8 THE MARKET FOR HEROIN BEFORE AND AFTER LEGALIZATION
Robert J. Michaels
In the space of one hundred and seventy-six years the Lower Mississippi has shortened itself two hundred and forty-two miles. That is an average of a trifle over one mile and a third per year. Therefore, any calm person who is not blind or idiotic, can see that in the Old Oolitic Silurian Period, just a million years ago next November, the Lower Mississippi River was upward of one million three hundred thousand miles long, and stuck out over the Gulf of Mexico like a fishing-rod. And by the same token any person can see that seven hundred and forty-two years from now the Lower Mississippi will be only a mile and three-quarters long, and Cairo and New Orleans will have joined their streets together, and be plodding comfortably along under a single mayor and a mutual board of aldermen. There is something fascinating about science. One gets such wholesale returns of conjecture out of such a trifling investment of fact.
If Mark Twain were alive today, he could probably be found studying either drug abuse or economics. Anyone wishing to predict the consequences of legalized opiates must first invest in some facts. The historical, biological, and statistical sources, however, yield little in the way of verifiable facts or properly constructed data. The conclusions typically drawn after a reading of this literature vary about as widely as the alleged facts and are frequently derived without the aid of elementary logic. In this paper I deal with the inferences about legalization that can reasonably be made on the basis of what is known about the market today. My mode of reasoning will be that of an economist.'
Economic analysis can help us to predict the qualitative outlines of a world in which heroin is legal or access to it is granted more liberally. If we wish to predict the outcome quantitatively, we clearly need numerical data about the present situation, summarized into relevant conceptual categories. In addition, we would probably like to have data at our disposal from reasonably similar societies or from adjacent historical periods that were characterized by different opiate regimes. Only with such data could we have much confidence in our estimates. For example, economists were able to predict the consequences of nationwide airline deregulation quite accurately from studies of the previously existing unregulated intrastate air travel market in California.2 In this case, accurate numbers were available, and one could reasonably assume that an unregulated market in Ohio would function much like the one in California.
As we will see below, none of the conditions for plausible inference are met in the case of heroin. Presently available "data" are produced almost exclusively by the federal government, largely for political purposes. Some of the data are attempts to quantify categories that are impossible to define, such as "addicts." Those data that are attempts to measure objective quantities (e.g., the purity of street heroin) are grossly inaccurate.' Heroin is traded in a market whose current characteristics are largely unknown, or known only through the self-serving and unverifiable testimony of captured or disillusioned participants. Patterns of use in other societies and at other times differ so drastically from those around us that even conjecture seems futile. Despite these formidable problems, we can draw some nontrivial conclusions about legalization.
In the remainder of the introduction, I outline the economic ideas a researcher would use in order to analyze legal and illegal heroin regimes. In Part II, I apply them to the supply side of the heroin market. The analysis depends heavily on recent theories of vertical integration, transaction costs, and information search. I will argue that current structural characteristics of the market stem largely from the illegal nature of the good traded, and that taken by themselves they give little guidance as to what a legal market would look like. This analysis of market structure also allows us to evaluate folklore about monopoly control of the market in the present and future.
In the foregoing analysis of the market, I will have assumed that buyers' preferences for heroin are qualitatively the same as for any other good. As will be seen in Part III, the stereotypes of addiction bear scant resemblance to actual use patterns. There I evaluate what is known about the elasticity of demand and the stability of demand for heroin. In reality, the population of users experiences both in-and out-migration, and habits of use are responsive to market forces. Even if we know enough about the response of users to price changes, however, we are also faced with a problem in which economic analysis appears to be of little value: The characteristics of user populations vary greatly over time, geography, and legal regimes, in ways that apparently cannot be predicted by a simple market model.
In Part IV, I explore the statistical literature on the size and scope of the market. Official statistics on "addicts" or "users" are invariably useless and frequently inconsistent with related data. They are typically constructed in an ad hoc and illogical manner, varying strongly in response to political exigencies. Little can be done to improve them. As such they are likely to be of no use as benchmarks from which to predict the consequences of legalization.
As shown in Part V, the relationship between heroin and crime is not an easy one to determine. Understanding it requires an explo-ration of what is known about the sources and uses of user incomes, along with information on the elasticity of demand for heroin. Con-trary to official statistics, few users are currently unemployed in any economically meaningful sense. The hope that legalization will sig-nificantly reduce urban crime, however, is probably misplaced. In Part VI, I summarize the paper and draw some conclusions.
What We Need to Know
In the competitive market of elementary economics, price and output are determined by the interactions of large numbers of anonymous suppliers and demanders of a homogeneous good. In this market, buyers and sellers exchange information on their personal valuations of the good and strike deals at mutually agreeable prices. Because people on both sides of the market are assumed to be well informed about the behavior of other traders, price quickly converges to its equilibrium value. To what extent can we apply this paradigm to heroin?
In any market, buyers make their decisions about the amounts they will purchase on the basis of their wealth and the price of the good in question relative to the prices of alternatives. At higher prices for our good, alternatives typically come to look more attractive, and fewer units of our good will be purchased. Certain goods, such as heroin, are sometimes alleged to violate this principle, known as the "first law of demand." Because it is "addictive," some claim that the quantity purchased by users will not vary with price and that rises in price will be accompanied by increases in crime as users at-tempt to acquire sufficient income to support their need for the substance.
The first important piece of information we need is thus some data on the responsiveness of quantity purchased to price, as summarized in the economist's concept of elasticity of demand. A demand that is completely unresponsive to price has an elasticity of zero; as re-sponsiveness to a given price change becomes proportionally greater, elasticity (formally defined below) takes on a larger value. To eval-uate legalization, we also need to know both short-run and long-run elasticities of demand. A user may be "addicted" and unresponsive to price over the short run, but in the long run (i.e., after information is acquired and adjustment costs incurred) he may find substitute drugs (methadone), lower his dosage, or learn abstinence. Similarly, a lower price may fail to induce much additional consumption by preexisting users, but new users may be attracted by it as time passes.
We also need to know about other influences on demand. For example, if incomes of purchasers increase, how will they respond as buyers of heroin, other things (e.g., its price) being equal? If the prices of substitute drugs or other enjoyable items fall relative to the price of heroin, how will buyers alter their consumption patterns? The foregoing are, respectively, questions about the income elasticity and cross-elasticities of demand. In addition, we need to know about the ways in which demand varies due to nonmarket factors ("tastes"). Do users "burn out" or drop their habits with increasing age?4 Does heroin use display massive unpredictable fluctuations that might be called "fads" or "epidemics"?'
The analysis of supply in a market is ultimately the analysis of production technology and the availability of resources. Additional units of a good will be produced in a competitive market if the price for which they can be sold exceeds the cost that supplying firms must pay in order to induce resource owners to leave other activities. Higher prices will accompany the expansion of production if there are resources for which supply limitations are of importance. Thus, in order to evaluate legalization, we would like to know about the production technology, limitational resources, and the behavior of costs as output levels change. Think of legalization as reducing the net price that must be paid by buyers and increasing the net price received by sellers, i.e., the analogue of removing a tax. If costs increase rapidly with industry output, there will be little change in quantity produced as we move from the illegal to the legal equilibrium, since few additional units will be producible profitably. Additionally, we are interested in the possibility that technological innovation or learning from experience will reduce productiori costs as time passes.
Another important question on the supply side concerns econ-omies of scale and scope. Regarding the first, we are asking about the extent to which average costs of a supplier's production decrease with increased output. If economies of scale are quickly exhausted, individual firms will be small relative to the market. If such econ-omies are extensive, competition among sellers might break down because too few of them remain. Alternatively, suppliers may be few because there are "barriers to entry" of newcomers, such as those stemming from monopoly control of an important resource. For ex-ample, some have alleged or assumed that control at the highest level of heroin supply is in the hands of a single "organized crime" en-terprise.6 Is this the case because both legal and illegal variants of the business would be characterized by extensive economies of scale, or is the enterprise an inefficient (high-cost) firm that maintains its monopoly through extralegal methods? Whether legalization implies competition or monopoly depends in part on the nature of scale economies.
We are also confronted with the question of evaluating economies of scope. Consider a good whose raw materials require processing and transport before they become useful to ultimate buyers. The activities may either take place within a single vertically integrated enterprise, or a series of transactions in the market may move the good from specialist to specialist. The degree of integration that arises in the market will depend on the relative costs of the two ways of doing business. These costs include the costs of finding trading partners, and negotiating and enforcing agreements. Traders in an uninte-grated situation may have an incentive to behave "opportunistically" if the returns from breaking a previously negotiated contract are high enough relative to the cost of enforcing it.' In the illegal heroin market, there might be incentives to disintegrate production. A primary supplier who would also do his own retailing in a legal market may find that illegality raises the cost of his own retailing so substantially that he contracts with others to do it. These contracts, however, may be more subject to opportunistic breach in an illegal market. Again, the shape of the firm and the market may change greatly under legalization.
In summary, supply and demand provide the central organizing principles for the analysis of the heroin market. The effects of legalization will be on both the size and the organization of the market. Both demand and supply determine price and output, and their characteristics must be jointly evaluated.' I turn first to supply.
SUPPLY IN THE HEROIN MARKET
In the heroin market, reliable information about both the internal structure of supplying units and the relationships among these units is hard to come by. The same forces that shape legal firms and their interactions should nevertheless be present here, despite the fact that illegality contrains their actions in certain important ways.
Vertical Relationships, Opportunism, and Transaction Costs
To organize a firm of any kind is not a trivial task. One can think of an entrepreneur laboriously shepherding various specialized productive inputs into a single organization. Alternatively, one can think of the firm as a set of resource owners who are bound together by elaborate sets of implicit and explicit contractual relationships, many of which are costly to negotiate or renegotiate. In an illegal firm, a complex organization will be more costly to achieve than in an equivalent legal enterprise, since the apprehension of a single component of such a firm may put an end to the entire operation. Thus the typical criminal "firm" consists of an independent solo practitioner, and "organized crime" consists of organizations with family or ethnic ties that are quite small relative to the size of legal organizations in the same or similar lines of business.9 Complex, multiperson production processes will be the exception.
