In last month’s article, I assumed that the incentive designer can easily measure performance and design incentives around these performance measures. For example, it is quite straightforward to determine how many windshields an installer at Safelite installs per day. It is also straightforward to trace a poor installation back to the original installer. However, in other cases, performance measurement is not so easy. For example, how does one objectively measure someone’s creativity, degree of cooperation, ability to adapt and be flexible, etc.? In addition, even if something is easy to measure, is it verifiable by a third party? If so, then a pay-for-performance plan would be legally enforceable, which can also impact the effectiveness of the plan. In general, I want to discuss three broad classes of performance measures:
1. Those that are objectively measured by a third party.
2. Those that are objectively measured but not disclosed to third parties (reduced transparency).
3. Those that are subjectively measured.
By “objectively” measured, I mean that it is possible to measure performance with little ambiguity. For instance, one can easily count number of windshield’s, measure the ton of tomatoes in a shipment, count the number of dead chickens in a flock, etc. However, this does not mean that one cannot manipulate outcomes that can be objectively measured. By “subjective” measurement, I mean that it is almost impossible to measure performance without ambiguity and subjectivity. For example, one’s degree of “cooperation” with peers is difficult to measure using an objective scale. Instead, one must rely on opinions of co-workers and supervisors.
Here are some specific examples:
- In the California processing tomato industry, processors write incentive contracts with growers. Some growers are paid bonuses based on performance measures such as sugar content. A third party, known as the Processing Tomato Advisory Board measures these outcomes.
- Feed conversion ratio and mortality rates in hog or broiler contracts are generally measured by the processor. Hence, these can be objectively measured but they are not necessarily verified by a third party.
- Subjective measures – cooperativeness, flexibility, motivation, good attitude, able to meet deadlines (rise to the occasion), and leadership.
Why is the type of performance measure important? Because it can dramatically impact the effectiveness and credibility of the incentive plan. Transparent, objectively measured performance can result in pay plans and agreements that are legally enforceable and have more credibility. Employees can be fairly certain that they will earn what they are promised when objective, verifiable criteria are used. Incentive plans built around these criteria tend to do “what they are supposed to do” and provide strong incentives to perform, without much room for manipulation and controversy. It is also easy to construct piece rates around these types of performance measures. In general, it is much easier to measure performance related to mechanistic or routine tasks such as picking apples, and data entry, and much more difficult to measure performance for tasks that require creativity, subtlety, and judgment.
Outcomes that can be objectively measured but are not undertaken by third parties can also be used for effective pay-for-performance plans. However, there is greater scope for controversy here and both parties have to develop a reputation for honesty and integrity to avoid conflicts. For example, there is great controversy in chicken broiler contracting because chicken farmers sometimes allege that integrators weigh birds and determine feed conversion ratios without making the performance data available to growers. These types of controversies undermine the incentive effects of the pay-for-performance plan because growers are not sure that they will be justly rewarded even when they deliver good birds or improve feed conversion. In fact, some states are considering creating legislation that mandates that integrators release performance data to any grower who wants them and provides growers with the right to be present at the time quality performance is measured. In contrast, these types of controversies rarely arise in the California processing tomato industry because a third party measures and verifies performance.
Finally, subjective performance measures tend to be the least reliable. However, they are subject to many perception errors (e.g. downgrading a person’s entire performance portfolio on the basis of one error or vice versa, overrating everybody, etc.), and political influence. If there is perceived bias in subjective performance evaluations, employees might rebel and consider the system too unfair or political so they may “game” the system. This undermines credibility and may even create a culture of mistrust that can be counterproductive. In addition, Baker, Jensen and Murphy report data showing that most people tend to overrate their own performance which can make supervisors reluctant to provide poor performance ratings to avoid conflict. Thus, these performance measures may contain less useful information about performance because there could be “performance inflation.”
While subjective performance measures have some pitfalls, they are useful for capturing outcomes that are difficult to measure objectively. Basing performance only on objective measures might be short sighted and incomplete. For those of you who follow football, think of Peyton Manning versus Joe Montana. Peyton Manning has better objective performance numbers (e.g. QB rating, TD passes thrown per season), but Joe Montana is known for his “subjective” attributes such as leadership and performance under pressure. It would not be difficult to imagine that many people would rather have Joe Montana rather than Peyton Manning at the QB position for an important game.
For subjective performance measures to be used effectively, there has to be a reasonable degree of trust between the boss and the employee, and there must not be so much performance inflation that the measures have little useful information content to distinguish “good” from “bad” performance. There are, however, variously methods developed by human resources professionals to improve the quality of subjective evaluations, but an adequate coverage of these methods would be beyond the scope of this article. For people who work in these areas, I would recommend consulting a good textbook such as the one by Milkovich and Newman (see reference section).
The bottom line is that, when considering pay-for-performance plans, employers should devote some resources toward the performance measurement system to ensure that the incentive system is credible. This is particularly crucial if significant bonuses or penalties are based on subjective measures.
References:
Baker, George P., Michael C. Jensen, and Kevin J. Murphy. “Compensation and Incentives: Practices vs. Theory.” Journal of Finance 3(1988):593-616.
Milkovich, George T. and Jerry M. Newman Compensation , 6 th Edition. Boston, MA: McGraw-Hill, 1999