Does policing the police increase crime? We avoid simultaneity effects of increased public oversight during a major scandal by identifying events in Chicago that only impacted officers' self-imposed monitoring. We estimate crime's response to self-and public-monitoring using regression discontinuity and generalized synthetic control methods. Self-monitoring, triggered by police union memos, significantly reduced serious complaints without impacting crime or effort. However, after a scandal, both civilian complaints and crime rates rise, suggesting that higher crime rates following heightened oversight results from de-policing and civilian behavior simultaneously changing. Our research suggests that proactive internal accountability improves policecommunity relations without increasing crime.Many thanks to Robert LaLonde for his guidance and support. Special thanks go to