The extensive academic literature on the growth of executive compensation has tended to polarise around one of two positions: the rents-capture view and the optimal contracting approach. These analyses lead to very different positions on the value of a ‘say on pay’ policy:
The rents-capture position, most strongly articulated by Bebchuk and Fried (2003, 2004), holds that CEOs can subvert the compensation-setting process through their ‘capture’ of the remuneration committee. In consequence, the secular rise in CEO rewards is treated as a reflection of a flawed governance process. On this view, the introduction of direct democracy in the form of a stockholder vote should help to rectify problems caused by the timidity of the board. In the Bebchuk–Fried formulation, executive pay is only constrained by the ‘outrage’ of stockholders and media. It follows that allowing an easy expression of this sentiment should limit excessive rewards – by withholding assent to egregious proposals and/or by constraining these proposals in anticipation of the vote.
By contrast, the optimal contracting view suggests that CEO compensation is determined by the operation of the market for managerial talent. The prevalence of external hires – at least in countries such as the US and UK – is taken as an indication that this market is relatively frictionless. Hence, any secular growth in executive rewards is suggestive of a long-term supply shortfall, perhaps in managers with general – i.e. non-firm-specific – human capital (Murphy and Zabojnik 2004). A recent, more sophisticated, variant of the argument by Gabaix and Landier (2008) models the executive labour market as a process of matching talent to productive opportunity, proxied by firm size. Their model gives rise to the prediction that among firms in the upper tail of the size distribution, rewards will track average firm value. This hypothesis appears reasonably well-supported either side of the financial crisis (see Gabaix et al. 2014).
The optimal contracting view sees say on pay as at best meddlesome and at worst disruptive. Requiring the shareholders’ retrospective endorsement of any pay offer introduces uncertainty into the executive appointment process. If this causes executives to see a risk premium – which Peters and Wagner (2012), for example, find has been a consequence of the rising probability of executive dismissal – the impact on compensation levels could even be perverse.
UK background and sample
The UK became the first country to adopt say on pay in 2003, and hence it provides a good test for the effectiveness of the policy. Before 2003, a minority of British companies had held shareholder votes, but on a voluntary basis. Since 2003, boards of quoted companies have been required to put a resolution to accept the report of the company’s remuneration committee to a vote of shareholders at that firm’s AGM. Until now the vote has had advisory status only; although the Combined Code on corporate governance has required that managements who go against a pay resolution explain their decision to the company’s stockholders.
Early evaluations of the policy (Conyon and Sadler 2010, Ferri and Maber 2013) reported that despite high levels of shareholder participation, no more than one or two pay resolutions per year were rejected, and the annual average vote against the board rarely exceeded 6%. However, the financial crisis of 2008 – with its attendant stock market fall and subsequent squeeze on real incomes – was widely believed to have altered shareholder and – perhaps as importantly – media sentiments. By 2011 both the popular and financial press were claiming a ‘shareholders’ spring’ reaction to executive rewards packages deemed to be excessive.
Our dataset comprises all non-financial companies that figured in the FT350 in any year between 31 December 1998 and 31 March 2012. Voting and executive characteristics data were supplied by Manifest Information Services Ltd, and financial data were derived from Thomson Datastream. A total of 4,090 pay resolutions were analysed.
Our raw data – summarised in Table 1 – confirm that shareholder voting has overwhelmingly supported the status quo. Very few report resolutions are rejected, and most secure very large majorities. There has been some reduction in ‘yes’ voting since 2008, but this has merely brought dissent back to the level experienced immediately after the start of say on pay. Moreover, this result is not obviously a product of shareholder apathy – ‘voter’ turnout has climbed to over 70%, a level comparable to that of national elections.
Table 1. Renumeration report resolutions
Notes: Proposals of the remuneration report became mandatory for UK incorporated and London Stock Exchange listed companies with the financial year ending on or after 31 December 2002.
* 2012 figures are up to and including 31 March 2012.
Source: Gregory-Smith et al. (2014: 29).
The paper next analyses the influence of CEO pay, company performance, and corporate governance characteristics on the level of dissent. In recognition of the high average degree of support for the board, we define dissent as both votes cast against the remuneration resolution and abstentions. We find that across the period as a whole – and having controlled for company performance and corporate governance factors – there is a positive correlation between dissent and CEO pay. Its magnitude is small however, equivalent to a 0.2% increase in dissent for a 10% increase in compensation – a result which suggests CEOs could extract much larger rewards still before they risk the remuneration resolution being overturned. Interestingly, we find no evidence of any structural break in the dissent–reward relationship in 2008 or 2009 – shareholders do not appear more anxious about pay since the crisis. Similarly, we find no significant effect played by the proportions of insider and non-executive directors. There is, however, some evidence that voting reflects shareholder returns, with the better rewarded shareholders more likely to endorse the pay proposal.
The paper also examines the influences of the above factors across quantiles of dissent. We find both the pay (positive) and shareholder return (negative) effects strengthen notably across the higher quintiles of dissent. This suggests that the very relaxed shareholder attitude to pay, implied by the coefficients at the mean, may substantially under-estimate hostility at higher quintiles.
Does dissent moderate compensation?
Finally, we explore the impact of last year’s dissent on this year’s suggested CEO rewards. We find, as expected, that lagged dissent does have a moderating effect across the sample as a whole, but again its overall magnitude is small. However, we find strong evidence that the effect is not uniform over the sample. Fitting a spline function suggests that levels of past dissent up to 10% had no moderating effect on current rewards and perhaps even encouraged higher pay. By contrast, lagged dissent above 10% exercised a clear restraining effect. This increased across the remuneration quantiles. That is, remuneration committees representing the more highly rewarded CEOs were quite sensitive to dissent, provided it had reached above a critical threshold of about 10%.
We could find no evidence of structural breaks in the pay-lagged dissent relation. Taken together with the absence of breaks in the dissent–compensation relationship, this suggests the anecdotal evidence for a ‘shareholder spring’ revolt in the aftermath of the crisis is no more than anecdotal.
Conclusion: Has the policy worked?
Say on pay does appear to have moderated CEO compensation awards, but the overall effect appears quite small. Indeed the effect only appears at levels of dissent which are historically high and is then substantial only across the higher pay quintiles. Since our study covers all the larger UK firms – firms where the growth in compensation has been most pronounced – it does suggest a limited role for say on pay.
What caveats should we add to our findings? First, the study covers a period when the pay resolution was non-binding and the institutional shareholders were largely unconstrained in their behaviour. Recent amendments to UK company law (HM Government 2013) make the voting outcome mandatory upon boards, and the Financial Reporting Council (2010a, 2010b) requirement for greater involvement by institutions on executive pay determination may well affect future pay proposals. Second, voting brings greater public scrutiny to executive compensation arrangements, which quite possibly constrains egregious behaviour. Finally, there is some evidence that executive turnover increases after displays of serious shareholder dissent. This possibility, which awaits empirical confirmation, would suggest a further mechanism by which voting might constrain excessive rewards.
Bebchuk, Lucian Arye and Jesse Fried (2003), “Executive compensation as an agency problem”, Journal of Economic Perspectives, 17(3): 71–92.
Bebchuk, Lucian Arye and Jesse Fried (2004), Pay without Performance: The Unfulfilled Promise of Executive Compensation, Harvard University Press.
Conyon, Martin and Graham Sadler (2010), “Shareholder voting and directors’ remuneration report legislation: Say on pay in the United Kingdom”, Corporate Governance: an International Review, 18: 296–312.
Ferri, F and D A Maber (2013), “Say on pay votes and CEO compensation: Evidence from the UK”, Review of Finance, 17(2): 527–563.
Financial Reporting Council (2010a), “Implementation of the UK Stewardship Code”, London.
Financial Reporting Council (2010b), “The UK Stewardship Code”, London.
Gabaix, X, A Landier, and J Sauvagnat (2014), “CEO Pay and Firm Size: An Update After the Crisis”, Economic Journal, 124: F40–F59.
Gregory-Smith, I, S Thompson, and P W Wright (2014), “CEO Pay and Voting Dissent Before and After the Crisis”, Economic Journal, 124: F22–F39.
HM Government (2013), “Large and Medium-sized Companies and Groups (Accounts and Reports) (Amendment) Regulations 2013”, Statutory Instruments 2013 No. 1981.
Murphy, K J and J Zabojnik (2004), “CEO pay and appointments: A market-based explanation for recent trends”, The American Economic Review, 94(2): 192–196.
Peters, Florian S and Alexander F Wagner (2012), “The Executive Turnover Risk Premium”, Tinbergen Institute Discussion Paper 12-021/2/DSF30.