06 August 2009

The Mystery of CEO Compensation

Over at the Daily Duck, a discussion of minimum wage laws slopped over into a discussion of CEO compensation because ... because ... well, it just did. Harry complained that CEOs are overpaid and I responded that, given sales and revenues, American CEOs are modestly paid. I offered to look at particular companies and see if their CEO compensation was high, or modest. Harry suggested looking at Duke Power, Bank of America and Four Seasons Hotels.

Before we get to that, though, I though that it might be useful to look at the role of the CEO and why CEOs are well-paid. The problem is that we -- and by "we," I mean people who study organizations -- don't really know, because we don't really know what CEOs do.

My own area of org. studies is Strategic Management, which expressly tries to determine why some firms outperform other firms. This is a relatively new area of study -- Strategy dates back to about 1980 -- but we do know some things. We know that the overall economy plays a large role in organizational performance and that the industry firms are in plays an even larger role. So, right away, we know of two important factors in organizational performance where CEOs, seemingly, can't make much of a difference.

There are at least two strands of Strategy, the deterministic school, that say, more or less, that the particulars of the individual firm have almost nothing to do with performance. The Industrial Organization school (representated by Michael Porter and pushed most energetically at Harvard) basically says that, over time, firm performance is a function of five industry level forces: competition, suppliers, customers, new entrants and substitutes. Over the long run, all the firms in an industry will earn the same return and look the same. Strategy is limited to industry level actions, like trying to get the government to prohibit new entrants into the industry. In this world, having one CEO or another doesn't make much difference.

The other deterministic school are the Pop Ecologists (most notably, Hannan & Freeman (1977), if you want to Google them), who basically apply biologically derived population rules to organizations. Again, performance results from the environment, rather than from any choice the firm makes.

I tend to believe, however, that strategic choice (Childs, 1972) does make a difference, and there is some good research to back me up. Although the research confirms that industry level and business level effects are important, they also show that corporate level effects made a difference. The best of these studies (Misangyi, et al, 2006) found that differences between businesses accounted for about 37% of observed variation in performance, while industry and corporate effects both accounted for about 14% of observed variation. (Although it might not be obvious, this is also how we know that related diversification, particularly if the related businesses can share a valuable, unique corporate resource, makes more sense than unrelated diversification. It's pretty clear that the returns from one business do not say anything about the expected returns from another co-owned business in an unrelated industry).

If corporate differences explain about 14% of the difference in returns, then we might think that choosing one CEO over another could only effect a portion of that 14%. I'd argue, though, that the CEO in a large, complex business -- and the businesses Harry suggested we look at are all large and complex -- is somewhat more influential than that. The CEO, after all, is ultimately responsible for deciding what industries the corporation will compete in (acquisitions and divestitures are decisions classically made by the CEO personally). Thus, the industry level effects are, in part, attributable to the CEO. Similarly, the CEO is ultimately responsible for steering corporate resources to one business or another, so the CEO is also responsible, in part, for the business level variation in performance.

On the other hand, the compensation question isn't really, how important is having a CEO to corporate performance. The question is, how important is having this CEO instead of some random, but lower paid, CEO. In other words, how much better is performance because the organization has this CEO and is it worth paying him a lot of money to keep him.

That's a question we can't answer (although my own research, which I'm going to be presenting to the Academy of Management next Monday, suggests that new CEOs reduce organizational returns for about 2 years after they take office because they just can't stop themselves from fiddling with things that aren't broken). There is a fairly large subset of Strategy that argues that the top-management team is the most important determiner of strategy, and thus performance (Hambrick & Mason, 1984). That research has looked at demographic characteristics of the CEO and TMT and has found consistent statistically significant effects on performance. Those effects, however, have been pretty small, explaining about 1% of the observed variance across organizations (Certo, et al 2006). Since we don't know which CEO characteristics, if any, really make a difference, and because organizations are reluctant to change CEOs as part of an academic experiment, we can't really say how much difference CEOs make.

Furthermore, it's not clear that the market has any better idea of what goes into making a good CEO. In fact, it's pretty clear that the market for CEOs is not efficient. The small number of CEOs hired every year, along with the opportunities for opportunism, ensure that (Arrow, 1974; Williamson, 1975). Given that the relationship between CEO identity and performance is, at best, causally ambiguous, it seems clear that the CEO (e.g., Steve Jobs, Jack Welch) can be a unique strategic resource that allows for above-average returns while other CEOs can have neutral or even negative effects on performance. This sort of small number, highly opportunistic, causally ambiguous, high transaction cost issue is exactly where we would expect the market not to function well and, in fact, we don't see companies bidding to pay a market-clearing price for CEOs. Rather, we'd expect to see what we do see, a bureaucratic, path-dependent, authoritarian mechanism for hiring and compensating CEOs. (This is also why government regulation won't make any difference; we already have a regulatory approach and no reason to think the government would do it any differently.)

