Read on Twitter
I have a working paper (currently R&R) with @ajc730 that simulates, using Chicago data, what happens when you fire officers that get the most complaints or use force the most (we do so using a few different percents of what constitutes "the most", https://nyti.ms/32O1PXY
like top 2%, top 10%), during a pre-period (1.5 years or 5 years), "replace" them with another officer who isn't in this top k% of complaints/force (we do this using several different ways of identifying the replacement officers and results don't change much) and see how citywide
complaints or uses of force change in a post period (10 years or 6 years). We find that there are only modest declines, even when simulating firing 10% of officers which would be a drastic increase in the number of officers fired than usual. Why do we find this? Officers who may
rank high in the pre-period may not do so in the post-period (i.e. not perfect stability over time) and replacement officers don't have that many fewer complaints or uses of force than the replaced officers. My takeaways from this single study (that should be replicated elsewhere
) are that it suggests that structural changes are needed to seriously reduce uses of force or complaints against officers. It's not evidence that you shouldn't fire bad officers, only that doing so won't solve the overall problem of complaints or uses of force citywide. If you
want to read the current draft of the paper, that is available here: https://jacobdkaplan.com/documents/UoF_concentration.pdf
