This version: March 23, 2015  Abigail S. Hornstein and Elif Sisli Ciamarra   We examine a unique characteristic of mutual fund governance: the role of overlapping 

advertisement
This version: March 23, 2015 Director Overlaps in Mutual Fund Families Abigail S. Hornstein and Elif Sisli Ciamarra* ABSTRACT We examine a unique characteristic of mutual fund governance: the role of overlapping boards of directors within mutual fund families. We find that overlapping boards are a mixed blessing. At first glance, such governance arrangements appear to benefit investors: they are associated with higher returns and higher managerial skill. However, we present evidence that such board structures work to benefit fund families rather than individual funds for which they have fiduciary responsibilities. Window dressing and strategic performance transfer from low‐
fee funds to high‐fee funds occur more often. In addition, families that have higher board overlap tend to have higher marketing fees. As a result of these coordinated actions, mutual fund families with greater degrees of board overlap enjoy larger asset inflows. *
Abigail S. Hornstein, Wesleyan University, Middletown, CT, ahornstein@wesleyan.edu; Elif Sisli Ciamarra, Brandeis University, Waltham, MA, esisli@brandeis.edu, corresponding author. We thank Daniel Bergstersser, Kathryn Graddy, Jens Hilscher, Aldo Musacchio, Debarshi Nandy and the participants at the Brown Bag Seminar in Brandeis International Business School. Hornstein thanks the Mellon Foundation for financial support. We also gratefully thank Deniz Civril, Yubing Cui, Eugene Kisselev, Geofrey Lordi, Brian Lau, Alex Moris, Jimmy Ong, Richa Sahay and Na Wang for excellent research assistance. All errors are our own. 1. Introduction The majority of mutual funds in the U.S. are set up by mutual fund sponsors, also referred to as mutual fund families. When a fund is initially incepted, its board of directors (trustees) is appointed by the fund sponsor that launches the fund. The board of the newly launched fund is composed of the same directors that serve the rest of the funds in the fund family. Therefore, a dominant board structure in the U.S. mutual fund families emerges: a common set of directors serving simultaneously on boards of multiple funds. For example, Fidelity's bond, money market, and asset allocation funds are seen by one board of directors, and another board oversees its equity and high income funds. Fidelity is not an exception in the mutual fund industry. The Investment Company Institute (ICI) reports that the majority of the mutual funds in the U.S. possess a unitary board structure, where a single board governs all of the funds operating under the fun family’s umbrella. The rest of the funds follow a cluster board model, where a few boards oversee multiple funds within the family like in the case of Fidelity (ICI, 2009; ICI, 2012). The main rationale for the prevalence of the overlapping board structure within the mutual fund industry is the presumption that it leads to economies of scale. For example, The Independent Directors Council Task Force Report on Director Oversight of Multiple Funds (May 2005) concludes that “mutual funds within a fund family share the same investment adviser and other key service providers and, as a result, significant efficiencies are realized when a single or limited number of boards oversee all of the funds.” It is also frequently contended that cluster boards that negotiate with fund service providers are able to drive down the funds’ expenses due to the increase in their bargaining power when they negotiate for multiple funds (Kong and Tang, 2007). The economies of scale and bargaining power arguments imply lower costs for mutual fund investors, providing support for overlapping board structures within fund families. However, the overlapping board structure is not free of criticism. One concern is that busy boards may not be effective monitors (Fich and Shivdasani, 2006). The workloads of mutual fund directors, particularly those that advise and monitor a significant number of funds, can be substantial because each fund has its own lengthy prospectus, regulatory filings and compliance issues to review. John Bogle, the founder and retired CEO of Vanguard, said “The required reading underscores the challenge. Mutual fund directors are either not being paid nearly enough for what they should be doing—or far too much for what they actually do.”1 In line with such concerns, Ferris and Yan (2007) present evidence that director busyness is associated with higher mutual fund fees. Another concern is that director overlaps may exacerbate the agency conflicts between mutual fund investors and fund families. Each fund that operates under the umbrella of a mutual fund family is a separate legal entity with its own set of investors, and the different classes of each mutual fund attract different types of investors. As each fund may have different objectives and dissimilar investors, it may be difficult for a single board to serve simultaneously and effectively the interests of the investors of each fund within a family. Moreover, a single board overseeing multiple funds may make it easier for mutual fund families 1
“Is Your Fund’s Board Watching Out For You?,” available at http://online.wsj.com/articles/SB10001424052702303753904577450243418998540. 2 to strategically transfer performance across member funds to favor those funds that are more likely to increase overall family profits (Gaspar, Massa and Matos, 2006).2 Given that the overlapping board structure may offer both benefits and costs, it is unclear if it is ultimately in the interest of mutual fund investors for boards to advise and monitor multiple funds within a family simultaneously. Board overlaps in mutual fund families has attracted attention of the media and legal experts3, but the academic research on the topic remains limited. In this paper, we perform a comprehensive analysis of the impact of overlapping boards within the mutual fund industry. We present evidence for the relationship between board structure and mutual fund characteristics such as fees, returns, net asset flows, as well as hidden actions of fund managers such as window dressing and cross‐fund subsidization. Our sample of mutual funds consists of 11,302 domestic U.S. equity mutual fund‐classes, which belong to 332 distinct fund families. The mean (median) fund family in our sample operates 164 (123) funds. We build a unique dataset of mutual fund directors using the certified shareholder reports and prospectuses filed by mutual funds that are available at the SEC Edgar database, and develop measures for the extent of director overlap in mutual fund families. Our first measure captures whether the fund family has a unitary board structure (i.e., complete overlap) as in Kong and Tang (2008). 60 percent of the funds in our sample of equity funds belong to families that follow a unitary board structure. 2
The SEC explicitly mandates that that mutual fund boards must monitor to guard against cross‐subsidization. These rules can be viewed at https://www.sec.gov/rules/final/finend.txt. 3
For example, see “On Board, at a Mutual Fund,” Wall Street Journal, September 3, 2014, available at http://www.wsj.com/articles/on‐board‐at‐a‐mutual‐fund‐1409757187. 3 Mutual fund families that do not have a unitary board structure still exhibit a considerable overlap of directors serving on the boards of their individual funds.4 Therefore, we develop additional measures that capture the extent of director overlap. We calculate the percentage of funds in a family that each individual director oversees, and obtain the average of this ratio across all directors in each individual fund. We name this variable the “Director Overlap Ratio.” For example, the Fidelity Blue Chip Value Fund has a director overlap ratio of 0.95, meaning that on average, the directors of this fund oversee 95 percent of all funds within the Fidelity family. Recognizing that some funds may be more important to a particular fund family because they manage larger assets, we also calculate an “Asset‐weighted Director Overlap Ratio,” where we express each director’s overlap by weighting the percentage of a family’s assets that he/she oversees. Continuing with our example, for the Fidelity Blue Chip Value Fund, the asset‐weighted director commitment ratio is 0.98, meaning that on average, the directors of the fund oversee 98 percent of the fund family’s assets. This measure is similar to Tufano and Sevick (1997)’s “board concentration” measure. We start our analyses by investigating the relationship between director overlap and mutual fund fees. If the overlapping board structure offers economies of scale and bargaining advantages with the fund service providers, and if these advantages are passed on to the fund investors, then there should be a negative relationship between measures of director overlap 4
Through talking with practitioners in the industry, we have learned that a departure from the unitary board structure occurs randomly. For example, in some instances a fund may need additional expertise on its board that no other director possesses. Or, it may need someone to assume an additional leadership role (e.g., leading the audit committee) that no current board member is willing to undertake. In other instances, a board member retires and is replaced by another director, but not all funds in the same family stand for election in that particular year. Finally, sometimes, the departure from the unitary board structure occurs as a result of a merger with another fund. 4 and fund fees. We use three measures of investor costs: expense ratios, total fees and marketing and distribution (12‐b1) fees. We present evidence for a negative relationship between board overlap measures and expense ratios and total fees – that is, funds that belong to mutual fund families with higher degrees of director overlap charge lower fees to their investors. This result is consistent with the findings of Tufano and Sevick (1997) and Kong and Tang (2008). However, marketing and distribution fees (12b‐1 fees) do not appear to be lower in the presence of higher director overlap. 12b‐1 fees include fees paid for marketing and selling fund shares, such as compensating brokers and others who sell fund shares, and paying for advertising, the printing and mailing of prospectuses to new investors, and the printing and mailing of sales literature.5 These fees have been criticized as being the least transparent cost component for mutual fund investors, and the SEC asked whether they result in “investors overpaying for services or paying for distribution services that they may not even know they are supposed to be getting.”6 It has also been shown that mutual fund companies selectively advertise their better‐performing funds (Koehler and Mercer, 2009). Next, we investigate the relationship between board overlap and returns to the fund investors. We analyze returns net of fees, gross returns, as well as fund alphas. We find no significant correlation between director overlap and fund returns.7 This evidence suggests that 5
A detailed description of these fees can be found at http://www.sec.gov/answers/mffees.htm. 6
On July 21, 2010, the SEC proposed “Measures to Improve Regulation of Fund Distribution Fees and Provide Better Disclosure for Investors.” See http://www.sec.gov/news/press/2010/2010‐126.htm 7
