Few board decisions are as scrutinised, or as awkward, as how much to pay the chief executive. The people setting the number sit across the table from the person receiving it, often owe their seat partly to that person, and have to do it all in public. Executive remuneration is the design of that pay package; say-on-pay is the mechanism that lets the owners of the company tell the board what they think of it.

The quick version

  • Executive remuneration is the total pay package for senior leaders, usually base salary, an annual cash bonus tied to short-term targets, and long-term incentives (typically shares that vest over several years), plus pension and perks.
  • The core problem is the agency problem: owners (shareholders) hire managers to run the firm, but managers' interests do not automatically match the owners'. Pay design is the main lever for aligning them, and a place where managers can quietly tilt the rules in their own favour.
  • Say-on-pay is a shareholder vote on executive pay. In the US (Dodd-Frank, 2010) it is advisory; in the UK (since 2013) the forward-looking pay policy carries a binding vote at least every three years, with an annual advisory vote on how it was applied.
  • The trap is mistaking the vote for a fix. Most say-on-pay votes pass comfortably; the value is in the conversation it forces a remuneration committee to have before the meeting, not the show of hands at it.

The idea in depth: why pay is a governance problem, not an HR one

The starting point is older than any pay code. In the separation of ownership from control that defines the modern public company, shareholders own the business but hire executives to run it. Economists call the resulting tension the principal–agent problem: the agent (the executive) has more information and different incentives than the principal (the owner), and will not automatically act in the owner's interest. Pay is the main tool for closing that gap, a bonus tied to profit, or shares that only pay off if the stock rises, are attempts to make the executive richer exactly when shareholders are.

So read every pay package as an alignment device and ask the blunt question: what behaviour does this actually reward? A bonus on this year's revenue rewards a sale even if it destroys next year's margin. Share options reward a higher price even if it came from cutting research the firm needed. Good remuneration design starts not with "how much" but with "for what, over what horizon, and what happens if it goes wrong."

The uncomfortable part is who designs it. In theory the board's remuneration committee, non-executive directors, negotiates with the executive at arm's length, the way an independent buyer haggles with a seller. In practice, the most influential critique of executive pay argues that arm's length is a fiction. In Pay Without Performance (Harvard University Press, 2004), Lucian Bebchuk and Jesse Fried set out the managerial-power view: powerful CEOs influence the boards that set their pay, so packages drift toward whatever managers can extract without provoking public "outrage," rather than whatever optimally aligns them. The practical reading: the structure of the committee, independent members, an independent adviser, a chair willing to say no, matters more than the elegance of the pay formula.

There is hard evidence that pay is not purely performance-driven. In "Are CEOs Rewarded for Luck? The Ones Without Principals Are" (Quarterly Journal of Economics, 2001), Marianne Bertrand and Sendhil Mullainathan found that CEO pay rises with profits driven by factors outside the CEO's control, oil prices, exchange-rate swings, almost as much as with profits the CEO actually generated. Crucially, this "pay for luck" was strongest where governance was weakest, and muted where a large shareholder was watching. For a director, that means separating the windfall from the work: design incentives around metrics the executive can influence, relative performance against peers, not the raw share price, so a rising tide doesn't quietly float the CEO's bonus.

Read every pay package by asking one question: what behaviour does this actually reward, and over what horizon?

An honest limitation. The managerial-power story is influential but contested. Other economists argue much of executive pay reflects a genuine market for scarce talent and the optimal contracting needed to attract and retain it, that big numbers can be rational, not just captured. The truth is almost certainly a mix, and it varies firm by firm. The lesson is not "all executive pay is theft" but the more useful "don't assume the arm's-length bargain happened, check that it did."

What say-on-pay actually does (and the limits)

Say-on-pay is the regulator's answer to that check: if boards can't be relied on to bargain hard, give the owners a vote. But the two big regimes work differently, and the difference is the whole point.

In the United States, the Dodd-Frank Act of 2010 (Section 951) requires public companies to give shareholders a periodic advisory vote on the pay of named executives, at least once every three years, plus a vote on how often that vote happens. Advisory means non-binding: the board can ignore a "no" and still pay what it planned. In the United Kingdom, the Enterprise and Regulatory Reform Act 2013 went further. Quoted companies must put their forward-looking remuneration policy to a binding shareholder vote at least every three years; if shareholders reject it, the company cannot pay outside the last approved policy. A separate annual vote on the implementation of that policy remains advisory.

flowchart TD
  A(["Remuneration committee
designs the package"]) --> B(["Disclosed in the
remuneration report"])
  B --> C{"Shareholder
say-on-pay vote"}
  C -->|"US: advisory"| D(["Board may proceed
even on a 'no'"])
  C -->|"UK: binding policy vote"| E(["A 'no' blocks pay
outside the old policy"])
  D --> F(["Real pressure = reputation,
proxy advisers, next election"])
  E --> F
Two regimes, one pressure point: the binding UK policy vote has teeth; the US advisory vote works through reputation and proxy advisers. Leaders Loop

So the move depends on which side of the Atlantic you sit. A UK committee has to win a binding vote roughly every three years, which makes early consultation with big shareholders a survival skill, not a courtesy. A US committee faces a softer instrument, but a heavily opposed advisory vote is still a public bruise that draws proxy advisers and governance ratings, so it is ignored at the board's peril.

