Influence of Public Opinion on Investor Voting and Proxy Advisors

Reena Aggarwal

McDonough School of Business, Georgetown University aggarwal@georgetown.edu

Isil Erel

Fisher College of Business, Ohio State University erel@fisher.osu.edu

Laura Starks

McCombs School of Business, University of Texas at Austin lstarks@mail.utexas.edu

April 2014

Abstract

We examine the evolution in voting patterns across firms over time. We find that investors have become more independent in their voting decisions, voting less with the recommendations of management or proxy advisors. Even when the proxy advisor recommends voting against a proposal, we find that over time investors are more likely to ignore the recommendation. Moreover, we also find that proxy advisory recommendations have become more supportive of shareholder proposals. Our main contribution is to examine the role of public opinion in the voting changes. We show that public opinion on corporate governance issues, as reflected in media coverage and surveys, is strongly associated with investor voting, particularly mutual fund voting. In addition, even proxy advisor’s recommendations are associated with public opinion. Our analysis suggests that public opinion influences corporate governance through the voting process.

JEL: G32; G34; G38; Keywords: Public Opinion, Media, Proxy Advisors, Institutional Investors, Proxy Voting

1

“We rely on several sources of information to inform our voting. We do our own analysis based on information published by companies, look at research done by proxy advisors, and follow media coverage. Media coverage helps us keep in touch with a broad range of constituents’ views on corporate governance issues."

Michelle Edkins, MD, Global Head, Corporate Governance and Responsible Investment, BlackRock

1. Introduction

Institutional investors have become the dominant shareholders in many firms. Their

ability to change the agenda through proposal submission and voting has evolved from having

little apparent importance to what many consider to be an important aspect of corporate

governance. In other words, institutional investors, whether activist investors or not, now have

the power of the vote to influence corporate policy. As considerations of best practices in

corporate governance have evolved, so have the opinions (and votes) of institutional investors.

Moreover, institutional investors do not operate in a vacuum; they have beneficiaries or their

own shareholders with opinions on how their investments should be managed and voted.

Because these beneficiaries’ and shareholders’ care about their institutions’ voting behavior, not

only mutual funds (which are required to do so by law) but also many other institutions now

publicly disclose their votes. Thus, there exists pressure or encouragement for institutions to vote

in particular ways, suggesting that institutional voting and its changes through time reflect public

opinion on corporate issues.

In this paper, we address the question of how public opinion influences the proxy voting

process. Our main contribution is to provide evidence suggesting a strong influence of public

opinion on the evolution in both investor voting behavior and proxy advisor recommendations.

Therefore, our results are consistent with the idea that a channel through which the public can

communicate with corporate management and potentially influence corporate behavior is

through the proxy voting process. Dyck, Volchkova and Zingales (2008) argue that media

attention can play a role in corporate governance by affecting the reputation of firms and their

executives. We provide evidence that the new channel, voting, also is a way in which media

coverage can influence firm behavior. This channel is important because media coverage

2

captures the attention of proxy advisors, institutional investors and individual investors, and is

thus reflected in recommendations and votes.

Large institutional investors face tens of thousands of proxy votes each year. For

example, BlackRock with more than $4.3 trillion under management, voted on 130,000

proposals, at more than 15,000 shareholder meetings, in 90 markets worldwide, and further

engaged with 1500 firms in 2013. Because of such large numbers of shareholdings and the

associated monitoring actions required, institutional investors need to gather information from

multiple sources in order to monitor firms and vote effectively. This information set includes

their own analysis, research conducted by sell-side analysts as well as proxy advisory firms, and

as indicated in the quote by Ms. Edkins of BlackRock, media coverage serves as an additional

external source of information about their portfolio firms. The media collects, aggregates, and

disseminates information as discussed by Dyck, Volchkova and Zingales (2008). Media

coverage would be expected to help calibrate the views of institutional investors about corporate

governance issues relative to general public opinion, or even to help them learn about the views

of other institutional investors. Ultimately, even large institutional investors are stewards of

capital for the wider public, and therefore need to pay attention to public opinion in addition to

their own research. Small institutional investors also need to rely on multiple sources of

information in their voting decision as they are more likely to lack the capacity to conduct their

own extensive research.

Previous studies have emphasized the influence of proxy advisory firms on voting by

institutional investors, finding a correlation between these firms’ recommendations and mutual

fund voting.1 Yet, the proxy advisory firms do not exist in a vacuum either. Both voting by

institutional investors and recommendations of proxy advisory firms can be influenced by the

economic and social climate of public opinion. In fact, this can be seen in the fact that one proxy

advisory firm, Institutional Shareholder Services (ISS), annually gathers feedback on their proxy

                                                            1 See, for example, Bethel and Gillan (2002), Ertimur, Ferri and Muslin (2011) and Iliev and Lowry (2013).

3

policies from institutional investors, corporate issuers and the wider corporate governance

community in advance of the proxy season. For example, the ISS U.S. Governance Policy 2012

updates dated November 17, 2011, state: “ISS reviews and updates its proxy voting guidelines

each year, taking into account emerging issues and trends, the evolution of market standards,

regulatory changes, and feedback provided by ISS' institutional clients.” Thus, as public opinion

changes so do the recommendations issued by ISS and voting by investors.

In order to better understand the role of public opinion in shaping the proxy voting

process, we examine trends in shareholder voting as well as changes in the underlying

influences. Our proxy for the public opinion that would be expected to be associated with

institutional voting is the magnitude and tone of media coverage about corporate governance. To

narrow this coverage down to a manageable number of news stories, we focus on the media

coverage of executive compensation. We also employ an alternative measure of public opinion

through Gallup surveys.

Although for many years there was little, if any, evidence showing direct consequences

of shareholder voting on proxy proposals (consistent with the Berle and Means (1932)

preliminary hypothesis on the role of shareholder democracy), in recent years several potential

influences have led to a higher importance of proxy voting. The potential influences include the

recently required disclosure of mutual fund voting records, the recent financial crisis with the

ensuing public attention to executive compensation, the apparent rise of influence of proxy

advisory firms, and better strategies on the part of sponsors of shareholder proposals. These

influences suggest higher votes in favor of shareholder proposals and against management

recommendations. In order to test the association between these influences and proxy voting, we

examine the changes in investor voting patterns in shareholder proposals from 2004 to 2010 with

a particular focus on the role of public opinion. Overall, our results indicate that shareholders

appear to take voting on proxy proposals seriously and that shareholder proposals are gaining

more positive votes. The increased votes in favor of these proposals suggest that shareholder

proposals are becoming a more important part of the governance process. Further, we find that in

4

recent years shareholders have become more independent in their voting. That is, they are less

likely to follow the recommendations of either management or proxy advisors. We argue that

part of this independence comes from receiving information through media coverage of firms.

Specifically, we find that while investor support for management proposals remains

relatively flat (from 81.6% in 2004 to 85.5% in 2010), investor support for shareholder proposals

has increased a large amount over our sample period. The mean support rate for shareholder

proposals increased from 23.6% in 2004 to 31.8% in 2010, after peaking at 37% in 2009. This

tendency is also observed in the rapid changes in the proportion of time investors voted with

management’s recommendation on shareholder proposals. In 2004, institutions voted in

agreement with management’s recommendation on shareholder proposals in 74% of all

proposals. However, by 2010, they voted in agreement with management in only 54% of the

proposals. These results at the firm level are consistent with those of Collins (2010) who

examines the 2007 to 2009 period at the mutual fund level and finds increasing support for

shareholder proposals by mutual funds. He concludes that one reason for this increasing support

rate is due to changes in proposal types. We find that even controlling for proposal types and

their changes, there has been an increase in support rate for shareholder proposals by all

investors as well as by mutual funds. We find the support rate was highest in 2009, following the

financial crisis. In contrast, the way in which investors vote on management proposals has not

significantly changed.

The changes in support for shareholder proposals are not driven by the recommendations

of the major proxy advisory firm, ISS. Strikingly, we find that while investors voted with ISS’s

recommendation 78.4% of the time in 2004, by 2010 the percentage in agreement with ISS was

down to 57.5%. This pattern is specifically driven by investors not following ISS’s

recommendation when the proxy advisor opposes shareholder proposals. In 2004, 60% of the

votes followed ISS’s opposition against proposals but by 2010 only 20% of the votes followed

the recommendation. Thus, shareholders appear to have become more independent of both

management and the proxy advisory firm. We find this decline in investor following of ISS’s

5

recommendation even though ISS’s support for shareholder proposals has increased over time. In

2004, the ratio of ISS’s “against” recommendations to “in-favor” recommendation was 159%,

and the ratio declined to 34% by 2010. One explanation for this result is that institutional

investors increasingly developed their own policies, requesting ISS to custom-design a voting

mechanism to implement the policy. At the same time, institutional investors have given more

attention to voting, and conduct their own analysis regarding the voting decision on a case by

case basis.

Our study is the first to show that public opinion matters in proxy voting by investors.

Specifically, we find that the increased support for shareholder proposals is associated with

public opinion. Since 2003, the voting for shareholder proposals, whether considered at the firm

level or the fund level, is related to public opinion. Further, this result holds for different

measures of public opinion, whether measured at the aggregate level through poll data from

Gallup surveys or media coverage, or measured at the firm-level through media coverage of

corporate governance news, specifically executive compensation. Moreover, our analysis

indicates that the recommendations of ISS are also related to public attention.