Ding and Wermers (2012) state that unitary boards affect fund performance negatively. However, they define “unitary board” differently ‐ as having at least two board members who are on at least one other fund board within the same management company. This may be a reason behind the differences in our results as their term “unitary board” is more similar to all of our other measures that capture incomplete board overlaps within a fund family. 5 lower fees are not passed onto fund investors in the form of higher net returns. The results we have summarized so far relate to the differences between funds with different director overlaps, but do not necessarily compare the funds that operate under the umbrella of the same fund family. Thus, next, we ask a related but different question that has not been investigated in the extant literature: If two funds that operate in the same family have different degrees of director overlap, are there differences in their fees and returns? We believe this is a more important question to answer from the fund governance perspective. A well‐functioning board should look after the fund’s investors and put the interest of the fund family before the interests of the individual fund’s interests. Recognizing this, the SEC rules that “ … the board should focus, among other things, on the relationship among the (fund) classes and examine potential conflicts of interest among (fund) classes regarding the allocation of fees, services, waivers and reimbursements of expenses, and voting rights.”8 To compare the funds with the rest of the funds in their families, we include family fixed effects in our regressions. If overlapping directors provide economies of scale, then we expect to see lower fees for funds that share more directors with the rest of the funds in the family (i.e., higher director overlap). Once we control for family fixed effects, we find no significant relationship between director overlap measures and fund expense ratios / total fees. The mutual funds that share more directors with other funds in their fund families do not seem to pass on to their shareholders any of the supposed cost savings that may stem from economies of scale and/or greater bargaining. Furthermore, we find that 12b‐1 fees are significantly 8
https://www.sec.gov/rules/final/finend.txt. 6 higher for funds with greater degrees of director overlap when compared to the rest of the funds in the family. We continue with studying the relationship between fund returns, net asset flows and director overlap within fund families. We find evidence for a positive relationship between director overlap and fund returns when we include family fixed effects. These results hold for both net returns and gross returns. Net asset flows, too, are positively related to our measures of director overlap. We recognize that the higher returns may reflect greater managerial skill, or it may stem from window‐dressing and/or cross‐fund subsidization. If we find evidence for managerial skill, then we can argue that overlapping boards are effective governance structures. However, if we find evidence for window dressing and cross‐fund subsidization, overlapping boards need to be approached with caution. Following Kacperczyk, Sialm and Zheng (2008), we construct the return gap to gain insight into whether the overlapping boards are able to help the funds employ more skilled managers. However, we observe a consistently insignificant relationship between the return gap (proxy for managerial skill) and director overlap in our regressions. This result does not support the claim that the unitary boards attract more skilled. We then look into window dressing. Agarwal, Gay and Ling (2014) develop a measure that captures the extent to which window dressing occurs at a fund – backward holdings return gap. Using Agarwal et.al.’s (2014) measure, we find a significant and positive relationship between director overlap and window dressing. 7 Another detrimental hidden action may be strategic performance transfer from one fund to another. Gaspar, Massa and Matos (2006) show that mutual fund families strategically transfer performance across member funds to favor those more likely to increase overall family profits. Families routinely charge different levels of fees on various member funds within the family, and also on different classes of each of those funds. Since investors tend to steer new monetary inflows to funds with higher past performance, mutual fund families may strategically transfer assets to those funds that have greater income generating potential for the family. Therefore, one example of strategic performance transfer would be from low‐fee funds to high‐
fee funds within a family. Overlapping boards may make it easier to steer investment opportunities to star funds that are attracting greater investor inflows or that generate more income for the fund family. For this reason, we study strategic performance transfer by closely following the empirical strategy of Gaspar, Matsa and Matos (2006), and find that this practice occurs more often when the high‐fee funds have higher board overlaps. Interestingly, we do not observe strategic performance transfer when the low‐fee funds have less director overlap with the rest of the funds in the family. We interpret this result to mean that having a director that is independent of the rest of the funds in a family protects the interest of shareholders of the low‐fee funds. To summarize, these results present a picture that overlapping boards appear to be a benefit for mutual fund investors as costs are unaffected while returns may be higher. However, by examining the hidden actions of funds, we present evidence for window dressing and cross fund subsidization at funds with greater degrees of director overlap. The cluster board structure appears to serve the interests of the fund family. Given that the SEC has 8 mandated that boards look out for the interests of investors in specific fund classes, there is now a clear mismatch between the board’s raison d’etre and fund performance. Interlocking directorates may include directors who have a greater sense of duty towards the management company that compensates them and not to the shareholders they represent. Our research contributes to the literature on mutual fund governance. Most of the attention in this literature has been on the independence of the board of directors, and we are only aware of two studies that examine the effects of board overlap. Kong and Tang (2008) study the impact of unitary boards, and they present evidence for a negative relationship between fund expenses and unitary boards. Since their measure of board overlap does not vary between funds in a fund family, they are only able to compare funds in families with and without complete board overlap. Tufano and Sevick (1997) is more similar to our study in that they develop several measures that capture incomplete overlap between boards within a single family. We extend their analysis to show that unobserved family heterogeneity changes the effects of the board overlap. Our significant contributions to the literature are to illustrate how different degrees of overlap impacts fund characteristics by developing multiple nuanced measures of board overlap within a fund family, and to include family fixed effects to capture previously unobserved fund family heterogeneity. 2. Data 2.1.
Sample Formation The sample consists of U.S. equity mutual funds with Lipper asset codes marked as “EQ” in the Center for Research in Securities Prices (CRSP) Mutual Fund Database in 2007. This database covers U.S. open‐end mutual funds and provides information on fund characteristics including 9 returns, total net assets, fees, and investment objectives. Since our focus is actively managed mutual funds, passively managed funds such as index funds are excluded from the sample. For each individual fund, we manually identify its ultimate fund family9. The final dataset contains information on 11,730 individual fund‐classes belonging to 332 mutual fund families. We collect the board data at the fund‐class level, because the SEC explicitly mandates that the mutual fund directors have fiduciary responsibilities for shareholders at the fund‐class level not the fund itself.10 2.2.
Database of Mutual Fund Directors We use publicly available certified shareholder reports (N_CRS) and prospectuses (485BPOS) to build a unique database of directors for each fund‐class in 2007. These reports are available at the SEC Edgar database. For each individual mutual fund we record the names of directors, their tenure on the board, whether the person is CEO, their independence status, the size of the board, and whether the chairman of the board is independent of the fund management company. As per the Investment Company Institute (ICI) we consider a director to be independent if the individual has not had a significant business relationship with the 9
Kuhnen (2009) provides detailed description of the fund family structure and we follow her methodology in identifying the ultimate fund families. 10
The SEC has stated, “Consistent with its oversight of the class system and its independent fiduciary obligations to each class, the board must monitor the use of waivers or reimbursements to guard against cross‐subsidization between classes. In making its findings, the board should focus, among other things, on the relationship among the classes and examine potential conflicts of interest among classes regarding the allocation of fees, services, waivers and reimbursements of expenses, and voting rights.” The complete text can be accessed at https://www.sec.gov/rules/final/finend.txt. 10 fund’s adviser, distributor or affiliates for at least two years, and does not own any stock of the investment advisor or certain related entities. 2.3.
Measures of director overlap in mutual fund families We develop five measures of the director overlap within a fund family. We now describe these measures in detail. In Appendix 1, we provide an example for how these board overlap measures are constructed. 2.3.1. Unitary Board Our first measure, unitary board, is an indicator variable that takes the value one if all of the individual funds in a fund family are overseen by the same group of directors (i.e. complete director overlap) and zero otherwise. This is the measure that is employed in Kong and Tang (2008). 60 percent of the fund‐classes in our sample had a unitary board structure in 2007. 2.3.2. Director Overlap Ratio Fund families that do not have unitary boards still exhibit a significant director overlap, with common directors serving on boards of multiple fund‐classes within the family. The next four measures are formulated to quantify the degree of director overlap for these funds. We first count the number of fund‐classes in the fund family that each individual director oversees and then scale the sum by the total number of fund‐classes in the family. We then obtain the average value across all directors in each mutual fund. Formally, for each fund‐
class with N individual directors, we estimate: 11 1
This variable corresponds to the average percentage of funds in the family that the directors of an individual fund oversee. Note that for unitary boards, this measure would be one. The mean value of “Director Overlap Ratio” is 0.90 in our sample, meaning that the directors of a fund, on average, serve together on 90% of all boards within their families. We note, however, that there is considerable heterogeneity among this variable: the minimum value of this variable is 0.47%. We also calculate the average Director Overlap Ratio for each fund family, “Director Overlap Ratio (Family)”. This is accomplished by first calculating the director overlap ratios for individual fund classes in a family, and then taking the averages. This family‐level board overlap measure has an average value of 90.3%, and is highly correlated with the director overlap ratio. 2.3.3. Asset‐weighted Director Overlap Ratio This measure focuses on the monetary value of the fund assets that are overseen by a director to capture the importance of the funds a director oversees within the fund family, since funds that attract larger assets might be more valuable to the fund families. We sum up the assets of funds in the fund family that each individual director oversees and scale it by the total assets of funds in the family, and then average the measure across all directors in each mutual fund. Formally, for each fund‐class with N individual directors, we estimate: 12 1
∑
This variable is estimated at the fund‐class level and corresponds to the average percentage of assets in the family that the fund’s directors oversee. Note that for unitary boards, this measure would be one. This measure is similar to Tufano and Sevick (1997)’s measure “board concentration.” The mean (median) value of this variable is 90% (100%). Finally, we look at the same concept of value‐weighting each director from the perspective of the fund family. The measure is estimated by averaging the asset‐weighted director overlap ratios across all funds within a fund family. This measure is similar to Tufano and Sevick (1997)’s measure “sponsor concentration.” The average value is 90.4% which is slightly higher than the raw frequency with which directors serve together on multiple boards within the family. This reflects both the fact that directors typically serve simultaneously on the boards of all classes of a particular mutual fund and that larger funds often have more classes, and that larger fund families are more likely to have overlapping boards. This variable has a mean (median) value of 90% (100%). 2.4.
