« Rockonomics | Main | Bryan Caplan speaks »
A contrarian look at CEO pay
Here is my latest New York Times column (non-gated). Excerpt:
Their [Gabaix and Landier) core argument is simple. If we look at recent history, compensation for executives has risen with the market capitalization of the largest companies. For instance, from 1980 to 2003, the average value of the top 500 companies rose by a factor of six. Two commonly used indexes of chief executive compensation show close to a proportional sixfold matching increase (the correlation coefficients are 0.93 and 0.97, respectively; 1.0 would be a perfect match).
By the way, Japanese CEOs are paid much more than many popular or Internet sources indicate. American CEOs are paid about three times more than their Japanese counterparts (on average), but not forty or so times more.
Posted by Tyler Cowen on May 18, 2006 at 07:45 AM in Economics | Permalink
Comments
Is it valid to relate CEO pay to market capitalization since they have little to do with the value of stocks?
why not just relate it to the earnings growth where they do play a major role in determining what happens.
For example if you look at the S&P 500 from the 1949 bottom to the 2000 peak the S&P 500 PE on trailing operating earnings rose from 5.9 to 26.5 and accounted for some 78.5% of the increase in the S&P 500 capitalization. Why should CEO compensation be related to inflation and/or monetary policy that largely determine the PE when their actions have little or no impact on them. EPS is what they influence strongly, and that accounted for only 21.5% of the rise in market capitalization.
If you look at a different period in the 1975 to 2000 period EPS only accounted for 27% of the rise in market capitalization.
We have had a sharp growth in CEO compensation but there is essentially no evidence that EPS growth has changed significantly from its long term trend -- 7.5% for the S&P 500, 7.8% for after tax profits reported by BEA, etc.
Interestingly, if you look at pre-tax profits the trend growth rate has actually slowed.
It looks to me like stock owners have a major agent problem.
Posted by: spencer at May 18, 2006 8:43:40 AM
It looks to me like stock owners have a major agent problem.
Yep.
Posted by: Matthew Cromer at May 18, 2006 8:50:03 AM
I should elaborate.
Anyone observant who works in corporate America knows that in a large corporation, agency problems and management self-incentives are a huge drain on corporate efficiency and profitability.
Posted by: Matthew Cromer at May 18, 2006 8:54:26 AM
One still has to show why executive pays grew with market capitalization while non-executive pay did not. Unless one wants to say that labor productivity do not contribute to market capitalization.
Posted by: Filter at May 18, 2006 9:14:20 AM
I also think the productivity argument is a complete red herring. We know that France's productivity is roughly the same as the U.S.', yet American executives are paid vastly more.
More important, American productivity growth was massive during the period 1946-1973, earnings growth was exceptional , etc., yet somehow CEOs were paid significantly less in relative terms than they are today. What changed?
Posted by: William Goodwin at May 18, 2006 9:45:01 AM
OK, I will say it -- labor productivity does not contribute to market cap. Walmart has 15x the market cap of GM. Does this mean that Walmart's employees are 15x more productive than GM's, or that a greeter deserves 15x the salary of a union welder?
If anything, as Matthew Cromer suggests, it is harder for labor to be productive at a larger company, due to efficiency and agency problems. I wouldn't be surprised if skilled workers (engineers?) were paid more at smaller companies than larger ones.
Posted by: DK at May 18, 2006 9:47:47 AM
Indeed, the efficiency of white collar workers decreases as the size of the company increases.
Posted by: Half Sigma at May 18, 2006 9:55:51 AM
Could you point toward some evidence to support your claim that Japanese CEOs make much more than the CW indicates?
Posted by: Kirk Larsen at May 18, 2006 10:02:04 AM
' More important, American productivity growth was massive during the period 1946-1973, earnings growth was exceptional , etc., yet somehow CEOs were paid significantly less in relative terms than they are today. What changed?'
Income tax laws. The top marginal rate in that period was between 90% and 70%. Since 1983 it's been as low as 28%, and as high as 50%.
Posted by: Patrick R. Sullivan at May 18, 2006 11:23:20 AM
I fail to see why this shows that CEO's are not overpaid. Spencer's point about P/E ratios and macroeconomic conditions is a good one, and there are other issues as well. If widespread technological improvements, for example, have increased the value of firms that was not the doing of the CEO. In any case it's far from obvious why CEO pay should rise linearly with market capitalization.
