Mutual Fund Voting And Pension Ties Finance Essay - Free Essay Example

Because mutual funds are the largest equity holders and because the retirement assets that are managed by mutual funds have been growing, mutual fund managers may have more incentives to support management in order to attract and retain pension business. I explore whether pension business ties have an impact on voting behaviors of mutual funds by examining the link pension business ties between mutual funds and the firms to actual mutual fund voting outcomes. At the fund family level, I find a positive relation between pension ties and mutual funds voting support for management. This relation becomes stronger when there is a voting divergence among funds within the same families. At the individual fund level, I find that individual funds are more likely to vote with management if they are included as one of the investment options of the pension plan of their portfolio firms. This suggests that the SEC should at least consider the recent petition from the AFL-CIO proposing that the SEC require mutual funds to disclose business ties with the firms in which they invest. Keywords: Mutual funds, Pension ties, Proxy voting * I thank Larry Dann and Diane Del Guercio for helpful comments and suggestions. All errors are mine. INTRODUCTION As institutional ownership has increased over the last few decades, there has been much academic interest in whether institutional investors take an active role in corporate governance. It is important to note, however, that due to potential conflicts of interest, all institutions do not necessarily have the same incentives to monitor management. In particular, mutual funds, the largest equity holders in the United States, were previously considered to be free of conflicts of interest because they did not do business with portfolio firms.  [1] According to the Investment Company Institute, as of the end of 2011, 61.2 percent of 401(K) plan assets and 45.1 percent of Individual Retirement Account (IRA) market assets were invested in mutual funds.  [2]  Since one half of the IRA market assets and 401(K) plan assets have been held by mutual funds since the late 1990s, there is a potential for conflicts of interest in mutual funds, and thereby fund managers have less incentive to exert an effort to monitor management and more incentive to support management in order to attract and retain the assets from the retirement market. Several studies document that mutual fund managers pursue their own interests at the expense of fund investors. For example, Cohen and Schmidt (2008) find that fund families acting as trustees systematically overweight their sponsor firms and even increase their holdings of the sponsor firms stock when other mutual funds are engaged in aggregate selling of sponsor firms shares. Other than through trading, mutual funds can support the management of the invested firms through voting for management proposals. Since proxy voting is the primary forum through which shareholders participate in the governance of corporations, examining mutual funds voting outcomes is one way to investigate the existence of mutual funds conflicts of interest. Davis and Kim (2007), Ashraf, Jayaraman, and Ryan (2010), and Pengfei Ye (2008) provide evidence tha t mutual funds are more likely to vote for management regardless of the best interest of the investors due to the existence of conflicts of interest. In this paper, I explore whether the pension business ties between mutual funds and their portfolio firms have an impact on mutual fund voting behavior. Through mapping mutual funds pension business ties to their portfolio firms onto mutual funds actual voting outcomes, I examine 61,336 voting decisions made by 3,130 mutual funds from 101 fund families for 172 management sponsored proposals and 129 shareholder sponsored proposals voted on at 257 firms over the sample period from June 2003 to December 2005. Using a probit model, I document a positive association between mutual fund voting and pension business ties. Analysis at the fund family level shows that fund families are more likely to vote for the ISS unfavorable management proposals and to vote against the ISS favorable shareholder proposals. The influence of pension ties to mutual fund voting becomes four times stronger when there is voting divergence among funds within the same families. However, there is no evidence that voting support of mutual funds is higher for the proposals voted at their clients firms compared to non-client firms. At the individual fund level, a fund included as one of the investment options of the pension plans of a firm in which it invests is more likely to vote against the ISS favorable shareholder proposals. The results of this study extend the findings of Ashraf et al. (2010) but are different in four aspects. First, to the extent that the marginal effect of the influence of business ties on mutual fund voting support for management is greater for close votes, the sample only includes votes in which the level of the voting support of the proposal is between 40% and 60%. Second, given the evidence of the impact of the Institutional Shareholder Services (ISS) advice on mutual fund voting support, I segregate the sample in to the proposals recommended by the ISS and the proposals not recommended by ISS. Third, as I observe non-trivial divergence in fund votes within the same families for both management proposals and shareholder proposals, I look into whether the association between business ties and fund voting is stronger with the existence of divergence. Furthermore, the finding of divergence among fund votes drives me to analyze the data at the fund level. Given