Are CEOs Overpaid?

Tom Bozzo of Marginal Utility argues, in contrast to the contrarian position taken by Tyler Cowen of Marginal Revolution, that "CEOs Are Overpaid":

Marginal Utility: News Flash: CEOs Are Overpaid: Over the last 25 years, corporate CEO pay has increased about sixfold, adjusted for inflation. This just happens to be about the same as the increase in firms' market capitalization. As Marginal Revolution's Tyler Cowen reported in his NYT column yesterday, two economists (Gabaix and Landier) have a recent working paper that purports to explain the latter as the cause of the former, without resorting to the usual array of CEO labor market failures. Mark Thoma and Brayden King are skeptical... Gary Becker — whose Becker-Posner contribution beat Cowen to the virtual presses by a couple days — suggests that grossly overpaid CEOs are just a few bad apples and that the CEO pay trajectory is just tracking the growth in the resources that CEOs manage.

My two cents is that this paper (which would-be clickers through should note uses a fair amount of math) presents some interesting results derived from fantasy fundamentals — you read a sentence like "Our [CEO] talent market is neoclassical and frictionless" and try to resist the urge to snort.

An appeal to the scarcity of CEO talent as an explanation for CEO pay levels and distributions likewise rings hollow. In the model, the market works to assign the most talented CEOs to the "largest" (market capitalization) firms, yet Gabaix and Landier really make no effort to establish that the largest firms actually employ the most talented CEOs. Among other things, the ability to exercise market power and to ride good market fortune make some CEOs wealthy beyond their demonstrated managerial skill. Or, from a different angle, look at Suzy Welch's stenography of Neutron Jack (BW subscription req'd)...

That said, I have even more substantial objections to the result.

First, using alternative measures of firm size, the "fundamentals" don't support the magnitude of the CEO pay increase as obviously as Gabaix and Landier suggest. The authors wave their hands ... and ... offer that profits could be an admissible measure of market size. ... the leadoff comment at Marginal Revolution ... highlights the excess growth of CEO pay relative to corporate earnings and the potential disconnection between profits and market valuations.

This points to a second, and arguably bigger, problem. Even if you were to accept market capitalization as the appropriate benchmark, the growth of market capitalization reflects various factors that are not causally attributable to CEO talent or effort. Investors' willingness to pay more for a dollar of earnings than they were in 1980 (for the time being, anyway) is Exhibit A.

Even the earnings growth, though, has substantial autonomous components. The recent spike in corporate profits seems to have a public "policy" component — it's hard to believe that other fundamentals have been that much better in the laggard recovery than at the peak of the '90s boom. ... Also, over the 25-year period in which real corporate profits have tripled, real GDP has increased by a factor of a bit over two, so much of the growth in profits is tracking general economic growth. CEOs are being rewarded for this, but not as obvious consequences of their efforts and talents; welcome back imperfect CEO labor markets.

Last, Gabaix and Landier's result would create a conundrum: why has executive pay diverged from that of the non-executive ranks? Their model setup offers no reason not to expect that any employee whose talent increases corporate profits shouldn't see pay tracking firm size. As an attempt to address the conundrum, some Marginal Revolution commenters argued that the value creation resides at the top of the corporate hierarchy. This is not especially consistent with the reality that many corporate decisions are developed or refined at lower ranks and 'sold' to various levels of upper management, often with easily digested action recommendations. So attributing all of the value to the final sign-off is arbitrary. Brayden reassures me that strains organizational theory view employees as "carriers of value," in conjunction with corporate institutions; embodying all value creation in the CEO doesn't hold water.

Where does that leave us? Back in imperfect markets territory, almost certainly.

Comments

[Alejandro]

This points to a second, and arguably bigger, problem. Even if you were to accept market capitalization as the appropriate benchmark, the growth of market capitalization reflects various factors that are not causally attributable to CEO talent or effort. Investors' willingness to pay more for a dollar of earnings than they were in 1980 (for the time being, anyway) is Exhibit A.

Also, over the 25-year period in which real corporate profits have tripled, real GDP has increased by a factor of a bit over two, so much of the growth in profits is tracking general economic growth. CEOs are being rewarded for this, but not as obvious consequences of their efforts and talents; welcome back imperfect CEO labor markets.

Do Gabaix and Landier claim that CEO caused the market capitalization to grow? Do they claim that they created growth in earnings? I don't think so. Market capitalization grew which increased the demand for CEO talent and that increased the CEO payments. The point of the paper is not to explain why did market cap grew. As mentioned it can be because there was an increase in the supply of funds or simple economic growth.