There have been a large number of studies on shareholder proxy voting and on mutual

fund voting.2 These studies have generally found that voting depends on firm and fund

characteristics. Many of these studies have limited the number of firms and the number of funds,

whereas we consider proposals at any of the Russell 3000 companies, and in our analysis of

mutual fund voting, we include 475 fund families and 13,313 individual funds. Thus, we

examine a broad array of firms and mutual fund families. Beyond our contribution using such a

large sample and examining changes over time, we also contribute by showing that investor

voting has become more independent of ISS recommendations, counter to what many

researchers and commentators believe. This change is surprising, given that over time the

support by ISS has increased for shareholder proposals.

                                                            2 For a review, see Gillan and Starks (2007). For a review of more recent papers, see Cotter, Palmiter and Thomas (2010).

6

Our most novel contribution is the examination of the relation between public opinion and

voting decisions at both the firm and fund level across time. Public opinion provides institutional

vote with power because there is a lot more than just the individual institutional investor behind

the vote, the vote reflects wider public sentiment. In that sense aggregate public opinion can play

the role of the activist investor envisioned by Levit and Malenko (2011) to aid in changes caused

by the nonbinding voting on shareholder proxies.

2. Shareholder Involvement, Voting and Public Opinion

An important question is whether shareholder involvement, such as through the voting

process, even matters to a firm’s valuation or managerial process. One argument maintains that if

shareholders have more voice through the corporate democratic process, agency costs will be

lowered.3 The counterargument is that shareholders should delegate their authority to the board

of directors due to the greater efficiency in the delegated model for the governance process.4 For

example, Bainbridge (2006) argues that in publicly held corporations there exist numerous

shareholders with diverse preferences, and these shareholders lack “both the knowledge and the

incentives necessary to exercise an informed vote.” In such a situation he argues that it makes

sense for there to be a complete separation of ownership and control. Similarly, Easterbrook and

Fischel (1983) state that this type of argument implies that shareholders’ interests are protected

not by voting but by the market for the firm’s stock and the managers’ need to raise capital, the

labor market, and the product markets. Further, “it would make little difference if shareholders,

like bondholders, could not vote at all. Funds spent providing shareholders with a more effective

voice are wasted at best and harmful beyond their costs if they hamper the firms’ effective

pursuit of profits.”

Despite these arguments, shareholder voting has become an increasingly important aspect

of corporate governance and shareholder ownership.5 For example, shareholders now approve a                                                             3 e.g., Bebchuk, 2005 4 See, for example, Hayden and Bodie (2010). 5 For an explanation of the operational issues of shareholder voting, see Kahan and Rock (2008).

7

number of managerial decisions such as the selection of the auditors, employee stock plans or

executive compensation plans. Similarly, many firms have revised their charters to require a

majority shareholder vote for directors.6 Maug and Rydqvist (2011) provide theoretical and

empirical evidence that shareholders vote strategically on management proposals. Christoffersen,

Geczy, Musto and Reed (2007) provide evidence of proxy vote trading and conclude that the

vote trading occurs more often for firms with more asymmetric information, worse performance,

and that it corresponds to support for shareholder proposals and opposition for management

proposals. Furthermore, federal agencies have stressed the need for institutional investors to

consider proxy voting as part of their fiduciary duties.7 In 1994 the Department of Labor, which

has oversight of corporate pension funds under ERISA, stated that a responsible pension fund

fiduciary should engage in active monitoring, which would include voting of proxies.8 Similarly,

the SEC issued a new rule in 2003 requiring mutual funds to disclose their proxy voting record

on an individual holdings basis.

Shareholder involvement encompasses the ways in which shareholder and investor input

are communicated to management. Such input can be communicated through direct engagement

with management. Institutional investors often consider the changes to a company’s governance

structure to be a long-term process. Consequently, they often work behind the scenes to effect

the changes.9 However, if management is not open to such communications or remains

intransigent (at least from the investors’ view), the shareholder has other tools. Another form of

involvement by shareholders is the submission of proxy proposals. Since the proxy proposal is a

way of communication with management and other shareholders, a majority vote is not always

the ultimate intention. In fact, shareholder proposals are often withdrawn because the shareholder

is able to reach an agreement with management.

                                                            6 For analyses of voting on directors or majority voting proposals, see Cai, Garner, and Walkling (2009; 2011). 7 See, for example, Lublin (2002). 8 See Department of Labor’s Interpretative Bulletin 94-2. 10 For example, management can influence the participation in the vote and the voting outcome through the efforts of hired proxy solicitors (Pound, 1988; Gillan and Starks, 2007).

8

One aspect of shareholder proposals that has been highlighted is that they are advisory to

the board’s deliberations. Thus, in concept, a firm’s board and management can ignore the voting

outcome even if the proposal receives a majority positive vote. Moreover, in some cases

shareholder proposals do not receive majority votes because management has the ability to affect

the voting outcome.10 However, given the increasing attention on many of the shareholder

proposals by the media and investors, it is not necessary for the proposal to win a majority vote

to have an influence.

Similarly, shareholders may expect that a significant vote in favor of a shareholder

proposal (or against a management proposal) is a signal that management should consider

making changes. For example, in a recent survey ISS found that managers expect to make

changes if a say-on-pay proposal receives more than 40% negative vote, but their institutional

investor clients set an even low bar, expecting to see the board or management make changes if

the proposal receives greater than 30% negative vote. Thus, it is not just a matter of whether a

particular proposal passes (receives majority vote), but rather it is a method through which

shareholders can communicate in aggregate with the board and management.

Aggregate changes in shareholder proposals and voting on those proposals can have

indirect effects as well. First, they can engender a regulatory response, for example, Congress

can change laws or the SEC can change rules, which can be costly for firms in terms of new

constraints or compliance costs. An example would be the Dodd-Frank requirements regarding

say-on-pay. On an individual firm level, changes in the submission of shareholder proposals and

the subsequent shareholder voting on those proposals can have negative effects on investors’

perceptions of and willingness to hold the firm’s shares, which can increase the firm’s cost of

capital. Similarly, lenders may also be less willing to provide capital if those investors view

management as being difficult. Thus, it is not just the vote itself, but the power of the vote and

what is implied behind it – institutional investor and regulatory muscle. Levit and Malenko

                                                            10 For example, management can influence the participation in the vote and the voting outcome through the efforts of hired proxy solicitors (Pound, 1988; Gillan and Starks, 2007).

9

(2011) argue that the nonbinding voting on shareholder proxies needs an activist investor to have

effects. We hypothesize that aggregate public opinion can itself play the role of activist investor

in forcing companies to make changes. Cvijanovic, Dasgupta, and Zachariadis (2014) in the

context of lawsuits conclude that investor attention plays a role in corporate governance via its

effect on institutional investors.

3. Changes in Proxy Proposals and Voting

3.1 Proposal Characteristics

We obtain data from several different sources. Voting records for each proxy proposal for

each firm in the Russell 3000 Index are obtained from ISS for the period January 2004 through

November 2010. The information in these records include the date of the meeting, a description

of the proposal, whether the proposal is sponsored by management or shareholders, and

recommendations regarding how investors should vote on the proposal from management and

from ISS. The records also provide the number of shares outstanding, the number of shares voted

for/against/abstain, requirements for the proposal to pass, and the final voting outcome.

Shareholder proposals are classified into four categories: 1) Board, 2) Corporate

Governance (e.g., supermajority voting, preemptive rights, and poison pills), 3) Executive and

Director Compensation, and 4) Other (e.g., proposals dealing with social issues, human rights,

health, and environment). For limited comparisons, we also examine five categories of

management-sponsored proposals: 1) Board, 2) Capitalization, 3) Mergers and Restructuring, 4)

Executive and Director Compensation, and 5) Anti-takeover. Any management-sponsored

proposal related to board elections or defined as “routine business” is excluded from our

analysis.11

                                                            11 Board proposals include issues such as board size, classified/staggered board, authority to appoint committees, term limits and proxy access; capitalization proposals deal with issues such as authorizing issuance of new stock, stock split and share repurchase; merger proposals address reorganization and mergers; compensation proposals relate to executive, director and employee compensation involving option plans, repricing of options, restricted stock, bonus, and loans; anti-takeover proposals deal with amendments to articles, charter, bylaws, written consent, supermajority voting and special meetings.

10

Table 1 provides descriptive statistics regarding the 18,438 proposals in our sample, of

which 14,448 (78.4%) are sponsored by management and 3,990 (21.7%) by shareholders.12 The

highest proportion of proposals that are sponsored by shareholders is in 2007 at 24.5% and the

lowest in 2005 at 19.9%. However, the maximum number of proposals is in 2009, following the

financial crisis. A major part of the increase dealt with issues related to board mechanisms such

as majority voting, the separation of the board chair from the CEO, and the right to call a

meeting. The participation rate, defined as the number of shares voted as a percentage of the

shares outstanding, varies within a tight range of 78% to 81% for management proposals and

70% to 75% for shareholder proposals. Institutional investors have a fiduciary responsibility to

exercise their vote in most circumstances, and given their dominance in share ownership, we

would expect to see a high participation rate.13

Table 2 reports the times series frequencies of the major categories of proposals. Panel A

shows that of the 3,390 proposals sponsored by shareholders during our sample period, 38% are

in the OTHERS category that aggregates proposals related to social, environmental and health

issues. The remainder consists of proposals addressing issues regarding the firm’s compensation,

board, and corporate governance. As shown in Panel B of Table 2, the majority of the

management-sponsored proposals fall in the compensation category, with other proposals on

capitalization, the board, anti-takeover amendments, and reorganization/mergers. As mentioned

earlier, routine and director election proposals are not included. The striking reduction in the last

category is primarily due to the fact that the financial crisis resulted in reduced M&A activity,

hence a reduction in related proposals.

