CHAPTER 11 ECONOMIC ISSUES IN ALCOHOL USE AND ABUSE Introduction O f the many effects associated with alcohol consumption and alcohol abuse, virtually all have an economic dimension. For example, chronic heavy consumption of alcohol affects the demand for medical services and influences individuals' spending patterns and behavior in the labor market. Similarly, increased risks of traffic crashes associated with drinking and driving affect insurance premiums and the demand for auto repair services. In addition to the effects of alcohol consumption on economic outcomes, economic factors influence drinking behavior and the incidence of alcoholrelated problems. For example, tax laws affect the price of alcoholic beverages relative to the price of other goods, which in turn may influence consumption levels and patterns. Finally, economics provides the analytic tools for measuring the effects of alcohol consumption using money as a common accounting unit. This review of economic research on alcohol focuses on three main areas. First, estimates of the economic costs of alcohol abuse attempt to quantify in dollar terms the burden that is imposed on society by the abuse of alcohol.1 Recent studies have brought new approaches to bear on this question. This area of research Of the many effects associated with alcohol consumption and alcohol abuse, virtually all have an economic dimension. helps to identify which aspects of alcohol abuse are most damaging. A portion of this research focuses on the costs that heavy drinkers impose on others. Results from a well-established body of economic theory (developed without specific reference to alcohol) suggest that these costs may be relevant in establishing appropriate tax rates on alcoholic beverages. Second, a growing body of research explores the relationships between alcoholism treatment and health care costs. Recent efforts have examined the potential for these costs to be offset by subsequent reductions in other health care costs and the costeffectiveness of alternative forms of alcoholism treatment. Research on the cost aspects of alcoholism treatment is important in designing health insurance coverage for such treatment and may ultimately affect the kind of treatment available in the market for health care services. Finally, recent theoretical studies have examined the economics of addictive behavior. This research As used in the literature on economic costs (and throughout this section), the term "alcohol abuse” means any costgenerating aspect of alcohol consumption. This differs from the clinical definition of the term, which involves specific diagnostic criteria. Thus, a single occasion of driving while intoxicated that leads to injury or property damage would qualify as alcohol abuse according to this definition, even though it would not meet the clinical criteria for a diagnosis of alcohol abuse. provides a new perspective on the behavioral mechanisms by which price changes may affect the consumption of alcohol by the heaviest drinkers. Although still in the early stages of empirical testing, this research may yield important insights for preventing alcoholism and alcohol problems. Domestic relationships may be disrupted because of alcohol abuse, thus leading to emotional distress and affecting purchasing patterns and labor market decisions. Economic Costs of Alcohol consumption and alcohol abuse gener- Alcohol consumption generates some benefits as well. Millions of people choose to drink alcoholic beverages and derive enjoyment from doing so. In 1990, consumers spent $70.3 billion on alcoholic beverages (U.S. Department of Commerce 1992), which reflects the value that people attach to these products. Part of these expenditures went to pay the wages and salaries of the more than 52,000 people who were employed in alcoholic beverages industries in 1990 (U.S. Department of Commerce 1991). Production of alcoholic beverages also generates employment and economic activity in other sectors of the economy, including agricultural products, packaging, shipping, wholesale and retail sales, and advertising services. In addition, taxes on alcoholic beverages contributed more than $14 billion in 1990 to the treasuries of Federal, State, and local governments (Distilled Spirits Council of the United States 1991).2 This catalog of costs and benefits associated with alcohol use is certainly incomplete, yet it illustrates the diversity of economic effects associated with alcohol. Economic methods provide a means of quantifying these various effects using money as a unit of measurement. Estimates of Economic Costs Any cost estimation must begin by identifying which effects will be quantified. Several studies (Cruze et al. 1981; Harwood et al. 1984; Rice et al. 1990) have estimated costs of alcohol abuse for the United States by using an approach employed in cost-of-illness studies for other disorders (Hodgson and Meiners 1982; Rice 1967; U.S. Department of Health and Human Services 