Nội dung

CFR-Working Paper NO. 05-10 Team Management and Mutual Funds M. Bär • A. Kempf • S. Ruenzi Team Management and Mutual Funds* Michaela Bär baer@wiso.uni-koeln.de www.wiso.uni-koeln.de/graduiertenkolleg Alexander Kempf kempf@wiso.uni-koeln.de www.wiso.uni-koeln.de/finanzierung Stefan Ruenzi ruenzi@wiso.uni-koeln.de www.wiso.uni-koeln.de/finanzierung September 2005 Keywords: Mutual Funds, Team Management, Performance, Risk-Taking, Investment Style, Fund Flows JEL-Classification: G23, M54 We thank Vikas Agarwal, Jason Greene, Jayant Kale, Ajay Khorana, Jerry Parwada, Laura Starks, Sheridan Titman and seminar participants at Stanford University, Georgia State University, University of Texas at Austin, University of Massachusetts at Amherst, and the CFR Cologne Research Seminar for helpful comments and suggestions. Bär is at the Graduate School of Risk Management at the University of Cologne, Albertus Magnus Platz, 50923 Koeln, Germany, Kempf and Ruenzi are at the Department of Finance, University of Cologne, Albertus-Magnus Platz, 50923 Koeln, Germany. Bär gratefully acknowledges financial support from the German Research Foundation (DGF). All authors are members of the Centre for Financial Research (CFR) Cologne. All errors are our own. Abstract In recent years, team management has become increasingly popular in the mutual fund industry. In this paper, we analyze team management along three broad dimensions. First, we examine potential determinants explaining a fund’s management structure. Second, we analyze potential effects of fund management structure on managerial behavior. Third, we address the consequences of team management on fund performance, performance persistence, and fund inflows. Findings show that the management choice is a strategic decision, made usually uniformly for all funds at the fund family level. In particular the extent and complexity of the tasks fund managers face determine management structure, with more teams in segments that require expertise in different fields. Regarding the effects of team management, we find that the management behavior of teams and individual managers differ systematically. Funds managed by teams exhibit significantly lower (unsystematic) risk than single manager funds and adjust their risk to a lesser extent as response to prior performance. In their investment style teams are less extreme and more consistent over time. Looking at fund performance, we find some, albeit only weak, evidence that team management has a negative impact on fund performance. However, team-managed funds are more persistent in their performance over time. Fund investors seem to care about fund management structure. Our findings show that team-managed funds experience significantly higher inflows. 1 I Introduction Conventional wisdom holds that two heads are better than one. But is this also true when it comes to managing a mutual fund? Recent developments in the mutual fund industry suggest so. Over the past several years, many mutual fund companies have converted their funds previously run by a single manager to team status. The percentage of US equity funds managed by teams multiplied between 1994 and 2003 from 5 % to about 46 %. Despite the growing importance of team management in the mutual fund industry, barely any empirical research has been conducted on this issue. This paper fills this gap. We address three broad research questions: 1. What are the determinants explaining the use of teams vs. single managers for the management of mutual funds? 2. How does the managerial behavior of teams and single-managers differ in terms of risk taking and investment style? 3. What are the consequences of team management on fund performance, performance persistence, and fund inflows? Answers to these questions have significant implications: When choosing among competing funds, fund investors should consider fund management structure if it has an effect on managerial behavior and fund performance. For investment companies, potential answers may be vitally important for how to organize and manage their funds. Furthermore, this issue is also of broader academic interest. Despite the fact that many economic, political as well as legal decisions are made by teams, much of economic theory has to date focused primarily on the behavior of individuals. The mutual fund industry is ideally suited to analyze the decisions of teams as compared to individuals as it constitutes a real-world setting, where the behavior and decision outcomes of managers can easily be observed and directly compared. There are two main reasons why fund management structure might matter: Firstly, several recent studies have stressed the important role of portfolio mangers in generating fund performance (e.g., Chevallier/Ellison, 1999b, Ding/Wermers, 2005). They find that manager characteristics, in particular their educational background and experience, have a significant influence on fund performance. However, if managers play an important role for fund performance, then it should also matter whether a single manager or a team of several managers decide about fund investments. Secondly, several (mainly psychological) experiments have shown that decisions made by individuals differ from decisions made by teams in various dimensions, in particular in terms of their riskiness, extremity, and quality (e.g. Adams/Ferreira, 2003, Cooper/Kagel, 2004). Thus, we would not necessarily expect team decisions to be the simple sum of individuals’ decisions. 