FIREARMS COSTS, FIREARMS BENEFITS AND THE LIMITS OF KNOWLEDGE
Daniel D. Polsby *

America’s intensifying dismay about violent crime has become so pervasive that one may well affirm that there is something of a “national crime crisis.” Yet there is something of a puzzle as well. Overall crime rates in the United States have been falling for nearly twenty years. Violent crime, declining on a national basis for the last three years, has not changed dramatically since 1980, especially in comparison to the startling run-up in serious crime that coincided with the maturation of the post-war birth cohort. The homicide rate has fluctuated to some extent, but despite recent increases it is still below the levels of the late 1970s and indeed, below the rates recorded though most of the 1920s.

To some extent the growth of public apprehension concerning violent crime can be explained by its cumulative nature: “[w]e experience the crime wave not as separate moments in time but as one long descending night.” [1] When serious crime touches oneself or one’s family, it is an event that is more or less present throughout one’s life. The direction of crime rates should be less important, therefore, than changes in the number of people whose lives have been touched by crime. This number may constantly increase through a generation or more though the crime rate falls. It should be obvious, however, that cumulative enlargement of the circle of people who have been victimized by crime can be at best an incomplete explanation for the change in public attitude that is taking place. Public attitudes about crime have changed much more rapidly than the size of its population of victims. “The crime crisis” is a crisis of confidence in the ability of the public sector to address the crime problem constructively. As such it is very much a part of the tide of skepticism about the role of government [Page 208] that has been an expanding feature of partisan political discourse in recent years. Liberalized carry concealed laws are essentially a response to intensifying doubt about the capacity of government–the police, the courts, and the corrections system–to deliver adequate levels of public or personal security. Serious questions remain, however, concerning the ability of private sector practices to deliver the goods where the public sector has failed.

Because the techniques of social science are clumsy, the information generated is often nebulous and hard to interpret. Seldom do social researchers have the luxury enjoyed by Rutherford or Michaelson, of performing a crucial experiment and then proclaiming that a definite increment in human understanding has been attained. Social scientists must sort through literatures filled with hints and intimations in order to get an idea of what is definitely known, and even then gains in knowledge are most often of the null variety, as we fail to find good evidence to support a hypothesis. The problem is ingrained in firearms research, because few if any criminologists believe that guns are the sole factor promoting anti-social behavior. It is not guns themselves, but guns plus additional variables, that lead to trouble. Getting much beyond that generality has been frustrating, and has often seemed something akin to biologists investigating microbes with binoculars. Firearms effects on crime rates–what good do guns do and what harm do they do–seem to be at most quite marginal, and of course available investigative techniques are of seriously limited power. Still, the game is worth the candle if undertaken with proper circumspection, for though accessions to knowledge from any given study may be small, the matter ultimately under scrutiny, that of personal and collective security, is of perennial concern.

In recent years a number of states, including Alaska, Arizona, Florida, Georgia, Idaho, Mississippi, Montana, Oregon, Pennsylvania, Tennessee, Texas, Virginia, West Virginia, and Wyoming, have relaxed their laws regulating civilians’ carrying of concealed firearms, thus joining Connecticut, Indiana, Vermont, and Washington in adopting regimes significantly more permissive than those typical in the rest of the country. A number of other states currently have similar modifications under consideration. The two questions that these amendments beg–and indeed that they may eventually help to answer–are: (1) whether widely permitted civilian handgun carriage risks turning every argument between strangers into a wild west shoot-’em-up, or, conversely; (2) whether increasing the prevalence of concealed handguns drives the crime rate down. These are the questions David McDowall, Colin Loftin, and Brian Wiersema seek to measure with interrupted time series analysis, looking at the experience in the largest [Page 209] urban areas of Florida (Miami, Jacksonville, and Tampa), Mississippi (Jackson) and Oregon (Portland). What they found was statistically significant enlargements in firearms homicides in three of the five cities and insignificant changes in Portland and Miami.

In order properly to evaluate the findings of studies like this one, it is useful to bear in mind the theoretical reasons that might be adduced either to believe or to doubt whether liberalizing civilian access to firearms actually will enhance either the general security of the public or (what is a distinct question) the private security of the person who arms himself. The argument in favor of liberalized gun laws would point to their usefulness as a means of deterring attackers. There is no a priori reason to believe that firearms should be any less useful to civilians, at least those properly trained to use them, than to police officers. Open carrying of a side arm tends to create a private security good (i.e., by “hardening” a particular target), whereas concealed carrying, if it is believed to be reasonably widespread, should tend to create a public security good because it will not be evident to a predator which potential victims or bystanders might have the means to resist attack.