Central to the rationale for a firm is the restraint it places on individuals' opportunistic behavior as regards physical resources or specialized information. An individual who supplies labor or other inputs to an illegal enterprise may have particularly strong incentives to act opportunistically, to the detriment of the firm, either by delivering goods or information to a competitor or by cooperating with enforcement authorities.") Deterrence of opportunism by illegal means within the firm is itself a risky activity as well." Firms that survive are thus likely to carry on only a limited range of activities because of these risks. They will be inefficiently small, relative to the size they would attain in an equally competitive legal market, and will engage in an inefficiently restricted set of stages of production and distribution. Because of the generally higher costs of illegal operations, the supply of and demand for heroin will be equated at a higher price and lower output than in a legal market.
Little reliable information exists regarding the structure of heroin supply, and what information exists deals only with New York City. There is agreement on the existence of a basic six-tiered system from importers to final dealers in that city,12 but it is unclear whether such a structure exists elsewhere, i.e., how the division of labor varies with the extent of the market. (Heroin is generally available to the knowledgeable in almost any part of the country.'3) The vertically unintegrated structure of the existing market is an efficient method of dealing with the risk of apprehension" at any stage of production and distribution, given that the activity is illegal." Any detained participant can supply only limited information about others, thereby lowering overall risk. Under legalization, one expects to see fewer stages, since in general there will be cost savings through the avoidance of additional middlemen. The stages are fewer than six between manufacturer and ultimate retailer for most legal consumable items, and the general trend in household goods has been toward a reduction in the number of stages. The chain grocer and drugstore displaced smaller independent retailers and jobbers in part because of cost advantages they were able to secure by acting as their own middlemen.
The cost of delivered heroin will likely fall for reasons other than a reduction in the number of redundant distribution stages. Illegality is equivalent to a per-transaction tax on a good, since the likelihood of apprehension increases with the number of such transactions one makes. Part of the transaction price thus consists of a premium for risk. As with any tax, some fraction of it will be borne by the buyer and some by the seller. Since the illegal market is frequently characterized by bilateral monopoly, economics offers us little guidance in identifying the gains to either side from lifting this tax. Even if there is a monopoly seller after legalization, we would expect to see price fall. After legalization, profits—at least in an account-ing sense—will be lower at any stage of the distribution chain. To the extent that they are above those obtainable on competitive investments of equal risk, they will be eroded by the investments of newcomers. As will be seen below, there are few likely sources of long-term supernormal profit in a legal heroin market, after adjust-ment for ordinary business risk.
Information and Reputation as Resources
The barrier to entry of newcomers that is of greatest importance un-der illegality stems from the ownership of resources for which it is costly to find alternatives in the short run. To the extent that such resources exist, they may generate longer-lived monopoly gains than would otherwise occur. In the heroin market, the most important of these resources is information, especially information about the identity and reliability of trading partners and the quality and pre-dictable availability of goods. Were such institutions of the legal market as advertising, brand names, etc., available, the value of this information and the associated monopoly gains would be less. In a legal market, contracts can be written and enforced against breach. In an illegal market, one must depend on reputation. Such a repu-tation is not easy to earn, and there is risk in spreading knowledge of that reputation among more than a handful of potential trading partners. The current market is thus characterized by highly person-alized trading relationships that are formed only after large amounts of search. Even the seemingly competitive final street markets are characterized by traders dealing predominantly with recognizable partners and by buyer loyalty to those sellers who are known to sell items of dependable quality.'6 The attenuation of informational flows shrinks the trading possibilities, while decreasing the riskiness of transactions actually made:7
In Nelson's terminology,18 heroin sold by a given seller is an "experience" good as opposed to a "search" good.19 An experience good is one that a purchaser can deem satisfactory or unsatisfactory only after actual use. (A search good is one for which unsuitable choices can be eliminated prior to actual purchase, e.g., a vehicle that I can determine from advertising to be too small for my needs.) It is in the experience-good situation that consumer uncertainty about quality is most easily resolved through branding and advertising of the brand name. While advertising in experience-good situations raises costs, it also increases the size of the set of choices perceived by an individual buyer and probably lessens the market power of an individual seller. Thus it has the potential to reduce price—and apparently does so in some situations.2° Heroin at any stage of illegal distribution is of widely varying quality, both in terms of potency and type of adulterant.2' One expects that legalization with advertising will stabilize quality and lower price relative to a market without advertising. More precisely, it would lead to a spectrum of qualities, priced roughly according to marginal cost. Given the understandable risk aversion of buyers, there is a good likelihood that both the price-lowering and quality-certifying effects of advertising will operate in ways that increase use.
The input mix chosen by those in the distribution chain would change noticeably with legalization in yet another way. It is estimated that between one-third and one-half of all pure heroin initially imported ends up being consumed in the distribution process rather than being purchased at street prices by final users.22 With heroin illegal, playing a part in distribution and thereby establishing dependable contact with a supplier is a method by which a user can cut the costs of searching for a source. Given that an addict middleman has few alternative sources of supply in the short run, he is less likely to turn in his supplier for a reward than a nonaddict who might otherwise be more efficient as a lower-stage agent. The observed method by which heroin is distributed thus reflects input choices and inventory wastage that are largely consequences of its illegal status. Distillers and alcoholics have little reason to seek employment relationships with each other. One expects the cost curve to fall after legalization because a given amount of heroin will be produced or transacted with a more competitive payment scheme and more productive personnel.
Under legalization, there will unquestionably be a radical change in the structure of heroin distribution. The nature of the illegal market is such that trading units in it are inefficiently small, restricted as to the scope of their operations, and dependent on personal rather than commercial channels for the acquisition and dissemination of information. Legalization will lower costs of both production and information and will break down informational barriers to entry and rivalry. Thus, one expects that the legal market will be characterized by a lower equilibrium price, higher output, and more dependable quality. There is no reason to believe that its distribution and retailing cannot be taken over by existing firms in, for example, pharmaceutical markets or that they cannot successfully be undertaken by new specialists. Can we be certain that this will be the long-run supply situation?
Competition and Monopoly in the Market
Despite its importance in the folklore of law enforcement, we know little about the role of "organized crime" as a possible monopolizer of the heroin market. In a production and distribution process that has well-defined "stages," a monopolist need only control one of them in order to extract all profit from the market. There is in general no gain to be made by expansion into other stages.23 If a mo-nopoly exists anywhere, it is presumably at the upper stages of production and distribution, where capital requirements might pose a barrier to entry and where transactions are highly personalized and infrequent.24 Note that even if there is a monopoly at some upper level, we need not see a reduced output relative to competition. Lack of contact between different sociological and geographic segments of the market may make resale among them difficult. In such cases, we might observe price discrimination based on differing elasticities of demand in the different submarkets. To the extent discrimination is finer, output will approach competitive levels.
Whether or not a monopoly or cartel controls some stage of the process depends on whose testimony one reads. Thus the assistant secretary-general of Interpol once stated before Congress that a truce among underworld elements had effectively cartelized the business.25 At about the same time the commissioner of the Federal Bureau of Narcotics reported to Congress that Cosa Nostra had altered its pol-icy and would henceforth not engage in the heroin business.26 On the bases of a shaky estimate of users and reports of transaction volumes determined from interviews with addicts,27 Moore estimated four to six major importers at the top of the pyramid in New York in 1973.28 Without identifying and observing them, however, we cannot infer whether they act competitively or collusively. Additionally, the factors that limit an importer's size have never been spelled out. Are there diseconomies of scale beyond those associated with the proliferation of illegal contacts?
We can probably infer that there is considerable turnover at the upper levels of the market and that it results from competition rather than law enforcement. Although the subject is partially folkloric, the sources of raw opium, locations of refining facilities, and ethnic identities of personnel, both inside the United States and abroad, have changed rapidly. The ethnic identification of arrested "Higher Echelon Narcotics Traffickers" during the 1950s and 1960s was overwhelmingly Italian.29 By the mid-1970s, Mexican heroin had largely replaced heroin of Middle Eastern origin in the eastern United States.3° The trade was apparently organized by Hispanics who had formerly operated only in the Southwest. During the Vietnam period, Southeast Asia came to compete with Mexico as a source.3' Finally, as the 1980s arrived, production from the "Golden Crescent" of Afghanistan, Pakistan, and Iran became a more important factor in the American market.32
It seems clear that these shifts in supply patterns and personnel abroad were accompanied by equally drastic shifts at upper levels of the domestic market. For other drugs, we have little evidence of sustained monopoly. According to anecdotal evidence, cocaine appears to be reasonably competitive at higher levels and is not dominated by any of the ethnic groups that at various times have allegedly controlled heroin. Marijuana and bootleg pharmaceuticals are clearly competitive at the primary sources of supply to the illegal market.
No other important input appears to act as a barrier to entry in the current production of heroin, and such barriers would be even less likely under legalization. As the drug trade has increased in volume and remained competitive as an illegal investment, financial capital has appeared, in part through retained earnings. The poppy crop required for America's estimated current annual consumption of heroin could be grown on 25 square miles of land," and the historical record of shifting sources makes it clear that usable land exists in many locations.34 There appear to be few relevant scale economies in processing opium into heroin, and no unusually skilled labor is required in production, transport, or distribution. Entry of a producer or distributor can probably be effected at only a single level of the chain, i.e., a processor need not also be a grower or distributor.
By the standards of ordinary manufacturing and distribution, a legalized heroin industry would have few of the elements likely to lead to monopoly or monopolistic behavior, and many likely to lead to competitive behavior. It is characterized by no discernible scale economies that would imply natural monopoly. Even when illegal, it has been characterized by few strong barriers to the entry of new producer or distribution units. There are no known inputs that are even in relatively inelastic supply. Costs in the industry are high today because of the risks imposed by illegality. Because there are no resources supplied inelastically to the industry, the costs of illegality are largely borne by users as higher prices. The inefficient distribution chains and input mixes are also thus borne. The price in a competitive legal market would settle at about marginal cost, adjusted for quality.
Currently used information about the characteristics of production and distribution is thus almost useless for predicting the long-run supply and price of legal heroin. According to one recent report," twenty 10-milligram tablets of heroin, legally prescribable for the terminally ill in England, cost "about $1.36 A liberal estimate of a typical addict habit would be 50 milligrams per day.37 Even if there is monopoly power on the part of drug producers, this comes to little more than 25 cents per day. Additionally, we do not know (but have reason to suspect) that large-scale commercial production will lead to cost-saving innovations. Such a figure gives one pause. The illegal market wastes incredible amounts of economic resources, apparently only because of illegality rather than monopoly. If the illegal price is $5-$15 for 10 milligrams of uncertain quality, it is clear that we need to investigate the responsiveness of demanders to the drastic changes in the market that legality would induce.
