One way criminals gets guns is to steal them, or to buy guns from gun thieves.

The question is if this is a major source of crime guns, and where is the leakage occurring?

The answers are serious though the data is convoluted.

Takeaways

• A net 154,746 guns (stolen minus recovered) enter the underground in a year.
• They are overwhelmingly taken from civilians.
• Most disappear from vehicles, not homes.
• Public carry is not a predictor of gun theft rates, but the adult population is.
• Gun theft rates are not covariant with gun homicide rates.

A crabby note about data

Data quality is the bane of research, and data quality on this issue is inconsistent.

We’ll include rants about data as we explain our findings, but for starters, we’ll note that snapshot data (i.e., a one-year picture) is divorced from long-term data, and the long-term data required us to triangulate factors. That latter bit leads to a mild caution about trend charts herein, and a not-mild (borderline screaming) complaint to the FBI for data design, collection and reporting.

The framing questions

We scripted the following framing questions.

• What is and has been the rate of guns thefts?
• How is it split between civilians and federal firearm licensees (FFL), a.k.a. gun dealers?
• Is the situation getting worse, better, or staying steady?
• Does the general degree of gun ownership correlate to gun thefts?
• Does public carry correlate to gun thefts, and if so, how?

We soon discovered that some of these questions produce fuzzy answers, some data is misaligned, and some sub questions (such as the percentage of stolen guns that end up “on the streets”) cannot be accurately estimated.

All that said, some things are clear and clearly not good.

The high-level national view

The Bureau of Alcohol, Tobacco, Firearms and Explosives (ATF) – which would be a good name for a Nevada convenience store – produced a report ¹ on the subject. Though some insights therein are important, they only covered five years, and two of those were during the pandemic when criminology data was wildly fluctuating. The report is also frustrating because they pulled data from the National Crime Information Center (NCIC), which is not readily available to the public, and thus not to Gun Facts (see Righteous Data Rant #1).

However, it gives us a starting point for both the degree of the problem and for sanity-checking other data.

The ATF report breaks down thefts from both civilian and Federal Firearm Licensees (FFLs). The latter are gun dealers ranging from your Uncle Harvey who trades a few pieces a year for profit, to every Bass Pro shop, all the pawn shops, et cetera. Combined, the ATF estimates that every year 211,934 guns get stolen. As we’ll explore in a moment, about 15% of stolen guns are recovered each year, so there is a net leakage of approximately 180,144 guns into criminal hands per the ATF’s snapshot.

That is a lot of guns.

Compare that gun loss rate with the 19,651 gun homicides in 2022. That is more than nine guns entering criminal hands for every gun used in a gun murder. But guns are used for other types of crime (i.e., attempted homicide, armed robbery, etc.), some are ditched after being used in a crime and then replaced, and others are not used. But still, that is a lot of guns entering the underground.

The good news is that the ATF’s private data source roughly agrees with published data from the not-yet-matured National Incident-Based Reporting System (NIBRS), the new scheme replacing the decades-old Uniform Crime Reporting (UCR) system. We used 2022 NIBRS data because they rolled out of NIBRS nationally in 2021 and the reporting compliance rate was not great. But the last year reported in the ATF was 2021, leaving a one-year gap.

That said, the numbers are “close enough.”

NIBRS 2022 reported 165,378 instances of gun thefts and the ATF 2021 report disclosed 159,422. This is a difference of less than 1% and given that the nation was still masking up and locking down in 2021, the non-pandemic related gun theft fluctuations are likely near zero.

However, NIBRS says that in the year 2022, there were 24,098 instances where stolen guns were recovered. The ATF report says that on average, 59,357 stolen guns were recovered. This is about 2.5 guns recovered for every instance of any stolen guns being recovered. This is significantly higher than the ATF’s reported average of 1.3 guns stolen in each gun theft instance. Summarized, police recover more stolen guns per instance than are stolen per instance. This discrepancy is likely due to busts of underground gun marketers and police obtaining small caches of guns when a criminal is charged.

For now, using just NIBRS instance numbers for 2022 and ATF guns/event ratios, we see:

• GUN STOLEN: 214,991
• GUNS RECOVERED: 60,245
• NET GUNS ENTERING UNDERGROUND: 154,746

For perspective’s sake, in 2022 there were 19,651 firearm homicides, ², or about seven guns stolen for every gun homicide. To add even more perspective, sundry local police statements have shown that a street gang member being held on a homicide charge typically is also the lead suspect in two to three other homicides. This resolves into a few thousand super killers. Per the Bureau of Justice Statistics decadal studies on crime gun sources, most of these guns are acquired from “street sources.” ³

But the gap between the number of guns stolen and the maximum number of guns chucked into a river after being used to commit a crime is way off. Perhaps crime guns not used in murder explains the gap. After all, if you wounded but didn’t kill someone, you might be motivated as well to dispose of the crime weapon.

In 2022, there were about 59,000 firearm woundings due to assaults or undetermined intents (though this includes the tiny fraction of gun accidents). Add this to the gun homicides, and we still only get to nearly 79,000 cases where someone was hit with a bullet in a criminal action, which is still well below the net 154,746 guns stolen (actual number stolen less the number recovered). That’s a nearly 2:1 ratio.

