Using a time regression discontinuity design, we estimate a 72.7 percent decrease in lower-level “quality of life” arrests, and a 69 percent decrease in non-index crime arrests in Minneapolis following George Floyd’s death. Our results also show that the decrease in arrests is driven by a 69 percent decrease in police-initiated calls for service. Using the same approach, we find a much smaller decrease of 2.7 percent in arrests and a 1.5 percent decrease in police calls following police-involved shootings. Our results, thus, suggest that the Ferguson Effect exists, and it is much larger following highly publicized events of police violence such as George Floyd’s death.
That is from the new AER, by Maya Mikdash and Reem Zaiour, “Does (All) Police Violence Cause De-policing? Evidence from George Floyd and Police Shootings in Minneapolis.” The title I find slightly Straussian, I hope not outright naive.