the fact that the size of mutual funds has significantly grown and is expected to grow more in the future, the findings of this paper suggest that the policy makers should consider new rules in order to mitigate the concerns from conflicts of interest. For example, the Securities and Exchange Commission (SEC) should seriously review the recent petition from the American Federation of Labor and Congress of Industrial Organizations (AFL-CIO) proposing that the SEC require mutual funds to disclose pension business ties with the firms in wh ich they invest. The paper is organized as follows: the next section reviews background; section III describes the sample data; section IV shows the empirical results; section V concludes the paper. II. BACKGROUND Conflicts of interest of mutual funds induced by pension plan business While institutional investors have grown in size and number over the last few decades, whether they play an appropriate monitoring role in corporate governance is still being debated. One possible method to reveal whether institutions exert effort in the monitoring of corporate governance is by examining proxy voting, because proxy voting is one of the primary mechanisms through which shareholders participate in corporate governance. Through management sponsored proposals, corporate management seeks approval of new and existing policies from shareholders. Likewise, through shareholder sponsored proposals, shareholders advocate changes in corporate policies. Brickley, et al. (1988) examine how large shareholders vote on management sponsored proposals for antitakeover amendment provisions during the 1984 proxy season, and they find that institutional investors are more likely to cast their votes against these p roposals. While institutional investors might have greater incentives to participate in monitoring management than do small shareholders, the incentives of all institutional investors are not necessarily the same, and there may be conflicts of interest. With respect to proxy voting, all institutional investors will not necessarily vote against management, even if they believe it will maximize share value. Some institutional investors may choose to vote for management strategically in order to facilitate business ties with companies they own. According to Pozen (2011), among institutional investors, mutual funds are the largest equity holders in the United States. Mutual funds have a fiduciary responsibility to act in the best interests of the fund owners. By sponsoring or supporting value-enhancing shareholder proposals and preventing value-destroying management proposals, mutual funds can influence corporate governance to maximize firm value in order to benefit shareholders. How ever, there is some evidence that mutual funds do not perform their fiduciary duties due to conflicts of interest. Mutual funds are able to benefit from pension fund business such as managing the 401(K) plans of their portfolio companies, providing the financial incentive for mutual funds to support management regardless of the best interests of fund owners. Cohen and Schmidt (2008) explain why being the trustee of a large 401(K) plan is attractive for mutual fund families: (1) the trustee fund family guarantees a large initial inflow invested in family funds, (2) the family receives additional inflows in retirement assets from employee savings each year, and (3) trustees of 401(K) plans rarely change. The empirical evidence that conflicts of interest for mutual funds have effects on proxy voting is as follows. Davis and Kim (2007) analyze conflicts of interest in proxy voting by mutual funds using aggregate voting outcomes and find that fund families with pension business ties a re more likely to vote with management compared to fund families without pension business ties. Moreover, by using records of actual mutual fund voting patterns on shareholder proposals, the authors compare mutual fund voting outcomes at firms in which the funds manage pension plans with their votes to firms that do not manage. The authors argue that funds are no more likely to vote with management at client firms than at non-client firms. However, their data is based on the 2003 proxy season, which is the first year of mandated voting disclosures of mutual funds. Therefore, under close public scrutiny, mutual funds might have incentives to appear to not be subject to the influence of management. Likewise, Ashraf et al. (2010) investigates whether pension ties between mutual fund families and the firms affect how fund families vote on shareholder sponsored compensation proposals. The authors find that while fund families that have business ties to the firms they own are less likely to support shareholder proposals for executive compensation, there is also no difference in voting support of fund families with pension ties between client firms and non-client firms. Other than voting for management, mutual funds can support the management of the invested firm through trading. Cohen and Schmidt (2008) find that fund families acting as trustees systematically overweight their sponsor firms and even increase their holdings of a sponsor firms stock when other mutual funds are engaged in aggregate selling of that sponsor firms shares. Duan, Hotchkiss, and Jiao (2011) investigate whether pension business ties enable mutual funds to gain informational advantage in trading. The authors find that there is a positive relation between mutual fund trading and future performance of the portfolios firms with which funds have pension business ties. In response to increasing concern that conflicts of interest allow mutual funds to vote for management regardless of the best in terest of