3.2 Changes in Support Rate for Proposals

We use two measures of proposal success: 1) the mean support rate for the proposals, and

2) the percentage of proposals that pass, i.e., have a shareholder vote of greater than 50%. We                                                             12 It is clear that a little less half of the firms have a substantive management proposal or a shareholder proposal in a given year. Although the vast majority of firms have their annual meetings in the spring, the coverage for 2010 is not complete because the data ends in November 2010. 13 If institutional investors lend out their shares on the record date then they cannot vote as discussed by Aggarwal, Saffi and Sturgess (2013).

11

calculate the support rate as the ratio of FOR votes to the total votes or outstanding shares,

depending on the voting criterion for the proposal. Total votes may include only FOR and

AGAINST votes or also ABSTAIN votes. We report the mean support rate in Table 3 for

shareholder-sponsored and management-sponsored proposals. The mean support rate for

shareholder proposals increased over time with an average of 29.4% for the full sample period,

implying that 29.4% of all votes were in support of the shareholder-sponsored proposal. This rate

can be contrasted with the average support rate for management proposals of 83.4%. Following

the financial crisis, shareholder proposals in 2009 obtain the highest support rate of 37%.

On average, only 16.9% of shareholder proposals pass, i.e., receive favorable votes of

greater than 50%. Again this can be contrasted with the pass rate for management-sponsored

proposals, 97.4% of all management-sponsored proposals pass.

The proxy season in 2008 was already over by the time the financial crisis was in full

force. Therefore, the response to the financial crisis is captured primarily in the 2009 proxy

season. The largest number of both management and shareholder proposals is in 2009. The

results of voting on these proposals suggest that there was less support for management and more

support for shareholders in response to the financial crisis. Moreover, the highest pass rate for

shareholder proposals occurs in 2009 with 25.9% of the proposals passing. That year also has the

largest pass rate for board and compensation proposals, in fact, almost half of all board-related

proposals pass. There is also a large increase in board and compensation proposals sponsored by

shareholders in 2009. There is evidence of less support for management in 2009 with lower mean

pass rate for board and anti-takeover proposals sponsored by management.

The effects of the financial crisis on the 2009 proxy season are underscored by the trend

reversal in 2010, when management proposals had their highest pass rate at 98.2% and the pass

rate for shareholder proposals dropped to 16%.

3.3 Influence of Management and Proxy Advisor Recommendations

Voting patterns of investors have exhibited influence of management as well as proxy

advisors. For each firm in our sample, we obtain the general recommendation issued by ISS for

12

each proposal on the ballot.14 We find that in the earlier years of our sample, investor voting

tends to be closely associated with the recommendations of management and proxy advisors.

However, Figure 1 and Table 3 show that over time investors have become more independent in

their voting, such that they are less likely to follow the recommendations of either ISS or

management when they vote on shareholder-sponsored proposals. From 2004 to 2010, the

percentage of votes following management’s recommendation has declined as has the percentage

of votes following ISS’s recommendation. The percentage of votes following management’s

recommendation in shareholder-sponsored proposals has decreased from 73.7% in 2004 to

53.6% in 2010, the lowest support being in 2009. Similarly, the percentage of votes following

ISS’s recommendation has declined from 78.4% in 2004 to 57.5% in 2010. In Figure 1b, we split

voting with ISS based on proposals that the proxy advisor supports or opposes. As shown in the

figure, voting independently of ISS is occurring for shareholder proposals that ISS opposes and

not for proposals that ISS supports. 60% of all votes followed ISS’s negative recommendation in

2004 but by 2010 only 20% of the votes followed the negative recommendation. This trend is not

observed for proposals that ISS supports. The pattern indicates that over time investors are

voting in favor of shareholder proposals even though ISS puts out a negative recommendation.

Interestingly, while ISS has veered sharply away from management’s recommendations on

shareholder proposals, shareholders have been less willing to support the negative

recommendation by ISS. That is, investors have become more independent of both ISS and

management recommendations.

Two potential explanations for the independence results are that investors have become

more sophisticated about proxy voting with many institutional investors conducting their own

analysis and that other proxy advisory firms are also providing independent analyses (e.g., Glass

Lewis, Egan-Jones, and Proxy Governance) and sometimes their recommendations differ.

Evidence tends to support the former explanation as during our sample period, more institutional                                                             14 Every year, ISS Governance Services makes its overall proxy voting guidelines on specific proxy issues available. As mentioned earlier, ISS may also provide a custom-based recommendation and analysis to individual clients based on the client’s proxy policies and procedures.

13

investors stopped relying on ISS’s generic recommendation, and instead ISS developed

customized voting guidelines based on the institution’s preferences.

Table 4 shows the number of ISS’s positive and negative vote recommendations, for each

year, for shareholder and management-sponsored proposals. These recommendations come in the

form of mostly FOR or Against. Other recommendations, which happen rather infrequently,

include withhold, abstain, and do not vote. Since there exist very few withhold

recommendations, they are not reported in the table. In 2004, ISS recommended voting against

366 shareholder-sponsored proposals and voting for 234 proposals. The proportion of proposals

that ISS opposed compared to those they supported was 156.4%. Similarly, in 2005, ISS

recommended voting against more shareholder-sponsored proposals than they had

recommendations to vote for the proposals. However, as shown in Figure 2, from 2004 onwards

there exists a clear trend, ISS has become increasingly supportive of shareholder-sponsored

proposals. The proportion of proposals opposed compared to those supported has steadily

declined from 156.4% in 2004 to 30.5% in 2010. ISS has become more supportive of shareholder

proposals while management rarely supports shareholder proposals, therefore it is not surprising

that the opinion of ISS and management has drifted apart over time. There exist several potential

explanations for ISS’ increasing support of shareholder proposals. First, the types of proposals

being submitted by shareholders have changed over time which means that newer proposals are

more acceptable to ISS. Second, it is possible that the shareholder proposers have become more

sophisticated about the wording of proposals, hence garnering support from ISS. Finally, public

opinion about corporate governance issues has changed, and is reflected in the proxy advisor’s

recommendations.

This change in ISS support for shareholder proposals can be compared to their relatively

steady support for management proposals over time as ISS has always supported a large

proportion of management proposals. The proportion of Against relative to FOR

recommendations for management-sponsored proposals varies from a low of 17.2% in 2008 to a

high of 24.4% in 2009. Again, we see an apparent impact of the financial crisis, with lower

14

support for management proposals in 2009, where the proportion of Against to FOR

recommendations peaked. Although we have not included the management-sponsored proposals

defined as routine in our sample, many of the remaining management-sponsored proposals are

fairly routine. Hence it is not surprising that ISS supports a large majority of management

proposals, and investors vote in favor of these proposals.

3.4 Public Opinion

As discussed earlier, we hypothesize that an explanation for the changes over time in

investor voting behavior and proxy advisory firm recommendations is the influence of public

opinion. In order to examine this hypothesis, we develop four proxies for public opinion. Three

of these proxies capture public opinion and general dissatisfaction with executives at a broad

economy-wide level, while the third measure is firm-specific. The first measure, CONFIDENCE,

is derived from the annual Gallup Poll Survey of confidence in different institutions of American

society, one of which reflects the percentage confidence in banks.15 This measure reflects a broad

opinion about the financial sector in general. The Gallup Survey asks how much confidence

individuals have in each institution with responses being: a great deal, quite a lot, some, or very

little. We combine the percentage of the top two responses: “great deal” and “quite a lot”. Figure

3 shows that during our sample period the highest confidence in banks occurred in 2004 when

53% had high confidence in banks. Not surprisingly given the financial crisis and the ill-will that

was engendered toward financial institutions, the lowest confidence level of 22% occurred in

2009.

The second measure of public opinion we develop is based on media coverage of

executive compensation. We focus the media coverage on executive compensation because it is

the general corporate governance proxy with a long history of being a concern of the public.16

We count the number of articles from major news and business publications from the Factiva

                                                            15 The institutions included are the church or organized religion, the military, the U.S. Supreme Court, banks, public schools, newspapers, Congress, television news, organized labor, the presidency, police, medical system, criminal justice system, big business, small business (added in 2007), and health maintenance organizations. 16 See, for example, Murphy (2013).

15

news database that contain at least one of the following keywords: "CEO compensation", "CEO

salary", "CEO pay", "executive compensation", "executive salary", and "executive pay". During

the sample period, 2004-2010, there are 31,802 articles on executive compensation. As shown in

Figure 4, the number of compensation-related articles peaked in 2009. Our measure is based on

the aggregate monthly count of these articles. That is, MEDIA is the natural log of the monthly

number of articles related to compensation based on a search of major news and business

publications on Factiva. The third measure, MEDIA-NEGATIVE, is the natural log of the

percentage of words in the articles related to compensation that match the negative words

proposed by Kuhnen and Niessen (2012) or Loughran and McDonald (2011).

Our fourth measure of public opinion, MEDIA-FIRM, is a proxy for public opinion at the

firm-level. In addition to the keywords used for MEDIA, we also include different versions of a

firm’s name in order to pick-up media coverage for the specific company. MEDIA-FIRM is

measured as the percentage of the number of articles related to compensation on a specific firm

in a given month, normalized by the total number of articles related to compensation on this firm

over the time period 2004-2010. As shown in Figure 5, the firm-level media count also peaked in

2009 in response to the financial crisis. The plot suggests more media coverage about

compensation around the proxy voting season in April. Firms must disclose executive and

director compensation in their proxies.

3.5 Firm Characteristics

We obtain firm characteristics from Compustat and CRSP. We use several firm-specific

control variables in our analysis: log of total assets in U.S. dollars (SIZE), debt to assets (DEBT),

cash holdings to assets (CASH), capital expenditure to assets (CAPEX), return on assets (ROA)

and annual stock return minus the value-weighted stock market return (EXCESS RET). We

winsorize firm-level accounting variables other than SIZE, namely DEBT, CASH, CAPEX, and

ROA as well as EXCESS RET, at the upper and lower 1% levels.

