Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Regulating the Supply of Alcoholic Beverages For much of U.S. history, commerce in alcoholic beverages has been regulated and subject to relatively high taxes. A disparate set of objec- tives has motivated this historical regulatory effort, the relative impor- tance of these objectives varying with shifts in public attitudes. The current regulatory structure reflects a paramount concern with main- taining an "orderly" commercial trade in alcoholic beverages and main- taining tax revenues. Also reflected in current regulations, though less visibly, is a concern for promoting temperance and protecting the public from adverse consequences of drinking. Minimum age restrictions on sales, limits on the number and nature of sales outlets, and (to some extent) high taxes are intended to limit the availability of alcohol and thereby reduce the harm engendered by high-volume or inappropriate drinking. Public enthusiasm for restrictive regulation of alcohol commerce peaked in the early decades of this century, when many states and then the nation adopted a prohibition on the supply of "intoxicating liquors," the term actually used in the 18th Amendment. While nearly all parties agreed that this meant potable distilled spirits, there were continual arguments during Prohibition about whether lower-proof alcoholic bev- erages should be covered. In addition, only "manufacture, sale or trans- portation," but not possession, consumption, or home production were covered (Aaron and Musto, in this volume). Thus, even during the period of greatest legal constraint on supply, the national laws governing alcohol use were by no means "bone-dry." 61
62 REPORT OF THE PANEL Since repeal in 1933 the regulatory emphasis has been on maintaining orderly markets and collecting the legal tax revenues. The national experiment with Prohibition has been deemed an overwhelming failure. But very recently there has been renewed interest in using commercial regulation of the alcohol beverage industry to advance prevention goals. This chapter reviews the regulations governing trade in alcoholic bev- erages. In each case, the main concern is with the potential efficacy of these regulations in preventing alcohol-related problems. We begin with a brief review of the effects of Prohibition, both as an influence on subsequent policy and as a source of evidence on the effectiveness of strategies to reduce supply in preventing alcohol-related problems. We then provide an overview of the current federal, state, and local regu- latory structure. Subsequent sections discuss taxation, the myriad of regulations governing retail outlets, and minimum drinking age restric- tions. THE LESSONS OF PROHIBITION It is widely believed that Prohibition was a failure and that it demon- strated once and for all the futility of attempts to legislate morality. The legacy of Prohibition has been to strongly discourage the use of alcoholic beverage control (ABC) laws as preventive instruments. There is no question that the Volstead Act was widely violated and that smuggling, moonshining, and speakeasies all thrived during the Prohibition era. The crime fostered by Prohibition was of paramount importance in the public mind at the time of repeal. Fosdick and Scott, after interviewing a number of opinion leaders in the early 1930s, found substantial support for their first "principle" (1933, p. 15~: At all costs even if it means a temporary increase in consumption of alcohol- bootlegging, racketeering and the whole wretched nexus of crime that devel- oped while the Eighteenth Amendment was in force must be wiped out. The defiance of law that has grown up in the last fourteen years, the hypocrisy, the break-down of government machinery, the demoralization in public and private life, is a stain on America that can no longer be tolerated. But while Prohibition was a failure in the sense that government authorities failed to suppress the emergence of a vigorous illegal supply network, there is nevertheless strong evidence to suggest that the "noble experiment" was an instructive failure. Law enforcement during this period was effective enough to raise alcohol prices and to reduce the ease of availability. Clark Warburton (1932), in his seminal study The Economic Results of Prohibition, demonstrated that consumption de-
63 Regulating the Supply of Alcoholic Beverages clined considerably, especially among the "working class"; various in- dications of the prevalence of heavy drinking, such as arrests for public drunkenness and admissions to mental hospitals for alcoholic psychosis, declined to low levels during the first few years of Prohibition. The most reliable indicators of this sort, acute alcohol overdose deaths and the rate of mortality due to liver cirrhosis, dropped well below their pre- Prohibition levels, reaching minima lower than any observed before or since in the 20th century. This partial result of Prohibition in preventing alcohol-related prob- lems has been overshadowed by its dramatic failures. As Room and Mosher note (1979-1980, p. 1 14: In American political and intellectual life since Repeal, Prohibition, an attempt at a structural and societal solution to alcohol problems, has been seen as an entirely negative experience, and those interested in alcohol problems and in helping alcoholics have often been concerned to dissociate themselves from the taint of temperance. In particular, it appears that ABC administrators have little faith in the possibility of promoting temperate drinking habits through supply reg- ulations. In a recent study, a number of these ABC administrators were interviewed by researchers for Medicine in the Public Interest (1979), to determine their perceptions of the goals of existing ABC laws (p. 27~. The overwhelming majority of those interviewed feel that their primary, if not exclusive, purpose is regulatory and not in any way related to the public health aspects of consumption. Interviewees stressed the impor- tance of revenue collection, maintaining orderly markets, and excluding criminal elements from the business. The report also concluded that state legislators are "generally skeptical about the effect of regulations, including taxation, on the incidence, patterns, or circumstances of use," (p. 31) and focus instead on issues related to tax revenues and economic regulation. The repeal advocates of the 1930s did not take this position. Fosdick and Scott report that the opinion leaders they talked to were very concerned with promoting temperate drinking practices through supply regulation. They advocated a state monopoly on retail trade and/ or moderate taxation to promote this end. To sum up, the real lessons of Prohibition are threefold: (1) Drinking customs in the United States are strongly held and re- sistant to frontal assault. It is well beyond the will or capacity of gov- ernment ever to eradicate the customary demand for alcoholic bever- ages. (2) A criminal supply network emerges if not instantly, then within