fund owners, on January 23, 2003, the SEC adopted new regulations: (1) the SEC required mutual funds to disclose actual voting results and (2) the SEC required mutual funds to publicly disclose a set of policies on how they will make decisions on proxy votes. The advocates of these rules expected that investors would be better able to monitor whether mutual funds engage in proxy voting in the interests of investors, preventing mutual funds from voting in support of management in order to facilitate other business relations with corporations whose shares they own. Cremer and Romano (2007) investigate the impact of the 2003 mutual fund voting disclosure regulation on voting outcomes by comparing voting outcomes of similar proposals sponsored at the same firms both before and after adoption of the regulation. The authors find no evidence that the rule decreases mutual funds voting in support of management. Indeed, they show that mutual funds appear to have increased their sup port for management on executive equity incentive plans (EEIC) since the rule change. Close votes Listokin (2008) finds that there is a large difference between the frequency of management sponsored proposals passing with votes just above 50% and the frequency of proposals failing with votes just below 50%. As shown in Figure 1, there is a discontinuity in the distribution of voting outcomes in management-sponsored votes at the 50 percent level. This discontinuity suggests that management has information about the outcome of voting at a time when management can do something to change the outcome of votes. One possible explanation for this pattern could be that managers, in order to achieve success, encourage mutual funds that manage their pension. Suppose that management predicts that the percentage of voting support for a management sponsored proposal is just below 50%. In that case, to pass the proposal, management attempts to pressure mutual funds that manage its pension pl ans to vote in favor of management. With a small shift, management can win the proposal as the marginal effect of mutual fund voting shift in close votes is large.  [3]  However, management will withdraw or alter the proposal if it predicts that the proposal would fail with far below 50% of voting support. Furthermore, management will not act if it predicts the proposal will win with far above 50% of voting support. To the extent that marginal effects of voting support of mutual funds for management proposals are greater for close votes than for non-close votes, I expect that the influence on proxy voting of pension ties is more likely to be detected in close votes rather than non-close votes. Therefore, the analysis of this study involves the voting outcomes for close votes. Although there is no such discontinuity in shareholder sponsored proposal, examining close votes for shareholder proposals is also meaningful given that marginal effects of mutual voting shift is large, wit h a small shift from voting against management to voting for management. The ISS recommendation As suggested by prior literature, the ISS has a non-trivial influence on mutual fund voting behavior. As a leading proxy advisory firm, the ISS gives recommendations on proxy voting by publishing general guidelines on each issue of each proposal or by providing case-by-case recommendations. Cotter, Palmiter, and Thomas (2011) find that voting support of mutual funds decreases by 68.3% for ISS unfavorable management proposals and increases by 53.1% for ISS unfavorable shareholder proposals. Choi, Fisch, and Kahan (2010) observe that the cases in which mutual funds always follow management of their portfolio firms are twice as many as cases in which mutual funds always follow the ISS. In considering the importance of the impact of the ISS recommendation on proxy voting, Ashraf et al. (2010) segregates its sample into shareholder proposals recommended by ISS and shareholder proposals o pposed by the ISS. They detect a negative association between voting support of fund families and pension business ties only for the first case. When the ISS opposes shareholder proposals, there is no association between fund family voting and pension ties. Considering that mutual funds likely follow the ISS advice for potential value reducing shareholder proposals regardless of whether they have business ties, their finding is somewhat predictable. If we assume that the ISS provides reliable advice and that mutual funds are the only shareholders, the ISS unfavorable shareholder proposals are comparable to the ISS favorable management proposals. This is because both mutual funds and management likely consider the former as value reduction proposals and consider the latter as value improving proposals. However, for the ISS favorable shareholder proposals, mutual funds consider them to be value increasing while management votes against them as it always opposes shareholder sponsored p roposals. When there is a disagreement between management and mutual funds, management acts to influence votes by using mutual funds ability to attract pension inflows as an incentive. Likewise, for the ISS unfavorable management proposals, shareholders consider them as value deceasing while management makes efforts for them to be passed by encouraging mutual funds with business ties to side with management. Therefore, I expect the association between business ties and voting support of mutual funds is more likely to be detected when the ISS endorses shareholder proposals and when the ISS opposes management proposals, compared to when the ISS opposes shareholder proposals and the ISS endorses management proposals. Divergence in fund votes within fund families Most prior studies on mutual fund voting analyze data at the fund family level since they consider that fund families vote their shares as a block.  [4]  However, Morgan, Poulsen, Wolf, and Yang (2011) argue that for shareholder sponsored proposals there is a large divergence in voting among funds within a fund family while there is a substantial uniformity in management sponsored proposals.  [5]  Such a divergence in shareholder proposals suggests that it may be important to perform individual fund level analysis in addition to fund family level analysis. If a fund votes for a proposal of its portfolio firm and this fund is included as one of the pension investment options of the firm, the fund likely has more incentive to support management compared to other funds in the same family which are not included as an investment option. If this is the case, dispersion in voting within fund families would be more likely to be observed when one of the funds is included as an investment option of the pension plan. Furthermore this dispersion may drive a stronger association between mutual fund votes and business ties. In other words, mutual funds more strongly support their portfolio firms with busi ness ties when dispersion within the same family exists compared to when there is uniformity. Therefore, I expect that mutual funds with business ties are more likely to vote for management when there is a divergence in voting among funds within fund families, compared to those funds when there is uniformity. In a similar vein, Burtler and Gurun (2011) find that educationally connected funds are 42% more likely to vote for management when there is a divergence in voting among funds within fund families (about 19% of the observation in their data), while those funds are 7% more likely to vote for management when there is uniformity. In addition, in order to support the achievement of managements goals, mutual funds with pension ties may be more likely to diverge from their family voting policy for close votes compared to non-close votes.  Ãƒ ¢Ã¢â€š ¬Ã‚ °Ãƒâ€šÃ‚ Ãƒ ¢Ã¢â€š ¬Ã‚ °Ãƒâ€šÃ‚ Ãƒ ¢Ã¢â€š ¬Ã‚ °. SAMPLE DESCRIPTIONS 3.1 Source of data Data on voting outcomes by mutual funds is obtained from the ISS Voting Analytics Database which includes Company Vote Results and Fund Vote Results (mutual fund N-PX filings). Company Vote Results contain aggregate voting outcomes for Russell 3000 companies shareholder meetings as well as the total number of shares outstanding, the total number of votes cast, and the ISS recommendation. Fund Vote Results show how individual funds of fund families vote on each proposal. Voting outcomes are recorded as being for, against, or abstaining per proposal per fund. I have access to voting outcomes of corporate governance related management sponsored proposals and shareholder sponsored proposals occurring from June 2003 through December 2005. Data on pension plans comes from the Form 5500 filings filed with the Department of Labor (DOL) and Internal Revenue Code and 11-K filings filed with the Security and Exchange Commission (SEC). The Form 5500 is required to be a nnually filed by plan administrators in order to report financial, investment, and operational information about pension and welfare benefit plans. The Form 5500 provides information on the fees paid to the service providers such as trustees and recordkeepers as well as the amount of assets held by investment providers. Although the Form 5500 covers most of the sample firms receiving proposals, the data is not as complete as the 11-K. The 11-K is an annual report dealing with employee stock purchases and savings plans which is required to be filed if a companys stock is offered as one of the investment options for plan participants. The 11-K filings include the identity of the trustee, total assets invested in the plan, and the proportion invested in each investment option. The analysis of this study first examines 354 firms that all have at least one corporate governance related proposal that has 40% to 60% of voting support. The percentage of votes in favor of a proposal is dif ferently computed according to the denominator of the calculation of both support level and the required threshold for each proposal.  [6]  For each firm in the initial sample, I gather pension data from both the Form 5500 and the 11-K by mapping the IRS Employer Identification Number (EIN) into CUSIP using the CRSP/Compustat Merged Database. Including firms with either the Form 5500 or the 11-K leaves me with 257 firms.  [7]  As my interest of in this study is whether mutual funds with business ties affect how mutual funds vote, by looking at the fund name and the family name. I check each sample firm to see whether the fund family of the individual fund that casts its vote is the trustee or the service provider for the firms pension plans and whether the individual fund is included as one of the pension investment options of the plans. Following Ashraf et al. (2010), I classify fund families in the sample sample as having a pension business tie if the fund family is a trus tee or service provider or if the individual fund is included among one of the investment options of the plans offered by a firm. The final step is to gather additional information about sample firms: the data on firm characteristics are from CRSP and Compustat; the data on governance index (G-index) is available freely from Metricks web-site; the data on institutional holdings is obtained from the Thomson-Reuters 13F database. The final sample of this study is constituted of 257 firms with close votes. As some firms have more than one proposal, 172 management sponsored proposals and 129 shareholder sponsored proposals are collected over the sample period from June 2003 to December 2005. The analysis at the fund family includes 13,120 voting decisions by 101 mutual fund families, and 41,407 fund voting outcomes on ISS favorable shareholder proposals are analyzed at fund level. 