2 For our empirical examinations, we use US open-end mutual fund data from the years 1994 to 2003, covering the years of the rapidly increasing popularity of team-managed funds. Our study delivers a broad array of new results on the determinants and consequences of fund management structure, as well as on the managerial behavior of teams versus single managers. Findings on the determinants of fund management structure indicate that the fund management choice is a strategic decision, usually made uniformly for all funds by the top management of the fund family. We show, that families following a team management approach (i) offer a higher number of funds, (ii) run larger funds and (iii) run more funds in segments that require expertise in different fields, e.g. in the balanced funds or global funds segment. Thereby, our results yield support to the view that teams are primarily employed for more extensive and complex tasks. Regarding their management behavior, our results show that the decisions made by individuals and teams differ systematically. Management teams take on less overall fund risk as compared to single managers. This difference is mainly driven by a lower level of unsystematic risk. Looking at changes in risk taking as response to prior performance, we find that teams alter their risk to a lesser extent than individual managers. With respect to their style characteristics team-managed funds show a less extreme investment style than single managed funds, i.e. they deviate less from segment specific style benchmarks. In addition, teammanaged funds are more consistent in their investment style over time. Overall, our findings support the idea that team decision-making represents a form of averaging among individual positions and ensures a higher continuity in management. Analysing the consequences of fund management structure on performance, we find some, albeit only weak, evidence that team management has a negative impact on fund performance. Fund management teams can either not realize potential benefits of having more than one manager running the fund, or these benefits are overcompensated by additional costs and team specific inefficiencies. Though differences in performance are generally small, team-managed funds exhibit a significantly higher persistence in their performance. This indicates that a higher continuity in management of team- as compared to single-managed funds eventually leads also to higher continuity in fund performance. Finally, fund investors seem to care about fund management structure. Analysing fund inflows we find that investors strongly prefer team-managed funds. This is a possible explanation for the increasing popularity of the team management approach in the mutual fund industry in recent years. 3 By providing a comprehensive view on team management and its potential effects in the context of the real-world setting offered by the mutual fund industry, our study contributes to the mutual fund literature as well as the general literature on group decision-making. While the literature has theorized that teams in money management can be motivated by risk-sharing considerations (Barry/Starks, 1984) and might offer benefits from specialization of team members and diversification among managers (e.g. Williams, 1980, Sharpe, 1981), there is barely any empirical evidence on the potential effects of team management. The only notable exceptions are two papers on the performance of team-managed funds that deliver contradicting results. While Chen/Hong/Huang/Kubik (2004) find that teams underperform singlemanaged funds, Prather/Middleton (2002) find no significant difference in performance. Our paper is the first to address the determinants of team management in the fund industry as well as the risk taking, investment style, performance persistence and inflows of team- versus single-managed funds. The paper proceeds as follows. Section II explores some of the implications for fund management that arise from the economic and psychological literature. We describe our data in Section III. Section IV presents empirical results on the determinants of fund management choice. In Section V, we examine differences in the behavior between management teams and single managers, while the consequences of fund management structure on fund performance, performance persistence and fund flows are analysed in Section VI. Section VII concludes. II The Management Structure of Mutual Funds and its Potential Effects In this section we analyze potential implications for fund management that arise from the economic and psychological literature. In particular, we focus on two aspects that have been studied in literature: first, the (managerial) behavior of teams versus individuals, and, second, the performance of teams versus individuals. Comparing