We use the firm-level corporate governance index GOV41 as in Aggarwal, Erel, Ferreira and

Matos (2011). GOV41 assigns a value of one to each of 41 governance attributes if the company

16

meets minimally acceptable governance guidelines on that attribute, and zero otherwise. Table 5

shows that the median firm in the sample with shareholder proposals is more than ten times

larger than the median firm with management proposals. Firms with shareholder proposals have

higher median governance, higher median returns on assets and higher median debt. Wilcoxon

rank sum test shows that differences in medians are significant at the 1% level.

4. Voting Behavior of Investors and Public Opinion

4.1 Changes in Voting Pattern over Time

Earlier in Figure 1, we saw changes in voting pattern over time for proposals, which indicate

that investors have become more independent in their voting, particularly with respect to

shareholder-sponsored proposals. That is, shareholders have been voting less with the

recommendations of proxy advisory firm ISS, and also with the recommendations of

management. We examine the relation between the support rate for a shareholder-sponsored

proposal and year dummies in order to examine the time trend in the voting. The dependent

variable is the support rate for the proposal and the independent variables consist of dummy

variables for each of the years from 2005 to 2010, with 2004 as the benchmark year. In Table 5

column 1 we report the results with industry fixed effects; in column 2 we add proposal fixed

effects; and in column 3 we include only firm fixed effects. We use the four proposal types

discussed earlier: Board, Corporate Governance, Compensation, and Others.

In column 1 of Table 6, each of the year dummies is positive and significant at the 1% level,

implying that relative to 2004, there is higher support for shareholder-sponsored proposals, in

subsequent years. The coefficient of the 2009 year dummy is the largest at 0.214, which is also

economically significant. In order to examine whether the increase in support rate for

shareholder-sponsored proposals is due to proposal type changing over time, in column 2 we

control for the four categories of proposal fixed effects. The coefficients continue to be

significant for the years, 2006 and 2010. Once we control for firm fixed effects in Column 3, we

17

find that all the year dummies are still significantly positive, indicating that for individual firms

voting in support of shareholder proposals is increasing over time.

In addition to year dummies, columns 4-6 of Table 6 include firm-level characteristics from

the preceding fiscal year as controls. The firm-level controls include SIZE, log of total assets;

ROA, return on assets; CAPEX, capital expenditure to assets; CASH, cash to assets; DEBT, debt

to assets; EXCESS RETURNS, annual stock market return minus market return; and GOV41,

firm-level measure of corporate governance. The results suggest less support for shareholder

proposals in larger firms as the coefficient of SIZE is negative and significant in two of the three

models. However, typically shareholder proposals are submitted only for the largest firms. Firm-

level corporate governance is not significant in explaining the support for shareholder proposals.

ROA becomes positive and significant when firm fixed effects are included. Even after including

firm-level characteristics as controls, we continue to find that the support rate for shareholder

proposals has increased relative to 2004. There is clearly a 2009 effect reflecting the response to

the financial crisis of 2008 and the declining confidence that year as indicated by the confidence

in banks. The inclusion of proposal and firm fixed effects increases the explanatory power of the

model as measured by adjusted R2.

4.2 Voting and the Role of Public Opinion

In the past decade or so there has been much public debate about corporate scandals and

more fundamentally, corporate governance issues. This debate started after the financial

scandals at Enron and Worldcom, and more recently picked up steam as a result of the financial

crisis of 2008. As anecdotal evidence and Figure 3 suggest, these scandals and crises appear to

have shaken investor confidence. We examine the impact of public opinion on voting patterns in

shareholder-sponsored proposals by including three different measures of public sentiment as

independent variables: CONFIDENCE, the percentage confidence in banks from the annual

Gallup Poll Survey, MEDIA, a measure of the public concern regarding executive compensation,

which is defined as the number of articles related to compensation, and MEDIA-NEGATIVE, is

the percentage of words in the articles related to compensation that match the negative words

18

proposed by Kuhnen and Niessen (2012) or Loughran and McDonald (2011). CONFIDENCE is

from the fiscal year just prior to the vote; MEDIA and MEDIA-NEGATIVE are from the month

preceding the vote month.

Columns 1-9 of Table 7 report the results. The sample includes types of shareholder-

sponsored proposals with at least 65 observations (the median count for director and executive

compensation-related proposals) between 2004 and 2010. However, results are similar using the

full sample of proposals. The primary independent variable in columns 1-3 is CONFIDENCE

and in columns 4-6 is MEDIA, and in columns 7-9 is MEDIA-NEGATIVE. In column 1, the

coefficient of CONFIDENCE is -0.309, significant at the 1% level. The negative coefficient

implies that there is more support for shareholder-sponsored proposals when confidence in the

banking system is low. The coefficient is negative and significant when proposal fixed effects

are included in column 2, suggesting that the significant effect of media continues even after

controlling for proposal types. The coefficient is negative and significant even when firm fixed

effects are included in column 3.

MEDIA is used as the measure of public opinion as shown in columns 4-6, has a positive and

significant coefficient in columns 4 and 6 at the 1% level. Similar results are reported in columns

7-9 using MEDIA-NEGATIVE as the proxy for public opinion. We find that more discussion

about executive compensation in the media results in less support for management, hence more

support for shareholder proposals. Votes in support of shareholder proposals are effectively votes

against management because it is rare to have management supporting shareholder proposals.

There is some indication that support for shareholder proposals is lower in larger firms as

indicated by the negative coefficient of the SIZE variable. Support for shareholder proposals is

higher at firms that have stronger corporate governance.

In Table 8, for robustness, we repeat the above analysis for only compensation proposals.

The sample size is much smaller when we limit the analysis to only this type of proposal. We

find an even stronger association between media coverage and voting for this sample compared

19

to the sample of all proposals. The coefficient of the media variables are larger for compensation

proposals relative to all proposals.

4.3 Voting and Firm-Level Media Coverage

We next examine the firm-level media coverage of compensation, the results of which are

provided in Table 9. Public opinion at the firm level, MEDIA-FIRM, is measured as the

percentage of the number of articles related to compensation on a specific firm in a given month,

normalized by the total number of articles related to compensation on this firm over the time

period 2004-2010. The dependent variable in these regressions is the support rate, the percent of

votes cast in favor of the proposal. Columns 1, 3 and 5 show the results with only industry and

year fixed effects while columns 2, 4 and 6 also include proposal fixed effects. We do not

include firm fixed effects because the variation in the public opinion proxy is at the firm level.

As can be seen from Table 9, the support rate for shareholder proposals is increasing in the

relative amount of firm-level media coverage, again suggesting that public opinion regarding

executive compensation is influencing the voting on shareholder proxies. The inclusion of the

explanatory variables GOV41 or ISS does not change this finding. As seen in columns 5 and 6, if

ISS supports a proposal then the proposal is more likely to receive higher support. However, the

coefficient of MEDIA-FIRM continues to be positive and significant. Therefore, even after

controlling for ISS’s recommendation and proposal type, we find that public opinion influences

voting. The coefficient is 0.40 in column 5 with industry fixed effects, and 0.34 in column 6 with

both industry and proposal fixed effects. The standard deviation of the variable, MEDIA-FIRM,

is 0.079 suggesting that, on average, support for shareholder proposals increases by 3.15%-

2.69% if there is a one standard deviation increase in media coverage. We consider these results

to be economically meaningful.

5. Public Opinion and the Recommendations of Proxy Advisors

In this section we examine whether public opinion is associated with the recommendations

of proxy advisors themselves. ISS has an extensive proxy policy formulation process which

20

includes conducting an annual policy survey of both the proxy voters (institutional investors) as

well as management and other market participants. They also receive additional feedback

through industry roundtables and feedback during the proxy season.

The draft policies are put out for comments that are then incorporated before the final update

is released. Therefore, ISS attempts to incorporate the views of the corporate governance

community and the market in formulating its policies. If public opinion reflects the views of the

market then this sentiment should get reflected in ISS’s recommendations. Not surprisingly,

earlier we saw that a positive recommendation by the proxy advisor is associated with a higher

support rate for the proposal. However, this correlation does not necessarily imply causation as

both ISS’s recommendations and investors’ voting decisions could be affected by other

influences such as public opinion.

We have shown the positive influence of public opinion and ISS’s recommendation on

voting. Next, we analyze whether there is any association between public opinion and the

recommendation of ISS itself. In Table 10 the dependent variable is a dummy equal to one if ISS

recommends voting for the proposal. The explanatory variables include firm characteristics as

controls. In columns 1-3, public opinion is measured by CONFIDENCE, in columns 4-6 by

MEDIA, and in columns 7-9 by negative media coverage MEDIA-NEGATIVE. As in our

previous regressions, the analysis is done with industry and year firm effects, with industry, firm

and proposal fixed effects and with year and firm fixed effects. The coefficient of

CONFIDENCE is negative and significant at the 1% level in all three estimations. The results

imply that low confidence in the banking system, which could be interpreted as low confidence

in corporations in general, is associated with ISS being more supportive of shareholder

proposals.

As seen in columns 4-6, the coefficient of MEDIA is positive and significant at the 1% level

in each model. Similarly, the coefficient of MEDIA-NEGATIVE is positive and significant.

Therefore, we conclude that more media coverage about issues dealing with executive

compensation is associated with ISS being more supportive of shareholder-sponsored proposals.

