Sunday, January 8, 2012

The Root of Some Evil

Pretty, ain't it?
“It has always seemed strange to me,” said Doc. “The things we admire in men, kindness and generosity, openness, honesty, understanding and feeling are the concomitants of failure in our system. And those traits we detest, sharpness, greed, acquisitiveness, meanness, egotism and self-interest are the traits of success.”

— John Steinbeck, Cannery Row

To greed, all nature is insufficient.

— Lucius Annaeus Seneca

Professor Ian Tonks—great name, by the way—put up an interesting column at vox today, in which he discusses recent research he and his colleagues have performed into the postulated links between banker compensation and the financial crisis. He cites a number of interesting results, including the fact that pay for all executives and directors at leading UK companies increased at a substantial rate during the decade preceding the crisis, and at a rate well in excess of pay for all employees, and that executives and directors at finance companies were second in total pay only to “non-cyclical services” firms (including food and drug retailers and telecom). But those looking for his research to confirm their belief that banker pay was tightly tied to company performance will be disappointed:

... contrary to the prediction that pay was over-sensitive to short-term performance, we find that the pay-performance sensitivity of banks is not significantly higher than in other sectors, and in general is actually quite low. Across all industries, we find a weak relationship between executive pay and company performance. The estimates suggest that a 10% additional increase in company share price performance leads to a 0.68% increase in the pay of the CEO, which translates into a £3,726 increase in CEO pay at the median level of £543,200.

We report that although the pay-performance relationship is slightly higher in the financial services sector for both total board pay and pay of the highest paid director, the additional sensitivity is not statistically significant, and is still economically very small. This tiny performance-related element of executive pay means that there is little evidence that executive compensation in the banking sector depended on short-term financial performance. In other words, executives were paid irrespective of performance. In which case, it seems unlikely that bankers were incentivised to take risks, and refutes the suggestion that incentive structures in banks could be blamed for the crisis.

In fact, Professor Tonks and his fellow Order of the Phoenix members1 do find a meaningful correlation between executive and director pay in finance and firm size, which is noteworthy, but the slavering hordes of Occupy Wall Street and well-meaning-but-dim regulators must look elsewhere for evidence that greedy bankster bonuses led to Grandma’s condo in Boca Raton being repossessed.

Problem sorted, right? Not so fast.

* * *

Now, uncredentialed peons like me, who merely work in the industry which everybody and their pet Chihuahua seems to have developed a fully formed opinion on nowadays, do not have pre-publication access to high-powered academic research like that produced by Messrs. Tonks and pals, which is being embargoed from all but fellow travelers in academia. Accordingly, I must read into the Doctor’s slender note some key assumptions about just exactly what sort of data it examined. But, if I read him correctly, I perceive at once a couple of key methodological assumptions which are clearly wrong, and which could have been avoided had the merry researchers simply called a couple of real-life bankers, rather than sallied forth to prove something which makes no sense.

The first problem may be inferred from the Professor’s passing reference to the insignificant correlation between increase in CEO pay and “share price performance.” But if the researchers truly measured “company performance” simply and solely by share price performance, they have got the relationship almost completely ass-backwards. For one thing, everyone who has an even passing acquaintance with the equity markets realizes that public company share prices have only a tenuous, intermittent, and volatile relationship with actual company financial performance.2 For another—and because of this—even the dimmest bulb on the compensation committee of a public company realizes that the CEO and other key executives have only limited direct influence on the evolution of the firm stock price, usually limited to jawboning the market that it is underpriced. Accordingly, they prefer to pay executives for performance entirely (or mostly) within their control. The metrics they use to measure performance are financial ones, including but not necessarily limited to net income growth, return on equity, and perhaps others like asset growth and credit strength, as appropriate. Simplifying greatly, the compensation discussion at most firms—public or private, financial or non-financial—usually boils down to a version of this: “Make a lot of money for the firm, chum, and you’ll get paid a lot of coin.” It is company financial performance which matters most to executive pay, not stock price performance.

This is a common failing of many real and pseudo-academic (i.e., consulting firm) approaches to measuring pay for performance among public companies in general. Researchers get confused by the fact that many firms pay executives with heavy allocations of restricted and unrestricted stock and stock options into believing that stock price performance is the chief or even a major criterion Boards use to pay them. But what you pay somebody does not necessarily have much to do with how much you pay them. This is particularly true in finance, where bankers have traditionally been paid oodles of funny money in order to conserve corporate cash, tie their wealth to the future performance of the firm, and prevent them from leaving the firm voluntarily without suffering material damage to their net worth. I also suspect researchers default to share prices as an input variable to their correlation studies because they are easily available. This is a misleading and lamentable bit of laziness which deserves to be stamped out.

Take it from me: stock prices are an unreliable way to measure corporate performance, and they are an absolutely shitty way to predict executive compensation.

* * *

The second methodological problem which this study seems to suffer from is perhaps more common to finance than other industries, especially in the more highly paid investment banking and corporate banking subsegments. For it is an absolute fact that a very large number of employees in your typical investment bank make enormous amounts of money. Not only do many more bankers than populate the executive suite bring home pay packages which could support small villages in Central Austria comfortably—that is, money which looks like “executive-level” pay anywhere else—but often the CEO and other executive officers of an investment bank are by no means the highest paid employees there. In a decent year, hundreds of employees at large investment banks make millions of dollars, and a substantial subsegment of those bring home tens of millions, if not more. If Messrs. Tonks and friends only collated and computed compensation data for named executive officers and non-executive directors—who, by the way, as non-producers are, relatively speaking, low-paid irrelevancies—then they missed the lion’s share of actual compensation going out the door in my industry. That is certainly the impression I get when I peruse Professor Tonk’s slim précis.

