Violence, Guns, and Drugs: A Cross-Country Analysis

The Journal of Law and Economics ^ | October, 2001 | Jeffrey A. Miron

[justlurking]

VIOLENCE, GUNS, AND DRUGS: A CROSS-COUNTRY ANALYSIS^*

JEFFREY A. MIRON
Bastiat Institute and Boston University

ABSTRACTViolence rates differ dramatically across countries. A widely held view is that these differences reflect differences in gun control and/or gun availability, and certain pieces of evidence appear consistent with this hypothesis. A more detailed examination of this evidence suggests that the role of gun control/availability is not compelling. This more detailed examination, however, does not provide an alternative explanation for cross-country differences in violence. This paper suggests that differences in the enforcement of drug prohibition are an important factor in explaining differences in violence rates across countries. To determine the validity of this hypothesis, the paper examines data on homicide rates, drug prohibition enforcement, and gun control policy for a broad range of countries. The results suggest a role for drug prohibition enforcement in explaining cross-country differences in violence, and they provide an alternative explanation for some of the apparent effects of gun control/availability on violence rates.

     ^* I thank John Lott for inviting me to prepare this paper for the American Enterprise Institute conference Guns, Crime, and Safety; Dan Kahan, Ed Glaeser, George Benston, Dennis Carlton, and an anonymous referee for comments on an earlier draft; seminar participants at Harvard University and the University of South Carolina for useful discussion; and especially my discussant, Mark Ramseyer, for a careful critique of an earlier draft.

I. INTRODUCTION

     VIOLENCE rates differ dramatically across countries. Each year there is roughly one homicide per 100,000 persons in England or Japan but nine per 100,000 in the United States. Moreover, these differences have persisted over long time periods; they suggest fundamental differences in the determinants of violence across countries, not just transitory variation.

     A widely held view is that these differences result from differences in gun control and/or gun availability, and certain pieces of evidence appear consistent with this hypothesis.¹ Most notably, England and Japan have restrictive gun control laws that border on complete prohibition with respect to handguns, and these countries appear to have relatively low numbers of guns per capita. The United States, in contrast, despite its many regulations and restrictions, allows legal ownership and use of a broad range of firearms and appears to have the highest rate of gun ownership in the world.

     A more detailed examination of the evidence, however, suggests a less compelling role for gun control/availability in explaining cross-country differences in violence rates.² Several countries (for example, Israel, Switzerland, and New Zealand) have relatively lax gun control laws and/or high firearms availability, yet have homicide rates that differ little from those in England or Japan. More generally, cross-country studies of the relation between homicide rates and gun control/availability suffer numerous deficiencies, including reliance on small samples, sensitivity to outliers, and an almost total lack of control variables. Perhaps most importantly, existing analyses fail to identify the direction of causation between guns and violence; a positive correlation between gun ownership and violence might indicate that violence creates a demand for guns rather than that guns cause violence. The existing literature thus fails to make a convincing case that guns and gun control are quantitatively important determinants of violence rates across countries.

     The question remains, however, What factors do explain cross-country differences in violence? One possible answer is that culturehistory, social norms, and the likemakes attitudes toward both violence and gun control substantially different across countries, simultaneously explaining both the differences in violence rates and the differences in gun control laws.³ Culture is likely an important determinant of cross-country differences in violence, but explanations based on culture are difficult to examine empirically. An alternative possibility is that various economic and social factors, such as demographics, ethnic diversity, education, income, inequality, deterrence, and the like, all contribute to the differences in violence across countries. There is empirical support for many of these effects, but it is not clear why such factors would explain the large differences in violence between the United States and other rich countries, nor is it clear how these factors can explain the existing patterns of gun availability, gun control, and violence.⁴

     This paper suggests that differences in the enforcement of drug prohibition are an important factor in explaining differences in violence rates across countries, and it argues that the omission of this factor from previous analyses explains much of the apparent effect of guns or gun control. The reasoning is as follows.

     In a black market, participants cannot resolve commercial disputes using lawsuits or battle over market share using advertising; they are thus likely to use violence instead. This means that the prohibition of drugs potentially causes increased levels of violence, even if prohibition reduces drug use and drug use itself causes violence. Moreover, the degree to which prohibition encourages violence is likely increasing with the degree to which it is enforced: higher levels of enforcement reduce the scope for legal circumvention of the prohibition, thus increasing the size of the black market, and higher levels of enforcement disrupt contractual arrangements and destroy reputational capital, thus increasing the level of violence for a black market of a given size.

     This reasoning suggests an explanation for many of the cross-country differences in violence rates. Virtually all countries have drug prohibition regimes that are similar in broad outline to that in the United States, but the degree of enforcement differs substantially. Conventional wisdom holds, in particular, that European countries rely far less on criminal sanctions than the United States, preferring medical or public health approaches.⁵ If this is accurate, then the elevated rate of violence in the United States compared with Europe is perhaps due to greater drug prohibition enforcement. Moreover, the fact that drug prohibition is lax in certain countries, implying low rates of violence, potentially explains why the restrictive gun laws in countries such as England or Japan have not themselves given rise to violent black markets in guns: the low rate of drug-prohibition-induced violence implies a minimal demand for guns and thus small or nonexistent black markets for guns. This reasoning also suggests why comparing violence rates with gun ownership rates might confuse the direction of causation; gun ownership rates might be high because violence rates are high, which in turn reflects the effects of drug prohibition.

     To determine the validity of this hypothesis, I examine data on homicide rates, drug prohibition enforcement, and gun control policy for a broad range of countries. The results suggest a role for drug prohibition enforcement in explaining cross-country differences in violence, and they provide an alternative explanation for some of the apparent effects of gun control/availability on violence rates.

