تخصیص حقوق تصمیم گیری و استراتژی های سرمایه گذاری در صندوق های سرمایه گذاری متقابل

The literature suggests that while decentralized decision making can allow for greater specialization in an organization, it heightens the cost of coordinating decisions. The mutual fund industry—in particular, sole- and team-managed balanced funds—provides an ideal setting to test the specialization versus coordination trade-off, as information on decision structures and fund actions is easily obtained. We show that sole-managed balanced funds, with centralized decision rights, exhibit significant market timing that requires reallocation across asset classes. However, consistent with coordination difficulties between managers specializing in particular asset classes, no market timing is evident in team-managed balanced funds. Team-managed funds exhibit greater returns from specialization, in the form of better security selection performance than sole-managed funds. These results hold cross-sectionally and for funds that switch management structures. The overall returns across different management structures are similar, indicating a market equilibrium. Investor flows reward market-timing performance for sole- but not team-managed funds.

How should decision rights be located within organizations? What are the consequences of centralized versus decentralized decision making? In their analysis of organizational structure, Jensen and Meckling (1992) emphasize the cost and distortion associated with communicating information within organizations. They posit that decentralized structures benefit from lower information transmission costs by giving decision rights to those with specific information. Decentralized organizations, however, face a challenge in coordinating actions between agents that could differ in terms of their information and incentives (also see Hart and Moore, 2005; Hart and Holmström, 2010). These coordination problems can be alleviated in a centralized structure in which the decision-maker has a greater span of control, while having less specific information. Similar insights emerge from the literature on teams. It is suggested that while teams can add value when members have specialized and complementary skills (Lazear, 1988), substantial costs are associated with monitoring and coordinating actions within teams (Alchian and Demsetz, 1972 and Becker and Murphy, 1992).1 Our objective in this paper is to propose and test predictions of the decision rights theory. We test whether centralized decision structures implement strategies consistent with lower coordination costs and a greater span of control and with less reliance on specific information. We identify a class of mutual funds—balanced funds—in which the investment strategies predicted under centralized decision making are different from those under decentralized decision making. The empirical predictions are developed within a simple framework that helps to clarify the trade-offs in centralized versus decentralized decision making in market equilibrium. Our empirical results are supportive of the decision rights theory: Investment strategies of sole-managed balanced funds are less reliant on specific information. They reflect the manager's greater span of control and lower coordination costs. Studying decision making in organizations in the context of investment strategy and performance of balanced fund managers has several distinct advantages. First, US mutual funds are managed both by individual managers and by management teams. Management teams are typically composed of two or more managers who, depending on fund objectives, have expertise in different asset classes or industry sectors. The structure of fund management gives a straightforward way to categorize the decision-making structure: being more centralized under a sole manager or less centralized when team-managed. Second, many decisions of fund managers, such as their portfolio holdings and trading activities, are observable, which is not the case in most other settings. Furthermore, the performance of mutual funds can be reliably measured. The finance literature has developed a large set of performance evaluation methodologies that allows for evaluation of the performance of mutual funds and the characterization of their investment strategies. Finally, our focus is specifically on balanced funds because, as we argue, they are particularly suited for testing decision-structure hypotheses. Balanced funds allocate investments across different asset classes, typically between stocks and bonds. They are usually required to maintain, with varying degrees of flexibility, a prespecified ratio of debt and equity investments.2 In broad terms, two types of investment strategies are available to balanced funds. Balanced funds invest in both stocks and bonds and, hence, can deliver performance through allocation decisions across asset classes (generally referred to as market-timing skills) or by identifying investment opportunities within each asset class (referred to as security-selection skills) or both. While both types of strategies can contribute to fund performance, the structure of decision rights that facilitates one or the other strategy is different. In team-managed balanced funds, individual managers are typically specialized in one of the asset classes.3 We should, therefore, expect security selection strategies to contribute to the performance of team-managed funds. A market-timing strategy is less suited to a decentralized team structure because it places substantial coordination demands on the managers. Allocation decisions across asset classes have to rely on the agreement and coordination among the various managers specialized in particular asset classes. These coordination problems could be exacerbated by conflicting incentives and information, the potential adverse impact of asset allocation decisions on security selection performance, and an unwillingness to reduce the assets under their control. Hence, a fund in which the decisions are made by a sole manager faces fewer coordination problems. In such a centralized structure, the manager can unilaterally change allocations across asset classes and, therefore, has an inherent advantage in terms of implementing market-timing strategies. At the same time, a sole manager might not have the specialized skills or resources necessary for successful security selection within an asset class. Even if the manager has subordinates who specialize in particular asset classes, issues of agency and communication, highlighted in Jensen