People don't get it on CEO pay

Recently I debated executive pay on Twitter with the FT's Kadhim Shubber, The Times's Daniel Finkelstein and others. I had written a letter to The Times arguing that according to a preponderance of evidence, CEOs are actually worth their huge salaries. Indeed, they probably create, and sometimes destroy, firm value worth orders of magnitude more than they are paid.

I used the example of Thomas Cook boss Harriet Green, whose £3m salary was dwarfed by the £400m wiped off the market capitalisation of her firm when she left. Shubber argued she was a bad example, because her departure was accompanied by worse forecasts that also hurt the firm. This is probably fair—perhaps the portion of the loss down to Green herself was small enough that she really wasn't worth it, though I doubt it.

But Shubber went on to argue that it was bad in general to use share price movements as evidence of executive performance, because departures are associated with uncertainty. It is true that investors are likely to value firms lower if the variance of their expected returns—the spread of possible profits and losses—is higher, even if the average returns expected are the same. This is rational given risk averse preferences. But there are a few reasons why this probably isn't the only, or even the main, factor driving the equity movements we see when bosses move.

Firstly, it has a hard time accounting for cases when shock CEO deaths or departures raise the stock price. When poor Steve Ballmer announced his resignation as Microsoft chief the company instantly got around 7.5%—or billions of dollars—more valuable. The stock price rose 39% in the year following. Sometimes executives are having a huge impact, but a negative one. Despite the uncertainty, markets rise.

Secondly, it's not just the stock price that reacts. When there is a shock death of a CEO, or even of a member of a CEO's family, this hurts profitability, investment and sales growth, particularly if the boss is relatively long-tenured, or if the family death is their spouse or child (not so much if it's their mother-in-law).

Thirdly, it's not just death or departure that hurts or improves prices. CEO hospitalisation dramatically hurts firm outcomes, particularly if the executive is young, highly educated, and if the firm is in a rapidly growing business environment—exactly when CEO influence would be expected to matter the most. Similarly, when the boss has more invested in the firm, or when they are measured as putting in more effort, the firm does much, much better. And when firm control is "inherited"—when CEOs dictate the choice of successor to a relative or friend, firms do substantially worse. All of this points to CEOs mattering a lot.

If this is all true, Shubber asks, then why has CEO pay risen so much over past decades. For a while, we only had hypotheses, suggesting that firms were growing ever larger in size, wider in scope, and more complex and globalised in organisation, making the decisions at the top ever more important. But a recent swathe of papers seem to confirm our intuitions and guesses: 1. CEO deaths, always a cost to firms, have become ever more costly recently; 2. bigger firms have always had more expensive CEOs; merely applying this relationship to the growth in firm size 1980-2003 is enough to explain the average pay rise for bosses.

Boards may use rules of thumb to decide on executive pay, but the reason these rules (and the firms using them) survive, is because they are adaptive for firms; they are good ways of setting pay. Small differences at the top end of talent make large differences for firm bottom lines, especially nowadays. Firms lose a lot when their star performers go, and when they don't bid for the best possible boss. People just don't get it: CEOs really matter!