Fund Characteristics We obtain data on fund characteristics from CRSP Mutual Funds Database for 2007‐
2010, including net asset values, expense ratios, front and rear‐load fees, net returns and portfolio holdings. We list the variable names and descriptions in Appendix II. In Table 2, we present the descriptive statistics for fund characteristics. 13 Fund size is measured as the total net assets under management (mtna). We calculate the average monthly total net assets over a calendar year for each fund. Fund family size is the total net assets under management of all funds within a mutual fund complex. Both fund and fund family size capture possible economies of scale as the size of a fund may affect its ability to make purchases or sales with minimal impact on market price (Ferris and Yan, 2007). The mean (median) size of a mutual fund is $431 mn ($32mn). Total asset inflows are measured as the annual change in total assets under management. The mean (median) value of net asset inflows is ‐$14mn ($0mn). We examine separately three types of fees: expense ratios, 12b‐1 or marketing and distribution fees, and total fees. First, the expense ratio is the ratio of total operating expenses to assets under management. We find that the sample mean (median) expense ratio is 135.18 bp (130 bp). Second, the mean (median) 12b‐1 fee is 58.91 bp (50.00 bp). Finally, to calculate the total fees faced by a representative investor in a mutual fund class, we add the annualized front and rear‐end loads to the expense ratios. Following Sirri and Tufano (1998), we assume that the average investor remains invested in the fund for a period of seven years. Higher fees are consistent with the management retaining a higher fraction of fund income and inflows (Kuhnen, 2009). The sample mean (median) total fees is 145.69 bp (144bp). To calculate the net returns, we calculate fund performance in year t by compounding monthly returns from CRSP (mret) over the entire year. Net returns are net of all management expenses and 12b‐fees, as well as front and rear load fees. The sample mean (median) fund 14 generated returns of 520 bp (1,217 bp). Gross returns, which are defined as net returns plus total fees, have a mean (median) value of 443 bp (1,168 bp). We estimate alpha using a four factor regression analysis of monthly data as this dataset consists exclusively of equity funds. We use the three Fama and French (1992) factors – excess return on the CRSP value‐weighted index, difference in returns between a small and large stock portfolio, and difference in returns between a high and low equity to book market portfolio – and we also include the Carhart (1997) momentum factor. Thus, the estimated alpha is a measure of the annual abnormal return associated with each mutual fund. The mean (median) value of alpha is 2.20 basis points (20.42 bp). The return gap is a proxy for managerial skill and is estimated as per Kacperczyk, Sialm, and Zheng (2008) and Agarwal, Gay, and Ling (2014). It is measured as the difference between a fund’s actual performance from the performance of the fund’s previously disclosed portfolio on the assumption the previously disclosed portfolio was actually held throughout the quarter. The mean (median) value of the return gap is 0.0003 basis points (0.0001 bp). The window dressing measure is based on Agarwal, Gay and Ling (2014). It is measured as the difference between the returns that would have been generated by the reported quarter‐end portfolio had it been held throughout the quarter and the fund’s actual performance. The mean (median) value of the window dressing measure is 0.0022 basis points (0.0011 bp). 3. Director Overlap and Mutual Fund Fees 15 We first investigate the relationship between the extent of board overlap in mutual fund families and mutual fund fees charged to investors. One of the main responsibilities of the board is to negotiate fees. Hence, lower fees may signal greater board effectiveness (Del Guercio et al., 2003). More importantly, unitary board structure is thought to be associated with lower fees due to the economies of scale and bargaining power advantages that such structure may provide (ICI (2009); Kong and Tang, 2007). To examine the relation between board overlap and mutual fund fees, we estimate the following equation: ,
.
,
[1] We employ three separate measures for fund fees: the expense ratio, 12‐b1 fees and total fees. The control variables include the size of the fund and the fund family, age of the fund and the fund family, and two other board characteristics that have been show to affect mutual fund fees – board size and board independence and CEO‐chairman duality. Fund and family size control for possible economies of scale, and have been shown to be inversely related to fund fees (e.g. Khorana, Servaes and Tufano, 2008; Kuhnen, 2009). The fund’s age is used to capture the lifecycle effect whereby the needs of a fund vary predictably according to its age (Tufano and Sevick, 1997; Del Guercio, Dann, and Partch, 2003; Ferris and Yan, 2007). Year and Lipper investment objective fixed effects are included in the regressions. We estimate Equation 1 using weighted least squares (WLS) estimations. Some funds have just one class while most funds have many classes. The weights used in our estimations are the ratio of number of classes of a focal fund family to the total number of fund classes 16 offered by any fund family. Using these weights mitigates potential heteroskedasticity. Standard errors are clustered to control for possible correlation across observations belonging to the same fund family. 3.1.
Expense Ratios Expense ratio represents the percentage of assets deducted for fund expenses and measures the annual expenses of a fund. These expenses include the management fees, administrative fees, operating costs, 12‐b1 fees (marketing and distribution costs), and all other asset‐based costs incurred by the fund. Controlling for observable characteristics that correlate with how difficult it is to operate the fund (such as its size and investment objective), a higher expense ratio indicates that more of the rents are captured by the management, and less by fund investors (Kuhnen, 2009). The results are presented in Table 2. We begin by examining the impact of a fund family adopting a unitary board structure for all funds within the family (Column I). The results indicate that expense ratios are lower when a fund family has greater director overlap. If a fund has a one standard deviation increase in the director overlap ratio, it will experience a decrease in total fund fees of ‐26.79*0.20 or 5.36 basis points. Given that the average fund expense ratio is 135 basis points, this would correspond to a 4 % decrease in expense ratios. 3.2.
Total Fees Next, we investigate the relationship between total fees – expense ratios plus total load fees – and board overlap. The results are presented in Table 3, and are parallel to the results 17 reported above about expense ratios ‐ that total expense ratios are lower when a fund has greater director overlap with the rest of the funds in its fund family. If a fund has a one standard deviation increase in the director overlap ratio, it will experience a decrease in total fund fees of ‐30.13*0.20 or 6.03 basis points. Given that the average fund fees are 154 basis points, this would correspond to a 3.9% decrease. 3.3.
Marketing and Distribution Costs (12‐b1 fees) We next analyze the relationship between board overlap and marketing and distribution fees (12b‐1 fees). 12b‐1 fees include fees paid for marketing and selling fund shares, such as compensating brokers and others who sell fund shares, and paying for advertising, the printing and mailing of prospectuses to new investors. These fees have been criticized as being the least transparent cost component for mutual fund investors, and the SEC asked whether they result in “investors overpaying for services or paying for distribution services that they may not even know they are supposed to be getting.” It has also been shown that mutual fund companies offer biased snapshots of their success by selectively advertising their higher performing funds (Koehler and Mercer, 2009), and that funds that advertise attract 20% more new inflows than comparable funds that do not advertise (Jain and Wu 2000). Therefore, in order to attract more asset inflows, fund families may find it beneficial to engage in selective advertising, but at the cost of funds’ existing investors. We present these results in Table 4. We find that the 12b‐1 fees are lower for funds that operate in fund families with greater director overlap (Columns I, III and V). For example, the coefficient on the unitary board dummy is ‐12.78 and is significant at the five percent level. 18 According to this coefficient, investors of funds that operate in families with a complete director overlap pay 12.78 basis points less in 12b‐1 fees. However, the coefficient on the director overlap ratios in columns II and IV are insignificant. Summary The findings so far about mutual fund fees are in general consistent with the conventional wisdom that overlapping boards can identify and exploit economies of scale and possess bargaining advantages. Kong and Tang (2007) present similar results. We add to the prior literature by analyzing the 12b‐1 costs separately. These are costs that are not very visible to the fund investors. Our results indicate that the mutual fund investors appear to be cognizant of the total costs associated with investing in a particular mutual fund but may not be sensitive to the breakdown of costs. 4. Director Overlap and Mutual Fund Performance In this section, we analyze fund performance. Overlapping boards may serve as a channel for conveying information to managers, or may possess bargaining advantages that enable the family to attract higher skill managers. If overlapping boards deliver informational or bargaining advantages to funds, then there should be a positive association between fund performance and our director overlap ratios. We estimate the following equation to examine the relationship between fund performance and director overlap: ,
.
19 ,
[2] We use three measures for fund performance: net returns, gross returns and fund alphas. Net and gross returns capture the year‐on‐year changes in mutual fund value with and without consideration of fees paid by the representative investor. Fund alphas capture fund performance on a risk‐adjusted basis and are calculated using Carhart’s (1995) 4‐factor model. We do not analyze style‐adjusted fund returns in either specification as per Gormley and Matsa (2014). In Table 5, we present the results for net returns. The evidence in the previous section and in the extant literature (Kong and Tang (2007)) suggests that overlapping board structure is associated with lower costs. If these cost savings are passed on to the mutual fund investors as a result of an effective governance mechanism, then director overlap would also be associated with higher net returns. However, we find that the coefficients on director overlap measures are not statistically significant, which runs contrary to this hypothesis. In Tables 6 and 7, we analyze the relationship between gross returns and alphas and our board overlap measures. Once again, we do not observe any significant relationship between gross returns and board overlap (Table 6) and alpha and board overlap (Table 7). These results are surprising in light of the widespread belief that the board may serve as a conduit for information. 5. Within‐Family Differences The results we have presented so far relate to the differences between funds with different degrees of director overlaps in general. However, they do not necessarily compare the funds that operate under the umbrella of the same fund family. Thus, next, we ask a related but different question that has not yet been investigated in the extant literature: If two funds that 20 operate in the same family have different degrees of director overlap, are there differences in their fees and returns? We believe this is a more important question to answer from the fund governance perspective. A well‐functioning board should look after the fund’s investors and put the interest of the fund family before the interests of the individual fund’s interests. Recognizing this, the SEC rules that “ … the board should focus, among other things, on the relationship among the (fund) classes and examine potential conflicts of interest among (fund) classes regarding the allocation of fees, services, waivers and reimbursements of expenses, and voting rights.” To compare the funds with the rest of the funds in their families, we include family fixed effects in our regressions. We note that once the family fixed effects are included in the regressions, we can only examine the two measures of director overlap that are estimated at the fund level (i.e., paralleling the results reported in Columns II and IV in Tables 2‐7). 5.1 Fees We start by analyzing fund fees. If the economies of scale and bargaining advantages are present, then we expect to find that the fees for the fund with a greater degree of director overlap with the rest of the fund in the family are lower. The difference in this regression equation from Equation [1] is the inclusion of family fixed effects. By including the family fixed effects, we are comparing two funds within the same family and answering the following question: If two funds within the same family have different degrees of director overlap, does the fund with higher overlap enjoy greater economies of scale, which is translated in lower fees? 21 ,
.
