Did Police Funding Cuts Cause the 2020 Crime Surge?

The 2020 homicide increase is one of the most thoroughly confirmed events in modern crime statistics. It shows up in two independent measurement systems that share no common data pipeline: police-recorded homicide counts (FBI UCR) and death-certificate mortality data (CDC/NCHS). When two separate systems using different methods agree, the signal is real.

Source: NCHS Data Brief No. 526 (March 2025) for 2019 through 2023; CDC WONDER 2024 mortality data for the 2024 count. Rates shown are age-adjusted per 100,000 standard U.S. population. ICD-10 codes *U01.4, X93–X95. No age-adjusted 2024 rate is given here because Data Brief 526 stops at 2023 and the crude 2024 rate CDC publishes in FastStats, 4.5 per 100,000, is a different measure that cannot be read against the rates above. The 2024 count is directly comparable: firearm homicides fell to 15,364, close to the 14,414 recorded in 2019.

Source: FBI CIUS press releases (Sep 2021, Oct 2023); FBI UCR 2024 summary (Aug 2025); FBI Reported Crimes in the Nation 2025 (Aug 2026). The murder figure is a change in the number of offenses; the violent crime figure is a change in the rate. The FBI computes each year-over-year percentage against a revised prior-year estimate, so 5.2% does not reproduce exactly from the two published rates above. Note: 2021 data has known comparability issues due to UCR→NIBRS transition.

The spike was severe, but it must be put in historical context. Firearm homicide peaked nationally in the early 1990s, fell through the 2000s to a low of 3.5 per 100,000 in 2014, climbed back to 4.6 by 2016 and held near that level through 2019, then spiked sharply in 2020–2021. Even at the 2021 peak (6.7/100k), the rate was below the early-1990s highs. The narrative of an "unprecedented" crime wave is accurate only for the speed of the increase, not its absolute level.

How to read it: the line is the number of people killed by firearm homicide per 100,000, age-adjusted, each year from 2010 to 2023; the darker points mark 2020 onward. Points: 3.6, 3.6, 3.8, 3.6, 3.5, 4.2, 4.6, 4.6, 4.4, 4.6, 6.2, 6.7, 6.2, 5.6. Source: CDC/NCHS Data Brief No. 526 (2025), Table 4, age-adjusted firearm homicide mortality rates.

The defund hypothesis requires that budgets actually were cut. A peer-reviewed study published in Social Problems examined police budget changes across 264 U.S. cities in the period surrounding the 2020 protests. The finding is unambiguous: there is no evidence that 2020 protest exposure reduced police budgets. In some contexts, police budgets actually increased in cities that experienced the most protest activity.

Two other counts point the same way. A research note in Urban Affairs Review examined about 100 of the most populous cities and found no strong evidence of efforts to defund the police. Bloomberg CityLab's compilation of the 50 largest cities' budgets found police budgets for fiscal 2021 down 5.2% in aggregate, mostly as part of broader pandemic cost-cutting, while police spending's share of general expenditure rose slightly, from 13.6% to 13.7%. Cuts happened in some places, and they were small and short-lived.

This is not a minor caveat, it removes the primary mechanism for the "defund caused crime" claim at the national level. If budgets weren't broadly cut, budget cuts cannot explain a national trend.

Even where formal budget cuts were absent, police departments faced real staffing pressure after 2020 through accelerated retirements, resignations, and reduced recruiting pipelines. In 2020, the FBI reported 696,644 sworn officers nationally across reporting agencies, a figure that masked significant vacancy rates in many large departments.

However, attributing the homicide spike to staffing attrition faces a key problem: the attrition was gradual and ongoing, while the homicide spike was sharp and sudden (occurring within 2020 itself, before most attrition effects would have fully materialized).

"De-policing" is distinct from budget cuts: it refers to reduced proactive enforcement: fewer traffic stops, pedestrian stops, officer-initiated contacts, and arrests for low-level offenses. This is measurable, and it did happen.