and Meckling (1992), tend to deliver weaker performance than if the subordinates had independent decision rights over the asset classes and were evaluated on that basis. For the empirical analysis, our sample of balanced funds includes two types of asset allocation funds from the Center for Research in Security Prices (CRSP) Survivor-Bias-Free US Mutual Fund Database: those that strive to maintain a balanced portfolio of stocks and bonds within a prespecified range (Rigid funds) and those that are generally more free to allocate resources across the two asset classes (Flexible funds). We identify team- and sole-managed funds for the full sample of balanced funds based on the fund manager information provided in the same database. We first study the asset allocation decisions, or market timing skills, in the sample of balanced funds based on fund returns and asset allocations. Results from the Treynor and Mazuy (1966) and the Henriksson and Merton (1981) market-timing models reveal significant market timing in the full sample of balanced funds. However, systematic differences exist in the market-timing performance between sole- and team-managed funds. In general, sole-managed balanced funds exhibit significant market-timing skills while little evidence exists of market timing in team-managed balanced funds. Further, sole-managed funds exhibit greater market-timing ability in Flexible funds, in which managers have greater discretion in their allocation decisions across asset classes, than in Rigid funds. The performance attribution methodology that employs information on both asset allocation and asset returns provides corroborating evidence. To better understand the impact of management structure on fund investment strategy and performance, we decompose the overall performance of funds into two components: that due to market timing and that due to security selection. We again do this based on fund returns as well as asset allocation decisions. The results indicate that, while team-managed funds outperform sole-managed funds in security selection, they under-perform in market timing, particularly among Flexible funds. As a result, the overall performance of the sole- and team-managed balanced funds is not significantly different, which is suggestive of a market equilibrium. The difference in investment strategy and source of performance between sole- and team-managed funds is consistent with the prediction that location of decision rights affects fund investment decisions. To further investigate the relation between fund management structure and fund investment decisions, we examine market timing activities in a sample of funds that change their management structure, from team to sole or from sole to team, over the sample period. The results suggest a causal link between management structure and investment strategy. Funds that change from sole to team management experience significant decline in market-timing activities; the converse holds for funds that change from team to sole management. For the management change sample, we also find that funds that switched from team to sole management experienced a decline in security selection performance. The results remain robust after controlling for past performance and fund fixed effects, as well as the time trend of increasing team management in the balanced fund industry. We evaluate alternative explanations, not based on decision rights, for the lack of market timing in team-managed balanced funds. For instance, one possibility is that, for reasons other than decision rights, fund companies assign managers with superior market-timing skills to be sole managers of balanced funds, while using team management when managers with specialized skills are recruited. To examine whether differences in fund objectives or managerial skills explain the lack of market timing in team-managed funds, we look for evidence of market timing within the equity portfolios (i.e., shifting the risk levels of equity portfolios according to market conditions) in both sole- and team-managed funds. Using information from the equity holdings of these funds, we find robust evidence that sole- and team-managed funds are equally likely to engage in market timing within their equity portfolios. This suggests that the lack of market timing across asset classes in team-managed funds might not be due to a lack of such skills on the part of team managers who specialize in equities. Instead, it could largely be the result of communication and coordination costs within a decentralized structure. Further investigation of the relation between market timing and security selection within sole- and team-managed funds reveals a strongly negative correlation between performance due to market timing and security selection. The negative correlation within sole- and team-managed balanced funds suggests that market-timing activity can, by itself, have an adverse impact on investment decisions associated with security selection. We also find that sole-managed funds typically hold far fewer stocks and more cash than team-managed funds, a finding that is further indicative of the costs associated with market timing. We next investigate whether balanced fund investors respond differently to market-timing performance depending on the management structure of the fund. Specifically, do they expect such strategies to be largely the domain of sole-managed funds? Examining the flow-to-performance relation, we find that investors do value market-timing performance, but the positive relation between market-timing performance and fund flows is largely driven by the sample of sole-managed funds. The lack of such a positive flow-to-performance relation for team-managed funds suggests that investors discount what appears to be market-timing performance as an indicator of managerial skills and future performance. This could be because it is not the strategy that the investors expect the team-managed funds to be implementing. Mutual fund companies could attempt to improve coordination within management teams by providing appropriate incentives to managers. For team-managed funds, for instance, fund companies might be able to promote coordination by rewarding managers for a fund's overall performance, in addition to rewarding their individual contributions. Further, to encourage market timing, fund companies could set performance evaluation benchmarks based on the fund's overall performance or based on explicit market-timing benchmarks to improve overall fund performance.4 It is, however, an empirical issue as to how