It's Early, But Murder Is Falling Even Faster So Far In 2024

Interpreting what an early decline in murder might mean.

This post needs to begin with an obligatory acknowledgment that it's early April. Any assessment of 2024 data at this point is, at best, an interpretation of imperfect and incomplete evidence that is could change over the course of a full year.

Caveat provided, murder is down around 20 percent in 2024 in more than 180 cities with available data this year compared to a comparable timeframe last year (as of the moment of this piece's publication). Murder is down 20.5 percent in 183 cities with available data through at least January, down 20.2 percent in 174 cities with data through at least February, and down 20.8 percent in 59 cities with data through at least March 20.

Analyzing a large sample of cities is an effective means of figuring out what is happening nationally, but that does not mean that murder is down 20 percent nationally in 2024. We could still see, and perhaps should expect to see the sample's murder decline to regress towards a more normal rate of decline as the year goes on. It's only April and there is a ton of time left in 2024 for these figures to regress, but murder is down roughly twice as much with a sample that’s twice as large as what we had last year at this time.

To recap how we got here: murder rose dramatically in 2020, a roughly 30 percent increase that was by far the largest one-year increase ever recorded. The nation's murder rate was roughly even or slightly higher relative to 2020 in 2021 before falling about 6 percent in 2022. The available evidence points to a larger — potentially even historically large — decline in 2023 though a more exact figure won't be known until October.

A murder decline of even half the magnitude suggested by the early 2024 data would place the US murder rate this year largely on par with or below where it was from 2015 to 2019 prior to the surge in murder in 2020. It’s far too early to guesstimate what that number for 2024 might be though.

Still, the figures from cities with available data so far this year are encouraging — or at least as encouraging as any data on hundreds or thousands of tragedies can be. Right now there are 16 cities that are down double-digits in murders this year compared to the same timeframe last year while there is just one city (Baton Rouge) with available data that is reporting a double-digit increase so far.

Murder is down more than 30 percent at the moment in Washington DC, New Orleans, Las Vegas, Cleveland, Milwaukee, Detroit, Columbus¹, Nashville, Philadelphia, and I could keep going. There are 18 cities with available 2024 data that reported 30 or more murders through February 2023…murder is down as of now in 16 of them. Murder is down in 22 of the 26 cities that reported 20 or more murders through February 2023 and it’s down 25 percent or more in 15 of those 26 cities.

Our YTD murder dashboard has now been updated to include 2024 data and hopefully can serve as a resource for following this trend as the year goes on. As a reminder, all data on the dashboard and in this analysis comes from publicly available sources (even if the original sources aren't always easy to find). If you see an error in calculation or know of a better source for an existing or new city, please let me know! The dashboard is regularly updated with links to sourcing so these numbers will change frequently! I’ve recreated the data as of this post’s publication on April 2nd as a snapshot².

Not every city is seeing a decline. Increasing murder has been reported so far in a few cities like Los Angeles³, Atlanta, and Charlotte, three cities where murder fell in 2023 but is up so far in 2024 (among a few others).

A few other cities aren't included in the sample — namely Memphis and Phoenix — because they have not published any aggregated murder data, but incident-level data from each police department shows fewer murders YTD in both place (-11 in Memphis and -16 in Phoenix).

It's not just murder data in cities pointing to a large decline. Shooting data from the Gun Violence Archive shows a decline of around 12 percent in terms of shooting victims through March compared to 2023. This matches the trend of declining shootings in 20 of the 25 cities with available shooting data through at least February this year.

Mass shootings are also down substantially per GVA data. There were 91 mass shooting incidents this year as of March 31, 2024 compared to 131 mass shootings at the same point in 2023 (-31 percent) according to GVA data.

Overall, a sample of this size tends to predict the final change reported by the FBI fairly well though the predictive value of the sample grows much stronger as the year goes on. To show this I grabbed monthly murder data between 2010 and 2019 for many of the cities in our sample as I could which was 117 cities (the other cities were either missing some or all of the monthly data or weren’t included because they were more recent additions).

For each year I calculated what the year end national change in murder was relative to the previous year as well as the percent change in the 117 city sample through March. In January, for example, the percent change in murder across the sample of cities was off by an average of 11 percentage points above or below the final estimated change made by the FBI (say +3 percent in the sample versus -8 percent in the year end report). It's decently accurate in February and March — especially for years with large changes. By the start of summer the average miss was down to around 2 to 3 percentage points and it stays there until the rest of the year.

The sample through March got the direction right in all but one year (2019 when it fell nationally a miniscule amount) and was spot on a couple of times while missing by a good bit in 2012 and 2013. A sample of this size did a really good job at predicting the national trend between 2014 and 2019, even this early in the year. The miss was enormous in 2020 because murder had not yet spiked in March that year and data for the full sample wasn’t available for 2021 because of the NIBRS switch (making 2022 also unusable for this purpose).

Remember that precision from the sample is not the goal at this point, rather the goal is to use the sample to interpret the likely direction and scope of the crime change. The direction so far seems crystal clear from the available data and will become clearer as the year goes along and the available data is joined by other data sources like the FBI’s quarterly data.

I first wrote about declining murder in 2023 on April 12th last year. At that point we had a sample of 73 cities with murder down 9.9 percent in those cities. The assessment of a large decline held up pretty well last year, but that doesn’t mean it has to hold up as well this year.

I also noted last year that we probably should be waiting until hurricane season starts (June 1 for those of you who aren’t traumatized by it annually) before drawing more firm conclusions about national crime trends from a large sample of cities. Just because the recent historical record suggests a decline of this magnitude should hold up through December as a large decline doesn't mean history can't be wrong or something can't happen to reverse the trend like it did in 2020.

Crime tends to trend up or down a few percentage points per year. A 15 or 20 percent annual decline in murder nationally seems fantastical, but so too did a 30 percent rise before 2020 happened. No team had ever blown a 28-3 lead in the Super Bowl until that one time when a team did.

The early data establishes another potentially large decline in 2024 as an important trend to follow as the year goes on. If the city sample is down like this in June then a massive decline in murder in 2024 will feel substantially more possible than it does now. But the city sample at this point suggests that the decline in murder that began in 2022 is continuing and even potentially accelerating in 2024. We'll see if it ho

1. Columbus is a bit tough because you have to add up how many homicides have occurred through each year based on the file number. There were 39 homicides through March 2023 based on that methodology compared to 17 through March 2024 (see here). Not perfect, but good enough!

2. Baltimore’s YTD total originally said the date was as of February 24. It is through March 28 and has been corrected now.

3. LAPD publishes a handy PDF of crime stats but it hasn’t been available for the last few weeks. I’m using their RMS data here through 3/18 which largely matches their offense totals to ensure such a big city is included. LAPD also includes shooting totals on the PDF but I don’t have any way of recreating them for the below table.