Board Structure, Mergers and Shareholder Wealth: A Study of the Mutual Fund Industry*

Transcription

1 Board Structure, Mergers and Shareholder Wealth: A Study of the Mutual Fund Industry* Ajay Khorana Georgia Institute of Technology Peter Tufano Harvard Business School and NBER Lei Wedge Georgia Institute of Technology Draft: February 15, 25 Abstract We study mutual fund mergers in to understand the role and effectiveness of the boards of directors in these decisions. Fund mergers are more likely when funds underperform and when their boards are composed of independent trustees. This strong interaction effect is consistent with more independent boards tolerating less poor performance before initiating a merger (especially for acrossfamily mergers). The effect is most pronounced when all of the fund's directors are independent, not at the 75 percent level of independence required by the SEC. Moreover, board compensation seems related to merger likelihood. While boards approve across-family mergers that lead to substantial reductions in their own compensation, more highly paid target fund boards are less likely to approve these mergers. Other structural board characteristics (board size and independent chairs) are not strongly related to fund merger likelihoods. Post-merger fund performance and fees revert to the mean of the objective, regardless of board structure. However, these improvements do not offset the substantial pre-merger underperformance experienced by funds overseen by less independent boards. In total, these results suggest board independence primarily leads to quicker action and less tolerance for underperformance, but cannot generate superior post-merger performance. Ajay Khorana Peter Tufano Lei Wedge College of Management Harvard Business School College of Management Georgia Institute of Technology and NBER Georgia Institute of Technology Atlanta, GA 3332 Soldiers Field Atlanta, GA 3332 Boston, MA 2163 ajay.khorana@mgt.gatech.edu ptufano@hbs.edu lei.wedge@mgt.gatech.edu * We would like to thank Tamar Frankel, Raghu Rau, Henri Servaes, Russ Wermers, Tracie Woidtke and seminar participants at the Ohio State/Federal Reserve Bank of New York/JFE Conference on Conflicts of Interest in Financial Intermediaries, Tulane University, University of Tennessee and Vanderbilt University for their comments. We would like to thank the Financial Research Corporation, which supplied some of the data used in this paper. This research has been supported by the Wachovia Professorship Fund at Georgia Tech and by the Division of Research of the Harvard Business School.

2 Board Structure, Mergers and Shareholder Wealth: A Study of the Mutual Fund Industry Abstract We study mutual fund mergers in to understand the role and effectiveness of the boards of directors in these decisions. Fund mergers are more likely when funds underperform and when their boards are composed of independent trustees. This strong interaction effect is consistent with more independent boards tolerating less poor performance before initiating a merger (especially for acrossfamily mergers). The effect is most pronounced when all of the fund's directors are independent, not at the 75 percent level of independence required by the SEC. Moreover, board compensation seems related to merger likelihood. While boards approve across-family mergers that lead to substantial reductions in their own compensation, more highly paid target fund boards are less likely to approve these mergers. Other structural board characteristics (board size and independent chairs) are not strongly related to fund merger likelihoods. Post-merger fund performance and fees revert to the mean of the objective, regardless of board structure. However, these improvements do not offset the substantial pre-merger underperformance experienced by funds overseen by less independent boards. In total, these results suggest board independence primarily leads to quicker action and less tolerance for underperformance, but cannot generate superior post-merger performance.

3 Mutual fund boards have come under extensive scrutiny in connection with the mutual fund trading scandals which have surfaced in recent months. The various scandals involving U.S. mutual fund companies have elevated concerns among fund regulators and pundits regarding the effectiveness of mutual fund boards, the so-called watchdogs of shareholder interest. 1 To better protect the interests of fund shareholders, on June 23, 24 the Securities and Exchange Commission (i.e. the SEC) voted to require that Chairpersons of mutual fund boards be independent of the investment advisory firm affiliated with the fund. In addition, the SEC required at least 75% of the mutual fund directors to be independent, a term that has been defined in regulation and practice over many years. 2 Previously, the requirement had been 5%. Two of the five commissioners objected to these changes, arguing that there was lack of empirical data to support this drastic change in fund governance policies. The U.S. Chamber of Commerce also objected to these changes by filing a lawsuit opposing this change. There's no empirical evidence that an independent chairman or a 75 percent majority will have a positive effect on the performance of mutual funds, said the general counsel of the U.S. Chamber of Commerce. 3 In reality, there is some but not very much evidence that addresses whether boards with certain structural characteristics perform differently. Examining open-end fund companies from the 5 largest fund complexes, Tufano and Sevick (1997) document the effectiveness of fund boards by showing that fund fees are smaller for those funds overseen by smaller boards and boards with a larger proportion of independent directors. Del Guercio et al. (23) document similar evidence in their study of board effectiveness for closed-end mutual funds. Zitzewitz (23) examines board structure in conjunction with funds decisions to adopt fair value pricing to protect shareholders from market timing. He concludes the adoption of fair value pricing is negatively related to the percentage of insiders on the board, suggestive 1 More than a dozen mutual fund companies have paid over $2 billion in damages to settle allegations that they improperly allowed some customers to engage in rapid trading at the expense of long-term investors. 2 Under the Investment Company Act, an independent director cannot be an employee of the adviser or a member of the immediate family of an employee, be an employee or a 5-percent shareholder of a registered broker-dealer, or have any affiliation with any recent legal counsel to the fund. 3 See for the Chamber of Commerce s position. This quote comes from (visited October 11, 24) 1

