Chapter 4 Economics of Tobacco Consumption in the Americas Preface 103 Economic Costs of the Health Effects of Smoking 105 Latency of the Health Consequences 105 Estimating the Economic Costs 105 General Considerations and Limitations 105 Prevalence- and Incidence-Based Studies 110 Application to Developing Countries 112 Financing of Health Care and Pension/Disability Funds 112 Costs of Smoking-Control Policies and Programs 114 Economics of the Tobacco Industry 214 The Tobacco Sector 114 Overview 114 Demand for Tobacco 115 Advertising 128 Supply of Tobacco 119 Manufacturing 122 Distribution 121 Trade 122 Subsidies to Tobacco Production 122 Contribution of Tobacco to Economic Growth and Development 125 Externalities 125 Price, Production, and Substitution 125 Future of Tobacco Production 127 Tobacco Taxation 127 Subnational Taxes 128 Effects of Excise Taxes on Smoking 129 Modeling Addiction 131 Analysis of Recent Tax Increases 132 Health Consequences of Tax Changes 133 Equity, Incidence, and Distribution of the Tax Burden 134 Use of Tobacco Taxes 135 Conclusions 136 References 137 Preface Although the economic aspects of smoking in North America have been extensively examined, detailed data are not availablefor Latin America and the Caribbean. For the latter region, a definitive analysis of the health costs of smoking and the economic configuration of the tobacco industry await more systematic reporting and collection of data. In the first part of this chapter, a generic approach to assessing the costs associated with the major adverse health effects of smoking is outlined. The background for this approach, which uses concepts introduced in Chapter 3, is described. Data and examples from the United States and Canada are provided, and the work done in these countries is summarized. In the second part, an overview of the tobacco sector of the economy is offered. Again, more data are available from North America than from Latin America and the Caribbean, but the economic issues (supply and demand, advertising, subsidies, taxation, and others) are relevant to all countries of the Americas. This overview provides a framework for weighing the relative costs and benefits of tobacco production and consumption. Economics 703 Economic Costs of the Health Effects of Smoking Latency of the Health Consequences Since 1964, when a report on the health conse- quences of smoking was released by the Surgeon General's Advisory Committee on Smoking and Health (Public Health Service 19641, extensive re- search has assessed the disability, morbidity, and pre- mature mortality attributable to tobacco use. The many effects of smoking on health were documented in the Surgeon General's twenty-fifth anniversary re- port on smoking and health (U.S. Department of Health and Human Services IUSDHHSI 1989). A detailed examination of smoking-attributable mortal- ity (SAM) in the United States summarizes these asso- ciations (Table 1). (See Chapter 3 for an assessment of SAM in Latin American and Caribbean countries.) As an epidemiologic transition occurs in Latin America and the Caribbean, noncommunicable dis- eases are expected to become increasingly prominent as causes of death. For example, although Brazil bears a burden from certain infectious diseases (such as Chagas' disease) and the growing incidence of human immunodeficiency virus infection, many other infec- tious and parasitic diseases have been brought under control. Many cases of lung cancer are now antici- pated in Brazil (The World Bank 1989a). Cardiovas- cular disease is the leading cause of death in Brazil (The World Bank 1989a), and the number of deaths due to cardiovascular disease is likely to increase sig- nificantly. Among Latin American women, for whom prevalence of smoking appears to have increased (see Chapter 3), an increased incidence of lung cancer may soon become apparent (Crofton 1990). Numerous studies have reported a 20- to 30-year latent period between the initiation of smoking on a regular basis and the development of lung cancer (USDHHS 1982), a phenomenon well documented in North America. In the United States, many men started to smoke as adolescents or young adults around World War I, and many women started as adolescents or young adults during or after World War II. The incidence of lung cancer in the United States began to increase for men around 1940 and for women around 1960 (USDHHS 1989). A similar lag occurred in Canada; from 1976 to 1986, the rate of lung cancer doubled (Millar 1988). An epidemiologic and economic result of latency is the continued rise in lung cancer deaths despite a decline in the prevalence of smoking. In the United States, the lung cancer mortal- ity rate for men did not begin to level off until 1985 (USDHHS 1989). For women, deaths from lung can- cer have not yet peaked, and lung cancer has become the most common cause of cancer mortality, surpass- ing breast cancer (USDHHS 1989). The correlation between the level of cigarette consumption in a population cohort when it enters adulthood and the lung cancer rate for that cohort when it enters middle age provides further evidence of the 20- to 30-year latency (Figure 1). In Brazil, lung cancer mortality among adult males has increased as a lagged response to the increase in tobacco consump- tion (Figure 2) that began during World War II. Thus, the consequences of tobacco consumption-including economic consequences-are long in developing, and the full impact of disease, disability, and death is measured over decades. Estimating the Economic Costs Many estimates have been made of the costs of smoking in the United States and Canada. A similar body of work is not available for Latin America and the Caribbean-in part because the data required for such analyses are often not available. In addition, a single estimate would probably not serve adequately because of the heterogeneity among countries of the region. An approach to estimating the health costs of smoking is described below, along with some esti- mates that have been made. General Considerations and Limitations Estimates of the economic effects of the health consequences of smoking generally consist of three components (U.S. Office of Technology Assessment [USOTAI 1985): o An attempt to identify an increased incidence of smoking-related illness in current or former smok- ers and attribution of that increase to smoking. o An application of these attribution ratios to esti- mates of the direct (health care) costs of caring for persons with smoking-related illness-to obtain an estimate of the direct costs of smoking. o An estimate of the indirect costs of smoking- related illness, which is made by measuring the increased rate of morbidity and mortality in current and former smokers and then valuing (1) time lost due to morbidity by their current wage rate and (2) excess mortality by discounted future earnings. Economics 105 Table 1. Relative risks* (RR) for death attributed to smoking and smoking-attributable mortality (SAM) for current and former smokers, by disease category and sex, United States. 1988 Men RR ~-_____- Disease category (ICD-9-CM)+ Current Former smokers smokers SAM Adult diseases (235 years of age) Neoplasms Lip, oral cavity, pharynx (140-149) 27.5 Esophagus (150) 7.6 Pancreas (157) 2.1 Larynx (161) 10.5 Trachea, lung, bronchus (162) 22.4 Cervix uteri (180) NA Urinary bladder (188) 2.9 Kidney, other urinary (189) 3.0 Cardiovascular diseases Hypertension (401404) 1.9 Ischemic heart disease (410-414) Persons aged 35-64 years 2.8 Persons aged 265 years 1.6 Other heart diseases (390-398, 415-417,42&429) 1.9 Cerebrovascular disease (43&438) Persons aged 35-64 years 3.7 Persons aged 165 years 1.9 Atherosclerosis (440) 4.1 Aortic aneurysm (441) 4.1 Other arterial disease (442448) 4.1 Respiratory diseases Pneumonia, influenza (480-487) 2.0 Bronchitis, emphysema (491-492) 9.7 Chronic airways obstruction (496) 9.7 Other respiratory diseases (Olcl-012,493) 2.0 Pediatric diseases (~1 year of age) Short gestation, low birthweight (765) Respiratory distress syndrome (769) Other respiratory conditions of newborn (770) Sudden infant death syndrome (798) Burn death& Passive smoking deaths5 Total Source: Centers for Disease Control (1991). *Relative to never smokers. 8.8 4,942 5.6 2.9 1,460 6,402 5.8 5,478 10.3 3.2 1,609 7,087 1.1 2,775 2.3 1.8 3,345 6,120 5.2 2,401 17.8 11.9 589 2,990 9.4 78,932 11.9 4.7 33,053 111,985 NA 0 2.1 1.9 1,246 1,246 1.9 2,951 2.6 1.9 963 3,914 2.0 2,729 1.4 1.2 363 3,092 1.3 3,441 1.7 1.2 2,254 5,695 1.8 29,263 3.0 1.4 9,105 38,368 1.3 41,821 1.6 1.3 27,990 69.811 1.3 27,503 1.7 1.2 14,638 42,141 1.4 5,121 4.8 1.4 4,504 9,625 1.3 11,554 1.5 1.0 5,134 16,688 2.3 4,644 3.0 1.3 3,612 8,256 2.3 5,798 3.0 1.3 1,435 7,233 2.3 1,874 3.0 1.3 1,111 2,985 1.6 11,580 8.8 9,670 8.8 29,838 1.6 828 2.2 10.5 10.5 2.2 1.4 8,098 19,678 7.0 5,269 14,939 7.0 16,884 46,722 1.4 690 1,518 1.8 344 1.8 261 605 1.8 351 1.8 233 584 1.8 1.5 384 422 850 1,330 286,824 1.8 1.5 277 661 280 702 453 1,303 2,495 3,825 147,351 434,175 Women _____~-~~ RR Current Former smokers smokers SAM Total SAM Tlnternational Classijication of Diseases, Nirlfh Revision, Clirzicnl Modificofiolr `Data from the Federal Emergency Management Agency, 1990. $Deaths among nonsmokers from'lung cancer attributable to passive smoking; hational Research Council (1986). 