Field Notes

question

What explains the 2020 US homicide spike?

StatusWorkingUpdatedStands on3 sourcesSupportsnotes

What explains the 2020 US homicide spike, and how much of it can be attributed to reduced proactive policing after the George Floyd protests rather than to the pandemic itself?

The question matters because it is the hinge between two incompatible policy readings. If the spike was driven by a police pullback — a “Ferguson effect” scaled nationally — then proactive enforcement is load-bearing and the reform agenda that curtailed it cost lives. If it was driven by pandemic disruption — social-program shutdowns, school closures, a record surge in gun purchases, economic dislocation — then the spike is a one-off shock with no lesson about enforcement levels. The video The most avoided number in America asserts the police-pullback reading as settled. The evidence does not settle it.

What is known

The spike itself is not in dispute: CDC vital statistics put firearm homicide up about 35 percent from 2019 to 2020, the largest single-year increase on record, concentrated among young Black men.1 Two timing facts constrain every explanation. First, the increase was specific to violence: López and Rosenfeld’s analysis of city data found property crime fell in 2020, so the cause must be something that raises lethal violence while leaving or reducing other offending.2 Second, homicide rose across the whole year and across cities with very different protest intensities, which fits a diffuse shock worse than it fits a localized enforcement response.

The candidate explanations

Pandemic disruption. School closures, the suspension of violence-interruption programs, court shutdowns, mass unemployment concentrated in young men, and a record surge in firearm purchases all moved in the violence-raising direction at once. This account explains the breadth of the spike but struggles to name which component did the work.

Police pullback. Devi and Fryer’s NBER working paper examined “pattern-or-practice” investigations following viral use-of-force incidents and estimated large homicide increases in investigated cities — on the order of 900 excess homicides across five cities over two years — attributed to reduced police activity.3 The study is a working paper, its design assumes the investigations themselves rather than the incidents or protests caused the change, and its magnitudes have been contested; it is evidence that the mechanism exists at pattern-or-practice scale, not a measurement of the 2020 national spike. The earlier Ferguson-effect literature on 2014-15 similarly found at most localized homicide increases.

Social legitimacy and trust. A third reading holds that the protests damaged police legitimacy and community cooperation, raising violence through the witness-and-clearance channel described in Homicide clearance in the United States rather than through officer withdrawal. This mechanism is consistent with the breadth of the spike and with the property-crime decline, but it has not been separately estimated.

What would settle it

A design that separates the three accounts needs city-level variation in protest intensity, police-activity measures (stops, arrests, staffing), gun-sales timing, and pandemic-severity measures, all in the same panel — plus a ruling on whether 2020 levels persist after pandemic conditions and protest intensity recede. The post-2022 homicide decline, if it continues toward pre-2020 levels without a restoration of pre-2020 proactive policing, weighs against the pullback account. Advocacy-side estimates such as Cassell’s should be weighted accordingly.

  1. Scott R. Kegler and colleagues, “Vital Signs: Firearm Deaths and Their Rates — United States, 2019-2020,” MMWR 71, no. 19 (2022), doi:10.15585/mmwr.mm7119a1. ↩

  2. Ernesto López and Richard Rosenfeld, “Pandemic, Social Unrest, and Crime in U.S. Cities,” Criminology & Public Policy 20, no. 3 (2021), doi:10.1111/1745-9133.12564. ↩

  3. Tanaya Devi and Roland G. Fryer Jr., “Policing the Police: The Impact of ‘Pattern-or-Practice’ Investigations on Crime,” NBER Working Paper 27324 (2020). Working paper, not peer-reviewed; its causal attribution and effect sizes are contested. ↩

Built on 3 sources (3 external).

Working out connections…