     The paper proceeds in two steps. Section II discusses the theoretical relation between prohibitions and violence; explains the implications of this reasoning for violence, drug prohibition, and gun control; and presents the estimation framework used to examine these issues. Section III then examines cross-country data on violence, drug prohibition, and gun control. Section IV concludes.

     ¹ See, for example, Martin Killias, International Correlation between Gun Ownership and Rates of Homicide and Suicide, 148 Can. Med. Ass'n J. 1721 (1993); and Martin Killias, Gun Ownership, Suicide, and Homicide: An International Perspective, in Understanding Crime: Experiences of Crime and Crime Control 289 (Anna Alvazzi del Frate, Uglijesa Zvekic, & Jan J. M. van Dijk eds. 1993).
     ² See Gary Kleck, Targeting Guns: Firearms and Their Control (1997).
     ³ See David B. Kopel, The Samurai, the Mountie, and the Cowboy (1992).
     ⁴ For work on the cross-country determinants of violence, see Pablo Fajnzylber, Daniel Lederman, & Norman Loazya, Determinants of Crime Rates in Latin America and the World: An Empirical Assessment (1998); and Pablo Fajnzylber, Daniel Lederman, & Norman Loazya, Inequality and Violent Crime (unpublished manuscript, World Bank 1999).
     ⁵ See, for example, Diana R. Gordon, Europe's Kinder, Gentler Approach, 252 Nation 128 (1991); Melissa Bull et al., Comparative Analysis of Illicit Drug Strategy (1992); or Peter Reuter, Mathea Falco, & Robert J. MacCoun, Comparing Western European and North American Drug Policies (1993).

II. PROHIBITIONS AND VIOLENCE: THEORETICAL CONSIDERATIONS

     The theoretical reasoning that underlies this paper consists of two parts: the hypothesis that prohibitions increase violence by creating black markets in which violence is used to resolve disagreements, and the hypothesis that this effect is increasing with the degree to which prohibitions are enforced. This section develops these two hypotheses in detail and then explains how to examine them empirically.

     The hypothesis that prohibitions can increase violence is based on the following reasoning. Prohibitions of goods for which there is substantial demand and imperfect substitutes generally give rise to black markets, and in such markets, participants cannot easily resolve disputes via standard, nonviolent mechanisms. For example, black-market producers of a good cannot use the legal system to adjudicate commercial disputes such as nonpayment of debts. Black-market employers risk legal penalties themselves if they report their employees for misuse of "company" funds or property. Purchasers of black-market goods cannot sue for product liability, nor can sellers use the courts to enforce payment. Along a different line, rival firms cannot compete via advertising and thus might wage violent turf battles instead. Thus, in black markets, disagreements are more likely to be resolved with violence.

     The hypothesis that prohibitions increase violence is consistent with a number of stylized facts. Numerous sources, anecdotal and otherwise, report the use of violence in the alcohol trade during alcohol prohibition in the United States (192033), but not before or after. Violence committed by pimps or johns against prostitutes is widely regarded as a feature of prostitution markets, in which prostitutes cannot easily report violence without risking legal sanctions themselves. Similarly, violence was an important feature of the gambling industry during its early years in the United States, when entry was prohibited in most places; the incidence of violence has decreased as legal gambling has mushroomed.

     Nevertheless, the hypothesis that prohibitions increase the use of violence to resolve disagreements is incomplete, since many prohibitions are associated with minimal levels of violence. For example, compulsory schooling laws are prohibitions against not attending school, yet there is little violence associated with this prohibition.⁶ Minimum-wage laws are prohibitions against hiring employees at subminimum wages, yet at least in the United States, there is little violence associated with this prohibition. More generally, a broad range of regulatory polices (environmental, Occupational Safety and Health Administration, and labor market) can be characterized as prohibitions and yet do not appear to generate violence, nor were the pre-1920, state-level prohibitions of alcohol or the 1940s and 1950s federal prohibitions of drugs associated with nearly the level of homicide experienced in the last several decades. And, of most relevance to this analysis, western European countries have drug prohibition laws similar to those in the United States, yet substantially lower rates of violence.

     There are several reasons why prohibitions might not generate violence, the analysis above notwithstanding. Some prohibitions, such as compulsory education, do not interfere with a substantial number of transactions. Other prohibitions, such as minimum-wage laws, prohibit actions for which there is not sufficient demand to generate large-scale black markets. And still other prohibitions outlaw goods for which there are reasonable substitutes.

     Most importantly, however, I suggest here that prohibitions are unlikely to create violence unless there is substantial enforcement, and the amount of violence caused will increase with the degree of enforcement. There are two parts to this argument.⁷ First, prohibitions are unlikely to create substantial black markets unless there is a substantial degree of enforcement, and the size of the black market will increase with the degree of enforcement. The reason is that prohibitions generally contain exceptions that permit legal or quasi-legal production and consumption of the good, thus allowing use of standard, nonviolent mechanisms to resolve many disagreements related to the prohibited product. Increased enforcement, however, in the form of new laws that decrease the scope of the exceptions, or increased monitoring of existing exceptions, places some additional transactions outside the realm of legal dispute resolution mechanisms.