106 Economics Several estimates have been made for the United with regard to the medical conditions attributed to States (Rice et al. 1986; Hodgson 19881, Canada (Col- smoking. Some studies include lung cancer only, lishaw and Myers 1984; Forbes and Thompson 1983a), while others include heart disease and chronic ob- the United Kingdom (Atkinson 1974), Sweden (Hjalte structive pulmonary disease (COPD). Other studies 19841, and Switzerland (Leu and Schaub 1984). Vari- compare differences in the overall use of health care ous factors should be included in a complete picture by smokers and nonsmokers. However, these esti- of the economic impact of smoking-related illness mates do not include nonmedical components of di- (Table 2), but few published studies have addressed rect costs, such as the costs of transportation to health all of these factors, and most studies have concen- care providers or of modifying an environment to trated on factors for which data are available. accommodate a person with a severe chronic illness. Most estimates of the costs of smoking-related illness calculate the direct costs of treating persons with smoking-related diseases, including the costs of hospital and nursing-home care, physicians` fees, and medications (Table 3). The specific items included in the estimates vary among studies, which also differ Estimates of the indirect costs of smoking-related illness attempt to measure the productivity lost or output forgone as a result of smoking-related illness or death (Table 4). This so-called human capital ap- proach has been criticized for placing a high value on losses sustained by young adults, men, and more- Figure 1. Correlation between cigarette consumption per person who entered adult life in 1950 and lung cancer rate for that generation as it entered middle age in mid-1970 100 - + United States so- Source: Doll and Pete (1981). Number of manufactured cigarettes consumed f Rate based on over 100 deaths. O Rate based on 25-100 deaths. 0 U.S. nonsmokers 1959-1972. Economics 107 Figure 2. Per capita rate of cigarette consumption in Brazil and lung cancer deaths for men in Rio Grande do Sul, Brazil t-- / 7 / \ / \ / \ / 14 /' / \ / / \ / \ / b-- d' / / / / / /* / / S-to 30-year lag d 1 , I I I I 1940 1950 1960 Source: The World Bank (1989a). educated persons (Markandya and Pearce 1989). In addition, earnings lost because of illness and mortality may have little relationship to the value people place on their life or health (Markandya and Pearce 1989). A more appropriate measure of that value may be the amount they are willing to pay to reduce the probabil- ity of death or disease. Although several attempts have been made to estimate willingness-to-pay for non-smoking-related illness (Viscusi 19901, this ap- proach has not been applied to cost-of-smoking stud- ies. In addition, no value has been assigned to intangible items, such as pain and suffering, prema- ture death, and loss experienced by relatives; accord- ingly, these intangibles have not been included in any published estimates of the costs of smoking. Some estimates include costs associated with the harmful effects on the fetus and on newborns of maternal smoking during pregnancy and of postnatal exposure to environmental tobacco smoke (Forbes and Thomp- son 1983b); however, most published estimates do not incorporate measures of external costs (those borne by persons other than smokers). The transfer payments (pension benefits and sick benefits) associated with smoking-related illness have / I, 30 20 10 0 1970 1980 been a source of confusion and controversy. Transfer payments reflect who pays for and who benefits from smoking-related illness; these transfers are not, strictly speaking, economic costs because they do not reflect resources consumed or lost due to smoking. How- ever, discussions of smoking-control policies have fre- quently asked whether smokers in economically advanced societies (with well-developed public or pri- vate health care financing, disability, and pension sys- tems) cover the costs of their own illness (Manning et al. 1989; Schelling 1986; Garner 1977). Accurate estimation of the cost of smoking is influenced by the quality of data available, current demographic circumstances, and competing mortality risks. Cost estimates require reliable data on smoking behavior, the incidence of smoking-related illnesses, and the prevalence of such illnesses at death. In many developing countries, vita1 statistics are unreliable or incomplete (see Chapter 3, "Smoking-Attributable Mortality in Latin America and the Caribbean"), al- though several Latin American and Caribbean coun- tries have well-established national statistical registries (World Health Organization [WHO] 1989) from which reliable estimates can be constructed. 108 Economics Table 2. Components of the costs of the health effects of smoking - Component Definition Direct costs Medical care Costs of treatment for smoking-related illness. Other Nonmedical costs of smoking-related illness. Indirect costs Morbidity costs Mortality costs Loss of earnings and/or housekeeping services due to smoking-related illness. Loss of earnings and/or housekeeping services due to premature death from smoking-related illness. Intangible costs Pain and suffering Premature death Relatives' loss Cost to individual of pain and suffering from smoking-related illness. Cost to individual of premature death due to smoking. Cost to smoker's relatives and friends because of concern for smoker's health, observation of sickness and suffering, and grief and suffering due to smoker's premature death. Transfer payments Taxes Pension benefits Sick benefits Reduced taxes paid by smokers due to illness-related reductions in earnings. Value of transfer payments such as pensions paid or forgone due to premature death. Health care costs paid by public or private insurance plans. Sick pay and disability benefits paid to smokers during illness. External costs Effects of smoking on nonsmokers, including deleterious health effects and the annoyance of exposure to environmental tobacco smoke. Includes the deleterious effects of maternal smoking on the fetus, on infants, and on children. A country's demographic configuration influ- ences the degree to which smoking-related illness be- comes manifest. Since many smoking-related illnesses do not have an important impact on persons under age 50, such illnesses do not significantly con- tribute to mortality in countries where life expectancy after infancy is low; however, low life expectancy affects only a small proportion of the population in Latin America and the Caribbean (Chapter 3, "Life Expectancy and Mortality"). The manifestation of smoking-related illness is also a function of competing morbidity and mortality. Latin American and Caribbean countries are at differ- ent stages of epidemiologic transition, and the chronic conditions associated with smoking may be obscured by the continued presence of infectious diseases and other disorders. Countries also vary in the extent to which background conditions (nutritional, genetic, or environmental) interact with smoking. Another limitation of cost-of-smoking studies is the method used to calculate attributable risk (AR).' Although quite useful, this calculation must be ap- plied judiciously; it attributes all differences between ever smokers and never smokers to smoking, and it may overestimate the level of smoking-related illness. Smokers and never smokers differ in several charac- teristics, including diet and level of alcohol consump- tion, exercise, and education KJSDHHS 1990), all of which may be associated with differences in health outcomes. Leu and Schaub (1983) developed the hy- pothetical construct of the "nonsmoking smoker- type," a person who is like a smoker in all ways except smoking, to serve as the standard of comparison in estimating costs of smoking. This construct was also used by Manning and associates (1989) to calculate the lifetime external costs of smoking in the United States. However, the concept may not be useful in many developing countries because of the variability of competing factors in different settings. In attempting to estimate tobacco-related dis- eases in developing countries, some researchers have used a single measure of AR for each of the major smoking-related illnesses, such as lung cancer, heart A detailed