GovernmentExecutive.com

With its proposal for a $500 million "human capital performance fund" in the fiscal 2004 budget, the Bush administration has associated itself with the view that pay for government employees should be based less, as Comptroller General David Walker puts it, on "the passage of time" and more on the quality of an employee's performance.

In addition to Walker, who has strongly promoted the idea of pay for performance in the context of his pioneering work on the government's human capital crisis, Sen. George Voinovich, R-Ohio, the leading congressional champion of improved personnel management in government, and Rep. Tom Davis, R-Va., the new chairman of the House Government Reform Committee, have expressed interest in moving government in the direction of tying compensation more closely to job performance.

This shouldn't be surprising. Good-government types, of whom I count myself as one, generally like the idea of pay for performance. First, it focuses government, and its employees, on the idea of performance and results, rather than regarding government service as a sinecure for the security-minded. Second, it seems to be consistent with the idea of making government management more businesslike.

However, I think we need to be careful about how we execute pay for performance plans. They have strong potential downsides, and if not done right, they could make government performance worse, not better.

The first problem involves the negative effect of extrinsic rewards, such as money, on intrinsic motivation. Paying people more money in exchange for better performance has the direct effect, of course, of creating an incentive to perform better. But there is strong, and repeatedly demonstrated, evidence from social psychology that extrinsic rewards exert a countervailing negative effect on the behavior of people who are intrinsically motivated to perform a certain task.

In the classic psychological experiments on this subject (which have been repeated in many different contexts over a 20-year period), children paid to play an enjoyable game played the game for less time after they stopped being paid than did children who weren't paid to play. The reason is that people see extrinsic rewards as controlling, reducing the self-determination benefits of engaging in behavior one has freely chosen and thus making the behavior less attractive.

For people who are intrinsically motivated to perform, the net effect of extrinsic rewards is indeterminate - it depends on whether the direct incentive is greater or less than the indirect de-motivating effect. In theory, increased extrinsic rewards might actually produce poorer performance among intrinsically motivated people.

This is likely to be a special problem in government, for two reasons. First, the evidence shows that a larger proportion of government employees are intrinsically motivated to perform their jobs than their private sector counterparts, meaning that the de-motivating effect is greater. Second, the extrinsic rewards government will give high-performing employees are likely to be modest by private sector standards, reducing their direct incentive effect. So the government could end up in the worst of both worlds. As one academic study of this tradeoff concluded, "Pay people well or don't pay them at all."

My own research on the spread of procurement reform among front-line contracting employees showed that, for employees who were very strong supporters of the reform effort, the more their supervisor was pro-reform, the less an employee changed his or her behavior in a pro-reform direction. For the highly intrinsically motivated, the de-motivating effects of the modest extrinsic rewards supervisors had available outweighed their incentive effects.

The second issue raised by pay for performance systems is the danger that individually based reward systems can hurt in situations where collaboration, teamwork and information sharing in a work group are crucial to good performance. If rewards are given to individuals, people have an incentive to keep information - such as tricks of the trade, advice or informal mentoring - to themselves. Again, research has demonstrated the existence of this problem. People even develop an incentive to make co-workers look bad compared with themselves. Of course, these problems will exist to some degree in any workplace, but pay for performance exacerbates them.

This problem can be ameliorated by making incentives team-based rather than individually based, but that does little to solve the problem of free-riding in teams (in which individuals slack off and hope to benefit from the efforts of others) that is one of the reasons for moving toward pay for performance in the first place.

A third issue involving pay for performance systems is fairness. When I first heard federal labor union leaders express the fear that pay for performance systems might be abused by managers playing favorites, I will confess that my reaction was that this sounded like an effort to protect poor performers from justified accountability. Shouldn't the presumption be that if a supervisor thinks poorly of an employee, this is because the employee isn't performing well?

However, if one looks at the literature on performance appraisal and pay for performance in private sector organizations, including non-unionized ones, it turns out that concerns about fairness and playing favorites are quite common among rank-and-file employees in private organizations as well. This isn't just a government phenomenon. Certainly, at a minimum these systems increase incentives to play office politics, an unconstructive waste of time and resources.

I do not believe we should design management systems simply around the fear of abuse, if they generally work well to improve performance. However, there is considerable evidence that employee morale influences employee performance, and employee perceptions of workplace fairness influence the willingness of employees to go above and beyond the specifics of their job description to demonstrate what researchers call good "organizational citizenship."

I don't believe the fact that there are risks associated with pay for performance means we shouldn't explore it. But we need to pay real attention to how the new systems are designed. Government already does an awful job of using its most important advantage compared with privatesector employers - the inherent meaningfulness of many of the jobs public servants perform. The last thing we should want is for government to abandon the effort to motivate employees based on mission in favor of a straight economistic carrot-and-stick approach.

Pay for performance needs to be melded with, and tied to, a greater emphasis on the importance of agency missions and public service. For example, why don't we call rewards "public service recognition" instead of "pay for performance"? Further, I am inclined to believe that monetary rewards should be tied to improvement along performance measures developed for the 1993 Government Performance and Results Act, and given to everyone in an organization or workgroup that achieves the performance improvement. Tying rewards to performance measures and making them group-based reduces worries about playing favorites and emphasizes the link to performance. This would, however, reduce individual incentives to perform better, since each individual's contribution to the improved performance would be small and the amount of money each would receive would be far more modest than in individually oriented pay for performance systems.

I like the emphasis on performance signaled by these new proposals. I don't like resistance to these changes based on any desire to retain a "good enough for government work" or "it's not in my job description" mentality. But I think we need to be careful so that pay for performance actually improves performance.