Having said all that, let's look at the three companies Harry suggested (which I'll probably do over time, since I'm incredibly busy getting ready from the AOM and then comps).

1. Duke Energy.

Duke Energy's 2009 Proxy Statement is here and includes, starting on page 22, a really excellent discussion of its executive compensation policy. This is, in fact, an odd choice for Harry to have made, because this is a state-of-the-art compensation policy that touches all the right bases. It also includes, starting on page 39, a separate discussion of the CEO's compensation. In 2008, it shows that the CEO, James Rogers, received a salary of, um, $0. He received a bonus of $0. Duke Energy has decided to base their CEO compensation entirely on long-term results by giving him various forms of stock grants that he's not allowed to sell for several years after receipt. Based on various objective and subjective factors, the compensation committee awarded him about 80% of the shares he could theoretically have received had Duke exceeded all of its goals. This included a 5% stock bonus due to the fact that no work-related employee or contractor fatalities in 2008, which strikes me as an astonishing achievement for a utility company. All in all, Mr. Rogers received stock worth about $5.5 million, none of which he could sell for years. His pension increased in value by $300,000, which Duke counts towards compensation but I probably wouldn't. He received other compensation worth almost $525,000 (health insurance, flying on the company planes, etc.) for total compensation worth $6.3 million. Duke Energy's sales in 2008 were $13 billion and net revenue was $1.3 billion. Paying the CEO $5.5 million in stock he can't sell strikes me as perfectly reasonable and certainly not outrageous. Also, I can't emphasize enough how impressive the report by the compensation committee is; this is gold-plated corporate governance.

Still to come, Bank of America and the Four Seasons.


Harry Eagar said...

'we know that related diversification, particularly if the related businesses can share a valuable, unique corporate resource, makes more sense than unrelated diversification'

Maybe you know it, and maybe I know it, but having lived through the '70s and '80s, I'd have to say that directors and CEOs don't know it. Or didn't know it.

I picked Duke Power because (I guessed, not knowing anything at all specifically about Duke Power) that its return would be about -- ta da! -- 10%. (Gee, why did I suspect that?)

And never much more, so that the CEO could hardly be compensated on the basis of high returns.

I have only recently read closely the goals that compensation committees set for CEO compensation targets, and the ones I have seen are convoluted in the extreme. That said, the Duke one about work-related deaths strikes me as odd. If a worker in a company truck gets killed in a highway wreck, not his fault, that determines the CEO pay?

That's more of a lottery than a system.

And you wonder why I do not admire American management?

Bret said...

Harry Eagar asks: "And you wonder why I do not admire American management?"

Do you admire anyone who organizes labor and other resources in order to produce things? Or do you just dislike production?

David said...


The rule at Duke Energy seems to be that, if the safety record falls below a certain point, the entire top management team compensation suffers a 5% penalty. If it is sufficiently good, they get a 5% bonus.

Susan's Husband said...

"And you wonder why I do not admire American management?"

No, actually, I don't wonder that at all. What I wonder is why you admire non-corporate American management (e.g., the people who write federal regulations).

David said...

I just took a look, and average ROS in Duke's industry, defined by its 4 digit SIC code, was just under 4% last year. However, Duke did a little worse than average on ROA and ROE.

Harry Eagar said...

Although I don't know this for fact, I presume that much of their business is profit-capped. That's why I picked a utility.

Although, curiously, few if any utilities make profits that bump up against the caps.

Another mystery for the mavens of markets to explain.

joe shropshire said...

An interesting but incomplete list of companies. You really do need to add Digital Equipment Corporation or Cray Research to this list, or we are not going to get anywhere.

Harry Eagar said...

I am not going to pile any more work on our genial host, and I was not looking for outliers, only typical establishments; but if you are going to look at outliers, you won't do better than Met Life and its CEO's $450 million lunch bill.

David said...

Nor am I ignoring you. I'll be back from my conference later in the week and will immediately put my nose to the grindstone.

Harry Eagar said...

But if I were to look for outliers, then Robert Benmosche, he of the $450 million lunches, would be a good choice.

He's not at MetLife now. He's at AIG. Sort of. Sort of.