discussion of the theory, limitations, and other methodologic issues concerning the calculation of AR and smoking-attributable disease and mortality is presented in the Surgeon General's 1989 report KJSDHHS 1989). Economics 109 Table 3. Medical care costs for smokers, by study type and author Year of Study type and author Country estimate Annual costs (prevalence-based estimates) Collishaw and Myers (1984)+ Canada 1979 Lute and Schweitzer (1978) United States 1976 Rice et al. (1986) United States 1984 Stoddart et al. (19861t5 Canada (Ontario) 1978 Thompson and Forbes (1983)+ Canada 1980 U.S. Office of Technology Assessment (1985) United States 1985 Lifetime costs (incidence-based estimates) Manning et al. (1989)" United States 1983 Oster, Cold&, Kelly (1984) United States 1980 Total cost (billions)* 1.64 52.02 24.85 0.34 3.04 12-35 cost per smoker* 164 868 444$ 127 302 214-870 6,113 2,474-6,576: 1,147-4,138 Hodgson (1990) United States 1985 501 .o 6,239++ Hjalte (1984)+ Sweden 1980 0.18 73 *Converted to 1985 U.S. dollars by using U.S. Bureau of the Census (1988) Table 738 consumer price index. +Markandya and Pearce (1989) report these estimates converted to 1980 U.S. dollars. STotal cost divided by 56 million smokers in the United States in 1985; U.S. Department of Health and Human Services (1989). %ublic expenditure only. 110.33 cost per pack x 16,300 packs = $5,379 (1983 U.S. dollars). _Men aged 4044 light (l-14 cigarettes per day) to heavy (235 cigarettes per day) smokers. "Women aged 4&44 light (l-14 cigarettes pe; day) to heavy (235 cigarettes per day) smokers. `+Lifetime cost for all smokers >25 years old. disease, and COPD (90,26, and 75 percent, respectively) (Pan American Health Organization [PAHOI 1989). Such use of AR can be misleading because the propor- tion of current and former smokers varies across countries and over time, and the relative risk is a function of smoking patterns (e.g., the number of cigarettes smoked daily and the duration of smoking), which also vary (USDHHS 1989). For example, Joly and colleagues (1983) reported that of all lung cancers for Cuba in 1984, 63 percent among women and 91 percent among men were caused by smoking; for U.S. women and men in the mid-1980s, the attribution proportions were 75 and 80 percent, respectively (Centers for Disease Control [CDC] 1987). Moreover, the relative risk for smoking is also determined by nontobacco causes of illness, and these differ among countries. Applying an exogenously determined set of AR proportions to any country's population may lead to unreliable estimates of the level and costs of smoking-related illness. However, for countries that lack endogenous data, this procedure is often the only alternative (see Chapter 3, "Smoking-Attributable Mortality in Latin America and the Caribbean"). Prevalence- and Incidence-Based Studies The prevalence-based approach to measuring the economic costs of tobacco-related disease has fre- quently been used, largely because of its relatively simple methodology, the availability of the data needed for the calculations, and the consistency of carefully made estimates (Rice et al. 1986) (Table 3). Several of these prevalence-based studies (Lute and Schweitzer 1978; USOTA 1985; Rice et al. 1986; Collishaw and Myers 1984) indicate that the costs of smoking in any one year are likely to be great and that the economic costs of smoking should be taken 110 Economics Table 4. Value of productivity lost due to mortality and morbidity, by study type and author _____ ___~~ .__ Mortality Morbidity Year of Total cost Cost er Total cost Cost er Study type and author Country estimate (billions) l! smo er (billions) K smo er Annual costs (prevalence-based estimates) Collishaw and Myers (1984) Canada U.S. Office of Technology Assessment (1985) United States Rice et al. (1986) United States Lifetime costs (incidence-based estimates) 1979 4.04 405 0.75 74 1985 27-61 484-1,080*+ 1984 9.63 172t 21.74 388t Leu and Schaub_ (1984) Switzerland Oster, Colditz, Kelly (1984) United States 1976 0.28-0.35 149-183 0.14-0.25 76-132 1980 24,221-68,316+" 5,894-21,765+' *Total cost divided by 56 million smokers in the United States in 1985; U.S. Department of Health and Human Services (1989). `Range includes both mortality and morbiditv losses. *Converted to 1985 U.S. dollars by using U.S.`Bureau of the Census (1988) Table 738 consumer price index. _Markandya and Pearce (1989) report these estimates converted to 1980 U.S. dollars. "Men aged 40-44 light (l-14 cigarettes per day) to heavy (235 cigarettes per day) smokers. _Women aged 40-44 light (l-14 cigarettes per day) to heavy (235 cigarettes per day) smokers. seriously. These studies estimate expenditures for medical care for tobacco-related diseases, workdays lost, and future productivity lost due to smoking- related deaths during the year. However, these stud- ies do not address other issues that most concern policymakers, including the economic impact of de- creased prevalence of cigarette smoking, the length of time before economic effects are realized, the eco- nomic benefits of not smoking, and a comparison of the lifetime illness costs of smokers with those of nonsmokers (Hodgson 1990). Health care expendi- tures tend to increase just before death, but smoking shortens life expectancy and changes the pattern of health care expenditures. The question arises whether the health care costs incurred by smokers, when ad- justed for the altered temporal pattern, exceed costs incurred by never smokers. Most cost-of-illness studies are based on esti- mates of the prevalence of illness in a particular year. Because many smoking-related illnesses are chronic and the latent period between initiation of smoking and onset of illness is long, prevalence-based cost estimates reflect the consequences of historical trends in smoking, which may differ among countries at different times. Accordingly, prevalence-based cost estimates cannot be used to predict the impact of smoking-control policies or to predict the impact of increases in smoking, except after long periods. For policymakers, incidence-based, or lifetime, estimates of the costs of smoking-related illness may be more useful than prevalence-based estimates (Leu and Schaub 1983; Manning et al. 1989; Oster, Colditz, Kelly 1984). In the incidence-based model, the eco- nomic costs of smoking are estimated as the average additional costs per smoker, due to smoking-related illnesses, incurred over the smoker's lifetime. Esti- mates can be made of direct (medical care expendi- tures) and indirect (e.g., lost wages, salaries, and housekeeping services) costs of smoking and of the benefits of quitting. For lung cancer, coronary heart disease, and emphysema, the discounted value of an- ticipated lifetime costs has been estimated for smoking- related diseases in persons who smoked in 1980 and continued to smoke (Oster, Colditz, Kelly 1984). The costs of the benefits of quitting can be estimated as the difference between the cost-of-smoking estimate and the expected costs of former smokers, which reflect the gradual rate of decline in risk for smoking-related diseases. Estimates of each smoker's lifetime cost of smok- ing differ by the person's age, sex, and quantity smoked (Oster, Colditz, Kelly 1984). For example, the lifetime costs of smoking for a 45-year-old man who is a heavy smoker are significantly greater than those of a 65-year-old woman who is a light smoker ($46,334 vs. $2,462; in 1980 U.S. dollars). Oster and colleagues suggest that estimates of the costs of the benefits of quitting are less than the costs of smoking and that benefits vary according to the characteristics of indi- vidual smokers. The expected costs of both smoking and the benefits of quitting were sizable for all groups of smokers foster, Colditz, Kelly 1984). Recently, Hodgson (1990) analyzed data on use and costs of medical care and on mortality for specific age groups in cross sections of the U.S. population to generate profiles of lifetime health care costs begin- ning at age 17. Because expenditures are higher for persons who die than for those who survive, the anal- ysis distinguished between the two groups within a given age range. The profiles, estimated for men and women by age and amount smoked, include the costs of inpatient hospital care, physician services, and nursing-home care. However, the cost of drugs and dental care, as well as morbidity and mortality costs, are excluded. Hodgson concluded that, despite the higher death rate for smokers, the cumulative impact of the excess medical care used by smokers while alive outweighs their shorter life span and that smokers incur higher medical care costs during their lifetime. For all smokers, excess medical care costs increase with the amount smoked. Hodgson (1990) estimated that the U.S. population of civilian, noninstitutional- ized