Most of the arguments advanced by economists to show that these levels are rational give a strong impression of being desperate attempts to justify a phenomenon that plainly does not fit into the standard formulas.
The fact is that CEO's are generally well-protected from market forces. Their pay is set by boards that they select, and whose members are well-rewarded for little work. Any explanation that ignores the real defects of corporate governance in favor of some Procrustean effort to fit things into a simple market model is unconvincing.
Posted by: Bernard Yomtov at May 18, 2006 11:49:25 AM
Sure private equity pays its people very well, but all of those pay packages are tied directly to performance -- most with some sort of clawback if you do really well one year and then suffer a terrible year the next. If you get rich in private equity, it's because your investors got rich along the way. Not necessarily the case with the executive comp that is now getting all the attention.
Posted by: q at May 18, 2006 11:54:29 AM
I'll just mention that Japan has a much higher marginal tax rate than the U.S., which some people may
or may not want to take into account.
Posted by: Ronald Brak at May 18, 2006 12:00:52 PM
The idea that CEOs get paid for performance is hard to empirically distinguish from the idea that they (with the help of their buddies on the board) pay themselves as much as they can get away with until someone squawks. A few years ago Marianne Bertrand and Sendhil Mullainathan came up with a very clever way to distinguish between these two stories in a paper entitled "Are Ceos Rewarded for Luck? The Ones without Principles Are." No pay-for-performance scheme would ever pay a CEO for something completely beyond his/her control (e.g., a change in oil prices in an oil-consuming industry), because that only adds risk to the CEO without improving incentives. But luck might play a role under the "steal until someone squawks" story--when there's more money around, even due to pure luck, there's more to help yourself to before someone gets excited about it. They find that CEOs do indeed get paid for luck, unless they work for a firm in which a single shareholder owns a substantial fraction of the firm. The idea is that in those cases, there is some individual who has a sufficient incentive to monitor CEO pay, whereas this is not the case when ownership is diffuse.
There is also a paper by Kevin Hallock called "Reciprocally Interlocking Boards of Directors and Executive Compensation." He finds that if you are on the board of the company that I'm the CEO of, and vice-versa, we both end up getting paid more.
I don't know about this new research, and I'm open to revising my opinion, but I am at least tentatively convinced that a lot of CEO pay really is rich guys paying themselves and each other lots of money.
Posted by: David J. Balan at May 18, 2006 12:04:20 PM
David, I'm perfectly willing to buy the agency problem you describe. But how do you reconcile "There is also a paper by Kevin Hallock called "Reciprocally Interlocking Boards of Directors and Executive Compensation." He finds that if you are on the board of the company that I'm the CEO of, and vice-versa, we both end up getting paid more." with the lower figure for Japanese CEOs? Interlocking boards have always been common there, should they have robbed their companies blind?
Posted by: Bernard Guerrero at May 18, 2006 12:41:46 PM
Bernard,
You have to do these comparisons ceteris paribus. And comparing companies in Japan vs. the US, that obviously does not hold true.
Posted by: Matthew Cromer at May 18, 2006 12:59:34 PM
I don't see why a firm's larger size directly leads to higher pay because the CEO's "talent has a longer reach". Why can't CEO No. 250 undercut CEO No. 1 until wages are different only by the amount of the difference in their talent?
Posted by: Ali Hasanain at May 18, 2006 1:08:40 PM
OTH, why doesn't CEO No. 1 hire CEO No. 250 to run a division, which might be larger than the 250th largest company?
Posted by: DK at May 18, 2006 1:25:51 PM
The way I described both of those papers, they say more about about the variation in American CEO pay across firms than they do about the average level. To make the Bertrand and Mullainathan paper say something about the average level, it would have to be the case that a large fraction of firms don't have a sufficient concentration of large shareholders to restrain the CEOs. I've looked at the paper again. One of their measures is whether or not there is someone on the board who owns a large share of the firm. There are 740 observations where there are and 3880 where there aren't (p. 924), so there is a large absolute number of firms that are not "well-governed" by this measure. As for the Hallock paper, 123 of 602 observations have any interlock, and only 45 have an interlock involving the current CEO (p.334). I'm not sure if this is a lot or a little, but it's not tiny, though it's less than I would have thought. Both of these results (although the first far more than the second) at least seem to suggest that my original point is right.