     For example, the United States did not treat the maintenance of opiate users by physicians as proscribed until several years after prohibition took effect.⁸ Similarly, England allowed doctors relatively free rein in dispensing heroin for the first several decades of its drug prohibition, but since the 1960s it has imposed greater limits on heroin maintenance. And the gun control systems in many countries have gradually become more restrictive.⁹

     Similarly, it was legal during alcohol prohibition to produce small quantities of alcohol for personal use, to produce certain kinds of low-alcohol wine and beer, to put alcohol in medicines and sacramental wines, and to use alcohol in industrial products. When monitoring and enforcement were lax, these exceptions provided substantial amounts of legal alcohol. In the case of drug prohibition, doctors can prescribe many otherwise prohibited drugs, and several countries operate treatment programs that provide prohibited drugs to certain consumers. Again, under lax enforcement, these sources of supply meet much of the market demand legally. In the case of prostitution, various escort services are legal, even though prostitution itself is illegal, so these services meet much of the demand without generating violence so long as enforcement is lax. In the case of prohibitory gun laws, exceptions for collectors or existing owners are common, and government use of prohibited firearms often remains legal. With little enforcement, these exceptions supply much of the market.

     The critical aspect of all these examples is that, when exceptions to the prohibition law exist, at least some manufacturing, transportation, and distribution of the good is legal; thus, this activity is unlikely to generate violence. There still might be violence associated with the illegal diversion of the good, but far less than if the good is prohibited entirely.

     The second reason that enforcement is critical to the degree of violence observed under a prohibition is that participants in black markets are likely to develop mechanisms for avoiding violence, but enforcement makes this more difficult. For example, rival suppliers might agree to cartelize a market, thus reducing the need for advertising, but enforcement leading to the arrest of one of these suppliers generates violence among the remaining suppliers, who attempt to capture new market share. Alternatively, black-market suppliers might create private, nonviolent mechanisms for resolving disputes, but enforcement that creates turnover among suppliers will destroy reputational capital and make such arrangements difficult to maintain. Still another mechanism is that given higher dispute resolution costs, participants in a black market will choose production and distribution methods that minimize transactions (for example, home production), but heightened levels of enforcement make this difficult. Likewise, consumers of the prohibited commodity might purchase repeatedly from a reliable supplier, but enforcement that generates turnover among suppliers makes this harder, increasing the scope for disagreements.

     Beyond the effects of increased enforcement discussed aboveincreasing the black market's share of the prohibited commodity and increasing the likelihood of violence for a black market of a given sizethere are several other mechanisms by which increased enforcement might increase the level of violence observed under a prohibition.

     First, increased enforcement of a prohibition might be accompanied by a redistribution of criminal justice resources away from other violence-reducing government policies, such as crime deterrence, the provision of an efficient system for protecting property rights, or suppression of other sources of violence. For example, increased enforcement of drug prohibition for a fixed-size police budget of implies reduced enforcement of laws against homicide, robbery, assault, and the like.¹⁰ This issue arises, for example, when violent prisoners are released early to make room for drug offenders. In places such as Russia, the resources devoted to drug prohibition enforcement might "crowd out" general enforcement of property rights, thus encouraging participants in other sectors to employ violence. And in countries such as Colombia or Peru, the resources devoted to drug enforcement are unavailable for fighting guerilla groups, who generate substantial violence for independent reasons.

     A different reason that prohibitions might generate violence is that prohibitions often raise the price of the prohibited commodity.¹¹ Elevated prices constitute a negative income shock to consumers of the prohibited good, which can encourage increased income-generating crime to finance purchases of the good. This mechanism does not necessarily imply violence directly, since many income-generating crimes are nonviolent (theft, shoplifting, prostitution). But some income-generating crimes are violent (robbery), and violence can occur incidentally as a result of otherwise nonviolent crimes. Assuming that increased enforcement implies higher prices, increased enforcement implies more income-generating crime and related violence.

     The higher prices caused by prohibition might also encourage violence by increasing the rents to certain factors. One model of what occurs under prohibition is that suppliers enter the prohibited market until the total return from black-market activity equals the total return from legal activity, taking into account the risks of incarceration, injury, or death and any stigma or glamor associated with working in a black market. Assuming homogeneity in the willingness to accept the special features of black-market activity, prohibition does not imply any excess profits in the prohibited as opposed to the legal sector. If there is heterogeneity in the willingness to work in the black market, however, then those more willing to do so select into this sector, earn rents to this characteristic, and are better off under prohibition. Such persons have more to protect under prohibition and might therefore have an additional reason to engage in violence, namely, protecting these rents. And the magnitude of this effect is likely increasing with increased enforcement, assuming prices increase with enforcement.

     Still another mechanism whereby prohibition might encourage violence is by making consumers or producers of the prohibited commodity less likely to use the official dispute resolution system for disputes not related to the prohibited commodity.¹² For example, a drug user or seller who has been robbed of nondrug items might not report this to the policesince this could risk sanctions related to possession or sale of drugsand instead attempt to sanction the perpetrator of the robbery himself, possibly using violence. And higher enforcement is likely to increase this effect; if police routinely overlook small quantities of prohibited substances, the effect is likely to be small; if police routinely hassle anyone thought to be associated with the prohibited good, the effect is likely to be large.

     The reasoning outlined above suggests the following hypotheses for empirical examination. First, differences in the degree of drug prohibition enforcement across countries might explain differences in violence. In addition, gun control might itself increase violence by driving gun markets underground, so differences in gun control across countries might also explain differences in violence.

     To examine the relation between violence, drug prohibition enforcement, and gun control empirically, I estimate



where v_i represents the level of violence in country i, e represents the degree of drug prohibition enforcement, e represents the degree of gun control, x_i represents other determinants of violence, and _i is an error term.