,
, [1’] and use three measures of fund fees as the dependent variable: expense fees, total fees, and 12b‐1 fees. First, we observe that there is no significant relationship between expense fees and director overlap. It could be that once we control for unobserved heterogeneity, there are no economies of scale that could be transferred to shareholders as lower fees. Alternatively, the lack of significance may mean that the economies exist but are not being shared among all of the funds, or fund classes, within the family. In any case, these results cast doubt on the argument that unitary boards are favorable for mutual fund investors. Next, we identify no significant relationship between total fees and board structure. Thus we conclude that mutual fund investors do not face different fee structures due to the nature of director overlaps within the fund family. Finally, we then observe that 12b‐1 fees are significantly higher when there is higher director overlap. These results suggest that boards of directors may approve higher 12‐b1 fees whenever the directors are cognizant of how the funds might be used by other funds within the family, given their inside knowledge of the needs of other funds. All of these results paint a consistent picture: when we control for family heterogeneity by including family fixed effects, we observe that the impact of board ownership changes due to the fact that directors may now be cognizant of current or future needs of other funds within the family. 5.2 Performance 22 To examine the relationship between fund performance and director overlap, we estimate the following model: ,
.
,
[2’] Once we control for family fixed effects, we observe a highly positive and statistically significant relationship between net returns and board overlap. However, when we control for family heterogeneity to capture the possibility that directors are a conduit of information between funds within a family or are able to negotiate preferential cost structures for the fund, we identify a highly positive and significant impact of board structure upon gross returns. Thus, we are able to observe that boards of directors can have both a direct and an indirect effect upon the performance of the fund with the direct effect working through the fees channel and the indirect effect occurring through information transmission. Alternatively, when family fixed effects are included in the model, the impact of family size disappears entirely while board size becomes positively associated with higher alpha. 5.3 Unobserved heterogeneity These results strongly suggest that each fund family has adopted policies that affect all constituent funds through multiple channels. We therefore believe it is appropriate to include family fixed effects in all subsequent analyses in order to ensure that all comparisons reflect the variation in fund characteristics within the family. We note that these unobservable characteristics may be transmitted across funds within the family through directors who serve 23 multiple funds simultaneously as directors have, among other responsibilities, the right to review managerial employment contracts. 6. Unobserved Actions of Fund Managers A fund’s disclosed performance record may correspond imperfectly with the hypothetical performance that would have been generated if the fund had actually held the publicly disclosed holdings for the reporting period. This reflects the fact that in between disclosure dates the fund managers have the ability to trade repeatedly and alter the composition of their portfolios. We therefore examine how funds’ performance deviates from their disclosed holdings. TKacperczyk, Sialm, and Zheng (2008) calculate the forward return gap and attribute the presence of this return gap to managerial skill. Meanwhile, the backwards holding gap allows us to understand the prevalence of window dressing at both the start and end of the quarter, as introduced by Agarwal, Gay, and Ling (2014). We follow Kacperczyk et al. (2008) and Agarwal et al. (2014) to construct the return gaps which are used separately as dependent variables in estimation of this model: ,
.
,
. [3] We first use the forward return gap as the dependent variable as in Kacperczyk et al. (2008) and then we use the backwards holdings return gap as the dependent variable as in Agarwal et al. (2014). In both cases the return gap is measured as the difference between actual performance of the fund and how the firm would have performed had the fund invested exclusively as indicated in the latest disclosure. The key difference is that the Kacperczyk et al. measure 24 presumes the end of quarter disclosure holdings were held for the subsequent quarter while the Agarwal et al. measure presumes the same holdings were held for the preceding quarter. 6.1 Managerial Skill Given that Kacperczyk et al. (2008) find that the return gap is persistent and affects fund performance, a positive impact of board overlap on the existence of the return gap would suggest that overlapping boards are able to recruit more skilled managers. However, managerial skill is time‐varying (Kacperczyk, van Nieuwerbergh, and Veldkamp, 2014), and thus we expect to find no relationship between the return gap and the degree to which boards overlap within a fund family. It is therefore not surprising that we find that there is no impact of board structure on the presence of the forward return gap. 6.2 Window Dressing The degree to which a fund manager feels it is appropriate to manipulate holdings through the practice of window dressing, as proxied by the magnitude of the backwards holding return gap, may reflect upon the culture of the mutual fund itself or the fund family. We find that the presence of overlapping boards is likely to lead to greater coordination of how each fund within a complex reports its holdings and returns and thus whether managers feel it is acceptable to engage in the practice of window dressing. Accordingly, managers of funds with overlapping boards are more likely to be encouraged to engage in window dressing in order to increase the perception of the fund family containing strong performers (Column I). We observe that the managers of the largest funds may have reduced ability to engage in 25 selective window dressing in the presence of board overlap as the asset‐weighted measure of director overlap is insignificantly associated with the degree of window dressing (Column II). 7. Board Overlap and Strategic Performance Transfer in Mutual Fund Families There is considerable evidence that the mutual fund families follow strategies to maximize the returns to the family as a whole. For example, within mutual fund families, there are often funds that out‐perform the rest of the family across time, and these can be termed “star funds”. Performance persistence is unusual within the fund industry as investment success is generally not likely to persist across multiple time periods (Brown and Goetzmann, 1995), therefore the existence of star funds is one sign that families may be purposefully allocating resources across funds in an unequal way (Guedj and Papastaikoidi, 2004). A star fund increases flows to all funds in the family (Nanda et.al., 2004), and is thus of significant marketing and public relations value to the mutual fund family. Strategic cross‐fund subsidization may occur whereby the fund family coordinates actions to systematically boost the performance of the funds with high family value at the expense of the funds with low family value. Gaspar et al. (2006) show that cross fund subsidization is a proxy for performance transfer as the mutual fund family reallocates “winners” to the better performing funds to ensure performance persistence. The fund family would benefit from future investor inflows if extra performance is directed towards the high‐
fee funds. This could happen through the fund family coordinating purchases and sales of investments made by particular funds within the family. It could also happen if the family allocates IPO stocks differentially to the individual funds under its umbrella. 26 We hypothesize that if a low‐value fund has a larger director overlap when compared to the high‐fee funds, there will be more opportunities and incentives for strategic performance transfer from the low‐fee fund to the high‐fee fund. That is because the low‐fee fund’s board would be less independent than the rest of the family, and may prioritize the fund family’s collective interests. Tufano and Sevick (1997) have shown that boards frequently discuss issues pertaining to multiple funds at a single board meeting. To test this hypothesis, we follow Gaspar et al. (2006) and test whether the observed differences in returns between high‐fee funds and low‐fee funds in a family systematically exceed the difference in returns of their investment styles. Specifically, we estimate the following regression: ,
,
,
,
. [4] “Same Family” is an indicator variable that takes the value 1 if the low value and high‐value fund pairs (on the left‐hand side of the above equation) belong to the same mutual fund family. These are called “actual pairs.” The indicator variable takes the value zero if the funds belong to different families. This is accomplished by replacing each low‐value fund in an actual pair with a low‐value fund from another family (matched pairs) that has the same investment objective. Gaspar et al. (2006) show that is positive and significant, and infer from this result that fund families strategically transfer performance from high‐value funds to low‐value funds. Our estimation results yield the same result – for our study sample, we too find a significant and positive coefficient on the “
” indicator (Table 7, Column 1). 27 Once we confirm the presence of such performance transfer from low‐fee funds to high‐fee funds within our study sample, we move on to investigate the effects of board overlap on performance transfer by estimating the following equation: ,
,
,
,
∗
,
∗
[5] This regression equation augments Equation 4 by incorporating the board structure. “
” is an indicator variable that takes the value one if the high‐fee fund in the actual pair has a greater director overlap ratio than the low‐fee fund, and zero otherwise. Similarly, is a variable that takes the value one if the low‐fee fund in the actual pair has a greater overlap ratio than the high‐fee fund, and zero otherwise. A significant and positive estimated coefficient would indicate that there is performance transfer from low‐fee funds to high‐fee funds when there is greater director overlap at the high‐fee funds. These would be the instances when the low‐fee funds do not have many directors that are independent of the high‐fee funds. On the other hand, a significant and positive would be evidence of performance transfer from high‐fee funds to low‐fee funds when the director overlap measure is higher at the low‐fee funds. In such cases, the low‐fee funds have additional directors that do not serve on the board of the high‐fee funds, and may 28 be more inclined to protect the interests of the shareholders of low‐fee funds rather than maximizing the family values by transferring performance to the high‐fee funds. The evidence supports our predictions. We find that the magnitude of the performance increase at high fee funds is both highly statistically significant and large in magnitude, 59.2 bp, while the low fee fund has a statistically insignificant change in performance (Table 7, column II). Altogether, these results suggest that high fee funds are consistently enjoying a 50‐60 basis point increase in net returns due to board overlap. Given that the average net return in our dataset is 520 basis points, this is roughly a 10% increase. These results suggest that the board overlap leads to within family performance transfers, which benefit investors in high‐fee funds through higher returns, while steering performance away from low‐fee funds. 8. Family asset inflows Given that the presence of a board overlap is associated with strategic performance transfer to higher fee funds, and that fund families selectively advertise star funds, we expect that board overlap should lead to higher asset inflows for individual funds within the family. We therefore examine a model that explores whether this is the case for individual fund‐classes: ,
.