21

In unreported results, we find the association between firm-level media coverage and ISS’s

recommendation to be far weaker and not significant in two of the three models. This result is

not surprising, we would expect the association to be stronger between the proxy advisor’s

recommendation and the broader proxies of public opinion because ISS obtains feedback from

market participants on its benchmark policies, and not on recommendations for individual firms.

6. Voting by Mutual Funds

If public opinion is primarily responsible for the changes in voting on shareholder

proposals, then we would expect to find it particularly important in the voting behavior of a

group of institutional investors most likely to be influenced by public opinion. Given that mutual

funds have direct dealings with a large number of retail investors and they are required to

disclose how they voted, they make an ideal sample in which to further test our hypotheses.

Fund management companies typically have an oversight process, with the fund’s board

involvement, to monitor the funds’ proxy voting. The passage of Rule 206(4)-6 of the Securities

and Exchange Commission (SEC) in 2003, requires funds to adopt and implement proxy voting

policies and procedures, and make voting record available to clients.17 Funds must file Form N-

PX, which identifies specific proposals on which the fund has voted their portfolio securities and

discloses how the fund voted on each, whether for, against or withhold. Proxy voting records for

the twelve-month period ending June 30, have to be filed by no later than August 31 of each

year. According to the SEC, “This disclosure enables fund shareholders to monitor their funds’

involvement in the governance activities of portfolio companies.”18

                                                            17 These policies include information on issues such as corporate governance matters (including

changes in the state of incorporation, mergers and other corporate restructurings, and anti-takeover provisions, i.e., staggered boards, poison pills, and supermajority provisions); changes to capital structure (including increases and decreases of capital and preferred stock issuance); stock option plans and other management compensation issues; and social and corporate responsibility issues. See http://www.sec.gov/rules/final/33-8188.htm 18 http://www.sec.gov/answers/mfinfo.htm

22

We merge the mutual fund NP-X voting data with the CRSP mutual fund data by

matching based on fund name. Our sample includes voting by 13,313 individual funds that are

part of 475 mutual fund families from 2004 to 2010.

Table 11 shows the total number of votes by mutual funds each year for both

management and shareholder proposals. There are a total of 4,283,930 votes, of which 1,844,636

are for shareholder proposals and 2,439,294 are for management proposals. The largest number

of votes cast is in 2009, driven by large increases in board and compensation proposals in that

year.

Table 11 shows that, consistent with our earlier analysis of voting by all shareholders,

over time mutual funds have become less likely to follow the recommendation of management or

the proxy advisor for shareholder-sponsored proposals. For these proposals, funds voted with

management’s recommendation 73.6% of the time but by 2010 the support had decreased to

53.9%. Funds are more likely to follow the recommendation of ISS than that of management but

here also we see that over time, funds are less likely to follow the recommendation of ISS. In

2004 they voted with ISS 78.6% of the time but by 2010 the number had dropped to 59.2%. The

results also suggest that, particularly during the financial crisis, mutual funds were even more

likely to vote differently from the recommendation of management and ISS. These results can be

contrasted with the results for management-sponsored proposals in which we find no consistent

pattern of changes in voting with management or the proxy advisor. However, it should again be

noted that many of the management proposals are noncontroversial so it is not surprising to

observe such high support rates.

We test our hypotheses regarding the association of mutual fund voting with public opinion

by regressing a favorable vote on a shareholder proposal against our four measures of public

opinion, CONFIDENCE, MEDIA, MEDIA-NEGATIVE and MEDIA-FIRM. The results of the

tests between voting by mutual funds on shareholder proposals and the role of public opinion are

reported in Table 12. The sample includes types of shareholders-sponsored proposals with at

least 65 observations (the median count for director and executive compensation-related

23

proposals) between 2004 and 2010.19 The dependent variable is a dummy equal to one if the

mutual fund voted in favor of the proposal. For each of the three proxies of public opinion, we

report results with and without proposal fixed effects. We include fund and firm characteristics

including the governance index as explanatory variables (although due to space limitations we do

not report the coefficients for the firm characteristics). Each estimation includes industry, fund

and proposal fixed effects. Standard errors are corrected for clustering of observations at the fund

level.

Table 12 shows that the coefficient of CONFIDENCE is negative and significant implying a

negative relation between mutual fund support for shareholder-sponsored proposals and public

confidence in banks. If investors have a low opinion about the banking system (and presumably

corporations in general), then this low opinion gets reflected in their voting behavior. In columns

2 and 3, the results show that more total media coverage (MEDIA) or negative media coverage

(MEDIA-NEGATIVE) about executive compensation is associated with more support for

shareholder proposals. The table also shows that fund characteristics matter in how mutual funds

vote. The coefficient of INDEX is negative and significant in all six estimations suggesting that

index funds tend to passive and are less likely to vote against management, therefore are less

likely to support shareholder-sponsored proposals.

Finally, in results not shown, we find that the recommendation of the proxy advisory firm

has a strong association with how funds vote. If ISS recommends voting for the proposal then the

proposal is more likely to receive the support of mutual funds, however, as pointed out earlier,

this correlation does not necessarily mean causation. Even after including the ISS

recommendation, the role of public opinion continues to be significant.

                                                            19 The median number of proposals during this time period is 149 for Board proposals, 98 for Corporate Governance proposals, and 67 for Others.

24

7. Conclusions

In this paper we consider the association between public opinion and the evolution in

corporate governance perceptions shown in proxy voting by examining voting on proxy

proposals in general and by mutual funds, in particular. We examine the changes in investor

voting patterns from 2004 to 2010 and find that, on average, shareholders, particularly

institutional investors, are participating in proxy voting at high rates and that their support for

shareholder proposals has been increasing. The high participation rates and increasing support

for shareholder proposals suggest that shareholder proposals are becoming a more important part

of the governance process. This result is consistent with the result of Aggarwal, Saffi, and

Sturgess (2013) that institutional investors are recalling shares from their securities lending

portfolios, at the time of proxy voting, in order to vote their proxies.

We also find that shareholders have become more independent of both management and

the proxy advisory firm. Investors are not following the proxy advisor’s recommendation

specifically when the recommendation is negative. Moreover, we find that the changes in voting

behavior can be partially explained by changes in proposal types, the financial crisis, and most

importantly, public opinion.

Our results suggest that public opinion, as measured through either Gallup Poll survey or

media coverage at the aggregate and firm level, influences shareholder voting. The implications

of these results are that financial intermediaries, such as mutual funds, pay attention to their

shareholders’ preferences regarding corporate governance. These results hold even after

controlling for the recommendations of the proxy advisor.

25

References

Aggarwal, R., I. Erel, M. Ferreira, and P. Matos, 2011, Does Governance Travel Around the World? Evidence from Institutional Investors, Journal of Financial Economics 100.1, 154-182.

Aggarwal, R., P. Saffi, and J. Sturgess, 2013, Does Proxy Voting Affect the Supply and/or Demand for Securities Lending? Working Paper, Georgetown University.

Bainbridge, S., 2006, The Case for Limited Shareholder Voting Rights, UCLA Law Review 53.

Bebchuk, L., 2005, The Case for Increasing Shareholder Power, Harvard Law Review 118, 833-914.

Berle, A., Jr., and G. Means, 1932, The Modern Corporation and Private Property, (Macmillan, New York, NY).

Bethel, B., and S. Gillan, 2002, The Impact of the Institutional Regulatory Environment on Shareholder Voting, Financial Management, 29-54.

Cai, J., J. Garner and R. Walkling, 2009, Electing Directors, Journal of Finance 64, 2389-2421.

Cai, J., J. Garner, and R. Walkling, 2011, Paper Tiger? An Empirical Analysis of Majority Voting, Working Paper, Drexel University.

Christoffersen, S., C. Geczy, D. Musto and A. Reed, 2007, Vote Trading and Information Aggregation, Journal of Finance 62, 2897-2929.

Collins, S., 2010, Trends in Proxy Voting by Registered Investment Companies, 2007-2009. Investment Company Institute Research Perspective 16.

Cotter, J., A. Palmiter and R. Thomas, 2010, ISS Recommendations and Mutual Fund Voting on Proxy Proposals, Villanova Law Review 55, 1-56.

Cvijanovic, D., A. Dasgupta, and K. Zachariadis, 2014, Ties that Bind: How Business Connections Affect Mutual Fund Activism, Working Paper, University of North Carolina at Chapel Hill.

Dyck, A., N. Volchkova, and L. Zingales, 2008, The Corporate Governance Role of the Media: Evidence from Russia, Journal of Finance 63, 1093-1135.

Easterbrook, F. and D. Fischel, 1983, Voting in Corporate Law, Journal of Law and Economics 26.

Ertimur, Y., F. Ferri, and V. Muslu 2011, Shareholder Activism and CEO Pay, Review of Financial Studies 24, 535-592.

Gillan, S., and L. Starks, 2007, The Evolution of Shareholder Activism in the United States, Journal of Applied Corporate Finance 19, 55-73.

Gillan, S., and L. Starks, 2000, Governance Proposals and Shareholder Activism: The Role of Institutional Investors, Journal of Financial Economics 57, 275-305.

Hayden, G., and M. Bodie, 2010, Shareholder Democracy and the Curious Turn Toward Board Primacy, William and Mary Law Review 51, 2071-2121.

26

Iliev, P. and M. Lowry, 2013, Are Mutual Funds Active Voters? Pennsylvania State University, Working Paper.

Kahan, M., and E. Rock, 2008, The Hanging Chads of Corporate Voting, Georgetown Law Review 96.

Kuhnen,C. and A. Niessen, 2012, Public Opinion and Executive Compensation, Management Science 58, 1249-1272. Levit, D. and N. Malenko, 2011, Nonbinding Voting for Shareholder Proposals, Journal of

Finance 66, 1579-1614.

