Association for Public Policy Analysis & Management

Are police officers more likely to make arrests at the end of their shift in order to take advantage of overtime pay? This story is a mainstay of public criticism of police departments and is the subject of lawsuits against police departments filed by litigants across the country. We assess whether police officers make collars for dollars using unique administrative data from Dallas, Texas, which link records on 911 calls, arrests and associated court outcomes, and officer shift assignments. While we find that officers are more likely to receive overtime pay when they make a late arrest, the frequency of arrests actually decreases at the end of the day. Further, the likelihood of a criminal conviction for each arrest increases later in the shift, suggesting officers set a higher threshold for making an arrest. Our findings hold for both arrests that are initiated by civilian calls for service as well as arrests that are officer-initiated. We conclude by estimating a simple dynamic model of an officer's arrest decisions throughout the shift. Because early arrests can incapacitate officers from making late arrests, changes to overtime policy have an ambiguous effect on the total frequency of arrests through differential changes to early and late behavior.