{"__content__"=>"\n Purpose\n \n \n Methods\n \n \n Results\n \n \n Conclusions\n \n ", "p"=>[{"__content__"=>"Estimate the causal effect of shift timing on discretionary police behavior. We test whether patrol officers’ enforcement activity changes near the end of shift and whether stop selectivity increases as shift-end approaches, evaluating these questions in the context of the “collars for dollars” hypothesis and principal–agent theory."}, {"__content__"=>"We analyze five years of Phoenix Police Department traffic citation data (291,352 citations by 2,924 officers). Outcomes are (i) traffic citation frequency and (ii) stop “quality,” measured as civil and criminal violations per stop. We estimate within-officer changes across the shift using officer fixed effects and extensive temporal and location controls, comparing the same officer at the same hour of day but at different points in their shift."}, {"__content__"=>"The end of shift is associated with a $$\\sim $$25% decline in traffic citation frequency relative to earlier shift-hours. We also observe increased selectivity near shift-end: officers make fewer total citation-containing stops, but those stops result in more cited violations, consistent with late-shift enforcement selectivity rather than overtime-seeking behavior."}, {"__content__"=>"Enforcement follows a predictable “discretion curve,” with proactivity peaking mid-shift and tapering as the end approaches, challenging the overtime-seeking narrative. Findings refine principal–agent accounts of policing by showing that time-within-shift constrains discretionary enforcement and suggest implications for staffing and shift design. Limitations include inferred shift assignment and the focus on a single large agency with a 10-hour schedule."}]}