64 REPORT OF THE PANEL a few yearsif production and sale of alcoholic beverages are outlawed. The prices and extent of this criminal supply depend on the degree of public support for the law and the resources devoted to law enforcement. (3) The quantity of alcohol consumption and the rates of problems varying with consumption can, however, be markedly reduced by sub- stantial increases in real price and reductions in the ease of availability. In contrast, the lesson that has apparently become ingrained in con- ventional wisdom is something like the following: It is futile and mis- chievous to legislate drinking morals. Prohibition and, by extension, even moderate supply restrictions create a criminal industry and are not effective in reducing the consumption or the problems of alcohol. Neither reading favors a return to prohibition. But there is much to be gained by unburdening discussion of other prevention strategies from the distorted image of the Prohibition experience that currently prevails. CURRENT INSTITUTIONS AND POLICIES Since the repeal of Prohibition the task of regulating commerce in al- coholic beverages has been left primarily to the states, but the federal government does play an important role. The U.S. Bureau of Alcohol, Tobacco, and Firearms (BATE) licenses importers, manufacturers, and wholesalers and regulates the advertising, size of containers, and labeling of alcoholic beverages. Although states have authority in this area of the market, they have largely left it to the federal government to manage. BATE is also responsible for collecting federal taxes on alcoholic bev- erages (which are ordinarily higher than state taxes) and suppressing illegal production. The U.S. Food and Drug Administration is respon- sible for overseeing the purity and cleanliness of alcoholic beverages. In addition to this general regulatory role, the federal government has direct control over the supply of alcoholic beverages on military reser- vations, a jurisdiction that is bigger than most states (8 million people are currently eligible to buy at outlets on military reservations) as well as direct control over national parks and waterways, rail, and air carriers. Other federal agencies have an indirect influence on the availability of alcohol: the Small Business Administration has been active in lending money to tavern and liquor store owners, and the Internal Revenue Service's rulings on business-related expense deductions have encour- aged the notorious three-martini lunch (Mosher and Mottl, in this vol- ume). State systems for controlling the alcoholic beverage trade regulate almost every aspect of retail sale. All states set a minimum age for legal
Regulating the Supply of Alcoholic Beverages 65 consumption, ranging from 18 to 21, and prescribe penalties for retailers who knowingly sell to underage customers. All states have enacted special excise taxes on alcoholic beverages and most have placed re- strictions on advertising, hours of legal sale, selling on credit, and so forth. The 21st Amendment left the "dry" option to the individual states. While all states now permit alcoholic beverage sales, most have passed the option on to local jurisdictions: in 1976, 3.5 percent of the population still resided in dry counties (DISCUS 1977~. During the 1930s, 18 states chose to separate private profit from one part of the trade by creating state or county monopolies to control both wholesale distribution and (except in Wyoming) retail sales of (at least) spirits. The remaining 32 states plus the District of Columbia have adopted a licensing system, whereby a state regulatory agency is em- powered to decide which wholesalers and package retailers will be per- mitted to operate in the state. This agency, called the Alcoholic Bev- erage Control Board in most states, promulgates detailed rules to implement the state's ABC laws and sets policy on the density, location, and nature of outlets through its licensing activities. These agencies also license outlets that sell alcoholic beverages for on-premise con- sumption, and the counterpart agencies in monopoly states have this responsibility as well. States differ in the extent to which local govern- ments are part of the licensing process. Fair trade laws govern pricing policies in some license states, and these are enforced by ABC boards. State and local governments have promulgated a wide variety of or- dinances governing the nature and operation of establishments that sell drinks. Of greatest interest here are the "dramshop" laws, which es- tablish civil liability for employees and owners of establishments in which a drunken or underage patron is served and then causes an accident. Civil liability has been established either by statute, by court decisions, or both in 27 states (Mosher 1979~. In this era of extensive economic regulation by federal and state government, the alcoholic beverage industry remains one of the most heavily regulated. But the trend in regulatory activity during the last two decades has been toward permitting increased supply; taxes and prices have declined (relative to the overall price level), and drinks are being sold at an increasing number and range of outlets. During the early 1970s a number of states lowered their minimum drinking age (although several of these have since reverted to a higher age). The ~ Some license states require that liquor be sold in stores that specialize in liquor or alcoholic beverages. One state (Missouri) limits liquor sales to stores that sell some other product line. The majority of license states do not place severe restrictions on the nature of package store outlets.