Since criminals notoriously decline to describe their activities, we are in the dark as to how many of the stolen guns:

1. Enter underground markets
2. End up in criminal hands
3. Are used in crime, even if not fired
4. Are used in crime and fired
5. Are used in crime, are fired, and someone is wounded or killed

But 154,746 new free-range guns a year is clearly a problem.

One quick aside is that not all stolen guns are acquired by street thugs. There are times where people acquire stolen guns though they are not active criminals (as in this report of an elderly man who shot a burglar, and the old fellow used a stolen gun). It is impossible to know the breakdown of who receives stolen guns, but other research ⁴ has shown the “street price” for a handgun is about 25–50% that of the retail price, which makes it affordable to gang members and disadvantaged seniors alike.

click to enlarge

click to enlarge

click to enlarge

click to enlarge

click to enlarge

click to enlarge

click to enlarge

To complete the high-level one-year snapshot, we need to look at the relative rate of guns “disappearing” (the real “ghost guns”) between civilians and gun stores.

In the five-year ATF report mentioned before, they concluded that only 3.2% of lost and stolen guns left the hands of FFLs. The 2022 data we gathered puts the figure at around 8%, which is a significant difference. Again, the ATF’s report was made using opaque data unavailable to us, and included pandemic-era fluctuations, so we are more content with the pie chart presented here.

State-level view

Keeping in mind that there was still an under-reporting problem with NIBRS for the year 2022, there are some disturbing and illuminating state-level elements in the data.

Foremost, we are concerned with the net number of stolen guns that enter the underground markets. This can be viewed for each state by the ratio of guns that are stolen and the number of stolen guns that are recovered. A lack of policing, which is associated with high gun homicide rates, also likely allows more stolen guns to not be recovered. Illinois (with Chicago’s Cook County, the leader in excess gun homicides) also has the worst stolen/recovered ratio of all states.

But that is not the whole story. If a state has a much higher rate of gun thefts, and an average theft/recovery ratio, then they will have more guns leaking into the wrong hands. For example, Michigan (think Detroit) is about average in the number guns stolen per capita, but their stolen gun recovery ratio is 20% worse than average. Texas has nearly double the number of guns stolen and has even a worse recovery rate (a recovery rate of 8%, compared to a national average of 13%).

Later on, we will get more detailed, but at a high level for states we see no meaningful covariance between rates for stolen guns and either the average gun ownership rates or public carry rates (“shall issue,” permitless carry, or the combination of public carry laws). It is, however, much more associated with how many adults there are in any state. Since you have to be an adult to buy a handgun, and since handguns are the primary choice of gun thieves, the intersections are unsurprising.

Is it getting better or worse?

Neither, though the data is maddening (see Righteous Data Rant #2).

Using the FBI’s data for the value of guns stolen, and adjusting for inflation, we see three interesting things about stolen guns.

• The gun homicide rate was falling as a byproduct of 24 states passing habitual offender laws in the early 1990s, even though the gun theft rate was static.
• For six years after the Great Recession of 2008, gun homicides continued to fall even though the gun theft rate was rising.
• Murder started spiking in 2015 (post Ferguson Michael Brown shooting and early police defunding drives) when the gun theft rate started falling.

The chart above includes both civilian and FFL thefts. Looking at just FLL thefts and losses (this is how the ATF clustered their data) we see a general downward trend in guns leaking into underground markets from gun stores. But recall that depending on which data sources are used, FFL guns account for only 3–8% of guns stolen (and for FFLs, lost). Good that the trend is downward, but the FLL contribution is the minor problem to consider.

The take-away here is that the gun theft rate does not appear to predict the gun homicide rate.

Did expanded public carry induce more gun thefts?

The short answer is “no,” but it really depends on how big of a change in public carry laws a state made.

From 1988 through the 2022 Bruen Supreme Court decision, the US went from 10 states that allowed public carry to 42 states. And by “allow” we mean that any adult not convicted of a felony would be issued a permit upon request, or within the increasing number of states that eliminated permitting all together.

From 1996 through 2020 (see Righteous Data Rant #3) 67% of states saw a net decline in the inflation and population adjusted value of stolen firearms. Only two states showed both an increase in firearm thefts and a significant 24-year covariance. In short, for most states, gun thefts did not get worse as public carry expanded from 10 to 42 states.

But there were hiccups.

How big of a leap a state made seems to matter. There are four modes of public carry legislation:

No Carry: Forget about it. No civilian is authorized to carry.

May Issue: The government may or may not grant anyone a permit to carry. This is often at the whim of local law enforcement and historically has been about as restrictive as No Carry.

Shall Issue: A carry permit will be issued to any adult who wants one providing they have no felony convictions. There may be other requirements (i.e., a written test, life-fire test, etc.).

Permitless: Lacking a felony conviction, any adult can carry a gun in public without the need to obtain a permit.