4 of agency problems inherent in fund organizations. These studies look at fee setting and pricing policies because regulation empowers fund boards and more specifically independent directors with making the decisions regarding these activities. In this paper, we examine another fund decision where boards play a critical role: the decision to merge a fund out of existence. Using a cross-section of fund boards in the period , we examine the causes and consequences of fund mergers, with particular emphasis on the relationship between mergers and board oversight. With respect to the causes of fund mergers, we examine factors that make mergers more likely, including poor performance, weak fund flows and the governance characteristics of the boards overseeing the target (i.e. the acquired) funds. Shareholders might find a merger attractive if the fund is earning poor returns, and the merger could reverse this poor performance without triggering a taxable event. Fund sponsors might find a merger attractive if the fund has been losing assets, and the merger could stem those losses and increase sponsor profitability. We seek to understand whether more independent boards tolerate less poor shareholder performance before a merger is initiated. With respect to the consequences of the mergers, we examine the ex post outcomes of mergers from the perspective of fund shareholders (changes in returns), fund sponsors (changes in revenues) and fund boards (changes in board seats and compensation). In brief, we seek to understand whether these consequences are related to the governance characteristics of the merged funds boards. We look at fund mergers because they are important decisions, not taken lightly, that affect the very existence of a fund. A fund may be merged for many reasons, but prior research shows that fund mergers are often the result of poor performance (Jayaraman, Khorana and Nelling (22), Zhao (25)). Poor performance could lead a board to suggest to the fund sponsor that the portfolio manager be replaced, could lead to a decision to merge the fund out of existence, or could lead to the removal of the management company altogether. The third almost never occurs, so fund mergers are the penultimate 4 but realistically most severe remedy for poor performance. Boards that are most closely aligned with 4 The ultimate remedy is almost never observed; this would include the board replacing the management company. 2

5 shareholder interests might be the most vigilant in merging a poorly performing fund out of existence, even when this results in the loss of compensation by the target fund s board members. We study two types of fund mergers. In-family mergers lead to a combination of two funds within the same family (or sponsor or complex) while across-family mergers occur when the acquiring and target funds belong to different fund families. The underlying dynamics of approving a fund merger in these two instances can be quite different. Because a number of fund families follow a unitary board structure, i.e. the same board oversees multiple funds in the fund complex, there tends to be significant overlap between the board members of target and acquiring funds in the case of in-family mergers. In such instances, the board members, as fiduciaries to the target and acquirer, owe duties of loyalty and care to both sets of shareholders. Reconciling their separate fiduciary responsibilities could be difficult, because target fund shareholders tend to be the primary beneficiaries of these merger events at the expense of acquiring shareholders. The potential conflict of fiduciary duties is absent for boards contemplating across-family mergers, where the target and acquiring fund s shareholders are represented separately. Our work broadly relates to studies which relate various governance characteristics and firm value, e.g., Gompers, Ishii and Metrick (23). We too care about value-consequences, but focus on a particular value-relevant decision: the decision to merge a fund out of existence. We also care about the governance structures that relate to this outcome, but the legal and regulatory structure of the open-end mutual fund renders many traditional corporate governance characteristics irrelevant or absent. Thus we focus exclusively on those governance measures that vary across funds. 5 In particular, along the lines of prior research on corporate boards and mutual fund boards, we study whether fund board structure, i.e., board size, the percentage of board seats held by independent directors and the presence/absence of an independent chair, are associated with the likelihood of a merger, conditional on poor performance. The second and third of these features are the ones mandated by the new SEC rules. To our knowledge, this paper is among the first to examine the effectiveness of having an independent chair on a fund board. 5 Most of the 24 IRRC governance provisions studied by Gompers, Ishii and Metrick (23) are not relevant for our analysis, including anti-greenmail provisions, poison pills, golden parachutes, cumulative voting, special supermajority rules, etc. 3

