Yesterday in “The Minneapolis Effect,” while citing City Crime Stats day-by-day data on homicides in a couple of dozen major American cities, I lamented that the University of Pennsylvania researchers hadn’t provided an aggregate graph summarizing all their cities, which could help distinguish between the validity of the crystallizing conventional wisdom that, well, yeah, there was a murder surge in 2020, but it was just bad luck due to covid, and nobody should even think that it might have anything to do with the George Floyd Racial Reckoning versus my view: of course BLM and the media and the politicians who encouraged the hundreds of Mostly Peaceful Riots are largely to blame.
In response, an old friend, the one who provided the data for my December Takimag column on excess deaths in the US during the pandemic, scraped the data for me from the City Crime Stats site.
The above graph shows pretty clearly the homicide trends over the course of 2020 vs. the average for the five previous years 2015-2019 across 19 cities for which CCS had day-by-day homicide stats through the end of September 2020. I created running totals for 2020 versus the average for 2015-2019. The red line represents the percentage change from the cumulative 2015-19 average to the cumulative 2020 total.
Because the the percentage difference during the earliest days of January bounces around wildly due to a small sample size, I didn’t start the graph until January 15, at which point the cumulative number of homicides in 2020 was 15% above the 2015-19 average.
The last few months of winter proved fairly violent, with the cumulative homicides so far in 2020 being 15% higher on March 16 than for the preceding five year average up through that data.
Why homicides were up in the winter of 2020 before the pandemic is hard to say. It could be a fluke — the winter sample size is small compared to the rest of the year. And, I suspect, that winter murder counts might tend to bounce up and down erratically depending upon how mild or harsh the weather is.
But, the day by day data shows that there was little direct cause-and-effect connection between the pandemic panic and 2020’s increase in murders.
San Francisco declared a lockdown on March 17 (blue vertical line on the graph), and much of the rest of urban America did too over the next couple of weeks.
Oddly, after an initial decline in murders, not much changed over the first ten weeks of pandemic panic. By Memorial Day, May 25, 2020’s cumulative homicides were running 14% ahead of 2015-19, down very slightly from the 15% before the first lockdown in mid-March.
This is not to say that the pandemic didn’t have long term effects on the murder rate. Perhaps lockdowns made people stir-crazy? Perhaps face masks contributed to the return of carjacking? Releases of prisoners due to fear of prison covid epidemics? One theory among the NYPD is that the aversion to the subways meant that bad people couldn’t get away from their bad neighborhoods when their bad neighbors started to get on their nerves. Domestic violence was expected to skyrocket, and that may have happened in Phoenix, but my impression is that it didn’t really happen as predicted most places.
It’s easy to come up with a lot of theories about the effect of coronavirus on murder, but the weird thing is that you can’t see them having much tangible effect through Memorial Day.
Memorial Day, of course, saw the fentanyl-assisted death of George Floyd (purple vertical line). By the time of the riots of the following weekend, the Great Murder Surge of 2020 was underway. The worst day for murders in 2020 prior to Memorial Day was 18, but 37 were murdered on Sunday, May 31, a day of massive Mostly Peaceful Protests. In Chicago alone, 17 were murdered that Sunday while looters ran amok. (My vague impression is that most of the increases in murdering in 2020 were not during riots per se, but due to the police being distracted by the MPP and then discouraged by the sudden decision on the part of influencers that every cop was a criminal and all the sinners saints.)
By Sunday, June 7, 13 days after George Floyd’s death, the difference between 2020 versus the average of 2015-2019 had risen from 14% to 20%.
Then the change in cumulative murders plateaued at this new high level through mid-June.
But a second national surge began about six days after the killing of Rayshard Brooks in Atlanta on June 12 after he stole a cop’s tazer and shot it at him.
By the Fourth of July, murders were up 23%, and by July 21, they were up 27%.
Then murders plateaued again until a third rise began about a week after the police shooting of the armed Jacob Blake in a domestic dispute in Kenosha on August 23.
For these 19 cities for which CCS has day by day data through the end of September, they ended the third quarter cumulatively up 30%.
Nationally, the largest ever one year increase in homicides was murderous 1968, when Martin Luther King’s assassination set off riots, at 12.7% followed by 2015 during the First BLM Era, when murders went up 12.1%. So, 2020 was the worst year for increase in homicides over the 60 years for which we have reliable data.
Here’s another way to graph the data, this time using 7-day moving averages:
I’m using a 7-day moving average of homicides across the 21 cities, with the average centered on the date: e.g., the 7-day moving average for July 4 would be the mean of the homicides counts from July 1 to July 7. The red line represents 2020, the blue line 2015-2019. The salmon-colored area is the difference the two time periods.
As you can see, the winter before lockdowns began on March 17th was modestly more homicidal than the previous half-decade’s average. The first few weeks of the pandemic saw a sharp fall in murders, followed by a big surge around April 7 (concentrated in Chicago, Houston, and Baltimore). Then nothing much happened out of the ordinary happened until late May.
The first weekend of Mostly Peaceful Riots saw a huge surge, especially in Chicago. Then, during the second week of June, the 7-day-moving average briefly fell to be no higher than the previous 5 years average. But then as the Rayshard Brooks Riots spread nationwide and the The Establishment doubled down on the idea that rioters and arrest-resisters were the Good Guys and the forces of rule of law were the Bad Guys, 2020’s murder surge returned. From June 15 through September 30, when the data collection period expires, the seven day moving average of homicides was worse than over the previous half decade in every single day.