$100 Billion Mistake
Loose factoids sink books.

By Dave Kopel, research director Independence Institute & Paul Blackman, research coordinator, National Rifle Association. The authors are speaking only for themselves, and not for any organization.
August 25-26, 2001

Gun Violence: The Real Costs By Philip J. Cook and Jens Ludwig (Oxford University Press, 242 pages, $25)

Printer-Friendly E-mail a Friend

id you know that gun violence costs American one hundred billion dollars a year? Most other people don't either. That wasn't the way things were supposed to happen. Gun Violence: The Real Costs was supposed to introduce a major new factoid into the American gun debate.

It's just as well, though, that the factoid never got off the ground, since it's based on some awfully shaky assumptions. And it's too bad that the book Gun Violence put so much emphasis on creating the $100 billion factoid, because the factoid dominates the book, and thus distracts attention from the book's genuine contribution to our knowledge about guns.

Professor Philip Cook, of Duke, and Prof. Jens Ludwig, of Georgetown, are the two best social scientists who are studying the gun issue from a pro-control perspective. In contrast to many of the "public health" antigun authors, Cook and Ludwig don't produce junk science.

And they're not afraid to report their results, no matter what they show. Last summer, Cook and Ludwig published their study in the Journal of the American Medical Association showing that the Brady Act had no impact on homicide rates.

A few years before, Cook and Ludwig conducted a study of defensive gun use (DGUs) for the Police Foundation (a think tank for police chiefs). Cook and Ludwig dutifully reported the results, finding over three million defensive uses annually. They did announce, though, that they had concluded that it was impossible to measure DGUs accurately, and that the true DGU number was not important to gun policy — a reversal of a position Cook had previously held.

In Gun Violence: The Real Costs, Cook and Ludwig set out to determine how much guns harm society, financially speaking. The book is purely a look at costs of "gun violence" not at cost/benefits of gun ownership. Thus, the authors make no attempt to quantify how much the firearms manufacturing industry helps the balance of payments, or how much hunting contributes to the economies of rural communities.

The authors do, briefly, acknowledge that guns could help financially by thwarting or deterring crime. But they quickly brush past any attempt to quantify self-defense benefits, because the data are supposedly insufficient.

In the "public health" literature, previous authors have offered two major claims about the costs of guns: First, the direct medical costs of treating gunshot victims. Second, the lost lifetime earnings of people who are killed with guns.

Cook and Ludwig analyze both subjects carefully, and find that earlier articles have grossly over-estimated both costs. The direct medical costs turn out to be very tiny percentage of total medical spending. Intentional self-inflicted gunshot wounds resulted in average medical costs of $5,400 (because most such injuries are instantly effective suicides, and thus cause few medical costs). Gun assault injuries averaged $18,400 in medical costs; accidental shootings averaged $22,400. Cook and Ludwig estimated that there were 113,000 gunshot wounds in 1997, with a total of $1.9 billion in treatment costs over the victims' lives. Sixty percent of these costs are attributable to the 2% of gunshot wounds that cause spinal-cord injury.

Cook and Ludwig also sought to determine the source of payment for these medical costs. It estimated that government (i.e., taxpayers) pays 49 percent of the total cost, private insurance pays for 18 percent and 33 percent comes from other sources such as self-pay. Because many victims of gun injuries are indigent, the government share of medical costs is higher than for medical costs in general.

As for lost lifetime earnings of gunshot victims, previous articles have assumed that gunshot victims are identical to the American (or Canadian) average population, even though a mountain of data shows that gunshot victims are far more likely to be criminals, mentally disordered, unhealthy, reckless, and/or poorly-educated than the general population. Similarly, suicides are much less likely to be economically productive people than the general population. The Cook and Ludwig note that suicides, particularly by the infirm, may well save society money. Because some gunshot victims (like smokers) die prematurely, they do not consume medical services which would otherwise have been spent on them, over the course of their lives.

Accounting for this unfortunate form of cost savings reduces the net medical cost of gun injuries by about half.

While the earning loss may be catastrophic to the victim's family, there may be no net economic effect on society. Cook and Ludwig explain that a person's economic benefit to society is the excess of his production over his consumption. Given the lower socioeconomic status of many firearms victims, the aggregate impact on society of lost economic productivity appears to be small. Depending on assumptions about worker replacement via immigration, there may be no net impact at all. Thus, Cook and Ludwig suggest that the "lost lifetime earnings" figure may, in the aggregate, be zero. (Yes, there's more to life than economics, but this is a book about economics.)

It is in this area of the book the Cook and Ludwig make a contribution to the gun debate, providing useful information and tables on the nature of gun-related violence, its perpetrators and victims. Although previous researchers have attempted to estimate gunshot costs, the Cook and Ludwig study appears to be more reliable because it employed the most up-to-date sources available and larger samples of patients.