6 Our findings suggest a nuanced story with respect to board structure, fund mergers and shareholder wealth. Consistent with prior studies, we find that mergers are more likely when the target s performance is poor, the target is small and young and its expenses are high. We also find that fund governance is statistically associated with the decision to engage in a merger. Fund mergers at least across-family mergers are more likely when the target underperforms and its board is composed primarily of independent directors, a structural characteristic of more independent and effective boards. Nominally more independent boards tolerate less underperformance (relative to other funds with the same objective) before initiating a merger. So, to the extent that mergers are reactions to poor performance, more independent boards are more likely to act, and to act more quickly, to stem shareholder losses from poor performance. The SEC has mandated that boards have 75% independent members, but our results suggest that it is boards composed wholly of independent members that seem to be most vigilant with respect to performance. Trustees approve across-family mergers even when this leads to a substantial loss of their board compensation, consistent with trustees putting the interests of shareholders before their own. But in our multinomial logistic regression analyses, we find that across-family mergers are less likely among target funds whose directors are paid more, especially for across-family mergers where trustees are likely to lose their seats and compensation as a result of the merger. This may suggest that directors private interests may come into conflict with their fiduciary duties. The above results are attributable primarily to the subsample of across-family mergers. This is not surprising given the potential dilemma faced by board members of in-family mergers (who tend to sit on both the target and acquiring fund board) who must simultaneously weigh the interests of the target and acquiring fund shareholders. Once a merger is initiated, the impact of having a more independent board is less pronounced. Generally, target underperformance is halted, but not reversed by subsequent overperformance, regardless of board structure. Once awakened, boards of all types seem to take steps that lead to roughly comparable net outcomes, i.e., they catch up to the mean fund of their type. However, given the differential levels of prior underperformance, fund shareholders who lived through the pre-period do not 4

7 enjoy comparable cumulative returns, with shareholders represented by less independent boards performing less well over the entire pre- and post-merger periods. While target funds experience outflows prior to the merger, many investors remain in the fund, possibly a reflection of behavioral biases. For sleepy investors (either due to biases or tax overhangs that lead to inertia), having a more alert board can have significant positive wealth consequences. We also study the consequences of mergers for the investment management companies and shareholders of acquiring funds. While the shareholders of acquiring funds do not enjoy any appreciable boost in performance or reduction in fees, both the manager of the acquiring fund and its board do enjoy post-merger gains. The management companies of the acquiring firms (in the case of across-family mergers) earn higher revenues after the merger and independent board members of the acquiring fund are paid higher compensation after the merger. (As we mention above, board members of target funds involved in across family mergers lose substantial director s compensation, but this is not the case for board members of in-family target funds who neither lose nor gain appreciably from a merger.) The remainder of this paper is organized in four sections. The first section provides a review of the related literature, institutional aspects of fund mergers and related hypotheses. The second section provides a description of the data and the methodology used. The third section of the paper reports our findings, including a discussion of the robustness of our results. A fourth section offers conclusions. I. Institutional Background and Hypotheses A. Fund Merger Background In the U.S., there has been a substantial increase in the number of mutual fund mergers and a modest increase in their level as a fraction of fund industry assets. Table I panel A shows that the number of mergers has roughly doubled since the early 199s, but most of this reflects an increase in the size of the industry. There are many possible explanations for the high level of fund mergers. An influential 1995 Goldman Sachs analysis (Hurley et al. (1995)) predicted a wave of fund mergers, arguing that the economies of scale in the fund industry would make consolidation a dominant strategy. Mergers could 5

8 also be strategic product management tools for fund families. Given the strong non-linear relationship between fund performance and subsequent asset flows (see Sirri and Tufano (1998)), fund families may wish to start or incubate many funds, then merge poorly performing funds out of existence. Strategic acquisitions across fund management companies could also create the need to realign the family's product portfolio. Many of these explanations would be observationally equivalent, in that we would see poorly performing or high fee funds being merged out of existence. Recent papers by Jayaraman, Khorana and Nelling (22), hereafter JKN, and Zhao (25) have studied the underlying drivers of fund merger and liquidation decisions. JKN (22) find that the likelihood of a fund merger is inversely related to fund size for both in-family and across-family mergers, but poor past performance is a significant determinant only for in-family mergers. The strong relation between the likelihood of merger and fund performance could be attributed to the desire of fund families to window dress their portfolio of fund offerings by eliminating a poorly performing fund and transferring its assets to another fund in the family. On the other hand, across-family mergers are more likely to be undertaken for strategic reasons. Mergers can have substantial consequences for fund shareholders. They can accompany a change in portfolio management personnel and objectives, and thereby affect future returns. They can accompany changes in fund fees, which in turn affect investor performance. On this latter point, it has been well established that fund fees are an important determinant of fund performance (for example, see Elton, Gruber and Busse (24)). Furthermore, mergers can influence fund fees (see JKN (22)). By either affecting portfolio returns or fees, mergers could affect shareholder wealth. We study not only wealth effects for shareholders, but for the management companies and boards as well. B. Mutual Fund Governance Mutual fund governance structures are somewhat different in law from what fund shareholders (and even scholars) may imagine them to be. Unlike financial products offered by financial institutions, such as bank deposits or certificates of deposit, a mutual fund is a corporate entity. Each and every fund is overseen by a board of directors, who then hire the service providers, which include the fund s auditor, 6

