _ structure + β Fund _ characteristics it, 0 1 it, 6 2 it, 6 + β Manager _ tenure + Year _ dummy + ε, 3 it, 6 it, Carhart Model Total risk β SMB HML MOM sd(error) Team-managed i,t *** *** *** *** *** *** (0.000) (0.000) (0.001) (0.000) (0.000) (0.000) Fund Size i,t *** *** *** *** (0.000) (0.929) (0.000) (0.000) (0.801) (0.000) Family Size i,t *** *** *** *** *** *** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Fund Flow i,t *** * *** ** (0.838) (0.000) (0.066) (0.187) (0.000) (0.001) Fund Age i,t *** *** *** *** *** (0.000) (0.000) (0.000) (0.851) (0.000) (0.000) Manger Tenure i,t *** *** *** *** *** *** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Gross Returni,t *** *** *** *** *** * (0.000) (0.000) (0.000) (0.000) (0.000) (0.079) Expense Ratio i,t *** ** *** *** *** (0.000) (0.004) (0.000) (0.000) (0.240) (0.000) Turnover ratio i,t *** *** *** *** *** *** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Constant *** *** *** *** (0.000) (0.000) (0.001) (0.187) (0.272) (0.000) Objective FE Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Number of Observations 400, , , , , ,963 Adjusted R square Note: Team-managed indicates a fund has more than one manager. Fund total risks is the standard deviation of the past 12 months return, market beta, the coefficient of size, book-to-market, and momentum are from Carhart model. The unsystematic risk using the standard deviation of fund i s residual fund returns from the same model. Manager tenure measures the year the fund managers have stayed in that fund. In case of the team-managed funds, we use the average tenure of all the managers to represent the overall tenure. p-values in parentheses, and are calculated by robustness standard errors. * indicates p < 0.10, ** indicates p < 0.05, and *** indicates p <

28 Table 5. Reason of Manager Number Change Pr( change = 1) = θ + θ Multi _ manager + θ Fund _ characteristics it, 0 1 it, 6 2 it, 6 + θ Manager _ tenure + Year _ dummy + ε, 3 it, 6 it, Change Change Change Change More managers Less managers Past Year return *** *** (0.001) (0.864) (0.000) Alpha i,t-6 (4-factor) * (0.096) CARi,t-6:t-1 (4-factor) ** (0.001) Underperform i,t *** (0.000) Fund Size i,t *** *** *** *** ** (0.000) (0.000) (0.000) (0.363) (0.000) (0.009) Family Size i,t *** *** *** *** *** *** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Fund Flow i,t *** *** *** ** * *** (0.000) (0.000) (0.000) (0.005) (0.070) (0.000) Fund Age i,t * * *** (0.080) (0.342) (0.186) (0.248) (0.092) (0.000) Manger Tenure i,t *** *** *** *** *** *** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Total Risk i,t (0.821) (0.772) (0.190) (0.594) (0.326) (0.413) Expense Ratio i,t ** *** *** (0.012) (0.000) (0.118) (0.529) (0.973) (0.000) Turnover ratio i,t *** *** *** *** *** *** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Constant *** *** *** *** *** *** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Objective FE Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Number of Observations 399, , , , , ,640 Pseudo R square Note: We use probit model to run this regression. Team-managed indicates that fund has more than one manager. The alphas are calculated from Fama French (1993) three-factor model, Carhart (1997) four-factor model, respectively. Total risk is the standard deviation of the past 12 months return. The underperforming indicator is based on the Morningstar overall rating, if the rating of fund i is below 3, we treat it as underperformed. Manager tenure measures the year the fund managers have stayed in that fund. In case of the team-managed funds, we use the average tenure of all the managers to represent the overall tenure. p-values in parentheses, and are calculated by robustness standard errors. * indicates p < 0.10, ** indicates p < 0.05, and *** indicates p <