ADDICTION AND THE DEMAND FOR HEROIN
The preferences of buyers ultimately determine what will be supplied to them, and in what quantities. The responsiveness of purchases to price and their predictability when external circumstances change must now be examined.
Elasticity of Demand
To the economist, a concept such as addiction is almost impossible to reconcile with a central tenet of economic theory. According to the "first law of demand," an increase in the relative cost of purchasing any good will lead to a decrease in the quantity of it demanded by buyers. The idea is general enough to include costs that take the form of time losses, risks of incarceration, risks of low quality, and social opprobrium. If different goods are substitute sources of well-being for an individual, then a given level of well-being can be more cheaply attained by substituting against those goods which have become costly relative to others. The economist's "second law of demand" is equally at variance with the popular concept of addiction. The law states that one's demand for a good becomes more elastic (price-sensitive) the longer the time that has elapsed since a change in its price. People become more aware of substitutes as time passes, and they minimize their long-term losses by taking their time in adjusting to changed circumstances. For an addictive good, however, an initial adjustment period is said to lead to a pattern of use for which substitution is impossible.38
If addiction is to be consistent with its popular version, we must interpret it as a vertical demand curve, i.e., one in which quantity purchased is held fixed, regardless of the cost the user must bear. Again, the cost includes a dollar price, time spent searching for a seller, a risk of apprehension for both narcotic and criminal viola-tions, and a risk of poor (or lethal) quality. Let us call their total the "full cost."39 The elasticity of demand with respect to the full cost of the good is a critical number for the evaluation of both short-run and long-run policy. If crime is a concomitant of addiction, effective policies to raise the price of heroin may increase crime if the demand for heroin is sufficiently inelastic. If new users have more elastic de-mands than do long-term addicts, policies that raise price to the for-mer and lower it to the latter may be in order.4° If the long-run elasticity of demand is very high, legalization (which lowers the full cost substantially) may turn us into a nation of users.
While the analysis of the market would be much simpler if in fact users had vertical demand curves, the world is not so obliging. De-manders have heterogeneous tastes, ingesting the drug in various manners and with varying frequencies. Although figures here are no-toriously unreliable, large-scale daily dependence probably charac-terizes no more than half of all users.'" Addiction can be broken by self-imposed detoxification or by imprisonment, and the odds of im-prisonment may be greater for those with larger habits. There is con-siderable evidence that in times when supplies are short, sporadic, or of uncertain quality, other drugs are substituted for heroin.42 In the process of substitution, "addiction" must vanish. Methadone maintenance substitutes another, equally habit-forming, substance for heroin, and there is some evidence that enrollments in methadone and detoxification programs vary inversely with the full cost to the user.43 Stories about the difficulty of withdrawal appear to be at variance with the facts. The interesting problem is not habitual use, but rather the frequent restarting of a habit by one who has previously gone through withdrawal."
Time patterns of population use and adjustment patterns of individual use give little support to a simple notion of addiction. There is general agreement that heroin use in the United States is predominantly a habit of the young, and that users "mature out" of their habits, in most cases without medical aid.45 Use of opiates in nineteenth century American was a middle-class phenomenon, with the majority of users being housewives who had acquired a tolerance for patent remedies that contained them. There is little record of withdrawal problems among them (and no record of criminal problems) following the passage of the Harrison Act, although their number may have been in the millions.46 Estimates of the percentage of American servicemen who used opiates while in Vietnam center around 30 to 40 percent's' Yet one study found that those defined by its author as addicted on or after their return constituted less than 3 percent of those sampled." Equally interestingly, they apparently functioned well while on active duty. As will be seen below, such evidence makes an estimation of postlegalization addiction rates either impossible or meaningless.
Algebraically, elasticity of demand is the ratio of the percentage change in quantity (units) of a good purchased to the percentage change in its cost to the buyer, holding other factors constant. Thus a low value of elasticity indicates relative unresponsiveness of quantity to price changes. If elasticity is 0.3, a 10 percent rise in price will result only in a 3 percent fall in quantity purchased. If elasticity of demand is less than one in value, an increase in the price of the good leads to a greater dollar amount spent on it. If it is greater than one, a rise in price leads to a fall in total spending on the good. Elasticity may vary as we move away from a certain range of prices, i.e., the relative responsiveness of users to market changes may depend on whether the initial price is $1 or $100.49 As noted in the previous section, we can reasonably expect that after legalization the price of heroin will be about 1 percent of today's full cost. Even with usable estimates of elasticity in the neighborhood of today's price, we can have little confidence in inferences about market situations so far removed from today's. As will be seen below, we also cannot make convincing inferences by examining past periods of legalization or the experiences of other countries.
The literature provides little in the way of elasticity estimates for heroin. This is not surprising, since estimation requires a fair amount of reliable data or market conditions at different times or places, as well as observations on other variables that might account for differences in buyer behavior in dissimilar environments. Fujii5° cites unpublished estimates of 0.007 and 0.09. The only estimates I have discovered that contain any respect for econometric theory or method were produced by the Center for Naval Analyses in the early and mid-1970s." Using samples based on street purchases in various cities, they corrected for the differences in purity among them. On average they found demand elasticities of about 0.25 with respect to dollar price. Full cost, however, varied widely among cities. Assume for the sake of argument that their estimate of elasticity is correct, and that the demand curve in question has a constant elasticity. If legalization led to a fall in the full cost of a pure gram to 1.5 percent of its current level, it would lead to a 187 percent increase in the amount purchased and a decrease in spending on heroin to 6.2 percent of its former leve1.52
Such a projection is no more than conjecture. With no experience in the world even remotely close to that of such a price drop, we can say no more. The elasticity estimate of 0.25, additionally, is a short-run figure. While we can learn little about the speed with which the market will adjust to legalization from the studies mentioned above, we can state with some confidence that theirs are low estimates of the long-run effect of reduced price on consumption. Habits of existing users do not expand instantly upon knowledge that price has fallen, and new users likewise require time to become attracted and habituated. Assume the far less likely alternative of a long-run elasticity of 1.5." The same assumed decrease in price now leads to a 560-fold increase in consumption and a twelvefold increase in expenditure on heroin. If we estimate conservatively (see below) that there are currently 200,000 users among American adults, about half of whom have large habits, we would be forced to conclude that under legalization nearly half of all adults in the country would be heavy users.
Both intuition and history tell us that such a conjecture is meaningless. No estimate cited by Musto for the pre-Harrison Act period comes remotely close, and the highest incidence estimate to which any credence can be given is 8-10 percent of physicians in 1899.54
The highest reported estimate in the contemporary world is 3 percent for adults in Hong Kong." Only 70 percent of America's adult pop-ulation uses alcohol.56 Once we abandon a simple idea of addiction and account for the diversity of potential users and the plausible range of price change after legalization, almost any outcome seems possible.
Shifts of the Demand Curve
Our problems in estimating the response to legalization are not com-pletely solved even if we have a reliable estimate of demand elasticity over the relevant range of prices. The entire demand curve for heroin will shift outward or inward as income,57 prices of other goods, and tastes of users or potential users change. Even with good price data, one cannot estimate the effect of legalization on consumption with-out also accounting for these other factors that will be changing con-temporaneously with price. In addition, changes over time in these other influences on demand may render it impossible to make pre-dictions about today on the basis of experience in the distant past. Rising wages, for example, will ambiguously affect use. Part of the larger income they give a person might be spent on heroin. Since leisure time is needed to consume it,58 however, rising wages might alternatively lead to decreased use because work is now more finan-cially rewarding." In the current world, it seems clear that the choice of a substance, at least initially, depends on the alternatives available. Whether a good is a realistic alternative to heroin or nbt depends on both of their prices.
Even more problematic is the question of the noneconomic characteristics of individuals that may predispose them to use. The changing demographics of American users seem to defy economic explanation. Middle-aged housewives and Civil War veterans apparently constituted the majority of American users in the nineteenth century,6° and most of them lived in small towns or rural areas.61 By the end of World War II, users had become predominantly male, split between an older group of white rural southerners and a younger group of urban blacks and Puerto Ricans.62 The southern whites have apparently since vanished. By the 1970s rhetoric about the spread of heroin to well-off suburban youth had become widespread, but hard evidence on its significance was lacking. Hispanics other than Puerto Ricans, however, had clearly become larger users and sellers.° Prior to the mid-1960s, the average educational attainment of users was the same as that of the general population, with few users characterized by either extremely high or low levels of schooling.64 Since then, average attainment has dropped somewhat.
Patterns of use elsewhere are equally paradoxical. In England heroin is almost exclusively a drug for middle-class whites. Despite extensive recent immigration of racial minorities from opium-supplying areas, addiction rates among them pose no significant problem.° In cultivating countries, use is greater in rural growing areas than in cities, contrary to the Western pattern. In Hong Kong, the habit is considered a prerogative of old age, and there is little use among the young. The latter, however, have turned to alcohol in increasing numbers.66 Massive increases in use by Iranians characterize the Khomeini years, despite extreme attempts at repression.67 On the other hand, harsh measures taken after the Chinese revolution are said to have been highly effective in ending opium use there.68
There is uniformly strong testimony that people are introduced to opiate use by their acquaintances.69 Famed though he was in folklore, the pusher who turned samplers into addicts probably never existed.7° To the extent that heroin use carries with it illegal status, one should expect that an ordinary process of market sampling by the general public will not be observed. Such a process might well be seen after legalization, but even this is by no means certain. Decriminalization of small amounts of marijuana possession in Oregon during the 1970s appeared simultaneously with an increase in the fraction of the population who claimed not to use it because they had no interest in doing so.' All of this makes it impossible to predict shifts in the demand curve resulting from shifts in the identities of groups likely to contain interested users. Both social convention and historical accident render predictions about future demand shifts effectively impossible.
ESTIMATING USERS AND ADDICTS
In the United States, governmental production of demographic and economic data is typically characterized by a fair degree of reliability and impartiality, given the cost of collection. While the variables measured (e.g., unemployment) may be imperfect realizations of the theoretical constructs of interest, the statistical methods used are generally of high quality. As one enters the realm of figures with a more obvious political purpose, both reliability and methodological quality drop off rapidly. The statistics on heroin are an ideal example. As the Nixon administration attempted to impress the public with its view of the seriousness of the problem, estimates of the addict population by the U.S. Bureau of Narcotics and Dangerous Drugs (BNDD) rose from 69,000 in 1969 to 560,000 in 1971. The effectiveness of newly instituted control programs was clearly in evidence as they fell to 160,000 in the election year of 1972.72 Describing the same period, another author claims that the government systematically attempted to understate the size of the addict population in its official estimates, in an attempt to minimize political controversy."