9 attorneys, distribution agent, administrator and its investment manager. Fund boards must renegotiate these contracts with the various service providers each year. While consumers think of Fidelity funds as products offered by Fidelity, under the law, the shareholders in these funds (represented by their boards) reselect Fidelity each year to manage the fund. On rare occasions, boards have exercised their right to take the management contract away from the original fund sponsor. The boards of the Navalier Funds (in 1997) and Yacktman Funds (in 1998) failed to grant the management contract to the founders for whom the funds were named, but in subsequent shareholder votes were defeated. In 22, the directors of the 4-year old Japan Fund were not satisfied with various aspects of their relationship with the existing management company and proposed to the shareholders that they sever the existing ties and award the management contract to Fidelity and the servicing and distribution contracts to SEI. Shareholders supported the board s recommendation. 6 This unusual outcome shows the difference between the corporate form of a mutual fund and the product form of a certificate of deposit. It is important to understand that these are extremely rare events, and in most cases, a board would indicate its dissatisfaction with a manager through the tone of its discussions with management, possibly through its fee renegotiations, or by the suggestion to merge a fund into another. The decision to merge two funds may or may not require formal approval from the fund's shareholders. Under Rule 17a-8 of the SEC, the target fund needs to obtain approval of a majority of its shareholders if: (1) any policy of the acquired fund that, under section 13 of the Act, could not be changed without a vote of a majority of its outstanding voting securities is materially different from a policy of the acquiring fund; or (2) the acquiring fund's advisory contract is materially different from that of the acquired fund, except for the identity of the funds as parties to the contract; or 6 See Ian McDonald, Japan Fund s Board Stages a Revolt Directors Ask Shareholders to Approve Severing Ties with Deutsche Bank, Wall Street Journal, July 15, 22, p. c17; Chuck Jaffe, Japan Fund shift shows who s boss. CBS Market Watch, November 27, 23 (visited via Factiva, 1/31/24). 7

10 (3) after the merger, directors of the acquired fund who are not interested persons of the acquired fund and who were elected by its shareholders will not comprise a majority of the directors of the acquiring fund who are not interested persons of the acquiring fund; or (4) after the merger, the acquiring fund will be authorized to pay charges under a plan that provides for use of fund assets for distribution ( rule 12b-1 plan ) that are greater than charges authorized to be paid by the acquired fund under such a plan. In practice, these rules exempt shareholders from voting only on some within-complex (i.e., infamily) mergers where the boards of the target and acquirer generally are the same. Even for mergers that must nominally be approved by shareholders, the proposal would not be advanced without the advice and consent of the board, as hostile proxy contests and takeovers are uncommon in the fund industry. Furthermore, the onus of making sure that the merger is in the best interests of the shareholders lies with the fund board. In addition to approving any mergers, the responsibilities of fund directors include (1) approval of contracts between the fund sponsor and distributor, (2) approval of fees paid to fund sponsors, (3) review of joint liability insurance policies and (4) approval of custodial contracts, among others. 7 C. Measuring Fund Merger Causes and Consequences We are interested in understanding the causes and consequences of fund mergers, and how both are related to the nature of fund governance. With respect to the causes of mergers, we seek to understand whether the unconditional likelihood of a merger, or the likelihood of a merger conditional on poor performance, is related to board characteristics. These hypothesized relationships are predicated on a few related arguments. First, mergers can be costly to implement on many levels, requiring the time and attention of boards, attorneys and the relevant fund companies. From the perspective of a board member, considering a merger proposal will likely involve extensive review of documents, deliberations and possible shareholder lawsuits. Given this effort, one might hypothesize that less effective boards (i.e., potentially those with less independence) might be less inclined to consider mergers, so the unconditional 8

