Testimony of the International Center for Law & Economics to the General Session of the National Council of Insurance Legislators
Rep. Carter and the Members of NCOIL,
Thank you for inviting me. My name is R.J. Lehmann, and I am the editor-in-chief and a senior fellow with the International Center for Law & Economics. ICLE is a think tank based in Portland, Oregon, dedicated to promoting the law & economics approach to legal analysis, and to issues of public policy more generally.
Some of you may know me from my prior work at the R Street Institute, which I co-founded in 2012. Among the hats I wore at R Street was running the institute’s insurance policy project, and I was the author of the first nine editions of R Street’s annual report card evaluating insurance regulation in the 50 states.
It was actually early in our days at R Street that I first encountered the topic before us today. After the tragic shootings at Sandy Hook Elementary in December 2012, there was a pressing call for new and creative thinking about ways to address the scourge of firearms violence. Being a research center that was, at that point, devoted almost exclusively to insurance issues, we explored whether mandatory insurance could be part of the solution to promote firearms safety, just as mandatory auto insurance has served to promote driving safety and mandatory workers compensation insurance has served to promote workplace safety. So, while I’m about to tell you why I think these mandates are a bad idea, I want to note at the top that I do understand the intuition.
What we concluded, after batting around various iterations of what a mandate might look like, is that it was fundamentally unworkable. That insurance could not possibly respond in the overwhelming number of cases that were of public concern and that in the limited set where it could respond – which is, basically, true accidents that befall third parties – coverage already exists, either through a homeowners policy or a renters policy.
The two central problems that limit the applicability of any firearms-insurance mandate are that intentional acts are uninsurable and that it is the nature of liability insurance that only harms to third parties are covered.
Taking those one at a time, the claim that intentional acts are uninsurable begs two other obvious questions, each of which, unfortunately, can take us down some rather unproductive detours. What does it mean for an act to be intentional and what does it mean for an event to be insurable?
On intentionality, there’s a whole rabbit hole one can head down on free will and determinism and whether all actions are intentional or whether no actions are intentional. This is not a philosophy class, so I’d like to rescue us from that particular rabbit hole.
The question of insurability returns me to a theme I found myself echoing a lot in another recent public policy discussion—which is whether business interruption for pandemics is insurable. What I said then and will say here is that insurability is a spectrum. Things may be more or less insurable, meaning, in a nutshell, that the willingness of capital to participate in risk-transfer solutions for any particular class of event will vary.
The framing that I think is most helpful for these purposes is to say the sorts of events that are most insurable are those that are fortuitious—which is to say, they happen by chance, rather than by design—and where there is a broad alignment between the goals of the insurer and the insured. When I step into my car, I would like to avoid getting into an accident. My insurer would also like me to avoid getting into an accident. If I do nonetheless get into an accident, it’s a fortuitous event. That event is insurable. If, rather than an accident, I willfully try to run someone down on the road, then we’re not aligned. That’s not insurable and claims for vehicular homicide are excluded—even though, in some places and some cases, the insurer may still be required by a judge or jury to pay a claim.
Applying that logic to the example of firearms incidents offers some context for just how many potential claims are excluded the realm of insurability simply from the fact that insurers are not willing to extend coverage to intentional acts. According to the Centers of Disease Control and Prevention, more than 70% of firearms injuries are the result of assaults, while less than 20% are unintentional. Among firearms-related deaths, the National Safety Council finds that 54% are suicides, 43% are homicides, and only about 1% are accidental.
We therefore start with proposition that only about one-fifth of firearms injuries, and only about 1 in 100 firearms deaths, are even potentially insurable. That universe of potentially insurable claims shrinks even further—although the data on this is harder to find—when you consider that it is the nature of liability policies that they only cover injuries to third parties. If a contractor slips and falls on your property, that might be covered under your homeowners insurance policy. If you slip and fall, it will not be. If your dog bites your neighbor, it might be covered. If your dog bites you, it will not be covered.
So, similarly, if there’s a firearms accident in your home and a third party is injured, that might be covered. Indeed, even if the accident is outside your home—say, you’re the vice president of the United States and you accidentally shoot your hunting partner in the face—your homeowners policy very well might cover that.
But the insured in a homeowners policy is the household, not an individual. If one member of your household accidentally shoots another member of your household—even in the very tragic incidents we hear about involving children—that’s not going to be an insured claim.
Another factor that likely shrinks the universe of claims even further is the language of the HO-3 policy itself. The policy has always excluded injuries or property damage that the insured “expected or intended.” But in 2000, the Insurance Services Office actually broadened that exclusion quite a bit, and the standard policy now states that coverage is excluded for an action that is “of a different kind, quality or degree than initially expected or intended” or “is sustained by a different person, entity, real or personal property, than initially expected or intended.” That’s a pretty broad exclusion and courts have tended to read it as covering even negligently careless actions that result in unintentional injuries.