As far as international comparisons are concerned, it really is hard to hold everything else constant. If we found that Japanese firms have governance that is just as bad, and CEOs that are just as interlocked, that wouldn't necessarily mean that the difference between U.S. and Japanese CEO pay is due to productivity differences. It might mean that there are other restraints on CEO pay (such as different social norms) that don't apply in the U.S.
Posted by: David J. Balan at May 18, 2006 1:51:16 PM
Matthew,
I don't think you were responding to me. I said nothing about Japan.
Posted by: Bernard Yomtov at May 18, 2006 2:07:13 PM
The P/E of the S&P 500 was ~8x in 1980. Now it is ~18x. Why should CEO's benefit from this trend?
Posted by: CD at May 18, 2006 2:09:57 PM
"One still has to show why executive pays grew with market capitalization while non-executive pay did not. Unless one wants to say that labor productivity do not contribute to market capitalization."
I'm perfectly willing to believe that CEOs are overpaid. But the reason why the labor productivity issue you raise isn't related to market capitalization is that in most cases as the firm expands more people will be hired so the per person contribution of non-executives is not increasing.
Posted by: Sebastian Holsclaw at May 18, 2006 3:18:03 PM
Isn't the most relevant question not "Are CEO's overpaid?" but rather "Are CEO compensation schemes well designed?". If you don't clearly specify what a well designed scheme is, it's impossible to answer the first question.
The answer to the latter question is clearly "No". Executives should be rewarded for how well they manage their shareholders' capital. To judge how well they manage this capital, you need a benchmark. The lowest possible benchmark is the return of a treasury bond or note. If an executive, over say a five year period, cannot return more to shareholders than what they could get with treasury, they should not be rewarded at all.
But in reality the vast majority of executive compensation is via long dated stock options that are struck at spot, not at the forward. For most companies the forward is much higher than spot. That means that the executive is almost guaranteed a large profit - even if he or she adds no value beyond what an investor could have gotten with a treasury bond. In fact, under current compansation schemes, all the executive has to do is generate nominally positive returns on the stock price (~stock value over along period)
in order to profit from their options.
Options also give the executive the chance to profit from volatility in the stock price - volatility that has most likely nothing to do with the performance of the business the executive is supposed to be running.
In a well designed package, options would be struck HIGHER than the forward and certainly not at spot, as they are now. I won't hold my breath for this, although my guess is that this is how Buffett's successor will be compensated.
Posted by: Jeffrey Miller at May 18, 2006 3:21:17 PM
I was talking to the other Bernard, Bernard. :-)
Posted by: Matthew Cromer at May 18, 2006 5:22:59 PM
Ok, Matthew. Sorry I wasn't paying attention.
Posted by: Bernard Yomtov at May 18, 2006 6:34:24 PM
“I also think the productivity argument is a complete red herring. We know that France's productivity is roughly the same as the U.S.', yet American executives are paid vastly more.”
No, the US has 30% higher productivity. France has the same productivity PER HOUR, because they are forced to work few hours, and because the least productive are unemployed and not part of the statistics.
The “agency theory” argument is not very strong. Have agency problems become worse? Remember you are trying to explain growth. By the way high salaries can also be a effective way to solve agency problems.
Three further points:
• All of the economy has seen skill biased wage growth and superstar economics, not only CEO:s. The share of the top 0.1% income that goes to professionals and entrepreneurs has gone up sharply. Maybe you need a more general explanation (larger markets and non-divisibility, value of top human capital higher as complement to rest of economy etc.)
• Do Japanese and European stockowners do worse than Americans? Looking at firm from start should give you the average return of capital, which is what you want. Does anyone want to bet there is no effect with CEO pay? (you could argue the CEO pay as a share of firm costs is small).
• The market capitalization argument is actually pretty interesting. The cost of making bad decisions (and value of good decisions) goes up when one person runs a larger firm. The one person part is the key.
Non-divisibility of superstar talent, you can’t hire 5 CEO:s, plus larger and larger firms may well be the key. The 300 billion dollar firm can only have one CEO, and wants the best available. Firms bid up, as they become larger the value of top management goes up.
Really CEO:s are underpaid if you think about it. 63% of American billionaires are self made. If you startup a firm and add value you can become a billionaire, shouldn’t the market give that as a salary to a CEO that can add a few percentage to firm value?
Posted by: Teller at May 19, 2006 3:58:15 AM