What we see is when states made a big jump from No Carry to Shall Issue, gun thefts jumped significantly and consistently. However, states that graduated from May Issue to Shall Issue were less at risk for new gun thefts, and states moving from Shall Issue to Permitless even less so.

Incidentally, that big spike on the right is New Hampshire, whose population in 2020 was about the same as Dallas, Texas. Being a small-population state means any smallish blip in the trending data looks like a huge covariance.

Between the long 24-year trends and the spot checks around specific public carry regime changes, no case can be made that expanded public carry, generally speaking, induced more gun thefts. This is somewhat intuitive because criminals have been in the gun theft business since guns were invented. Their MOs were well established before public carry came to their state. Given that the population has grown and the number of handguns in circulation has too, the lack of uptick in gun thefts is a good sign.

Because the FBI did not track either the number of guns stolen or where they were stolen from until 2021 (at least not in public accessible datasets), we don’t know for certain if gun thefts from cars has increased with pubic carry.

This is important because some states and the federal laws disallow carrying guns in certain “sensitive” locations. For example, in most states you can carry a gun in the supermarket, down the sidewalk, in a movie theater, and just about anywhere. But when you enter a post office, you have to leave it in your car.

Also, people are lazy. When we posted some data about car gun thefts, one person on Twitter/X chimed in with “Your car is not a holster!” People get lax and leave guns in cars. Sometimes they forget, sometimes they think “Nobody knows my gun in in the console.”

click to enlarge

click to enlarge

Most of the time (52%), guns stolen from civilians are taken from a residence, for the top 10 locations, which account for 95% of civilian gun thefts. This might be the gun owner’s residence, or it might be someone they are visiting.

Contrast that with about 36% of civilian gun thefts happening in parking lots or in roadway settings. Initially one might think that 36% of gun thefts from cars is pretty bad, but the number is actually 73%.

A theft of a gun from a car at a residence gets logged initially as a residential theft, because your driveway is part of your residence. Fortunately, we have additional data in NIBRS that shows added detail.

About 72% of residential gun thefts are thefts from cars. Again, if you are parked in a friend’s driveway and leave a gun in your car that is then stolen, that is a “residential” theft.

When you combine the residential car gun thefts with parking lot and roadway thefts, the total of civilian gun thefts from cars is about 73%… at least for the top ten location sources (95% of all gun thefts are in that top 10).

Using the ATF’s annualized averaged data, that means civilian car gun thefts account for approximately 149,874 stolen guns. This may well be the core of the supply of underground armament.

Undeniable problem, onery solutions

154,746 guns is a lot of guns, by anyone’s accounting… outside of the U.S. Army.

The data shows the sore points. Handguns stored in cars is the biggest source. But the solutions will be hard and/or contentious.

People who carry have lives just like everyone else. They have to go to the post office like everyone else. But federal law says they have to leave their gun in the car. One aspect of reducing gun thefts is to reduce the number of not-really-sensitive places where you cannot carry a gun down to just those which are actually sensitive (court houses, where people who are already in conflict congregate).

But gun owner laziness has the bigger payoff. The firearm industry and local governments might educate people about the rate of vehicle gun thefts and encourage new gun buyers to not be lazy.

And, of course, more police on the streets will prevent gun thefts, especially if local police analyze their data and identify which places in their cities have car gun thefts happen the most… which might lead to declassifying some “sensitive places.”

Acknowledgements

Special thanks to Gun Facts research volunteer Steve S. for gathering and codifying a set of ATF files on FFL gun thefts.

Also, a special thanks to the crew at Outdoor Analytics who compiled historical gun prices that helped cross-validate our normalization of FBI UCR data about stolen and recovered guns.

Righteous data rants

#1: Non-public data

The National Crime Information Center (NCIC) “is a computerized database of documented criminal justice information available to virtually every law enforcement agency nationwide.”

But not to researchers.

There may be good reason for this as there may be sensitive data about ongoing investigations or personal info about victims therein. But it raises suspicions when researchers cannot verify reporting.

#2: FBI data is ugly

The old FBI UCR system is a bit of a mess. For example, if you use the raw data for homicides, some states will show nearly none. When you dig in, you discover that the number of agencies not reporting their crime data at all to the FBI is extreme. Both the FBI and the criminology data warehouse NACJD interpolate the missing data to produce the annual crime stats you read.

For gun thefts, data quality sank into the realm of ridiculous.

The FBI collected not the number of guns stolen for each instance of gun thefts, but the subjective estimate of the value of the property taken. Unless the officer creating the report was an expert in retail and resell values for a wide array of firearms, they likely either looked it up in a valuation book (which might be wrong, outdated, unadjusted for condition, etc.) or made a guess.

#3: Even the archives are crappy

While loading annual FBI data into our databases, we notice a sudden 40,000% drop in the reported value of stolen firearms in the 1995 data. We documented this for NACJD, the archivist of crime data. They reviewed the situation and said (paraphrased) “Yep. Pre-1996 data has a data anomaly. We are unsure what happened, but we’ll look into it.”

This is why our analysis does not go back further than 1996. The data is not available, and there is no telling when it might be.