teams and individuals with respect to the riskiness and extremity of their behavior, findings in literature are differing. There is some evidence that team decisions entail a diversification-of-opinion effect and tend to be less extreme than individual ones (e.g. Adams/Ferreira, 2003). As discussed by Moscovici/Zavalloni (1969), a natural hypothesis is that final group decisions represent a compromise, an averaging among individual positions. In order to reach a consensus, group members have to balance their individual opinions. Consequently, one would expect decisions of teams to be less extreme and less volatile over time than decisions of single managers (e.g. Sah/Stiglitz 1986, 1991). In the fund management con4 text, this should be reflected in less volatile fund returns of team-managed funds and less extreme, more consistent investment styles of management teams. However, there are also theories predicting an increase of the extremeness and riskiness of group decisions. Some psychological experiments provide evidence for a group-polarization effect which refers to the tendency that group member’s individual judgements become more extreme after group discussion (e.g. Myers/Lamm 1976). One potential reason for this effect is that team members become comfortable with more extreme positions when they realize that the other team members also generally support this position. Furthermore, some psychological studies document a risky-shift phenomenon which indicates that groups tend to make riskier choices than the average group member (e.g. Kogan/Wallach, 1965, Kahneman/Tversky, 1979). A potential explanation lies in the process of team deliberation which might create an illusion of control and leads to a higher confidence, or even overconfidence, of team members. Persuasive arguments favouring the dominant position might convince doubtful members and help to reach a consensus around the riskier choice. These two phenomena can attenuate or even reverse a moderating affect of teams described above and lead to a more extreme investment style and higher risk of funds managed by more than one manager. Regarding the performance of teams versus individuals, the literature also provides differing evidence. Several studies find that teams act more rational and perform better (e.g. Bone/Hey/Suckling, 1999, Blinder/Morgan, 2000, Rockenbach/Matauschek, 2001, Cooper/Kagel, 2004). Management literature suggests that team decisions may benefit from two sources. Firstly, team members can correct each others errors in the process of team deliberation (e.g. Shaw, 1932, Sharpe, 1981). Given the bounded rationality of individuals, this is particularly important under conditions of high uncertainty and complexity, which regularly characterise investment decisions of mutual funds. Secondly, teams may profit from a broader resource of knowledge and capabilities, particularly when specialists with complementary skills are integrated in teams (Pelled/Eisenhardt/Xin, 1999). Consistent with these ideas, we would expect team-managed funds to make better and more rational decisions than individually managed funds. As a result, team-managed funds should deliver a better performance. However, additional costs associated with group decision-making (e.g. coordination and communication costs) may reduce potential benefits. Furthermore, experimental studies have documented inefficiencies and biases that are specific to group decision-making. For example, some studies find that team members become less motivated and reduce effort as compared to situations where they work individually (e.g. Williams/Nida/Baca/Latané, 1987, Weldon/Gargano, 1988). Hölmstrom (1982) argues that this is due to moral hazard in teams. 5 Moreover, groupthink may lead highly cohesive teams to strive for unanimity, even at the expense of decision quality (Janis 1982). Mutual fund management teams may also be subject to these phenomena. In addition, management teams may be used by fund companies as “training grounds” for inexperienced managers, as adversaries of the team management approach argue (e.g. Pizzani, 2004). As a result of these effects, potential benefits of team decisionmaking can be reduced or even overcompensated. Ultimately, whether investment decisions taken by management teams are less or more extreme and risky, better or worse than individual decisions are open empirical questions that we address in the following. III Data Our primary data source is the CRSP Survivor Bias Free Mutual Fund Database.1 This database covers U.S. open-end mutual funds and provides information about fund returns, fund management structures, total net assets, investment objectives, and other fund characteristics. We focus on actively managed equity funds which invest more than 50 % of their assets in stocks, excluding bond, money market and index funds. We use the ICDI objective codes, identified by Standard & Poors’s Fund Services to define the market segments in which funds operate. This leaves us with 10 different segments. We aggregate multiple classes of the same fund to avoid multiple counting. Although multiple share classes are listed as separate funds