There are also theoretical reasons for skepticism. One is that if an argument blows up between two people, resort to a handgun would confer what is sometimes called a “first mover” advantage. Hostile confrontations between latent antagonists, each of whom estimates that the other is (with some probability) armed, may catastrophically degenerate into gunplay as each recognizes the advantage of beating the other to the draw and the detriment of being beaten. Environments in which “first movers” possess a strategic edge–what in international arms reductions talks would be called a “first strike capability,” are well understood to be intrinsically unstable. [2] Another theoretical reason that points in the same direction was suggested separately by Frank Zimring [3] and Philip Cook [4], who explained why higher levels of civilian armaments (what is called in jargon “gun density”) would be associated with higher levels of homicide. This theory is based on the observation that many homicidal assaults are not accompanied by a specific intention to kill but rather [Page 210] are mercurial outbursts whose lethality will depend on the virulence of the weapons at hand. Guns are much more lethal, wound for wound, than other weapons. [5] When the ratio of firearms to non-firearms weapons increases, one should expect to see increases in the rate of homicide, and conversely, holding all else constant.

McDowall, Loftin, and Wiersema do not actually test the Zimring-Cook hypothesis, because their study focuses on the relationship between the liberalization of carry concealed weapons laws and, not murder rates, but firearms murder rates. Hence, their study does not measure the size of the margin in which homicidally tending persons will substitute non-firearms weapons when firearms become more difficult to get. But it does reach conclusions consistent with those in a number of studies over the past decade that find connections between firearms laws and rates of homicide and suicide. The most widely publicized contributions have appeared in the New England Journal of Medicine, authored by John Sloan and collaborators, [6] Arthur Kellermann and collaborators, [7] and Loftin, McDowall, and Wiersema themselves. [8] Taken together, these papers have affected to establish not only that restricting civilian access to firearms saves lives (i.e., is a sound public health measure analogous to abating a disease vector from the environment), but also–an especial theme of Dr. Kellermann’s–that firearms do not increase the personal security of those who possess them but, on the contrary, actually constitute a “risk factor” for becoming a homicide victim. Such results have received wide media coverage and have considerably gratified those who disdain firearms, the “gun culture,” and the people and artifacts taken to be associated with them. However, the actual results of this research look far more modest if one actually reads the text of the monographs rather than the many press notices that accompanied their publication. As a threshold point, it is seriously debatable whether either the Sloan or the Kellermann results should be credited at all, because the data on which their work rests was neither deposited with the New England Journal nor otherwise made available to independent [Page 211] researchers. [9]

Even if one takes the papers at face value, however, they do not make a trustworthy case either for the Kellermann et al. “victimization-risk factor” hypothesis or for the Sloan et al. “firearms density relates to homicide” hypothesis. Kellermann et al. should not have been surprised to find that homicide victims are armed in disproportionate numbers: a large and growing proportion of victims are criminals themselves; [10] it is rational (if not usually legal) for people who might be embarrassed to call upon official police services to arm themselves, especially if, as is true of the narcotics demi-monde, the risk of being killed by competitors, creditors and others in the business are great [11] and no doubt known to be great. It is unpersuasive to maintain that homicide victimization follows handgun ownership in a causal sequence, as though possession of a weapon could somehow magnetize murderers to ones’ doorstoop. Sloan et al. made a great deal out of differences in firearms murder and suicide rates in the assertedly matched cities of Seattle, Washington, with its relatively slack gun control laws, and Vancouver, B.C., with its relatively strict laws. Among several important questions discounted by this methodology is whether and to what extent the state of the law is a satisfactory proxy for the actual density of firearms. There are, of course, large conceptual and practical distinctions between “how has the law changed” and “how has gun-possession behavior changed.” The second of these questions cannot be answered by assumption rather than with data. At least one researcher, moreover, has expressed doubts that the assumption is correct. [12]

It is noteworthy that the authors of Concealed Firearms Laws stop short of making the claim that relaxing concealed carry laws “will” lead to higher rates of firearms homicides. They do, however, end [Page 212] their paper with a warning against repealing restrictive carry concealed laws. The benefit of this form of gun control is discreetly characterized as the weaker of two conclusions that emerge from the analysis (the stronger being that liberalized concealed carry laws do not have a measurable crime-fighting effect), but the result can hardly be thought a strong foundation upon which to build public policy.