11 likelihood of a merger might be lower. However, we would expect to see this effect most strongly among poorly performing funds, i.e., the effect of independence would be strongest conditional on poor performance. Furthermore, in the case of across-family mergers, board members might lose their seats (and their board fees) as a result of a merger. Given this consideration, one might hypothesize that better paid boards might be less likely to consider across family mergers. If mergers are intended to benefit shareholders of poorly performing funds, then we would expect that ceteris paribus mergers are more likely among boards which are more effective and who are therefore more willing to bear these personal costs. 8 Similarly, more effective boards would be quicker to respond to poor performance. One measure of the timeliness of fund mergers is the degree of underperformance experienced by funds prior to merger. Funds acting in the best interests of shareholders might tolerate lower levels of underperformance before acting, i.e., initiating a merger. With respect to the consequences of mergers, our goal is to document these consequences and to see if they differ based on the characteristics of the boards. While our focus on alignment is primarily concerned with whether different board structures lead to different levels of post-merger shareholder performance, we also examine the consequences from the perspectives of the investment manager and board members as well, in order to understand their incentives to undertake mergers. In particular, we examine post-merger management fees and board fees. Our goal is to better understand the causes and consequences of mergers, but in doing so to determine whether board structure in particular board structures alleged to be better aligned with shareholder interests is related to the likelihood, performance-based timing and outcome of fund mergers consistent with the fiduciary duties of the boards. 9 We examine a number of board structure variables as well as various controls. 7 See Tufano and Sevick (1997) and the Investment Company Institute's Investor Awareness Series on Understanding the role of mutual fund directors for a discussion of the legal responsibilities of fund boards. 8 The truth of this statement depends on our ability to hold all else equal more effective boards might have better performance, and be less likely to require a merger in the first place. 9 We study the fund s boards, not the board of the management company, as it is the fund s directors that owe a fiduciary duty to the fund shareholders. The fiduciary duties of the management company board to its fund s shareholders are more indirect, at best. 9

12 Makeup of the fund's board. Along the lines of recent corporate finance and mutual fund literature, we hypothesize that the effectiveness of a mutual fund board may be driven by the degree of independence of its constituent members. 1 The presence of a larger fraction of independent directors is a key measure of board independence, and one that the SEC has focused upon in its recent rule changes. Until SEC rules were recently changed, funds were required to have no less than 5% of its board composed of independent members. However one set of funds were subject to considerably more stringent provisions. Section 32 of the Glass-Steagall Act effectively barred bank officers from holding seats on the boards of funds managed by subsidiaries of the bank. It did so by restricting employees, directors, and officers of member banks from serving as employees, directors, or officers of firms that are primarily engaged in the issuance, underwriting or distribution of securities (Wills (1995)). By defining funds as firms that meet this criteria, the Act tended to make bank-funds more independent than non-bank funds. In 1999, the Glass-Steagall Act was overturned and bank officers were free to sit on their funds boards. While bank funds were no more restricted than other funds during our sample period, we keep track of bank funds separately, as they might have the legacy of an exogenously-imposed independent board structure than those of non-bank funds. In our sample, bank funds do have significantly higher percentages of their boards composed of independent trustees. The median bank fund comprises of 8% independent directors versus 75% for non-bank funds, but obviously bank funds availed themselves of the opportunity to have more diverse boards. Independence of the fund's chair. The second lever of fund governance that the SEC is using is to mandate that the chair of a fund board be an independent member. In many instances, senior personnel of the investment advisory firms are represented on the mutual fund board along with the independent members, 11 creating a potential for conflicts between the interests of the investment adviser and the fund's shareholders for the board as a whole. In addition, recent public debates suggest that the lack of 1 Under the 194 Investment Company Act, at least 4% of a fund's board members should be independent. In an effort to enhance the effectives of mutual fund boards, effective February 15, 21 the SEC increased the percentage of independent board members to 5%. 11 For example, Edward C. Johnson III, the Chairman of Fidelity Management and Research, the investment adviser for Fidelity funds, sits on the board of all Fidelity funds in his capacity as Chairman of the board. 1

13 independence of the fund's chair can jeopardize effective fund governance and consequently hurt the interests of the shareholders of the fund. We use the independence of the board Chair as a possible measure of board independence. In a hearing of the Senate Banking Committee on February 24, 24, Jack Bogle, Chairman and former CEO of the Vanguard group, stressed the importance of having an independent chair on the fund's board to better align the interests of the board with those of the fund's shareholders. He said in written testimony that despite the express language of the 194 Act that arguably calls for all of the weight to be on the side of fund shareholders, it is the manager's side of the scale that is virtually touching the ground. The CEO of Charles Schwab Corp on the other hand, indicated that his firm does not support the presence of an independent chair. 12,13 In response, Senator Jon Sununu, a committee member said that he had not seen evidence to suggest that having an independent chair would improve the board's responsibility to shareholders. Hence, using fund mergers, we explicitly examine whether the presence of an independent chair of the mutual fund board is beneficial or detrimental to the interests of the shareholders. Board size. Prior research has suggested that one measure of board effectiveness is board size, with larger boards hypothesized to be less effective. Along the lines of Tufano and Sevick (1997) and Del Guercio, Dann and Partch (23) who document that funds with smaller boards enjoy lower fees, we hypothesize that larger boards are less effective monitors, an attribute partly driven by the inherent problems associated with managing and coordinating the efforts of larger teams of individuals. Elsewhere in the corporate finance literature, larger boards have been shown to be less effective. For example, Yermack (1996) finds an inverse relation between board size and firm value. He also finds that smaller boards are more likely to initiate CEO dismissal due to poor performance. Director compensation. Higher compensation for directors could reflect greater skill and effectiveness, if compensation serves as a reward for good performance. However, lawsuits have alleged 12 Source: Securities and Investments Weekly, Week of March 1, The issue of independent board chairs is highly contentious in the fund industry. Edward C. Johnson III, the Chairman of Fidelity states in a recent Wall Street Journal editorial that: mandating an independent chairperson is akin to requiring that every ship have two captains. I don't know about you, but if a ship I was sailing on were 11

