One point on which almost everyone interested in drug prohibition agrees is the existence of a connection between drugs and violent crime. The disagreement is in the form of the connection and the sign of the correlation. Supporters of drug prohibition typically argue that drug use leads to violent crime and should be illegal in part for that reason. Critics of the war on drugs argue that the attempt to prohibit drug use leads to violent crime and that that is one of the reasons drugs should be legal. A glance at the figures for U.S. murder rates over the course of this century provides some support for the critics’ position (Figure 1). Murder rates were high during the period of alcohol prohibition, fell after repeal, rose again with increased efforts to prohibit illegal drugs, and remain high.
The impression given by the graph is confirmed by more sophisticated analysis. Jeffrey A. Miron has analyzed the relation between violent crime in the U.S., as measured by the murder rate, and the enforcement of drug prohibition (including alcohol prohibition) as measured by expenditures by the federal agencies in charge of enforcing prohibition (Figure 2), over the entire period for which murder rates are available on a national basis. His statistical results “suggest the homicide rate is currently 25%-75% higher than it would be in the absence of drug prohibition.”The case of the U.S. is particularly interesting for at least two reasons. One is that the U.S. murder rate is anomalously high relative to other countries that are otherwise similar–about 8 to 10 murders per 100,000 population over the past two decades, compared to 1 to 2 for countries such as Canada, Australia, the U.K., and countries in western Europe.
Prices start at $12
Prices start at $11
Prices start at $12
The other is that the U.S. provides data on both the murder rate and enforcement of drug prohibition over a fairly long period of time. The high U.S. murder rate is frequently attributed to the high rate of gun ownership in the U.S., relative to most comparable nations. One problem with that explanation is that, while it is true that there is a significant correlation in international comparisons between gun ownership and murder rates, that correlation is driven by a single observation–the U.S. Regressions with the U.S. omitted show much weaker results, despite the existence of other countries with relatively high gun ownership rates–and without anomalously high murder rates. A second problem is that the behavior of murder rates over time, both in the U.S. and elsewhere, does not seem to be closely linked to gun ownership or legal restrictions thereof.
That suggests the possibility that U.S. murder rates are due to something other than gun ownership, with the gun ownership rate either unrelated to or consequence rather than the cause of, the murder rate. Since U.S. drug prohibition, while similar on paper to the laws in most of the countries it is compared to, is much more strongly enforced, it provides a possible explanation. Professor Miron has attempted to investigate the relationship between violent crime and drug law enforcement across countries, in work that is not yet published but is available on the web. While the results are consistent with the U.S. results, the evidence is very much weaker, in part perhaps because of the lack of good data to measure drug law enforcement across countries. The purpose of this paper is to investigate not the statistical evidence but the economic theory. I consider the various causal mechanisms by which drug use and drug prohibition might be linked to violent crime and try to determine, in each case, how various enforcement mechanisms might be expected to influence violent crime through that mechanism.
One reason to do so is to provide the theoretical framework for further statistical work–to suggest ways in which analysis of the relevant data might tell us more about exactly how drug prohibition and violent crime are related. A second reason is to provide a framework for considering what the effects of different enforcement strategies might be expected to be, and why. The paper consists of this introduction, three main sections, and a conclusion. The first section sketches the possible causal relations between drugs and violent crime. The second briefly sketches possible enforcement strategies for preventing drug use. The third combines the previous two and attempts to analyze the effect that alternative strategies might have on violent crime via each of the causal relations.
Part I: How Drugs Might Influence Violent Crime. Broadly speaking, the link between drugs and violent crime could occur in three ways: violent crime by consumers of drugs, violent crime associated with the production and distribution of drugs, or violent crime directly associated with the attempt to enforce drug prohibition. For the case of crime by drug consumers, two mechanisms are commonly asserted, with opposite implications. One is drugs as in input to violent crime–people committing crimes under the influence of drugs that they otherwise would not commit. This claim is made both for drugs for which it is pharmacologically implausible, such as heroin and marijuana, and for ones for which it is plausible, such as alcohol. If it is correct, the obvious implication is that drug prohibition, by reducing consumption of drugs, can be expected to reduce violent crime.