The issue, in brief, is what is called the specification problem, a chronic pitfall of time series studies. It is often difficult to know how to interpret findings when the direction taken by a dependent variable (e.g., the homicide rate) exhibits great sensitivity to the time period selected for examination. [13] McDowall, Loftin, and Wiersema’s own previous work on the 1976 District of Columbia gun control law illustrates the objection clearly. [14] The authors’ conclusion there–that the gun control law prevented an average of forty-seven deaths per year from homicides and suicides–seems doubtful because not only did they fail to give a measure of firearms density that was independent of the ordinance itself, but, to their professed surprise, the numbers seemed to them to indicate that the law had operated virtually instantaneously to suppress the rate of homicides by firearms. In other words, their study finds a beneficial effect from simply implementing the law, that shows up in the homicide statistics before it could possibly have had an appreciable effect on the actual distribution of firearms in the jurisdiction. It is not clear why this should be the case, but the effect is plain to see in the data. [15]

Or is it? In the D.C. study, for example, the authors looked at a number of years in which homicides and suicides by firearms were generally declining. If different periods had been selected for measurement–the D.C. study stopped in 1988–it would have been necessary to explain the long-term association between restrictive gun control laws and increased homicide rates as the District’s numbers, both for homicide and firearms homicide, climbed steadily higher in the late 1980′s. Currently these rates much exceed those in any state notwithstanding the District’s more stringent gun control laws. [16] Or, [Page 213] had one chosen to measure firearms homicides from mid-1974 through mid- 1980–a period in which the rates were trending lower (though with plenty of month to month variance)–it might have seemed more natural to associate the decline in the murder rate with some event more nearly coincident with its beginning–for example, the resignation of President Nixon or the election of the 94th Congress–rather than with a variable like the gun control law, that commenced some two years after the decline in the murder rate had begun. The null hypothesis actually fits the facts rather exactly: the murder rate in a given jurisdiction and the state of its statute law relating to firearms possession have little or nothing to do with one another.

“Easing Concealed Firearms Laws” exhibits specification difficulties similar to those of the earlier effort. For example, in four of the five cities studied, 1973 was used as the baseline for beginning the study, but in Miami, 1983 was used as the baseline because “of an unusually sharp increase in homicide rates in May 1980 after an influx of Cuban refugees. In late 1982 the rates appeared to stabilize.” [17] Had Miami been treated uniformly with the other jurisdictions studied rather than given its own time period, evidently it would have exhibited a statistically significant decrease in homicides instead of no significant change. The authors might justify this special treatment as an appropriate effort at noise reduction in the data, but because the outcome in this sort of inquiry is so sensitive to changes in the time period being studied, the exercise is bound to awaken qualms in the minds of skeptics, and to further weaken a conclusion that the authors concede to be weak to begin with. This is not at all meant as a reproach to the authors, but rather a cautionary appeal to those unaccustomed to the limitations of this kind of research about the quality of information that it is capable of generating. [18] It is often difficult to tell the difference between a spurious and a meaningful association [Page 214] among data.

There is, however, at least one reason that is not indwelling in methodology to be restrained about basing policy conclusions on “Easing Concealed Firearms Laws.” The paper selects for study three cities in Florida–but why so? The change in the concealed carry law, after all, was statewide, and during the period studied, the murder rate for the state as a whole declined. [19] If it increased somewhat in two of the state’s most populous cities and held constant in a third, it must have decreased considerably elsewhere in the state. Is it the case that concealed carry laws “caused” matters to get worse in some places while “causing” them to get better in others? Here one must punt. It is not impossible for one and the same variable to produce a certain effect in some circumstances and the opposite effect in other circumstances. But it is just as plausible to think that the independent variable (that is, the change in the state’s concealed carry law) was actually “causing” nothing as to think that it was causing both of two antithetical effects at once.