,
. [6] We find that there is a positive effect of board overlap upon asset inflows. This result implies that the family is able to use the board to coordinate the actions of individual funds through strategic performance transfer and window dressing to advertise the family as a whole. Accordingly, an overlapping board structure enjoys and leverages informational advantages as it can act as a conduit for sharing information across funds and managers within the family. 29 This result suggests that the board is more focused on aggregate family benefits than on the interests of investors in each class of each mutual fund, as is the official mandate of the board from the SEC. 9. Robustness Checks 9.1.
Number of funds: In our first round of robustness tests, we replace family size with the number of funds in a family and rerun all regressions. When we do this analysis it is necessary to exclude family size, because family size and number of funds in a family are highly correlated. This round of results allows us to control for fund family scope as proxied by number of funds. In a sense we are now testing the extensive margin of a mutual fund family, family scope, vs. the intensive margin, family size. These results are qualitatively similar to those described in this paper, and are available in our appendix. 9.2.
Using funds instead of fund classes It is common in the mutual fund literature for all empirical analyses to be conducted at the level of the fund, not the fund class. However, we conducted our analyses at the fund class level consistent with the SEC requirement that board of directors have a fiduciary responsibility to investors at the class level. While that is the case, it is also true that the managers comingle funds across classes and thus make decisions at the fund level. Accordingly, we replicated all results at the mutual fund level and obtained results that are qualitatively similar to those discussed earlier, and are available in our appendix. 30 9.3.
Nonlinearity We also estimate an alternative specifications of the regression equation that allow for non‐
linearity in the relation between dependent variables and board overlap. Our results do not indicate the presence of such non‐linearities. 10. Conclusion Conventional wisdom suggests that a greater extent of board overlap within a mutual fund family will benefit investors through the presence of economies of scale and bargaining power advantages. However, as documented by Tufano and Sevick (1997) and Kuhnen (2009), boards of directors are initially appointed by the fund sponsor, and are generally asked to stay on for long periods of time. Accordingly, the directors may prioritize the interests of the fund family even as the SEC mandates that their fiduciary responsibilities lie at the fund‐class level. Our first contribution is that we examine how variations in the extent to which boards overlap within a fund family affect fund characteristics by examining board overlap from angles that directly reflect both the busyness of the board to capture the scope of the demands on their time and the number and size of the funds to capture the intensity of these demands. Prior research has focused primarily on the case of complete board overlap, a unitary board, as in Kong and Tang (2008). Our second contribution is to acknowledge that boards of directors often have connections that are not observed, and that each mutual fund family has a unique institutional culture that may foster different types of communication. Accordingly, we include family fixed effects in our analyses to reduce the impact of unobserved heterogeneity. This is similar to the approach 31 adopted by Tufano and Sevick (1997). We illustrate that the conventional expectations for the impact of an overlapping board are not observed if we control for this heterogeneity. Finally, by examining multiple dimensions of board overlap, we identify clear relationships between board overlap and fund performance. Specifically, we find that funds with a higher degree of board overlap have higher marketing fees but not higher total fees. This is consistent with the fact that mutual fund families advertise star funds more than others, and board overlap is associated with higher net and gross returns. However, performance transfer to high fee funds is more likely to occur when there is board overlap. Overlapping boards may make it easier to steer investment opportunities to star funds that are attracting greater investor inflows or that generate more income for the fund family, because of their compensation practices. We conclude that unitary boards are a mixed blessing. 32 References Agarwal, Vikas, Gerald D. Gay, and Leng Ling, 2014, “Window dressing in mutual funds,” Review of Financial Studies, 27(11), 3133‐3170. Brown, Stephen J., and William N. Goetzmann, 1995, “Performance persistence,” Journal of Finance, 50(2), 679‐698. Chen, Joseph, Harrison Hong, Ming Huang, and Jeffrey D. Kubik, 2004, “Does fund size erode performance? Liquidity, organizational diseconomies and active money management,” American Economic Review, 94(5), 1276‐1302. Cremers, Martijn, Joost Driessen, Pascal Maenhout, and David Weinbaum, 2009, “Does skin in the game matter? Director incentives and governance in the mutual fund industry,” Journal of Financial and Quantitative Analysis, 44(6), 1345‐1373. Del Guercio, Diane, Larry Y. Dann, M. Megan Partch, 2003, “Governance and boards of directors in closed‐end investment companies,” Journal of Financial Economics, 69, 111‐152. Ding, Bill, and Russ Wermers, 2012, “Mutual fund performance and governance structure: The role of portfolio managers and boards of directors,” working paper. Ferris, Stephen P., and Xuemin (Sterling) Yan, 2007, “Do independent directors and chairmen matter? The role of boards of directors in mutual fund governance,” Journal of Corporate Finance 13 (2‐3), 392‐420. 33 Fich, Eliezer M., and Anil Shivdasani, 2006, “Are busy boards effective monitors?,” Journal of Finance, 61(2), 689‐724. Gaspar, Jose‐Miguel, Massimo Massa, and Pedro Matas, 2006, “Favoritism in mutual fund families? Evidence on strategic cross‐fund subsidization,” Journal of Finance, 61(1), 73‐104. Gormley, Todd A., and David A. Matsa, 2014, “Common errors: How to (and not to) control for unobserved heterogeneity,” Review of Financial Studies, 27(2), 617‐661. Guedj, Ilan, and Jannette Papastakaikoudi, 2004, “Can mutual funds families affect the performance of their funds?,” MIT working paper. Investment Company Institute (ICI), 2009, “Overview of fund governance practices 1994‐2008,” http://www.ici.org/pdf/pub_09_fund_governance.pdf, accessed on January 14, 2012. Investment Company Institute (ICI), 2012, website, http://www.ici.org, accessed on January 12, 2012. Jain, Prem C., and Joanna Shuang Wu, 2000, “Truth in mutual fund advertising: Evidence on future performance and fund flows,” Journal of Finance, 55(2), 937‐958. Kacperczyk, Marcin, Clemens Sialm, and Lu Zheng, 2008, “Unobserved actions of mutual funds,” Review of Financial Studies, 21(6), 2379‐2416. Kacperczyk, Marcin, Stijn van Nieuwerburgh, and Laura Veldkamp, 2014, “Time‐varying fund manager skill,” Journal of Finance, 69(4), 1455‐1484. 34 Khorana, Ajay, Henri Servaes, and Peter Tufano, 2008, “Mutual fund fees around the world,” Review of Financial Studies, 22(3), 1279‐1310. Koehler, Jonathan J., and Molly Mercer, 2009, “Selection neglect in mutual fund advertisements,” Management Science, 55(7), 1107‐1121. Kong, Sophie Xiaofei, and Dragon Yongjun Tang, 2008, “Unitary boards and mutual fund governance,” Journal of Financial Research, 31(3), 193‐224. Kuhnen, Camelia M., 2009, “Business networks, corporate governance, and contracting in the mutual fund industry,” Journal of Finance, 64(5), 2185‐2220. Nanda, Vikram, Jay Wang, and Lu Zheng, 2004, “Family values and the star phenomenon: Strategies of mutual fund families,” Review of Financial Studies, 17(3), 667‐698. Sirri, Erik R., and Peter Tufano, 1998, “Costly search and mutual fund flows,” Journal of Finance, 53(5), 1589‐1622. Tufano, Peter, and Matthew Sevick, 1997, “Board structure and fee‐setting in the U.S. mutual fund industry,” Journal of Financial Economics, 46, 321‐355. Yan, Xuemin (Sherman), 2008, “Liquidity, investment style, and the relation between fund size and fund performance,” Journal of Financial and Quantitative Analysis, 43(3), 741‐768. 35 Appendix. Variable Descriptions
Variable Expense Ratio
Definition
Source
Expense ratio of the fund in year t. Percentage of assets deducted CRSP
for fund expenses, including 12b‐1 fees, management fees, administrative fees, operating costs, and all other asset‐based costs incurred by the fund. Portfolio transaction fees, or brokerage costs, as well as initial or deferred sales charges are not included in the expense ratio. Converted to basis points.
Management Fee
Fees paid out of fund assets to the fund's investment adviser or its CRSP
affiliates for managing the fund's portfolio.Converted to basis points.
12 b1 Fees
Reported as the ratio of the total assets attributed to marketing and CRSP
distribution costs. Represents the actual fee paid in the most recently completed fiscal year as reported in the Annual Report Statement of Operations. Converted to basis points.
Non‐distribution Expenses
CRSP
Annual fund expenses (expense ratio) minus 12‐b1 fees. The nondistribution expenses measure the costs of operating the fund (e.g., investment management, transfer agency, audit, etc.), excluding the costs of distributing the fund.
Annual fund expenses plus the front and rear‐end loads. CRSP
Fund’s net return in year t, computed by compounding monthly net CRSP
returns (retm). Expressed in basis points.