66 REPORT OF THE PANEL central question of this chapter is whether the regulatory apparatus that has gradually been weakening might be used to improve alcohol prob- lems. In the next section, we state the general case for using regulatory powers in this way. In subsequent sections, we review the available evidence on specific kinds of control mechanisms. THE SINGLE DISTRIBUTION THEORY A fundamental postulate of economics states that, other things being equal, an increase in the cost of a commodity to consumers will reduce the quantity they demand. For a given cost increase, the extent of this reduction in quantity demanded may vary across a broad range of "elasticities" for different individuals. Nevertheless, an increase in taxes or a general reduction in the availability of alcoholic beverages should reduce overall alcohol consumption for both moderate and heavy drink- ers. Such a reduction, particularly to the extent that it involves heavy drinkers, would in turn yield public health benefits. This line of argument is rejected by those who equate alcohol prob- lems with alcoholism and believe that the alcoholic is insensitive to economic (market) factors such as price and availability. In this view, alcoholic beverage control measures simply make alcoholics' lives more difficult without helping to cure them of their disease. In response, we make two general observations: (1) As we have argued above, alco- holism is only one of several consequences of alcohol consumption de- serving of public concern, and (2) there is considerable evidence, both direct and indirect, to suggest that the volume of heavy drinking is sensitive to economic factors. The indirect evidence is reviewed here and the direct evidence in subsequent sections. The indirect evidence in support of the influence of market forces on the prevalence of heavy drinking rests on an analysis of the distribution of alcohol consumption. If a group of alcoholics exists whose drinking habits are not sensitive to alcohol-related features of their socioeconomic environment, then we would expect the amount of alcohol consumed by the top 5 or 10 percent of drinkers to have little relationship to the drinking habits of the remainder of the population. This hypothesis has been evaluated and rejected by Ledermann (1956) and his followers. Ledermann's studies of the distribution of consumption persuaded him that there exists a precise relationship between per-capita con- sumption and the proportion of drinkers consuming in excess of a spec- ified level (e.g., three ounces of pure alcohol per day). Ledermann's evidence for the existence of such a "distribution law" was survey and other evidence on the drinking habits of several groups of people; he found that these groups differed widely with respect to average con-
Regulating the Supply of Alcoholic Beverages 67 gumption (and in other respects), but in each case the distribution of consumption among the group's members was approximated by a "log normal" curve, with approximately the same variance for each group. Ledermann concluded that a sufficient and virtually necessary condition for reducing the volume of heavy drinking in a population was to reduce per-capita consumption. Since Ledermann's pioneering work, additional data on the distri- bution of drinking have been analyzed for a number of population groups in North America and Europe (Bruun et al. 1975~. These distributions, like those of Ledermann, for the most part are reasonably approximated by a log normal curve. In all cases they exhibit a single mode and are highly skewed to the right' with the top 10 percent of drinkers consuming in the neighborhood of 40-50 percent of the total. However, Skog (1971) and others (Popham et al. 1976) who have followed Ledermann's path express their conclusions in terms of tendencies rather than rigid laws of nature. The definitive statement on this subject by an international panel (including both Skog and Popham) was expressed as follows (Bruun et al. 1975, p. 45~: 1. A substantial increase in mean consumption is very likely to be accom- panied by an increased prevalence of heavy users.... 2. If a government aims at reducing the number of heavy consumers, this goal is likely to be attained if the government succeeds in lowering the total consumption of alcohol.... As the distribution of consumption is highly skewed, a substantial proportion of the total amount drunk by a population is consumed by heavy drinkers, so that a sizeable reduction in total consumption will not occur unless some of the heavy drinkers reduce their consumption. This conclusion is stated conservatively' since it claims simply that per-capita consumption is a good indicator of (covaries substantially with) the prevalence of heavy drinking; this claim is entirely responsible, given the arithmetic importance of heavy drinkers in determining the mean consumption level. But the results suggest a stronger conclusion, i.e., a close link between the median consumer's drinking level and the consumption level of the consumer at the 90th or 95th percentile of the distribution. That is, to some degree the same factors that determine the intake of the relatively moderate median drinker also appear to influence the intake of the relatively heavy consumer. The drinking level of the typical, relatively extreme consumer is sensitive to many of the same cultural, socioeconomic, and legal forces as influence other con- sumers. This relationship between median and 95th-percentile con- sumption has not been directly tested, but its probability is strong enough to force consideration in designing policy. This conclusion does not preclude the possibility of designing policies