There is one quite startling result from the Florida experience that is given little credence by the authors but that deserves some notice. It appears that not a single one of the homicides studied by McDowall, Loftin, and Wiersema was committed by someone who had obtained concealed carry permits under liberalized laws but would have been denied such permits under prior law. In other words, the authors concede that the increase in the population entitled to carry concealed did not contribute directly to any of the excess firearms mortality found by the study. In that case, then, why was there any increased killing at all? Easing Concealed Firearms Laws proposes that “greater tolerance for legal carrying may increase levels of illegal carrying as well. For example, criminals have more reason to carry firearms–and to use them–when their victims might be armed.” [20]

Possibly so–but not obviously so. Indeed, the authors’ reasoning points to the farfetched conclusion that legal carrying of firearms by police officers would increase illegal carrying by criminals. It also implies that increasing the police presence on the streets would lead to higher levels of firearms crime. Surely not. In principle, the expected value of the sort of street crime that is facilitated by brandishing a firearm should decrease as the chances of being shot by a victim, a good Samaritan or a police officer, increases. The use of a firearm confers a decisive tactical advantage on a criminal predator whether [Page 215] his victim is armed or not, but from the predator’s point of view, use of a firearm would undoubtedly have greater net utility in a world in which he had the only gun than it would in a world where some potential targets were secretly armed. Increasing the chances that a predator may encounter armed prey–or may have to deal with an armed Good Samaritan–might very well diminish the value of a firearm to him rather than increase it.

When discussing the effects of concealed carry laws, one should not overlook the evident fact that such laws are capable of putting the Zimring-Cook hypothesis to a practical test. This hypothesis is undoubtedly the most important conjecture in the literature connecting increasing firearms availability with exacerbated rates of mortality from interpersonal violence. Indeed, precisely because the Zimring-Cook argument is a cogent one, it calls for explanation why it is not borne out more convincingly in the data. One would certainly expect that legalizing concealed carriage would increase by some amount–probably a large amount–the number of people who actually do carry firearms on their persons in public. A number of states have had permissive concealed carry laws for several years now, and hundreds of thousands of people have applied for and received legal authority to carry guns in these jurisdictions. There must certainly be, in a sample so large over a period so long, many hundreds of cases (and indeed, many tens of thousands) where people legally carrying guns got drunk, lost their tempers, were in traffic accidents, had domestic quarrels, and in short, experienced all of the psychological sturm und drang which modern life is capable of dishing out–yet so far as a diligent Nexus search discloses, there seems to be not a single case, anywhere in the country, of someone who was legally carrying a concealed handgun using that weapon in a criminal homicide. Thus, the best direct evidence currently available that bears on the Zimring-Cook hypothesis seems to be inconsistent with it.

Perhaps this experience points to nothing more exciting than the fact that the most extreme categories of social violence are still quite rare in the general population. On the other hand, more interesting game may be afoot. Suppose, counterbalancing the Zimring-Cook effect, there were something that might be called the Robert Heinlein effect, after the science-fiction author who coined the aphorism “an armed society is a polite society.” [21] There are plenty of historical examples, [22] and a sophisticated theory to organize them, [23] that demonstrate [Page 216] the ability of heavily armed populations to live together in peace. [24] Furthermore, what if there were also something that might be called the Gary Kleck effect–the principle that while firearms wounds are more serious than those inflicted by other weapons, hostile confrontations that are mediated by firearms are less likely to lead to woundings in the first place–true, whether the person with a gun is a predator or a defender. In this case, the extra lethality of firearms wounds might be negated by the fact that fewer wounds were inflicted in the first place. [25]

The proposition that private firearms may possess some such compensating utility has provoked massive public controversy in recent years. This comes as a result of the fact that the program of the American gun control movement, to restrict private firearms to recreational activities and de- legitimatize them as a means of self-defense, will seem misguided. It is not too grandiose to claim that if one could verify the compensating utility hypothesis, or falsify it, one would possess the passe-partout to national firearms policy. On the firing line of this controversy Kleck and his co-author Marc Gertz find that privately owned firearms are used in as many as two and one-half million annual instances of lawful self defense. This estimate is twenty or thirty times higher than the figure given by the Justice Department’s National Household Victimization Survey, which for most purposes is considered to be the best available measure of American crime statistics. Which estimate is the better? Kleck and Gertz’s argument against too heavy a reliance on survey numbers is sensible enough: in a world in which defending oneself and one’s property is increasingly stigmatized as barbaric–if not tortious or criminal–there is good reason to suppose that survey respondents might not account for themselves honestly, especially not to a government auditor.