Total Fees
Net Return Gross Return Alpha
Total Net Assets Net Asset Flows
Fund Age Fund Size Return Gap
Window Dressing Board Size
Board Independence
Director Overlap (Fund)
Director Overlap (Family)
Asset‐weighted Director Overlap (Fund)
Asset‐weighted Director Overlap (Family)
Fund’s gross return in year t, computed by adding total fees and Net CRSP
Return. Expressed in basis points.
Fund's alpha estimated using returns over 2007‐2009 using Fama‐
CRSP
French 4 factor model
Natural logarithm of fund total net assets (tna) in USD millions.
CRSP
TNA_flow = mtna ‐ (mtna[_n‐1]*(1+mret))
CRSP
Log of the age of the mutual fund, measured as the current year CRSP
minus the year at which it was first offered
Log of the total net assets for each fund in each year available, CRSP
reported in millions of dollars
Natural logarithm of number of directors on a fund's board of Hand‐collected
directors.
Percentage of independent directors serving on a fund's board.
The average percent of funds overseen by directors in this mutual Hand‐collected
fund, measured as the number of the funds overseen by a director in a family divided by total funds in the family, and then averaged across all directors in each mutual fund.
Board overlap averaged over all funds in a family.
Hand‐collected
The average percent of assets overseen by directors in this mutual fund, measured as the total net assets overseen by a director in a fund family divided by total fund family assets, averaged over all directors in the fund and all years available
Asset‐weighted board overlap averaged over all funds in a family.
Hand‐collected
Hand‐collected
Table 1. Summary Statistics
This table presents the summary statistics for board characteristics and fund characteristics for U.S. equity mutual funds for 2007‐2010. All variables are defined in the Data Appendix.
Panel A. Board Characteristics Unitary Board
Director Overlap Ratio (Fund)
Director Overlap Ratio (Family)
Asset‐weighted Director Overlap (Fund)
Asset‐weighted Director Overlap (Family)
Board Size Board Independence Mean
Std. Dev.
Median
0.59
0.90
0.90
0.90
0.90
8.61
0.82
0.49
0.20
0.18
0.21
0.18
2.61
0.10
1.00
1.00
1.00
1.00
1.00
8.00
0.83
Mean
Std. Dev.
Median
p25
p75
min
max
‐
0.89
0.92
0.93
0.90
7.00
0.75
1.00
1.00
1.00
1.00
1.00
10.00
0.89
‐
0.00
0.33
0.00
0.24
2.00
0.44
1.00
1.00
1.00
1.00
1.00
16.00
1.00
p25
p75
min
max
5
5,809
47.00
5.00
95.00
35.00
99.00
25.00
(2,166)
(2,300)
(0.11)
(0.0010)
(0.0025)
171
48,386
237.00
12.00
181.00
86.30
195.00
100.00
2,231
(2,417)
4.63
0.0015
0.0062
Panel B. Fund Characteristics Fund Size ($mn)
Family Size ($mn)
Number of funds in family
Fund Age
Expense Ratio (BP)
Management Fees (BP)
Total Fees (BP)
12‐b1 Fees (BP)
Net Return (BP)
Gross Return (BP)
Alpha (BP)
Return Gap (BP)
Window Dressing (BP)
Net Asset Flows
461
96,350
171.39
9.37
135.18
49.21
145.69
58.91
301
443
2.20
0.0003
0.0022
2,534 31
219168.7 21 248
158.47 126.00
7.79 8.00
61.23 130.00
88.40 67.30
64.84 144.00
34.92 50.00
2,791 999
2,828 1,168
4.51 20.42
0.0043 0.0001
0.0104 0.0011
(14) 3,297
0.08
0.40
2.00
1.00
7.00
(685.00)
8.00
2.24
(5,259)
5,242
(27.30)
(0.0544)
(0.0720)
(0) (71) 68 (91,017)
89,887
1,026,294
628.00
87.00
280.00
145.40
293.14
100.00
6,324
6,617
55.85
0.0556
0.1120
91,278
Table 2. Expense Ratios
This table presents the results from weighted least squares (WLS) estimation of Equation 1 in the paper. The dependent variable, Expense Ratio, is the ratio of the fund’s expenses divided by the value of the fund’s assets in year t (item exp_ratio in CRSP Mutual Funds). Year, ICDI investment objective (style), and fund family fixed effects (columns VI and VII) are included. All variables are defined in Appendix Table. Standard errors are adjusted for heteroskedasticity and correlation across observations belonging to the same fund family, and are presented in parantheses. * , ** and *** indicate significance at the 10%, 5% and 1% level, respectively.
(I) Unitary Board (II) (III)
(IV)
(V)
‐10.67
(8.305)
Director Overlap (Fund)
‐26.79*
(14.00)
Director Overlap (Family) ‐34.95**
(16.08)
Asset‐weighted Director Overlap (Fund)
‐23.13*
(13.24)
Asset‐weighted Director Overlap (Family)
‐33.97**
(14.93)
Fund Age
18.04***
(4.239)
17.91***
(4.256)
17.72***
(4.231)
17.86***
(4.316)
17.70***
(4.227)
Fund Size ‐7.210***
(1.237)
‐7.328***
(1.229)
‐7.311***
(1.225)
‐7.284***
(1.234)
‐7.311***
(1.229)
Family Size ‐6.465***
(2.224)
‐5.867***
(1.926)
‐5.805***
(1.920)
‐5.808***
(1.954)
‐5.756***
(1.951)
Board Size 19.50
(12.37)
22.08*
(12.46)
22.32*
(12.31)
21.46*
(12.56)
21.87*
(12.28)
CEO is the Chairman
‐1.533
(6.709)
3.108
(6.828)
3.124
(6.657)
2.773
(6.945)
2.946
(6.754)
Board Independence
‐16.04
(31.88)
‐13.35
(28.11)
‐16.02
(28.19)
‐11.24
(27.75)
‐14.36
(27.47)
Constant 184.5***
(37.68)
189.0***
(38.34)
197.5***
(39.14)
184.4***
(37.75)
195.8***
(38.11)
Year Fixed Effects
Style Fixed Effects Family Fixed Effects r2
r2_a
N
Yes
Yes
No
0.314
0.313
38134
Yes
Yes
No
0.315
0.314
38134
Yes
Yes
No
0.316
0.315
38134
Yes
Yes
No
0.314
0.313
38134
Yes
Yes
No
0.316
0.315
38134
Table 3. Total Fees
This table presents the results from weighted least squares (WLS) estimation of Equation 1 in the paper. The dependent variable is the total fund fees. Year, ICDI investment objective (style), and fund family fixed effects (columns VI and VII) are included. All variables are defined in Appendix Table. Standard errors are adjusted for heteroskedasticity and correlation across observations belonging to the same fund family, and are presented in parantheses. * , ** and *** indicate significance at the 10%, 5% and 1% level, respectively.
(I) Unitary Board (II) (III)
(IV)
(V)
‐13.56
(9.480)
Director Overlap (Fund)
‐30.13*
(15.46)
Director Overlap (Family) ‐38.52**
(17.93)
Asset‐weighted Director Overlap (Fund)
‐26.23*
(14.53)
Asset‐weighted Director Overlap (Family)
‐37.06**
(16.73)
Fund Age
17.98***
(5.410)
17.88***
(5.449)
17.69***
(5.413)
17.83***
(5.524)
17.68***
(5.412)
Fund Size ‐6.347***
(1.589)
‐6.497***
(1.578)
‐6.478***
(1.574)
‐6.446***
(1.580)
‐6.477***
(1.579)
Family Size ‐7.797***
(2.549)
‐6.936***
(2.161)
‐6.854***
(2.175)
‐6.877***
(2.187)
‐6.792***
(2.202)
Board Size 22.32
(14.22)
24.65*
(14.59)
24.77*
(14.48)
24.01
(14.71)
24.20*
(14.43)
CEO is the Chairman
1.061
(7.895)
6.612
(8.179)
6.576
(8.017)
6.255
(8.295)
6.355
(8.100)
Board Independence
‐7.577
(37.28)
‐2.004
(32.93)
‐4.609
(32.84)
0.258
(32.46)
‐2.604
(32.09)
Constant 195.4***
(40.43)
197.1***
(39.64)
205.8***
(40.71)
192.2***
(38.66)
203.7***
(39.38)
Year Fixed Effects
Style Fixed Effects Family Fixed Effects r2
r2_a
N
Yes
Yes
No
0.268
0.267
38134
Yes
Yes
No
0.268
0.267
38134
Yes
Yes
No
0.269
0.268
38134
Yes
Yes
No
0.267
0.266
38134
Yes
Yes
No
0.269
0.268
38134
Table 4. 12‐b1 Fees
This table presents the results from weighted least squares (WLS) estimation of Equation 1 in the paper. The dependent variable is the 12‐b1 fees. Year, ICDI investment objective (style), and fund family fixed effects (columns VI and VII) are included. All variables are defined in Appendix Table. Standard errors are adjusted for heteroskedasticity and correlation across observations belonging to the same fund family, and are presented in parantheses. * , ** and *** indicate significance at the 10%, 5% and 1% level, respectively.