1988). These studies attempt to measure the total costs of alcohol abuse, although some of the effects identified above are excluded (e.g., costs associated with pain and suffering generally are not quantified). Other studies have used elements of the cost-of-illness framework to examine the costs of alcohol abuse for individual States (Gorsky et al. 1988; Parker et al. 1987) and other countries (Adrian 1988; Collins and Lapsley 1991; McDonnell and Maynard 1985). Additional studies focus on particular components of the total costs of alcohol abuse, such as the effects of alcohol on productivity (Berger and Leigh 1988; Mullahy and Sindelar 1989) or the costs associated with fetal alcohol syndrome (Abel and Sokol 1987, 1991a, 1991b; Harwood and Napolitano 1985). Finally, some recent research has adopted an approach that differs from 2 Evidence suggests that there may be some beneficial effects on cardiovascular health associated with moderate consumption of alcohol, although these benefits may be offset by other (adverse) health effects associated with alcohol consumption. For more discussion of these health effects, see Chapter 8, Effects of Alcohol on Health and Body Systems. cost-of-illness studies by focusing on the total costs that heavy drinkers impose on others (Manning et al. 1989, 1991; Pogue and Sgontz 1989). A number of conceptual problems arise in estimating the economic costs of alcohol abuse. As outlined above, one problem involves identifying which costs should be counted and categorizing them in a way that is comprehensive and amenable to estimation while avoiding double counting. The cost-of-illness accounting framework provides one solution to this problem. Another problem involves assigning money values to consequences that are not intrinsically monetary in nature. A leading example is the problem of assigning dollar values to alcohol-related deaths. Further difficulties arise in attributing causation for particular cost items to alcohol abuse, since cause-and-effect relationships are generally difficult to ascertain. Finally, many problems arise in producing estimates of specific cost categories, both because of deficiencies in data and because of difficulties in estimating unobservable quantities (such as lost productivity). The main components of the costs of alcohol abuse under the cost-of-illness framework are shown in table 1. These costs include (1) the costs of treatment for alcoholism, (clinical) alcohol abuse, and the medical consequences of alcohol consumption; (2) the value of lost productivity due to the health effects of alcohol abuse and alcoholism (termed morbidity costs); (3) costs associated with premature deaths due to alcohol consumption (called mortality costs); and (4) various nonhealth costs resulting from alcohol-related crime, property damage, and costs of administering the social welfare system (e.g., social security benefits, unemployment insurance) that are attributed to alcohol problems. As shown in table 1, these costs are classified as either core costs, which result from adverse effects on the health of the drinker, or other related costs, which comprise various nonhealth costs, including both those bome by drinkers and those imposed by drinkers on others. Another way of classifying these costs, also shown in table 1, distinguishes between direct costs, which are payments that are actually made, and indirect costs, which are benefits lost or forgone. Direct costs are mostly the costs of medical treatment plus support costs, including research, training, and administration attributable to alcohol abuse, and some nonhealth expenditures. Indirect costs include costs associated with premature mortality and productivity losses attributed to alcohol. Finally, under the framework employed by Rice et al. (1990), the costs associated with fetal alcohol syndrome (FAS) are estimated separately. The FAS cost estimate comprises costs for medical treatments and residential care that otherwise would be classified as core direct costs. The total costs of alcohol abuse in the United States have been estimated at $70.3 billion for 1985 (Rice et al. 1990). By adjusting for inflation and demographic changes, the 1985 estimate was projected to $85.8 billion for 1988. These estimates contrast with earlier estimates (Harwood et al. 1984) of $89.5 billion for 1980, projected to $116.7 billion for 1983. Differences between the estimates reflect differences in estimation methods and data sources employed; they also highlight the inexact nature of such estimates. The data and methods employed in estimating the various components of the economic costs of alcohol abuse ultimately determine the magnitude of the final estimate. Each category of costs is estimated using data and methods suited to that particular component. As a result, a number of different issues arise in estimating the various components of the economic costs of alcohol abuse. Treatment costs include expenditures for