Loughran, T. and B. McDonald, 2011, When is a Liability not a Liability? Textual Analysis, Dictionaries, and 10-Ks, Journal of Finance 65, 35-65.

Lublin, J., March 21, 2002, Proxy Voting is a Fiduciary Duty, SEC Chief Says in Letter to Group, The Wall Street Journal, C20.

Maug, E. and K. Rydqvist, 2011, Do Shareholders Vote Strategically? Voting Behavior, Proposal Screening, and Majority Rules, Review of Finance 13, 47-79.

Murphy, K., 2013, Executive Compensation – Where We Are and How We Got There, Handbook of the Economics of Finance, Edited by G. Constantinides, M. Harris and R. Stulz, Elsevier, 211-356

Pound, J., 1988, Proxy Contests and the Efficiency of Shareholder Oversight, Journal of Financial Economics 20, 237-265.

Rothberg, B., and S. Lilien, 2005, Mutual Fund Proxy Votes, Working Paper, http://papers.ssrn.com/sol3/papers.cfm?abstract_id=669161.

0

Figure 1 Voting with Management and Proxy Advisors

The figure shows the total percentage of votes cast following the recommendation of management and the proxy advisory firm, ISS for the period January 2004 to November 2010. Percentages are shown separately for shareholder-sponsored and management-sponsored proposals.

 

Figure 1b Voting with Proxy Advisor’s FOR and AGAINST Recommendation

The figure shows the total percentage of votes cast following ISS’s FOR and AGAINST recommendations for shareholder-sponsored proposals during the period January 2004 to November 2010.

40%

50%

60%

70%

80%

90%

100%

2004 2005 2006 2007 2008 2009 2010

Percentage Votes with Management and ISS Recommendations

% of Votes with MGMT Rec: Management‐Sponsored Proposals% of Votes with MGMT Rec: Shareholder‐Sponsored Proposals% of Votes with ISS Rec: Management‐Sponsored Proposals% of Votes with ISS Rec: Shareholder‐Sponsored Proposals

0%

10%

20%

30%

40%

50%

60%

70%

2004 2005 2006 2007 2008 2009 2010

Percentage Votes with Proxy Advisor FOR and AGAINST Recommendations

%ofVoteswithISSFORRecommendation %ofVoteswithISSAGAINSTRecommendation

1

Figure 2 Change in ISS’s Recommendations

The figure shows the percentage of proposals that ISS supports and recommends voting in favor for the period January 2004 to November 2010. Percentages are shown separately for shareholder-sponsored and management-sponsored proposals.

30%

40%

50%

60%

70%

80%

90%

2004 2005 2006 2007 2008 2009 2010

% of Proposals Supported by ISS

Management-Sponsored Proposals Shareholder-Sponsored Proposals

2

Figure 3

Public Opinion over Time

Public opinion is measured yearly as the percentage confidence in banks from the Gallup Poll Survey of confidence in institutions. The Gallup Survey asks Americans how much confidence they have in each institution – a great deal, quite a lot, some, or very little. The institutions included are the church or organized religion, the military, the U.S. Supreme Court, banks, public schools, newspapers, Congress, television news, organized labor, the presidency, police, medical system, criminal justice system, big business, small business, and health maintenance organizations. We combine the % of “great deal” and “quite a lot” responses for banks.  

 

  

0%

10%

20%

30%

40%

50%

60%

2004 2005 2006 2007 2008 2009 2010 2011

Public Opinion - Confidence in Banks

3

Figure 4 Aggregate Media Coverage about Compensation

The figure shows the number of articles per month, discussing executive compensation. The results are based on searching major news and business publications on Factiva. The keyword searched include CEO compensation or CEO salary or executive compensation or executive salary.

Figure 5 Firm-Level Media Coverage about Compensation

The figure shows the number of articles per month, discussing executive compensation at the firm. The results are based on searching major news and business publications on Factiva. The keyword searched include CEO compensation or CEO salary or executive compensation or executive salary.

0

200

400

600

800

1000

1200

1400

1600

Aggregate Media Coverage about Compensation

02468

101214161820

Firm-Level Media Coverage about Compensation

4

Table 1 Number of Proposals by Year

The sample consists of 18,439 proxy proposals during the period January 2004-November 2010. Each proposal is sponsored either by shareholders (SH) or by management (MGMT). We do not include director election proposals sponsored by management that tend to be routine. The percentage of proposals that are shareholder-sponsored is also reported. Two measures for voting outcome are reported for shareholder-sponsored and management-sponsored proposals. Mean support rate based on percentage of votes in favor of the proposal, and the percentage of proposals that passed.

Year # of Firms

# of Proposals

# of SH Sponsored

# of MGMT Sponsored

% SH Proposals

Participation Rate for SH- Sponsored

Participation Rate for MGMT- Sponsored

2004 1,638 2,926 605 2,321 20.7% 75.4% 81.2% 2005 1,456 2,534 504 2,030 19.9% 74.0% 78.1% 2006 1,474 2,599 556 2,043 21.4% 74.9% 78.0% 2007 1,323 2,469 604 1,865 24.5% 74.8% 79.1% 2008 1,389 2,539 559 1,980 22.0% 73.6% 78.8% 2009 1,625 3,074 653 2,421 21.2% 70.2% 77.6% 2010 1,290 2,297 509 1,788 22.2% 74.1% 78.7% Total 10,195 18,438 3,990 14,448 21.7% 78.80% 73.78%

5

Table 2 Type of Proposal by Year

Shareholder-sponsored proposals during the period January 2004-November 2010 are categorized into four sub-categories: board, corporate governance, compensation and others, as shown in Panel A. The sample of management-sponsored proposals is categorized into five sub-categories: board, capitalization, reorganization/merger, compensation, and anti-takeover, as shown in Panel B. Proposals in the others category include proposals related to social issues, human rights and general economic issues.

Panel A: Shareholder-Sponsored Proposals

2004 2005 2006 2007 2008 2009 2010 Total Board 63 96 141 124 141 244 136 945Corp Governance 118 84 106 92 66 92 84 642Compensation 159 120 96 181 146 144 125 971Others 265 204 213 207 206 173 164 1,432Total 605 504 556 604 559 653 509 3990

Panel B: Management-Sponsored Proposals

Board 126 126 121 160 214 225 133 1,105Capitalization 293 243 258 221 244 286 196 1,741Reorg/Merger 178 73 108 130 75 40 56 660Compensation 1,634 1,531 1,445 1,226 1,333 1,734 1,252 10,155Anti-Takeover 90 57 111 128 114 136 151 787Total 2,321 2,030 2,043 1,865 1,980 2,421 1,788 14,448

6

Table 3 Voting with Management and Proxy Advisors

The table shows the support rate and percentage of proposals that pass. Support rate is the ratio of FOR votes to the total votes or outstanding shares, depending on the proposal-type. Also included is the percentage of votes that followed management’s recommendation, and the percentage of votes that followed the proxy advisory firm, ISS’s recommendation. Each proposal is sponsored either by shareholders (SH) or by management (MGMT).

Support Rate % of Proposals that Pass Voting with Management Voting with

ISS

Year SH-

Sponsored MGMT-

Sponsored SH-

Sponsored MGMT-

Sponsored SH-Sponsored

MGMT-Sponsored

SH-Sponsored

MGMT-Sponsored

2004 23.6% 81.6% 15.5% 97.2% 73.7% 78.9% 78.4% 83.9% 2005 25.4% 82.1% 14.1% 97.2% 69.8% 80.5% 71.9% 85.4% 2006 28.8% 83.7% 15.4% 97.9% 60.8% 82.9% 69.6% 86.8% 2007 28.4% 83.7% 14.5% 97.2% 60.4% 86.1% 66.4% 88.8% 2008 29.3% 85.5% 15.1% 97.6% 61.2% 85.1% 62.8% 88.3% 2009 37.0% 83.4% 25.9% 96.8% 51.3% 81.4% 60.6% 85.1% 2010 31.8% 85.5% 16.0% 98.2% 53.6% 85.5% 57.5% 87.7%

7

Table 4 Proxy Advisory Firm’s Recommendation

The proxy advisory firm, ISS, provides a recommendation for each shareholder and management sponsored proposal on the ballot. ISS mainly recommends “FOR” or “Against” vote for a given recommendation. Other recommendations that happen rarely include withhold, do not vote, and abstain. A recommendation FOR a shareholder proposal typically implies that ISS is making a recommendation against management’s recommendation. The table shows the number of recommendations Against, FOR, and the ratio of Against to FOR recommendations, by year for shareholder and management-sponsored proposals. Year Shareholder-Sponsored Management-Sponsored

Against FOR % Against/FOR Against FOR % Against/FOR

2004 366 234 156.4% 354 1,938 18.3% 2005 252 248 101.6% 345 1,668 20.7% 2006 235 319 73.7% 309 1,729 17.9% 2007 247 349 70.8% 318 1,529 20.8% 2008 198 326 60.7% 285 1,661 17.2% 2009 146 440 33.2% 462 1,897 24.4% 2010 116 380 30.5% 304 1,478 20.6%

8

Table 5 Firm Characteristics by Proposal Type

The table provides descriptive statistics for firms with shareholder and management-sponsored proposals. Median values are reported for total assets, return on assets, capital expenditure to total assets, cash and short-term assets to total assets, debt to total assets, excess return, defined as annual stock return minus the value-weighted market return, and GOV41. GOV41 is the percentage of the 41 governance attributes that a firm meets, an index of 1 means that a firm has adopted all 41 governance provisions. *** represents statistical significance at the 1% level, using Wilcoxon rank-sum test for the difference in medians.