Unitary Board (I) ‐12.78**
(5.270)
Director Overlap (Fund)
(II) (III)
(IV)
(V)
‐11.43
(7.943)
Director Overlap (Family) ‐18.65*
(10.10)
Asset‐weighted Director Overlap (Fund)
‐9.212
(7.120)
Asset‐weighted Director Overlap (Family)
‐17.94*
(9.417)
Fund Age
7.977*
(4.298)
7.439*
(4.373)
7.435*
(4.398)
7.431*
(4.372)
7.415*
(4.404)
Fund Size ‐4.294***
(1.082)
‐4.155***
(1.182)
‐4.168***
(1.172)
‐4.135***
(1.183)
‐4.158***
(1.176)
Family Size ‐2.693
(2.264)
‐0.178
(2.439)
‐0.180
(2.437)
‐0.109
(2.434)
‐0.0876
(2.438)
Board Size 6.261
(7.583)
3.138
(8.600)
4.177
(8.434)
2.707
(8.580)
3.797
(8.382)
CEO is the Chairman
0.955
(5.689)
5.519
(6.239)
5.761
(6.131)
5.372
(6.295)
5.658
(6.160)
Board Independence
3.993
(25.67)
23.69
(28.99)
20.47
(28.27)
25.34
(28.97)
21.77
(28.18)
Constant 67.48***
(19.13)
50.32**
(20.40)
56.66***
(20.14)
47.50**
(19.85)
55.36***
(19.77)
Year Fixed Effects
Style Fixed Effects Family Fixed Effects r2
r2_a
N
Yes
Yes
No
0.0973
0.0950
24040
Yes
Yes
No
0.0782
0.0759
24040
Yes
Yes
No
0.0814
0.0790
24040
Yes
Yes
No
0.0774
0.0750
24040
Yes
Yes
No
0.0811
0.0787
24040
Table 5. Net Returns
This table presents the results from weighted least squares (WLS) estimation of Equation 2 in the paper. The dependent variable is net return in year t, computed by compounding monthly net returns, expressed in basis points. Year, ICDI investment objective (style), and fund family fixed effects (columns VI and VII) are included. All variables are defined in Appendix Table. Standard errors are adjusted for heteroskedasticity and correlation across observations belonging to the same fund family, and are presented in parantheses. * , ** and *** indicate significance at the 10%, 5% and 1% level, respectively.
Unitary Board (I) 7.574
(34.20)
Director Overlap (Fund)
(II) (III)
(IV)
(V)
68.82
(60.90)
Director Overlap (Family) 18.85
(75.16)
Asset‐weighted Director Overlap (Fund)
75.92
(59.08)
Asset‐weighted Director Overlap (Family)
20.05
(76.25)
Fund Age
20.80*
(12.43)
21.45*
(12.58)
20.90*
(12.41)
21.90*
(12.65)
20.94*
(12.42)
Fund Size 1.630
(3.523)
1.851
(3.595)
1.708
(3.547)
1.722
(3.620)
1.710
(3.547)
Family Size 12.65
(9.212)
13.58*
(7.959)
12.07
(7.758)
13.88*
(7.936)
12.08
(7.705)
Expense Ratio
‐0.713***
(0.225)
‐0.697***
(0.226)
‐0.713***
(0.225)
‐0.695***
(0.226)
‐0.712***
(0.226)
Board Size ‐9.404
(70.87)
‐23.57
(75.21)
‐10.27
(73.97)
‐26.19
(75.16)
‐10.35
(73.83)
CEO is the Chairman
89.37**
(35.19)
81.65**
(38.24)
86.47**
(38.14)
81.06**
(38.09)
86.45**
(37.94)
Board Independence
92.52
(192.8)
118.1
(202.3)
89.53
(200.6)
121.6
(203.0)
89.44
(200.4)
Constant 2162.7***
(149.3)
2105.6***
(150.0)
2159.5***
(158.3)
2100.7***
(150.9)
2158.6***
(156.9)
Year Fixed Effects
Style Fixed Effects Family Fixed Effects r2_a
N
Yes
Yes
No
0.866
38121
Yes
Yes
No
0.866
38121
Yes
Yes
No
0.866
38121
Yes
Yes
No
0.866
38121
Yes
Yes
No
0.866
38121
Table 6. Gross Returns
This table presents the results from weighted least squares (WLS) estimation of Equation 2 in the paper. The dependent variable is the gross returns, computed by adding total fees and Net Return, expressed in basis points. Year, ICDI investment objective (style), and fund family fixed effects (columns VI and VII) are included. All variables are defined in Appendix Table. Standard errors are adjusted for heteroskedasticity and correlation across observations belonging to the same fund family, and are presented in parantheses. * , ** and *** indicate significance at the 10%, 5% and 1% level, respectively.
(I) Unitary Board (II) (III)
(IV)
(V)
5.067
(34.98)
Director Overlap (Fund)
66.46
(61.22)
Director Overlap (Family) 16.57
(75.45)
Asset‐weighted Director Overlap (Fund)
73.66
(59.46)
Asset‐weighted Director Overlap (Family)
18.22
(76.68)
Fund Age
20.10
(12.25)
20.77*
(12.41)
20.22*
(12.25)
21.21*
(12.47)
20.26*
(12.26)
Fund Size 2.750
(3.664)
2.951
(3.738)
2.810
(3.685)
2.826
(3.763)
2.812
(3.686)
Family Size 11.54
(9.320)
12.72
(7.983)
11.24
(7.775)
13.03
(7.957)
11.25
(7.720)
Expense Ratio
0.323
(0.225)
0.340
(0.227)
0.324
(0.226)
0.342
(0.227)
0.325
(0.226)
Board Size ‐7.281
(72.63)
‐21.81
(77.09)
‐8.642
(75.85)
‐24.43
(77.04)
‐8.826
(75.70)
CEO is the Chairman
92.02**
(36.12)
85.05**
(39.29)
89.81**
(39.22)
84.44**
(39.13)
89.76**
(39.02)
Board Independence
101.6
(196.4)
129.9
(206.1)
101.5
(204.2)
133.6
(206.8)
101.7
(204.0)
Constant 2167.1***
(150.4)
Yes
Yes
No
0.866
38121
2106.8***
(149.8)
Yes
Yes
No
0.866
38121
2160.7***
(157.8)
Yes
Yes
No
0.866
38121
2101.7***
(150.9)
Yes
Yes
No
0.866
38121
2159.2***
(156.4)
Yes
Yes
No
0.866
38121
Year Fixed Effects
Style Fixed Effects Family Fixed Effects r2_a
N
Table 7. Alphas
This table presents the results from weighted least squares (WLS) estimation of Equation 2 in the paper. The dependent variable is the fund's alpha, estimated using the Fama‐French 4‐Factor model with monthly fund returns over 2007‐2009. Year, ICDI investment objective (style), and fund family fixed effects (columns VI and VII) are included. All variables are defined in the Appendix Table. Standard errors are adjusted for heteroskedasticity and correlation across observations belonging to the same fund family, and are presented in parantheses. * , ** and *** indicate significance at the 10%, 5% and 1% level, respectively.
Unitary Board (I) ‐0.0711
(0.229)
Director Overlap (Fund)
(II) (III)
(IV)
(V)
0.300
(0.395)
Director Overlap (Family) 0.180
(0.543)
Asset‐weighted Director Overlap (Fund)
0.319
(0.377)
Asset‐weighted Director Overlap (Family)
0.155
(0.541)
Fund Age
0.122
(0.0936)
0.130
(0.0926)
0.128
(0.0927)
0.132
(0.0930)
0.128
(0.0929)
Fund Size 0.104***
(0.0225)
0.103***
(0.0225)
0.103***
(0.0224)
0.103***
(0.0225)
0.103***
(0.0224)
Family Size 0.118**
(0.0517)
0.135***
(0.0432)
0.130***
(0.0419)
0.136***
(0.0430)
0.130***
(0.0417)
Board Size ‐0.108
(0.453)
‐0.206
(0.480)
‐0.168
(0.477)
‐0.215
(0.483)
‐0.162
(0.479)
CEO is the Chairman
0.564***
(0.176)
0.551***
(0.196)
0.566***
(0.195)
0.549***
(0.195)
0.568***
(0.194)
Board Independence
‐0.781
(1.233)
‐0.503
(1.235)
‐0.578
(1.242)
‐0.494
(1.230)
‐0.596
(1.235)
Constant
4.917***
(1.524)
4.453***
(1.444)
4.574***
(1.513)
4.444***
(1.423)
4.603***
(1.502)
Style Fixed Effects Family Fixed Effects r2_a
N
Yes
No
0.444
39405
Yes
No
0.444
39405
Yes
No
0.444
39405
Yes
No
0.444
39405
Yes
No
0.444
39405
Table 8.Fees and Family Fixed Effects
Expense Ratios
Director Overlap (Fund)
‐2.495
(13.63)
Asset‐weighted Director Overlap (Fund)
Total Fees
‐6.100
(15.50)
‐0.646
(13.95)
12b‐1 Fees
10.81**
(4.912)
‐4.172
(15.87)
10.45***
(3.568)
Fund Age
15.03***
(3.883)
15.03***
(3.901)
14.52***
(5.013)
14.50***
(5.042)
6.938
(4.692)
6.966
(4.710)
Fund Size ‐6.826***
(1.138)
‐6.824***
(1.136)
‐5.705***
(1.547)
‐5.697***
(1.542)
‐4.207***
(1.139)
‐4.218***
(1.140)
Family Size 4.537*
(2.554)
4.524*
(2.549)
3.741
(2.766)
3.711
(2.763)
.
.
.
.