     In estimating (1), I treat differences in drug prohibition enforcement and gun control as exogenous with respect to the level of violence. In practice, some countries might choose greater enforcement of drug prohibition or impose stricter gun control laws in response to higher levels of violence. Thus, a positive relation between drug prohibition enforcement and violence does not prove that there is a causal effect of enforcement on violence, and the estimated relation between gun control and violence might understate the true causal effect of gun control in reducing the level of violence. There are likely several factors that contribute to differences in drug prohibition enforcement and gun control regimes other than the homicide rates themselves, however. For example, the strong degree of drug prohibition enforcement in Latin America results in part from U.S. attempts to address its own drug or crime problems, not just from events in Latin America. Thus, although not strictly exogenous, the differences in drug prohibition enforcement and gun control are plausibly predetermined relative to homicide rates over the time horizons considered here, in which case a causal interpretation of the results is likely to be approximately correct.

     ⁶ Some might characterize episodes such as the Columbine shootings as violent resolution of a disagreementwhether the perpetrators had to be in schoolthat could not readily be adjudicated by other means. Whether one adopts this interpretation or not, the total number of such incidents is trivial compared to the number of persons affected by compulsory education laws.
     ⁷ J. Mark Ramseyer, Guns and Drugs and Rock and Roll: A Comment on Miron (unpublished manuscript, Harvard Univ. L. Sch. 2000), discusses conditions under which the level of enforcement might not be critical. His framework omits the two key considerations emphasized below: that the level of enforcement affects the size of the black market and that the level of enforcement affects the ability of participants to establish reputations that help discourage violence. Ramseyer also notes that some kinds of enforcement are more likely than others to generate violence. In particular, locking up the small-fry traffickers leaves the bosses' reputational capital in place; locking up the kingpins does not.
     ⁸ David F. Musto, The American Disease: Origins of Narcotics Control (1973).
     ⁹ See, for example, Kopel, supra note 3; or Joseph E. Olson & David B. Kopel, All the Way Down the Slippery Slope: Gun Prohibition in England and Some Lessons for Civil Liberties in America, 22 Hamline L. Rev. 399 (1999).
     ¹⁰ Bruce L. Benson, David W. Rasmussen, & Iljoong Kim, Deterrence and Public Policy: Trade-offs in the Allocation of Police Resources, 18 Int'l Rev. L. & Econ. 77 (1998), examines the relation between the Index I crime rate and the rate of drug arrests using panel data for a cross section of Florida counties during the mid-1980s. Benson and coauthors find a marginally significant positive association between drug arrests and the Index I crime rate, controlling for a variety of demographic factors and the levels of police resources. They do not address the breakdown of their results for different kinds of Index I crimes, so the results shed no direct light on the effect of drug prohibition on homicide. Bruce L. Benson et al., Is Property Crime Caused by Drug Use or by Drug Enforcement Policy? 24 Applied Econ. 679 (1992), and David L. Sollars, Bruce L. Benson, & David W. Rasmussen, Drug Enforcement and Deterrence of Property Crime among Local Jurisdictions, 22 Pub. Fin. Q. 22 (1994), provide evidence that increased drug enforcement diverts police resources from deterring property crime.
     ¹¹ Jeffrey A. Miron, Do Prohibitions Raise Prices? Evidence from the Market for Cocaine (unpublished manuscript, Boston Univ. 2001), argues that the effect of prohibitions on the price of the prohibited commodity is less clear-cut than generally believed. Nevertheless, it appears that many, perhaps most, prohibitions are associated with higher prices than would otherwise exist, so I accept this presumption here.
     ¹² I am indebted to Bjorn Frank for suggesting this hypothesis.

III. RESULTS

     This section presents evidence on the relation between violence, drug prohibition enforcement, and gun control across countries. I first present the basic data used in the analysis and then consider regression results that relate these and other variables.

A. Violence Rates across Countries

     Column (1) of Table 1 presents vital statistics data on the average homicide rate in countries for which data are available in at least one of the years 199396.¹³ This period is used because it corresponds roughly to the period for which data on drug prohibition enforcement and gun control laws are available.

TABLE 1     HOMICIDE RATES, DRUG SEIZURES PER CAPITA, AND GUN PROHIBITION

     The data show first that homicide rates differ substantially across countries. The U.S. homicide rate averages approximately nine per 100,000, which is five to nine times the average rate in most Western-style democracies. At the same time, the homicide rate in the United States is similar to or less than the rate in many countries. Seven Central or South American countries have homicide rates in excess of the U.S. rate, and several others have rates that are close. Every country in this group has a homicide rate in excess of the average for the rich countries other than the United States. Similarly, 10 of the 20 former Soviet bloc countries have rates that exceed the U.S. rate, with practically every country in this group having a homicide rate in excess of the non-U.S., rich-country average. Thus, the level of homicide in the United States stands out in comparison to other rich, democratic countries, but not in comparison to the world as a whole.

     This overview of the data on homicide rates is partially suggestive of the main hypothesis of this paper. On the one hand, violence rates are high in the countries of the Caribbean and Latin America, most of which are key producers of, or transit points for, illegal drugs. On the other hand, violence rates are also high in the countries of the former Soviet bloc, which are less obviously important producers or transhippers of illegal drugs.

     ¹³ For details, see Jeffrey A. Miron, Violence, Guns, and Drugs: A Cross-Country Analysis 11 (Working Paper No. 107, Boston Univ., Dep't Econ., Industry Stud. Prog. 2001).

B. Drug Prohibition Enforcement

     Constructing a measure of drug prohibition enforcement is difficult for two reasons. First, none of the standard measures of enforcement (for example, drug arrests) is necessarily well correlated with the conceptually correct measure, the degree to which enforcement raises the cost of using nonviolent dispute resolution mechanisms. Second, there are few reliable data on the standard measures in any event.¹⁴

     The one variable for which consistent data are available is seizures of illegal drugs. This is not a perfect measure, but it plausibly captures the main effect of enforcement in disrupting dispute resolution within the black market for drugs. Further, a high seizure rate is likely to correlate positively with other aspects of enforcement, such as a high drug arrest rate, and reflect a stricter attitude toward enforcement.

     The main negative of the seizure rate is that differences across countries might reflect differences in the use of drugs, holding enforcement constant.¹⁵ If drug use rates differ substantially, then the amount of seizures can differ even if enforcement does not. And if drug use independently causes violence, there will be a positive correlation between violence and drug seizure rates even without an effect of enforcement on violence. There is little convincing evidence that drug use has an independent effect in precipitating violence,¹⁶ however, and the existing data on drug use across countries show if anything a negative correlation between drug abuse and homicide rates.¹⁷ But existing data on drug consumption across countries are too crude to test this hypothesis convincingly, so this issue deserves consideration in future work.^18,19

     In this paper, I employ the drug seizure rate as the main measure of drug prohibition enforcement.²⁰ The available data tabulate the quantities of drugs seized in each of 22 different categories for most countries in the world over the period 199496.²¹ The difficulty that arises in using these data is that no one category of seizures is likely to capture a country's enforcement practices fully, yet adding seizures in different categories is problematic. I address this problem by considering several different measures of seizures.

     The data on three of these measures, presented in columns (2)(4) of Table 1, show that per capita seizures differ substantially across countries. Some of the variation is consistent with the main hypothesis of this paper, but there are also many anomalies. Roughly, seizure rates for several categories are high in Central and Latin America, which is consistent with this paper's thesis given the high observed homicide rates in these countries. But seizure rates in the former Soviet bloc countries tend to be low, which is not consistent given the high observed homicide rates there.

     ¹⁴ See id. at 13, n.9.
     ¹⁵ See Ramseyer, supra note 7.
     ¹⁶ Steven B. Duke & Albert C. Gross, America's Longest War: Rethinking Our Tragic Crusade against Drugs (1993).
     ¹⁷ The drug use data are from United Nations, Office of Drug Control and Crime Prevention, Global Illicit Drug Trends 1999 (1999).
     ¹⁸ Differences in drug use rates also imply differences in the size of black markets and therefore differences in the amount of violence because of a greater number of black-market transactions. This factor by itself implies a negative correlation between the degree of enforcement and the level of violence.
     ¹⁹ A different limitation of the seizure rate is that it takes no account of differences in drug treatment and drug maintenance. In practice, inclusion of such information might strengthen the results, since several European countries with high seizure rates (for example, the Netherlands) also have generous treatment or maintenance policies.
     ²⁰ The data are from United Nations International Drug Control Programme, Supply of and Trafficking in Narcotic Drugs and Psychotropic Substances 1996 (Tech. Series No. 7, U.N. Int'l Drug Control Programme 1998).
     ²¹ See Miron, supra note 13, at 14.

C. Gun Control across Countries

     To measure differences in gun control laws across countries, I use the United Nations International Study on Firearm Regulation.²² This survey asked United Nations member countries to provide information on their gun control laws and outcomes related to guns. The information was requested from participating countries in 1996, so the information roughly corresponds to 1995.

     The survey questions most relevant to this paper ask whether a country has regulations that prohibit the ownership of long guns or handguns. A potential problem is that the questions are broad, allowing countries with substantially different gun control laws to provide the same responses even if their laws differ substantially. In particular, the possible answers to the questions about gun prohibition are "none," "certain," or "all," with responses coded 0, 1, or 2, respectively. Thus, many countries ban certain guns but permit others, so many countries belong in the middle category even though there are major differences in this middle group with respect to the kinds of guns prohibited. An additional difficulty is that this data set measures differences in laws, not the degree to which they are enforced.

     Column (5) of Table 1 provides the sum of each country's responses to questions about prohibitions of long guns and prohibitions of handguns. As hypothesized, there are certain anomalies: the United States and the United Kingdom receive the same score, for example. Overall, however, the responses are consistent with other information about gun control regimes.

     ²² See United Nations, Crime Prevention and Criminal Justice Division, United Nations International Study on Firearm Regulation, Guidelines and Questionnaire (1999).

D. Regression Results

     Table 2 presents regressions of the homicide rate on each of nine different drug seizure rates, the gun prohibition variable, and a set of control variables.²³ Because the sample size is small and control variables are often missing for countries in the basic sample, specifications with a complete set of control variables are problematic.²⁴ I focus on control variables that exist for relatively broad samples of countries: gross national product (GNP) per capita, the percentage of the male population that is aged 1524, the percentage of the population living in urban areas, population density, educational attainment, and use of the death penalty.²⁵

TABLE 2     CROSS-COUNTRY REGRESSIONS OF HOMICIDE RATE ON DRUG SEIZURES, GUN PROHIBITION, AND CONTROL VARIABLES

     The results show first that drug seizure rates and homicide rates are positively related in six of nine cases. Three of the positive results are strongly statistically significant, and the three remaining positive coefficients have t-statistics of 1.3 or greater. None of the negative coefficients is significant, and two of these (Heroin and Opiates) are for categories with small quantities of seizures.²⁶

     The results also suggest that greater prohibition of guns is associated with a higher homicide rate; the coefficient on gun prohibition is positive in eight of nine cases, although the coefficient never enters significantly. Results that use handgun prohibition rather than all-gun prohibition are extremely similar.²⁷

     The control variables themselves are almost never individually significant and never jointly significant, so I omit any detailed discussion here. In general, the signs are consistent with results in earlier work on cross-country determinants of violence.²⁸

     The results in Table 2 rely on the seizure measure of drug prohibition enforcement. Existing data on alternative measures are problematic,²⁹ but I have estimated the specifications in Table 2 for two alternative measures of drug prohibition enforcement, the drug trafficking arrest rate and the total drug offense arrest rate. In these regressions, the coefficient on gun prohibition is positive but not significant. The coefficient on the total drug arrest rate is negative but not significant, while the coefficient on the trafficking arrest rate is positive but not significant. The trafficking arrest rate is likely the better measure of the degree to which enforcement causes violence in drug markets. Given this, along with the data problems and the small samples involved, these results suggest only a mild weakening of the conclusions suggested by the seizure measure of enforcement.

     The empirical results are therefore consistent with the overall hypotheses of the paper. The measure of drug prohibition enforcement highlighted here, per capita seizures of drugs, enters positively and sometimes statistically significantly in a substantial fraction of the regressions. Likewise, the measure of gun prohibition usually enters positively although not significantly.

     Although the results are subject to several caveats, they are consistent with other evidence that suggests an important role for drug prohibition in increasing the level of violence. Paul J. Goldstein and coauthors, using police reports and police evaluations, examine the causes of all homicides in a sample of New York City precincts during part of 1988.³⁰ They determine that more than half of the homicides were due to drug-related factors, but of these, almost three-quarters were due to "systemic" factors, meaning disputes over drug territory, drug debts, and other drug-trade-related issues. Thus, approximately 39 percent of the homicides resulted from the inability of drug market participants to settle disputes using the official dispute resolution system; only 7.5 percent resulted from the psychopharmacological effects of drugs or alcohol.

     Harold J. Brumm and Dale O. Cloninger compare homicide offense rates, homicide arrest rates, and drug prohibition arrest rates across cities.³¹ They find that drug prohibition arrest rates are negatively associated with homicide arrest rates and that homicide arrest rates are negatively associated with homicide offense rates, implying that higher drug prohibition arrest rates are associated with higher homicide offense rates. They interpret these results as suggesting that increased enforcement of drug prohibition takes resources away from deterrence of other criminal activity, such as homicide.

     David W. Rasmussen, Bruce L. Benson, and David L. Sollars find that a higher drug arrest rate is positively associated with the violent crime rate in a cross section of Florida jurisdictions in 1989.³² They also find that a higher drug arrest rate implies a higher violent crime rate in neighboring jurisdictions, presumably because increased drug enforcement in one jurisdiction disrupts the market equilibrium in neighboring jurisdictions.

     Jeffrey A. Miron documents that increases in enforcement of drug and alcohol prohibition over the past 100 years have been associated with increases in the homicide rate, and auxiliary evidence suggests that this positive correlation reflects a causal effect of prohibition enforcement on homicide.³³ Controlling for other potential determinants of the homicide ratethe age composition of the population, incarceration rate, economic conditions, gun availability, and the death penaltydoes not alter the conclusion that drug and alcohol prohibition have substantially raised the homicide rate in the United States over much of the past century.

     Finally, Pablo Fajnzylber, Daniel Lederman, and Norman Loazya regress crime statistics measures of homicide rates for the period 197094 on a broad range of variables, including GNP per capita, Gini indices, average years of schooling, urbanization rates, deference measures (for example, the death penalty), religious composition, and region dummies, plus a dummy for whether a country produces drugs and the drug possession arrest rate.³⁴ Across a broad range of specifications, they find that being a drug-producing country or having a high drug possession arrest rate is positively associated with a higher homicide rate. They also consider panel regressions of 5-year average homicide rates and again obtain a consistently positive relation between the drug production or arrest variables and homicide rates. They obtain a similar result for the drug producer dummy using vital statistics data on homicide rates.³⁵

     ²³ The seizure rates are defined as follows: Cocaine Base, Cannabis Herb, and Heroin are the kilogram data for these three drugs. Cannabis is the sum over all kilogram seizures of cannabis products; Coca is the sum over all kilogram seizures of coca products; Opiates is the sum over all kilogram seizures of opium products; Pills is the sum over all units seizures of depressants, hallucinogens, LSD, methaqualone, and stimulants; and Cannabis Plants and Opium Plants are the units seizures of cannabis plants and opium plants, respectively. I calculate per capita amounts using the 1995 population.
     ²⁴ For examples of studies that attempt to include comprehensive sets of control variables, see Fajnzylber, Lederman, & Loazya, Determinants of Crime Rates in Latin America and the World, supra note 4; and Fajnzylber, Lederman, & Loazya, Inequality and Violent Crime, supra note 4.
     ²⁵ The sources of these data are identical to the sources employed by Fajnzylber, Lederman, & Loazya, Determinants of Crime Rates in Latin America and the World, supra note 4; and Fajnzylber, Lederman, & Loazya, Inequality and Violent Crime, supra note 4. The precise data that I employ differ slightly because of differences in samples.
     ²⁶ See Miron, supra note 13, table 2.
     ²⁷ See id., table 8. Regressions that omit the control variables, and thus utilize a larger sample, do find a statistically significant coefficient on gun prohibition in most cases; see id., tables 7 and 8.
     ²⁸ See Fajnzylber, Lederman, & Loazya, Determinants of Crime Rates in Latin America and the World, supra note 4; and Fajnzylber, Lederman, & Loazya, Inequality and Violent Crime, supra note 4.
     ²⁹ See Miron, supra note 13, at 13, 20
     ³⁰ Paul J. Goldstein et al., Crack and Homicide in New York City, 1988: A Conceptually Based Event Analysis, 16 Contemp. Drug Probs. 651 (1989).
     ³¹ Harold J. Brumm & Dale O. Cloninger, The Drug War and the Homicide Rate: A Direct Correlation? 14 Cato J. 507 (1995).
     ³² David W. Rasmussen, Bruce L. Benson, & David L. Sollars, Spatial Competition in Illicit Drug Markets: The Consequences of Increased Drug Law Enforcement, 23 Rev. Regional Stud. 219 (1993).
     ³³ Jeffrey A. Miron, Violence and the U.S. Prohibitions of Drugs and Alcohol, 1 Am. L. & Econ. Rev. 78 (1999).
     ³⁴ Fajnzylber, Lederman, & Loazya, Determinants of Crime Rates in Latin America and the World, supra note 4.
     ³⁵ Fajnzylber, Lederman, & Loazya, Inequality and Violent Crime, supra note 4.

IV. DISCUSSION

     The empirical results presented above provide a possible explanation for the large differences in violence rates across countries, and they suggest that previous analyses might have spuriously attributed these differences to gun control or availability. According to the analysis here, differences in drug prohibition enforcement explain differences in violence, which in turn explain differences in gun ownership that correlate positively with violence but do not cause that violence. Further, the results provide a hint that restrictive gun control regimes can themselves increase violence. As noted above, these results should be considered suggestive rather than conclusive. Future research on these issues will need to exploit time-series rather than cross-sectional data.

BIBLIOGRAPHY

• Benson, Bruce L.; Rasmussen, David W.; and Kim, Iljoong. "Deterrence and Public Policy: Trade-offs in the Allocation of Police Resources." International Review of Law and Economics 18 (1998): 77100.
• Benson, Bruce L., et al. "Is Property Crime Caused by Drug Use or by Drug Enforcement Policy?" Applied Economics 24 (1992): 67992.
• Blumstein, Alfred, and Cork, Daniel. "Linking Gun Availability to Youth Gun Violence." Law and Contemporary Problems 59 (1996): 524.
• Brumm, Harold J., and Cloninger, Dale O. "The Drug War and the Homicide Rate: A Direct Correlation?" Cato Journal 14 (1995): 50717.
• Bull, Melissa, et al. Comparative Analysis of Illicit Drug Strategy. Canberra: Australian Government Publishing Service, 1992.
• Cook, Philip J. "The Influence of Gun Availability on Violent Crime Patterns." In Crime and Justice: An Annual Review, Vol. 4, edited by Michael Tonry and Norval Morris, pp. 4989. Chicago: University of Chicago Press, 1983.
• Duke, Steven B., and Gross, Albert C. America's Longest War: Rethinking Our Tragic Crusade against Drugs. New York: G. P. Putnam's Sons, 1993.
• Fajnzylber, Pablo; Lederman, Daniel; and Loazya, Norman. Determinants of Crime Rates in Latin America and the World: An Empirical Assessment. Washington, D.C.: World Bank, 1998.
• Fajnzylber, Pablo; Lederman, Daniel; and Loazya, Norman. "Inequality and Violent Crime." Unpublished manuscript. Washington, D.C.: World Bank, 1999.
• Goldstein, Paul J., et al. "Crack and Homicide in New York City, 1988: A Conceptually Based Event Analysis." Contemporary Drug Problems 16 (1989): 65187.
• Gordon, Diana R. "Europe's Kinder, Gentler Approach." Nation 252 (1991): 12830.
• Kellerman, Arthur L., et al. "Gun Ownership as a Risk Factor for Homicide in the Home." New England Journal of Medicine 329 (1993): 108491.
• Killias, Martin. "Gun Ownership, Suicide, and Homicide: An International Perspective." In Understanding Crime: Experiences of Crime and Crime Control, edited by Anna Alvazzi del Frate, Uglijesa Zvekic, and Jan J. M. van Dijk, pp. 289302. Rome: United Nations Interregional Crime and Justice Research Institute, 1993.
• Killias, Martin. "International Correlation between Gun Ownership and Rates of Homicide and Suicide." Canadian Medical Association Journal 148 (1993): 172125.
• Kleck, Gary. Targeting Guns: Firearms and Their Control. New York: Aldine de Gruyter, 1997.
• Kopel, David B. The Samurai, the Mountie, and the Cowboy. Buffalo, N.Y.: Prometheus Books, 1992.
• Krug, E. G.; Powell, K. E.; and Dahlberg, L. L. "Firearm-Related Deaths in the United States and other High- and Upper-Middle-Income Countries." International Journal of Epidemiology 27 (1998): 21421.
• Lott, John R., Jr., and Mustard, David B. "Crime, Deterrence, and Right-to-Carry Concealed Handguns." Journal of Legal Studies 26 (1997): 168.
• Miron, Jeffrey A. "Violence and the U.S. Prohibitions of Drugs and Alcohol." American Law and Economics Review 1 (1999): 78114.
• Miron, Jeffrey A. "Do Prohibitions Raise Prices? Evidence from the Market for Cocaine and Heroin." Unpublished manuscript. Boston: Boston University, 2001.
• Miron, Jeffrey A. "Violence, Guns, and Drugs: A Cross-Country Analysis." Working Paper No. 107. Boston: Boston University, Department of Economics, Industry Studies Program, 2001.
• Musto, David F. The American Disease: Origins of Narcotics Control. New Haven, Conn.: Yale University Press, 1973.
• Olson, Joseph E., and Kopel, David B. "All the Way Down the Slippery Slope: Gun Prohibition in England and Some Lessons for Civil Liberties in America." Hamline Law Review 22 (1999): 399466.
• Ramseyer, J. Mark. "Guns and Drugs and Rock and Roll: A Comment on Miron." Unpublished manuscript. Cambridge, Mass.: Harvard University Law School, 2000.
• Rasmussen, David W.; Benson, Bruce L.; and Sollars, David L. "Spatial Competition in Illicit Drug Markets: The Consequences of Increased Drug Law Enforcement." Review of Regional Studies 23 (1993): 21936.
• Reuter, Peter; Falco, Mathea; and MacCoun, Robert J. Comparing Western European and North American Drug Policies. Santa Monica, Cal.: Rand, 1993.
• Sollars, David L.; Benson, Bruce L.; and Rasmussen, David W. "Drug Enforcement and Deterrence of Property Crime among Local Jurisdictions." Public Finance Quarterly 22 (1994): 2245.
• United Nations. Demographic Yearbook 1996. New York: United Nations, 1998.
• United Nations. Crime Prevention and Criminal Justice Division. United Nations International Study on Firearm Regulation, Guidelines and Questionnaire. Vienna: United Nations, 1999.
• United Nations. International Drug Control Programme. Supply of and Trafficking in Narcotic Drugs and Psychotropic Substances 1996. Vienna: United Nations International Drug Control Programme, 1998.
• United Nations. Office of Drug Control and Crime Prevention. Global Illicit Drug Trends 1999. Vienna: United Nations International Drug Control Programme, 1999.
• White, Halbert. "A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity." Econometrica 48 (1980): 81738.
• Zimring, Franklin E. "Firearms, Violence, and Public Policy." Scientific American 265 (1991): 4858.

[Lazamataz]

I hate people who introduce facts into the debate about guns.

If we don't ban all guns, then the terrorists will have won.

[justlurking]

I forgot to add this comment:

The entire October 2001 issue of the Journal of Law and Economics is on the web at:

If you don't want to read the entire thing, the introduction by John Lott is a good place to start:

http://www.journals.uchicago.edu/JLE/journal/issues/v44nS2/012212/012212.html

Here's the full list of articles:

• Guns, Crime, and Safety: Introduction
• Violence, Guns, and Drugs: A Cross-Country Analysis
• The Impact of Gun Laws on Police Deaths
• Safe-Storage Gun Laws: Accidental Deaths, Suicides, and Crime
• The Impact of Banning Juvenile Gun Possession
• Guns, Crime, and Academics: Some Reflections on the Gun Control Debate
• Privately Produced General Deterrence
• Right-to-Carry Concealed Weapon Laws and Homicide in Large U.S. Counties: The Effect on Weapon Types, Victim Characteristics, and Victim-Offender Relationships
• Does the Right to Carry Concealed Handguns Deter Countable Crimes? Only a Count Analysis Can Say
• Testing for the Effects of Concealed Weapons Laws: Specification Errors and Robustness

[justlurking]

The entire October 2001 issue of the Journal of Law and Economics is on the web at:

Hmmm. For some reason, it ate the URL (and I didn't see it until after I posted). Here it is:

http://www.journals.uchicago.edu/JLE/journal/contents/v44nS2.html

[Khepera]

The fact of the matter is there is no such thing as Gun Violence.

[Lazamataz]

I beat up my Ruger 10/22 the other day. Gave it a black eye and a busted lip.

Therefore, there is such a thing as Gun Violence.

I am a Gun Nut. You can affix me to a Gun Screw to keep a Gun Part in place.

[Khepera]

Hey the violence didn't come from a Gun it came from you when you beat the gun and gave it a black eye.

Keep your nuts out of it!

[Atlas Sneezed]

Bang

[justlurking]

Bang

Thank you. I couldn't remember the list name.

Does someone keep a list of lists? I think there may be at least one other that needs to be added.

[Blood of Tyrants]

Interesting hypothesis. The more they enforce the "War on Drugs", the more violent the dealers get, and thus the greater the cry (at least from the liberal media and the sheeple) for more gun control. The liberals NEED the WOD even though they don't give a sh#% about the people who use and sell them.

It would have been good if he could have weeded out the number of homocides known or suspected to be caused by drug dealings.

[justlurking]

It would have been good if he could have weeded out the number of homocides known or suspected to be caused by drug dealings.

He mentions it:

Paul J. Goldstein and coauthors, using police reports and police evaluations, examine the causes of all homicides in a sample of New York City precincts during part of 1988 (footnote 30). They determine that more than half of the homicides were due to drug-related factors, but of these, almost three-quarters were due to "systemic" factors, meaning disputes over drug territory, drug debts, and other drug-trade-related issues. Thus, approximately 39 percent of the homicides resulted from the inability of drug market participants to settle disputes using the official dispute resolution system; only 7.5 percent resulted from the psychopharmacological effects of drugs or alcohol.

But, that's a separate study. You would have to follow the footnote and read it.

I found the paragraphs immediately following this particularly interesting:

• Enforcement of drug prohibition appears to shift law enforcement resources away from deterrence of other crimes, such as homicide.
• Increased enforcement of drug prohibition in a specific jurisdiction appears to increase violent crime in neighboring jurisdictions.
• The enforcement level of alcohol and drug prohibition in the US has a positive correlation with the homicide rate.

[Blood of Tyrants]

Thanks. That's what I thought. Almost half of all murders involving guns were caused by disputes over drugs.

[ActionNewsBill]

One year bump