treating the medical consequences of alcohol consumption as well as for treating alcoholism itself. The most recent study (Rice et al. 1990) estimates hospital costs using discharge records on the number of days of care for patients with primary alcohol-related diagnoses and on the extra days of care for patients with other medical conditions when accompanied by a secondary diagnosis indicating alcohol involvement (Rice and Kelman 1989). These methods replace techniques used in previous studies (Cruze et al. 1981; Harwood et al. 1984) that were criticized by Heien and Pittman (1989). Although the new procedures represent improvements in some respects, they have important limitations as well. For example, relying on hospital discharge records to identify health problems caused or potentially exacerbated by alcohol consumption is likely to underestimate the role of alcohol in contributing to medical problems. Recent research (Umbricht-Schneiter et al. 1991) shows that hospital discharge records "markedly underestimate" the prevalence of alcohol abuse in hospitalized patients, perhaps by as much as two-thirds. This finding suggests that Rice et al. (1990) may have underestimated treatment costs related to alcohol abuse by a considerable margin. Productivity losses due to alcohol abuse present special estimation problems because of the difficulty in measuring the value of goods and services that go unproduced because of alcoholrelated disorders. These costs have been estimated at $27 billion in 1985, or 39 percent of the total economic cost of alcohol abuse (Rice et al. 1990). On the assumption that individuals' incomes correspond to the value of what they produce, these loss-of-productivity costs are measured by the reductions in income suffered by individuals with alcohol abuse disorders. The estimate for 1985 found that individuals satisfying the clinical criteria for alcohol abuse or dependence at any time in their lives suffered income reductions ranging from 1.5 percent to 18.7 percent depending on age and sex compared with those with no such diagnosis. The mechanisms by which this effect might be realized include lower occupational achievement, either as a direct result of alcohol problems or because of the effects of lower educational attainment; reduced labor force participation; increased unemployment; reduced full-time work; absenteeism and tardiness; and reduced productivity on the job (Rice et al. 1990). The magnitude and even the existence of productivity losses has been questioned (Cook 1991; Heien and Pittman 1989). A review (Cook 1991) concluded that ". . . there is little credible evidence... that heavy drinkers are less productive than other members of the labor force (once adjustment is made for differences in education and demographic characteristics)." One study using data from the early 1970s (Berger and Leigh 1988) found that drinkers earn significantly more than abstainers. In contrast, an earlier estimate of the total economic costs of alcohol abuse (Harwood et al. 1984) found that problem drinking reduced household income by 21 percent. However, a replication of that analysis using a slightly different statistical procedure found no significant effect of alcohol problems or alcohol consumption on income (Heien and Pittman 1989). Another study (Mullahy and Sindelar 1989) reported that although alcoholism has no significant direct effect on income, onset of alcoholism symptoms by age 18 affects earnings indirectly by reducing educational achievement. These indirect effects are incorporated in the recent estimate by Rice et al. (1990). Mortality costs due to alcohol for 1985 were estimated at $24 billion, or 34 percent of the total economic cost of alcohol abuse (Rice et al. 1990). In the cost-of-illness framework, the cost of a premature death is approximated by the present value of the future earnings that are lost (Cooper and Rice 1976; Hodgson and Meiners 1982; Rice 1967; Rice and Hodgson 1982). Critics contend that this “human capital” approach understates the value of human lives, especially for women and retired people (Cohodes 1982; Landefeld and Seskin 1982). An alternative approach to valuing lost lives, referred to as the "willingness-to-pay" approach, is based on the amounts that individuals are willing to pay to avoid small increases in risks to life (Landefeld and Seskin 1982; Mishan 1971; Schelling 1968). This approach is more firmly grounded in economic theory than the human capital approach and generally produces larger values for lost lives, but it is more difficult to implement in practice. Nevertheless, valuations based on willingness to pay have been used in studies of alcohol-related costs (Manning et al. 1989, 1991; Phelps 1988). Other components of the estimated costs of alcohol abuse include costs of care for people born with FAS; property damage resulting from alcohol-related fires and automobile crashes; alcohol-attributable costs of administering the criminal justice