Comment: What we should take from decline in crime stats
Published 1:30 am Monday, September 1, 2025
By Justin Fox / Bloomberg Opinion
The U.S. Department of Justice has reportedly launched an investigation into whether the Washington, D.C., police have been manipulating data to show crime falling in the nation’s capital.
I can’t say this with 100% confidence because there have been cases in the past of police manipulating data to reduce crime rates, but I’m pretty sure this “investigation” will follow the path of most such efforts undertaken during the second presidency of Donald Trump and fizzle out once the president’s attention has moved elsewhere. Also, while I normally hesitate to pretend to know the motivations of others, I think that by now it is clear that President Trump does not care whether government statistics are accurate, and if Washington crime rates continue to drop in coming months, as they probably will, he will trumpet this as evidence of the success of his partial takeover of city law enforcement.
With all that out of the way, though, the question remains: Are the statistics from D.C. and other big cities that have showed sharp drops in crime last year and this year really telling the truth? The short answer is yes, mostly, probably. A longer one is that you really should be skeptical of crime statistics, but that’s not the same as completely discounting them.
Questioning the accuracy of reported crime statistics is a time-honored and bipartisan tradition in the U.S., as it should be. There are opportunities at every level — from people reporting crimes to police, to police recording them, to police releasing the statistics, to the system by which these are conveyed to the keepers of national crime data at the FBI — for inconsistency and error. For example, the FBI and local police departments bungled the transition to a new crime-reporting system in 2021 just as a pandemic-era crime wave seemed to be gaining strength, making the national statistics for that year unreliable and those for subsequent years still somewhat suspect.
On the other hand, the U.S. crime-reporting system is quite decentralized, making it hard to orchestrate any kind of national cover-up, and has become more transparent in recent years. Some big cities post incident-level crime data online — making misreporting or manipulation easier to discover than it used to be — and for some crimes there are alternative, non-police data sources that allow for reality checks. These mostly back up the story told by police data, but not entirely.
The most important alternative data source is the National Crime Victimization Survey conducted by the U.S. Census Bureau on behalf of the Bureau of Justice Statistics. Each year, surveyors contact 150,000 households to ask about crime victimization. These surveys show an even sharper decline in violent crime since the early 1990s than the FBI’s statistics (the same is true with property crime) but some movements in recent years that aren’t reflected in the FBI’s crime reports.
The disagreement between the two sources of crime statistics made headlines in autumn 2023, when the BJS survey showed a 42% increase in violent victimization in 2022 while the FBI’s annual crime report showed a 1.7% decline in violent crime. The FBI has since revised that to a 5% increase and also warns that the 2021 numbers from which the increase is calculated “are below a statistically acceptable level to be nationally representative and are not comparable to other yearly estimates,” so I think the BJS wins that internal fight.
However, the sharp decline in violent victimization in the 2020 BJS survey, without which the 2022 increase would have been much less striking, may have had something to do with the large share of Americans who stayed home for most of the year because of covid-19. Compare 2022 with 2019, and the BJS violent crime increase wasn’t much steeper than the FBI’s and may not have been an increase at all, given that it was within the survey’s margin of error. With the FBI’s 2024 report already out, showing a 4.5% decrease in violent crime, it will be interesting to see whether the 2024 victimization survey due out next month backs that up.
The victimization survey also asks victims whether they reported the incidents to the police, offering another gauge of how much crime is being missed by police statistics and whether that’s been getting worse over time. White House Deputy Chief of Staff Stephen Miller claimed recently that “real rates of crime, chaos & dysfunction are orders of magnitude higher” in U.S. cities than crime statistics show. Going by the reporting rates from the BJS survey, “real rates of crime” appear to be about two times higher for violent crime and three times higher for property crime; significant, but a lot less than Miller’s purported disparity of 20 times or more.