persons aged 25 years or older who ever smoked cigarettes will incur lifetime excess medical care costs of $501 billion (1990 U.S. dollars discounted at 3 per- cent) or $6,239 per current or previous smoker (Table 3). This excess is a weighted average of the costs incurred by all smokers, whether or not they develop smoking-related illness. For smokers who do develop such illnesses, the personal financial impact is much higher. Lifetime or incidence-based cost-of-illness esti- mates are preferred over prevalence-based estimates for measuring the costs of changes in, and trends affecting, the incidence of disease. However, lifetime cost estimates require knowledge of the natural his- tory of disease, the pattern of medical care use, and the occurrence of co-morbidity. Lifetime costs are often estimated from current profiles for cross sections of populations at different ages and at different stages of disease. To measure the potential impact of changes in public policies and demographics on future health care costs, projections of cost estimates must be made. Changes in parameters, such as technologic change and its rate of diffusion, must be considered, or esti- mates may be biased and misleading (Hodgson 1988). The incidence-based approach is better suited than the prevalence-based approach for estimating the costs of smoking because the former relates current changes in smoking behavior to future changes in the costs of smoking-related illness. The incidence-based approach, however, suffers from the limitations of transferability between countries (mentioned above); it does not directly address intangible costs and externalities; and it values mortality and morbidity by measuring forgone earnings rather than willingness- to-pay. Moreover, even for economically advanced countries, including the United States, the incidence- based approach is limited by the lack of adequate and comprehensive data; for less-developed countries, this limitation may be exacerbated. Application to Developing Countries The cost-of-illness studies conducted in the United States and other developed countries reflect health care rendered in technologically sophisticated, expensive health care systems. In many other parts of the world, health care delivery systems are less tech- nologically advanced, and access to sophisticated therapy is frequently limited to residents of large met- ropolitan areas. Thus, the costs and benefits of health care services in one area may differ significantly from those found in other areas. Using the experience of North American and European countries to predict trends in health care for much of the rest of the world is speculative because both the future development of medical technology and the rate of its transference across national boundaries are largely unknown. Few estimates are available on the costs of smoking- related illness in Latin American and Caribbean coun- tries. In one report, an average of 19,000 deaths were attributable to smoking-related diseases in Venezuela during 1980 to 1984 (PAHO 1992). The costs of medi- cal care and employee absenteeism associated with smoking-related illness in Venezuela increased signif- icantly from 1978 to 1985 (from US$69 million to US$llO million). Because of the wide variation among countries in demographic structure, morbidity and mortality, health care systems, and prevalence of smoking, these results cannot be generalized to all of Latin America and the Caribbean. Financing of Health Care and Pension/ Disability Funds Considerable attention has been focused on not only the size of the economic burden of smoking-related illness but also on how societies will bear that burden. Miscalculations of economic burden have been de- rived by dividing prevalence-based estimates of the costs of smoking-related illness by the quantity of 112 Economics cigarettes sold. The resultant quotient has been re- ported as the per cigarette cost of smoking borne by society. For example, in the United States, $2.17 is frequently quoted as the cost of smoking per pack of 20 cigarettes (USOTA 1985). This overall cost fails to distinguish between the costs of smoking borne by smokers (internal costs) and those borne by others (external costs). The discussion of taxation (later in this chapter) explains how the magnitude of the bur- den imposed on nonsmokers by smokers is as much a function of the institutional arrangements for financ- ing health care, sick pay, disability, and retirement pensions as it is of the costs of smoking-related illness. Therefore, the incidence of the health costs of smoking varies among countries depending on the structure and scope of each country's social insurance system. Different national systems finance health care, disability, and retirement within the Americas. In some countries, participation in benefit programs is financed by payroll taxes or job-related insurance pre- miums. These types of programs are limited to per- sons who participate in the formal economy. Although national health insurance systems are man- dated in some countries, a Iow level of funding may limit the scope of public systems and lead to the cre- ation of private markets for health services. Informa- tion on the formal health care system may be inadequate for measuring the external costs of smok- ing-related illness; data may be needed on the actual source and disposition of funds. The U.S. health care system is financed by vari- ous government and private payment sources. In the United States in 1985, direct payments accounted for 24 percent and private insurance-principally pro- vided by businesses for their employees-ac- counted for 33 percent of the total personal health care expenditures. The federal government paid for 30 percent, mostly through Medicare (a federal program for disabled persons and persons aged 65 or older) and Medicaid (a program that provides health care for the poor). State and local governments paid for 11 percent of health care expenditures, largely through contribu- tions to the Medicaid program. Government health programs are financed by various mechanisms, in- cluding a payroll tax. The cost of employer-financed health insurance is included in total payroll costs and is reflected in prices; profits, and wage rates. Public old-age pensions and disability payments are financed through the federal Social Security Administration for most persons in the work force, but private plans account for a substantial proportion of benefits (Lazenby and Letsch 1990). In Canada, health care is financed through a national system separately administered by each province, with some direction and funding from the federal government. The Canadian government finances a comprehensive set of medical benefits and restricts funding by private sources, but Canadian citizens can select their own health care providers. Physicians' fees and hospital budgets are negotiated by the government, and savings are achieved in part through the administrative simplicity of the insurance plans. In 1987, Canada spent US$1,483 per person for personal health services, and the United States spent US$2,031 (Igelhart 1989). In 1987, personal health ser- vices accounted for 8.6 percent of the total gross do- mestic product (GDP) in Canada and 11.2 percent in the United States (Igelhart 1989). These comparisons suggest that, on a per capita basis, Canada spends less on smoking-related illness than the United States does. Brazil has a mixed public and private system for financing health care but is moving toward a new constitutionally mandated, unified, and decentralized health system (The World Bank 1989a). Brazil spends approximately 5 to 6 percent of its total GDP on health care, an amount divided almost equally between the private and public sectors. About half of all public financing for health care is channeled through the National Institute for Medical Assistance and Social Security and is tied to employment (The World Bank 1989a). Health services, primarily basic services for the urban and rural poor, are funded by the Ministry of Health through the general budget. State and local governments, which also finance health care, ac- counted for 27 percent of public expenditures on health in 1986. Private health care is financed by indi- vidual persons, who directly pay fees for services, and private insurance, largely financed by employers, which features various capitation and reimbursement- for-expenditures insurance plans. In a recent survey of the Brazilian health care system, The World Bank concluded that "resources have been poorly allocated; little is spent on prevention and much on curative care (70 percent on hospitals alone); little is spent on the poor, and much on the middle class" (The World Bank 1989a, p. 44). In Venezuela, as in Brazil, access to health care is constitutionally guaranteed, but