Private and Public Performance Reports as Drivers of Performance and Determinants of Performance Measure Information Content

Abstract

This dissertation addresses how private and public performance reports affect performance and the information content of performance measures. First, I show how disclosing consumer ratings to the general public affects performance and biases raters. Using data from a health care system, I find that publicly disclosing patient ratings of physicians leads to: 1) performance improvement by the ratings and by objective measures of quality, and 2) a bias among raters, who positively weight a physician’s published average rating in deriving subsequent ratings for the physician. To understand the moderating effects of public attention, I use variation in web traffic to a physician’s disclosed rating. I find evidence consistent with public attention reinforcing raters' bias toward concurring with a physician’s published average rating, thus impeding rating improvement. Within a national distribution of ratings, the disclosure leads to an improvement in ratings by 17 percentile points and a bias in a given physician’s ratings toward his or her published average rating by 24 percentile points. These findings demonstrate that consumer-rating disclosure is a means of performance management, and that resulting bias is a reason to interpret subsequent trends in ratings as understated signals of trends in service. The second section of the dissertation shows an understudied and low-cost way of customizing private performance reports to best drive reported performance, and warns that the private reporting causes reported performance to diverge from unreported performance. A field experiment reveals the performance benefit of customizing a private performance report to include the peer-performance reference point that will most motivate improvement. The below-average performers improve most when shown the median as a reference point. The 50th-75th percentile performers improve most when shown the top-quartile as a reference point. The top-quartile performers improve most when shown the top-quartile as a reference point, but only when reported performance is outcome-based as opposed to process-based. Neither the median nor top-quartile reference point has a more positive performance effect overall. With regard to the performance measure’s content, privately reporting a measure causes the measure to become less correlated with unreported performance. These findings have the following implications. First, the optimal reference point for peer performance comparison depends on 1) an individual’s initial performance relative to each reference point, and 2) whether the performance measure regards an outcome or process. Second, a performance measure, once reported, becomes a less informative signal of unreported performance.