Monday, July 20, 2009

Pay for Performance

With center-right policies on offer from the Obama Administration in nearly every direction one looks--e.g., Afghanistan, detainees, health care reform, financial regulation, gay rights--it may be hard to recall that during the 2008 Presidential campaign, one of the more successful lines of attack by the right and center-right went like this: Obama talks about being a post-partisan pragmatist, but on the issues he's an old-school liberal. Sen. McCain challenged then-Senator Obama to name an issue on which he had bucked the Democratic orthodoxy and Obama came back effectively: He, Obama, supported merit pay for public school teachers, thus taking on the teachers' unions. And lately, the President and his Education Secretary have been talking up just this issue.

Here I want to point out the oddity of choosing the current political moment as the time to push merit pay as part of the solution to what ails (some of) our public schools. We are now in the midst of two financial crises fueled in large part by merit pay run amok.

The public outcry over bonuses paid to executives at TARP fund recipients was in part scapegoating caused by misunderstanding of the nature of "bonuses" on Wall Street: These are often de facto salary. But it wasn't all scapegoating. Informed critics believed quite rightly that the system of rewards for short-term paper profits, without any real penalty for back-end failure, led to wildly excessive risk taking--indeed, not just risk, but often the certainty of losses, albeit losses that would register as spectacular gains for just long enough to generate equally spectacular bonuses.

Meanwhile, the single greatest driver of long-term government deficits, not to mention a drag on the economy as a whole, is the high cost of American health care. Much of that cost, as explained in this wonderful article by Atul Gawande in the June 1, 2009 New Yorker, can be understood as the product of a kind of "merit pay" for doctors--a system that financially rewards doctors for performing expensive procedures but not for making patients healthier.

There is, of course, a ready response to both of these examples: They are perversions of merit pay, because the financial incentives are not properly aligned with social benefit. Yet that observation is easier to make than to implement. Corporate managers can be given bonuses in stock or stock options that can't be sold or exercised for some long period of time, but if the period is sufficiently long to prevent incentivizing shortsightedness, it may also become decoupled from actual performance. Likewise, paying doctors more for maintaining their patients' "health" can lead to very difficult measurement issues and adverse selection of patients. My point isn't that pay should never be correlated with performance but rather that this is much easier said than done.

For public school teachers, the problem can be acute. The most obvious objective measure of performance by teachers and schools is student test scores, but this has well-known perverse effects: teaching to the test and even official support for cheating. Here again, the point isn't that these are insuperable obstacles to measuring teacher and administrator performance. The point is simply that merit pay has serious costs.

Among those costs is also demoralization. On the most recent Planet Money podcast, my fellow Cornellian, economist Bob Frank, explains that one problem with merit pay is the Lake Wobegon phenomenon: Most people in most enterprises believe themselves to be performing above average; yet most people, by definition, cannot be among the best compensated; any system of merit pay will therefore leave many, perhaps most, people in the enterprise, disgruntled.

To be sure, there can also be demoralization costs from lockstep systems. I recall a few of my colleagues on a faculty at which I used to teach constantly griping about the fact that a small number of our other colleagues were, in the words of one of the gripers, "stealing their salary," i.e., shirking. Part of what kept this griping from getting completely out of control was the knowledge that the shirkers were not receiving the research stipends that went more or less automatically to productive faculty. Even then, however, the real problem from the organization's standpoint was not the "stealing" but the shirking. That's why, in an environment in which workers can't easily be fired (such as a faculty with tenure) a good organizational leader would find a way to re-energize the shirker. The non-payment of the research stipend did not do the trick, but project-oriented initiatives did, at least in one case I recall of a dean reaching out to an erstwhile shirker.

The bigger point is that a carefully designed and nuanced system of merit pay can be part of a larger system for structuring a workplace, but by itself, merit pay can be useless or counter-productive.

Posted by Mike Dorf