The question of the debate is whether executive compensation contributed to the crisis. Does contribute mean that executive compensation affected the behavior of executives, or does it mean that executive compensation made the crisis significantly worse? If it means that it affected the behavior of executives, surely that is the case. With different compensation schemes, the financial system would have been different and the crisis would have been different. If it means that it made the crisis worse, how could we know that executive pay made the crisis worse? It is an empirical issue and only empirical work can resolve it. Unfortunately, there is very little empirical work so far because it is so difficult to determine ex ante what excessive risk-taking is and data on executive compensation is limited.
Most of the studies of compensation arrangements in the financial industry look at the named executive officers in banks (i.e., the CEO, CFO, or President) because we have data on the compensation arrangements of these individuals. There has been much conjecture about compensation arrangements of traders in banks, but there is no data and thus no empirical evidence. In addition, even if we had such data, it could not be understood independently of the risk management practices of the institution. To see this, consider a trader: Risk-taking incentives for the trader can be affected by his compensation but also by how his performance is defined. If his funding cost reflects the risks he is taking, his attitude towards risk will be very different from a trader who pays the same cost that his institution pays regardless of the risks he takes.
Turning to the evidence on compensation arrangements for top executives, the published work is the paper that Rudiger Fahlenbrach and I published in the Journal of Financial Economics and the paper that Lucian Bebchuk published with Alma Cohen and Holger Spamann in the Yale Journal on Regulation. Although I discussed my paper already in my first blog post, it is useful to show how different researchers can draw seemingly contradictory conclusions from the same data.
Let’s take the example of Dick Fuld, long-time CEO of Lehman Brothers. Dick Fuld sold equity and options worth approximately $470 million during 2000-2006. At the end of 2006, he held Lehman shares and options worth approximately $950 million, all of which he lost when Lehman went bankrupt. In my paper with Rudiger Fahlenbrach, we focus on the $950 million in vested equity that Dick Fuld voluntarily held at the end of 2006, and argue that this, together with his reputation, provided powerful incentives to maximize long-term shareholder value going forward. We say, given relatively low cash compensation in investment banks, it is not too surprising that the CEO cashed out one third of his equity holdings to diversify his stake and reduce his exposure to the company. In contrast, Bebchuk, Cohen, and Spamann focus on the $470 million liquidated between 2000 and 2006, and indirectly argue that this, together with short vesting conditions, gave Dick Fuld such a large wealth that he would be willing to take on excessive risks. What Bebchuk, Cohen, and Spamann cannot explain is why Dick Fuld voluntary held $950 million in equity in his own bank if he knew that he was taking excessive risks. It is simply not true that Dick Fuld had no skin in the game.
Professor Bebchuk is an outstanding exponent of the view that executive compensation aggravated the crisis. I take the various arguments he has made in several papers seriously. We need to be thinking about those issues and make progress on them. However, in the absence of supportive empirical evidence, I cannot agree with his view that executive pay contributed to excessive risk taking. This view presumes that we know that excessive risk taking was taking place before the crisis and that alternative compensation arrangements would have reduced it. Since we had a horrible financial crisis, it is easy to argue ex post that there was excessive risk taking. However, realizations of risk are not by themselves evidence of excess risk taking. Whenever one takes risks, bad outcomes are possible, even if they have an extremely low probability of occurring. A bad outcome can occur even if its ex ante probability is infinitesimal. One cannot infer from the occurrence of a bad outcome the level of risk that was taken ex ante.
Ultimately, to support the argument that executive compensation aggravated the crisis by encouraging excessive risk taking, it has to be shown that risk taking was in fact excessive and that compensation practices made it more so. Excessive risk taking can be understood in two separate ways. First, risk taking could be excessive if a reduction in risk would increase shareholder wealth. Second, risk taking could be excessive even though it maximizes shareholder wealth because it imposes negative externalities on the financial system. The first type of risk taking could be avoided by better governance, including better compensation schemes. The second type of risk taking cannot be avoided by better governance, but requires mechanisms that internalize externalities created by excessive risk taking.
Irrespective of which definition of excessive risk taking one adheres to, alternative compensation schemes would have been helpful before the crisis in limiting excessive risk taking only to the extent that there was a concern that excessive risk taking was taking place. If we point to the amount of leverage as evidence of excessive risk taking, it is important to note that most banks had room to increase leverage before the crisis given the existing regulatory requirements and chose not to do so. Further, there was no pressure from regulators or from shareholders for banks to reduce leverage before the crisis. In any case, leverage is observable, so that if the board of a financial institution finds it excessive, it can control it by imposing a limit on it. There is no reason to use executive compensation to manage leverage. If we look at the types of assets that institutions were buying before the crisis as evidence of excessive risk taking, we also saw that typical board members and regulators of financial institutions did not have the view that executives were taking excessive risk. Banks made large losses on highly rated tranches of securitizations that were viewed before the crisis as extremely safe. Changing compensation schemes to make executives less willing to take risks before the crisis might possibly have led to a worse crisis as executives might have chosen to have their banks hold more highly rated assets from securitizations.