Figures on the numbers of addicts are frequently characterized by gross inconsistencies with harder facts. Singer argued that the state government's estimate of 200,000 addicts in New York City at the time was probably an overstatement by a factor of three, since the volume of property crime (assumed to be a major source of addict support) was far lower than that required to support so many addicts, even after allowance for unreported crime:4 If we take Governor Rockefeller's estimates of addicts and habit sizes seriously, we are forced to conclude that in 1973 "the average resident of New York would be robbed, mugged, or murdered approximately seven times a year."" Despite an astoundingly complex estimating formula, estimates by the National Institute on Drug Abuse of about 500,000 addicts nationwide in 1978 are equally inconsistent with what is known about property crime." In part because of internal bureaucratic criticism,' and in part because of the decreasing importance of heroin on the national political agenda, official estimates of the addict or user populations seem to vanish after 1980.'8 Despite the possibilities for making the estimates more accurate,79 there is virtually no constituency to demand such accuracy.8°
In 1928, Terry and Pellens collated a set of estimates of the Amer-ican opium "addict" population made by various methods between 1913 and 1920. The highest estimate (782,000) exceeded the lowest (102,000) by a factor of 7.7.8' More recently, Moore constructed several estimates of heroin "users" in New York City in 1970.82 The figures vary by a factor of five, from 54,000 to 252,000. The upper estimate for New York City exceeds some low estimates for the entire United States. The results become more interesting at lower levels of aggregation. The Federal Bureau of Narcotics reported a total of one active addict in Utah in 1959, despite 117 narcotics arrests and 101 convictions in the state between 1953 and 1959.83 Apparently the law enforcement effort failed to deter crime, since by 1963 he had been joined by a federally estimated second addict.84
In part, the problem of identifying and enumerating heroin "users" and "addicts" is a definitional one. As noted above, the distribution of consumption varies smoothly rather than moving ab-ruptly from zero to some addiction-maintenance level. Moore esti-mates that about 60 percent of all active users have habits of over 50 milligrams of pure heroin per day." Below some threshold, which probably varies with the individual user, heroin is an object of oc-casional consumption rather than daily obsession. Serious symptoms of physical withdrawal are apparently seldom encountered, but those who withdraw for a period often relapse into using it again.86 There is also reason to believe that users overstate average habit sizes or the frequency of days on which they consume claimed amounts.87 Researchers have frequently found that the distributions of alcohol and other recreational drugs consumed daily are logarithmically normal in survey data, characterized by low modal amounts and long upper tails.88 There is some evidence on heroin that leads to similar conclusions. 89
If the distribution of users is thus characterized, an estimate of their total number tells us little in the absence of information about the circumstances of individuals in it. If all heavy users are wealthy (as they are not), then crime as a correlate of use will be a small problem at worst. Estimates of users and "problem users" are made in a variety of ways, each of which reduces to a simple idea: Find a population of known users, or one whose size is assumed to be highly correlated with that of the user population. Then find an inflation factor and multiply the known population by it in order to estimate the overall user population. As an ideal example, assume that a perfectly random (in all relevant characteristics) subset of the entire population is injected with truth serum and its members told to describe their drug use. If the sample is large enough we will be able to bound our estimate of users in the population with a considerable degree of confidence. The critical inflation factor will simply be the ratio of population size to sample size.
Surveys of drug use in the real world carry with them little guarantee of truthfulness, and the likelihood that the understaters balance out the exaggerators is unknown.9° If we are dealing with an infrequently encountered attribute, the sample required for reliability will be very large. Glenn reports that it would require $500,000 to produce a statistically accurate estimate for North Carolina alone.' To drive home the point, he describes a survey of 2,000 randomly selected persons in that state actually taken by the North Carolina Drug Authority in 1974. Only two of the individuals surveyed admitted to regular heroin use (and they may have been lying). On the basis of this survey, the state officially estimated 2,760 regular users. From the characteristics of the respondents, the state concluded that 1,320 were male and 1,440 female, 1,320 were black and 1,440 white, and all 2,760 were moderate smokers.
In the absence of good information from truly random questionnaire samples, inferences are usually made on the basis of samples whose habits are less in doubt. From such samples we can frequently derive interesting facts about the sampled group, but not necessarily about the population. For example, in 1981 74.0 percent of the 93,678 heroin user admissions92 to federally funded drug clinics were male and 43.6 percent of the total were black,93 both of which might sound reasonable and consistent with intuition as descriptions of the population. On the other hand, 56.7 percent of heroin admissions claimed they had not been arrested within twenty-four months of admission.94 Is this "reasonable" or isn't it? According to an unpublished source cited by Moore,95 45 percent of total users are arrested in any given year. If the population is stationary, the fraction never arrested in twenty-four months will be (1 — .45)2 = 30.2 percent. The 45 percent figure, of course, was drawn out of thin air by a police official, and arrest histories given at clinics are usually unverified self-reports. One thing is certain: On the basis of figures for admissions to treatment alone, we cannot identify the size of the total user population without making probably untenable assumptions about the representativeness of the clinical sample.96
Most estimates of the overall population simply apply a multiplier to a known population. Moore's low estimate of 54,000 New York City users in 1970 applies a multiplier of 2 to the number of New York names on the BNDD register of active users, whose activeness is unverifiable.97 The multiplier is the unpublished opinion of a second expert, as reported by a third.98 Moore's high figure of 252,000 is an inflation of the 151,000 names on the New York City Register by a factor of 1.67. The adjustment factor reflects the "finding" that 60 percent of dead users were known to the Narcotics Register. How "user" was defined or determined is unknown. The actual practice for determining and reporting heroin-related deaths violates proper sampling and reporting practice in ways that render the estimates valueless.99 There is no known way of establishing a stable ratio of overdoses'°° or serum hepatitis episodes to users.'°' Attempts to extrapolate population sizes by comparing criminal and medical populations have also shown little promise, in large part because we still have no idea of the number of individuals who have not been noticed by either authority.102
NIDA had two basic methods of estimating addicts in the late seventies. The first was a discredited inflation of the BNDD Register. t°3 The second was an astonishingly sophisticated use of assumptions and unreliable data. According to the GAO,'" estimates in later years are adjustments of some assumed 1974 and 1975 figuresm by an elaborate simultaneous-equation model that incorporates estimated heroin-related deaths, estimated heroin-related injuries, and the purity of purchased samples of commercial heroin. NIDA does not specify why it believes that these are the only three relevant factors determining addiction, or how it defines an addict. As noted above, heroin-related deaths and injuries are neither reliably reported nor estimated, and the data on purity are characterized by erratic geographic coverage and overly small sample sizes.'" Most interestingly of all, over a recent period in which all three indicators rose in value, NIDA's estimate of addicts fell. 1°7
We cannot reasonably deny a conclusion that published addict and user estimates are meaningless. I have attempted to correlate certain NIDA estimates with obvious political variables such as budget size requests and changes, but have been unable to obtain significant re-sults owing to sample size limitations.m8 If the problem is the size of the user or addict population, we have no ability to measure it. Neither we nor anyone else can say much that is useful about how it will change under legalization. The "problem," however, is gen-erally characterized as one of dependency and crime, which expand with the user population. It is to these matters that I turn next.
HEROIN, CRIME, AND EMPLOYMENT
While some Americans are certainly offended by the fact that others consume heroin and derive pleasure from doing so, it is likely that very few would claim that their disgust stems from the sight of con-sumption per se. Rather, the general distaste for heroin users is usu-ally attributed to their criminality and low economic productivity. In this section I examine these two matters.
As noted above, the distribution chain in the heroin market is char-acterized by an inefficiently large number of stages and by enter-prises whose sizes are too small to exploit what scale economies do exist. Highly personalized exchanges are a method of coping with uncertain quality and availability. One consequence of these facts is that users will frequently attempt with success to participate as agents, i.e., to become their own middlemen, to an extent not seen in legal markets. Those selling at higher levels are likely to find users more acceptable as customers, since a dependable supply and quality of heroin acts as a method of eliciting desirable conduct at lower levels. A significant part of the crime associated with access to heroin is thus committed in the process of making consensual, albeit illegal, business transactions in the material itself.
Moore, citing unpublished surveys of street users, concluded that 51 percent of the income of "addicts" and "drug dependents" (median and larger dose users) in New York was earned by work in heroin distributioni09 Other studies and surveys cited by Goldman"° concur in stating that a majority of higher-volume users play some role in distribution. Smaller users appear more likely to finance their consumption through employment, welfare, and borrowing.1" An unknown fraction of remaining income (probably less than 25 percent) is derived through other consensual crimes, including prostitution and sale of other drugs. According to federal treatment statistics, 71 percent of primary heroin users entering drug treatment in 1980 reported themselves as unemployed, and 75 percent of them claimed not to be seeking work."' Seventy-three percent of admit-tees claimed to have daily habits."3 Most of their income was apparently derived through illegal work,"4 as opposed to the victimization of others.
Statements by users about their habit sizes are typically not given under conditions likely to elicit honest answers. An individuals' self-esteem may increase with the size of the habit he claims he can afford."5 At methadone clinics the size of dosage allowed may depend on self-reported use. There is likewise a strong incentive on the part of law enforcement or clinical personnel to accept such figures as an inflated justification for their own activities."6 The problem is that estimates of the volume of addict property crime are typically based on such statements, combined with assumptions (little hard data seem to exist here) about discounts given by buyers of stolen goods."' The breakdown of addict crimes with victims between violent crimes (mugging) and property crimes is also uncertain. Although violent crime carries a higher physical risk for the criminal, if successful it is more likely to produce cash rather than goods."8 As a Task Force Report on Narcotics for the President's Commission on Law Enforcement noted, "There is no reliable data to assess properly the common assertion that drug users or addicts are responsible for 50 percent of all crime."9
Indirect evidence on the weakness of the link between heroin and crime is easy to find. Differences in crime rates among cities are typically too small to be accounted for by differences in addiction rates.'2° While about half of all crime is unreported, differences in reporting rates between cities are too small to account for known differences in heroin use. Street crime has not risen by an amount that could possibly be consistent with reported "epidemics" of heroin use, and the broad trend in the temporal pattern of reported crime is not consistent with that of any estimated shift in the user population.I21 Some have claimed that drops in the quality and availability of heroin in those cities that have experienced epidemics have been accompanied by drops in the crime rate,'22 while others claim to have found the opposite.'23 The former finding implies either that the demand for heroin is highly elastic, i.e., addiction is a fiction, or that the link between heroin and crime is too weak to be of much value for policy planning.