14 that high board compensation can lead to lack of independence of a fund board, because board members might be reluctant to rock the boat. (Note that a fund sponsor has no power to fire a director, with the exception of a controversial recent case in which the SEC and the fund s investment manager agreed to replace a fund board without the vote of the shareholders in conjunction with an SEC settlement. 14 ) A fund merger can result in loss of compensation for the independent board members of the target fund, hence we might think that only the most effective fiduciaries might fire themselves to protect the interests of their fund shareholders. 15 Compensation loss is less likely in the case of an in-family merger since (1) there is no reduction in total family assets, and (2) a fund board is common to multiple funds in the family (i.e., unitary board structure); hence a fund merger is less likely to cause board members to lose their board seat and consequently their compensation. We hypothesize that given the higher potential for loss in compensation in the event of an acrossfamily merger (since this would typically result in the board member losing his seat), better compensated boards of the target fund would be less willing to approve the merger. However, to the extent that better paid boards reflect higher quality board members, we might observe the opposite result. Control variables. In this section, we outline a number of control variables in explaining the likelihood of a fund merger. A more detailed explanation of the underlying rationale for these variables in explaining the merger decision can be found in JKN (22). Based on the performance hypothesis, mergers are motivated by the desire to eliminate a poorly performing fund. Hence, we hypothesize an inverse relation between fund performance and the likelihood of a merger. The performance effect is likely to be stronger for in-family mergers, where the target and acquirer are likely to be represented by the same boards and management companies. Acrossfamily mergers are more complicated and are more likely to occur for strategic reasons, where target and/or acquiring fund families are attempting to realign the set of underlying product offerings via a headed for an iceberg, I'd want one and only one captain giving orders. I'd like to know that he'd spent some time at sea and knew what he was doing and if he owned the ship, so much the better. 14 Beth Healy, Regulators Overstep with Mutual Fund Trustees, Boston Globe, July 18,

15 merger. Hence, the impact of fund performance on the merger likelihood may be a secondary rather than a primary motive for across-family mergers. Fund flows could also be an important driver of the decision to undertake a fund merger. Sirri and Tufano (1998) document that superior fund performance results in significantly higher asset flows in the subsequent period, but inferior performance does not result in asset outflows. To the extent that the lack of additional inflows causes a loss of relative market share for the fund families and higher costs for the fund shareholders, a fund merger can stem these losses. Hence, after controlling for fund performance, we argue that net asset flows into the fund can also be an important determinant of the fund merger decision: a fund with higher asset inflows (or smaller outflows) is less likely to be merged out of existence. An important rationale for undertaking a merger is the need to exploit economies of scale. If this is indeed the case, we hypothesize (consistent with JKN (22)) that the likelihood of a fund merger is inversely related to size of the fund and positively related to the fees and expenses charged to fund shareholders. If the merger does indeed lead to economies of scale, we would expect this to result in a reduction in fees and expenses in the post-merger period. 16 We also control for the number of objectives in the fund family. The number of objectives in the family is used as a proxy for family size and the extent of product diversification in the family. Larger fund families are likely to be more sensitive to poor performance. In addition, they tend to have a greater number of funds within a given objective which creates a need for ongoing consolidation of poorly performing funds. Smaller families with fewer objectives are likely to exhibit dramatically different behavior, i.e., expand the choice of product offerings by acquiring funds from outside the family. This behavior would be consistent with Khorana and Servaes (24) who document that families with a higher degree of product differentiation have higher market share. This suggests a positive relation between the 15 Typically, the non-independent members of the board are not paid for board service; rather, their compensation comes from their roles in the management company. Occasionally, individuals who are deemed not independent, say former executives, do serve as trustees and are compensated. 13