Shareholder-Sponsored

(1)

Management- Sponsored

(2)

Difference (1) – (2)

Total Assets (MM) $12, 680 $1,238 $11,442***

Return on Assets 8.13% 6.75% 1.38%***

Capital Expd. to Assets

3.53% 2.63% 0.91%***

Cash to Assets 4.88% 6.29% -1.41%***

Debt to Assets 24.73% 18.49% 6.24%***

Excess Return -4.69% -2.40% 2.29%***

GOV41 0.68 0.63 0.05***

9

Table 6 Changes in Voting Pattern

This table shows estimates of panel regressions for shareholder-sponsored proposals, during the period January 2004 to November 2010. The dependent variable is the support rate, percent of votes cast in favor of the proposal. In columns 1-3, the only independent variables are the year dummies for each year from 2005 to 2010. Columns 4-6 include firm-level attributes from the preceding fiscal year as controls. SIZE is log of total assets; ROA is return on assets; CAPEX is capital expenditure divided by total assets; CASH is cash and short-term assets divided by total assets; DEBT is debt to total assets; and EXCESS RET is annual stock return minus the stock market return. They are all winsorized at 1%. GOV41 assigns a value of one to each of 41 governance attributes if the company meets minimally acceptable governance guidelines on that attribute, and zero otherwise. Column 1and 4 reports results without any fixed effects; columns 2 and 5 include proposal type fixed effects; and column 3 and 6 include firm-level fixed effects. The models reported in columns 4 and 5 also include industry fixed effects. Errors are corrected for clustering at the firm level. *, **, *** reflect statistical significance at the 10 percent, 5 percent, and 1 percent levels.

(1) (2) (3) (4) (5) (6)

2005 0.055*** 0.026** 0.044*** 0.043*** 0.023 0.041** (3.95) (2.06) (2.63) (2.69) (1.63) (1.97) 2006 0.102*** 0.052*** 0.098*** 0.091*** 0.050*** 0.105*** (6.44) (3.63) (5.10) (4.76) (2.97) (4.22) 2007 0.075*** 0.044*** 0.073*** 0.064*** 0.042** 0.083*** (4.94) (3.32) (4.17) (3.28) (2.50) (3.05) 2008 0.129*** 0.069*** 0.108*** 0.114*** 0.066*** 0.130*** (4.68) (3.21) (5.40) (3.66) (2.61) (4.30) 2009 0.214*** 0.114*** 0.164*** 0.184*** 0.100*** 0.185*** (7.83) (5.49) (8.12) (7.49) (4.41) (5.92) 2010 0.110*** 0.051*** 0.109*** 0.100*** 0.051*** 0.134*** (7.10) (3.37) (5.84) (4.79) (2.74) (4.53) SIZE -0.030*** -0.019*** -0.018 (-5.85) (-4.34) (-0.61) ROA -0.186 -0.124 0.440*** (-1.53) (-1.30) (3.19) CAPEX 0.115 0.223 -0.464 (0.45) (1.11) (-1.28) CASH -0.117 -0.054 0.036 (-1.04) (-0.56) (0.25) DEBT -0.048 -0.068* 0.022 (-0.99) (-1.72) (0.21) EXCESS RET -0.006 0.008 0.018 (-0.31) (0.54) (1.19) GOV41 0.107 0.030 -0.201 (1.16) (0.41) (-1.39) Constant 0.129*** 0.222*** 0.247*** 0.373*** 0.391*** 0.534* (2.82) (5.54) (18.58) (4.93) (6.12) (1.74)

Fixed Effects Industry Industry & Proposal

Firm Industry Industry & Proposal

Firm

Observations 2425 2425 2462 2237 2237 2237 Adj. R2 0.091 0.320 0.387 0.129 0.339 0.372

10

Table 7 All Proposals: Influence of Public Opinion on Voting

This table shows estimates of panel regressions of voting support rate on role of public opinion, for all shareholder-sponsored proposals, during the period January 2004 to November 2010. The dependent variable is the support rate, percent of votes cast in favor of the proposal. Public opinion is measured using three proxies from the month or year preceding the vote. CONFIDENCE is the percentage confidence in banks from the annual Gallup Poll Survey of confidence in institutions. MEDIA is the natural log of the number of articles related to compensation based on monthly search of major news and business publications on Factiva. MEDIA-NEGATIVE is the monthly percentage of words in the articles related to compensation (MEDIA) that match the negative words proposed by Kuhnen and Niessen (2012) or Loughran and McDonald (2011). The keyword searched included CEO compensation or CEO salary or executive compensation or executive salary. Errors are clustered at the firm level. *, **, *** reflect statistical significance at the 10 percent, 5 percent, and 1 percent levels.

(1) (2) (3) (4) (5) (6) (7) (8) (9) CONFIDENCE -0.309*** -0.145*** -0.251*** (-5.73) (-3.12) (-4.26) MEDIA 0.054*** 0.026** 0.042*** (3.74) (1.98) (3.71) MEDIA-NEGATIVE 0.057*** 0.026* 0.039*** (3.80) (1.92) (3.35) SIZE -0.034*** -0.020*** 0.009 -0.035*** -0.021*** 0.028 -0.035*** -0.021*** 0.030 (-6.44) (-4.75) (0.28) (-6.64) (-4.84) (0.93) (-6.61) (-4.82) (1.01) ROA -0.189 -0.124 0.515*** -0.199 -0.128 0.442*** -0.199 -0.128 0.444*** (-1.45) (-1.26) (3.34) (-1.59) (-1.35) (3.15) (-1.58) (-1.35) (3.12) CAPEX 0.261 0.305 -0.086 0.254 0.299 -0.207 0.257 0.303 -0.188 (0.93) (1.47) (-0.22) (0.95) (1.48) (-0.53) (0.96) (1.50) (-0.48) CASH -0.128 -0.059 0.070 -0.089 -0.041 0.140 -0.093 -0.042 0.143 (-1.09) (-0.60) (0.43) (-0.78) (-0.42) (0.86) (-0.81) (-0.44) (0.86) DEBT -0.043 -0.067* 0.102 -0.041 -0.066 0.098 -0.040 -0.065 0.101 (-0.87) (-1.66) (0.89) (-0.83) (-1.63) (0.81) (-0.80) (-1.61) (0.83) EXCESS RET -0.008 0.007 0.025 -0.009 0.007 0.021 -0.008 0.007 0.023 (-0.40) (0.49) (1.59) (-0.43) (0.44) (1.33) (-0.40) (0.48) (1.42) GOV41 0.279*** 0.129* 0.183 0.329*** 0.150** 0.264** 0.337*** 0.155** 0.286** (3.07) (1.87) (1.62) (3.60) (2.17) (2.33) (3.77) (2.29) (2.55) Constant 0.514*** 0.454*** 0.149 0.048 0.231*** -0.451 0.249*** 0.330*** -0.313 (6.17) (6.31) (0.45) (0.55) (2.86) (-1.54) (3.61) (5.45) (-1.07)

Fixed Effects Industry Industry & Proposal

Firm Industry Industry & Proposal

Firm Industry Industry & Proposal

Firm

Observations 2237 2237 2237 2237 2237 2237 2237 2237 2237 Adj. R2 0.103 0.332 0.357 0.099 0.331 0.356 0.098 0.331 0.355

11

Table 8 Compensation Proposals - Influence of Public Opinion on Voting

This table shows estimates of panel regressions of voting support rate on role of public opinion, for shareholder-sponsored compensation proposals, during the period January 2004 to November 2010. The dependent variable is the support rate, percent of votes cast in favor of the proposal. Public opinion is measured using three proxies from the month or year preceding the vote. CONFIDENCE is the percentage confidence in banks from the annual Gallup Poll Survey of confidence in institutions. MEDIA is the natural log of the number of articles related to compensation based on monthly search of major news and business publications on Factiva. MEDIA-NEGATIVE is the monthly percentage of words in the articles related to compensation (MEDIA) that match the negative words proposed by Kuhnen and Niessen (2012) or Loughran and McDonald (2011). The keyword searched included CEO compensation or CEO salary or executive compensation or executive salary. Errors are clustered at the firm level. *, **, *** reflect statistical significance at the 10 percent, 5 percent, and 1 percent levels.

(1) (2) (3) (4) (5) (6) CONFIDENCE -0.618*** -0.417***

(-7.89) (-2.87) MEDIA 0.077*** 0.061*** (5.30) (3.12) MEDIA-NEGATIVE 0.085*** 0.062*** (5.50) (2.94) SIZE -0.023*** 0.118** -0.021*** 0.153*** -0.021*** 0.156*** (-3.58) (2.37) (-3.29) (3.50) (-3.33) (3.58) ROA 0.027 0.590* -0.068 0.454 -0.051 0.471 (0.37) (1.73) (-0.80) (1.43) (-0.60) (1.51) CAPEX 0.097 0.204 0.216 -0.119 0.216 -0.146 (0.53) (0.30) (0.98) (-0.17) (0.99) (-0.21) CASH -0.124 -0.112 -0.050 0.070 -0.058 0.061 (-1.43) (-0.34) (-0.53) (0.21) (-0.62) (0.18) DEBT -0.105* 0.036 -0.089 0.021 -0.088 0.021 (-1.94) (0.15) (-1.58) (0.08) (-1.57) (0.08) EXCESS RET -0.006 0.015 -0.009 0.009 -0.009 0.009 (-0.29) (0.47) (-0.41) (0.27) (-0.41) (0.25)GOV41 0.406*** 0.336 0.560*** 0.485 0.572*** 0.504* (4.65) (1.10) (6.39) (1.62) (6.50) (1.67) Constant 0.402*** -1.014** -0.453*** -2.012*** -0.167 -1.813*** (3.86) (-2.16) (-3.57) (-4.73) (-1.60) (-4.42) Fixed Effects Industry Firm Industry Firm Industry Firm Observations 526 526 526 526 526 526 Adj. R2 0.243 0.432 0.170 0.421 0.171 0.419

12

Table 9 Firm-Level Media Coverage and Voting

This table shows estimates of panel regressions of voting support rate on role of public opinion, for all shareholder-sponsored proposals, during the period January 2004 to November 2010. The dependent variable is the support rate, percent of votes cast in favor of the proposal. Firm-level media coverage, MEDIA-FIRM, is measured as the percentage of the number of articles related to compensation on a specific firm in the month preceding the vote, normalized by the total number of articles related to compensation on this firm over the sample time period. The keyword searched included CEO compensation or CEO salary or executive compensation or executive salary, and the firm’s name. Errors are clustered at the firm level. *, **, *** reflect statistical significance at the 10 percent, 5 percent, and 1 percent levels.