To some extent it is an inherent limitation of survey research that one can never be sure that self-reported instances of using firearms in defense of self or property have been correctly or even defensibly tallied by respondents. Nor can one be certain of how many correctly self-reported cases involve criminals defending their contraband inventory from other criminals. [26] Nor can we be confident that cases in [Page 217] which some sort of self defense might be privileged were also cases in which the use of a firearm would be sanctioned. [27] Kleck and Gertz have made every effort to design their survey around the resulting problems, but a more fundamental objection to their project remains. Even if such survey research instruments could be made perfect–so that every respondent accurately self-reported necessitous defensive firearms use–it is questionable what information one would have acquired. It is far from evident that counting self-defensive uses of firearms is a meritorious index of private firearms’ crime-suppressive utility. Surely the real question is how much predatory behavior is discouraged assuming a given diffusion of private firearms. The focus of concern should not be the number of times a firearm was displayed, let alone how often one was discharged, and certainly not how often someone was wounded or killed. What one should really want to know is the number of times overt threats were made unnecessary because the existing distribution of firearms set up an implicit threat potent enough to make explicit threats unnecessary, [28] In other words, one must ask what distribution of private arms is optimal in society. Of course this question presents variables one cannot count (e.g., how many times did a burglary or a rape or a murder not-happen), and asks one to value these non-events in relation to how money might be invested alternatively in firearms in order to produce comparable levels of personal security. As no one can possibly perform this calculation, abstruse even if it were not beset with unknowns, one is consigned to trying to count instances in which “good guys” used guns to run off “bad guys” who were threatening them unlawfully. One is reminded here of that ancient joke about the man who lost his keys in a dark alley but looked for them under a lamp post because there the light was better.

The connection between the data concerning the dispersion of [Page 218] firearms on the one hand, and the homicide statistics on the other is strikingly irregular. For example, we know that firearms ownership correlates positively with household income, [29] and recent figures from a survey of Chicago and its suburbs show a marked inverse correlation between the density of handgun ownership in a given neighborhood and the rates of homicide and other serious crimes. [30] To generalize about the American population as a whole, homicide has been trending downward over the past decade or so; [31] the rebound of the homicide rate to its very high, late-1970s level of nine or greater in 100,000 looks to be essentially explainable by the rapid increase in rates of both commission and victimization of homicide among populations of urban, African-American, teenage boys. It is not widely appreciated that this spike in the statistics dates only from the mid-1980s and is not at all associated either with the relaxation of firearms regulation laws or with increases in the supply or practical availability of firearms. Something changed around 1984 or 1985–but what? Alfred Blumstein pins the rap just where it chiefly belongs, namely, on changes in the retail market for drugs and in law enforcement efforts to control the drug market. Not only did the bull market for crack cocaine greatly multiply the number of illegal retail transactions that occurred (this being the point in the distribution chain that is most vulnerable to police intervention), but it also seems to have involved a new and younger class of distributors, taking advantage, perhaps, of the relative lenity of the juvenile justice system. [32] Juveniles’ debut in this marketplace demanded a gun for protection of self, remittances, and inventory, and for sustaining a reputation that would facilitate treating one’s elders on terms of proper respect. [33] To this parvenu class of criminals, the economic opportunities in the cocaine market were perceived to be greater, and may actually have been greater, than those that were available in the straight world, notwithstanding the enormous risks to life, limb, and liberty that resulted. [34]

With respect to the firearms side of this problem, it cannot be emphasized too strongly that one is dealing with a demand-led rather than a supply-led phenomenon–young men demanding guns as a means of self defense and self- realization. These young men are not merely using [Page 219] guns because large numbers of them are floating around, as mayors and police chiefs insinuate when they tell reporters that “there are too many guns out there.” Recognizing this problem as a demand-side situation predicts the limited usefulness (if not futility) of public policies that seek to “dry up” the supply of guns. The most ludicrous policies of this type are “turn-in-your-gun days” or rules that prohibit police departments from selling surplus weapons. But many kinds of regulatory interventions that place burdens on legal markets embrace the same faulty premise.