A peer-reviewed study published in Perspectives on Politics used a regression discontinuity-in-time (RDiT) design to measure what happened to policing and crime in four cities (Austin, Los Angeles, Philadelphia, and Seattle) immediately after the onset of 2020 protests. Key findings:

A compilation of monthly crime data across 40 large U.S. cities through December 2025, published by the Council on Criminal Justice, provides the most current city-level picture. These cities are not nationally representative, but they show the post-peak trajectory.

Download this conclusion: de-policing

This section as one image: the four-city study of police pullback and violent crime, and the Denver study that runs the other way.

Download the card (PNG, 1,200 × 630)

Source: Council on Criminal Justice multi-city compilation, published January 2026. In the same cities homicide was 44% below its 2021 peak and 25% below 2019. The FBI's preliminary figures for the first half of 2026 show the fall continuing: murder down 23% and violent crime down 10.6% on the same period of 2025. Note: city sample is not nationally representative; offenses vary by city; incident data subject to revision.

One of the clearest correlates of the 2020 homicide spike is a historic surge in firearm purchases. NICS firearm background checks jumped from about 28.4 million in 2019 to 39.7 million in 2020, a 40% increase in a single year and the highest annual total on record. Checks stayed elevated at 38.9 million in 2021 before falling to 31.6 million in 2022.

The CDC data provides critical evidence here: the 2020–2021 surge was almost entirely a firearm homicide increase. Non-firearm homicide did not rise proportionally. This pattern (where the firearm-specific harm increases while non-firearm harm does not) is exactly what a weapon availability hypothesis would predict.

How to read it: each bar is the number of firearm background checks run through the FBI's NICS system in that year, in millions: 25.2, 26.2, 28.4, 39.7, 38.9, 31.6, 29.9, 28.1. Source: FBI NICS firearm background checks, month/year totals. A check is not one-to-one with a gun sold: a single check can cover several firearms, and several states run permit checks through NICS that involve no immediate sale.

The 2020 crime surge coincided with the most severe social disruption the United States had experienced in decades. Schools closed. Courts halted proceedings. Social services were interrupted. Economic stress peaked suddenly. Routine activities that structure social life (work, school, community organizations) were suspended or radically altered.

Criminology research has long linked disruptions to routine activities and social controls with increases in violence. The 2020 spike was also preceded by massive civil unrest: protests, fires, curfews, and the social and psychological aftermath of George Floyd's killing in late May 2020. The homicide increase appears in data starting in June 2020, coinciding with the protests and their aftermath rather than with any specific budget vote.

The strongest version of this argument focuses not on formal budget cuts but on the legitimacy crisis and behavioral withdrawal that followed the Floyd killing and protests. The argument runs: even without formal defunding, officers reduced proactive activity, the data confirms this. Reduced deterrence and reduced clearance rates would be expected to increase violence. The Ferguson Effect (reduced policing following high-profile use-of-force controversies) has some supporting evidence from earlier years. The timing of the June 2020 onset is consistent with a change in officer behavior rather than a budget change alone.

The strongest version of this argument is that the pandemic, not policing, was the dominant driver. COVID-19 disrupted every institution simultaneously: courts stopped processing cases, social services shut down, schools closed, and millions of people lost jobs overnight. Gun purchases nearly doubled. These shocks overwhelmed normal social controls with no connection to police behavior. The same period saw homicide increases in countries that had no "defund" movement. The rapid decline after 2021, while policing controversies continued, suggests the pandemic-era shocks were temporary rather than policing-related.

On the strongest causal and mechanism evidence available, policing changes after 2020 are not supported as the dominant national cause of the 2020–2021 violence surge. "Defund the police" as a national budget-cutting program largely did not happen. De-policing did occur, but the best quasi-experimental evidence finds no clear resulting increase in violent crime in treated cities.

The cross-system evidence most strongly supports a multi-factor pandemic-era explanation with increased firearm availability as a central component. This does not mean policing is irrelevant to crime. Decades of research show it matters. It means the specific "defund caused crime" causal chain is not supported by the available evidence for the 2020–2021 national surge.