68 REPORT OF THE PANEL that are specifically focused on heavy consumers. For example, a tax policy designed to raise the price of the least expensive source of alcohol (cheap fortified wine) would probably have a greater effect on destitute alcoholics than on other drinkers. Our assessment of the distribution literature does suggest that overall trends in price and availability are likely to affect the entire distribution of consumption; there is no evi- dence that a large fraction of drinkers is entirely immune to such forces. It is worth noting that the evidence is comprised of econometric studies of aggregate indicators. Firmer guides to policy may be provided by additional, less aggregated studies, such as the effects of "naturally occurring" price changes on longitudinal panels of individual consumers. Direct evidence on these matters must be derived from evaluations of specific changes governing price and availability within a jurisdiction. Some evidence of this sort is available and is included in the review of alternative control strategies presented below. TAXATION AND PRICES Alcoholic beverages were first subject to federal taxation in the United States in 1791 (Hu 1950~. Indeed, a liquor excise was the first internal revenue law enacted by Congress under the Constitution. Alcohol tax revenues were a major source of income for the federal government until Prohibition; these revenues constituted 80 percent of all federal internal tax collections in 1907 and about 10 percent at the beginning of World War II. Currently the federal tax on alcoholic beverages has a large effect on alcohol beverage prices but figures very lightly (less than 1 percent) in the federal budget. Federal tax rates on alcoholic beverages were last changed in 1951? when they were set at $.29 per gallon of beer, between $.17 and $3.40 per gallon of wine depending on alcohol content and type, and $10.50 per proof gallon for distilled liquor. (A "proof gallon" is defined as 1 gallon of 100 proof t50 percent] liquor, or 1.2 gallons of 80 proof liquor, or, in general, the volume of liquor of any proof that contains 2 quarts of ethanol.) The states also levy taxes on alcoholic beverages; in license states, the liquor tax rates range from $1.50 to more than $4.00 per proof gallon. Total tax collections from liquor sales constituted about 0.5 percent of total expenditures by consumers in 1977; much less went for beer and wine taxes. Of all consumer expenditures for alcoholic beverages, just over one-third went to government as tax receipts (Hy- man et al. 1980~. States influence alcohol prices through fair trade laws (in all but a few license states) and set them by administrative fiat in monopoly states. In recent years, tax and price decisions by state governments, combined
Regulating the Supply of Alcoholic Beverages 69 with federal inaction with respect to federal tax rates, have contributed to a steady downward trend in the prices of beer, wine, and spirits, when compared with the consumer price index. Between 1960 and 1980, the "real" cost to the consumer of a bottle of liquor declined by 48 percent, of beer by 27 percent, and of wine by 19 percent (Cook, in this volume). If drinking is responsive to price, then declines of this mag- nitude should have provided considerable stimulus for increased con- sumption. Obtaining reliable measures of the responsiveness of alcohol con- sumption to changes in prices is difficult for a number of reasons, as discussed by Cook (in this volume) and Ornstein and Levy (no date). A number of econometric studies provide estimates of price and income elasticities for each of the three beverage types, for both the United States and other countries. A comprehensive review of these studies is presented by Ornstein (1980~. Most of these studies conclude that the demand for alcoholic beverages, like other commodities, is responsive to price. The demand for beer is probably moderately inelastic in the United States today, meaning that an increase in price will result in people budgeting more money for beer but buying less of it in total, all other things remaining equal. The demand for spirits appears to be somewhat more responsive to price than the demand for beer. There has only been one study of the demand for wine in the United States (Niskanen 1962), but evidence from other countries suggests that it tends to be fairly sensitive to price (Ornstein 1980.) Sulkunen (1976) has speculated that the less popular beverage types in a given country will generally prove to be more price-elastic than the national favorite. The results of econometric studies differ widely, due to differences in the data and methods of analysis as well as national, regional, and local variations in preferences; therefore, more precise quantitative conclu- sions cannot now be drawn. The effect of changes in alcohol prices on consumption is of interest only if there are corresponding effects on at least some of the adverse health and social consequences of drinking. For example, it is logically possible that alcoholic or problem drinkers could be relatively insensitive to alcohol prices in their consumption decisions, so that price-induced changes in aggregate consumption could result exclusively from a subset of drinkers who do not in any case cause or experience drinking prob- lems. This logical possibility was examined and rejected for the Canadian province of Ontario (Seeley 1960), and similar findings, consistent with Seeley's result, have been derived from other data series (Popham et al. 1978~. The most direct test of the connection between tax-increased alcohol prices and consequences of drinking is the analysis by Cook (in this