care is delivered both privately and through various government programs (Morgado 1989). The Ministry of Health is responsible for providing health care, and approximately two- thirds of the country's physicians are employed by the Ministry in some capacity. In addition, largely unreg- ulated private insurance reimburses both physicians and private hospitals on a fee-for-service basis. The physician-to-population ratio is high; however, as in other Latin American countries, physicians are con- centrated in the large urban centers. Ecolromics 113 The costs of smoking-related diseases may be substantial in Brazil, Venezuela, and other countries of the Americas with similar health care systems. The concentration of health care resources for curative care (mainly hospital and fee-for-service physicians' care) in urban, middle- and upper-class areas suggests that these groups consume a disproportionate share of the resources and that smoking-related diseases in these groups are treated aggressively. Smoking-related dis- eases may also be a more important source of illness in urban, high-income groups than in low-income groups because persons of high income are likely to have a longer, more intense exposure to tobacco use and a longer life span during which smoking- associated diseases may become manifest. Costs of Smoking-Control Policies and Programs Knowledge of the dangers of tobacco use and concern for public health have led to the development of smoking-control policies in several countries. (See Chapter 6 for a discussion of control efforts.) Many of these policies-such as restrictions on advertising, warning labels on tobacco packages and in advertise- ments, restrictions on smoking in public places, and increases in tobacco taxes-use few direct resources, but hidden or intangible costs may be associated with such policies. However, other smoking-control policies-such as public and school education pro- grams, lobbying efforts of smoking-control advocates, and enforcement of restrictions on cigarette sales, ad- vertising, and smoking in public places-use re- sources that can be considered part of the costs of smoking. The 1989 report of the Surgeon General presents a detailed analysis of smoking-control activities in the United States (USDHHS 1989). Such activities have Economics of the Tobacco Industry recently increased significantly in Canada, where the federa& provincial, and municipal governments have moved to increase tobacco taxes, restrict tobacco ad- vertising, strengthen product warnings, restrict smok- ing ir public places, and help tobacco growers diversify and produce other crops (Collishaw, Kaiser- man, Rogers 19901. Except for the program to dis- courage tobacco cultivation, these policies and programs use few direct resources. These programs reflect, in part, the health advocacy of more than 30 voluntary agencies working individually and collec- tively (as the Canadian Council on Smoking and Health). Such advocacy activities, although rarely costed-out, consume resources that should be in- cluded in estimates of the costs of smoking-control activities. Through the initiative of local medical leaders and health and education authorities, Brazil's first antismoking campaign began in Port0 Alegre in 1976 (The World Bank 1989a), spread to other regions, and gained support. In 1985, the Ministry of Health began to develop a national program to control smoking. A recent evaluation by The World Bank (1989al cited the Brazilian program as a success, although the effects of the program on smoking patterns have not been for- mally assessed. Health planners from The World Bank found that "public information and personal smoking-cessation services," which cost only 0.2 to 2 percent of per capita gross national product (GNP) for each year of life gained, were the most cost-effective of the preventive and therapeutic interventions re- viewed. In contrast, treatment for lung cancer cost 200 percent of per capita GNP per year of life gained. This comparison suggests that public information pro- grams designed to control smoking in Brazil are ex- tremely cost-effective. The Tobacco Sector Overview From an economic perspective, the existence of a market for tobacco indicates that tobacco produces some economic benefits, including (1) consumer satis- faction from smoking and other forms of tobacco use and (2) income to producers in excess of the cost of resources for tobacco production. Tobacco produc- tion also generates costs-principally the value of re- sources used to manufacture tobacco products. Confusion about the costs and benefits of tobacco pro- duction has been spawned by tobacco industry ana- lysts who label the value of the land, labor, and capital used in tobacco production as a benefit of such pro- duction (Tobacco Growers' Information Committee, nd.; 114 Economics Agro-economic Services Ltd. and Tabacosmos Ltd. 1987). In fact, because the resources used in tobacco production are not being used for other products, the cost of these resources is the true resource cost of tobacco production. The value of the alternative goods that could be produced with the resources allo- cated to tobacco production is a measure of the oppor- tunity costs of producing tobacco. A tobacco industry may also generate tax revenues, which are neither benefits nor costs to a society. Rather, taxes are trans- fers of resource claims from one segment of society to the government for redeployment. Subsidies, such as agricultural support programs, are also transfer payments. The cultivation of tobacco is prima facie evi- dence of tobacco's net contribution to growers' in- comes. Although tobacco production may be very profitable for the individual producer, it is not neces- sarily beneficial economically. Subsidies and exter- nalities associated with the production of tobacco may lead to a divergence between what is best for produc- ers and what is best for society as a whole. Demand for Tobacco Worldwide consumer demand for tobacco prod- ucts drives the market for tobacco. In the economist's view, this demand originates from consumer efforts to satisfy exogenously determined wants, which are sub- ject to constraints on consumer resources. Such con- straints include limits on time and disposable income. By using information about products and prices, each consumer purchases a mix of goods to maximize con- sumer satisfaction. One of tobacco's benefits is the avoidance of nicotine withdrawal symptoms by addicted smokers. This benefit and other pleasurable sensations, called "utility" by economists, may have many components, including status, enjoyment, relaxation, a sense of se- curity, affihation with other smokers, and perhaps in certain cultures, a sense of being modern or progres- sive. However difficult these attributes are to mea- sure, economists posit that when consumers choose to spend some of their own limited resources on tobacco, they reveal their preference for purchasing tobacco than for engaging in other forms of consumption or savings. Price is a measure of the amount of alternative goods forgone to purchase tobacco products. (The effects of variation in cigarette price on tobacco con- sumption are discussed later in this section.) Tobacco products, as well as most consumer goods, tend to obey the law of downward sloping demand-as price falls (rises), quantity demanded increases (decreases). Factors that increase the retail price of cigarettes, in- cluding taxes, tariffs, and import quotas decrease con- sumption. The cost of raw tobacco is generally not an important factor in the retail price of tobacco products. In addition, although the supply of cigarettes does not affect demand directly, supply influences consump- tion through the market price: as supply increases, price tends to decrease, which stimulates consump- tion until the additional sales clear the market. Factors other than price that influence the demand for ciga- rettes and other tobacco products are cited in Figure 3. Income determines a consumer's command over resources and limits consumption options. In general, the consumption of most goods increases as income increases, but at a decreasing rate as consumers reach satiety for a particular good. The income elasticity of demand is defined as the percent change in the quan- tity demanded divided by the percent change in in- come that caused the demand change. The relation of consumption to income can be observed for individu- als, groups, and countries, for which income and con- sumption fluctuate over time, and for variations in income and consumption among groups at a particu- lar time. For countries in the Americas, the correlation is positive between per capita cigarette consumption and per capita GNP (Figure 4 and Table 5). This relation is stronger in less-developed countries in Figure 3. Factors, other than price, that affect the demand for tobacco products Reducing Factors Restrictions on sales to minors Restrictions on places for smoking Public education on harmful effects of tobacco use Health warnings on packaging and in advertising I Perception of harm from tobacco use 1 Demand t Augmenting Factors Disposable income of smokers and potential smokers Smokers preference for attributes of tobacco products Advertising and promotion Addiction to nicotine Economics 115 Figure 4. Per capita cigarette consumption and annual per capita gross national producF (GNP) in 24 countries of the Americwt 1985 r . . O ? o ? ? ? 