13:10 - 14:50
P3
Room: Club D
Panel Session 3
Barbara Piotrowska - The effects of banning "extremists" on selection into and job satisfaction in the civil service
Niklas Harder, Samir Khalil - What we do in the shadows: A large scale analysis of biased policing in Germany.
Rasmus Schjødt - Making sense of mandatory activation - vulnerable young people's motivation for participation in job placements in Denmark.
Elisa Wirsching - Political Power of Bureaucratic Agents: Evidence from Policing in New York City
Brian Kisida - The Effect of Charter Schools on School Segregation
What we do in the shadows: A large scale analysis of biased policing in Germany.
P3-01
Presented by: Niklas Harder, Samir Khalil
Niklas Harder 1, 2, Samir Khalil 1, 3
1 Deutsches Zentrum für Integrations- und Migrationsforschung
2 Immigration Policy Lab, Stanford
3 University of Potsdam
Through interacting with civilians, police officers serve as “everyday conduit[s] of state power” (Knox et al. 2020) and thus determine who gets protected, suspected, prosecuted, and punished. Racial profiling or other forms of discriminatory policing are therefore a form of systematic inequality in the provision of public safety, one of the most important public goods.
We present the first large scale analysis of biased policing in Germany. We apply the so-called veil-of-darkness test (Grogger & Ridgeway 2006) to extensive administrative data from the register of traffic violations. This test overcomes the several data limitations and takes advantage of the fact that visual characteristics of drivers are likely to be less visible to police officers at night when darkness obscures them. If the number of recorded visually different drivers drops with darkness, this indicates biased policing during daylight. To account for simultaneous time-dependent effects like group-specific driving-patterns, we restrict to times of day that are either dark or light depending on the season. With this data, we can estimate the effect of darkness on the nationality, gender, and age of recorded drivers.
We find robust evidence for biased policing in Germany with considerable variance over states. We present further descriptive evidence to discuss different explanations for these differences.
We also perform various robustness checks. That is, we compare records that are likely not recorded by human officers but machines, we perform efficiency analyses and use an alternative identification strategy by regarding within-day discontinuities in recorded group-rates after sunrises and sunsets.