Next, Cook and Ludwig address the costs of avoiding gun violence — such as airport security systems, longer commutes from people who flee urban cores because of violence, window bars, and the like. The authors admit that their estimates here are somewhat speculative, but they are not implausible.

So far, we have total "gun violence" costs in the annual range of about $5-10 billion. So how do we get $100 billion in "real costs" of gun violence? Here's where the book goes off-track.

Cook and Ludwig turn to "contingent valuation" — using an opinion poll to ask how much people would pay to reduce gun violence. It's ironic that Cook and Ludwig rely on this single poll to prove so much — since Cook had previously argued that we should not rely on the fifteen different polls (including the poll that Cook and Ludwig conducted) which all suggest that the number of annual defensive gun uses is at least half a million. But Gun Violence hinges on a single survey's single question about hypothetical support for higher taxes to reduce gun violence.

In a wide-ranging poll conducted by Johns Hopkins and the National Opinion Research Center, Cook and Ludwig convinced the pollsters to ask respondents an additional question:

"Suppose that you were asked to vote for or against a new program in your state to reduce gun thefts and illegal gun dealers. This program would make it more difficult for criminals and delinquents to obtain guns. It would reduce gun injuries by about 30% but taxes would have to be increased to pay for it."

Cook and Ludwig then figure that if people would pay X to reduce gun violence 30%, then they would pay 100/30 of X to reduce gun violence 100%. Add other estimates for suicide and accident costs, and we get the $100 billion "cost" of gun violence.

Before asking people about how much they would be willing to pay to reduce gun violence, the Johns Hopkins survey warmed them up with some questions focusing attention on the amount and nature of gun violence. Of course warm-up questions are likely to get the respondent in a mindset of wanting to show the pollster that they care about the problem, and are thus likely to make respondents claim they are willing to outspend the House Democrats to "do something" about gun violence, endangered sea turtles, AIDS, cancer, drunk driving, carnival-ride safety, or almost any other good cause.

Cook and Ludwig reject all the surveys finding huge numbers of defensive gun uses, because Cook and Ludwig think that since defensive gun use is a socially acceptable act, respondents will invent DGUs. Yet Cook and Ludwig forget that a willingness to pay to reduce gun violence is also something that decent folks might feel obligated to tell a pollster they support. Would a socially minded respondent tell a pollster that she opposes a measure to achieve a substantial reduction in a real problem?

Cook and Ludwig assert that their survey is reliable because respondents are presumed to have acquired information on the risks of gun violence on their own, making them less dependent on the survey interviewer for information about baseline risks of gunshot injury. The only problem with that statement is that there is no evidence that such information is known to respondents.

In general, opinion surveys show unrealistically high levels of fear of violence and, for that matter, fear of pretty much any other problem covered by the news media, including air safety, the Y2K computer bug, nuclear war, identity theft, food poisoning, and global warming. The public tends to overestimate virtually any risk that the media harp on year after year.

Thus, the Cook/Ludwig survey may be reflection not of the "real costs" of gun violence, but of the imagined costs created by media sensationalism and hysterical coverage of gun crime. More adolescents die from high school football than from school shootings, but only the latter gets guaranteed front-page national coverage every time it occurs.

Since the survey asked about what people would pay to reduce gun violence in "your community," people who live in more dangerous communities would be expected to be willing to pay more. Since New Orleans has a gun violence problem that is vastly worse than the problem in Colorado Springs, one would expect people in New Orleans to be much more willing to pay for a 30% reduction. But, as Cook and Ludwig admit, willingness to pay is distributed in the survey far more evenly is actual gun violence. This even distribution suggests that people are not, in fact, knowledgeable about the levels of gun violence in their community; but people are instead responding to bogus, media-induced fear. Actual gun violence varies widely from city to city, but media panic mongering about gun violence is constant wherever you go.

Media reporting on crime has no necessary relation to actual crime levels. For example, according to Berkeley Media Studies' report "Off Balance: Youth, Race and Crime in the News," homicides declined 33% from 1990 to 1998, but network news coverage of homicides soared 473%. Thus, even if a particular tax increase did lead to a 30% reduction in gun violence, there is no reason to believe the media panic mongering on guns would decline 30%. Panic mongering might even increase — thus raising the (inaccurate, media-driven) "cost" of gun violence even higher.

Other evidence suggests that "willingness to pay" for additional gun control may not be very great:

1. Opinion polls typically report about 70 percent support for gun registration with no costs, but support falls to the mid-40s if the cost will be several billion dollars.

2. In 1976, a 69% landslide of voters in Massachusetts rejected a measure to confiscate all handguns. Supporters of the confiscation proposal blamed the cost of buying the handguns, which was $40 million. Since Massachusetts has one of the lowest rates of gun ownership in the country, and has been, politically, one of the states most hostile to gun rights, the supporters' theory that costs were the fatal flaw was not implausible. Again, no estimates were made as to what percentage of handgun violence would have been prevented by the confiscation law, but the campaign certainly suggested a large share — probably most — gun violence would be eliminated.