16 number of objectives in the family and the likelihood of an in-family merger and a negative effect for across-family mergers. Consistent with Tufano and Sevick (1997) we include fund age as a control variable in our regressions. They argue that younger funds may have higher startup costs. On the other hand, fund families may subsidize the fees of newly originated funds to make them attractive to investors. Hence, the ultimate effect of fund age on fee changes is an empirical issue. II. Data and Methodology A. Data We use the CRSP Mutual Funds database as of 22 to obtain monthly returns, expense ratios, loads, dollar value of fund assets under management, investment objective of the fund, fund inception dates and the merger and liquidation dates (using the last_date variable in CRSP) for funds during the period. 17 Fund management fees are gathered from the Financial Research Corporation (FRC). Information to categorize bank versus non-bank funds is obtained from both FRC and COMPUSTAT. Data for funds with multiple share classes is combined using a value-weighted approach. Mergers occurring within share classes are excluded from the analysis. Consistent with JKN (22), we categorize our sample into within and across-family fund mergers recognizing that the mechanics and motives for these mergers can be dramatically different. To keep the data collection process manageable, we gather detailed information on the underlying board structure and compensation of the board members for the acquiring funds, target funds and a matched sample of control funds (described below) from Form 485APOS and Form 485BPOS (for mergers occurring during ) with a filing date closest to the date of the merger. The following 16 In their research on mutual fund families, Khorana and Servaes (1999) examine the tendency of fund families to originate new funds. They find that one of the important determinants of fund starts is the ability to exploit economies of scale. 17 We spot checked approximately 2% of the CRSP merger dates for accuracy. The mean difference between the merger date reported in the CRSP database and the merger filing date is 15 days and the median is 9 days. Our tests treat the year of the merger as a separate period (year ), so this level of inconsistency in merger reporting dates does not affect our results. 14

17 information is gathered: the name of the trust the fund belongs to, the number of funds which are a part of the trust, the size of the board (i.e. the number of persons that sit on the board), the number of independent board members unaffiliated with the fund, the number of compensated board members, the name of each member of the board, the name of the chair, and whether the chair is independent. In addition, we gather data on the total compensation received by board members from the fund, the trust the fund belongs to, and the fund complex. 18 We also rely on the use of Morningstar data to supplement any missing data in the CRSP database. Data on industry assets are obtained from the Investment Company Institute, and on detailed objective/sector returns from Datastream and CRSP. B. Variable Construction and Methodology We utilize two basic measures of fund performance in our analysis. First, we analyze fund performance in the years surrounding the merger by compounding monthly returns to compute annual holding period returns for each of the two years preceding the effective date of the merger and for two years after the merger. Objective-adjusted annual holding period returns (OARs), defined as the annual return of the fund less the corresponding annual return of the median fund within the same investment objective, are then computed for each fund as follows: OAR i = t= 1 t= ( 1+ Ri, t ) 1 ( 1+ Ro, t ) 1 where R i,t is the return of fund i in month t and R o,t is the median return of all funds in the same investment objective in month t. These OARs measure fund performance relative to other funds in their peer group, and are computed for years 2, 1, +1, and +2 relative to the year of the merger. This performance measure implicitly adjusts for sector, industry, or style-specific factors that may exogenously affect all funds in the same investment category. Second, we also measure fund performance based on a 18 Since 1996, mutual funds have been required by law to report information on the composition of the board at least once every 18 months. Any changes occurring to the board in between that period also need to be reported via a special filing. (see Almazan et. al (24)). 15

18 single-factor and a multi-factor model. 19,2 Monthly returns for the year under consideration and the preceding year (i.e., 24 monthly return observations) are used to estimate the regression parameters. Consistent with prior studies, we measure the magnitude of asset flows in the pre- and post-merger periods by taking out the effect of portfolio returns on the size of the fund. Hence, we compute flows net of returns (Net Asset Flow i,t ), using the following approach: Net Asset Flow i,t = [Assets i, t - Assets i, t-1 (1 + R i, t )]/ Assets i, t-1 where Assets i,t is total assets in fund i at the end of year t and R i,t is the return of fund i during year t. We also compute net asset flows on an objective-adjusted basis to capture assets flows into a fund relative to the underlying flows in the median fund in the corresponding investment objective. Compensation per board member is computed as the total fund compensation divided by the number of compensated board members. In instances where this information is not available, we use the compensation of the entire trust, and divide this by the number of funds included in the trust as a proxy for board compensation at the level of the individual fund. 21 To study the relationship between the likelihood of a merger and the underlying fund-specific factors including various board level characteristics, we estimate a multinomial logistic regression specification 19 The following benchmarks are used in computing single-factor alphas: aggressive (CRSP equally weighted NYSE/AMEX/NASDAQ Index), balanced (.65*S&P 5 Index +.35*Solomon Brothers long-term, high-grade corporate bond index), corporate bond (Solomon Brothers long-term, high-grade corporate bond index), equity income (.7* CRSP value weighted NYSE/AMEX/NASDAQ Index +.3*Solomon Brothers long-term, high-grade corporate bond index), government bond (Lehman U.S. aggregate government bond index), government mortgage (Solomon Brothers government mortgage index), growth (CRSP value weighted NYSE/AMEX/NASDAQ Index), growth and income (CRSP value weighted NYSE/AMEX/NASDAQ Index), international bond (Solomon Brothers world bond index), international equity (MSCI AC World Index), municipal bond (Lehman municipal bond), small cap (CRSP first decile index of NYSE/AMEX/NASDAQ firms), specialty environment (S&P 5 environment services index), specialty finance (S&P 5 Financials Index), specialty health (.5*S&P 5 ES health care index +.5*S&P 5 pharmaceuticals index), specialty metal (S&P 5 gold index), specialty natural resources (.5*S&P 5 ES energy index +.5* S&P 5 oil and gas index), specialty real estate (national REIT index), specialty technology (.5*Goldman Sachs technology index +.5* S&P 5 ES technology index) and specialty utility (S&P 5 utility index). 2 Alphas based on a four-factor model are computed for the subsample of equity and bond funds. For equity funds, we use the three Fama and French (1992) factors: excess return on the CRSP value-weighted index, difference in returns across small and big portfolios controlling for the same weighted-average book-to-market equity in the two portfolios, the difference in returns between high and low book-to-market equity portfolios, and the momentum factor used by Carhart (1997). For bonds funds, we use the excess return on the Lehman Brothers government/corporate bond index, the excess return on the mortgage-backed securities index, the excess return on the long-term government bond index, and the excess return on the intermediate-term government bond index. These factor-model specifications are consistent with those in Blake, Elton, and Gruber (1993). 16