There is some evidence that delinquency often precedes heroin use, frequently by several years.'24 What drug use actually does, accord-ing to Chein et al.,'26 is to alter the objectives of one who is already predisposed to and possibly involved in criminal behavior. The mix of crimes committed will change to favor those which facilitate the acquisition of heroin. Finestone'26 found higher rates of larceny and robbery for addict offenders, and higher rates of sex crimes, auto theft, and assault for nonaddict offenders, consistent with this the-ory. According to Coate and Goldman's survey-based simultaneous equation mode1,127 criminal earnings, if one can obtain them, pre-cede drug use.
Observing his sample of street users, Goldstein concluded that "the data suggest that, in many cases, subjects who have opportun-istically and successfully seized a chance . . . simply use their gains to produce drugs. . . . Drug use is a major form of conspicious con-sumption among society's outsiders."128 Although the folkloric lit-erature frequently states that crime only begins after a well-defined onset of addiction,'29 the proposition would have little empirical backing, even if we could construct usable criteria for the onset of addiction.'" If heroin use is a movable feast, engaged in by individ-uals who can switch among crimes, employment, and drugs, then we cannot draw any conclusions, favorable or unfavorable, about the impact of legalization on crime rates. According to one study, individuals entering methadone treatment have lower arrest rates for nondrug crimes during treatment, but afterward arrest rates rise to about 70 percent of pretreatment levels.'3' Epstein cites an unpub-lished study's findings that the only criminal charges (against those younger users admitted into a methadone program) that declined were for drugs, forgery, and prostitution.132 I have been unable to collect reasonable data on the amount of law enforcement resources that would be freed for the apprehension of other criminals after legalization.'33 More important, it is hard to estimate the marginal deterrent effect on predatory crime of an increase in police resources devoted to controlling it.
The work incentives of opiate users, as reported by hostile parties, have changed radically with the politics of their times. According to a contemporary authority, an addict "characteristically becomes lethargic, slovenly, undependable, and devoid of ambition."34 By contrast, at the turn of the century opium became illegal in the United States, in large part due to the efforts of Samuel Gompers. He claimed that its use by Chinese immigrants so increased their productivity that whites were at a disadvantage in the labor market. Additionally, it increased their risk-taking proclivity, as virtually all Chinese were alleged to be heavy gamblers.'" In reality, the period between the Civil War and World War I was the period of America's greatest sustained proportional economic growth. It was characterized by low unemployment rates, legal opium, and a large population of habitual users. The same was true of Victorian England.
People's choices of work and leisure, or of occupations or employers, are fundamentally dependent on the alternatives available. An individual who works legally gives up the possibility of some illegally obtained income. One who can support a $30/day habit and $10/day in living expenses'36 from illegal activity is earning an income of $14,600 per year, which puts him considerably above any reasonably defined poverty level. An obsession with the concept of addiction leads us to forget that in all but the shortest of runs the money is voluntarily spent on this bundle of goods. Relative to legitimate alternatives, the street may offer a superior financial incentive and work environment, if this is one's preferred consumption bundle. Given the uncertainty associated with the supply and quality of drugs, a more public life-style is probably a method of reducing one's search costs for drugs and becoming more cognizant of earning opportunities.
Unemployment by the federal definition is not synonymous with leisure for the typical heroin user. As Goldstein notes in his detailed study of the daily lives of 51 of them,137 they are always on the alert for peculiar circumstances of time and place that will enable them to "get over," i.e., succeed economically for the day. The workplace in which this occurs is unquestionably a competitive one. If heroin use produces an all-consuming lethargy, it is difficult to explain how an addict can support his habit.138 Those with better opportunities away from drug use are more likely to attempt treatment for their habits. In 1980, 56 percent of all enrollees in heroin treatment programs claimed to have twelve or more years of education.'39
Even today, those with appropriate skills can support their habits by working, although regular employment probably increases the time-cost of a heroin purchase. The relatively high frequency of use in the health professions'4° may reflect their members' lower full cost of obtaining drugs. For users in less renumerative positions, crime supplements work income. Additionally, size of habit appears to be positively correlated with work income.'4' Although the above gives some reason to be hopeful, we cannot even qualitatively predict the effects of legalization on work time and effort. Cheapening heroin implies that an individual will be more disposed to consume it, an activity that requires both heroin and time. Lowering the value of heroin search time relative to the value of work time, taken by itself, will lead to an increase in work. Which of the two effects will dominate cannot be determined a priori. The actual consumption pattern of Chinese opium smokers was typically one of moderate use after the day's work was finished.'42 Their world, however, was characterized by minimal uncertainty about quality or availability.
SUMMARY AND CONCLUSIONS
In several important respects, the prohibition on heroin has clearly been effective. It has imposed a structure of supply and distribution whose complexity and costs are very high relative to those that would be seen in an unregulated market. What we know about the technology and history of supply indicates that after legalization the market would be workably competitive. With no patented production processes or monopolized inputs, the market would be open to any newcomer willing to make an investment in it. With no discernibly great economies of scale or scope in the unregulated market, it is unlikely that monopoly would result. In the existing market, illegality makes the economically optimal distribution chain a peculiarly high-cost one, i.e., one dependent on the use of addicts as agents. This too would end with legalization.
Illegality may have imposed high costs on the supply side of the market, but without knowledge of demand conditions we cannot say anything about how effectively it has curbed the use of heroin. If users truly are addicts and must support their habits regardless of price, the illegality will have done no more than raise the expense of obtaining the same quantity that would have been sold to them under a regime of legality. It is clear that such a simplistic characterization of users is incorrect, and that demand in the market is not perfectly inelastic. Existing estimates of demand elasticity, however, are unable to give us usable predictions about the consequences of a price drop of the magnitude that would follow legalization. Additionally, demographic patterns of use change rapidly and vary widely among cultures. Such instability of demand is difficult to capture in the economist's market model and adds an even greater hazard to predicting the consequences of legalization.
To predict those consequences numerically requires that we have a reasonably accurate statistical picture of the existing market at our disposal. While they are frequently circulated and quoted with alarm, figures on the numbers of users and the volume of crime for which they are responsible are meaningless political constructs. They are highly sensitive to the use of arbitrary assumptions and are depend-ent on surveys or registers whose methodology is questionable and whose coverage is poor. Judicial and legislative testimony about the market invariably comes either from self-interested former partici-pants or self-styled experts whom only a television reporter could take seriously. Some of the literature alleges that the statistics over-state the problem, while the remainder claims that they conceal a much larger one. No one outside of elective office assumes they are correct.
If we know little about the extent of heroin use, we know even less about the careers of users, in particular about their criminal hab-its and other sources of income. The case for almost any type of public policy toward heroin depends in large part on the strength of the correlation between crime and heroin use. Such a correlation is both hard to find in the literature and inconsistent with the actual experience of crime during "epidemics" of heroin use. Because we know so little about the user population, we cannot say much about that fraction of users known to support their habits through legal employment, or how that fraction would change after legalization. It is hard to think of an unskilled individual as unindustrious if he is attempting to support his habit at today's prices.
All statistics are imperfect, but those related to drug use are egre-giously bad. There is some room for improving the latter at reason-able cost,"3 but little reason to expect that this will be done.'" The suppliers and demanders of meaningless statistics are both heavily concentrated in government, and what numbers appear in newsprint do so as the consequence of a carefully orchestrated process. As a typical voter, I have so little influence on electoral outcomes that it is not worth my while to ascertain the actual magnitude of the heroin problem, if one exists. As a comparison, consider the activity of the U.S. Census Bureau. A large number of individuals outside of gov-ernment (e.g., marketing experts) have an interest in accurate demo-graphic statistics, and ultimately a biased or incomplete federal census imposes on them costs that will lead to pressure for accuracy. There is no such well-defined set of losers from poor heroin statistics.
The pattern has been seen time and again. If the specious figures on oil reserves announced at the start of the "energy crisis" were correct, America would have by now been immobilized and frozen. The "crisis," of course, was no more than an attempt to politicize and bureaucratize the allocation of a good for which market conditions had changed. As an issue loses political saliency, the crisis fades away. We thus move from one statistical crisis to another—from vanishing farmland to environmental cancer to heroin epidemics to illegal aliens. Some of these matters may actually be problems on which accurate statistics could shed considerable light. The policy-making process, however, is unlikely to produce either good statistics or good policy.
1. I do not discuss the normative literature that centers on the "cost to society" of heroin use. Such costs are said to include lost economic output (because users seldom work 9:00 to 5:00 at legal jobs) and crime. The former "loss" is no more than a judgment that only market output is of value, and the latter (see Sec. V below) generally assumes, erroneously, that in the absence of heroin most of this crime would vanish. For examples of such methodology, see Ralph E. Berry and James P. Boland, The Economic Cost of Alcohol Abuse (New York: Free Press, 1977); Brent L. Rufener, J. Valley Rachal, and Alvin M. Cruze, "Costs of Drug Abuse to Society," in Irving Leveson, ed., Quantitative Explorations in Drug Abuse Policy (New York: Spectrum, 1980).
I also do not deal with the normative theoretical literature that attempts to establish rationales for legalization or prohibition. See A. J. Culyer, "Should Social Policy Concern Itself with Drug Abuse?" Public Finance Quarterly 1 (October 1973): 449-56; J. Patrick Gunning, "Notes on Two Abuses: Drugs and the Pareto Criterion," Public Finance Quarterly 4 (January 1976): 43-49.
2. William A. Jordan, Airline Regulation in America (Baltimore: Johns Hopkins University Press, 1970).
3. U.S. General Accounting Office (GAO), Report to the Attorney General: Heroin Statistics Can Be Made More Reliable, Report GCD-80-84, July 1980.
4. Charles Winick, "Some Aspects of Careers of Chronic Heroin Users," in Eric Josephson and Eleanor E. Carroll, eds., Drug Use: Epidemiological and Sociological Approaches (Washington, D.C.: Hemisphere, 1974).
5. Robert L. DuPont and Mark H. Greene, "The Dynamics of a Heroin Addiction Epidemic," Science 181 (August 24, 1973): 716-22.