 

(1) (2) (3) (4) (5) (6) MEDIA-FIRM 0.272** 0.225*** 0.287** 0.236*** 0.400*** 0.341*** (2.36) (2.74) (2.46) (2.85) (2.87) (3.18) SIZE -0.028*** -0.018*** -0.030*** -0.018**** -0.031*** -0.022*** (-6.11) (-4.49) (-5.90) (-4.13) (-6.16) (-5.27) ROA -0.155* -0.085 -0.163* -0.097 -0.190** -0.138* (-1.71) (-1.11) (-1.73) (-1.23) (-2.08) (-1.80) CAPEX 0.085 0.166 -0.042 0.088 0.064 0.123 (0.39) (0.86) (-0.22) (0.53) (0.33) (0.73) CASH -0.138 -0.069 -0.160 -0.086 -0.196* -0.128 (-1.24) (-0.71) (-1.40) (-0.86) (-1.91) (-1.40) DEBT -0.044 -0.064 -0.063 -0.081** -0.112*** -0.113*** (-0.92) (-1.64) (-1.34) (-2.10) (-2.61) (-3.02) EXCESS RET -0.003 0.013 -0.002 0.015 -0.009 0.004 (-0.14) (0.83) (-0.12) (0.91) (-0.43) (0.25) GOV41 0.069 0.004 0.153* 0.106 (0.80) (0.06) (1.75) (1.46) ISS 0.271*** 0.219*** (18.68) (14.06) Constant 0.403*** 0.386*** 0.392*** 0.397*** 0.323*** 0.323*** (6.13) (6.62) (5.09) (5.97) (4.12) (4.96)

Fixed Effects Industry, Year Industry, Year

& Proposal Industry, Year

Industry, Year & Proposal

Industry, Year Industry, Year &

Proposal Observations 2209 2209 2137 2137 2137 2137 Adj. R2 0.138 0.343 0.135 0.342 0.414 0.502

13

Table 10

Influence of Public Opinion on Proxy Advisory Recommendation

This table shows estimates of panel regressions of recommendations by ISS and role of public opinion, for all shareholder-sponsored proposals, during the period January 2004 to November 2010. The dependent variable is a dummy variable that equals one if ISS recommends voting for the proposal. Public opinion is measured using three proxies. Public opinion is measured using three proxies from the month or year preceding the vote. CONFIDENCE is the percentage confidence in banks from the annual Gallup Poll Survey of confidence in institutions. MEDIA is the natural log of the number of articles related to compensation based on monthly search of major news and business publications on Factiva. MEDIA-NEGATIVE is the monthly percentage of words in the articles related to compensation (MEDIA) that match the negative words proposed by Kuhnen and Niessen (2012) or Loughran and McDonald (2011). The keyword searched included CEO compensation or CEO salary or executive compensation or executive salary, and salary. Errors are clustered at the firm level. . *, **, *** reflect statistical significance at the 10 percent, 5 percent, and 1 percent levels.

(1) (2) (3) (4) (5) (6) (7) (8) (9) CONFIDENCE -0.974*** -0.674*** -1.137*** (-7.65) (-5.35) (-8.00)MEDIA 0.122*** 0.073** 0.156***

(4.24) (2.47) (5.51)MEDIA-NEGATIVE 0.111** 0.057* 0.143*** (3.71) (1.82) (4.84) SIZE 0.002 0.015 0.061 0.004 0.016* 0.186*** 0.004 0.017* 0.202*** (0.24) (1.59) (0.83) (0.39) (1.77) (2.63) (0.44) (1.82) (2.84) ROA 0.202 0.292 1.243*** 0.132 0.246 0.908** 0.128 0.240 0.919** (1.19) (1.50) (3.23) (0.77) (1.24) (2.41) (0.75) (1.21) (2.39) CAPEX -0.790** -0.747* 0.241 -0.705* -0.681 0.047 -0.674* -0.649 0.164 (-2.25) (-1.83) (0.26) (-1.85) (-1.60) (0.05) (-1.80) (-1.54) (0.16) CASH 0.047 0.111 0.510 0.191 0.214 0.906** 0.188 0.214 0.928** (0.26) (0.59) (1.28) (1.08) (1.15) (2.14) (1.05) (1.15) (2.13) DEBT 0.173** 0.141 0.632** 0.169* 0.139 0.628* 0.173** 0.143* 0.641* (1.97) (1.62) (2.11) (1.93) (1.61) (1.87) (1.98) (1.66) (1.86) EXCESS RET 0.017 0.041 -0.007 0.012 0.039 -0.017 0.015 0.042 -0.013 (0.47) (1.05) (-0.17) (0.33) (1.04) (-0.40) (0.41) (1.11) (-0.29) Constant 0.794*** 0.623*** 0.210 -0.332* -0.095 -2.462*** 0.154 0.211 -1.973*** (5.33) (3.61) (0.28) (-1.81) (-0.46) (-3.42) (1.22) (1.36) (-2.74)

Fixed Effects Industry Industry & Proposal

Firm Industry Industry & Proposal

Firm Industry Industry & Proposal

Firm

Observations 2369 2369 2369 2369 2369 2369 2369 2369 2369 Adj. R2 0.067 0.173 0.218 0.041 0.160 0.198 0.036 0.157 0.194

14

Table 11 Voting by Mutual Funds

The total number of mutual fund families and funds are shown in the table for each year 2004 to 2010, the data for 2010 ends in November and does not include the full year. For shareholder-sponsored and management-sponsored proposals, the number of proposals voted, percentage voted with management, and percentage voted with the proxy advisory firm ISS are also shown.

Shareholder Proposals Management Proposals

Year # of Fund Families

# of Funds Voting

# of Votes Cast Voting with

MGMT Voting with Proxy

Advisor # of Votes Cast

Voting with MGMT

Voting with Proxy Advisor

2004 110 3,585 255,676 73.6% 78.6% 329,918 80.2% 84.8% 2005 291 5,335 194,841 69.1% 72.6% 276,230 81.3% 85.5% 2006 259 5,989 258,995 60.1% 69.7% 304,785 83.2% 87.2% 2007 265 4,950 286,434 59.8% 68.3% 315,099 85.7% 88.8% 2008 282 5,445 287,381 61.0% 64.0% 377,078 85.7% 88.8% 2009 275 5,120 310,942 52.0% 61.9% 489,684 82.1% 85.9% 2010 219 3,624 250,367 53.9% 59.2% 346,500 86.3% 88.2%

15

Table 12 Voting by Mutual Funds and Public Opinion

The table shows the relation between voting by mutual funds and public opinion. The sample includes types of shareholders-sponsored proposals between 2004 and 2010. The dependent variable is a dummy equal to one if the mutual fund voted FOR the proposal. Public opinion is measured using three proxies from the month or year preceding the vote. CONFIDENCE is the percentage confidence in banks from the annual Gallup Poll Survey of confidence in institutions. MEDIA is the natural log of the number of articles related to compensation based on monthly search of major news and business publications on Factiva. MEDIA-NEGATIVE is the monthly percentage of words in the articles related to compensation (MEDIA) that match the negative words proposed by Kuhnen and Niessen (2012) or Loughran and McDonald (2011). Firm-level media coverage, MEDIA-FIRM, is measured as the percentage of the number of articles related to compensation on a specific firm in the month preceding the vote, normalized by the total number of articles related to compensation on this firm over the sample time period. Mutual fund characteristics include FUNDSIZE, measured as log of total assets under management, EXPRATIO is expense ratio, TURNOVER is fund turnover, INSTITUTION is a dummy if the mutual fund has institutional business and INDEX is a dummy equal to one if the fund is an index fund. In addition, firm-level controls are included but not shown. Standard errors are corrected for clustering of observations at the fund level. *, **, *** reflect statistical significance at the 10 percent, 5 percent, and 1 percent levels.

(1) (2) (3)

CONFIDENCE -0.228*** (-6.38) MEDIA 0.037*** (8.73) MEDIA-NEGATIVE 0.035*** (6.92) FUNDSIZE -0.004 -0.003 -0.004 (-0.77) (-0.63) (-0.66)EXPRATIO -1.288 -1.727* -1.730 (-1.32) (-1.75) (-1.75)TURNOVER 0.004 0.008 0.008 (0.70) (1.26) (1.28) INSTITUTION 0.013* 0.012 0.012 (1.87) (1.64) (1.65) INDEX -0.057*** -0.046*** -0.044*** (-4.83) (-4.24) (-4.02)Constant 0. 903*** 0.568*** 0.725*** (24.18) (10.74) (16.56)

Fixed Effects Industry, Fund & Proposal

Industry, Fund & Proposal

Industry, Fund & Proposal

Observations 413133 413133 413133

Adj. R2 0.368 0.368 0.368