Boards’ Response to Shareholders’ Dissatisfaction: The Case of Shareholders’ Say on Pay in the UK

Abstract

In 2002, the UK adopted a regulation allowing shareholders to cast non-binding (advisory) votes on their firm's Directors' Remuneration Report during annual general meetings (the ‘Say-on-Pay’ rule). This study evaluates a decade of this regulation and examines how it affected the behavior of shareholders and boards in a sample of Financial Times Stock Exchange 350 firms during the period 2002–2012. I find evidence that shareholder dissatisfaction increases with excess Chief Executive Officer (CEO) compensation. This relationship does not exist for the expected level of compensation, suggesting that shareholders take a sophisticated approach when casting their vote. Boards do not appear to respond to shareholder dissatisfaction systematically; however, they do respond selectively by reducing the excessiveness of CEO compensation when performance is poor. Boards also seem to respond swiftly to shareholder dissatisfaction. There is evidence that the probability of CEO turnover increases with shareholder dissatisfaction. Overall, the evidence suggests that ‘Say-on-Pay’ regulation addressed regulatory concerns about transparency, accountability, and performance linkage.

Acknowledgements

I thank Fabrizio Ferri (guest editor) and an anonymous reviewer for their constructive feedback and guidance throughout the review process. This paper is part of my dissertation at the Pennsylvania State University. I am grateful to members of my dissertation committee for their advice and guidance: Paul Fischer, Karl Muller (chair), Amy Sun, and Arun Upneja. I also thank Sam Bonsall, Vedran Capkun, Christine Cheng, Scott Collins, Kevin Koharki, Jim McKeown, Delphine Samuels, Kristy Schenck, and seminar participants at HEC Paris, INSEAD, and the Pennsylvania State University for their insightful comments and suggestions. Walid Alissa is a member of GREGHEC, Unit CNRS UMR 2959.

Funding

The author gratefully acknowledges financial support from the Smeal College of Business at the Pennsylvania State University and HEC Paris.

ORCID

Walid Alissa http://orcid.org/0000-0002-6436-2372

Notes

¹In the UK, ‘director’ refers to both executive and nonexecutive board members.

²House of Commons, Delegated Legislation Committee Debates, Session 2001–2002, ‘Draft Directors’ Remuneration Report Regulations 2002'.

³This finding is consistent with observations by Bryan-Low (2012). The author argues that

shareholders have played a role in the departure of several high-profile corporate chiefs in the UK, including Andrew Moss, the CEO of insurer Aviva PLC; David Brennan, chief executive of AstraZeneca PLC, and Sly Bailey, the CEO of Trinity Mirror PLC

when talking about voting on pay packages.

⁴The Combined Code sets out standards of good governance practice in a non-binding ‘comply or explain’ fashion.

⁵ Prior activism research also suggests that activists can (1) participate in proxy contests for control of the firm; (2) participate in or lead shareholders' suits and class actions; (3) steer the filing of shareholder-sponsored governance proposals; (4) vote against the re-election of certain board members; (5) participate in or lead ‘just vote no’ campaigns; and (6) use the media to embarrass board members (see Gillan & Starks, 1998, 2007, for references).

⁶Bebchuk and Fried (2004) provide that ‘collegiality, team spirit, a natural desire to avoid conflict within the board, friendship and loyalty, and cognitive dissonance’ are reasons why boards are biased against engaging in arm's length negotiations when it comes to CEO pay.

⁷Core, Guay, and Thomas (2005) agree that CEOs' contracts reflect their power and that the higher the CEOs' power, the higher the pay; however, they disagree that this necessarily reflects suboptimality for shareholders. The purpose of this study is not to test the optimality of UK pay practices, but to test shareholders' perception of pay practices.

⁸‘Just vote no’ campaigns are organized communication attempts by activists to shareholders, encouraging them to withhold their vote from one or more directors at the AGM. In doing so, shareholders can express their dissatisfaction and put pressure on the board to respond to their demands.

⁹Larcker, Ormazabal, and Taylor (2011) find no significant market reaction to the announcements related to executive pay events, but find a negative cross-sectional relationship between abnormal returns on event days and CEO compensation.

¹⁰Alternatively, if shareholders believe that board members are not going to respond, they can vote down board members up for re-election at the same time as they are expressing their dissatisfaction with boards' pay decisions. Lo et al. (2014) find evidence consistent with this argument. They find that board turnover is positively associated with contemporaneous dissatisfaction.

¹¹The FTSE 350 index is a market capitalization weighted stock market index incorporating the largest 350 companies with their primary listing on the LSE. The FTSE 350 index represents about 96.67% of the UK's market capitalization. Because the FTSE index is a weighted stock index, firms are added to the index or dropped from the index on a regular basis. I only follow the firms that were listed in the index when I first started hand-collecting the data.

¹²Note that I do not require firms to exist for the entire period of the sample and I do not continuously update the sample according to the FTSE 350 index. I keep the same firms throughout the sample unless they violate one of the conditions listed above.

¹³See Conyon and Murphy (2000) for a discussion on the potential drawbacks from using this formula.

¹⁴There has been some debate over how to account for . UK law does not consider an ‘abstain’ vote as a vote when it comes to the passage of a resolution. That being said, the approval (or lack of approval) of the remuneration report is non-binding and a shareholder decision to elect (vote) to abstain may carry a signal to the board. Note that an ‘abstain’ vote is not the same as not voting. When the shareholder does not select any of the available choices (i.e. do not cast a vote), the shareholder is giving the chairman of the board, when the chairman is selected as a proxy, the discretion to vote their shares. Across all the firms in the sample a ‘discretion’ vote results in a vote for a resolution (never against it) when the chairman is selected as a proxy. Given this discussion, and as a robustness check, I construct a measure where I take the ABSTAIN out of the total vote cast and another measure where I add ABSTAIN to the AGAINST in the numerator. Both measures yield qualitatively the same inferences.

¹⁵In untabulated analysis, I split the sample based on whether or not the firm is part of the HD subsample and whether the firm belongs to FTSE100 or FTSE250. CEOs in firms in the HD subsample have lower tenure (a mean five years vs. seven years and a median four years vs. five years). The HD firms have lower mean and median returns and returns on assets. FTSE100 CEOs are older, better paid, have less tenure, and smaller holdings relative to their boards' total holdings. FTSE100 firms have more growth opportunities, lower stock returns, and less institutional ownership.

¹⁶One potentially omitted variable in the model above is proxy advisors' recommendations. This is less of a concern in this setting since the majority (about 78%) of institutional investors use proxy advisors to supply research and not to obtain voting recommendations, according to a 2012 survey of adherence to the Stewardship Code, Investment Management Association.

¹⁷A potential concern is that can be endogenous in the regression if boards perceive shareholder dissatisfaction to be costly. Boards might proactively reduce excess compensation in year in order to avoid upsetting shareholders in year and incurring the costs associated with shareholder dissatisfaction. To the extent that this happens, variation in CEOs' excess compensation across firms with greater shareholder dissatisfaction will be smaller, because shareholders would vote less against the remuneration report than if boards ignored the shareholder dissatisfaction when they design the compensation package. I expect this reduction in excess compensation and the resulting reduction in votes against the remuneration report to make it harder to detect significance in .

¹⁸A potential concern with interpreting the insignificant coefficient on in Column 3 of Table 2 is that many of the independent variables that were used to determine expected compensation in Model (1) are used to explain the level of shareholder dissatisfaction. It is possible that is significantly explained by the control variables in Model (1), and that it becomes insignificant when it is added to Model (2). To alleviate this concern, untabulated results indicate that inferences are unaffected when the control variables are dropped. I also run model (3) with only with and without the control variables, and the lack of significance is still there.

¹⁹In another test, I follow Ferri and Maber (2013) and create two industry-adjusted ROA and two industry-adjusted RET variables by splitting them into positive and negative variables. The positive (negative) variables are the values of the original variables if they are above (below) industry-adjusted RET and ROA, and zero otherwise. I interact them with excess and rerun Model (3). Neither negative industry-adjusted ROA nor negative industry-adjusted RET are significant.

²⁰For completeness, I follow prior research (see Core et al., 2008) and consider the change in excess compensation from year to year and from to year . The lack of significance continues.

²¹An interesting question to test is whether boards respond to continuous dissatisfaction. I argue that the pressure on the board is expected to be at its highest when shareholder dissatisfaction is persistent. I develop a persistence measure, , which is equal to 1 if firm belongs to the HD subsample for two consecutive years. I rerun Model (4) with interacted with . I expect the coefficient on the interaction term to be negative and significant. Untabulated results do not support this expectation.

²²Although the Combined Code states that the CEO should not move to the chairmanship of the board of the same company, a CEO can become a chairman if the board consults major shareholders in advance and provides reasoning to shareholders at the time of the appointment and in the next annual report (see Provisions A.2.1 and A.2.2 of the Combined Code for more information).

²³If it was disclosed that the CEO was retiring but he/she accepts a CEO role in another firm within six months after his or her departure, I interpret this as an indication that the CEO was forced out of office.

²⁴When Model (4) is estimated without , the results are unchanged. In addition, when Model (4) is estimated without DISSATISFACTION, the negative sign and significance on is still there, consistent with prior research.