6. Simon Rottenberg, "The Clandestine Distribution of Heroin, Its Discovery and Suppression," Journal of Political Economy 76 (January 1968): 78-90.
7. Oliver E. Williamson, "Transaction-Cost Economics: The Governance of Contractual Relations," Journal of Law and Economics 22 (October 1979): 233-61.
8. There is a fair-sized body of economic literature that attempts to model the illegal heroin market. None of it contains more than a peripheral attempt to compare the legal and illegal markets. Many of the conclusions reached by the various authors depend on such doubtful assumptions as monopoly control of supply or highly inelastic demands by addicts. See Godwin Bernard, "An Economic Analysis of the Illicit Drug Market," International Journal of the Addictions 18 (October 1983): 681-700; Billy J. Eatherly, "Drug Law Enforcement: Should We Arrest Pushers or Users?" Journal of Political Economy 82 (January 1974): 210-14; Daryl Hellman, The Economics of Crime (New York: St. Martin's, 1980), chap. 9; John F. Holahan, "The Economics of Heroin," in Patricia Wald et al., eds., Dealing With Drug Abuse (New York: Praeger, 1972), pp. 255-99; James V. Koch and Stanley E. Grupp, "The Economics of Drug Control Policies," International Journal of the Addictions 5 (December 1971): 571-84; Mark H. Moore, Buy and Bust (Cambridge, Mass.: Lexington, 1977), part 1; Rottenberg, "The Clandestine Distribution of Heroin"; Michael D. White and William A. Luksetich, "Heroin: Price Elasticity and Enforcement Strategies," Economic Inquiry 31 (October 1983), pp. 557-64.
9. Annelise Anderson, The Business of Organized Crime: A Cosa Nostra Family (Stanford: Hoover Institution Press, 1979); Peter Reuter and Jonathan B. Rubinstein, "Fact, Fancy, and Organized Crime," The Public Interest 53 (Fall 1978): 45-67.
10. The problem is complicated by the fact that a long-term career in organized criminal enterprises is generally available only to a person with a reputation for dependability, much as in ordinary business. See Benjamin Klein and Keith Leffler, "The Role of Market Forces in Assuring Contractual Performance," Journal of Political Economy 89 (August 1981): 615- 41. The youth and short time horizons of many in the heroin market may render reputation a poor enforcer of noncontractual relationships.
11. Information on the paucity of violent activity in organized crime appears in Reuter and Rubinstein, "Fact, Fancy, and Organized Crime."
12. Moore, Buy and Bust, pp. 95-115; Edward Preble and John J. Casey, Jr., "Taking Care of Business—The Heroin User's Life on the Street," International Journal of the Addictions 4 (March 1969): 1-24, esp. 9-12.
13. Evidence for this is provided by the existence of nontrivial enrollments in heroin treatment programs in nearly all states. See U.S. National Institute on Drug Abuse (NIDA), Data from the Client-Oriented Data Acquisition Process, State Statistics 1980, Statistical Series E, No. 22 (Washington, D.C., Government Printing Office, 1982). Note that the studies covering New York identify the same structure both before and after the market growth of the 1970s.
14. Since statistics on users and law enforcement activity are uniformly poor (see Section IV below), I cannot make a reasonable estimate of the probability of arrest or the probability that an arrestee will be imprisoned. Citing a 1972 news report, Levin et al. found that in New York "the fraction of people arrested for possessing a pound or more of heroin (presumably for sale) who ended up in prison ranged from 32.2% in the Bronx, 15.7% in Manhattan, 14.3% in Queens, to only 4.4% in Brooklyn." See Gilbert Levin, Edward B. Roberts, and Gary B. Hirsch, The Persistent Poppy: A Computer-Aided Search for Heroin Policy (Cambridge, Mass.: Ballinger, 1975), p. 59.
15. Personalized and frequently repetitive transactions in heroin also render the corruption of law enforcers more likely than in other areas of illegal activity. See Peter K. Manning and Laurence Redlinger, "Working Bases for Corruption: Organizational Ambiguities and Narcotics• Law Enforcement," in Arnold S. Trebach, ed., Drugs, Crime, and Politics (New York: Praeger, 1978). In 1975 Baltimore's chief narcotics officer claimed that 800 criminal narcotics informers worked with police there. See Edward Jay Epstein, Agency of Fear (New York: Putnam, 1977), p. 101. In the absence of some understandings between these individuals and the police, it is difficult to explain why any users were still on the street.
16. Richard Stephens and R. Smith, "Copping and Caveat Emptor: The Street Addict as Consumer," Addictive Diseases 2 (December 1976): 585-600.
17. The market does contain some channels of information. The employment of "touts," who discreetly publicize the availability and quality of certain sellers' heroin on the street, constitutes a highly restricted form of advertising. See Moore, Buy and Bust, p. 49. While street-level packages of heroin sometimes carry "brand names," these names are typically short-lived and undependable, since they do not carry any quality-guaranteeing investment behind them that would serve as a performance bond for the user of the name. (These names can also, of course, be counterfeited.) See Klein and Leffler, "The Role of Market Forces in Assuring Contractual Performance"; Paul J. Goldstein et al., "The Marketing of Street Heroin in New York City," Journal of Drug Issues 14 (Summer 1984): 553-66.
18. Philip Nelson, "Advertising as Information," Journal of Political Economy 82 (August 1974): 729-47.
19. In Sec. IIIB below, I discuss the introduction of users to heroin in general, as opposed to the process of sampling a given seller's wares.
20. Lee Benham, "The Effect of Advertising on the Price of Eyeglasses," Journal of Law and Economics 15 (October 1972): 337-52.
21. In those situations where heroin is in short supply, personalized trading relationships are more likely to break down as a result of constraints on availability. This increases the risk to buyers associated with each transaction, because of seller incentives to act opportunistically when repeat purchase is less likely. DuPont and Greene ("The Dynamics of a Heroin Addiction Epidemic," p. 719) claim that supply interdiction during the Washington, D.C., "epidemic" of the early 1970s helped increase the actual average price paid per milligram of pure heroin from $1.53 in February 1972 to $5.80 in March 1973. Sellers diluted the amount in the average bag from 16.4 to 5.2 mg over the period, and the coefficient of variation of heroin content per bag rose from 1.11 to 1.43.
22. Moore, Buy and Bust, chap. 2.
23. Robert H. Bork, The Antitrust Paradox (New York: Basic Books, 1978), chap. 11. The statement of the text is subject to a variety of minor qualifications. Prime among them is that there may be gains in allocative efficiency under some circumstances if a monopolist at one stage ventures into other stages of production.
24. Rottenberg, "The Clandestine Distribution of Heroin." One difficulty with a model of monopoly is that it does not explain the observed fluctuation of price or potency with short-term supply availability. If profits are concave in price (as in the case under reasonable as-sumptions about demand) and inventories are not too costly to store, the monopolist makes more under a steady price-output strategy than under an unstable one.
25. U.S. Senate, Permanent Subcommittee on Investigations, Committee on Governmen-tal Operations, 88th Congress, Hearings Pursuant to S. Res. 178, Parts 3 and 4. Organized Crime and Illicit Traffic in Narcotics (Washington, D.C.: Government Printing Office, 1964), pp. 3028-30.
26. U.S. Senate, Subcommittee to Investigate Juvenile Delinquency, 87th Congress, Hear-ings Pursuant to S. Res. 265, May 9, 17, 20 and August 6, 7, 1962 (Washington, D.C.: Gov-ernment Printing Office, 1963), pp. 645-46. Commissioner Giordano's testimony cites Cosa Nostra informer Joe Valachi as his source of the assertion.
27. The range for the user estimates is 70,000-150,000. The source of the interviews is Preble and Casey, "Taking Care of Business." The authors, however, do not characterize their sample statistically.
28. Moore, Buy and Bust, pp. 95-100.
29. U.S. Senate, Hearings Pursuant to S. Res. 178, parts 3 and 4. Organized Crime, pp. 772-89. Using subjective judgment, I identified 60 out of 86 names on the congressional list as Italian, 9 as Jewish, 2 as Hispanic, and 15 as "other" or unidentifiable. I have found no comparable lists for subsequent years.
30. Mark H. Greene, "The Resurgence of Heroin Abuse in the District of Columbia," American Journal of Alcohol and Drug Abuse 2, no. 2 (1975): 141-64.
31. Catherine Lamour and Michael Lamberti, The International Connection (New York: Pantheon, 1974).
32. U.S. House of Representatives, Appropriations Committee, Hearings on Department of Justice Appropriations for 1983, part 7, March 9, 1982 (Washington, D.C.: Government Printing Office, 1983), p. 37; David J. Bellis, Heroin and Politicians (Westport, Conn.: Greenwood Press, 1981), pp. 84-88.
33. John Kaplan, "A Primer on Heroin," Stanford Law Review 27 (February 1975): 801- 26, esp. 817.
34. Thomas Szasz, Ceremonial Chemistry (Garden City, N.Y.: Doubleday Anchor, 1975), p. 52. Quoting President Nixon's special adviser on drugs, the author states that only 5 percent of the world crop would supply America's users. Although this may indicate a high elasticity of supply to Americans, it should be remembered that virtually all estimates of unknown quantities (crop size, number of users) are worthless.
35. CBS, "Sixty Minutes," 2 December 1984.
36. The figure is roughly consistent with the statement by Kaplan ("A Primer on Heroin," p. 813) that "the morphine equivalent of $30 worth of heroin [presumably in 1975 prices] is available through legal medical channels for about $0.20." Figures on "pharmacy cost" quoted by Szasz (Ceremonial Chemistry, p. 192) are considerably lower.
37. Moore, Buy and Bust, p. 84.
38. Economists have recently begun work on theories of habit formation and self-control.
Their application to "addiction" is uncertain. See Thomas C. Schelling, "Self-Command in Practice, in Policy, and in a Theory of Rational Choice," American Economic Review 74 (May 1984): 1-11; Richard H. Thaler and Hersh M. Shefrin, "An Economic Theory of Self-Con-trol," Journal of Political Economy 89 (April 1981): 392-406; Gordon C. Winston, "Addic-tion and Backsliding: A Theory of Compulsive Consumption," Journal of Economic Behavior and Organization 1, no. 4 (1980): 295-324.
39. Additionally, time that might be spent working may be necessary to consume the good in a satisfying way. I discuss heroin and employment in Sec. V.