The other claim is that drug users commit crimes in order to get the money to pay for drugs. If that is correct, the effect of marginal changes in enforcement is theoretically ambiguous. Making drugs more expensive increases the expenditure of drug users per unit of drug consumed, but decreases consumption, so the net effect on the expenditure of drug users depends on the elasticity of demand. If, however, as is widely believed, the price of currently illegal drugs would be very low if they were legal, then the effect of legalization via this mechanism is unambiguous, since consumption has an upper bound set by non-pecuniary constraints. A heroin user who maintains his level of expenditure on heroin when its price falls by a factor of a hundred will no longer be able to commit crimes to pay for his habit because he will be dead.
Violent crime by people involved in the distribution network for drugs might also come about by a variety of mechanisms. One possibility is that violence occurs because people in that industry have wealth in highly portable forms–drugs and cash–which make them obvious targets for theft or robbery, and since calling the police is not a practical option they must use private violence instead to protect themselves.. A second, suggested by Jeffrey Miron, is that violence occurs as a form of dispute resolution among people who cannot use legal channels because their disputes are occurring in an illegal industry.
A third, and rather different, possibility is that violent crime represents rent-seeking in the competition among suppliers. Suppose, as much anecdotal evidence suggests, that drug distribution often occurs through local monopoly providers. Their profits depend in part on the area they control. So we would expect competition between adjacent firms for territory. One form such competition might take would be violence–by agents of one firm against agents, or possible customers, of another. The final source of violence is the enforcement of drug prohibition. Part of enforcement is arresting people, seizing drugs, and similar activities–all of which carry with them the risk of a violent confrontation between law enforcement agents and people who they suspect of violating the drug laws.
Part II: Possible Strategies for Reducing Drug Use. There are a variety of ways in which a government might try to reduce the use of illegal drugs. Roughly speaking, they can be categorized as ways of reducing the demand for illegal drugs, ways of reducing the import of illegal drugs, and ways of reducing the (domestic) production and distribution of illegal drugs. One way of reducing demand is by making substitutes, such as methadone for heroin addicts, more readily available. A more extreme version of that approach would be to legalize some drugs in order to reduce the demand for others. A different approach is to subsidize drug treatment centers; whether that works depends on whether drug users are actually helpless addicts who would quit if they only had a little help, or rational consumers choosing to use drugs because they like the effects they produce. Another way of reducing demand is by enforcing drug laws against users–spending law enforcement resources on identifying consumers of illegal drugs, prosecuting them, and punishing them.
A weaker version of this approach is to make drug use more costly by encouraging drug testing by employers. What about discouraging the import of illegal drugs? This strategy might take a variety of forms, ranging from more careful customs inspection to waging war against producing nations–all of which involve activities either outside the U.S. or on the border. From the standpoint of effects inside the U.S., any such policy has roughly the same effect–it increases the cost of drugs to the distributors. The final alternative is one that appears to consume a large fraction of the domestic law enforcement resources devoted to the war on drugs in the U.S.–actions against domestic production and distribution. Here it is useful to distinguish between actions against small-scale domestic producers, including those producing for their own use and attempts to identify, arrest and prosecute people in the business of mass-producing and/or distributing illegal drugs.
Part III: Economics of Drug Prohibition. In this section, I will try to look at each of the mechanisms discussed in part I of this essay, and see what we can say about the likely effect on the violence of the various enforcement approaches, assuming that only that mechanism is involved. Violence by Consumers. The most plausible mechanism linking drug use to violent crime, and one routinely observed in the context of alcohol use, is as a side effect. People who are drunk often have less control over themselves than they would when sober, so quarrels can become violent. In terms of economic analysis, this means that there is the desired output–the pleasure from alcohol consumption–that is produced by two costly inputs. One input is alcohol, the other is a risk of violent confrontation and the associated costs. If we assume a production function such that a given amount of alcohol necessarily produces a certain amount of pleasure and a certain amount of risk, then increasing the cost of alcohol via prohibition unambiguously reduces consumption, hence reduces risk, hence reduces violent crime.
