By Rohit Acharya and Rhett Morris
In 2020, the average U.S. city experienced a surge in its homicide rate of almost 30%—the fastest spike ever recorded in the country. Across the nation, more than 24,000 people were killed compared to around 19,000 the year before.
Homicides remained high in 2021 and 2022, but in 2023 they began to fall rapidly. Projections suggest the national homicide rate in 2024 is on track to return to levels close to those recorded in 2019, prior to the COVID-19 pandemic. Yet that spike in murders continues to deliver major costs in terms of the lives lost, the people incarcerated, and the perception of decreased safety across the country.
Some commentators have suggested the increase in homicides during 2020 was a response to the murder of George Floyd by a Minneapolis police officer in May of that year. Others hypothesized that it was caused by a police “pull back,” in which officers chose to do less work in reaction to the protests that followed Floyd’s death.
As more information has become available, these theories appear to be less supported by evidence than some initially thought.1 The evidence indicates that the national homicide rate was already on track to reach a peak far above the previous year even before Floyd was killed.As Figure 1 demonstrates, murders began to rise rapidly in mid-April of 2020. Cell phone data show this is when residents started leaving home more often as lockdown policies eased and the weather grew warmer. During the 6-week period from April 12 to May 23 (weeks 16 to 21 in Figure 1), homicides went up by an average of 17 murders each week.
After Floyd was killed on May 25, the national homicide rate continued to follow this trend, with additional increases during the 2 weeks around Memorial Day and the 2-week period around July 4. But even the highest point of these additional increases was less than 40 murders above the pre-existing trend. While it’s true that homicides did temporarily rise more than they were already on track to following Floyd’s death, these additional increases are unlikely to explain the 5,000 additional murders seen during the year.
This leaves us with a question: What happened that could have caused homicides to spike in 2020, remain high for 2 years, and then start to decline rapidly in 2023?
New data offers a potential explanation. In this report, we analyze thousands of police records and compare them to changes that occurred in U.S. cities just before homicides started to surge. This showed that the spike in murders during 2020 was directly connected to local unemployment and school closures in low-income areas. Cities with larger numbers of young men forced out of work and teen boys pushed out of school in low-income neighborhoods during March and early April, had greater increases in homicide from May to December that year, on average. The persistence of these changes can also explain why murders remained high in 2021 and 2022 and then fell in late 2023 and 2024.
These analyses point to new answers and offer focused solutions that civic leaders can use to prevent future spikes in murder and reduce current levels of violence. We present these findings in detail below in three main sections:
First, we review evidence from existing research that reveals a formula for how violence develops in cities.
Second, we analyze detailed administrative data from a single city with multiple spikes in murder prior to the COVID-19 pandemic to illustrate how this formula applies to local communities.
Finally, we use lessons from this analysis to offer a new explanation for the national spike in homicides in 2020 and make recommendations on how to avoid future increases in murder.
Brookings, 2024. 20p.