Nevertheless, optimists remain who believe that supply-reduction techniques might hold the key to the reduction of gun violence. Among these people, none is more incorrigibly optimistic than Philip Cook. Along with his collaborators Stephanie Molliconi and Thomas Cole, Cook concedes the difficulties of trying to regulate the possession and use of the two-hundred- plus million firearms already in place in the United States. He reasons that because guns used in crimes will tend to have been recently acquired, “[a]n effective transfer-regulating scheme that prevents guns from going to dangerous people would be nearly as successful [in suppressing the homicide rate] as a much more intrusive scheme targeted at current gun owners.” [35]

Indeed, this premise is almost tautologically true, but the trick is: how does one successfully regulate a market for a commodity for which there exist no satisfactory substitutes and which is demanded by people whose disposition to comply with the law is less than that of average citizens, and whose willingness to take risks is greater? Despite the hopefulness of Cook et al. on this score, it is by no means clear that introducing additional restrictions into (currently legal) firearms markets–cutting down on sales through classified ads or gun shows or the like–will have much effect disarming those prone to firearms abuse. For illustration: it has been widely noticed that the effectiveness of local gun control laws can be defeated by gun runners, who load up on guns in low regulation jurisdictions and sell them–illegally–in high regulation jurisdictions. Several years ago it was widely noticed that many of the guns seized from criminals by Washington, D.C., police had originally been procured in Virginia. Public indignation led to Virginia enacting a law that would allow only one gun per month to be purchased legally by a given individual. [36]

This exercise damaged the Capital’s illicit gun market little if at all; the gun runners simply moved a few states south, to Georgia, [Page 220] where no such rationing is practiced. Of course it is easy to turn this observation into a plea for Georgia now to do something (which would move the action to Texas or Arkansas) and thence into a polemic for a national firearms law; but that sort of twist in the argument tends to obscure the source of the difficulty, which does not lie in the disuniformities or inadequacies of various states’ firearms laws but in the fundamental economics of the crime business. Of course gun runners will seek the least cost and most convenient source of supply, whatever it may be, legal markets, if available, but if they cannot deliver what is demanded, the turn to illegal markets, of smuggled guns or guns manufactured in cottage industry, is a simple operation. The acquisition behavior of illicit retail customers should be discouraged modestly at best by piling costs on gun runners. These customers are seeking to invest in capital plant for which there exists no ready substitutes. Licit buyers, on the other hand, usually are shopping for items of personal consumption, for which a number of obvious substitutes (e.g., archery; B-B guns; and for that matter, going to the movies) evidently exist. The implication of this situation, though usually ignored, is very important: the price sensitivity of firearms buyers will diminish as their motive for owning a firearm becomes more sinister. The price sensitivity of buyers will increase as their motive for owning a firearm becomes more innocuous.

The expectation that the sorts of market interventions described by Cook et al. would have a beneficial effect on the homicide rate embeds the assumption of monotonicity, that is, that there are constant returns (in the form of lowered homicide rates) to reductions in the number of firearms in private hands. Those who in any degree credit the possibility of Heinlein or Kleck effects operating, however, and who understand the implication of the distinction between “firearm as capital” and “firearm as toy,” will regard this assumption as rather naive. Such students of the problem will consider the question of how firearms are distributed in society as much more important than how many there are. They will also reject as inherently counter-productive efforts to adopt policies that aim at reducing the number of arms in the hands of criminals by imposing regulatory costs in licit markets. [37]

* Kirkland & Ellis Professor of Law, Northwestern University. Grateful acknowledgement is made to the Kirkland & Ellis Research Fund and the William M. Trumbull Research Fund, each of which partly subsidized the preparation of this paper. The research assistance of James K. Fitzpatrick, M.D., is also acknowledged with thanks. None of the above are to be held responsible for any of the ideas or attempted ideas expressed herein.

[1]. Adam Walinsky, The Crisis of Public Order, 276 ATLANTIC MONTHLY, July 1995, at 41, 44.

[2]. JEROME H. KAHAN, SECURITY IN THE NUCLEAR AGE 273 (1975); HERMAN KAHN, ON THERMONUCLEAR WAR (1969); HERMAN KAHN, THINKING ABOUT THE UNTHINKABLE IN THE 1980′s (1984); THOMAS C. SCHELLING, THE STRATEGY OF CONFLICT 232 (1960); THOMAS C. SCHELLING, ARMS AND INFLUENCE 224-25 (1966); Wolfgang K.H. Panofsky, The Mutual Hostage Relationship Between America and Russia, 52 FOREIGN AFFAIRS 109 (1973); Albert Wohlstetter, The Delicate Balance of Terror, 37 FOREIGN AFFAIRS 211 (1959).