Board Size 43.58***
(10.68)
43.47***
(11.23)
45.51***
(13.61)
45.66***
(14.31)
6.527**
(3.036)
5.732*
(3.073)
CEO is the Chairman
‐36.02***
(12.83)
‐36.65***
(13.11)
‐31.13**
(15.05)
‐31.80**
(15.34)
‐9.751*
(5.665)
‐9.518
(5.916)
Board Independence
60.32**
(26.36)
60.61**
(25.45)
61.27*
(34.69)
60.82*
(33.56)
0.173
(15.05)
1.914
(14.77)
Constant ‐44.29
(50.10)
‐45.56
(48.21)
‐29.77
(59.40)
‐31.00
(57.28)
47.12***
(13.12)
48.01***
(12.67)
Year Fixed Effects
Style Fixed Effects Family Fixed Effects r2
r2_a
N
Yes
Yes
Yes
0.403
0.397
38134
Yes
Yes
Yes
0.403
0.397
38134
Yes
Yes
Yes
0.370
0.363
38134
Yes
Yes
Yes
0.370
0.363
38134
Yes
Yes
Yes
0.194
0.184
24040
Yes
Yes
Yes
0.194
0.184
24040
Table 9. Fund Performance with Family Fixed Effects
Director Overlap (Fund)
Net Returns
138.1**
(66.42)
Asset‐weighted Director Overlap (Fund)
Gross Returns
134.5**
(67.44)
124.9**
(55.55)
Alphas
0.109
(0.487)
121.4**
(56.32)
0.228
(0.444)
Fund Age
19.23
(12.48)
19.68
(12.53)
18.34
(12.35)
18.78
(12.37)
0.120
(0.0904)
0.121
(0.0906)
Fund Size 0.954
(3.227)
0.728
(3.227)
2.242
(3.348)
2.022
(3.346)
0.0968***
(0.0210)
0.0965***
(0.0209)
Family Size ‐307.9**
(128.9)
‐307.2**
(129.0)
‐308.8**
(128.8)
‐308.1**
(128.9)
‐0.181
(0.597)
‐0.181
(0.598)
Expense Ratio
‐0.904***
(0.212)
‐0.905***
(0.210)
0.120
(0.210)
0.120
(0.208)
Board Size 161.9***
(52.83)
153.3***
(51.73)
162.7***
(52.75)
154.4***
(51.66)
0.790*
(0.433)
0.761*
(0.435)
CEO is the Chairman
167.0
(102.8)
171.9*
(101.3)
172.8*
(102.4)
177.7*
(100.7)
0.941***
(0.359)
0.899***
(0.334)
Board Independence
‐185.2
(198.3)
‐161.4
(189.3)
‐185.7
(199.6)
‐162.6
(190.5)
‐1.919
(1.571)
‐1.844
(1.513)
Constant 5535.1***
(1480.9)
5541.0***
(1478.7)
5550.7***
(1478.2)
5556.7***
(1475.9)
7.073
(7.323)
6.988
(7.312)
Year Fixed Effects
Style Fixed Effects Family Fixed Effects r2_a
N
Yes
Yes
Yes
0.866
38121
Yes
Yes
Yes
0.866
38121
Yes
Yes
Yes
0.866
38121
Yes
Yes
Yes
0.866
38121
Yes
Yes
Yes
0.460
39405
Yes
Yes
Yes
0.460
39405
Table 10. Managerial Skill and Window Dressing
This table presents the results from weighted least squares (WLS) estimation of Equation 2 in the paper. In Columns I and II, the dependent variable is the fund's return gap, calculated using the procedure developed by KSZ . In Columns III and IV, the dependent variable is the backward holdings return gap developed by AGL. Year, ICDI investment objective (style), and fund family fixed effects are included. All variables are defined in the Appendix Table. Standard errors are adjusted for heteroskedasticity and correlation across observations belonging to the same fund family, and are presented in parantheses. * , ** and *** indicate significance at the 10%, 5% and 1% level, respectively.
(I)
Director Overlap (Fund)
(II)
16.73
(13.43)
Asset‐weighted Director Overlap (Fund)
(III)
(IV)
54.75**
(27.58)
19.17
(13.71)
0.600
(22.96)
Fund Age ‐1.057
(1.042)
‐0.942
(1.035)
15.61***
(2.844)
15.68***
(2.842)
Fund Size
0.181
(0.705)
0.104
(0.688)
‐8.444***
(1.765)
‐8.487***
(1.796)
Family Size
1.136
(2.983)
1.316
(2.950)
14.66***
(5.495)
14.98***
(5.517)
Board Size 3.046
(9.079)
2.132
(9.160)
‐15.64
(19.16)
‐7.052
(17.22)
CEO is the Chairman
‐13.33
(10.14)
‐16.25
(11.28)
‐38.46*
(21.91)
‐8.021
(21.78)
Board Independence
33.91
(26.02)
39.34
(27.48)
‐300.1***
(74.17)
‐299.8***
(69.57)
Constant
‐50.38
(47.59)
‐55.00
(47.54)
51.93
(100.5)
64.38
(90.02)
Style Fixed Effects Family Fixed Effects r2_a
N
Yes
Yes
0.0771
6622
Yes
Yes
0.0777
6622
Yes
Yes
0.183
7215
Yes
Yes
0.181
7215
Table 11. Strategic Performance Transfer
(I) Same Family (II)
(III) 49.66*
(27.90)
Director Overlap Measures
(Director Overlap of High Fee fund >= Director Overlap of Low Fee Fund) * Same Family 59.20**
(28.56)
(Director Overlap of Low Fee fund > Director Overlap of High Fee Fund) * Same Family ‐7.576
(22.04)
Asset‐weighted Director Overlap Measures
(Director Overlap of HIgh Fee fund >= Director Overlap of Low Fee Fund) * Same Family 57.78**
(28.52)
(Director Overlap of Low Fee fund > Director Overlap of High Fee Fund) * Same Family 0.520
(25.55)
Size of the high‐fee fund
10.54**
(4.619)
10.77**
(4.588)
10.70**
(4.589)
Size of the low‐fee fund ‐7.777***
(2.803)
‐7.678***
(2.737)
‐7.638***
(2.741)
Age of the high‐fee fund 44.73**
(21.86)
48.77**
(20.65)
48.07**
(20.80)
Age of the low‐fee fund ‐12.46
(13.98)
‐13.13
(14.59)
‐12.76
(14.58)
‐28.76
(17.86)
‐33.68*
(17.21)
‐32.89*
(17.34)
Constant 30.11
(64.42)
28.25
(59.82)
28.04
(60.70)
Year Fixed Effects
Style Fixed Effects Family Fixed Effects r2
r2_a
N
Yes
Yes
No
0.0172
0.0168
749471
Yes
Yes
No
0.0175
0.0171
749471
Yes
Yes
No
0.0174
0.0170
749471
Same style Table 12. Net Asset Flows
This table presents the results from weighted least squares (WLS) estimation of Equation 2 in the paper. The dependent variable is the fund's net asset flows, calculates as TNA_flow = mtna ‐ (mtna[_n‐1]*(1+mret)) . Year, ICDI investment objective (style), and fund family fixed effects (columns VI and VII) are included. All variables are defined in the Appendix Table. Standard errors are adjusted for heteroskedasticity and correlation across observations belonging to the same fund family, and are presented in parantheses. * , ** and *** indicate significance at the 10%, 5% and 1% level, respectively.
Unitary Board (I) ‐111.9
(125.3)
Director Overlap (Fund)
(II) (III)
(IV)
(V)
349.1
(257.8)
Director Overlap (Family) (VI)
(VII)
1063.7*
(603.6)
159.6
(215.3)
Asset‐weighted Director Overlap (Fund)
426.1
(273.1)
Asset‐weighted Director Overlap (Family)
1081.8*
(591.9)
185.8
(199.2)
Fund Age 633.6**
(272.3)
637.8**
(273.7)
636.4**
(273.4)
640.4**
(274.4)
636.8**
(273.3)
651.9**
(272.9)
655.7**
(274.6)
Fund Size ‐1
266.9***
(87.52)
267.0***
(87.50)
266.6***
(87.44)
266.3***
(87.06)
266.6***
(87.40)
300.3***
(95.80)
298.3***
(94.84)
Family Size‐1 ‐158.7**
(76.96)
‐134.9*
(70.25)
‐141.1*
(72.36)
‐132.4*
(71.98)
‐140.8*
(73.15)
‐27.74
(93.32)
‐26.35
(93.41)
Expense Ratio ‐1
‐2.645
(1.691)
‐2.473
(1.702)
‐2.526
(1.715)
‐2.456
(1.713)
‐2.521
(1.723)
‐3.133*
(1.604)
‐3.134*
(1.598)
Fund Return ‐1
0.00298
(0.0133)
0.00300
(0.0134)
0.00324
(0.0133)
0.00240
(0.0136)
0.00319
(0.0133)
0.00608
(0.0131)
0.00513
(0.0133)
Board Size 329.8*
(191.4)
210.9
(188.8)
262.3
(185.9)
189.1
(199.8)
259.1
(188.3)
‐575.0
(609.3)
‐661.4
(638.8)
CEO is the Chairman
‐104.4
(134.6)
‐116.6
(119.1)
‐96.20
(130.4)
‐122.5
(121.6)
‐97.30
(132.2)
‐424.1
(410.3)
‐440.7
(428.0)
Board Independence
‐1516.6*
(820.6)
‐1149.8
(706.1)
‐1261.7*
(749.6)
‐1107.8
(711.7)
‐1254.7*
(750.5)
‐763.4
(1144.2)
‐521.5
(1121.2)
Constant
‐226.0
(761.4)
‐866.1
(759.7)
‐658.5
(778.8)
‐939.2
(792.3)
‐683.4
(787.3)
‐1415.5
(1485.2)
‐1417.9
(1496.2)
Style Fixed Effects Family Fixed Effects r2_a
N
Yes
No
0.0519
36146
Yes
No
0.0519
36146
Yes
No
0.0518
36146
Yes
No
0.0520
36146
Yes
No
0.0518
36146
Yes
Yes
0.0487
36146
Yes
Yes
0.0489
36146
Download