Last, Gabaix and Landier's result would create a conundrum: why has executive pay diverged from that of the non-executive ranks? Their model setup offers no reason not to expect that any employee whose talent increases corporate profits shouldn't see pay tracking firm size.

Lets look at delivery boys. They do increase the company's profit but when the company gets bigger and the marginal product of delivery boys increases the company hires more of them. Now lets look at CEOs. The company gets bigger but it still only hires 1 CEO.

My two cents is that this paper (which would-be clickers through should note uses a fair amount of math) presents some interesting results derived from fantasy fundamentals — you read a sentence like "Our [CEO] talent market is neoclassical and frictionless" and try to resist the urge to snort.

Well, some people can't resist to snort when they read things like what this person wrotes. It is funny how the revier think the paper is based on fantasy fundamentals but at no point makes any description on what they are.

I repeat you can't understand Gabaix and Landier with understanding Rosen's theory of Superstars. The key assumption is that a CEO with talent T cannot be replaced by two CEO's with talent T/2 each.

[tom]

I hate to rewrite comments, but I feel compelled to point out again that there is no way that Tyler Cowen's viewpoint could possibly qualify as "contrarian".

The word suggests that there is an orthodoxy out there that Cowan is opposing -- but there isn't. The word also suggests opposition to an establishment of some kind -- some kind of independent minded, never-say-die, bucking of the sytem attitude -- but CEO's *are* the system.

To say that CEO's deserve the lavish rewards they are getting is the ultimate establishment viewpoint, and not its opposite.

[Tom Bozzo]

Tom: Excellent point.

Alejandro:

1. My understanding of their model is that the effect of CEO talent on corporate profits introduces the causal link between CEO talent and profits/firm size that drives the conjunction between CEO pay and firm size. As the saying goes, this is a blog and not the AER, but I will correct the post if I've misinterpreted their model.

You're right that Gabaix and Landier don't purport to explain the general growth of market capitalization as a byproduct of CEO talent or effort. However, look at the PR justifications of enormous CEO payouts and casual observers could get the impression that they did.

The issue is, it's a problem for shareholders if CEO pay institutions are compensating CEOs for profit growth they don't cause.

2. I might argue that it's a matter of institutional convention that there's only one CEO. The alternative of having division heads report directly to the board isn't incomprehensible. While the actual co-CEO exceptions are rare (and relate to special conditions like temporary post-merger arrangements), more common are spinoffs of divisions or breakups of companies.

3. I thought a "neoclassical and frictionless" CEO talent market was a "fantasty fundamental." The other one, of note, is assuming away the agency problems that are commonly viewed as justifying CEOs' incentive pay arrangements.

As I noted, a problem with the application of 'superstar' theory as well as with the applicability of G&L's model is that there's no demonstration that the market actually is sorting CEOs by talent to the largest firms.

[Robert]

The operative word of Rosen is "Theory". I have a no less potentially valid theory I call Robert's "Theory of Crony Capitalism in Regards to CEO Pay". It posits a "Scratch my back & I'll scratch yours" between directors and executive management combined with a pernicious "Don't ask, Don't Tell" policy between the largest ultimately passive insitutional investors that leads to feedback loop of compensation that only raises eyebrows when the disregard is flagrantly illegal (e.g. VTSS's backdated lookback option awards) or brought to the limelight by other events (e.g. Richard Grasso or Bob Raymond).

The gap between Anglo-Saxon vs. Asian and European exec comp remains large, and it's hard to argue that their enterprises are somehow employing inferior talent as a result of their relative parsimony. It's certainly not borne by the statistics as I've seen them.....

[Lord]

Similar sized industry competitors will have to offer similar compensation or risk losing talent to their competitors. A larger competitor will have subordinates at the same level as the top of smaller competitors so one can expect compensation to rise monotonically with size. One cannot conclude this should rise linearly, and in fact a more natural distribution would logarithmically. If it rises linearly, it is because it can, not that it should.

[Alejandro]

One cannot conclude this should rise linearly, and in fact a more natural distribution would logarithmically.

I quote from the paper

The model predicts a cross-sectional constant-elasticity relation between pay and firm size.

Translation: Firm A is 2 times bigger than firm B. CEO A pay is not 2 times bigger than CEO B. It is 2 to some power less than one.

If it rises linearly, it is because it can, not that it should.

It also predicts that the level of CEO compensation should increase one for one with the average market capitalization of large firms in the economy.

Translation as the "market" capitalization increases (not the individual firm) then the CEO compensation increases increases 1 to 1 with the "market" capitalization. This is actually not exactly the result. Compensation increases to some power gamma which is less than or equal to 1. Gamma is a parameter and they "estimate" it to be equal to 1, it is not a prediction of their model.