Here is the limit, and it matters. Say-on-pay rarely produces a "no." Drawing on Semler Brossy's tracking of Russell 3000 companies, average support has sat around 90% for years, and outright failures run at roughly one to two percent of companies (Semler Brossy, 2025 say-on-pay reports; figures vary by year and index). A vote that almost always passes looks like theatre. It isn't, but its power is indirect. The vote is the smoke alarm, not the fire brigade: its job is to make the committee behave so the alarm never sounds.

A worked example

Take a mid-cap listed company, call it Harlow Industrials, whose remuneration committee is finalising a new three-year pay policy for its CEO. (Illustrative figures throughout; this is a teaching example, not a real company.) The draft package: a base salary of, say, £700,000, an annual bonus worth up to 150% of salary on hitting profit targets, and a long-term share award worth up to 250% of salary vesting after three years if the share price rises.

A director who has read the evidence above asks two questions. First, the Bertrand–Mullainathan one: is the long-term award rewarding luck? As drafted, it pays out on the absolute share price, so if the whole sector rises on falling input costs, the CEO is enriched for the weather. The committee changes it to vest on relative total shareholder return against a peer index, plus a return-on-capital hurdle the CEO can actually move. Second, the say-on-pay one: would the company's largest shareholders support this in a binding vote? Rather than find out at the meeting, the chair consults them in advance. Two flag that the bonus and the long-term award reward overlapping things and risk paying twice for the same year. The committee trims the annual bonus and lengthens the share holding period to five years.

flowchart LR
  A(["Draft: bonus + LTIP
on absolute share price"]) --> B{"Rewards luck or
genuine performance?"}
  B -->|"Luck risk"| C(["Switch LTIP to relative TSR
+ return-on-capital hurdle"])
  C --> D(["Consult top shareholders
before the binding vote"])
  D --> E(["Trim overlap, extend
holding period to 5 yrs"])
  E --> F(["Policy passes; pay aligned
to controllable, long-term value"])
The work happens before the vote: fix the incentive design, then consult, so the binding vote ratifies a package shareholders already shaped. Leaders Loop

Notice what did the work. Not the vote, by the time it happened, the package passed easily. The value came from the committee using the evidence and the threat of the vote to negotiate a better-aligned package. That is the discipline this Toolkit is pointing at: say-on-pay is most useful as a forcing function for the conversation, not as a referendum on the day.

Frequently asked questions

What's actually in an executive pay package?

Usually three layers plus extras. A fixed base salary; a short-term incentive (an annual cash bonus tied to that year's financial and operational targets); and a long-term incentive, normally shares or options that vest over three to five years against longer-horizon goals. On top sit pension contributions, benefits and perquisites. The design question is the balance between them, too much fixed pay weakens alignment, too much short-term bonus rewards short-term thinking.

Is say-on-pay binding or just advisory?

It depends where the company is listed. In the US, under Dodd-Frank, the vote is advisory, the board can override it. In the UK, the forward-looking remuneration policy carries a binding vote at least every three years, while the annual vote on how the policy was implemented is advisory. "Binding" means the company legally cannot pay outside an approved policy; "advisory" means a defeat is reputationally costly but not legally decisive.

If most votes pass, does say-on-pay do anything?

Yes, but indirectly. High pass rates, typically around 90% support, reflect that the real negotiation happens before the meeting, when committees consult large shareholders and proxy advisers to avoid a public rebellion. The vote's deterrent effect shapes the package; the recorded result is the visible tail of an invisible process. Judge it by the consultation it forces, not the tally.

Why is high CEO pay such a recurring controversy?

Because the numbers have grown far faster than typical pay, and the link to performance is contested. The Economic Policy Institute estimates that CEO pay at large US firms rose 1,085% between 1978 and 2023, against a 24% rise for typical workers, leaving a ratio near 290-to-1 (EPI, 2024). Whether that reflects a market for scarce talent or weak governance is exactly the debate that pay design and say-on-pay try to referee.

What should a non-executive director check before approving a package?

Four things. That the committee is genuinely independent with its own adviser. That incentives are tied to metrics the executive can influence, ideally relative to peers, so windfalls aren't rewarded as skill. That horizons are long enough to discourage short-termism. And that the major shareholders have been consulted before the vote, not surprised by it. If any of those is missing, the arm's-length bargain probably didn't happen.

Related in the Toolkit

Setting pay is a committee job, so it sits inside the wider question of board roles & committees, and it leans heavily on the independence of the people doing it, the distinction between executive and non-executive directors is what makes an arm's-length bargain even possible.

Where to go next