The case for pay for performance

With the Obama administration close to making decisions about the fate of pay for performance in government, I was interested to come across a summary of research about the approach’s effectiveness in improving overall organization performance, and I found there is relatively strong evidence for a generally positive effect.

The study — “Pay and Performance: Individuals, Groups, and Executives” — is in the most recent Academy of Management Annals, which is published each year by the Academy of Management, the primary association of professors who study organizations. The paper is by Barry Gerhart from the University of Wisconsin, Sara Rynes from the University of Iowa (and former editor of the prestigious Academy of Management Journal), and Ingrid Smithey Fulme from Georgia Tech.

The authors divide pay for performance into three types:

1. Incentives for individuals based on objective performance measures.
2. So-called merit pay, with incentives based on more subjective evaluations by an employee’s boss.
3. Group-based incentives.

Analyses of a large number of studies show strong evidence that individual incentives that use objective performance measures produce large output increases. The limitation is that for many jobs, developing such measures is not feasible.

The most common pay-for-performance system in the private sector, as in government, is merit pay. In companies, the approach is plagued by the same problems as in government: grade inflation and insufficient differentiation. However, despite those limitations, studies show that better performance is generally associated with higher merit pay increases, particularly when the effect of higher ratings on promotion is considered.

There are fewer studies of group-based incentives, and the largest number of them look at gain-sharing plans at the level of a large office or even a whole organization. Those studies show positive effects on productivity. There is not much research on group incentives at a workgroup level.

However, the findings come with caveats. Some studies — though not all — involve companies that had pay-for-performance programs at the time of the study. That might create bias because unsuccessful pay-for-performance programs might have been abandoned and thus do not appear in study samples.

Also, in government, an intrinsic motivation to work based on commitment to an agency’s programs might be important, and there is evidence money rewards might reduce intrinsic motivation, decreasing the impact of pay for performance.

Nonetheless, I think a fair reading of the evidence in the Academy of Management Annals article suggests that pay for performance should be considered as a tool to improve government performance.

I am not naïve enough to believe — and this is coming from somebody inclined to regard naïveté as almost as much of a virtue as a vice — that evidence will determine the administration’s decision in this politically hyper-charged area. But it would be nice if evidence played at least some role.