The question of whether armed citizens deter violent crime or exacerbate it has been controversial in academia since at least the mid-1990s—not to mention the debate it continues to fuel in American politics. Conflicting studies have informed polarized lawmakers in the parallel battles over gun regulation.

The case for less restrictive gun laws generally boils down to this: Law-abiding citizens have a right to protect themselves and their communities, full stop. The academic backing for this argument can be traced to a 1997 study by University of Chicago economists John Lott and David Mustard. After analyzing the impact of “right-to-carry” laws, the umbrella term for various legislation that allows citizens to acquire a concealed-carry gun permit, the authors concluded that these regulations were “the most cost-effective method of reducing crime thus far.”

In the roughly two decades since, additional academic studies have strongly suggested that the opposite is true: that these laws lead to higher rates of violent crime. The latest—and, at least according to one of its authors, most comprehensive—was released earlier this month by the non-partisan National Bureau of Economic Research. Like the Lott and Mustard report, the new working paper analyzes the laws’ effect on violent crime rates. But researchers used an unusual method to imagine what crime trends would have looked like in right-to-carry states had they not adopted those policies.

The researchers built fictional, or “synthetic,” states as near-identical counterparts to the 33 that passed right-to-carry laws between 1981 and 2014. Using the states’ crime rates prior to the laws’ adoption, as well as national crime data from before and after, they created an algorithm to estimate what trends would have been prevalent had these areas never passed right-to-carry. The researchers then compared crime in the real states with findings from their synthetic versions.

“Ten years after the adoption of RTC laws, violent crime is estimated to be 13 [percent to] 15 percent higher than it would have been without the RTC law,” the authors concluded. Just five years after, it’s about 7 percent higher. “There is not even the slightest hint in the data that [these] laws reduce violent crime,” they write.

I spoke with one of the lead authors, Stanford Law School economist John J. Donohue III—who personally supports stricter gun-control laws—about the report and its implications. Our conversation has been edited for length and clarity.

Maura Ewing: Could I sum up your new study with the phrase “fewer guns, less crime”?

John J. Donohue III: It’s probably too general a statement. But if states are contemplating, “Should we move to the more permissive regime of allowing citizens to carry concealed handguns?” I do think we can say that it’s not a good idea. That will lead to higher levels of violent crime.

Ewing: Why didn’t you focus on states that don’t require a gun permit at all?

Donohue: I could have done that, although it’s a very recent phenomenon that states have moved in that direction. Usually we’d like to have 10 years of data after a [legislative] change gets made. Most of the states that have eliminated permits have done so just in the past couple of years.

Ewing: So you’re specifically tracking the loosening of gun laws, which is the predominant trend among states?  

Donohue: Indeed. When John Lott wrote his paper, there were just a handful of states that had adopted these permissive [right-to-carry] laws. Over time, more and more states have moved in that direction. Clearly the trend is in the direction that the National Rifle Association wants.

Ewing: Why was your study necessary? A 2004 National Research Council report essentially debunked Lott and Mustard’s claim that these laws reduce violent crime.

Donohue: It did. I was probably the first one, with various co-authors, to come down hard on the Lott study [in previous papers]. The NRC report essentially adopted all of our criticism on the Lott study, but was much more muted in their expression of what they actually thought the effect of these laws was. The authors essentially said: “There are no statistical supports for the claims that Lott has made, but we are not drawing any inference as to whether they are good or bad or crime goes up or down.” So it was essentially a dismantling, but not an affirmative statement.

This paper, I think, will change the literature. It is the most thorough documentation of a violent-crime increase associated with the adoption of these laws.

Ewing: To be clear: It wasn’t a definitive increase—it was a lesser decrease. Is that correct? Broadly speaking, crime went down across the country, but to a lesser degree in states that have a right-to-carry law?

Donohue: It depends on the state. There was a downward trajectory in crime in the ’90s due to a number of other factors. We are documenting how much a difference it made that states had this law.

In this study, we compare what things would have looked like using two different approaches. One is called the panel-data evaluation, which is what Lott initially used [and which compares trends over time]. And the second is this new synthetic-controls approach. What was nice for me is that they both seem to point in the same direction of higher violent crime.

Ewing: The synthetic-controls approach is what differentiated your study from the 2004 NRC study, is that right?

Donohue: One factor [that makes our research different] is that the 2004 study data ended in 2000, so I have 14 more years of data and 11 more states adopting the right-to-carry laws. The other thing is that we used this new statistical technique. No one that I know of has looked at the question of the impact of right-to-carry laws using synthetic controls.

Ewing: So using a synthetic-control methodology, you predicted what a state’s violent crime rate would have looked like if it had not adopted right-to-carry laws?

Donohue: In a sense. We can look at every state that adopted these laws—we know what happened before they were adopted, we know what happened after they were adopted.

Let’s say the state is Texas. The synthetic control will match what happened in Texas prior to its right-to-carry law in 1996, and then we can see what happened in Texas [after the law’s adoption] compared with this synthetic Texas. That difference dictates whether the particular intervention elevates crime or reduces crime.

Ewing: Is the general takeaway that gun owners in these states are more likely to commit crimes because they are allowed to be armed all the time?

Donohue: The one thing that the paper puts most of its focus on is estimating what the net impact is. There could be some beneficial use of these guns, but overall the harm outweighs the benefit. And the harm comes in many different forms.

For example, the Philando Castile case in St. Paul, Minnesota. [After he was stopped by police,] he immediately told the officer that he was a right-to-carry holder and had a gun, which you’re advised to do. And then the officer shot at him seven times. It scares the hell out of people when they think someone has a gun. Obviously, that right-to-carry holder wasn’t doing anything wrong, but he ended up getting killed anyway.

When more people are carrying guns, things can get more heated. There are times in which the gun could be involved in a way that thwarts a crime, but for the same reason that the officer shot Castile, guns tend to escalate the situation.

The NRA offers a very simplistic view to the public in the way in which the world works, which is: There are all these bad guys out there, but now we’re going to give you a gun, and that means you’re going to be able to be the good guy who saves your life and the lives of other people.

But [with more] people carrying around guns—they’re going to be losing them, they’re going to be stolen, there are going to be more criminals with guns, and the criminals are more likely to carry guns because they know there are guns out there. For a whole array of reasons, more concealed-gun-carrying outside the home pushes up violent crime.