70 REPORT OF THE PANEL volume), using an elaboration of the quasi-experimental approach in- troduced by Simon (1966~. Cook examined changes in liquor taxes among 30 license states between 1961 and 1975, in order to ascertain whether liquor tax increases led to statistically discernible changes not only in the indicator of liquor consumption, but also in two alcohol problem indicators: highway crash fatalities and deaths due to cirrhosis of the liver. There were 39 cases of tax increases, ranging from $.25 to $1.75 per proof gallon, during this lS-year period. One can view each tax increase as a "test" case and use these 39 tests (a few happening each year) to see whether such small tax increases are effective, as hypothesized. To do this, Cook first calculated the percentage changes in annual rates of liquor consumption, highway fatalities, and cirrhosis deaths in each of the 30 states in each of the IS years, then rank-ordered these changes from largest decrease (rank 1) to largest increase (rank 30) for each year. If the tax increases had had no independent effect on the drinking or death rates, then one would expect the test cases to be evenly dis- tributed above and below the middle; that is, on each indicator, about half of the 39 test cases would have been above the median rank, half below it. The result is summarized in Table 9 (see Cook, in this volume, for additional tests and details). On each indicator, an excess of cases falls below the median. The excess is very pronounced for liquor consumption (77 percent ranked below), as one would expect. But the differences extended to highway fatality and cirrhosis death rates; the results hover about the conventional border of statistical significance (p 0.054. Even a marginally significant influence from a relatively small pressure (1 percent to 10 percent) on retail price is a suggestive result' since: (a) liquor represents less than half the alcohol consumed in these states, generally; (b) it is estimated by Reed (in this volume) that only about one-fourth of highway fatalities are causally dependent on alcohol use; and (c) cirrhosis death rates are a complicated precipitate of a long drinking history, in addition to the effect of the most recent consump- tion. These results indicate that even relatively small changes in prices may influence not only the quantity of consumption but also the most serious health effects as well. The few historical instances in which there has been a large sustained price change lend credence to this conclusion. A tenfold increase in the price of aquavit in Denmark (the national beverage at the time) during World War I, resulting from tax increases, transformed Denmark into a predominantly beer-drinking country, and in the near term greatly reduced total per-capita alcohol consumption and the prevalence of heavy drinking (Popham et al. 19764. The sub-
Regulating the Supply of Alcoholic Beverages 71 TABLE 9 Ranking of 39 "Test Cases" of Alcohol Tax Increases by Net Percentage of Change in Rates of Liquor Consumption, Highway Deaths, and Cirrhosis Deathsa Rank Among 30 States Liquor Highway Cirrhosis During the Test Year Consumption Deaths Deaths Below median 30 25 24 Above median 9 13 14 Number of tests 39 38h 3sb a States in which a liquor tax increase occurred. b One case, in 1975, was deleted for lack of sufficient data. Source: Adapted from Cook (in this volume). stantial reductions in the prevalence of heavy drinking, cirrhosis mor- tality, and alcohol overdose deaths in the United States during the 1920s could reasonably be viewed as, in large part, a response to sustained Prohibition-induced market price increases, roughly triple to quadruple prices in 1915 (Warburton 1932~. We conclude that alcohol consumption and the problems caused by it respond to the price of alcoholic beverages, and we infer that the large reductions in the real cost of alcohol to consumers in recent years are likely to have exacerbated drinking problems. The downward trend in alcohol prices could be reversed by indexing federal excise taxes to inflation or by making the tax proportional to wholesale price rather than volume. A more extreme action would be to restore the federal tax in real terms to, for example, the 1951 level, which would mean an increase from $10.50 to roughly $30.00 per proof gallon.2 If alcohol taxes are to be viewed as principally a preventive rather than a revenue measure, the appropriate structure of tax rates across different types of alcoholic beverages needs review. Currently, liquor (distilled spirits) is taxed much more heavily than beer or table wine, not only per gallon of beverage, but also per gallon of alcohol content. State and federal ABC laws reflect the widely held view that beer and wine are more temperate or innocuous than liquor; that beer, in par- ticular, is the "drink of moderation," while liquor is more likely to cause problems. However, epidemiological research evidence raises questions about the appropriateness of this distinction. These studies suggest that 2A question of relevance to tax strategies is whether increased tax rates will lead to massive criminal evasion and fraud on the part of buyers. manufacturers, or distributors. There is no single answer to the question. Insofar as the taxes are considered legitimate by the public, as the additional revenues are used to support proportionately intensified enforcement efforts, and as increases are applied uniformly. it seems doubtful that new major criminal activity would be incurred by the relatively modest increases that could be practically considered during an inflationary peacetime.
