Metrics Management and Bureaucratic Accountability: Evidence from Policing

Abstract

Abstract Bureaucracies increasingly use quantitative measures to monitor personnel behavior. I develop a model of the incentives created by metrics management , a bureaucratic accountability technique, using policing as a case to show that monitoring can lead public‐interest motivated bureaucrats to focus on work not in the public interest. Second, I develop a new measure of data manipulation in crime statistics: although theory predicts the presence of manipulation, researchers observe only the altered data. I solve this using the fact that police departments can reclassify rapes (but not other violent crimes) as “unfounded,” concluding the reported crime did not occur. Finally, I test the effects of metrics management in policing using a novel data set. CompStat is associated with at least 3,500 additional minor arrests per city‐year, substantial data manipulation, and no decrease in serious crime. These results have implications for bureaucracies implementing metrics management, scholarship using administrative data, and legal implementation.