Performance Measure Properties and Incentives

We examine the effects of performance measure properties on incentive system design, using data on incentive contracts for auto dealership managers. The data include information on five properties: two indicators of risk; two indicators of distortion; and one indicator of potential manipulation. We find that these properties have important effects on incentive system design. First, firms appear to choose the "best" performance measure available along these dimensions, and use it for the most important (primary) formula bonus. Second, the properties of this primary performance measure are important determinants of the weight placed on the measure for explicit. Third, firms appear to use other performance measures to balance multitask incentives relative to the primary performance measure. Specifically, we find evidence that second and third bonuses are used to provide better incentives for cooperation and to reduce incentives for manipulation. Fourth, we present evidence that subjectivity, through discretionary bonuses based on subjective performance evaluations, and also through implicit incentives for promotions and other rewards, also appear to play the same roles