[3]. Frank Zimring, Is Gun Control Likely to Reduce Violent Killings?, 35 U. CHI. L. REV. 721, 735 (1968).

[4]. Philip Cook, The Technology of Personal Violence, 14 CRIME & JUST. 1, 13 (1991).

[5]. GARY KLECK, POINT BLANK: GUNS AND VIOLENCE IN AMERICA 164 (1991).

[6]. John H. Sloan et al., Firearm Regulation and Rates of Suicide, 322 NEW ENG. J. MED. 369, 373 (1990);John H. Sloan et al., Handgun Regulations, Crime, Assaults, and Homicide: A Tale of Two Cities, 319 NEW ENG. J. MED. 1256, 1261 (1988).

[7]. Arthur L. Kellermann et al., Gun Ownership as a Risk Factor for Homicide in the Home, 329 NEW ENG. J. MED. 1084, 1090 (1993); Arthur L. Kellermann & Donald T. Reay, Protection or Peril?: An Analysis of Firearm- Related Deaths in the Home, 314 NEW ENG. J. MED. 1557, 1559 (1986).

[8]. Colin Loftin et al., Effects of Restrictive Licensing of Handguns on Homicide and Suicide in the District of Columbia, 325 NEW ENG. J. MEd. 1615, 1620 (1991).

[9]. Brandon Centerwall notes that Sloan has not made his data available. Brandon Centerwall, Homicide and the Prevalence of Handguns: Canada and the United States, 1976-1980, 134 AM. J. EPIDEMIOLOGY 1245, 1246 (1991). Professor Henry Schaeffer of the North Carolina State University Genetics Department made a telephone request and Professor Lawrence Southwick of the SUNY-Buffalo School of Management Sciences a written request to Dr. Kellermann to inspect the data on which his 1993 study was based. Neither was honored.

[10]. Chicago Police Department reports show that the percentage of Chicago homicide victims with police records has been as high as 65% in recent individual years. See, e.g., DETECTIVE DIVISION, CHICAGO POLICE DEP’T, MURDER ANALYSIS 23 (1992) (65.53% of homicide victims in 1992 had criminal records). In the early 1970s, the corresponding figures were 40%-45%. Id.

[11]. Peter Reuter & Mark A.R. Kleiman, Risks and Prices: An Economic Analysis of Drug Enforcement, 7 CRIME & JUST.: ANN. REV. RES. 289, 305 (1986).

[12]. Robert J. Mundt, Gun Control and Rates of Firearms Violence in Canada and the United States, 32 CAN. J. CRIMINOLOGY 137, 137-54 (1990).

[13]. Gary Kleck et al., The Emperor has No Clothes: Using Interrupted Times Series Design to Evaluate Social Policy Impact (Oct. 30, 1993) (unpublished manuscript, presented to the American Society of Criminology).

[14]. Loftin et al., supra note 8, at 1619-20.

[15]. Glenn Pierce and William Bowers found something similar in their study of Massachusetts’ gun control law, and proposed that some sort of “announcement effect” may have been at work to produce this result. Glenn Pierce & William Bowers, The Bartley-Fox Gun Law’s Short-Term Impact on Crime in Boston, 455 ANNALS AM. ACAD. POL. & SCI. Sol. 120, 130-31 (1981). As this proffer amounts to the proposition that publicity about the new gun law discouraged murderers from shooting people, a bit of skepticism is warranted.

[16]. The authors believe that the growth of the crack cocaine market and other intervening variables eventually overcame the benefits of the gun control law, and of course, no one can say for sure if matters would have been worse had the District’s gun control law not been toughened. The fact remains, however, that neither crack cocaine nor other social ills were unique to Washington, D.C.: what was unique was its strict gun control law and a homicide rate that in recent years has been as high as 80 in 100,000 of population per year–a rate nearly five times higher than that in the next most-sanguinary American jurisdictions, gun-loving Texas and Louisiana. Loftin et al., supra note 8, at 1620.