0 2poO 4poO 6poO ~Poo 10,000 12,000 14poO 16JXKl 18,000 20,tUbO Per capita GNP *Using a model that compares the annual per capita consumption of cigarettes to the log of the GNP, the n$ationship is expressed by the following linear regression equation: Consumption = -3241+ 616 ln(GNP per capita) (I? = 68). This equation was used to calculate the elasticities discussed in the text. `See Table 5. which rising incomes frequently lead to increased cigarette consumption due to an increase in the per- centage of the population that smokes and in the amount each smoker smokes and to a shift from homemade and roll-your-own cigarettes to more- expensive, factory-made, higher-quality tobacco products. Several studies indicate that income elasticity measured for multiple countries is higher than that measured for a single country (Table 6). The estimates reported by Chapman and Richardson (1990) and Townsend (1990), and the estimate based on the data in Figure 4, cluster around 0.50 (0.45 to 0.55). How- ever, elasticity tends to fall as income rises, and near- zero estimates have been reported for developed countries (Table 6). In the model that compared consumption to the logarithm of GNP (Figure 41, esti- mated income elasticity of demand is approximately 2.0 at the lower end of GNP but falls to almost zero (0.04) at the upper end. Restrictions on cigarette sales or on where smok- ing is permitted make smoking more difficult. These restrictions raise the total effective price of cigarettes for consumers and reduce cigarette consumption. In- creased perception of the harm of cigarette smoking also depresses demand by increasing the total price of cigarettes (including health-associated costs) or by affecting taste. Physical characteristics of cigarettes, such as fil- ters, and aspects of taste, which include strength, flavor, and smoothness, augment demand. In many countries, the modern tobacco industry developed 116 Economics Table 5. Per capita* cigarette consumption and income in the Americas Per capita cigarette GNP+ per capita Change in Average annual growth in consumption KJS$) consumption (%) GNP (%I Country (1985) (1987) (1970-1985) (1965-1987$ North America United States 3,370 18,530 -15 1.5 Canada 2,392 15,160 -30 2.7 Latin America Argentina 1,780 2,390 3 0.1 Bolivia 330 580 10 -0.5 Brazil 1,700 2,020 30 4.1 Chile 1,000 1,310 -7 0.2 Colombia 1,920 1,240 15 2.7 Costa Rica 1,340 1,610 -20 1.5 Cuba 3,920 -2 Dominican Republic 930 730 -11 2.3 Ecuador 880 1,040 26 3.2 El Salvador 750 860 -21 -0.4 Guatemala 550 950 -26 1.2 Haiti 240 360 -55 0.5 Honduras 1,010 810 7 0.7 Mexico 1,109 1,830 2.5 Nicaragua 1,380 830 10 -2.5 Panama 894 2,240 2.4 Paraguay 1,000 990 4 3.4 Peru 350 1,470 -10 0.2 Uruguay 1,760 2,190 14 1.4 Venezuela 1,890 3,230 -4 -0.9 Caribbean Barbados 1,380 20 Guadeloupe 1,080 -1 Guyana 1,000 390 -26 -4.4 Jamaica 1,190 940 -34 -1.5 Suriname 1,660 60 Trinidad and Tobago 1,600 4,210 -16 1.3 Source: The World Bank (1989b); U.S. Department of Health and Human Services (1989); Chapman and Wong (1990). Aged 18 years or older. `GNP = Gross national product. $1982-1988 data. because of a shift in consumption from traditional forms of tobacco to modern, machine-made, quality- controlled, flavored cigarettes made from blends of tobacco, including fabaco rubio, a flue-cured tobacco. Some authorities have suggested that the develop- ment of filter-tipped cigarettes and long, slim ciga- rettes has increased smoking among women (see Chapter 2, "The Emergence of the Tobacco Compa- nies"). The addictive nature of tobacco, another demand-augmenting factor, is discussed in a prior report (USDHHS 1988). The degree of competitiveness or structure of the market for tobacco products can also affect the de- mand for cigarettes by operating on retail price, prod- uct differences, and product promotion. In many countries, the market for tobacco products may be reserved for a government-operated or sanctioned monopoly, but cigarette markets in the Americas are characterized by oligopoly-dominance of the market by several large firms (see Chapter 2, "The Emergence of the Tobacco Companies"). Prices tend to be lower and aggregate advertising and promotion expenditures Economics 117 Table 6. Estimates of income elasticity of demand for cigarettes Study Chapman and Wong (1990) Chapman and Wong (1990) Walsh (1980) Witt and Pass (1981) Lewit and Coate (1982) Townsend (1990) Data in Figure 4 Data Worldwide, 1980 Countries with gross national product ~$5,000 per capita, 1980 .55* Ireland, 1953-1976 United Kingdom, 1955-1975 .33 .13 United States, 1976 .08 Europe, 1987-1988 24 countries of the Americas, 1985 Elasticity .45* .46 .49 `Estimates calculated for this report from data provided in Chapman and Wong (1990). tend to be higher in oligopoly markets than in monop- oly markets, because of competition. In addition, oli- gopoly markets are characterized by greater variety as firms attempt to capture market niches for specific products. Cigarette advertising and the sponsorship of en- tertainment, sporting, and cultural events are intended to increase the demand for particular cigarette brands. Measuring the effect, if any, of such advertising on aggregate demand is problematic. Accordingly, pub- lic policy toward cigarette advertising and promo- tional activities is controversial in many countries. Assessment of the impact of tobacco advertising and advertising restrictions was presented in the Surgeon General's 1989 report (USDHHS 1989) and is updated below. Advertising In the United States, cigarettes are one of the most heavily advertised products, and the mix of ad- vertising and promotion has changed over time. Cig- arette commercials have been prohibited from television and radio since 1971. In 1975,75 percent of expenditures were directed toward traditional print advertising media (newspapers, magazines, bill- boards, and point-of-sale posters) and 25 percent to- ward promotional activities, such as coupons, free samples, public entertainment, and allowances to re- tailers (CDC 1990). By 1988, when total expenditures reached $3.27 billion, promotional activities ac- counted for more than two-thirds of all advertising and promotional expenditures. Despite the sizable decline in the use of traditional print media from 1975 to 1988, cigarettes were in 1988 the product most heav- ily advertised on outdoor media, the second most heavily advertised in magazines, and the sixth most heavily advertised in newspapers (CDC 1990). In many other countries of the Americas, tobacco advertising expenditures are substantial (Table 7), despite restrictions on advertising activities (see Chapter 5). The Canadian Tobacco Products Control Act banned all tobacco advertising in the Canadian print media beginning January 1,1989, and required that outdoor advertising on billboards and spon- sorship of sporting and cultural events be phased out (Collishaw, Kaiserman, Rogers 1990). This advertis- ing ban is currently being contested by Canadian tobacco companies in a protracted court case (Col- lishaw, Kaiserman, Rogers 1990). Advertising aims to increase profit by increasing demand for a particular product (Scherer 1980). In oligopoly markets, advertising is used to differentiate Table 7. Estimated advertising expenditures* of tobacco industry in selected countries of the Americas Country Cost United States $3,270.0 Canada 88.0+ Argentina 18.5 Brazil 68.0 Costa Rica 1.8 Dominican Republic 2.4 Ecuador 1.0 El Salvador 0.9 Guatemala 1.8 Mexico 19.8 Panama 1.8 Uruguay 0.7 Source: Philip Morris International Inc. (1988); ERC Sta- tistics International Limited (1988); Centers for Disease $2ontrol(1990); Chapman and Wong (1990). Estimates are for 1986,1987, or most current year available; in millions. `A phased-in ban on tobacco advertising began in January 1989 and is scheduled for completion by January 1993. A court ruling declared the law unconstitutional, but it remains in effect pending appeal (RJR-Macdonald Inc. v. Attorney General of Canada 1990; Imperial Tobacco Limited v. Attorney General of Canada 1990). 