72 REPORT OF THE PANEL the key factor in the incidence of most major diseases associated with lifetime consumption is simply the daily intake of ethanol, irrespective of the type of beverage (Popham et al. 1976~. Research on the physiology of drunkenness has failed to find important differences among beverage types, and studies of the beverage choice of drivers with high blood alcohol content and of alcoholics demonstrate that all three beverage types are well represented (Borkenstein et al. 1974, Popham et al. 1976~. For poor alcoholics, at least, relative prices are the major determinant of beverage choice (deLint 1962, Makela 1971~. These results cast doubt on such grounds for tax discrimination among beverage types as the assertion that liquor is intrinsically more dangerous than beer. A more appropriate tax strategy might be to tax alcoholic beverages not on the basis of beverage type, but according to ethanol content or whatever other characteristics might be shown to create prob- lems.3 A different type of concern generally taken into account in evaluating tax strategies is the relative distribution (incidence) of costs and benefits across the citizenry. While we cannot here present a comprehensive analysis of this complex issue, the following observations seem basic and germane. The bulk of alcohol taxes are paid by a small fraction of the population. Half or somewhat more than half of the alcohol is bought by one-tenth of the U.S. consumer population. Members of this fraction suffer pro- portionately greater health damage, trauma, and personal financial loss associated with the adverse effects of alcohol consumption, and hence would also be major beneficiaries of reduced consumption that resulted from an increase in taxes. Thus, the incidence of direct benefits and costs would both be correlated and concentrated. These statistical as- sociations would be modified to the extent that Medicaid, Medicare, and health insurance collections from all consumers might be reduced as a result of health cost savings. Measuring the incidence of cost and 3 Each of the following three tax strategies could be viewed as neutral with respect to beverage type: 1. A tax simply proportional to ethanol content. 2. A tax structure designed to equalize the average wholesale or retail price of an ounce of ethanol from each beverage. 3. A tax structure designed to equalize the price of an ounce of ethanol from the cheapest brand of each beverage type. Even more complex alternatives (from the point of view of computation) emerge if the tax structure is designed to reflect that demands for beer, wine, and liquor may differ with respect to their sensitivity to price.
Regulating the Supply of Alcoholic Beverages 73 benefit resulting from taxes on alcohol is a difficult research problem for which current data are thin and methodological issues require much more attention. However, the current imprecision does not seem to us a major stumbling block for policy evaluation. Poor households that are burdened by the heavy drinking habit of a member may be further deprived if tax-led price increases induce an increase in alcohol expenditures. By some estimates (see Cook, in this volume, for a review), the price elasticity of demand for alcoholic bev- erages is high enough that expenditures would not increase much, if at all, in the average family. To the extent that the heavy-drinking member reduces his (in most cases) or her alcohol consumption in response to higher prices, earnings may increase and health care expenditures may be reduced, thus tending to improve the net economic position of the household. However, alcohol taxes are probably regressive, in the sense that they constitute a higher fraction of poor households' budgets than other households. The net incidence of alcohol taxes depends on how the revenues are used. If they were used, for example, to create an income deduction in the social security tax or to underwrite additions to mini- mum income maintenance programs, the net incidence would be more progressive. It is probably already the case that the external costs of medical care incident to alcohol use, supported by public expenditure, are balanced by current tax receipts. Finally, a comprehensive pricing policy for alcoholic beverages would take notice of mechanisms other than the excise tax by which govern- ment regulation influences the cost of alcoholic beverages to consumers. For example, the income tax code subsidizes consumption by allowing tax deductions for alcoholic beverages purchased in connection with business-related meals (Mosher 19804. A second example are the armed services, which sell alcoholic beverages at greatly discounted prices at post exchange stores and at clubs on base. This policy encourages drink- ing, not only by uniformed and civilian employees and their families, but also by reservists and others eligible to shop at post exchange stores, about ~ million people in all (Mosher and Mottl, in this volume). The armed forces have long been concerned with the high incidence of al- cohol problems among career personnel, the rate of which is clearly exacerbated by these pricing policies. CONTROLLING OFF-PREMISE RETAIL SALES Since World War II, there has been a secular increase in the fraction of alcohol purchased for consumption at home (off-premise), rather