19 with three outcomes: (1) no merger, (2) an in-family mutual fund merger and (3) an across-family mutual fund merger. Our sample includes a control group that did not experience a merger during this period. For each target fund in the merger sample, we construct a randomly chosen objective-matched control sample after excluding those funds which experienced a merger or liquidation in the -2, +2 years surrounding the merger. We also ensure that the control fund does not come from the same family as the target or acquiring fund. The probability of engaging in a merger is estimated as a function of a number of fund characteristics (performance, fund size, asset flows, expense ratio, fund age), family/complex characteristics (size of complex or number of objectives covered by funds in the complex) and fund governance characteristics. Performance is measured as the annual holding period return of the fund in excess of the median return for all funds in the same investment objective (measured over the year -3, -1 window relative to the merger), and based on alphas obtained from a single-factor and a multi-factor model. Asset Flows are the objective-adjusted net asset flows in a fund. Fund Size is the log of total net assets for the fund. Expense Ratio is measured as the fund's expense ratio minus the median expense ratio for all funds in the same investment objective. Fund Age is the logarithm of fund age computed in years. Number of Objectives in the Family is measured as the number of investment objectives in a family in the year of the merger. Family size is the log of total net assets managed by the complex. Board size is the number of directors on the board of the fund. Compensation per board member is average compensation earned by a board member from the fund over the course of a year. Independent chair is a dummy variable equal to one if the chair is classified as independent of the fund. Interaction effects are included to ascertain the marginal impact of board characteristics and fund performance. Finally, in some specifications we include in a dummy variable for funds offered by banks, which might have had the legacy of independent boards as a result of the Glass-Steagall Act. 21 Funds with similar investment objectives within a fund family are typically placed in a trust. For instance, in 2, Fidelity had 287 funds placed in 53 different trusts. 17

20 In theory, each fund has a separate board, but in practice the same board members often oversee multiple boards within one complex. In theory, they should make decisions about each fund separately. In practice, they may not. To account for the possibility that merger decisions involving multiple funds within a complex may not necessarily be independent, we estimate both multinomial and clustered logistic models. The clustered models take into account this possibility and adjust upwards the standard errors of the coefficients. While mergers within and across families may be legally alike, in practice they may be quite different. For in-family mergers, the target and acquirer both use the same management company, and usually have the same board. This would not be true for across-family mergers. Therefore, we analyze these two types of mergers separately, and in addition, we also examine the likelihood of an infamily versus an across-family merger conditional on the fact the decision to merge has been made. To understand the consequences of mergers, we examine a number of measures. To capture the economic impact on the acquiring fund sponsor, we examine the change in management fees before and after the merger. We decompose the fee change into (1) the percentage change in assets under management and (2) the percentage change in management fees charged (as a fraction of AUM). To capture the economic impact on the boards, we calculate the trustees fees received by the pre-merger board members before and after the merger, both for the acquiring and target boards. To examine the economic impact of mergers on fund shareholders, we measure fund performance before and after the merger. We study the level of annual post-merger performance and the improvement (or deterioration) in performance relative to the pre-merger level, and relate it to various board level characteristics. First, we examine whether the objective-adjusted fund performance in year +1 is statistically different across various subgroups of target funds classified based on the percentage of independent board members, and the presence/absence of an independent chair on the board. Second, we compare the post-merger fund performance across the sample of target funds and our objective-matched control sample. Third, we compute cumulative objective-adjusted returns for the target funds starting in year -3 and ending in year +1 to study the relationship between board structure and the longer run effect of mergers on shareholder wealth. 18