40. Mark H. Moore, "Policies to Achieve Discrimination on the Effective Price of Her-oin," American Economic Review 63 (May 1973): 270-77.
41. Moore, Buy and Bust, p. 90.
42. John R. Pekkanen, "Drug Law Enforcement Efforts," in Drug Abuse Council, The Facts About "Drug Abuse" (New York: Free Press, 1980), pp. 63-94, esp. p. 92.
43. James Q. Wilson, Mark H. Moore, and I. David Wheat, Jr., "The Problem of Heroin," The Public Interest 29 (Fall 1972): 3-28; DuPont and Greene, "The Dynamics of a Heroin Addiction Epidemic."
44. Kaplan, "A Primer on Heroin," p. 807. The difficulty of withdrawal is frequently cited as a justification for methadone programs or medical dispensation of heroin. (See, for example, Trebach, ed., Drugs, Crime, and Politics.)The argument seems strained. In the short run, many of my demands will be inelastic, i.e., the fact that there are high costs of "withdrawing" from electricity use does not constitute either an economic or an ethical rationale for its collective production or allocation.
45. Dan Waldorf and Patrick Biernacki, "Natural Recovery from Heroin Addiction: A Review of the Incidence Literature," Journal of Drug Issues 16 (Spring 1979): 281-89; Winick, "Some Aspects of Careers of Chronic Heroin Users."
46. David F. Musto, The American Disease (New Haven: Yale University Press, 1973); Charles E. Terry and Mildred Pellens, The Opium Problem (New York: Bureau of Social Hygiene, 1928).
47. Jim Mintz, Charles P. O'Brien, and Beverly Pomerantz, "The Impact of Vietnam Service in Heroin-Addicted Veterans," American Journal of Drug and Alcohol Abuse 6, no. 1 (1979): 39-52; Lee N. Robins, "Estimating Addiction Rates and Locating Target Populations," in NIDA Research Monograph 16 (Washington, D.C.: Government Printing Office, 1977), pp. 25-39.
48. Robins, "Estimating Addiction Rates," p. 29.
49. Consider a linear demand curve. Although the slope of such a curve is constant, elasticity will vary as we move along it, since elasticity is a ratio of percentage rather than absolute changes.
50. Edwin T. Fujii, "Public Investment in the Rehabilitation of Heroin Addicts," Social Science Quarterly 55 (June 1974): 39-51.
51. George F. Brown, Jr. and Lester P. Silverman, "The Retail Price of Heroin: Estimation and Applications," Journal of the American Statistical Association 69 (September 1974): 595-606; Lester P. Silverman and Nancy L. Spruill, "Urban Crime and the Price of Heroin," Journal of Urban Economics 4 (1977): 80-103.
52. The logic of the calculation is as follows: I use Moore's estimate (Buy and Bust, p. 90) of average daily use over all classes of users as 46 milligrams of pure heroin per day. Brown and Silverman ("The Retail Price of Heroin," p. 597) estimate an average price per pure gram of $231 in Manhattan in 1974. Such a purchase, however, would not normally be made in a day by a final user. Assume that the average purchase is made each day, and that the search time and various risks have a cash value of $5 to the purchaser. Then the full cost per pure gram is approximately $340. Consequently, let legalization reduce the price of a pure gram to $5. (In 1974 dollars, this is considerably higher than the prices mentioned in Sec. II.) Such action will also eliminate search time and most of the risk components of consumption. A drop in price of this magnitude (to 1.47 percent of its former level) leads to the results of the text if elasticity is 0.25.
53. The long-run elasticity assumed here is roughly equivalent to that estimated by others for alcohol. See Rodney T. Smith, "The Legal and Illegal Markets for Taxed Goods: Pure Theory and Application to State Government Taxation of Distilled Spirits," Journal of Law and Economics 19 (August 1976): 393-430. It is unlikely that the elasticity along the curve would be everywhere that high. See Note 49 above.
54. Musto, The American Disease, p. 275.
55. James M. N. Chi'en, "Voluntary Treatment for Drug Abuse in Hong Kong," Addic-tive Diseases 3 (1977): 99-104.
56. I. D. McIntosh, "Population Consumption of Alcohol and Proportion Drinking," British Journal of Addiction 76 (1981): 267-79.
57. Using BNDD Register data as the dependent variable in a cross-state multiple regres-sion, Leveson estimated an income elasticity of addiction of 1.7. (See Irving Leveson, "Drug Addiction: Some Evidence on Prevention and Deterrence," in Leveson, ed., Quantitative Ex-plorations in Drug Abuse Policy, p. 91.) Since the median income of users bears no necessary relation to median income in the state (the measure Leveson used), the results are questionable. Income elasticity is the ratio of the percentage change in addicts to the percentage change in income as we move between two otherwise identical situations.
58. This assumes more hours of work by the individual in response to rising wages, which need not be the case in general.
59. Although we think of heroin today as consumed by the poorly-off, opium importation in the United States between 1870 and 1914 increased significantly with increasing per capita income. (For import figures, see Terry and Pellens, The Opium Problem, pp. 50-51.) The positive "income elasticity" of demand might, however, actually be due to other influences. The price of opium, for example, might have been falling at the same time incomes were rising. Were price data available, multiple regression techniques would allow further analysis.
60. John C. Ball and Carl D. Chambers, "Overview of the Problem," in Ball and Chambers, eds., The Epidemiology of Opiate Addiction in the United States (Springfield, Ill.: Thomas, 1970), p. 9; Terry and Peliens, The Opium Problem, pp. 428-67.
61. Terry and Peliens, The Opium Problem, chap. 1.
62. John C. Ball, "Two Patterns of Opiate Addiction," in Ball and Chambers, eds., The Epidemiology of Opiate Addiction, pp. 81-94. The stories of addiction as a peculiar problem for blacks seem inadequate. According to Johnson, "the propensity of blacks to addiction is explained by the fact that larger proportions of blacks than whites experiment with heroin; but very seldom does such use translate into recent use or addiction for either race." See Bruce D. Johnson, "The Race, Class, and Irreversibility Hypotheses: Myths and Research about Heroin," in NIDA Research Monograph 16 (1977), pp. 51-57, esp. p. 52.
63. Greene, "Resurgence of Heroin Abuse."
64. Ball and Chambers, "Overview of the Problem," p. 10.
65. Horace Freeland Judson, Heroin Addiction in Britain (New York: Harcourt Brace Jovanovich, 1974), pp. 46-49.
66. K. Singer, "The Choice of Intoxicant Among the Chinese," British Journal of the Addictions 69 (September 1974), pp. 257-68.
67. Arnold S. Trebach, The Heroin Solution (New Haven: Yale University Press, 1982), pp. 15-16.
68. Ernest Drucker and Victor Sidel, "The Communicable Disease Model of Heroin Addiction: A Critique," American Journal of Drug and Alcohol Abuse 1, no. 3 (1974): 301-11, esp. pp. 308-9.
69. Wilson, Moore, and Wheat, "The Problem of Heroin," pp. 10-12; Denise B. Kande! and Deborah R. Maloff, "Commonalities in Drug Use: A Sociological Perspective," in Peter K. Levison, Dean Gerstein, and Deborah Maloff, eds., Commonalities in Substance Abuse and Habitual Behavior (Lexington, Mass.: Lexington, 1983), pp. 3-27.
70. The mythology carries with it some interesting inconsistencies. Why, for example, would a profit-making pusher single out the least affluent part of the community for attention?
71. Robert R. Carr and Erik J. Meyers, "Marijuana and Cocaine: The Process of Change in Drug Policy," in Drug Abuse Council, The Facts about "Drug Abuse" (New York: Free Press, 1980), pp. 153-89.
72. Epstein, Agency of Fear, pp. 109, 177.
73. Richard Ashley, Heroin (New York: St. Martin's, 1972), pp. 40-47.
74. Max Singer, "The Vitality of Mythical Numbers," The Public Interest 23 (Spring 1971): 3-9. As will be seen in Sec. V, the link between drug use and crime is also doubtful.
75. Epstein, Agency of Fear, p. 44.
76. Peter Reuter, "The (Continued) Vitality of Mythical Numbers," The Public Interest 75 (Spring 1984): 135-47. We might try going the other way and attempt to estimate user populations from crime statistics. This seems equally hopeless. A detailed list of the assumptions that must be made in order to do so appears in John A. Newmeyer and Gregory R. Johnson, "The Heroin Epidemic in San Francisco: Estimates of Incidence and Prevalence," International Journal of the Addictions 11(1976): 417-38, esp. 430-32.
77. GAO, Report to the Attorney General, chap. 5.
78. President Carter's 1980 State of the Union message reported an addict population of 380,000. I have been unable to find an official source for the figure. See Trebach, The Heroin Solution, p. 243.
79. Estimates of amounts of heroin seized and unseized might provide additional infor-mation, but estimates of total traffic are currently provided only for internal federal use. (See U.S. House of Representatives, Appropriations Committee, Hearings, Part 7. Department of Justice Appropriations for 1983, pp. 15, 37.) Publicly available seizure estimates for past years are duplicative (some Customs and DEA seizures are double-counted) and inaccurate. (Gross weights rather than potency-adjusted weights are reported.) (See GAO, Report to the Attorney General, pp. 36-39.) In any case, estimates of the amounts of drugs not seized, regardless of who makes them, are pure conjecture.
80. Reuter, "The (Continued) Vitality of Mythical Numbers," pp. 145-46.
81. Terry and Pellens, The Opium Problem, p. 42.
82. Moore, Buy and Bust, p. 72.
83. William Butler Eldridge, Narcotics and the Law, 2d ed. rev. (Chicago: University of Chicago Press, 1967), p. 76.
84. U.S. Senate, Hearings Pursuant to S. Res. 178, part 3. Organized Crime, p. 769.
85. Moore, Buy and Bust, p. 90.
86. Kaplan, "A Primer on Heroin," pp. 807-8.
87. Fred Goldman, "Drug Abuse, Crime and Economics: The Dismal Limits of Social Choice," in James A. Inciardi, ed., The Drugs-Crime Connection (Beverly Hills, Calif.: Sage, 1981), p. 164.
88. Diane McDermott and James Scheurich, "The Logarithmic Normal Distribution in Relation to the Epidemiology of Drug Abuse," Bulletin on Narcotics 29 (January-March 1977): 13-19; Reginald G. Smart, "The Distribution of Illicit Drug Use: Correlations between Extent of Use, Heavy Use, and Problems," Bulletin on Narcotics 30 (January-March 1978): 33-41.