[17]. David McDowall et al., Easing Concealed Firearms Laws: Effects on Homicide in Three States, 86 J. CRIM. L. & CRIMINOLOGY 193, 198 (1995).

[18]. Claims about what this study demonstrates, many of which are plainly more intrepid than any made by its authors, have received wide currency in the newspapers. Among the 83 mentions of this study in the Nexis data base, for example, was a column by Chicago Sun-Times editorial writer Cindy Richards, bearing the headline: “Legalizing Hidden Guns Will Increase Slaughter.” Cindy Richards, Legalizing Hidden Guns Will Increase Slaughter, CHI. SUN-TIMES, March 12, 1995, at 36.

[19]. Clayton E. Cramer & David B. Kopel, “Shall Issue”: The New Wave of Concealed Handgun Permit Laws, INDEPENDENCE INST. 15 (Oct. 1994) (on file with author).

[20]. McDowall et al., supra note 17, at 196.

[21]. James P. Pinkerton, Column Right: Nullification: Wrong in 1832 and in 1995, L.A. TIMES, October 12, 1995, at 9 (quoting Robert Heinlein).

[22]. ROGER D. MCGRATH, GUNFIGHTERS, HIGHWAYMEN AND VIGILANTES (1984); Robert Day, “Sooners” or “Goners,” They Were Hell Bent on Brabbing Free Land, 20 SMITHSONIAN 192, 202 (1989).

[23]. John Umbeck, Might Makes Rights: A Theory of the Formation and Initial Distribution of Property Rights, 19 ECON. INQUIRY 38, 40-46 (1981).

[24]. Daniel D. Polsby, Equal Protection, REASON MAG., Oct. 1993, at 34, 37.

[25]. KLECK, supra note 5, at 163-64.

[26]. So far as the law is concerned, criminals have just as much right to defend themselves from unlawful force as anyone else, but it seems safe to suggest that as far as public policy is concerned, this interest of criminals should be lightly regarded. Thought-experimental proof: if the only utility of firearms were to allow criminals to use them in lawful self-defense, we should have no difficulty embracing and trying to enforce a universal gun ban because, ex hypothesis, the only people who would be made worse off thereby would be criminals.

[27]. Defenders are never privileged to use deadly weapons when it would be reasonable to use non-deadly weapons or none at all. Furthermore, a number of states follow the Model Penal Code rule that certain circumstances require retreat before deadly force may lawfully be used.

[28]. Some scholars have taken a crack at pondering the imponderable. For example, David Kopel writes:

Only for one category of violent crime, namely, burglary, does the British rate exceed the American rate. Burglary is a more socially destructive crime in Britain because most British burglars attack houses when a victim is present. A 1982 survey found 59 percent of attempted burglaries take place against an occupied home, compared to just 13 percent in the United States. Fear of being shot convinces most American burglars to strike empty targets. [citations omitted.]

DAVID KOPEL, THE SAMURAI, THE MOUNTIE AND THE COWBOY 92, 127 (1992).

[29]. KLECK, supra note 5, at 23.

[30]. Metro Chicago Information Center, Summary of Neighborhood Crime Perceptions, in 1991-1995 MCIC METRO SURVEY (1995).

[31]. E.g., BUREAU OF JUSTICE STATISTICE, UNITED STATES DEPARTMENT OF JUSTICE, SOURCEBOOK OF CRIMINAL JUSTICE STATISTICS 1993, at 388, figure 3.7 (1994).

[32]. Jean Rimbach, Crime Without Punishment – Many Youths See System as “a Joke,” BERGEN COUNTY REC. (N.J), Feb. 2, 1994, at A1.

[33]. Elijah Anderson, The Code of the Street, 273 ATLANTIC MONTHLY, May 1994, at 81.

[34]. Reuter & Kleiman, supra note 11, at 303.

[35]. Philip Cook et al., Regulating Gun Markets, 86 J. CRIM. L. & CRIMINOLOGY 59, 63 (1995).

[36]. E.g., Rene Sanchez, Building an Arsenal One Gun at a Time, WASH. POST, Nov. 3, 1993, at A1; Donald Baker, Wilder Signs Gun Control Bill, WASH. POST, Mar. 24, 1993, at C1.

[37]. Daniel D. Polsby, The False Promise of Gun Control, 273 ATLANTIC MONTHLY, Mar. 1994 at 57.