The same argument would apply to any other drug with similar effects. Risk, however, is not a function only of consumption, since there are precautions that consumers can take to reduce the risk of violence, such as consuming their alcohol alone or with friends with whom they are unlikely to quarrel, rather than in a rowdy bar. Casual evidence suggests that much of the low-level violence associated with alcohol use is actually viewed by consumers not as a cost but as a benefit, and occurs for that reason. People go to a bar to get drunk and have fights. To the extent that violence from the use of alcohol (or other drugs) depends on choices other than whether to consume the drug, it can be reduced by enforcement mechanisms that target the violence rather than the drug use. Doing so reduces violence in two ways. Most obviously, it makes it in the interest of people who consume alcohol to do so in contexts where consumption is unlikely to lead to violence.
Less obviously, it makes it in the interest of people not to consume alcohol at all, because doing so may result in violence, which may result in punishment. Consider a very simple model in which one act of drinking leads to a ten percent probability of committing assault, in which all assaults are due to alcohol, and in which the only cost of law enforcement is the cost of imposing punishment, and in which everyone is risk-neutral. We can spend law enforcement resources imposing an expected punishment of Pd for drinking or Pa for committing assault or some mixture thereof. Suppose we decrease the penalty for drinking by one dollar per act of drinking and spend the punishment costs saved on increasing the penalty for assault. Since each act of drinking leads to a tenth of an assault, we can increase the penalty for assault by ten dollars. Someone deciding whether to drink now faces one dollar less expected punishment for drinking, but he also faces a .1 chance of committing assault, for which the punishment has increased by $10. So the combined expected cost of the penalties is unchanged.
Now modify the model by allowing drinkers to reduce, at some cost, the chance that they will commit assault. It is then straightforward to see that the same expenditure on punishment will produce a greater reduction in assault if used to punish assault than if used to punish drinking. Why? Individuals who drink, faced with penalties for assault, will bear some cost, call it Cp, in precautions against committing assault. The result will be to reduce the probability of assault–say to .05 instead of .1. The legal system then imposes an expected penalty of (say) $100 for assault instead of an expected penalty of $5 for drinking; the enforcement cost is the same either way since the smaller penalty must be imposed on twenty times as many offenses. With a penalty of $5 for drinking, the cost to the individual of drinking is the price of the drink plus $5 in expected penalties for drinking.
With a penalty of $100 for assault, the cost of drinking is the price of the drink plus $5 in an expected penalty for assault (.05 chance of an assault, which leads to an expected penalty of $100 if it happens) plus Cp in cost of precautions. So the total cost of drinking is now higher than before, leading to less drinking. But, because drinkers are taking precautions, the probability that a drink will lead to an assault is also lower than before. We are spending the same amount on enforcement and reducing the number of assaults–in the example, to less than half what it was under the previous strategy. [Add more explicit mathematical treatment of this?] This is not a new argument, merely a new application of an old argument for one advantage of ex post punishment, punishing the undesirable output, over ex-ante punishment, punishing one of the inputs to that output.
I have demonstrated the result for a simple model, but it holds more generally. It is not, however, true in all circumstances. It would not be true if the cost of imposing penalties on someone guilty of assault happened to be substantially higher than the cost of imposing penalties on someone guilty of drinking. An obvious example is a tax on alcohol, since that both penalizes drinking and brings in revenue. And one could imagine circumstances in which prohibition of alcohol penalized drinking at a lower cost per unit punishment than laws against assault penalized assault. Another reason the result might not hold is that not all violence is due to alcohol. If much violence is due to other sources, and if for some reason the demand for such violence is relatively inelastic (the amount of violence does not fall very much as we increase the penalty), while the demand for alcohol is relatively elastic, then a policy of punishing assault rather than drinking wastes most of the enforcement resources on punishing people who will not be deterred by punishment (people whose violence is not due to drinking). In this situation, resources devoted to punishing drinking can produce a greater reduction in violence than the same resources devoted to punishing violence.
These arguments suggest that if the objective is to reduce violent crime, there is a presumption, although a rebuttal presumption, that drug prohibition is an inefficient way of achieving that objective–that one can get a greater reduction at the same cost by targeting violent crime directly. Of course, drug prohibition may have other objectives as well, in which case the conclusion, although interesting, does not settle the question of whether we should have it or how strongly we should enforce it. We are then left with the observation that if violent crime is occurring as a side effect of the consumption of a drug, reducing that consumption via enforcement of the prohibition of that drug can be expected to produce some benefit in the form of a reduction in violent crime. This result must be qualified once we consider a world of multiple drugs. Different drugs are to some degree substitutes for each other. Different drugs are likely to have different side effects; some (alcohol) may make violence more likely, some (heroin) less. Enforcement of prohibition of the latter sort of drug may result in a substitution of the former sort and thus an increase in violent crime.
In a world of many alternative drugs, the argument for targetting violent crime rather than drug use becomes stronger, because one of the “precautions” that users can take in order to avoid committing violence and being punished for it is shifting to a drugless likely to produce violence. Reducing or eliminating enforcement of drug prohibition makes that more likely to happen, since it increases the range of alternatives faced by the user, and increases the incentives for producers to create and market drugs less likely to produce violent behavior. So far I have been discussing violence by consumers of drugs as a side effect of drug usage. Similar arguments would apply to the closely related case where the violence is deliberate and the drug use facilitates it–where drugs are in input to (desired) violence, rather than violence an (undesired) effect of drug use. Here again, there are grounds for a presumption that targetting the violence reduces it more than targetting the drug use, but the presumption is again rebuttable.
The general result is that any enforcement strategy that increases the cost to users of drugs whose use is associated with violent behavior can be expected to reduce such behavior, although it is likely to be a less cost-effective way of achieving that particular objective than directly targetting the behavior. We are left with the case of violence committed by users in order to obtain money for drugs. To the non-economist, this seems like an obvious and plausible scenario. For economists, the situation is not quite so clear. The problem is the central economic assumption of rationality. If mugging people produces a higher income for me than alternative occupations–driving a cab, say–then I ought to be mugging people already, whether or not I need the money for drugs. There are, after all, plenty of other things to spend money on. That argument suggests that the existence of expensive but desirable drugs should simply result in an increase in effort. I have more uses for money, so I work harder to earn more. If my best paying activity is driving a cab, I work harder at that; if my best paying activity is mugging people, I work harder at that.
The result is an increase in mugging with increased expenditure on drugs only if there is some reason why drug users are already supporting themselves by mugging people. To get a stronger result than that, we need stronger assumptions about the production functions associated with alternative ways of making money. We might assume, for example, that the cost of earning more money driving a cab is more time, and that there is an upper limit to how many hours a day you can drive without being too tired to do so safely. We might also assume that part of the cost of earning more money mugging people is an increased risk of getting killed by one of your victims. That risk you can increase as much as you want, at least up to the point where you actually do get killed. The more valuable money is to you, the greater the risk you are willing to take to get more of it. Generalizing the argument, we observe that an increase in the value of money to the worker will result in a greater increase in output in some activities than others.
If a violent crime happens to be an activity which people shift into when money becomes more valuable to them, or if people who consume illegal drugs happen to be people who in any case support themselves by violent crime, we would expect increased expenditure by consumers of drugs to lead to an increase in violent crime. Under such circumstances, the effect of increased drug prohibition on violent crime depends on the elasticity of demand for the prohibited drugs. If demand is elastic increasing the price results in a decrease in expenditure, hence a decrease in violent crime. If demand is inelastic increasing the price results in an increase in expenditure, hence an increase in violent crime. A further point worth mentioning is the relation between the cost of drugs and the value of leisure. Most forms of drug use tend to reduce productivity in most income-earning activities. Most ways of earning an income tend to reduce the pleasure from consuming most drugs.
Hence there is a tendency for drug use to be a leisure time activity. To the extent that it is, lowering the price of drugs increases the value of leisure to drug users and so reduces their willingness to trade leisure for income. If the user happens to be a mugger, that means less time spent mugging people. This effect strengthens the argument in the case where drug demand is inelastic. If stricter enforcement drives up the price of a drug, consumers spend more money on the drug, increasing their need for money and willingness to work, and have less of the drug, decreasing their enjoyment from leisure time. Both effects result in more muggings if the user happens to be a mugger. The effect weakens the argument in the case where drug demand is elastic, since then increased enforcement reduces the need for money but also the value of leisure. What about the effect of efforts to reduce the demand for illegal drugs, whether by arresting consumers, providing treatment programs, or legalizing substitutes? A reduction in demand should unambiguously reduce user violence, whether it comes as a side effect of drug use or as a means of obtaining money to pay for drugs.
Violence by Distributors. Violent crime due to distributors of illegal drugs might occur through at least three different mechanisms: Violence associated with attempts to steal and defend valuable assets, such as cash and drugs, violence associated with dispute resolution, and violent competition for territory. The first case is the easiest. Suppose increased enforcement drives up the cost of drugs to distributors. The price of drugs rises. If demand for drugs is inelastic, total expenditure on drugs goes up; if demand is elastic, it goes down. The total value of drugs and cash held by people in the distribution industry will be roughly proportional to the total value of drugs being sold. The amount of violence associated with attempts to seize and protect those drugs and cash will be an increasing function of their value. So we would expect violence due to this mechanism to increase with increased enforcement effort if the demand for drugs is inelastic, and decrease with increased enforcement effort if the demand for drugs is elastic.
The conclusion becomes more complicated if we consider alternative forms of prohibition effort. Suppose, for example, that almost all of the effort is devoted to preventing the import of illegal drugs; once the drugs are inside the country, domestic dealers are free to do as they like with only minimal risk of arrest. Since there is little risk, there is little reason to devote effort to concealing activities from the police or minimizing the amount of evidence lying around. But the relatively open nature of the illegal activity should make it easier for other criminals to observe it and attempt to profit by hijacking drugs and cash. Now imagine that law enforcement switches its effort from import to distribution, while keeping the overall effect, as measured by the street price of drugs, unchanged. Drug dealers become more careful, drug hijacking is reduced, associated violence falls. Finally, imagine that law enforcement itself goes into the hijacking business–as, via civil forfeiture, it has done on a large scale in the U.S.
The incentive to take precautions becomes even higher for the drug distributors, and the amount of hijacking and violence (not counting violence by or against law enforcement agents) decreases. There is one more factor that remains to be considered. For private protection of property as for public protection, the preferred outcome is not conflict but deterrence. If I can persuade all potential hijackers that if they try to steal from me they will die, nobody will try to steal from me and nobody will be killed. In a well-functioning private system, as in a well-functioning public system, people are protected mostly by the threat of violence rather than by actual violence. How well such a system works depends to a considerable extent on the stability of the industry; the longer my firm is in business, the more opportunities it has to create commitment strategies and build a reputation. Hence to the extent that enforcement strategies successfully target and destroy well-established criminal firms, increasing the instability of the industry, they are likely to shift private protection from threat to violence, increasing the total amount of violent crime.
So far I have been considering violence associated with attempts to seize and protect property unprotected by the law. What about violence associated with ordinary business disputes–the sort of disputes that, in other industries, end up being settled by courts or private arbitration? Here again, we would expect (as Jeffrey Miron has suggested) that law enforcement efforts that reduce the stability of the industry and increase the information costs of its members in dealing with each other would tend to increase violence. We would also expect violence to increase with the amount at stake in such disputes. One determinant ought to be the total value of drugs being sold, since at least some disputes will be associated with such transactions; to that extent, the conclusion will be the same as in the previous case. A second determinant might well be the total profit of the industry, since profits are what are at stake in some contract disputes, such as disagreements over market sharing agreements, or mergers, or the like.
Enforcement effort aimed at the import of drugs should raise the input cost of the industry, hence unambiguously reduce profits. Enforcement effort aimed at preventing small-scale production, on the other hand, and especially production by consumers for their own use, may well result in increasing the demand for the services provided by the drug distribution industry, and thus increasing profits. Enforcement efforts directed against the distribution industry itself have somewhat more ambiguous results. To the extent that they simply raise costs, they can be expected to reduce profits. Suppose, however, that enforcement has a much larger effect on new entrants to the industry, who have not yet built up the necessary network of trust and expertise and the necessary portfolio of corrupt police officers and judges, than on existing firms.
In that case, it might increase profits, just as a tax on new construction of housing might increase the revenue of apartment owners by more than it increased their costs. If so, it might also increase the amount at stake in contract disputes and the associated violence. On the other hand, since such enforcement would tend to reduce industry turnover, it might also reduce violence for reasons discussed earlier. Third and last, consider the possibility that, for some illegal drugs, the usual industry structure consists of local distribution monopolies, and that the violence associated with distribution is a form of rent-seeking as firms compete for territory–turf wars. In that scenario, anything that decreases industry profits ought to decrease the stakes in such conflicts and so decrease the amount of violence. Hence we would again expect to see increased effort against imports associated with reduced violence and increased effort against home production associated with increased violence. [Add formal mathematical treatment of equilibrium here, a la monopolistic competition with violent conflict included?]
What about increased enforcement effort against the distribution network? Here again, the conclusion is ambiguous. On the one hand, such an effort should make distribution less profitable, reducing violence. On the other hand, the violence is a conflict over monopoly profit, and the degree of monopoly may well depend, among other things, on the level of enforcement. One major source of economies of scale in the industry of distributing illegal drugs may be corruption–and owning your own police chief or judge is more valuable the more energetic the enforcement that he protects you against. [More analysis needed here?] What about the effect on violence by distributors of policies that reduce the demand for illegal drugs? Here again, the effect appears to be unambiguous. Reductions in demand reduce revenue and reduce profit, hence should reduce violence via any of the mechanisms I have discussed.
Violence in Enforcing Prohibition. Violence may be due to drug consumers, it may be due to drug distributors fighting among themselves or with other criminals. It may also be due to conflicts between drug distributors and law enforcement. The most obvious determinant of the amount of such violence is the amount of law enforcement directed against drugs within the U.S. This should include both efforts against the distribution network and efforts against small-scale and home production since either can lead to violent conflict. In addition, one would expect that the willingness of drug distributors to employ violence against law enforcement would depend on the amount at stake and so increase with both total revenue and profit. Similarly, to the extent that law enforcement efforts are aimed at seizing property, either for civil forfeiture of private theft, one would expect such efforts to increase with the value of the property available to be seized.
So policies that raise the cost of bringing drugs into the country ought to increase such violence if demand is inelastic, decrease it if demand is elastic. Here again, one might expect well-established firms to successfully protect themselves, either by bribing law enforcement or by developing networks of trust. If so, enforcement efforts that destabilized the industry could be expected to increase the level of violence. As in the previous cases, efforts that reduce demand for illegal drugs ought to reduce both revenue and profit, hence reduce violence. The one exception is demand reduction via law enforcement efforts targeted at users. Such efforts might result in a violent conflict between law enforcement agents and suspects.
Conclusion. My purpose in this essay has been to try to sketch out the possible mechanisms relating illegal drugs to violent crime, and how various enforcement strategies might affect each. The clearest result is that policies that reduce the demand for illegal drugs can usually be expected to reduce the violence associated with the sale and use of such drugs. Policies that increase total revenues or total profits can generally be expected to increase violence, policies that decrease them to decrease it. Policies that decrease the stability of the illegal distribution industry are likely to increase violence. Generally speaking, increased enforcement of prohibitions on imports can be expected to decrease profits; increased enforcement of prohibitions on home products can be expected to increase it.
Two more general points are worth making. The first is that, if one regards reductions in drug use as desirable, the associated violence is not entirely a bad thing. Much of it is among people involved in the illegal distribution of drugs. The resulting risk is a cost for that industry, and so raises the price and decreases the consumption of drugs. That effect must be balanced against the risk that violence creates for bystanders and the costs of violence committed against outsiders. The final point is to observe that my discussion has been aimed almost entirely at the effect of marginal changes in the enforcement of drug prohibition. If we consider instead the effect of shifting from prohibition to legalization, the results are much more straightforward. With one exception, legalization eliminates all of the sources of violence I have been discussing.
That exception is violence by drug users as a side effect of drug use. Legalization can be expected to increase drug use, hence it could well increase such violence. While it could increase it, it could also decrease it–for two reasons. The first, as noted earlier, is that different drugs are substitutes for each other. Legalization would improve both information and availability, making it easier for users to select drugs with fewer undesirable side effects–including the side effect of causing violent behavior. The second reason is that, in an illegal market, quality is likely to be more variable than on a legal market, making severe unanticipated effects, including violent effects, more likely. [Should I add formal models of both the competitive and the monopolistic competitive industry models to this, perhaps as an appendix?] Nils Christie, “Conflicts as Property,” British Journal of Criminology 17(1) The very beginning of the graph is probably unreliable since in the first few years only a fraction of the states were providing data. Alcohol prohibition began in … and ended in … .  Jeffrey A. Miron, “Violence and the U.S. Prohibitions of Drugs and Alcohol,” American Law and Economics Review, 1, Fall 1999, 78-114.  Both of Miron’s papers are available from his home page at http://econ.bu.edu/miron/.
“Violence, Guns, and Drugs: A Cross-Country Analysis,” the unpublished piece, measures drug enforcement by drug seizures. It finds a strong relationship between enforcement and murder rates, but the relationship is mostly driven by a single case–Columbia, which has a murder rate of …, … times higher than the U.S. Since Columbia is a major producer of illegal drugs, the high seizure rate cannot be taken as a good measure of enforcement effort; there are, after all, a lot more illegal drugs there to seize than almost anywhere else.  But see Jeffrey A. Miron, “´Do Prohibitions Raise Prices: Evidence from the Market for Cocaine,” (unpublished, on his web page) for some evidence to the contrary.  Of course, the number of users might increase, but given the non-pecuniary costs of drug use, it is hard to see how they could increase rapidly enough with a fall in price to keep the demand elastic at low prices. For some evidence on this point, it would be interesting to investigate the fraction of people in places where marijuana is legal who smoke it–or usage rates for various drugs that were legal early in this century. Is there data?
 This fits the description in The Cocaine Kids, ….  The economics of such a situation are similar to those of the competition of nations for territory and the associated tax base described in David Friedman, “An Economic Theory of the Size and Shape of Nations,” JPE … . In that paper, however, my concern was with the equilibrium outcome, not the costs of getting and maintaining it.  Presumably, precautions would take forms such as drinking at home when it was more fun to drink at a bar, or locking up a gun and giving someone else the key before drinking, or merely exercising self-control while drunk, at some costs in enjoyment. While none of these precautions has a cost paid in money, each of them is, considered as a cost, equivalent to some amount of money.  This issue is discussed at some length in David Friedman, Law’s Order, Chapter 7, pp. 74-83.
 Presumably, prohibition makes sense only when you want to impose a cost higher than any tax you can collect; putting the argument differently, a tax so high that nobody pays it and all alcohol is smuggled is equivalent to a prohibition.  Note that the demand for violence by people who are violent because of drinking cannot be less elastic than their demand for alcohol, because of the argument we have just sketched. But the total demand for violence can be if most of it comes from other people who are harder to deter.  This argument requires, of course, that there are no legal activities that provide similar opportunities to accept risk in exchange for money and provide a more attractive way of earning money than illegal activities.  Grossman, Michael, “The Economics of Substance Use and Abuse: The Role of Price,” April 2000 provides empirical evidence of on-demand elasticity for a variety of addictive drugs. He concludes that demand elasticity may well be greater than 1, meaning that total expenditure may fall as price rises. Becker, Gary S., Michael Grossman, and Kevin Murphy, “An Empirical Analysis of Cigarette Addiction,” American Economic Review, 84 (no. 3): 396-418, June 1994 find a long-run price elasticity for cigarettes above .7.
 Drug use by musicians may be an exception.  This is a point raised by Jeffrey Miron in a closely related context.  In equilibrium in a perfectly competitive industry, economic profit is zero. The profits I am considering here need not be economic profit in that sense. They include returns to sunk costs in human and organizational capital (strictly speaking, quasirents), monopoly profits, returns to specialized human abilities in scarce supply, and the like.  The obvious case is marijuana growing.  The Last Testament of Lucky Luciano claims to be based on first-hand information from one of the leading criminal entrepreneurs of the prohibition period. While there seems to be no way of confirming the author’s claim, I suspect it is true on internal evidence–the picture presented of the illegal market appears economically plausible. One interesting feature of the account is that criminal firms which were otherwise independent appear to have pooled assets for the purpose of purchasing the services of corrupt judges and law enforcement agents, suggesting that that was, perhaps the major source of economies of scale in that industry.
 Are there any sources for data on this? In 1995, 131 law enforcement agents were killed in the line of duty (stat abstract). Figures go back to 1980; there is no clear pattern, although the 1980 figure is high.  I should perhaps add that I do not regard reductions in drug use as inherently desirable. Following out the usual assumptions of economics, I assume that drug users, like other people, are rational, and so tend, on the whole, to make the decisions that best serve their interest. Hence their decision to use drugs is evidence that doing so, on the net, benefits them. That implies that it is more generally desirable, except to the extent that it imposes costs on others. Since most such costs at present are the result not of drug use but of drug prohibition, I see no reason to regard reductions in drug use as an unambiguously desirable outcome.