in CRSP, they are backed by the same portfolio of assets and have the same portfolio manager(s). Following the approach in Daniel/Grinblatt/Titman/Wermers (1997), we identify classes by matching fund names and characteristics, such as fund management structure, turnover, and fund holdings in asset classes. To examine the consequences of specific management structures, it is crucial to clearly classify a fund’s management structure. CRSP reports management structures in several ways. First, for funds managed by an individual, the manager is reported by name. We classify these as “single manager” funds (SM). Second, if CRSP reports “team” or “management team”, we label these funds team-managed (T). A third category lists the names of two or more managers or reports a manager name and “et al.” or “and team”. As it is not quite clear, how this 1 Source: CRSPSM, Center for Research in Security Prices. Graduate School of Business, The University of Chicago. Used with permission. All rights reserved. For a more detailed description of the CRSP database, see Carhart (1997) and Elton/Gruber/Blake (2001). 6 classification differs from the team-managed and single manager funds, respectively, we exclude these funds from the final sample. A fourth category reports the name of a management company. These funds are also excluded from the sample since the precise management structure remains unclear. Our final sample spans the period from January 1994 to December 2003 and includes 14,848 yearly observations on US equity funds. It covers the years of the rapid growth of teammanaged funds, as it can be seen from Figure 1. This figure plots the percentage of team and single manager funds in our sample between 1994 and 2003. – Insert FIGURE 1 about here – In 1994, team-managed funds represent only about 5 % of the total number of equity funds. In the following years, this percentage grows rapidly, reaching about 46 % in 2003. When looking at assets under management of single and team-managed funds during the same period, we find a similar development. Assets held by team-managed funds increase from 7 % in 1994 to about 50 % of total assets held by single and team-managed equity funds in 2003. Summary statistics of our final sample are given in Table 1. – Insert TABLE 1 about here – The second column shows the characteristics of all funds. On average, sample funds are 9.7 years old and manage over 840 million USD. The mean turnover rate is slightly above 1.14 and the average fee burden is 1.4 % p.a.2 To better understand the characteristics of single and team-managed funds, we report summary statistics of the two sub-samples in columns 3 and 4. The respective differences are reported in column 5. Team-managed funds are significantly younger (8.9 versus 10 years), have higher TNAs (997 versus 791 million USD), a higher turnover ratio (146 % versus 102 %), and lower fees (1.29 % versus 1.44 % p.a.) as compared to single-managed funds. 2 Following Sirri/Tufano (1998), we calculate total fees as the sum of a fund’s expense ratio and 1/7 of its total loads. 7 Looking at the distribution of team- and single managed funds across sections we see considerable cross-sectional variation, as shown in Figure 2. – Insert FIGURE 2 about here – – This figure plots the share of team- and single-managed funds in each of the ten market segments for the year 2003. The balanced funds sector has the highest percentage of teammanaged funds (61% in 2003), followed by the global equity funds and international equity funds segments (about 50 %). The share of team-managed funds is lowest for sector funds and in the utility funds segment (33 % and 39 %, respectively). The comparison between the two sub-samples provides preliminary evidence that team and single manager funds differ significantly with respect to their characteristics and are unequally spread across market segments. We will examine the management choice decision more formally in the next section. IV Determinants of Funds Management Structure 1 Analysis on the Individual Fund Level In this section we explore possible determinants of a fund’s management structure. We hypothesize that funds are managed by a team rather than a single manager if, first, their tasks are more extensive and complex, and, second, their families in general promote the team approach. To investigate the potential effects of these two aspects on fund management we relate the probability of a fund being team-managed, Prob(Team Management), to fund specific variables that characterize a fund’s task and the management policy of its family. We estimate the following logit model: ( Prob(Team Management)i,t = F β1 ( Size )i ,t −1 + ∑ β k ( Segment )i ,t + β 2 ( Family Policy )i ,t −1 k + β 3 ( Age )i ,t −1 + ∑ α y ⋅ D( y )i ,t + ε i ,t ) . (1) y Here, i is the index for an individual fund. Size proxies for the extent of a fund’s tasks and is computed as the logarithm of assets under management of fund i. To capture the complexity of a fund’s task, we add a set of dummy variables, Segment. These variables adjusts for the fact that managing a fund in certain market segments is more complex in the sense that exper8