89. Erik J. Meyers, "American Heroin Policy: Some Alternatives," in Drug Abuse Council, The Facts about "Drug Abuse" (New York: Free Press, 1980), p. 198.
90. The state of the art, such as it is, is adequately summarized in Adele V. Harrell, "Validation of Self-Report: The Research Record," in NIDA Research Monograph 57 (1985), pp. 12-21.
91. William Glenn, "Problems Related to Survey Sampling," in NIDA Research Monograph 10 (1977), pp. 154-59, esp. p. 155.
92. There is a potentially important bias in the reporting of heroin admissions. Federal policy makers apparently have made it clear that "drug abuse" clinics are not to concern themselves with marijuana and alcohol users, and that the funding depends in part on reported heroin treatments. See C. James Sample, "Institutional Data—CODAP," in NIDA Research Monograph 10 (1977), pp. 197-205.
93. NIDA, Data from the Client-Oriented Data Acquisition Process, Annual Data 1980, Statistical Series E, No. 21 (Washington, D.C.: Government Printing Office, 1982), p. 6.
94. Ibid., p. 9.
95. Moore, Buy and Bust, p. 72, note i.
96. John A. O'Donnell, "Comments on Hunt's Estimation Procedures," in NIDA Research Monograph 16 (1977), pp. 96-102; S. B. Sells, "Reflections on the Epidemiology of Heroin and Narcotic Addiction from the Perspective of Treatment Data," also in NIDA Research Monograph 16 (1977), pp. 147-76, esp. p. 148.
97. Moore, Buy and Bust, p. 72. The BNDD Register has no criteria for defining a user or verifying one's status as a user. The General Accounting Office declared that reporting by local police to the BNDD was "limited, incomplete, and erratic" (Report to the Attorney General, p. 43). In any case, compilation of the Register ceased in December 1977.
A few other estimates, along the same methodological lines, are given by Marc D. Brodsky, "History of Heroin Prevalence Estimation Techniques," in NIDA Research Monograph 57 (1985), pp. 94-103. A more sophisticated multiplier technique is examined by J. Arthur Woodward, Douglas G. Bonnett, and M. L. Brecht, "Estimating the Size of a Heroin-Abusing Population Using Multiple-Recapture Census," in NIDA Research Monograph 57 (1985), pp. 158-71.
98. Moore, Buy and Bust, p. 72.
99. GAO, Report to the Attorney General, chap. 3.
100. In general, a death associated with heroin use is indiscriminately referred to as an overdose. It may stem from unexpectedly high potency in a street purchase or from reaction to an adulterant. It may also be a consequence of multiple drug use or of the health status of the user. Some are purposive suicides. For a summary of the possibilities, see David P. Desmond, James F. Maddux, and Aureliano Trevino, "Street Heroin Potency and Deaths from Overdose in San Antonio," American Journal of Drug and Alcohol Abuse 5 (March 1978): 39-49.
101. Louis A. Gottschalk and Frederick L. McGuire, "Psychosocial and Biomedical Aspects of Deaths Associated with Heroin and Other Narcotics," in NIDA Research Monograph 16 (1977), pp. 122-29; Leon Gibson Hunt, "Prevalence of Active Heroin Use in the United States," also in NIDA Research Monograph 16, pp. 61-86, esp. p. 77.
102. The basic idea here is a statistically sound one, frequently used for counting wildlife populations. Assume that we wish to estimate the number of fish in a closed lake. First, we catch an appropriately sized sample of them, tag them, and wait while they swim around randomly. Using the hypergeometric distribution, we can then estimate the entire population after we learn the number of tagged fish that appear in a subsequent catch. A city is not a closed lake, users do not necessarily remain users, and the probability of a first arrest or treatment may differ from the probability of a second. For a good summary of statistical techniques, which have not since been improved, see Alfred Blumstein, Philip C. Sagi, and Marvin E. Wolfgang, "Problems of Estimating the Number of Heroin Addicts," in U.S. National Commission on Marijuana and Drug Abuse, Technical Papers of the Second Report, vol. 2 (Washington, D.C.: Government Printing Office, 1973), pp. 201-11. See also Hunt, "Prevalence of Active Heroin Use," pp. 68-71.
103. GAO, Report to the Attorney General, pp. 42-43.
104. Ibid., p. 47.
105. To determine the 1974 and 1975 figures, NIDA calculated an "index" of heroin problems for each of twenty-four metropolitan areas. The five variables included in the index were heroin-related injuries and deaths, retail price, retail purity, and admissions of users to federally funded treatment programs. All except the last are measured with great error, and NIDA has not released the index formula or a rationale for it. Once the index is computed, it must still be inflated to an estimate of users in each area, and then that estimated total must be inflated to a national value. In both inflation procedures, the size of the multiplier is arbitrary. While a large amount of computation is needed, it is fair to say that the resultant "estimates" are really assumed figures.
106. GAO, Report to the Attorney General, pp. 11-12.
107. Ibid., p. 48.
108. A summary of recent budgetary data appears in Peter Goldberg, "The Federal Gov-ernment's Response to Illicit Drugs, 1969-1978," in Drug Abuse Council, The Facts about "Drug Abuse," pp. 56-60.
109. Moore, Buy and Bust, p. 94.
110. Goldman, "Drug Abuse, Crime and Economics," pp. 174-75.
111. Moore, Buy and Bust, p. 92.
112. NIDA, Data from the Client-Oriented Data Acquisition Process, Annual Data 1980, p. 6.
113. Ibid., p. 7.
114. Preble and Casey, "Taking Care of Business."
115. Paul J. Goldstein, "Getting Over: Economic Alternatives to Predatory Crime Among Street Drug Users," in Inciardi, ed., The Drugs-Crime Connection, p. 82.
116. Epstein, Agency of Fear, p. 296.
117. A survey of Oakland pawnbrokers by James Roumasset and John Hadreas ("Addicts, Fences, and the Market for Stolen Goods," Public Finance Quarterly 5 [April 19771: 247-72, esp. p. 257) yielded the conclusion that, on average, the drug buyer receives 15 percent of the value of a newly sold item. Compare U.S. Senate Hearings Pursuant to S. Res. 265, p. 2692, which estimates (without data) that a "fence" gives one-third to one-half of "cost." Cost is undefined.
118. Preble and Casey, "Taking Care of Business," pp. 18-19.
119. U.S. President's Commission on Law Enforcement and Administration of Justice, Task Force Report: Narcotics and Drug Abuse (Washington, D.C.: Government Printing Office, 1967), p. 11; Epstein, Agency of Fear, p. 40.
120. Using BNDD Register data, "addicts" per 1,000 population in 1969 in major cities ranged from 0.43 (St. Louis) to 5.33 (Paterson, N.J.). (See Leveson "Drug Addiction," p. 76.) The Register, however, is sufficiently unreliable and nonrandom as to call these findings into question. See GAO, Report to the Attorney Genera!, pp. 42-43, and also note %.
121. Adding the BNDD-measured addiction rate to Ehrlich's cross-state multiple regression model of property crime rates yields a significant increase (4 percent) in the explanatory power of the equation (R2). The implied addiction-elasticity of property crime is a relatively low .04 to .09. See Isaac Ehrlich, "Participation in Illegitimate Activities: A Theoretical and Empirical Investigation," Journal of Political Economy 81 (May 1973): 521-65; Leveson, "Drug Addiction," p. 93; DuPont and Greene, "The Dynamics of a Heroin Addiction Epidemic"; Newmeyer and Johnson, "The Heroin Epidemic in San Francisco."
122. DuPont and Greene, "The Dynamics of a Heroin Addiction Epidemic," p. 721.
123. Michael Alexander, "The Heroin Market, Crime and Treatment of Heroin Addiction in Atlanta," in Proceedings of the Fifth National Conference on Narcotic Treatment (New York: National Association for the Prevention of Addiction to Narcotics, 1973), pp. 733-51.
124. Harold Finestone, "Narcotics and Criminality," Law and Contemporary Problems 22 (1957): 69-85, esp. 74, 80-81; Carl D. Chambers, "Narcotic Addiction and Crime: An Empirical Review," in James Inciardi, ed., Drugs and the Criminal Justice System (Beverly Hills, Calif.: Sage, 1974), pp. 125-42.
125. Isidor Chein et al., The Road to H (New York: Basic Books, 1964), p. 65.
126. Finestone, "Narcotics and Criminality."
127. Douglas Coate and Fred Goldman, "The Impact of Drug Addiction on Criminal Earnings," in Leveson, ed., Quantitative Explorations in Drug Abuse Policy, chap. 4.
128. Goldstein, "Getting Over," p. 82.
129. David P. Ausubel, Drug Addiction (New York: Random House, 1958), pp. 67-69.
130. Goldman, "Drug Abuse, Crime and Economics," p. 177.
131. Robert M. J. Haglund and Charles Froland, "Relationship between Addict Crime and Drug Treatment: Two Cohorts Examined," American Journal of Drug and Alcohol Abuse 5 (1978): 455-62.
132. Edward Jay Epstein, "Methadone: The Forlorn Hope," The Public Interest 36 (Summer 1974): 3-24, esp. 11.
133. Fujii estimates the cost of enforcing the heroin laws at $20.8 million in California in 1970. The assumptions he uses to make the estimate are untenable. See his "Public Investment in the Rehabilitation of Heroin Addicts," p. 50.
134. Ausubel, Drug Addiction, p. 13.
135. Szasz, Ceremonial Chemistry, pp. 69-75.
136. The proportionate breakdown between drug and other expenses is that cited by Moore, Buy and Bust, p. 85.
137. Goldstein, "Getting Over."
138. Economic principles of specialization also seem to hold for the generation of user income. Over 50 percent of the income from predatory crime acquired by the users Goldstein studied ("Getting Over," p. 70) was earned by less than 20 percent of the group. For those active in the distribution chain, there are a variety of specialized roles (ibid., pp. 71-74).
139. NIDA, Data from the Client-Oriented Data Acquisition Process, Annual Data 1980, p. 6.
140. Eldridge, Narcotics and the Law, p. 27.
141. Goldman, "Drug Abuse, Crime and Economics," pp. 170-72.
142. John C. Kramer, "Speculations on the Nature and Pattern of Opium Smoking," Journal of Drug Issues 12 (Spring 1979): 247-55.
143. GAO, Report to the Attorney General.
144. Reuter, "The (Continued) Vitality of Mythical Numbers."