Thus, if the confiscation law failed mainly because of costs, then Massachusetts voters were not willing to spend $40 million to end handgun violence (which constitutes a large majority of total gun violence). Nationally, handgun confiscation could cost $3-4 billion in direct compensation costs; it is reasonable to believe that America as a whole is less supportive gun banning than is Massachusetts. Thus, the Massachusetts referendum (projected nationally) suggests an unwillingness to spend $3-4 billion to end handgun violence. This casts doubt on Cook and Ludwig's claim that the public would be willing to spend $100 billion to end gun violence. Voting to actually raise one's taxes is different from telling a pollster that you support a tax increase.

3. Another real-life test for measuring willingness to pay for gun control is by looking at willingness to contribute to campaigns on the issue. In 1982, California had a "handgun freeze" initiative on the ballot. The measure would have forbidden new handgun purchases, but would have allowed current owners to keep their guns; thus the Massachusetts compensation problem was eliminated. Again, proponents promised that the initiative would drastically reduce gun violence. Based on contributions to a California handgun freeze initiative, Californians were willing to pay about a quarter per person for gun control.

In Colorado in 2000, voters faced an initiative on gun shows. Proponents claimed that very large numbers of crime guns — as much as 70% — came from gun shows, and that the gun show initiative would have prevented the Columbine murders. In-state campaign contributions amounted to less than 20 cents per voting-age Coloradan. (This doesn't count a $12 million contribution from California billionaire.)

Now suppose that, instead of expecting a 70% in reduction in gun crime, Coloradans were expecting only a 2% reduction in gun crime. (Studies show that 2% or less of crime guns come from gun shows.) This assumption makes Colorado voters appear more willing to pay for reducing gun violence, since they were willing to spend 20 cents to get a mere 2% reduction, rather than a 70% reduction. Extrapolating this figure nationwide, we get an actual willingness to pay of $3-4 billion to eliminate all gun violence — again, far below the Cook/Ludwig figure of $100-billion.

4. Another way to measure actual "willingness to pay" is to compare the willingness of people to pay for gun control compared to their willingness to pay for firearms freedom. Presumably, people who contribute to antigun groups believe that gun control works, and the success of these groups will lead to reduced gun violence.

To start with, let us eliminate from both sides of the equation the grants from foundations and bigwigs, and instead focus on ordinary memberships. This removes some money from the NRA side, such as some of the large gifts that help fund the NRA Firearms Museum. This also removes a huge fraction of the money from the antigun groups.

Now there are plenty of people who join the NRA for reasons unrelated to Second Amendment rights. They might join because they are competitive shooters, and the NRA supervises much of the high-level competitive shooting in the U.S. Or they might join because they are police officers, and the NRA conducts extensive training for police.

So let's narrow the focus down to the pro- or antigun lobbying side, for which people's primary motivation is clearly gun rights or gun control. If you compare the total budget of the Brady Campaign (by far the leading antigun lobby) with the budget of the Institute for Legislative Action (the NRA's lobbying arm), which is funded with contributions of ordinary folks, you find the willingness to pay for gun control is rather small, roughly a third or a quarter of the willingness to pay for firearms freedom, counting only specific contributions to ILA.

Neither figure approaches $100 billion; neither figure approaches $100 million. But if willingness to pay is done as a cost-benefit approach, the demonstrated willingness to pay for firearms freedom is at least three times the willingness to pay for gun control.

These figures ignore the two billion or so dollars spent every year actually buying guns and ammunition, which may also be indicative of a willingness to pay.

Finally, whatever the "willingness to pay," quantifying the problem does not solve it, or even hint at a solution. Whatever the solution, it has to affect all guns, since the Gun Violence figure is based on a one-hundred percent reduction, since a smaller reduction still leaves us with the costs of fear — a significant part of the $100 billion.

The survey question was for a proposal which was so vague and innocuous — cracking down on illegal gun dealers, gun theft, and preventing high-risk people from acquiring guns — that it was consistent with policies advocated by the NRA. Cook and Ludwig favor broader gun controls than the NRA does, but respondents could have believed that they were supporting funding for something like Project Exile (the Department of Justice program, strongly supported by the NRA, to enforce existing laws against felons possessing guns).

To the extent that a gun control proposal restricts the freedom of legitimate gun-owners (as opposed to merely harming thieves and delinquents), then the "willingness to pay" equation changes dramatically. Many gun rights supporters would pay in order to prevent the policy from being implemented.

The official purpose of Gun Violence, the authors explain, is to "move gun policy forward in America." To the extent that the authors provide us with high-quality data about real costs of gun violence, they have done so. It is unfortunate that the authors obscured this contribution by attempting to create a sensational hundred billion dollar factoid.