Nonetheless, despite these manifest limitations on what an insurance mandate could possibly cover, we have watched such proposals perennially introduced in various states in the decade since Sandy Hook, with New York and Connecticut being two of the most frequent states where legislation was considered. Until this past year, when the City of San Jose and the State of New Jersey both adopted differing versions of a mandate, they never went anywhere.
But interestingly, in 2018, we saw regulatory action that, rather than mandate liability insurance for gun owners, actually would appear to forbid it, and this contradiction is important and underappreciated in the current discussion.
For a recap, back in 2018, New York State Financial Services Superintendent Maria T. Vullo brought complaints against the broker Lockton, the underwriter Chubb, and the National Rifle Association over their respective roles in administering the Carry Guard insurance program for NRA members. Some of the charges concerned alleged violations of the declinations requirements to place policies in the surplus-lines market and that the NRA was marketing policies as an unlicensed producer. Those violations aren’t of much interest here. But the core charge was that, because Carry Guard would pay legal defense costs for insureds who face civil or criminal charges related to the use of firearms (that is, where the insured pleads innocent, claims self-defense, or asserts that they are not liable in a civil proceeding) the coverage itself was fundamentally contrary to public policy establishing that criminal acts cannot be insured.
Now, as many insurance lawyers in this room could testify, it’s not always quite as simple as that. It is not unusual at all for an insurer in, say, the directors and officers, or errors and omissions, or environmental-liability lines to find themselves on the hook for the defense costs of an insured accused of a criminal act. And where they are adjudicated guilty, the insurer may try to claw back those costs. But until that point, there are fiduciary duties an insurer owes to its policyholders, and refusing to pay defense costs on a liability policy is usually a quick ticket to a bad faith lawsuit.
But more fundamentally, paying defense costs is a if not the fundamental purpose of liability insurance. So, if the Carry Guard program was contrary to public policy, that’s another way to say that liability insurance for firearms is illegal. And the primary reason I think that has to be considered in this discussion is that one of the states that filed follow-on actions in the Carry Guard case was New Jersey. Which suggests the absurd scenario that New Jersey is now requiring a form of insurance that is illegal to sell in New Jersey.
I am not a constitutional lawyer—or any kind of lawyer for that matter—so I’m going to refrain from saying too much about how these mandates would be treated under the rubric the Supreme Court promulgated in last year’s Bruen decision, although I reserve for myself the right to chime in with my amateur opinion if the subject comes up in the Q&A, which I imagine it will. I would recommend a paper by Adam Schniderman of the University of Michigan Law School that I believe is the first to look at the question, and he makes what I think is a compelling case that neither the New Jersey statute nor the San Jose ordinance would survive under Bruen analysis.
But more generally, I think it’s clear that what these proposals seek is a kind of end-run around the Second Amendment; i.e., that you can outsource to the insurance industry, through its underwriting and rate-setting processes, vetting of firearms owners that existing Second Amendment jurisprudence would appear to deny to state and local governments.
There are various problems with this, but one that I think is most important is that it’s grounded on a theory of what insurers would do to manage firearms risk that appears to be fundamentally untrue. In other words, as mentioned, we already have coverage for firearms accidents in homeowners policies. But insurers don’t charge different rates to different homeowners based on their risk of firearms accidents. Based on my understanding, there aren’t even any insurers who ask whether a policy applicant owns a firearm, so it doesn’t even appear in the underwriting side of the equation.
Now, maybe this is because liability is a relatively small part of the risk underwritten in a homeowners policy, and as mentioned, firearms incidents are an even smaller proportion of liability claims. But it should be noted that, even the NRA’s Carry Guard policy—which was a standalone policy for firearms liability—didn’t charge variable rates. It charged a flat fee.
Bespoke, targeted risk-based underwriting is such a ubiquitous part of our modern insurance markets that we sometimes take for granted just how new and novel it is. In auto insurance, it really only dates back to George Joseph’s Mercury General in the 1960s. There have always been underwriting criteria, such as Benjamin Franklin’s Philadelphia Contributionship refusing to insure homes with trees because they were likely to spread fire. But the assumption that, for any given risk, insurers will automatically have and know how to use the relevant data sets to segregate high risk from low risk, is naïve. The use of this data is actually a historical aberration.
Even if insurers do find that data, the variables that provide actuarially credible projections may not be the ones that you assume or hope for. For instance, it may be that the thing that best predicts whether you’re going to have a firearms accident is your income. That sort of correlation is always problematic and controversial, but it should be particularly concerning if what it implicates is a constitutionally protected right.
I look forward to your questions.