3.2 Summary sStatistics Table 1 provides information on the number of proposals and the number of mutual fund families that cast their votes for each type as well as information on how uniformly funds within the same family vote with management for each proposal type. Voting with management is equivalent to funds voting in favor of management proposals. This is also equivalent to voting against shareholder proposals given the fact that management always takes a position against shareholder sponsored proposals.  [8]  Given the importance of the effect of the ISS advice, I divide each proposal type according to whether the ISS endorses it or not and according to who sponsors the proposal (ISS favorable management proposals, ISS unfavorable management proposals, ISS favorable shareholder proposals, and ISS unfavorable shareholder proposals). Panel A of Table 1 shows that for the ISS favorable management sponsored proposals, fund families uniformly vote in favor of management 78% of the time and uniformly vote against management 11% of the time. The divergence in fund votes w ithin the family is, on average, 11%. Interestingly, individual funds show the greatest divergence (50%) surrounding the issue of Declassify the Board of Director, known as one of the most significant value reducing effects. Compensation related proposals such as Approve Omnibus Stock Plan, Amend Omnibus Stock Plan, and Amend Stock Option Plan have the next highest level of divergence as well as the next highest frequency. For the ISS unfavorable management proposal at in Panel B of Table 1, funds uniformly vote against management 64% of the time and vote for management 19% of the time. The average divergence of funds within fund families is registered at 17%. Unlike previous studies, I observed a non-trivial voting divergence among funds within a fund family. Compensation related proposals (Approve Omnibus Stock Plan, Amend Omnibus Stock Plan, and Amend Stock Option Plan) are again the most frequently occurring types of proposals. Panel C of Table 1 shows that for the ISS fav orable shareholder sponsored proposal types, funds are less likely to vote uniformly within families, compared to management sponsored proposals. On average, funds in a family vote identically on 73% of the proposals while votes of funds in a family differ in 27% of the proposals. Board related proposals (Require a Majority Vote for the Election of Directors and Declassify the Board of Directors) are the most common shareholder sponsored proposals. Compensation related proposals (Expense Stock Options and Performance- Based/Indexed Options) have the next highest divergence as well as the next highest occurrence. Table 2 presents the number of and the percentage of business ties of funds families with pension ties to the any of the sample firms and fund families without any pension ties. It also shows the number of proposals for which fund families cast their votes. The top five mutual funds families with business ties are Fidelity Investment, John Hancock Funds, Vanguard Group, S SGA Funds, and Oppenheimer Funds/MassMutual, and these families account for 35.8% of ties in the sample. The top 15 fund families account for 67.6% of pension ties, which is more than twice the level of pension ties of the next 50 fund families managing any pension assets of the sample firms. As shown in Panel B of Table 2, Jackson National, Thrivent Investment Management, USAA, and Munder Funds cast their vote on more than 200 proposals while these families do not engage in any pension business to any firm in my sample. Given the findings of in the previous literature showing that firm characteristics affect how institutional investors vote, I include firm characteristics such as past performance, market capitalization, market-to-book ratio, G-index, and institutional holdings as control variables. Past one year performance is definied as the one year buy and hold market adjusted abnormal returns from the meeting date of proxy voting, and market capitalization is defined as the total number of shares outstanding times the closing price at the end of the year of the proposal. Market-to-book ratio is obtained by dividing the sum of market capitalization of equity and the book value of debt by the book value of total assets. As the sample period of this study is June 2003 to December 2006, I use the G-index published in 2004. Institutional holdings are calculated according to ownership held by institutions as of the end of the meeting year. Table 3 provides information on firm characteristics in the sample. On average, past one year returns for the sample firm are 13.1%, market to book ratio is 2.9, and market capitalization is 26,536 million dollars. Average institutional ownership of the sample firm is 71.4%, and average G-index for the sample firm is 9.4. IV. EMPIRICA ANALYSIS 4.1 Analysis at fund family level The main analysis of this paper explores whether business ties have an impact on mutual fund voting behavior by comparing the association between pension business ties and mutual fund voting support. I begin with the analysis at the fund family level and then extend my analysis to the individual fund level. At the fund family level of analysis, the dependent variable is the percentage level of votes in support of management. Following Ashraf et al. (2010), the percentages of voting support in favor of management for each mutual fund family are calculated by dividing the number of individual funds within the same fund families that vote with management for a proposal by the number of funds in the family that are eligible to vote on the proposal.  [9]  Therefore, the higher voting support indicates that funds are more likely to vote with management by voting for management proposals and by voting against shareholder proposals. As the distribut ion and the level of shareholder support of management proposals differ from those of shareholder proposals, I separately estimate the impact of pension ties on management sponsored proposals and on shareholder sponsored proposals. In addition, given the previous findings on the influence of the ISS recommendation to mutual fund voting, I segregate my sample into the ISS favorable proposals and the ISS unfavorable proposals. Therefore, through segregating the proposals according to sponsors and according to whether the ISS recommends passage of a proposal, the results of this study are separately presented for management proposals recommended by ISS, for management proposals not recommend by ISS, and for shareholder proposals recommended by ISS. The sSmall sample size of shareholder proposals not recommend by ISS does not allow me to include them in the analysis. All analysis is estimated using robust standard errors (Whites heteroscedasticity consistent standard errors). The result s also hold if I use standard errors with clustered by fund family, year, and firm. Table 4 compares the percentages of voting support for management among mutual fund families with pension ties and mutual fund families without pension ties using a simple t-test. For the ISS favorable management sponsored proposals, there is no significant difference in voting support between mutual fund families with ties and mutual fund families without ties. However, for management proposals opposed by the ISS, I find that the percentage of voting support in mutual fund families with pension ties accounts for 32.7 % of the proposals while fund families without pension ties accounts for 22.6 % of the proposals. In order words, voting with management of mutual funds with pension ties is significantly greater than that of mutual funds without pension ties for the ISS unfavorable management sponsored proposals. As discussed, to the extent that shareholders are likely to consider ISS unfavorable ma nagement proposals as value reducing, management may ask affiliated institutional investors such as mutual funds managing their pension plans to support it in order to win a victory. Likewise, management may pressure its trustees of pension plans not to support shareholder proposals as shareholders are likely to consider the ISS favorable shareholder proposals as value improvement. Consistent with this hypothesis, when the ISS recommends passage of a shareholder proposal, fund families with pension ties vote with management 31.5 % of the time while fund families without pension ties vote with management 28.1% of the time, and the difference is statistically significant. Table 5 is the main analysis of this study. As the finding thatSince previous studies have found that firm size, prior performance, institutional holdings, and corporate governance affect the behavior of institutional investors, I include these factors as control variables. The dependent variable is one if the per centage of voting support for management is equal to or more than 50%; otherwise it is zero. The independent variable equals one if a fund family has business ties with the firm in which they invest and is thereforeso eligible to cast their votes; otherwise it is zero. As the dependent variable is the indicator variable, I use a probit model to examine whether pension business ties of mutual fund families affect mutual fund voting behavior. The coefficient estimates of Table 5 present marginal effects of independent variables. I begin by examining whether there is a relation between the likelihood that a fund family votes with management and business ties. As shown in column (1) of Table 5, the marginal effect of the pension ties is negatively significant for the ISS favorable management proposal, indicating pension ties decrease the probability that fund families support management by 11.3%. Although this result opposes the hypothesis that mutual funds are more likely to vote wi th management in order to attract assets from pension inflows, it is not surprising. Suppose that there are two different management proposals: one is recommended by ISS and the other is not recommended. To the extent that shareholders are more likely to vote for the ISS favorable proposal compared to for the ISS unfavorable proposal, the probability that the former proposal is passed is greater than the latter. Therefore, the managers need to exert more effort in order to pass the ISS unfavorable proposal by contacting its affiliated shareholders such as mutual funds managing their pension plans. As expected, for the ISS unfavorable proposals in column (2) of Table 5, there is a strong positive relationship between business ties and voting support of fund families. This indicates that a business tie increases by 11.6% the probability that a fund family votes with management. Regarding the ISS favorable shareholder proposals in column (3) of Table 5, the probability that the fund fa mily supports management by voting against shareholder sponsored proposals is increased by 4.3% when a business tie exists. As discussed in the previous section, the ISS favorable shareholder proposals are comparable to the ISS unfavorable management proposals given that the disparity in voting decisions between management and shareholders is large. Therefore, it is expected that