Performance-based compensation and novel financial instruments: an explosive combination

Econbrowser draws a connection between known problems with performance-based compensation, given to CEOs to “align” their interests with those of shareholders, lack of regulatory oversight, and financial innovation (emphases mine) …

Econbrowser: Executive compensation

… That the incentives for CEOs need not necessarily coincide with those of the shareholders is a well understood phenomenon that is a special case of what economists call the principal-agent problem. This arises in situations when an agent (in this case, the CEO) has better information about what is going on than the principals (in this case, the shareholders) who rely on the agent to perform a certain task. One way to try to cope with these problems of asymmetric information is to tie the agent's compensation directly to performance.

What caused that principle to go so badly awry in the present instance? I believe there was an unfortunate interaction between financial innovations and lack of regulatory oversight, which allowed the construction of new financial instruments with essentially any risk-reward profile desired and the ability to leverage one's way into an arbitrarily large position in such an instrument. The underlying instrument of choice was a security with a high probability of doing slightly better than the market and a small probability of a big loss. For example, a subprime loan extended in 2005 would earn the lender a higher yield in the event that house prices continued to rise, but perform quite badly when the housing market turned down. By taking a leveraged position in such assets, the slightly higher yield became an enormously higher yield, and while the game was on, the short-term performance looked wonderful. If the agent is compensated on the basis of current performance alone, and the principal lacks good information on the exact nature of the risks, the result is a tragically toxic incentive structure…

In point form then the contributing factors were:

1. the well known principal-agent problem
2. poorly designed performance based compensation (an attempt to mitigate the principal-agent problem, but the time scales are not aligned)
3. leverage without regulatory oversight
4. novel financial instruments with a no more than average* (probably below average when expenses are incorporated) expected value but a skewed probability distribution – so there was a high probability of slightly above average returns and a low probability of catastrophic failure

It’s a persuasive argument, consistent with Lewis and Einhorn on repairing the financial world, Lewis (again) reporting on the “end of Wall Street” and Henry Blodget writing for the Atlantic – Why Wall Street Always Blows It.

My take away lessons are:

1. We need to be very careful with performance based compensation. The problem isn’t that it’s not an effective incentive, the problem is that it’s too effective an incentive. The physician “pay for performance” crowd should pay attention to this, but they won’t.
2. We need regulatory oversight. Yeah, we knew that.
3. Novel financial instruments, whether they provide new ways to skew probability (2008), or new forms of leverage (1929), need to set off red flashing lights and screaming alarms in the Treasury, Congress, the White House and the Federal Reserve. Our government and regulatory agencies need to be doing continuous “war gaming” about how new technologies will transform finance, and constantly about how to avert future catastrophic scenarios.

See also:

1. Jumping the canyon of Great Depression II
2. Stimulus and the scale of under-utilized global productive output
3. Lewis and Einhorn: repairing the financial world
4. The role of the deadbeats
5. Complexity collapse
6. Disintermediating Wall Street
7. The future of the publicly traded company
8. Marked!
9. Mass disability and income skew
10. The occult inflation of shrinking quality
11. You get what you pay for. The tragedy of the incentive plan.

* In this sense “average” refers not to the averaging over the lifespan of the instrument, but rather all the probable outcomes – an expected value calculation really.