effective management incentives might be in mitigating coordination problems with teams. Our results suggest that, despite the possible use of incentives, coordination remains a substantial problem in funds. Lastly, we explore possible explanations for the increasing trend toward team management in the balanced fund sector. In particular, we investigate whether the trade-off between specialization benefits and coordination costs has shifted. This could be caused by, for instance, a decline in the ability of funds to deliver market-timing performance or in investor demand for market-timing skills. Our evidence suggests that market-timing performance does appear to have declined in the latter half of our sample period. However, no evidence shows that investor demand for market timing has faded. Our paper contributes to several different strands of the literature. First, our paper is related to the literature on decision rights and organizational structure (Jensen and Meckling, 1992, Hart and Moore, 2005 and Hart and Holmström, 2010) as well as to the literature on the economics of teams (Becker and Murphy, 1992). We conduct a direct test of the decision rights theory in the specific context of balanced funds. We show that the investment strategy and performance of balanced funds is consistent with the location of decision rights and the trade-off between specialization benefits versus coordination costs. As predicted by the hypothesis, team-managed (sole-managed) balanced funds reallocate investments less (more) frequently across asset classes but are more (less) specialized, as reflected in their security selection performance. The management structure and allocation of decision rights in the money-managing industry has attracted considerable academic interests. Our paper is related to recent papers that study the allocation of decision rights in fund management. van Binsbergen, Brandt, and Koijen (2008) and He and Xiong (2013) theoretically examine allocation of decision rights in delegated portfolio management. Chen, Hong, Huang, and Kubik (2004) explore whether organizational structure of mutual funds affects their performance. Blake, Rossi, Timmermann, Tonks, and Wermers (2013) analyze the pension plan sponsors' choice of money managers and show a trend toward decentralization. They find that pension plan sponsors are switching from generalist managers to specialist managers and using multiple managers within one asset class. We analyze the determinants as well as the performance implications of such allocation of decision rights within a mutual fund. Our study also contributes to the mutual fund literature on market timing. While a substantial theoretical literature exists on market timing (such as Treynor and Mazuy, 1966, Merton, 1981 and Henriksson and Merton, 1981; Admati, Bhattacharya, Pfleiderer, and Ross, 1986), empirical studies on market timing typically focus on equity mutual funds and generally find little evidence of market timing.5 Our paper, however, studies mutual funds with investment objectives that encompass market-timing decisions and finds significant market-timing ability in this group of funds. More important, our analysis reveals that various factors such as organizational and management structure can have a significant impact on whether fund managers pursue market-timing strategies. The rest of the paper is organized as follows. Section 2 presents a simple theoretical framework for deriving testable predictions, and Section 3 describes the data on balanced funds. Section 4 introduces the methodologies for evaluating market-timing skills as well as provides evidence on the investment decisions and performance of team- versus sole-managed mutual funds. Section 5 studies the coordination costs in teams and also analyzes the trade-off between specialization benefits and coordination costs within the team structure. Section 6 explores the trend toward team management in the balanced fund sector, and Section 7 concludes.

Using a sample of balanced mutual funds, we study how the allocation of control rights in mutual funds affects their investment strategies. We find that sole- and team-managed funds differ with respect to both investment strategies and investment performance. Sole-managed funds exhibit significant market-timing ability while team-managed funds do not. Team-managed funds show greater security selection ability than sole-managed funds. We evaluate various alternative explanations for this evidence. Our analysis shows that coordination costs in teams, not managerial ability, explain the lack of market timing in team-managed funds. The paper presents direct evidence on the coordination costs in teams or in decentralized organizations. The results provide empirical support for the argument that specialization benefits and coordination costs jointly determine the degree of specialization and the structure of organizations (Becker and Murphy, 1992). In the context of balanced funds, the trade-off between specialization benefits and coordination costs between fund managers helps to explain the use of teams in mutual fund management, funds' investment strategies, and their performance along different dimensions. We argue that the trend toward greater use of team-managed funds can be explained, in part, by the apparent decline in the market-timing performance of sole-managed funds in recent years. Our results have implications for the management structure in the mutual fund industry. The evidence suggests that there is little joint production in teams within balanced funds because of the significant coordination costs. How important are coordination costs in fund management in the overall mutual fund industry? While it is conceivable that the coordination costs could be lower in pure equity or bond mutual funds because they hold more homogeneous assets (within the same asset class), such costs can still be substantial if the investment decisions are made jointly. Because investment research involves the processing of soft information (Stein, 2002), fund managers could have a harder time convincing others of their investment ideas, making it more difficult to incorporate the soft information in joint investment decisions. Chen, Hong, Huang, and Kubik (2004) find that sole-managed funds are significantly more likely than team-managed funds to invest in local stocks and to do better than team-managed funds at picking local stocks. Such findings are consistent with the notion that coordination costs can still play a significant role in the overall mutual fund industry.