74 REPORT OF THE PANEL than at taverns and restaurants (on-premise). State, municipal, and fed- eral jurisdictions can regulate retail sales for home consumption in two major respects. First, they control the number and location of retail outlets, either directly, as in the monopoly states, or through licensing or other regulation. Second, the operation of retail outlets is regulated by specifying legal hours of sale, minimum and/or maximum purchases per customer in any one visit, prohibition of sale to inebriated customers, and so forth. Retail outlets also differ among jurisdictions with respect to allowable merchandising practices. With respect to the number and location of retail outlets, it is doubtful that variation in the density of outlets, within the ranges observed in North America, has much effect on beverage sales. A number of studies (Harford et al. 1979, Smart 1977b) have attempted to measure the effect of outlet density on drinking by correlation or regression analysis, com- paring, for example, consumption and outlet density across states. As Ornstein and Hanssens (no date) point out, it is difficult to interpret the results of these studies due to the confusion of cause and effect. We would expect outlet density to be primarily a result of the demand for alcoholic beverages. A recent quasi-experimental study of retail sales in a rural area of Ontario compared sales to residents of two cities located some miles apart, both of which were serviced by a package store located in one of them. Per-capita sales were about equal for these cities, despite the differences in accessibility (Popham et al. 1976~. More extreme distances may make a difference; rural alcohol consumption in Finland increased appreciably when state monopoly stores were first opened in rural areas in 1968 (Beauchamp, in this volume). The same phenomenon would be expected to occur when outlets first open in formerly "dry" counties in the United States. For most consumers living in "wet" counties, the opening of a new package store in an especially frequented shopping zone does not appear to change their consumption levels measurably, although it may cause them to keep a smaller home inventory while making more frequent purchases. Still, there is not a great deal of detailed knowledge about variations in availability. The effects of per- mitting sales of alcoholic beverages in food stores and other high-density outlets as well as more specialized locations deserve more study. Very little is known about the effect of merchandising practices on consumption. A recent study of Ontario found that customers made larger purchases on the average in a self-service outlet than in a nearby outlet in which orders were filled by clerks; the customers in t~self- service store were also more likely to report to interviewers that they made unplanned purchases. However, it is possible that the larger in-
Regulating the Supply of Alcoholic Beverages 75 ventories thus acquired by the self-service customers did not influence their actual consumption. Other merchandising practices have not been studied with respect to alcoholic beverage sales. CONTROLLING ON-PREMISE SALES Public drinking places have two distinct images. On one hand are the neighborhood bar or tavern, the sophisticated cocktail lounge, the Brit- ish pub, the German beer hall places where neighbors, friends, and workmates gather for relaxation and conviviality. On the other hand is the residual temperance-era image of the saloon, where criminal activ- ities are hatched, where working men drink up their weekly paychecks and then some, often fighting with other customers and creating a public nuisance for the neighborhood- then stagger home (or worse, in current times, attempt to drive home) and abuse their families. Both of these images have clearly had their real-life exemplars. Regulation of public drinking places is in part an effort to control the excesses of the saloon, while preserving the legitimate pleasures of the neighborhood tavern. Typically, such regulation includes restrictions on operating hours, a ban on extending credit to customers, zoning restrictions to prevent taverns from operating near schools or churches or in residential areas, a requirement that food be served with the drinks, and so forth. The laws uniformly prohibit serving minors or drunks and enjoin owners/ managers to maintain orderly premiseswith the threat of civil penalties and license revocation for noncompliance. Indeed, tavern owners can be liable for injury and property damage caused by inebriated customers, an issue that arises most commonly in connection with auto accidents. Mosher (1979) summarizes the current status of the law with respect to civil liability (p. 7824: . . . 27 states and the District of Columbia impose some type of dram shop liability on servers of alcoholic beverages.... Most states with dram shop acts and all states recognizing dram shop liability under common law have as a matter of social policy established that commercial servers of alcoholic beverages have a broad obligation to protect the public from injuries caused by their intoxicated or underaged patrons. The legal standard for assigning negligence (in cases that do not involve underage patrons) is a demonstration that the patron was "obviously intoxicated. ' Mosher points out that this standard is inordinately vague and conducive to capricious judgments and recommends that the factual inquiry in such cases be expanded to include an assessment of the "de- gree of care" exercised by the server particularly with respect to the
76 REPORT OF THE PANEL training given to employees and the host's concern for patrons' trans- portation arrangements. Such a reorientation may serve to enhance the preventive effectiveness of dramshop laws. At this time, in the absence of research data, the effectiveness of such laws or variations therein is entirely a matter of anecdote, a priori speculation, and common sense argument. Accidents and violent crime that may result from acute episodes of intoxication in public drinking places are a central concern of on-premise control of alcohol. There is also the question of whether widespread availability of public drinking places increases the total quantity of con- sumption. It seems reasonable to suppose that increased availability of alcoholic beverages in restaurants, cafeterias in workplaces, sports are- nas, theaters, and so forth would have an effect on per-capita con- sumption; generally speaking, if the practice of drinking is integrated into a wider range of day-to-day customary activities, the quantity of consumption will increase. The question of how many and what types