The problems with the traditional model of banker compensation are well known. To simplify, if a trader (or CEO) is paid a year-end cash bonus based on his performance that year (such as a percentage of profits generated), he will have an incentive to take excess risks because the payout structure is asymmetric; the bonus can’t be negative. That way the trader/CEO gets the upside and the downside is shifted onto shareholders, creditors, or the government.
I was talking to Simon this weekend and he said, “Why a year? Why is compensation based just on what you did the last year? That seems arbitrary.” When I asked him what he would use instead, he said, “A decade,” so I thought he was just being silly. But on reflection I think there’s something there.
Most approaches to solving the compensation problem focus on changing the way the bonus is paid out, not the way it is calculated in the first place. Many people think that bonuses should be paid out in retricted stock that (a) vests over five years and (b) cannot be sold for some period of time after it vests. (This is already the case for top executives–but not most employees–at many big banks.) The goal here is to tie the eventual payout to the long-term performance of the company, via its stock price.
This is better than all cash, but it still has a few problems. For one, the top executives at Bear Stearns and Lehman Brothers already had this type of bonus payout, and it didn’t help. For another, tying managers’ incentives to shareholders isn’t necessarily what you want when it comes to highly leveraged banks, since shareholders also have the incentive to take on too much risk (since they can shift losses to creditors or the government). (For this reason, Lucian Bebchuk has suggested tying long-term compensation to a basket of securites that includes not just common stock but also preferred stock and some kinds of debt.) In addition, this type of payout structure wouldn’t change the incentives for individual traders, since their bonus is calculated based on the one-year performance of their individual book, and then paid out depending on how the entire company does; even if those trades blow up the next year, chances are they won’t affect the stock price that much.
Clawbacks in future bad years are another idea, but at least as proposed by the Obama administration, I think they would only apply in cases of material misrepresentation. Also, it’s hard to claw money back from people who have left your company. In the business world generally, sales compensation agreements often have clawback provisions, but it’s generally understood that you’re not going to sue your former employees to collect on them. (Besides, often the money has already been spent.)
Simon’s idea (at least as I’ve thought it out) is to change the way the bonus is calculated in the first place, instead of (or in addition to) the way it is paid out. He’s right: why should your bonus be calculated every 365 days and based on the last 365 days’ results, weighted equally? For one thing, this kind of lumpiness encourages behavior that is bad for the company; think about all those discounts that salespeople give at the end of a quarter or a year trying to make their quota. More important, Simon’s main point is that December 31 is just too soon to determine how well an individual did during that year.
Instead, what about making your 2009 year-end bonus based on your performance in 2006, 2007, and 2008? That is, by the end of 2009 you would have better information about whether the trades placed in those years had turned out well or badly. There are all sorts of variations possible: you could weight the years differently; you could include 2009 (with a low weight because it’s too early to tell); you could do it on a quarterly basis to smooth out the lumps; you could pay out on a quarterly basis; and so on. But the basic principle is that you don’t calculate the bonus until enough time has elapsed to ensure that the employee deserves it. If you wait long enough, you could even just pay it out in cash instead of restricted stock.
The first objection will be that new employees get screwed, since they will get lower bonuses until they have been with the company for a few years. There are a couple possible solutions to that. The one I like less is that for someone who joins on 1/1/2009, his 2009 bonus could have a full-size target and be based on 2009 performance; but his 2010 bonus would be based on two years of results, his 2011 bonus on three years of results, and his 2012 and later bonuses on four years of results.
My preferred solution, though, is that people simply get smaller bonuses (and maybe somewhat higher salaries to compensate) when they switch companies, and only get the big bonuses after they’ve been around a few years. (You can imagine a new equilibrium where bonuses for employees in their first years are lower than today, but bonuses for long-term employees are higher. I’m not saying in this post that total compensation has to go down; that’s a separate issue.) In the technology startup industry, employees get stock options that vest over four years (with a one-year cliff). If you leave after two years, you give up your last two years of options (and unless the company is already public, you have a difficult choice about whether or not to exercise your options). This increases the cost to the employee of switching companies, which is good on two levels. First, as far as the division of the pie is concerned, it benefits employers (shareholders) relative to employees, which would be a good thing for the banking industry (as opposed to, say, the fast food industry). Second, it makes the pie bigger, since companies are more productive if they have more stable workforces. For these reasons, banks should actually want to move to this type of bonus calculation.
Now, how do we get there from here? Like many markets, if one firm changed its policy and the others didn’t, it would be at a competitive disadvantage. (Of course, this problem exists with all proposals to change compensation practices, including the restricted stock ideas.) First, the banks could possibly get there on their own, just like airlines do when they raise prices; one announces a change, and then (sometimes) the others copy it. This could be helped along if one of those famous self-regulatory bodies would recommend this as a new compensation structure. Or if Goldman took the lead; since (I think) it has longer average tenure than most banks, if it changed its policy today, fewer employees would be affected than at other banks; and if you’re just two years into your banking career at Goldman, would you really leave for Citigroup now rather than sticking it out at Goldman until you have the tenure to take advantage of the new policy? Second, we could have legislation. Or third, we could have regulatory action, since the Fed has already said that compensation practices fall under its jurisdiction as a guarantor of the health of individual bank holding companies and the financial system as a whole.
Changing the basis of the bonus calculation is a more direct way of dealing with the current incentive problems than changing the form of the payout. How about it?