And here is the problem with that: all those uncounted flow traders, M&A bankers, structured products professionals, prop traders, leveraged finance bankers, and derivatives marketers—not to mention all the non-executive group and division heads above them—get paid buckets of simoleons for making money for the firm.

* * *

And this is where I part ways with our dear Herr Professor Doktor regarding his conclusion. If I have correctly identified his study’s methodological weaknesses, not only has he measured the wrong independent variable, but he failed to apply it to the entire set of relevant dependent variables. He doesn’t collect the proper financial performance data—the gross revenue and gross profit metrics upon which investment bankers are paid in the real world—and he doesn’t correlate it against the revenue-producing employees who are producing them. Based upon how my industry actually conducts business and pays its employees, he hasn’t proved anything.

Sadly, Your Dedicated and Evenhanded Bloggist, like many others, would still like to see a comprehensive, data-based investigation of the question which Professor Tonks addresses. Unfortunately, I do not know how one could go about this without at least acquiring time series of aggregate payroll data for all revenue-producing employees at each financial firm, correlated against preferably group or divisional level revenue and profit results. You can just imagine how well that request would go over in the offices of Jamie Dimon or Lloyd Blankfein.

For my part, I continue to believe some banker bonuses were indeed contributory to the financial crisis. My industry’s pay practices and culture were built over decades when the vast majority of business investment banks conducted was agency business. Business like M&A, where you earn a fee for helping a client buy or sell a company, or security underwriting, where you earn a fee for placing client securities with outside investors, or securities market making, where you earn a spread for standing between buy- and sell-side investors as a middleman and temporary warehouser. None of these businesses entailed any material amount of persistent or hidden financial risk to investment banks: we did the deal, we got paid, and we moved on. There are no meaningful, dangerous “tail” exposures from such activities. Accordingly, investment banks got used to toting up the profit and loss for each banker and each business line at the end of each year and paying out a percentage of that as compensation to the people who either brought the money in or who could argue most persuasively they had. Simple.

The problem arose when investment banks (and their bastard cousins and often ultimate owners, commercial or universal banks) began conducting business as principals, either explicitly and in full knowledge, or—most dangerously—in total ignorance. Mouthwateringly profitable leveraged lending, structured products, complex derivatives, and proprietary investing of all kinds meant that investment banks no longer conducted business as short-term conduits of temporary risk, but began accumulating long-term financial risks on or off their balance sheet, often without their own knowledge. But when this happens, the old view that Joe in Structured Products should get a massive bonus in February because he brought in $100 million of fee revenue to the firm this year cannot cope with the fact that Joe’s fabulous trades expose the firm to $1 billion in potential losses over the next five years. Even if some investment banks did develop robust and accurate risk-pricing models which accurately tallied and kept track of the massive tail risks metastasizing on their balance sheets—and recent history puts this assertion in considerable doubt—almost none of them drew the connection to compensation practices. Projected firm profits on trades like Joe’s should never be totaled up front when determining Joe’s pay; they should be amortized over the life of the potential risks the ongoing trade poses to the firm. Most banks just didn’t seem to get this important point.3

* * *

There really is a story to be told in here, somewhere, about exactly how and how much banker bonuses contributed to the aggregation of huge hidden and misunderstood risks in the global financial system. From what I can glean from limited evidence, Professor Tonks’ study is not it. Perhaps one day some academic will actually make the effort to understand how my industry works before they design a study to explain it.


Related reading:
Ian Tonks, Bankers’ bonuses and the financial crisis (vox, January 8, 2012)

UPDATE January 10, 2012: Subsequent to the initial publication of this piece, certain readers inside the sanctum sanctorum of the academic priesthood (or their acolytes) were so exceedingly kind as to direct me to the prepublication version of Professor Tonks et al.’s paper, here. As I suspected, this merry band of scholars only looked at aggregate director pay and highest director pay (usually, but not necessarily, the CEO) as dependent variables, and did not examine compensation to revenue producing ranks within financial institutions. This, as I explain above, is simply and irrevocably wrong. I also can confirm these scamps measured company performance primarily by calculating total shareholder return, based upon the following logic:

The most important measure of company performance is the total shareholder return, since the purpose of performance-related pay is to align the interests of the directors with those of the shareholders.

But this, as I outline above, completely begs the question of how bankers actually are paid and replaces it with the hoary old shibboleth about how they should be paid. This is not research; this is theology. The academics also did try to correlate director pay to slightly less silly measures, like earnings per share, return on assets, and revenue growth, but presumably they found little enough correlation between these variables and Board pay either. As I explain at nauseating length above, they were simply looking in the wrong place(s).

My arguments and conclusions remain unchanged.

1 Gratuitous Harry Potter reference. Sorry.
2 Consider, for example, what happens when a company posts impressive, even record, financial results (most usually net income growth) but fails to meet or exceed investor expectations: the stock price goes down. Consider, as well, a company which posts exceptional results in a falling market: more likely than not, the stock price falls then, too. Stock price is a lousy short-term and even intermediate-term indicator of absolute financial performance, if for no other reason than stock price is (supposed to be) a forward-looking measure, and financial performance is backward looking. Lots of academics seem to have trouble grasping this distinction.
3 And paying Joe 30–50% of his total compensation in unvested stock and options didn’t help much either. Sure, he had to stick around to cash it in, and therefore he was concerned with the continued existence and good stock price performance of his employer, but neither of those things are much that Joe, or anyone else not in the executive suite (and sometimes even there), can do much about. Long-term stock compensation is a pretty weak disincentive to risk-taking; producers like Joe focus much more on booking huge profits—and hence huge bonuses—right now, and devil take the hindmost. I confess I have occasionally argued the opposite side of this too strenuously in the past.

© 2012 The Epicurean Dealmaker. All rights reserved.