of public places should be permitted to accommodate drinking then becomes in part an issue of public health, albeit one that can neither be readily quantified nor simply resolved. The current trend toward increases in the number and variety of drinking premises deserves at- tention and thoughtful analysis, for the cumulative effect on drinking practices may be substantial. MINIMUM AGE RESTRICTIONS While only a small fraction of the United States continues to prohibit the sale of alcoholic beverages, the prohibition of sales to one large segment of the population youths is currently mandated by every state. The age thresholds all lie between 18 and 21. As of 1979, 23 states set the minimum age at 18 or 19 years, 3 set the limit at 20, and 24 set the limit at 21 (12 of these, however, allowed beer sales to 18- or 19- year-olds). There was considerable flux in these legal thresholds during the 1970s: between 1970 and 1973, 24 states reduced their minimum drinking ages (Williams et al. 1975), while a number of states have raised the minimum in the last few years. These changes have provided the basis for quasi-experimental analyses of the consequences of varying minimum age restrictions. Williams et al. (1975) performed a short-term follow-up of minimum age reductions legislated in the early 1970s in Michigan, Wisconsin, and Ontario. Douglass (1979-1980) and his colleagues performed short-term follow-up studies of minimum age reductions in Michigan, Maine, and
Regulating the Supply of Alcoholic Beverages 77 Vermont and a longer term follow-up for Michigan. These and other studies have focused on a single dimension of concern drunken driving (Haddon 1979~. They found consistent evidence that the age reductions resulted in an increase in the rate of auto crashes and fatalities involving youthful drivers. Williams et al. estimate that during the first year of reduced age in the three jurisdictions they studied, the number of drivers 15-20 years old who were involved in fatal crashes was about S percent greater than would be expected in the absence of the change. Smart (1977a) found that in 25 states in which the drinking age was lowered, beer was the only beverage type showing a discernible increase in consumption. Douglass refined this result in his analysis of the Mich- igan experience, concluding that only draught beer consumption in- creased significantly as a result of the minimum age reduction in that state. However, Williams et al. found an increase in youthful auto fa- talities in Wisconsin following a reduction in the drinking age for spirits and wine, while beer had remained constant at 18suggesting that beer is not uniquely responsible for teenage drinking problems. While the minimum drinking age does have an effect on alcohol con- sumption by youths, underage youths still drink a great deal. National surveys in the 1970s have consistently shown that over 80 percent of high school seniors have had a first drink before age 18; over one-third of high school students, including half of all students 16-17 years old, reported drinking within the past 30 days (Abelson et al. 1977, Blane and Hewitt 1977a, Johnston et al. 1979~. The prohibition on sales to youths thus may reduce availability to them somewhat, but it falls far short of imposing total abstinence on this group. Minimum age restrictions in this country reflect widely accepted be- liefs that drinking tends to be more harmful for youths than for adults and that we cannot trust youths to make good decisions about when, where, and how much they should drink. While the legitimacy of this type of restriction is widely accepted, the question of precisely where the age line should be drawn remains alive in many areas. State and federal laws currently gives 18-year-olds most of the rights and respon- sibilities associated with adulthood, the right to purchase alcohol being the only major exception. If 18-year-olds are mature enough to vote, seek many elective offices, enter into contractual arrangements, serve in the armed services, and so forth, it would seem logically consistent to also confer the remaining symbol of adulthood, the right to drink, on this group. The response to this argument is that as a group, people aged 18-20 are extraordinarily prone to auto accidents, as well as violent crime and other forms of socially destructive activity, and it is simply
78 REPORT OF THE PANEL foolish to exacerbate these tendencies by legalizing drinking for this age group.4 We do not attempt to resolve this debate, but simply to note that there is reasonable evidence that prohibition for youths does have some effect on their drinking and in particular that the choice of a minimum drinking age has a small but consistently exacerbating effect on the auto accident and fatality rates. CONCLUSION The common belief that alcohol control measures (government action to regulate the supply of alcohol and drinking premises) are ineffective as prevention instruments is unfounded. This belief has been engendered in part by a misunderstanding of the lessons of the Prohibition experi- ence. There is good evidence from econometric studies that alcohol pnces, as affected by excise taxation, can affect consumption levels, and probably the consequent rates of alcohol- related problems. Reductions in the minimum drinking age slightly but consistently increase auto accident involvement by younger drivers. The effects of merchandising practices, outlet density, civil liability for servers, and so forth have not been established with reliability, in part because these control mecha- nisms are intrinsically very difficult to study. It is possible but as yet hypothetical that the cumulative effect of a number of changes in these areas of regulation has been substantial. 4 If the concern is centered about lowering youthful traffic accidents, one might think of raising the driving age rather than the drinking age. European countries generally have lower drinking ages and higher driving ages than the United States.