228 Economics among similar products and to build sales or to sustain the price of a particular product (Scherer 1980). Ad- vertising attempts to associate smoking with attri- butes generally considered positive, such as high-style living, healthful activities, and economic, social, and political success; it fails to voluntarily provide infor- mation on the substantial hazards of cigarette con- sumption. In emphasizing the positive attributes of a product, advertising may increase demand for both a particular brand and a class of products. Much of the debate over tobacco advertising has focused on whether such advertising increases cigarette sales and, consequently, has a negative impact on public health, or whether advertising is strictly a competitive device tobacco companies use to determine relative market share in a stable or declining market, in which case such advertising would have little effect on public health (USDHHS 1989). The results of many analyses of the effects of advertising on cigarette consumpt-ion were reviewed in the Surgeon General's 1989 report, which cited the conclusion that it is "more likely than not that advertising and promotional activities do stimulate cigarette consumption" (Warner et al. 19861, although precisely quantifying the influence of these activities on the level of consumption may not be possible. Evidence from the Canadian advertising ban and the continuing debate over increasing restrictions on advertising in the United States (Koop 1989) and other countries suggest that focus has shifted from the impact of advertising per se to the effects of advertis- ing restrictions on consumption. An extensive study of this issue was performed by the New Zealand Toxic Substances Board (1989) in support of its recommen- dation for a total ban on tobacco promotion in that country. The relation between tobacco advertising bans and tobacco consumption was examined from 1976 to 1986 in 33 countries. The study demonstrated that "government tobacco advertising bans and con- trols are accompanied by enhanced rates of fall in tobacco consumption" (page xxiii) and that "the greater a government's degree of control over tobacco advertising and promotion, the greater the annual average faI1 in tobacco use in adults and young people" (page xxiv). As a follow-up to the New Zealand report, Laugesen and Meads (1990) examined the effects of tobacco advertising restrictions, price, and income on tobacco consumption between 1960 and 1986 in 22 economically developed countries. They found that a total ban on tobacco advertising would have lowered average consumption by 5.4 per- cent in 1986 in countries without a total ban at that time. However, these studies have limitations-pri- marily a failure to account for the potential bias that antitobacco sentiment may be stronger in countries that ban advertising than in countries that do not. Accordingly, restrictions on tobacco advertising are, to some extent, markers of antitobacco sentiment, and a portion of the decline in consumption in countries with bans may be attributable to this sentiment rather than to advertising restrictions. In addition, both studies primarily included developed countries with a high but declining level of tobacco consumption. Extrapolation of these findings to less-developed countries with different patterns of tobacco consump- tion may be inappropriate. Supply of Tobacco Tobacco, which is grown in more than 120 coun- tries, is the most widely grown nonfood crop. It is grown in most developing countries, and the share of tobacco production in developing countries has in- creased steadily from 50 percent of world production in 1961 to 1963 to 58 percent in 1972 to 1974 to 69 percent in 1987 (Stanley, in press) (also discussed in Chapter 2, "The Emergence of the Tobacco Compa- nies"). In the past decade, most of the increase in worldwide tobacco production has been in China, which accounts for about 34 percent of total world production (Table 8). Major producers in the Ameri- cas include the United States (almost 10 percent of Table 8. Share of world tobacco production, 1990 Country Major producers China United States India Brazil USSR Other producers in the Americas Canada Argentina Mexico Cuba Colombia Dominican Republic Paraguay Venezuela Chile Production* 33.5 9.8 7.3 6.3 5.4 1.1 1.0 0.9 0.6 0.6 0.4 0.3 0.2 0.1 Source: Food and Agriculture Organization of the United Nations (1990). *As percentage of world output; computed from weight of crop. Economics 119 total world production) and Brazil (about 6 percent). Worldwide, about 22 percent of tobacco leaf by weight is grown in the Americas. Tobacco production is in- creasing more rapidly in developing than in devel- oped countries and is expected to increase in developing countries to more than 72 percent of world production by the year 2000 (Food and Agriculture Organization of the United Nations [FAO] 1990). In the Americas, tobacco production is expected to de- cline from 23 percent of world production in 1984 to 1986 to 21 percent by the year 2000 (FAO 1990). Considerable differences exist between the quality and, hence, the price of tobacco leaf produced in different countries. For example, tobacco grown in the Americas is worth almost four times as much as tobacco produced in China, although by weight, the American crop is only 65 percent of the Chinese crop (Agro-economic Services Ltd. and Tabacosmos Ltd. 1987). Tobacco production is mainly concentrated on small farms in limited geographic areas. The value of the typical tobacco crop frequently makes tobacco an important source of income not only for growers but for local agricultural workers, even though tobacco is often grown in rotation with other crops. Compared with most other crops, tobacco uses little arable land (about 0.3 percent worldwide), but tobacco cultivation is labor intensive (Table 9) (Muller 1978). The tobacco industry's ability to create employment is valued in areas where labor is plentiful and production alterna- tives are few. Millions of persons are involved in or dependent on some stage of the tobacco-production process for a portion of their livelihood (Agro- economic Services Ltd. and Tabacosmos Ltd. 19871, Table 9. Labor* and land use in tobacco growing, processing, and manufacturing in the Americas, 1983 - Country North America United States Canada Latin America Argentina Bolivia Brazil Costa Rica Chile Colombia Cuba Dominican Republic Ecuador El Salvador Guatemala Haiti Honduras Mexico Nicaragua Panama Paraguay Peru Uruguay Venezuela Caribbean Tamaica Growing No. FTE+ 66.80 105.40 43.90 600.00 288.90 3.76 1.93 302.00 100.50 20.00 17.00 24.20 6.55 1.23 1.23 351.00 117.00 10.00 95.00 59.68$ 77.00 228.08 75.80 20.40 8.10 31.18 9.58 3.50 22.90 9.73 43.87 1.95 9.35 40.10 1.48 0.44 4.81 1.44 3.57 215.76 352.00 42.00 108.00 23.20 55.02 12.20 197.50 22.00 100.00 Processing and manufacturing Distribution FTE - No. FTE 7.70 120.20 2.60 30.30 13.40 0.93 1.52 25.90 1.90 6.70 Arable land used (%I 0.21$ 0.20 1 .oo 0.50 0.20 0.10 0.40 2.10 1.10 0.10 0.50 0.40 0.10 0.50 0.10 0.20 0.20 1.70 0.10 0.10 0.20 0.40 Source: Agro-economic Services Ltd. and Tabacosmos Ltd. (1987); Chapman and Wong (1990). *In thousands of workers. +FTE = Full-time equivalent. SFor 1989; U.S. Department of Agriculture unpublished estimates. 120 Economics and persons in certain regions may substantially de- pend on tobacco. Tobacco farming is also highly seasonal. If the work could be spread evenly throughout the year, the average-sized tobacco farm could be managed by one full-time farmer, with some time remaining (Stanley, in press). However, because many workers are needed for harvesting and planting, tobacco farming provides many countries with part-time, seasonal em- ployment for many laborers (Table 9). The average number and full-time equivalent (FTE) number of workers employed in tobacco growing and other aspects of the tobacco industry vary widely in the Americas. After tobacco is harvested, the crop is processed in various ways before being made into cigarettes and other consumer products. This processing includes sorting and grading, curing and drying, and destemm- ing the raw tobacco leaves. In most countries, these activities occur in agricultural areas and are included in statistics for the agricultural sector. In other coun- tries, some of these activities are associated with the initial stages of the manufacturing process and are included in statistics for that sector. Many features of the tobacco market make to- bacco particularly attractive to growers in many coun- tries. First, and most important, when tobacco is grown extensively, it yields a higher net income per unit of land than most other cash crops and substan- tially more than most food crops. In addition, price does not fluctuate substantially for tobacco as it does for other cash crops. Moreover, in most countries, tobacco growers protect themselves from the unex- pected price fluctuations that plague other crops by negotiating sales prices for crops before