The reporting rate for violent crime also shows no clear trend: It has declined since the early 2000s but rebounded in recent years and is higher than in the 1990s. The property victimization rate is definitely headed downward, but when you look under the hood, recent declines are almost all about burglary, which could be because residential burglary is a disappearing art in the U.S. Apart from a few high-end specialists who target, say, professional athletes on road trips, it just doesn’t pay to be a burglar anymore, so not only has the burglary rate plummeted in both victimization surveys and police data, but I suspect that a growing share of burglaries are amateurish jobs where so little of value is taken that calling the police is more trouble than it’s worth.
One thing the victimization surveys don’t provide is timely sub-annual information. The FBI hasn’t been great on that either, although the new National Incident Based Reporting System that caused all those problems in 2021 is now allowing it to providing monthly crime data. The statistics for recent months are woefully incomplete, though, warns crime analyst Jeff Asher, and should be used with caution.
In the meantime, many cities publicly report at least some crime statistics weekly — Washington does it daily — and the Real-Time Crime Index that Asher’s team at AH Datalytics assembled last year with backing from Arnold Ventures, a philanthropy that funds much criminal justice research, now provides a monthly view of crime in 400-odd cities accounting for almost a third of U.S. population. One requirement for inclusion in the index is the availability of consistent data back to 2018, meaning that the index provides the clearest view available of when violent crime actually peaked during the pandemic, which appears to have been the summer of 2022.
The data here make something else clear, which is that violent crime is mainly robbery and aggravated assault, with murder and rape having little effect on overall violent crime rates. I have in the past tried adjusting the numbers using the Canadian system of assigning greater weights to more serious crimes, but it didn’t make enough difference to be worth the continued effort.
This is one of several reasons it’s usually worth considering the gravest of crimes, murder, separately. Others are that murders are much less likely to go unreported than nonfatal crimes, plus there’s a mandatory reporting system separate from the voluntary police-FBI one that routes data on homicides and other deaths from funeral homes to the Centers for Disease Control and Prevention’s National Center for Vital Statistics. Homicide in the CDC statistics is not exactly the same as murder and manslaughter in the FBI’s; the biggest difference being that the CDC includes negligent manslaughter, which covers many traffic deaths. The national homicide rates calculated by the CDC have thus always been higher, and the gap has grown somewhat in recent years. But the two measures paint a similar picture through 2024 of homicide rates that are much lower than in the mid-1960s through mid-1990s and seemingly headed back toward pre-pandemic levels.
Continuing holes in the FBI’s data explain some of the recent gap. The 2021 crime report was based on statistics from law-enforcement agencies covering just 65% of U.S. population, and while the FBI attempts to adjust for missing information in its annual national crime estimates, it’s hard to get the adjustment right when so much information is missing. For the 2024 report, the national population coverage was back up to 96%, but one of the states with the lowest coverage percentages in 2024, at just 57% of population, was the one that according to the CDC had the highest homicide rate, Mississippi. The state’s capital and largest city, Jackson, has probably been the U.S. city with the highest murder rate over the past few years, but the FBI’s most recent crime statistics from there date to November 2020. Another possible explanation for some of the recent growth in the gap is the rise in traffic fatalities that began in 2015, as traffic deaths classified as negligent homicides aren’t included in the FBI’s totals.
To bring things back to Washington, a very large gap opened up there from mid-2021 through mid-2023 between police- and CDC-reported homicide numbers. DC’s homicide totals exclude negligent vehicular homicide, which may explain some of the difference but surely not all. Whatever the cause, the gap has since shrunk but remains larger than before the pandemic.
Crime analyst Asher has also found disparities between D.C.’s publicly reported numbers on certain crimes (robbery and assault, mainly) and those it reports to the FBI, but no indication that these were deliberate. Crime statistics in the U.S. are far from perfect. But when they’re almost all pointing in the same direction, as has been the case lately, they’re probably telling us something real.
Justin Fox is a Bloomberg Opinion columnist covering business, economics and other topics involving charts. A former editorial director of the Harvard Business Review, he is author of “The Myth of the Rational Market.”