and social welfare systems; and the value of productivity lost by victims of alcohol-related crimes and those imprisoned for alcohol-related crimes (Rice et al. 1990). These components are estimated by various techniques and often without great precision. One of these components, the costs of FAS, has attracted significant research attention that has yielded widely varying results. This variation reflects differences in assumed FAS prevalence rates and differences in the cost elements included in each study. (Further discussion of the prevalence of FAS may be found in Chapter 9, Effects of Alcohol on Fetal and Postnatal Development.) Harwood and Napolitano (1985) based estimates of FAS costs on three alternative rates of FAS prevalence per 1,000 live births, ranging from a low of 1.0 to a midrange estimate of 1.67, as well as an upper limit prevalence estimate of 5.0 (acknowledging that the high rate was not supported by the literature on FAS). A study by Abel and Sokol (1987) employed an estimated prevalence rate of 1.9 cases per 1,000 births; this study also served as the basis for the FAS cost estimate reported by Rice et al. (1990). A recent analysis by Abel and Sokol (1991b) applies a much lower prevalence estimate of 0.33 per 1,000 births to arrive at a conservative estimate of FAS costs. Different studies include different cost elements, which lead to further variation in the estimated costs. The studies by Abel and Sokol (1987, 1991a, 1991b) consider only medical treatment and care for FAS patients through age 21. Rice et al. (1990) add the costs of residential care for FAS patients over 21; these costs account for 80 percent of their FAS cost estimate. Harwood and Napolitano (1985) include costs for lost productivity as well as for treatment and care for FAS patients of all ages. Cost estimates range from $75 million (Abel and Sokol 1991b) to $3.2 billion (Harwood and Napolitano 1985, midrange estimate), not counting the upper bound estimate of Harwood and Napolitano of $9.7 billion. The estimate by Rice et al. (1990) found FAS costs to be $1.6 billion, or 2.3 percent of the total costs of alcohol abuse. The great variation in cost estimates for this relatively modest component of the overall costs of alcohol abuse illustrates the uncertainty involved in estimating costs and the significance of critical assumptions for the final estimates. Limitations and Caveats Apart from specific problems of measurement and estimation, estimates of the total economic costs of alcohol abuse based on the cost-ofillness approach have several limitations. One set of concerns involves the reliability and comprehensiveness of the estimates, caused in part by the difficulty in discerning cause-and-effect relationships. Thus, identifying a particular death, illness, crime, or income loss as resulting from alcohol abuse carries inherent uncertainty. Further, certain consequences are generally omitted from cost-of-illness studies. Among these are the costs of pain and suffering among alcohol abusers and their families, friends, and coworkers; effects of alcohol abuse on domestic stability; and secondary effects of alcohol abuse on the markets for other goods and services, such as insurance, automobiles, and nonalcoholic beverages. In general, cost estimates exclude these elements because they cannot be quantified easily or reliably. Another concern is that the cost-of-illness approach combines costs that drinkers accept voluntarily with costs that drinkers impose on others. Critics contend that only the latter qualify as social costs that are relevant to public policy, while costs borne voluntarily amount to legitimate expressions of individual choice (Heien and Pittman 1989; Wagstaff 1987). In particular, the standard economic theory of externalities shows that excise taxes that are based on the costs imposed on others may control these costs and so generate a gain to society (Boadway and Wildasin 1984; Dasgupta and Heal 1979; Pigou 1918; Varian 1978). There is no parallel theoretical justification for taxing commodities on the basis of their total costs including those costs that are accepted willingly by consumers. Apart from specific problems of measurement and estimation, estimates of the total economic costs of alcohol abuse based on the cost-of-illness approach have several limitations. Finally, cost-of-illness estimates make an implicit comparison between the current level of alcohol abuse and a hypothetical world of no alcohol abuse. It is unreasonable to interpret the estimates as a measure of the potential gain to society from eliminating alcohol abuse, since it would be costly to accomplish that objective, and some benefits associated with alcohol consumption (e.g., enjoyment, tax revenues) would be lost in the process as well (Heien and Pittman 1989; Wagstaff 1987). Thus, the components of a particular estimate of the economic costs of alcohol abuse both reflect and limit the purposes for which the results |