planting; growers are paid in cash immediately upon sale (Econ- omist Intelligence Unit 1983). The combination of prenegotiated price and quick sale makes tobacco growing easy to finance. The extremely favorable conditions of sale offered to tobacco farmers are not usually offered to growers of other crops. Various combinations of government and transnational tobac- co company activities, including controls on planting, production quotas (guaranteed prices, incentives, and subsidies), import duties, state tobacco monopolies, state trading in tobacco, foreign aid programs, and limitations on marketing, benefit tobacco growers in many countries. As a result, much of the risk of tobac- co growing is shifted from the farmer to the purchaser. Although tobacco provides most farmers with higher gross returns per hectare than many other crops do, considerable costs are associated with to- bacco growing. In addition to being labor intensive, tobacco cultivation requires large amounts of fertiliz- ers and pesticides, and in many areas, fuel (wood, gas, or oil) is needed for tobacco curing. The U.S. Depart- ment of Agriculture (USDA) estimated that, excluding land and quota cost, the cost of growing flue-cured tobacco in the United States in 1990 amounted to 70 percent of the value of the crop produced (Clauson and Grise 1990). In examining the opportunity costs of tobacco growing in Brazil in terms of alternative crops, Barrows (unpublished) found that the value that labor employed in tobacco growing would have in alter- native activities is the most important factor in deter- mining the profitability of tobacco. Barrows estimated that in 1986 in Rio Grande do Sul, total returns to land, labor, and management for tobacco were 130 percent of those for maniac and 118 percent for potatoes. However, cultivation of tobacco re- quired 7.5 times as many man-hours of labor as man- ioc did and 5.3 times as many man-hours as potatoes did. Accordingly, all of the apparent additional re- turns to the tobacco grower were in fact returns to the additional labor invested, and the actual profitability and net social benefit of the tobacco crop depended on the wage rate and the potential alternative uses of the labor employed in tobacco growing. Manufacturing Most of the tobacco grown worldwide is flue- cured and processed on the farms. Tobacco is then manufactured into cigarettes, cigars, smokeless tobac- co products, and loosely cut smoking tobacco. About 85 percent of worldwide tobacco production is used for cigarettes. Flue-cured tobacco accounts for almost 60 percent of the tobacco in American-style cigarettes and all of the tobacco in British-style cigarettes. The manufacturing of cigarettes provides sub- stantial employment in many countries, but the labor intensity of cigarette manufacturing varies consider- ably by country. In the United States, production is highly automated; seven factories produce enough cigarettes for the domestic market and for the large and growing export market. In Latin America, ciga- rette manufacturing is less automated and more labor intensive (Table 9). Cigar manufacturing is more labor intensive than cigarette manufacturing, which is reflected in the employment figures for countries that are important producers of cigars (e.g., Cuba and the Dominican Republic). Distribution Tobacco is distributed in many forms. Ciga- rettes are sold in cartons of 10 packs and in packs of 10, 15,20, and 25 cigarettes. In many areas, street vendors Economics 121 sell cigarettes individually from broken packs. In some countries, cigarettes are sold by tobacconists; however, cigarettes and other tobacco products are typically sold by retail merchants who also sell a variety of other consumer goods. Accordingly, in most countries, total employment in tobacco distribu- tion is many times FTE employment because tobacco sales represent a small part of the employees' jobs (Table 9). Distribution in the tobacco sector is a small com- ponent of larger distribution activities in most econo- mies. Although attributing some proportion of employment to tobacco distribution activities is statis- tically appropriate, such attribution may be inappro- priate for analytic reasons. In the absence of tobacco products, consumers would purchase alternative goods, and the production of these goods would result in employment-not only in the distribution sector but in the manufacturing and farming sectors as well. Although the level and type of employment generated by alternative consumption patterns may change with changes in the tobacco sector, total employment would not change significantly. Some persons, how- ever, may be affected by shifts in consumption pat- terns; some persons may become unemployed, and some may change jobs or job activities. The tobacco industry also creates output in other parts of the economy-both directly, by creating de- mand for products such as fertilizers, fuel, and paper used in the manufacture of tobacco products, and indirectly, when persons employed in the industry spend their earnings for their own consumption. Every economic activity, however, has both direct and indirect links to other economic activities. The exact nature of the links differs among industries and coun- tries, but the net aggregate effect of shifts in demand into or out of specific industries is small, except per- haps for some transitional costs. Exceptions may occur, however, for factors that receive higher-than- normal returns (called "rents" by economists) from a specific activity. Such factors are particularly disad- vantaged by a reduction in rent-producing activity; however, even their losses are balanced by gains to other factors of production or to consumers. Trade Most tobacco is consumed within the country of production; only 25 percent of world production is traded internationally, primarily as a raw commodity. Only the United States, the United Kingdom, and the Netherlands are important exporters of cigarettes, and the United States is the leading cigarette exporter-at 25 percent of the worldwide total. In addition, the United States exports much high-quality tobacco, which in several countries, is blended with tobacco from other sources to make the increasingly popular American-style cigarettes. The United States imports oriental tobaccoand other less-expensive filler tobacco to blend with U.S.-produced tobacco to make ciga- rettes for domestic consumption and export. Brazil, another major tobacco exporter sells much of its crop in Europe. On the whole, countries in the Americas have a substantial balance-of-trade surplus in tobacco (Table 10). Subsidies to Tobacco Production Subsidization may be used in an attempt to de- velop or protect a domestic tobacco industry or to control the importation of cigarettes or tobacco to conserve foreign exchange. The growing and curing of tobacco is frequently controlled and directed by the main tobacco purchasers-either large, private com- panies or government agencies. In many areas, these organizations set the price of tobacco before planting and provide seeds or seedlings to tobacco farmers, who are thus guaranteed a minimum income for their crop at harvest time. These production controls are primarily designed to encourage the production of a limited amount of high-quality, marketable tobacco (Lewit 1988). The situation in southern Brazil exemplifies an industry-sponsored support program for tobacco growers that has fostered the development of a tobacco- growing sector. The cigarette manufacturers provide the growers with all purchasable inputs-including seed, pesticides, and fertilizers-at wholesale prices, and maintain agricultural extension programs to de- velop tobacco plants and technology appropriate for the area. Farmers are visited regularly by technical advisers provided by the tobacco companies. The purchasers also control the chemicals used in growing tobacco so that the crop will conform to U.S. and European standards and be exportable (about 37 per- cent of the Brazilian crop is exported) (Economist Intelligence Unit 1983). The value of the extension services rendered to farmers is estimated at 30 to 35 percent of the prices paid to farmers for the tobacco (Economist Intelligence Unit 1983). A similar relationship exists in Venezuela among the government, two tobacco processors, and several hundred tobacco farmers. The farmers receive finan- cial and technical aid from the companies, along with guaranteed prices for crops. As a result, the compa- nies have some control over the quality and quantity of the tobacco crop, but the companies can also set retail cigarette prices. The Venezuelan government 122 Economics Table 10. International trade in tobacco, 1984 and 1985* Country Imports Exports Total Percentage of Total Percentage of Trade value all imports value all exports balance 0.1 1.3 0.6 1.7 0.1 0.6