management puts exerts more effort to achieve a victory through pressuring mutual funds that manage its pension plans. Although the table is not included due to inconclusive results, I provide conditional logit analysis as including fund family fixed effect to investigate whether there is a difference between voting support by fund families at their client firms versus non-client firms. Contrary to my expectation that narrowing down the sample to close votes where the marginal effects of voting shift is large would allow me to examine different voting behavior of mutual funds for their client firms and f or non-client firms, I find no difference in voting support of mutual funds across the two groups. This indicates the probability that mutual funds that manage retirement assets vote with management of their client firms is similar to the probability that those mutual funds vote with management of their non-client firms. This result is consistent with the previous findings on mutual fund voting behaviors. As argued above, there is a non-trivial divergence in fund votes within the same families. It is likely that an individual fund has stronger incentive to support management if the fund is one of the pension investment options of the firm in which it invests, compared to other funds in the same families that are not included as investment options. Therefore, I expect the influence of pension ties to be stronger when there is divergence within fund families. To test this, in Table 6 I analyze the subsample with the proposals voted by mutual funds which have divergence within fund families. When there is disparity in fund votes on the ISS favorable management proposals, I find no relation between pension ties and voting support of fund families. However, for the ISS unfavorable management proposal, a pension tie increases the probability that a fund family votes with management by 43.3%. Compared to Table 5, the marginal effect of pension ties is four times greater if there is a disparity in fund voting. Similarly, the marginal effect of pension ties becomes four times stronger when fund families do not uniformly vote for the ISS favorable shareholder proposals. 4.2 Analysis at fund level As shown in Table 7, some funds vote uniformly with other funds in the same family while other funds vote in a diverse fashion. Although most of the previous literature argues that funds within the same family tend to vote uniformly, the evidence in Table 1 suggests that mutual fund voting should be analyzed at the individual fund level rather than at the fund family l evel. To investigate how support of individual funds toward management in proxy voting is influenced by business ties, I use a probit model. Due to limited observation availability, I can only test for the ISS favorable shareholder sponsored proposal. To do this, I examine fund voting at the individual level. The dependent variable equals one if a fund votes with management, zero otherwise. The independent variables is 1 one if a fund is included as one of the pension investment options of the firm in which it invests, and zero otherwise. Table 6 illustrates that the coefficient estimate of the pension tie is positively significant, which indicates that a business tie increases by 15.3% the probability that a fund supports management. This result suggests that individual funds are more likely to vote with management if they are included as investment options in the pension plans of their portfolios firm, compared to other funds not included. Individual funds are more likely to vote against shareholder proposals to support management of the firms with large size and high market to book ratio, and with high institutional ownership. V. CONCLUSION AND FUTURE WORKS This paper investigates whether pension business ties have an influence on mutual fund voting. As pension assets managed by mutual funds have been growing since the late 1990s, the concern that mutual fund managers could potentially pursue their own interest at the expense of fund owners is raised. This is because fund managers might have more incentive to support management in order to attract and retain pension business. In fact, recent studies provide evidence that mutual funds tend to support management of their portfolio firms through proxy voting or trading. Using the mutual fund voting outcomes fromoccurred the period from June 2003 to December 2005, I find a positive association between mutual fund voting and business ties. Analysis at the fund family level show that fund families are more likely to vote for the ISS unfavorable management proposals and to vote against the ISS favorable shareholder proposals. The influence of pension ties to mutual fund voting becomes four t imes stronger when there is voting divergence among funds within the same families. At the individual fund level, the a fund included as one of the investment options of the pension plans of the firm in which it invests is more likely to vote against the ISS favorable shareholder proposals. Overall, my findings provide the evidence that mutual funds are conflicted due to their pension business ties with the firm in which they invest.Only close votes have been included in the sample proposals at in this analysis. In that Since the probability that management causes an upset by encouraging mutual funds to support management in close votes is higher than in non-close votes, I expect that the difference in voting support between mutual funds with pension ties and mutual funds without pension ties in close votes would be greater than the difference in the voting support across two groups in non-close votes. I leave comparison between close votes for a future analysis.