Tobacco Documents Online

Curbing the ~ Govern~_r~ts and the Econor, ',,, Dra~4 • February I: Draft only, noi foreitation o_~_r ci_~culation Thtadocume~lthasarestrtcted~d:tr::nltioP'n~~~1~'~'~'~ '-'-~''~ ..... ~a~ed ~ recipients o~ ~f~ma, we of their o~cial duti~. Its conten~ m~ not othe~R¢ be disci~~ authorization in the form of written permi3sion ~om Prabhat Jha, T~k Team Leader 321422960

rbmg the eptdem~c: : Govern)~t__s and the Economics of TobacCo Control February 1999 Draft only, not)~e~culation Documen~the World Bank ~ This do~ment h~ a r~tricted distribution an~ ~ed by recipients ~ ~~mance of their o~eial d~ti~. Its contcnt~ m~ hoe oth~is~ b~ discl~ Ban~ authorization, in the form of wr~tt~n perm~sion from Prabhat Jha, r~k re~ Leader 321422961

Foreword and team members 1. Introduction: the purpose of this report 2. Executive Summary 3. Global trends in tobacco use 3. I Rising consumption in low-income and middle-income countries page xx 3.2 The impact of trade liberalization page xx 3.3 Tobacco use is increasingly associated with poverty page xx 3.4 Patterns of tobacco use by region, gender and age page xx 4. The health consequences of smoking page xx 4. I The addictive nature of tobacco use page xx 4.2 Long-term disease cause~by smoking page xx 4.2. l Damage to smokers themselves page xx 4.2.2 Damage to non-smokers page xx 5. Do smokers know their risks?page xx 5.1 Awareness of the health hazard~ is a deterrent to tobacco use page xx 5.2 Factors that complicate the relaticmship between knowledge and demand page xx 5.3 Predicting the impact ofinforrnafion in developing countries page xx 6. Is there an economic rationale for intervening in the tobacco market? page xx 6.1 Counting the costs of smoking page xx 6.2 Market failures and the grounds for intervention page xx 6.3 What types of intervention should be considered? page xx page xx page xx page xx page xx 7. Measures to reduce demand can succeed--and raise dollars page xx 7.1 Reducing demand by raising the price: tobacco taxes page xx 7.2 Reducing demand with improved health information page xx 7.3 Reducing demand with restrictions on smoking in public places and the workplace page xx 7.4 Nicotine replacement therapy page xx 7.5 Conclusion 8. Most measures to reduce supply will fail page xx 8.1 Policies with a low probability of success page xx 8.2 Control of smuggling remains a priority page xx 9. Is tobacco control ~vorth paying for? page xx 9.1 The impact of falling demand on consumer welfare and jobs page xx 9.2 The cost-effectiveness of interventions page xx 10. An agenda for action page xx 10.1 Overcoming barriers to change page xx 10.2 Research priorities page xx 10.3 Recommendations for action page xx Three components of tobacco control at national level page xx An agenda for international action page xx 321422962

List of Boxes List of Tables List of Figures Background papers, meetings and acknowledgements Annexes 321422963

Foreword Pending The Report Team This Report has been prepared by a team led by Prabhat Jha and Frank J. Chaloupka, and comprising Phyllida Brown, Son Nguyen, Jocelyn Severino-Marqucz, Rowena van der Merwe and Ayda Yureldi. Phillip Musgrovc, Thomas E. Novotny, Kent ganson and Abdo S. Yazbcck provided valuable contributions and advice. This work benefited from substantive early work on tobacco at the World Bank by Howard Barnum. Input from the World Health Organization was coord'~nated by Derek Yach, and input from the US Centers for Disease Control and Prevention was provided by Michael Eriksen. The work was carried out under the general direction of Helen Saxenian, