Executive compensation: investor preferences during say-on-pay votes and the role of proxy voting advisers

Abstract

Shareholder say-on-pay voting allows institutional investors to influence the incentives of managers and, consequently, corporate behaviour. Surprisingly, the preferences of investors on executive compensation have been largely overlooked in the ongoing debates on the role of say-on-pay in corporate governance and the impact of shareholder stewardship on sustainable corporate behaviour. The analysis of investor disclosed explanations of say-on-pay votes in the FTSE 100 companies during 2013–2021 shows that institutional investors rely repeatedly on several dominant themes aimed at improving the incentives of corporate managers and controlling managerial rent extraction. But shareholder interests remain the core focus of say-on-pay votes, with only few investors demanding that companies reward executive directors for protecting the interests of a broader range of affected stakeholders. Additionally, most investors can be grouped into several clusters formed around the voting recommendations of proxy advisers. A group of UK-based institutional investors stands out by taking a more individualistic and diverse approach to the stewardship of executive compensation. These findings highlight the role of local investors in the oversight of executive pay, the growing influence of proxy advisers along with the increasing share of foreign institutional investors, and the influence of best practice governance codes in driving investor stewardship preferences.

Introduction

In May 2021, more than 60% of votes at the annual shareholders’ meeting of Rio Tinto Group, one of the world's largest metals and mining corporations, went against the company remuneration report.¹ Shareholders were furious that the total annual remuneration of the company's former CEO, who had stepped down earlier after Rio Tinto's blast of ancient Juukan rock shelters in Australia to clear the way for a mining project, increased despite the destruction of the sacred Aboriginal site and the ensuing reputational damage.² Only a month later, 70% of shareholders voting at the annual meeting of Wm Morrisons voted against the remuneration report of one of the UK's largest supermarket chains.³ The reason this time was the failure of the company's board to adjust the remuneration of top managers to account for the windfall profits during the Covid-19 pandemic when the hospitality sector was closed.⁴ More recently, the UN-convened Net Zero Asset Owner Alliance, which includes some of the world's largest pension funds and insurance companies, has pledged to cut up to 65% of emissions linked to their investments by 2030, including by using shareholder voting rights.⁵

Say-on-pay votes, as these examples illustrate, give shareholders of a company the right to vote on the remuneration of executive directors, often also referred to as executive compensation or executive pay, thereby allowing shareholders to influence and exercise oversight over executive compensation contracts.⁶ The design of these contracts takes a central place in corporate governance because it has long been considered an effective way to align the incentives of corporate managers with the interests of shareholders by linking executive compensation with company performance.⁷ Compensation contracts then are one of the principal reward strategies for addressing the problem of managerial misbehaviour.⁸ The proponents of this idea, known as the shareholder value theory (or the efficient contracting theory), consider executive pay as a tool to mitigate the agency problems between managers and shareholders.⁹

This pay-based reward strategy relies on two fundamental ideas. One is the power of financial incentives summarised famously by Charles Munger, vice chairman of Berkshire Hathaway: ‘Show me the incentive and I will show you the outcome.’¹⁰ Another is the dominant narrative that the company's interest can be reduced to the interests of its shareholders.¹¹ Corporate managers, accordingly, must act towards increasing shareholder value.¹²

Both these ideas have come under serious criticism from different directions. While the power of incentives remains unchallenged, the influential rent extraction theory (also known as the managerial power theory) of executive compensation casts doubts on the ability of conflicted corporate boards to design optimal pay-based incentive structures.¹³ This theory explains high-profile scandals and the increasing level of pay, which made executive compensation a controversial topic in many countries, especially after the 2008 financial crisis, by the failure of captured and conflicted corporate boards to constrain the interests of powerful CEOs.¹⁴ Governments have gradually strengthened the disclosure of pay practices¹⁵ and shareholder voice on pay by introducing ‘say-on-pay’ votes. The premise is that the transparency of executive compensation and more active (and better) oversight by shareholders can constrain excessive pay and improve its structure.¹⁶

Similarly, the second idea about the primacy of shareholder interests has encountered major objections since the outset of the twenty-first century and especially during recent years.¹⁷ Changing societal preferences regarding the purpose of corporations and increasing attention on their societal impact have posed difficult questions for the optimal design of pay-based incentive structures and for the established practices of executive compensation.¹⁸ In line with these changing tides, experts have focused attention on the possibility of using incentive-based reward design to address the interests of a broader group of stakeholders instead of single-mindedly targeting shareholder value creation.¹⁹ Incentives of managers for considering environmental and social interests can be strengthened by incorporating various non-financial targets – like metrics related to reductions in corporate CO₂ emissions, investments in green energy sources, workforce diversity, health and safety, and employee engagement – in the design of executive compensation contracts. The challenge, of course, is that different actors exert influence over the executive pay design. On one side are corporate board remuneration committees and their remuneration consultants; on the other side are shareholders who vote on management-sponsored remuneration proposals and proxy advisers consulting shareholders regarding their votes. The introduction of say-on-pay votes has reinforced the position of shareholders – especially of large asset owners, like pension funds, and asset managers responsible for managing collective investments accumulated by investment funds – in this power dynamic.²⁰

The questions of how shareholders use compensation-related voting rights and what objectives they pursue when doing so lie at the very heart of both criticisms. Shareholders may rely, for instance, on say-on-pay rights to rein in managerial power and control potential rent extraction by entrenched powerful executives or they may aim for a better alignment of managerial interests with shareholder value. But if this is all that matters for shareholders, then say-on-pay votes constrain the ability of corporate boards to design incentive rewards aligned with more inclusive and responsible business practices, which runs counter to the changing preferences in society.²¹ Alternatively, say-on-pay votes can be a powerful tool in the hands of shareholders in their ongoing efforts – if we are to believe the sustainability rhetoric of many institutional investors – to promote more responsible business practices regarding major global challenges, like climate change and social and economic inequalities.²² Investors can use their influence during voting on remuneration proposals to demand the alignment of the incentives of corporate managers with more responsible corporate behaviour. Say-on-pay votes then, instead of being an obstacle, can be a catalyst for changing corporate behaviour.

This study seeks to understand how and why shareholders use their compensation-related voting rights. It does so by analysing the voting records and vote explanations of institutional investors on all say-on-pay proposals put for a vote by the FTSE 100 companies during 2013–2021 to identify investor preferences during say-on-pay votes, changes in these preferences over time, and the key influencers of say-on-pay votes. This is the first study to use voting explanations of institutional investors on a large scale to offer systematic evidence on the determinants of investor decisions during say-on-pay votes. This information is crucial for understanding how shareholder stewardship of executive compensation is steering corporate behaviour. Shareholder stewardship is a complex phenomenon with multiple conceptions determined by the identity of stewards and the targets of stewardship.²³ Given that institutional investors control and vote on most of the shares in UK listed companies, this study defines stewardship as voting engagement by institutional investors on corporate governance and environmental, social, and governance (ESG) matters. Accordingly, the goal of this paper is to develop a better understanding of the role and limits of institutional investors in dealing with managerial opportunism and promoting more ESG-conscious corporate behaviour through the stewardship of executive compensation.

The findings show how shareholders influence pay design by communicating their preferences to corporate boards (and to compensation consultants). Institutional investors rely repeatedly on several dominant themes of engagement on compensation, including the quantum and structure of pay, the link between pay and performance, and better communication and disclosure. Investor attention to these topics suggests that investor stewardship of pay is driven by both the need to design optimal pay structures and the urgency of addressing the agency problems of pay through the oversight of pay levels and demanding better disclosure. But say-on-pay is largely an unused tool in the quest to steer companies towards a more sustainable path: investors do not at present demand that executive pay be linked with environmental or social targets on a broader scale. Nevertheless, say-on-pay is also not an obstacle for promoting more responsible corporate behaviour because investors do not oppose remuneration contracts that include clearly defined and material ESG targets.

Additionally, the findings show that say-on-pay preferences of institutional investors are not homogenous: preferences differ depending on the particular investor. Nevertheless, institutional investors can be grouped into several clusters based on their preferences. Investors within a cluster vote similarly to each other, rely on shared reasons to explain their votes, and give priority to similar aspects of executive compensation. The major clusters are formed around the voting recommendations (stewardship and voting guidelines) of proxy advisory firms, specialised service providers that offer their clients recommendations on how to vote at shareholders’ meetings. Another large cluster of primarily UK-based institutional investors takes a more firm-specific approach by engaging with companies individually and relying more evenly on a broader spectrum of engagement topics.

The study has important theoretical and policy implications. The findings equip us with a better sense of how say-on-pay votes influence executive compensation, as well as which mechanisms and major actors are the drivers of this influence. But they also reveal knowledge gaps that require future research, particularly in relation to the effectiveness and impact of pay-linked ESG targets as a governance tool. Additionally, the findings lead to several recommendations that can promote diversity of investor perspectives during shareholder voting on executive pay, strengthen the accountability of influential proxy advisory firms, increase the potential of their recommendations for improving corporate performance, and advance the practices of integrating various ESG targets into the design of executive pay.

This paper proceeds by providing a brief overview of shareholder say-on-pay votes in the UK (Section 1), explaining the methodology and data sources (Section 2), and presenting the key findings (Section 3). Section 4 discusses the results and highlights their implications, including future legal reform proposals.

1. Shareholder say-on-pay votes in the UK

Shareholder voting on pay has important functions in corporate governance. It allows shareholders to enhance the optimality of executive compensation contracts by deciding whether an adopted pay package is appropriate in a particular context.²⁴ Additionally, Professor Bebchuk and his co-authors argue that shareholder outrage (voting) on pay arrangements constrains managerial rent extraction.²⁵ The impact of shareholders in influencing the design of executive compensation plans and constraining the power of corporate boards to make pay awards has been strengthened after a number of countries introduced say-on-pay votes in response to a widespread perception that executive pay has become increasingly disconnected from the long-term performance of companies.²⁶

The UK became the first country to introduce shareholder voting on director compensation, adopting say-on-pay in 2002.²⁷ The current legal framework has been in force, with minor modifications, since 2013. Shareholders of any UK publicly traded company are entitled: (1) to an annual advisory say-on-pay vote on the directors’ remuneration report (backward looking implementation report);²⁸ and to (2) a binding vote on remuneration policy at least every three years (forward looking remuneration strategy document).²⁹ If a company fails to pass the annual advisory vote on remuneration in a year during which shareholders do not vote on the binding remuneration policy, the company must put the remuneration policy to a shareholders’ vote the following year.³⁰

Theoretical models predict that optimal executive compensation incentive structures are highly firm- and manager-specific. Factors like company size, risk, and CEO wealth and intrinsic motivation are expected, at least in theory, to have an important impact on compensation structures.³¹ For example, CEO effort has a larger monetary effect in a bigger company, and so a smaller incentive linked to the company value is required to motivate the CEO.³² Similarly, a highly motivated CEO who is willing to leave a legacy needs less incentives to work hard compared to a wealthier CEO whose outside option is consuming their existing wealth.³³ But industry best practices have gradually led, for various reasons, to a high-level of standardisation of executive pay structures in practice.³⁴ A typical executive compensation package includes five components: (1) a fixed base salary; (2) an annual bonus or short-term incentive plan; (3) payments from long-term incentive plans (LTIPs), including restricted option and share grants; (4) defined-benefit pension plans; and (5) benefits in the form of company-sponsored goods and services like health benefits, club membership, housing, and transportation (commonly referred to as perks or perquisites).³⁵ Many companies also reward executive managers through discretionary bonuses based on qualitative evaluations of executive performance and one-off awards upon appointment and departure.³⁶

The responsibility for setting the pay of executive directors and senior managers in the FTSE 100 and other companies with a premium listing of shares in the UK lies with the remuneration committee of the board of directors.³⁷ The remuneration committee must consist fully of independent non-executive directors.³⁸ In practice, remuneration committees work closely with executive compensation consultants in the process of designing executive compensation.³⁹ Accordingly, the design and practices of pay in publicly traded companies are strongly influenced by pay consultants.⁴⁰

Say-on-pay votes, which give shareholders a binding or advisory vote on executive pay packages, partly shifted power from directors (and their pay consultants) to shareholders. At minimum, these votes have opened lines of communication between management and shareholders regarding the concerns of shareholders on executive compensation.⁴¹ This has created a new power dynamic by adding an additional important constraint that boards need to consider when designing and setting executive pay. Indeed, a recent survey of the directors of UK publicly traded companies confirms that the need to obtain shareholder support is the strongest constraint on CEO pay design.⁴²

Existing evidence thus suggests that shareholder say-on-pay votes do matter. But the exact impact of say-on-pay votes depends on shareholder attention to voting and, more importantly, the subject matter of shareholder demands during voting and the forces that shape these demands. The evidence is not very clear on the level of attention that shareholders pay to say-on-pay votes and the extent to which their voting decisions are informed.⁴³ For example, small shareholdings of diversified institutional investors weaken their incentives to study each pay proposal individually beyond few large companies. As a result, shareholders may tend to focus more on the actual level of pay or the maximum pay opportunity, rather than on other, arguably even more important, aspects of compensation design.⁴⁴ In consequence, Professor Edmans and his co-authors are concerned that boards may make inefficient changes to executive pay in response to ill-informed or inattentive demands of shareholders to secure their support during say-on-pay votes.⁴⁵

We know even less about the topics of shareholder demands during say-on-pay votes. This information, meanwhile, is critical for understanding the impact of say-on-pay voting on managerial and general corporate behaviour. Shareholders can use say-on-pay votes as a tool for achieving different goals, such as restraining managerial rent extraction though high levels of payments, creating stronger incentives for managers to put shareholder interests first, or stressing the need for a more inclusive and responsible corporate behaviour that considers the impact of business on a broader range of affected stakeholders. Furthermore, information on forces that influence shareholder preferences during say-on-pay votes is required to come up with effective solutions for changing the flaws of established practices. The rest of this paper relies on the explanations of shareholders' votes in the FTSE 100 companies to offer evidence on the preferences of institutional investors during say-on-pay votes and the forces influencing these preferences.

2. Methodology, data sources, and sample overview

(a) Methodology and data sources

The sample consists of all advisory remuneration reports and binding remuneration policies (collectively, say-on-pay proposals) of the FTSE 100 companies voted on during 2013–2021. Accordingly, the study covers only shareholder votes related to the compensation of top executive managers and directors; shareholder votes on the remuneration of employees, such as the approval of employee share ownership plans and other similar schemes, are omitted. The sample includes 913 remuneration reports and 358 remuneration policies. The total number of say-on-pay proposals in the sample is thus 1,271.

Companies are added to the sample if at the date of voting on a say-on-pay proposal a company was part of the FTSE 100 index. The historical composition of the FTSE 100 index on specific dates during 2013–2021 comes from FTSE Russell, the provider of the index, and from the Financial Times Markets Data section.⁴⁶

As explained next, the analysis proceeds in two steps.

(i) Identifying institutional investor clusters using investor voting records

At the first stage of the analysis, institutional investors are grouped into clusters based on the similarity of their votes on say-on-pay proposals. Insightia's One Voting platform (formerly Proxy Insight) provides access to the voting records of institutional investors. The data set includes the votes of different types of investment funds (aggregated at the fund family level), as well as pension funds. All votes in favour, abstain (withhold), and against say-on-pay proposals were coded as 0, 1, and 2, respectively. All other outcomes, such as shareholder decision not to vote, unknown votes, and split votes (where shares controlled by an investor are voted differently), were dropped from the sample. In total, One Voting platform includes 239,108 votes on say-on-pay proposals of the FTSE 100 companies cast by 1,455 institutional investors during the sample period. After cleaning the data manually by merging the voting records of same investors recorded under different names (for example, Insightia's original data set includes two separate records for BlackRock and BlackRock, Inc), the sample includes 238,203 votes by 1,359 unique institutional investors.

Many institutional investors cast their votes independently and thus have unique preferences when voting on say-on-pay proposals. The definition of those preferences at a more general level across many investors requires transforming the large set of variables into a smaller one that makes data aggregation and comparisons feasible. The study uses principal components analysis (PCA) of the voting records in the data set to identify the principal components of the data – the clusters of institutional investors that vote similarly. For the purposes of this analysis, different institutional investors were the variables; votes cast by these investors on say-on-pay proposals were the observations. In other words, the complete data set includes 1,359 variables (investors) and 238,203 observations (votes) split among 1,271 proposals.

One feature of PCA software is that variables with missing observations are automatically dropped from the analysis. To illustrate this in the context of this study, if one investor did not vote on one of the say-on-pay proposals, the analysis software would drop this proposal from the sample. The presence of many relatively small investors in the sample with votes on a limited number of say-on-pay proposals means that running the analysis on the full data set would yield no results. This challenge can be overcome by using an algorithm that can impute the missing voting records based on the mean of each attribute (proposal) or the predicted votes of investors.⁴⁷ This, of course, comes at the cost of the accuracy of the inserted observations because it is not clear how an investor would have voted in real life. An alternative strategy is to reduce the analysis to only those investors that have a large number of recorded votes. Although this approach leads to a smaller sample size, the results reflect accurately the actual votes of investors in the sample. The analysis at this stage thus included only those investors that voted on at least 80% of all say-on-pay proposals included in the sample (at least 1,017 proposals out of the total of 1,271). Only 49 institutional investors satisfy this criterion. The 80% voting threshold was chosen to ensure that the PCA has a large number of observations.

In addition to the actual voters, the analysis also adds the voting recommendations by two leading firms offering advice to institutional investors about how to vote on various matters at shareholders’ meetings – Institutional Shareholder Services (ISS) and Glass Lewis (GL) – as two additional voters. ISS and GL, known as proxy advisers, inform institutional investors about executive compensation practices and are thus key actors in corporate governance.⁴⁸ Numerous studies show that they have big influence on voting outcomes, including say-on-pay votes.⁴⁹ Proxy advisers also encourage communication and engagement between investors and firms.⁵⁰ The addition of the voting recommendations of proxy advisers increases the number of voters included in PCA to 51 in total.

After defining the number of clusters through PCA, the study proceeds to populate the clusters by other institutional investors by building the correlation matrix of the voting records of investors. This stage relies on a more relaxed selection criteria and includes all institutional investors that voted on at least 30% of all say-on-pay proposals in the sample (at least 382 proposals out of the total of 1,271). Again, the cut-off voting threshold was selected with the aim of having enough observations for a robust correlation analysis. As reported in Table 1 below, this restriction on the data set drops many institutional investors from the analysis but does not lead to a substantial reduction in the recorded shareholder votes. In particular, the correlation analysis includes 152,659 votes (64.09% of the original data set) by 198 unique institutional investors that include 120 investment fund families and 78 pension funds and endowments (14.56% of the original data set). This is because many institutional investors have only a small number of votes on say-on-pay proposals recorded by Insightia's One Voting platform (many of those are relatively small investors or actively select in which companies to invest). In addition to the actual voters, the correlation analysis includes also three additional ‘voters’ – ISS, ISS SRI Funds,⁵¹ and GL. The correlation analysis groups voters with highly correlated voting records together within the same cluster.

Table 1. Sample summary

To be sure that the grouping of institutional investors within clusters is correct and not the result of a chance, the study then relies on text similarity analysis to verify the accuracy of the composition of investor clusters. It is possible that two investors may, independently from each other, arrive to the same voting outcome but for different reasons. The analysis of the similarity of language used by institutional investors in vote explanations can thus be used as a robustness check of the results of the correlation analysis. If the clustering of investors based on the similarity of their voting records is correct, then we would expect investors within the same cluster to use similar language when explaining vote reasons. This analysis was conducted using free online text comparison tools that are usually used to detect plagiarism (Copyleaks).

(ii) Identifying say-on-pay preferences of institutional investors using vote explanations

As a second step, the study uses voting explanations disclosed by institutional investors to define the reasons for their say-on-pay votes and investor preferences on executive compensation. Many investors provide a detailed rationale for high-profile votes – as a rule, when votes are cast against management voting recommendations – to explain how the application of the stated voting priorities has led to the voting decision. In this way, investors can influence the practices of remuneration even where they do not engage with the board of directors directly. The primary audience of this disclosure is thus the board of the affected company, which is expected to rely on the disclosed voting reasons to revise or improve executive remuneration in the future.

Similar to voting records, Insightia One Voting provides access to the vote explanations (reasons) of the votes where institutional investors disclose such information. Insightia's platform includes 11,282 vote explanations on all types of pay-related proposals voted on in the FTSE 100 companies during 2013–2021, of which 9,762 (86.53%) are disclosed in connection with say-on-pay proposals (Table 1).

All these explanations were coded by the author manually to determine the preferences of institutional investors during say-on-pay votes. Investors may explain their votes by reference to one or more factors, such as the structure of the compensation, the link between the compensation and performance (whether financial or non-financial), the quantum (level) of pay, or other. All vote explanations were coded based on the underlying reason(s) and then grouped under seven broad categories of say-on-pay preferences. These categories were identified during an initial coding of a sub-sample of vote explanations. Some of those broad categories, like the quantum and the structure of compensation, include several sub-categories which can offer more granular information on the say-on-pay preferences of institutional investors.

The frequency of references to specific categories can help identify the topics that dominate the preferences of institutional investors during say-on-pay votes. The key challenge during this step is that only a limited number of institutional investors disclose the explanations behind their votes. In total, the sample includes 9,762 vote explanation disclosures by 207 unique institutional investors. Therefore, the recorded data on vote explanations offer information only on the fraction of the say-on-pay votes of institutional investors in the FTSE 100 firms. Major investors like Fidelity Investments, Vanguard, and Capital Group do not explain their voting decisions. This means that the data can be skewed by the preferences of few institutional investors that are more transparent.

To overcome this challenge, the study relies on the clustering of institutional investors. Voting preferences of institutional investors that belong to a specific cluster are extrapolated to other investors in the same cluster. In other words, if one or several investors within a cluster explain their votes, then it is fair to assume that others that belong to the same cluster and voted similarly relied on similar reasoning for their voting decisions. The intuition behind this is the strikingly similar – and often fully identical – language used by groups of different institutional investors in their say-on-pay vote explanation disclosures. The similarity of used language suggests that one of the investors or a third party, most likely a shared proxy adviser, is the original source of the voting decision and explanation behind the decision.

(b) Sample overview

The original sample includes 1,271 say-on-pay proposals voted on in 151 different FTSE 100 companies during 2013–2021, 238,203 votes cast by 1,359 institutional investors, and 9,762 explanations accompanying those votes disclosed by 207 unique institutional investors. These say-on-pay proposals represent almost the three-quarters (74.07%) of all pay related proposals voted on in the FTSE 100 companies during 2013–2021. Table 1 offers a summary of the sample.

Compensation proposals tend to attract lot of attention from shareholders, as suggested by the relatively high disapproval rate – the highest among all different proposal types. But the overall opposition rates of advisory remuneration reports and binding remuneration policies, as illustrated in Figure 1, do not differ substantially. There is also no clear trend on the changing patterns of shareholder support for remuneration-related proposals over time. The instances when a say-on-pay proposal fails to pass are rare. As shown in Figure 1, only 12 advisory remuneration reports failed to receive the minimum required support for approval during the entire sample period. Binding remuneration policies never failed to receive the minimum required votes for approval, but on 20 occasions more than 30% of votes were cast against remuneration policies. More than half of those instances came during the post-Covid years.

Figure 1. FTSE 100 companies with high shareholder rebellion against say-on-pay proposals

3. The preferences of institutional investors during say-on-pay votes

The analysis reveals two key findings. First, the study shows the interconnected patterns of institutional investor voting on say-on-pay proposals. In other words, institutional investors can be grouped into several clusters based on the similarity of their say-on-pay votes. Secondly, the preferences of institutional investors during the stewardship of executive compensation vary across those clusters. The following two sections report these findings.

(a) Say-on-pay vote clusters

The findings show that institutional investors differ in their voting patterns but can nevertheless be grouped into several clusters based on the way investors exercise their say-on-pay rights. The first step in the identification of these clusters, as explained above, is principal components analysis using a smaller sample that includes all institutional investors that voted on at least 80% of all say-on-pay proposals (actual voters) and the two leading proxy advisory firms. Table 2 below reports the results of PCA analysis.

Table 2. Principal components analysis results

Note: The table reports PCA results for institutional investors that voted on at least 80 per cent of say-on-pay proposals in the FTSE 100 companies during 2013–2021. PCA variables are institutional investors; investor votes on say-on-pay proposals included in the sample are the observations. The 80 per cent cut off leaves only 49 investors in PCA (Vanguard is dropped because of zero variance); two additional ‘voters’ are added based on the voting recommendations of proxy advisors. PCA results include 33 components but most of those components have a small weight in the variance. The table reports only the first six components with eigenvalues that are above the average eigenvalues of all components (1.55).

PCA reduces the variables in the data set by computing new variables – principal components. The principal components are correlated with one or more original variables (investors) but are uncorrelated with each other. Thus, PCA allows the grouping of institutional investors into several distinct clusters by aggregating information about investor votes. Investors with similar voting records on say-on-pay proposals appear in the same cluster. Results reported in Table 2 show that six principal components together carry 69.01% of the variance. These six new variables are the clusters of institutional investors based on their say-on-pay votes. The remaining components are dropped because of their little weight in the variance (the eigenvalues of the components dropped from the analysis are below the average eigenvalues of all components together). Remarkably, the first component stands out, as alone this component accounts for 43.90% of the variance in the data set. As such, this component is correlated with the largest number of the original variables. In other words, the first component shows the cluster with the largest number of institutional investors.

The correlation of the original variables with one of the six new components shows the information that the original variables share with the principal component. If one or more original variables are strongly correlated with a principal component, then these variables vary (change) together. Component 1, which explains most of the variance in say-on-pay voting patterns, includes ISS and other institutional investors that vote similarly to ISS recommendations. The second and third components include GL and Hermes EOS, respectively. The remaining three components include smaller groups with different combinations of other institutional investors that do not vote similarly to ISS, GL, or Hermes EOS.

At the next step, these clusters were populated with more institutional investors by relaxing the cut-off criterion for the inclusion of investors in the analysis. Figure 2 presents the correlation matrix of institutional investors based on the similarity of their voting records on say-on-pay proposals.

Figure 2. Institutional investors say-on-pay clusters in FTSE 100 companies, 2013–2021

Note: The figure shows the correlation matrix of institutional investors based on the similarity of their voting records on say-on-pay proposals. The analysis covers the voting records of 198 institutional investors (actual voters) and the recommendations of three proxy advice service providers (additional voters). Each square denotes a pair of two voters. Colour coding is used to illustrate the extent of similar voting. The darker the colour is, the more similar the pair votes. Lighter colours, by contrast, show weak or no correlation between the voting records. Voting records that are negatively correlated at 1 per cent significance level are in orange colours.

The analysis of the similarity of language used by institutional investors in the explanations of say-on-pay votes further validates the grouping of institutional investors into clusters. As expected, voting explanations by investors within the clusters formed around the recommendations of proxy advisers show striking similarity. Most investors within each of those clusters explain their votes by using fully identical language, often with even the same orthographic and punctuation errors; a few other investors in the cluster may use differing language, but nevertheless refer to similar problems. By contrast, institutional investors with idiosyncratic voting patterns use highly distinctive language not shared with others. The text similarity analysis also shows that some investors, like a group of Dutch investment managers and pension funds, belonged to one cluster at the beginning of the sample period but moved to another later. This additional information allows more accurate grouping of those investors because reliance on the voting records alone, due to deviations from the recommendations of any one proxy adviser over the full sample period, could wrongly place them among the independent voters. In sum, the text similarity analysis largely confirms and helps to strengthen the accuracy of the composition of investor clusters identified earlier.

The clustering of institutional investors shows that voting patterns on say-on-pay proposals are not homogeneous. In other words, institutional investors do not have one standard set of preferences for executive compensation. But it is also clear that most investors are not unique voters. They can be clustered into several major groups around different proxy adviser recommendations or the votes of major investors. Moreover, a large number of institutional investors in the sample (over 28%) almost blindly follow one of the three proxy advisers – ISS, GL, or Hermes EOS – across all say-on-pay proposals with vote similarity very close to or even at 100%.⁵² This clustering is in line with the findings of two earlier studies of institutional investor preferences showing that investors do not have unanimous voting preferences but can be grouped based on their ideology and/or corporate governance priorities.⁵³

The two largest clusters are around the recommendations of major proxy advisory firms, ISS and GL. The ISS cluster is the biggest and the most influential by the number of institutional investors. Remarkably, most investors in this cluster, with few exceptions, are based outside the United Kingdom. Similarly, the GL cluster includes primarily foreign fund managers, but the cluster is more uniform in its composition as it includes primarily investment managers associated with Australian, Canadian, and U.S. pension funds. This finding is consistent with the evidence that proxy advisers have a substantial influence on the say-on-pay voting outcomes.⁵⁴ The clustering around the recommendations of proxy advisory firms also explains why the directors of UK publicly traded firms believe that proxy advisers have an excessive influence on CEO pay.⁵⁵ The composition of clusters is also in line with earlier studies finding that North American pension funds tend to vote along with GL recommendations, whereas the votes of mutual funds are associated more strongly with the ISS recommendations.⁵⁶

The Big Three – a term commonly used to describe BlackRock, Vanguard, and State Street, the three largest fund groups – are close to the ISS cluster by voting similarity but fall outside it. Although the votes of the Big Three are significantly correlated with the ISS voting recommendations, both BlackRock and Vanguard deviated from the ISS recommendations on more than 10% of say-on-pay proposals. State Street differed from ISS on just below 10% of say-on-pay proposals. The frequency at which the Big Three did not follow ISS recommendations is much higher compared to investors in the ISS cluster. For comparison, BNY Mellon, Goldman Sachs Asset Management, the University of California, and Wells Fargo Funds Management, all members of the ISS cluster, voted in line with the ISS recommendations on more than 99% of say-on-pay proposals. The Big Three also differed from the ISS cluster by the language used to explain their votes. This means that the Big Three were likelier to rely on in-house teams of governance experts for defining how to vote on say-on-pay proposals than investors in the ISS cluster. This is an important difference because the Big Three cluster, although small in terms of investor numbers, has considerable power to influence votes. Collectively the Big Three held 11.77% of the FTSE 100 index in December 2019, which represents around 15.7% of the votes cast at the annual meetings of the FTSE 100 companies.⁵⁷

Last, a relatively large group of institutional investors that vote independently from others are grouped together. The group of independent voters includes mostly UK-based large and medium-sized fund managers – abrdn, Aviva Investors, Baillie Gifford & Co, Columbia Threadneedle UK, Eden Tree Investment Management, Fidelity International, Kames Capital, Legal & General Investment Management, M&G Investments, Newton Investment Management, Royal London Asset Management, Sarasin & Partners, and Schroders – and pension funds – like Railpen and local government pension funds. Most independent voters approach say-on-pay proposals uniquely by conducting careful in-house analysis. As a result, their voting outcomes are different from others, including from the voting recommendations of proxy advisers. These investors, as suggested by more frequent references to private engagements with corporate boards in vote explanations, are also the most likely to conduct irregular direct meetings with companies before voting.

Unlike the clusters formed around the recommendations of proxy advisers, the voting patterns of independent voters are very heterogeneous. Whereas some independent voters often oppose management recommendations on say-on-pay proposals, others diverge from management recommendations less frequently. This group is thus diverse, with preferences that are not necessarily shared. The grouping of independent voters then shows the aggregated say-on-pay preferences of diverse institutional investors that vote differently from others. By contrast, other clusters are more homogenous.

It is worth clarifying that being associated with a certain cluster does not mean that an investor always votes identically to other cluster members. Some investors do vote in a highly similar way, but more often there are some deviations within clusters, especially in relation to voting in the largest companies, which tend to receive special attention compared to relatively smaller companies.⁵⁸ These deviations are normal because some institutional investors consider the recommendations of proxy advisers as an information source and deviate where they have other sources of information, for example, following a private meeting with the target company's directors.⁵⁹ What the clustering shows is that overall institutional investors within a cluster vote more similarly to each other compared to investors outside the cluster.

(b) The dominant topics of engagement over say-on-pay proposals

This study exploits the tendency of institutional investors to cluster around few centres to overcome the problem of limited disclosure of vote explanations when establishing investor say-on-pay preferences. The clustering of institutional investors shows that many investors vote similarly on say-on-pay proposals. The analysis relies on this finding to establish investor say-on-pay preferences based on the disclosure of vote explanations by some investors within clusters. If several investors within a cluster rely on a specific reason to explain their say-on-pay votes, it is reasonable to expect that other investors in the same cluster vote based on the same reasoning. The rest of this section reports the findings on the dominant topics of voting engagement over say-on-pay proposals.

Figure 3 illustrates the popularity of topics mentioned by institutional investors in explanations that accompany their say-on-pay votes. Four topics stand out in investor discussions: (1) the structure of executive pay; (2) the quantum or the level of pay; (3) the link between pay and performance; and (4) communication and disclosure by corporate boards of their pay practices. In addition, the Big Three tend also to focus on the use of discretion by corporate boards when awarding bonuses or other variable pay.

Figure 3. The preferences of institutional investors during say-on-pay votes

Figure 3 also displays differences in say-on-pay preferences across investor clusters. Voting engagement by the members of the Hermes EOS cluster almost universally targets the link between pay and performance. Other clusters refer to more diverse topics when evaluating say-on-pay proposals. The structure of compensation stands out as the most discussed aspect of compensation in most clusters. This is consistent with survey data showing that most investors rank the structure of pay as the most important element when setting CEO pay.⁶⁰ This is also in line with the common criticism that large institutional investors and proxy advisers impose standard best practice solutions across companies.⁶¹ In this case, the Big Three, investors in the Glass Lewis and ISS clusters, as well as many independent voters promote standard ideas about the ideal structure of pay in all FTSE 100 companies.

The second most common ground for engagement by investors in the Glass Lewis and ISS clusters and among independent voters – but clearly not for the Big Three – is the level of pay. Accordingly, executive compensation from investor perspective is not just about pay and performance; concerns about the failure of board oversight procedures in controlling executive compensation and the societal consequences of pay – to the extent that high executive pay also deepens inequality – matter for investors as well. The broadly shared nature of this topic across clusters reveals the link between the preferences of investors and the themes dominating local discourse. Voting engagement over the level of executive compensation closely reflects the extensive coverage of the high level of executive pay in media and at political and other forums in the UK. Ironically, this also means that fears about limited interest of overseas investors – which now hold the majority of the value of the UK stock market – in the level of executive pay and its disconnect from wider UK society are overstated.⁶² The effective outsourcing of say-on-pay voting by overseas investors to proxy advisers blurs the lines between the preferences of local and most of the overseas investors. Only overseas investors that vote independently, like the Big Three, export values across borders.

Another popular investor voting rationale, especially in the ISS cluster, is better communication by corporate boards and improved transparency of pay reporting. Transparency allows shareholders to oversee executive compensation better. This is in line with the focus of the ISS group on more disclosure to assess firms in terms of their compliance with best practices.

Figure 4 compares changes in the preferences of investor clusters over time. The figure shows that investor attention to the level of pay in the UK is not a recent phenomenon. Although there was some increase in this trend during the later years, most clusters focused on the level of pay during earlier years as well. The Big Three stand out as the only cluster where the level of pay was never a dominant aspect explaining votes on executive compensation.

Figure 4. The dominant topics of investors say-on-pay engagement over time

Notably, investors in the ISS cluster moved away from a substantial focus on disclosure during the early years towards paying more attention to the level of executive compensation during the later years. Improving disclosure practices, as well as changes in the UK company laws and corporate governance standards that were introduced in 2018 explain this shift in preferences. The Companies (Miscellaneous Reporting) Regulations 2018 amended company reporting requirements to introduce reporting on the ratios between CEO and average employee pay for financial years beginning on or after 1 January 2019; the first reporting year was thus 2020.⁶³ The UK Corporate Governance Code 2018, in turn, recommends the alignment of executives’ pension contributions with those of the workforce.⁶⁴

Figure 4 also reveals a major change in the preferences of the Big Three towards the end of the sample period. The exercise of upward discretion by the board remuneration committee gradually replaced compensation structure as the main say-on-pay engagement topic. This change is linked to the attempts of remuneration committees in some companies to ease pay targets during the Covid-19 lockdowns to maintain the level of executive compensation stable against declining performance. Coupled with staff furloughs, state relief, and dividend cuts during the pandemic, this practice caused shareholder outrage.⁶⁵ The Big Three protested against the use of board discretion to reward executive directors whilst leaving intact their policy of tolerating the high level of pay.

Institutional investor vote explanations also offer more detail on the determinants of investor engagement on the level and the structure of pay – the two most popular engagement topics. Figure 5 shows the common determinants of investor votes on the quantum of pay. Institutional investors use several benchmarks for assessing the level of pay in their investee firms. The five most popular benchmarks are: pay levels at peer firms; workforce compensation levels; the pay of a newly hired executive director's predecessor; firm performance; and firm size. Common references to peer firms means that investors regularly compare pay with peers or industry practices for incumbent CEOs and not only for a new CEO. Regarding firm size, an increase in compensation may be justified if it comes with more responsibilities, due to an acquisition, for example, or more complex business conditions. References to the US market by independent voters and investors in the ISS cluster are noteworthy as well: investors perceive CEO pay in the US to be higher and are ready to agree to a higher pay in the FTSE firms if the CEO is hired from or the firm's main operations are concentrated in the US.

Figure 5. Determinants of investor votes on the quantum of pay

Figure 6 presents similar findings regarding the structure of pay. Clearly, the choice of performance conditions for the variable component of executive pay is the dominant topic of engagement by investors in all clusters. Investors oppose pay structures that rely on one or a small number of performance conditions or if the selected performance conditions are not stretching enough, thereby guaranteeing high rewards for average performance. Two other structural aspects that commonly come up during voting engagement by investors across clusters are the short-term bias in the variable pay structure and uncapped or high pay opportunity. Remarkably, relative performance is almost absent from investor preferences on the structure of pay. This means that investors are content with rewarding executives for good industry-driven performance. This may be driven by fairness considerations: if shareholders benefit from good industry performance, it is fair to reward executives for the same reason as well.⁶⁶

Figure 6. Determinants of investor votes on the structure of pay

The broad topics of say-on-pay engagement mean that institutional investors do pay attention to less graphical and headline aspects of the remuneration report and the remuneration policy. As shown above, this also applies to different clusters of investors that effectively delegate voting to a central hub within investor groups, like ISS or GL. Where these groups differ is whether voting engagement is delegated or individual and whether engagement is best practice minded or acknowledges specific company needs.

To be clear, the stewardship preferences of institutional investors on say-on-pay proposals are inferred from the voting behaviour of investors and do not necessarily reflect their intrinsic values. Investors may vote and explain their votes strategically to communicate information to different audiences. The main audience for this information, however, is corporate boards, who are supposed to use the vote outcomes and explanations to make adjustments in the existing pay practices. Accordingly, the voting preferences reflect what institutional investors expect firms to do when setting and paying executive compensation.

(c) The role of ESG targets during voting on say-on-pay proposals

Institutional investors increasingly want corporate boards to link executive pay with ESG targets. Cevian Capital, an activist hedge fund based in Europe, and Allianz Global Investors, one of the largest European asset managers, have announced plans to vote against say-on-pay proposals in large European firms if executive pay does not include ESG performance metrics.⁶⁷ A recent joint study by London Business School's Centre for Corporate Governance and PwC, an audit and consulting firm, reports that 45% of the FTSE 100 companies have an ESG target in the variable component of executive pay.⁶⁸ Against this background, it is important to understand whether investors use references to ESG targets during voting engagement over say-on-pay proposals and, if they do so, which cluster is the driving force behind this movement.

Figures 5 and 6 offer evidence that investors in the ISS cluster and independent voters occasionally rely on poor non-financial performance to question the level of pay and encourage the inclusion of non-financial targets, including ESG targets, into the structure of pay. But non-financial targets are practically absent from the say-on-pay preferences of investors in other clusters.

Figure 7 zooms in on ESG targets by presenting data on investor references to ESG targets over time. The figure shows the growing popularity of ESG targets. The growth in ESG-related explanations in say-on-pay proposals is especially pronounced in 2021 during which independent voters and investors in the ISS cluster referred to non-financial performance aspects in 7.21% and 20.77% of vote explanations on the level and structure of pay, respectively. But only a small group of investors is behind the increasing trend of integrating ESG targets into the pay design. Among independent voters, this list includes Amundi Asset Management, Legal & General Investment Management, Northern Ireland Local Government Officers’ Superannuation Committee, and Sarasin & Partners. The promoters of ESG targets in the ISS cluster were AXA Investment Managers and DWS Investment, which started voting against pay proposals that do not have ESG performance metrics.

Figure 7. ESG-related explanations in investor say-on-pay votes

Negative references to ESG targets in say-on-pay vote explanations offer some clues why many investors are taking a cautious approach to the integration of non-financial metrics into the design of executive compensation. The opacity of those metrics is the main argument against their use. Institutional investors prefer incentive awards that are based (only or predominantly) on financial targets because financial targets are ‘tangible’, ‘quantifiable’, and offer ‘a high level of transparency’. By contrast, opaque qualitative targets can be manipulated to increase the level of pay, thus weakening investor oversight. As noted by one institutional investor, ‘[i]t is hard to judge whether the targets are stretching or not’. This cautious approach corresponds with the warning against using poorly defined and non-transparent ESG targets in executive pay put forward in the academic literature.⁶⁹

To conclude, notwithstanding the growing importance of ESG factors in investor stewardship, ESG targets have not become a dominant and broadly shared topic of investor engagement on say-on-pay proposals. The inclusion of ESG targets in the structure of executive compensation is clearly not a factor that determines the outcome of say-on-pay votes for most institutional investors. Other elements of the structure of compensation, as well as the level of pay and disclosure, are much more likely to explain investor votes. The adoption of ESG targets by almost half of the FTSE 100 companies is thus more the result of a pull factor from corporate boards and their consultants than a push from investors. So far, only a handful of investors have taken the lead in the promotion of ESG targets in the context of executive pay. Their number is too small to draw valid and meaningful conclusions about the characteristics of investors that promote the inclusion of ESG targets into executive compensation other than that all are headquartered in Europe (the UK, France, and Germany). That said, institutional investors are not against the use of non-financial targets in principle. Accordingly, corporate boards willing to use ESG metrics in pay can secure investor votes by using metrics that are clearly defined and measurable and by explaining their relevance for the firm.

4. Discussion and implications

(a) Explanation of the findings

The topics of investor voting engagement during the stewardship of executive compensation offer support for both theories of executive compensation. Investor attention to the structure of pay is in line with the prediction of the efficient contracting theory, according to which the design of pay is critical for mitigating managerial agency problems and improving shareholder value creation.⁷⁰ Investor attention to the level of pay, demands for better disclosure of pay, and the exercise of board discretion, meanwhile, support the predictions of the rent extraction theory, according to which executive compensation as a tool to deal with the managerial agency problem suffers from an agency problem of its own.⁷¹ By closely monitoring the level of pay and improving pay transparency, institutional investors address pay-related agency problems and improve the effectiveness of executive compensation as a tool for improving shareholder value creation.

The significant share of disclosure demands in investor stewardship preferences during voting on say-on-pay proposals is noteworthy. This aspect of engagement has almost been neglected in the literature. But investor focus on better disclosure and communication makes lot of sense in the context of the rent extraction theory of executive pay. Shareholder voting on pay has an important function in corporate governance by constraining managerial rent extraction.⁷² To avoid shareholder outrage, firms are expected to use pay practices with obscure information on the amount and form of executive compensation. This is what Professor Bebchuk and his co-authors term ‘camouflaging’ of pay by its designers.⁷³ Disclosure rules play an important role in mitigating the risk of pay ‘camouflage’. But these rules give significant discretion to corporate boards in deciding how to disclose information; it is also not always possible to ensure ideal compliance with minimum regulatory requirements. As such, shareholders can strengthen the standards of disclosure by demanding more transparency on pay level and design during voting and engagement efforts. Thus, transparency (more disclosure) explanations have an important function in shareholder stewardship of pay by helping shareholders to perform their oversight role.

Differences in the voting patterns and the stewardship preferences of institutional investors over say-on-pay proposals suggest that not all investors agree on how executive compensation can be used to improve shareholder returns. Some investors, as evidenced by references to non-financial ESG performance criteria when assessing the level and structure of pay as well as by demands to better align the level of executive compensation with payments to workforce, may even disagree on whether the exclusive focus of investor stewardship on shareholder returns is appropriate. These investors may consider the protection of the interests of a broader group of stakeholders as a better goal. Alternatively, they may regard the consideration of stakeholder interests as an integral element of managing company exposure to risks, for example, by precluding possible social backlash or unfavourable regulatory interventions, thereby improving shareholder returns in the long run.⁷⁴ Nonetheless, ESG integration is clearly not a broadly shared topic during the stewardship of say-on-pay proposals by the institutional investors of the FTSE 100 companies.

The level of pay is the only topic with potential stakeholder implications that receives broadly shared investor attention. Yet, some investor clusters – most remarkably the Big Three – clearly do not consider the level of pay as a stewardship priority. Both cultural and incentive-related factors can explain this finding. The greater tolerance of the high level of executive pay in the US, where the main investor discourse has focused on the link between pay and performance,⁷⁵ as well as large compensation packages of the top managers of many US-based asset managers,⁷⁶ contribute to the reluctance of the Big Three to oppose high CEO pay. But even in investor clusters where the level of pay receives considerable attention it is hard to draw clear conclusions on whether this attention is driven by the negative societal consequences of high levels of pay or is an attempt by investors to control rent extraction by powerful executives.

The clustering of investors’ say-on-pay preferences around the recommendations of proxy advisory firms means that many investors effectively delegate say-on-pay engagement to proxy advisers. Because the economies of scale give proxy advisers an advantage in supplying their clients with information on voting compared to the efforts of individual investors, this delegation does not necessarily lower the quality of investor voting.⁷⁷ Voting by investors in clusters dominated by proxy advisers is often well informed. Indeed, investor vote explanations in these clusters, particularly in the ISS cluster, are detailed and are based on a careful analysis of say-on-pay proposals. Moreover, the dominant topics of compensation stewardship in the ISS cluster have similarities with the engagement topics of independent voters. All these suggest that clusters dominated by proxy advisers accommodate local market practices and cast informed votes. This finding is consistent with prior studies on the quality of the recommendations of proxy advisers on say-on-pay votes in US firms.⁷⁸

The bigger problem rather is the inflexible approach to compensation stewardship adopted by the major proxy advisers. Vote explanations in investor clusters formed around the recommendations of proxy advisers contain little evidence, if any, on the willingness of proxy advisers to tolerate deviations from best practice standards. Some level of standard voting geared towards best practices is common for all investors: after all, most investors publish their voting guidelines and analyse voting items in the light of their pre-declared preferences. But where the clusters formed around the recommendations of proxy advisers differ from independent voters is in the adoption of an uncompromising engagement approach.⁷⁹ This conclusion corroborates with the complaints of the chairs of major UK listed companies that proxy advisers are inflexible and refuse to engage with companies on critical issues.⁸⁰

The rigid stewardship of executive compensation through proxy advisers contravenes the ‘comply or explain’ principle of the UK Corporate Governance Code, which assumes that there is no one standard of governance that suits all companies. As such, companies in particular circumstances may have good reasons to diverge from best practices.⁸¹ The obvious negative consequence of voting engagement through proxy advisers geared towards best practices is thus the inefficient standardisation of pay across companies and sectors.⁸² In order to satisfy the demands of investors and their proxy advisers, the remuneration committees of corporate boards may be forced to design compensation packages in line with widely adopted market practices even though this may not be the best solution for the company. Indeed, many directors of UK publicly traded companies admit that they are often forced to offer inferior pay structures by following market practices preferred by investors and proxy advisers.⁸³

Last, this study shows the propensity of well-resourced local investors to engage in more individualistic oversight of say-on-pay proposals. The tendency for home and local bias in financial markets is well-known in finance literature. Investors tend to overweight shares issued by domestic and local firms.⁸⁴ Similarly, banks tend to lend more to local businesses.⁸⁵ Factors contributing to this bias can be classified into two groups. First, investors may be subject to behavioural biases such as familiarity and salience,⁸⁶ loyalty,⁸⁷ or patriotism.⁸⁸ Secondly, local investors may have informational advantages, such as strong local relationships and better access to private information,⁸⁹ better ability to process publicly available information,⁹⁰ and industry specialisation where domestic and local markets are dominated by specific industries.⁹¹ Disproportionate asset allocation towards domestic and local shares strengthens further the incentives of investors to invest in information gathering and analysis in relation to proximate firms.

These factors combined can explain why UK-based institutional investors pursue more individual approaches towards the stewardship of say-on-pay proposals of the FTSE 100 firms than, for instance, institutional investors from Germany or the Netherlands. In addition to possible behavioural biases, UK investors are likelier to outperform their international peers in getting superior access to private information from the FTSE 100 firms, building stronger relationships with their managers, processing better public disclosures, and becoming more specialised in oil and gas, financials, and consumer goods industries that dominate the FTSE 100 index. Coupled with disproportionate asset allocation towards the FTSE 100 firms, the information advantage of UK-based institutional investors strengthens their incentives to engage with these firms.⁹² Foreign institutional investors, by contrast, are likely to rely on the services of proxy advisers for voting and engagement on say-on-pay proposals of UK firms. Private conversations with investors confirm that the practices of investor voting and engagement in domestic and international markets differ.

(b) Implications and policy recommendations

The findings of this study have several important implications and reveal areas of say-on-pay votes that require further regulatory reforms. First, the role of proxy advisers, especially ISS, grows with the increasing share of foreign institutional investor ownership in a market. Except for the few largest asset managers that have the capacity to conduct in-house analysis of say-on-pay proposals, overseas institutional investors tend to rely on the recommendations of proxy advisory firms. Accordingly, regulators need to closely follow and subject proxy voting advice service providers to greater scrutiny in countries where local equity markets are dominated by overseas institutional investors, including in the UK where the shareholdings of domestic institutional investors are in decline. The least regulators need to do is: (1) address potential conflicts of interest in the business model of proxy advisers who, in addition to issuing voting recommendations, may also advise companies on how to deal with shareholders; and (2) impose on proxy advisers an expectation of meaningful engagement with companies they cover. Such engagement includes sharing in advance voting recommendations that diverge from management recommendations and the underlying analysis with companies, giving companies enough time to respond, and considering company responses when reviewing the original recommendation (with explanation in the final recommendation of why the proxy adviser's recommendation has remained unchanged notwithstanding explanations provided by the company). These reforms can be implemented through either the direct regulation of proxy advisers or a code of conduct for proxy advisers. In the latter case, reporting expectations for institutional investors regarding the use of the services of proxy advisers can add pressure on investors to demand compliance with the code of conduct from their service providers.

Secondly, given that the pay preferences in the ISS group are largely tailored towards the best practice recommendations of corporate governance codes, the growth of foreign ownership also promotes box ticking regarding compliance with corporate governance codes. Box ticking is reduced when there is a critical mass of well-resourced local investors who are ready to engage with companies individually and consider their explanations for deviation from best practices. In their absence, regulators need to encourage global proxy advisers to work with local advisers to develop voting and engagement recommendations that are better tailored to the needs of local firms and consider their particular circumstances. As an example, until 2015, ISS relied on the Pensions and Lifetime Savings Association (formerly the National Association of Pension Funds), the trade association of UK pension funds, to develop voting recommendations for the UK market.⁹³ Additionally, regulators can require proxy advisers to develop menus of recommendations for clients with different voting preferences. In this way, clients can choose menus that are similar to their own voting policies by, for example, following a menu of recommendations that gives priority to shareholder interests or a different menu that places more weight on the interests of non-shareholder stakeholders.

Thirdly, the importance of best practice corporate governance codes for the recommendations of proxy advisers highlights the crucial role of such codes and industry standards as a tool for promoting certain values and influencing corporate behaviour. For example, if regulators consider stakeholder protection as an important goal of corporate governance, then a best practice recommendation on the integration of non-financial ESG targets in the design of executive compensation can lead to quick changes in the pay practices of local firms by directing voting and engagement efforts of investor clusters. The impact of best practice recommendations is bigger in markets dominated by foreign institutional investors where proxy advisers have more influence. This leads to two important conclusions. First, if a country lacks a corporate governance code with best practice standards adopted for a local market, proxy advisers are likely to base their recommendations on best practices from elsewhere, with outcomes that may be a poor fit for local companies. Secondly, this also means that special care is needed in designing those soft law standards because, once in place, they tend to be promulgated by institutional investors (through their reliance on proxy advisers) and quickly become the market standard. The ‘stickiness’ of the recommendations of best practice corporate governance codes must be kept in mind during the planned review of the UK Corporate Governance Code in the not-too-distant future, which is expected to put more emphasis on corporate sustainability.⁹⁴

Fourthly, foreign investor reliance on proxy advisers means that institutional investor preferences do not move freely across borders. Consider, for example, empirical evidence that European institutional investors are more socially responsible than investors from the US.⁹⁵ US institutional investors, meanwhile, are perceived to be more tolerant towards high levels of executive compensation than UK investors.⁹⁶ But many of those investors do not use in-house analysis for their foreign holdings and, as shown in this study, are in the ISS cluster for the purposes of stewardship of say-on-pay proposals of the FTSE 100 firms. The lack of independent engagement by foreign investors limits or delays the transfer of the values and preferences of institutional investors across borders. Since only a handful of investors are engaging independently with foreign companies, the impact of exported stewardship values remains limited. Again, this stresses the need to design governance standards locally instead of relying on the power of global investors in promoting desirable business practices.

Fifthly, investor engagement is most effective when it is part of a broader movement. Therefore, investor coalitions are thought to strengthen investor engagement. So far, the literature has focused on explicit coalitions through the Collaboration Platform provided by the United Nations-supported Principles for Responsible Investment (PRI),⁹⁷ other transnational investor networks like the European Fund and Asset Management Association (EFAMA) and Climate Action 100+,⁹⁸ and national investor forums and investor associations like the Investor Forum in the UK,⁹⁹ the Canadian Coalition for Good Governance in Canada,¹⁰⁰ Assogestioni in Italy,¹⁰¹ and Eumedion in the Netherlands.¹⁰² But, as this study shows, there are also implicit or tacit coalitions around cluster centres. These can be large coalitions, like ISS, Glass Lewis, and Hermes EOS clusters, but also smaller groups where a major local asset manager attracts other investors and asset owners, like the coalitions formed around the voting decisions of LGIM and Schroders. Tacit coalitions leverage the voting power of lead investors or service providers. In the most extreme cases, lead investors may effectively exercise substantial voting power with little economic interest in a company, thus leading to a form of ‘empty voting’. These coalitions have been overlooked in the literature and operate outside the reach of share ownership disclosure rules and regulatory restrictions on acting in concert. Regulators need to assess the risks of tacit investor coalitions and intervene where the decoupling of economic interests and voting power can lead to perverse incentives.

Sixthly, investor clustering around a few centres leads to reduced engagement perspectives in the market. The lack of engagement perspectives is not concerning where there is strong consensus about the best practices of compensation and the dominant investor group is aligned with this consensus. But standard solutions are not optimal where efficient (or the ‘right’) remuneration structures are likely to differ over time, across sectors, countries, and companies.¹⁰³ Under the current voting and engagement practices clustered around few centres with certain pre-defined best models in mind, directors responsible for pay design may come under a heavy criticism for deviating from the best practice standards even though this may be in the interests of the company, its shareholders, and stakeholders.¹⁰⁴ This standardisation not only promotes executive remuneration models that are not optimal for individual companies, but also reduces governance innovation by limiting the emergence and serious consideration of potentially better alternatives.¹⁰⁵ The lack of diverse investor viewpoints then constrains company choice in selecting a suitable governance model from a variety of options.

Direct regulatory intervention cannot make investor engagement more diverse across firms. If anything, the observed standardisation of engagement is, to some extent, driven by the recommendations of corporate governance codes, which is a form of soft regulatory intervention.¹⁰⁶ Accordingly, there is a need for regulatory tools that can encourage the emergence of different engagement perspectives indirectly. This study's finding on the presence of local investors who are supplying diverse perspectives points out the solution: stewardship codes can exploit the expertise of domestic investors and reinforce their impact by encouraging advance information sharing by the most actively engaging investors. Pre-declaration of voting intentions and voting reasons, as well as subsequent changes in these intentions, can amplify the voice of an informed investor by allowing other ‘satellite’ investors to free ride on this information supply. Pre-declaration can thus strengthen the role of actively engaging domestic investors, whose expertise is now often wasted because of their small shareholdings, by turning them into the central nodes of newly emerging uncoordinated investor clusters.¹⁰⁷ This, in turn, is likely to make investor clusters more dispersed.

The emerging practices of voluntary advance voting intention disclosure show that this proposal can be implemented realistically without creating unreasonable costs and risks. Some institutional investors, like Norges Bank Investment Management, the Norwegian giant oil fund, are already releasing their voting intentions (but not voting reasons) a few days ahead of shareholders’ meetings.¹⁰⁸ At the same time, other investors have weakened their communication practices with corporate boards recently,¹⁰⁹ illustrating the weaknesses of voluntary disclosure and underlining the need for a regulatory intervention. Crucially, pre-declaration of voting intentions ahead of shareholders’ meetings allows affected companies to respond before the actual vote takes place. This, in turn, allows other shareholders to weigh the reasons provided by both the disclosing investor and the company in making an informed decision.¹¹⁰

The problem of standardisation can also be mitigated if corporate boards improve the communication of decision-making on compensation matters, especially where they intend to deviate from the established best practices. To improve the communication between investors and corporate boards, say-on-pay laws should include a requirement for boards to explain in the remuneration report how they addressed large shareholder dissent expressed during the previous meeting. This is already the practice in the UK where the Investment Association, a trade body representing investment managers, maintains a public register tracking significant shareholder opposition (20% or more of votes cast during the meeting) to any resolution, including the company's response to the dissent.¹¹¹ Similarly, the EU has made welcome progress in this regard by requiring Member States to enact laws according to which companies shall explain in the following remuneration report how the shareholder vote has been taken into account.¹¹²

Conclusion

This paper analyses the voting records and accompanying voting rationales of institutional investors on 1,271 say-on-pay proposals put for a vote by the FTSE 100 companies during 2013–2021 to identify the stewardship preferences of institutional investors during say-on-pay votes, changes in these preferences over time, and the main influencers of the votes. The findings show that institutional investors rely repeatedly on several dominant themes of investor voting engagement on compensation – the quantum of pay, the link between pay and performance, the structure of pay, and better communication and disclosure. Investor attention to these topics suggests that investor stewardship of pay is driven by both the need to design optimal pay structures and the urgency of addressing the agency problems of pay through the oversight of pay levels and ensuring better disclosure. In terms of more responsible corporate behaviour, investor efforts focus mostly on the level of pay; investors do not at present demand the linking of executive pay with environmental or social targets on a broader scale but also do not seem to discourage boards from using such targets. Additionally, the findings show that say-on-pay preferences of institutional investors are not homogenous: preferences differ depending on which investors we study. Nevertheless, institutional investors can be grouped into several clusters based on their voting preferences. The major clusters are formed around the voting recommendations of proxy advisers. Another large cluster of primarily UK-based institutional investors takes a more firm-specific approach by engaging with companies individually and relying more evenly on a broader spectrum of topics for engagement.

These findings highlight the role of local investors (market expertise) in the oversight of executive pay, the growing influence of proxy advisers along with the increasing share of foreign institutional investors, and the influence of best practice governance codes in driving investor stewardship preferences.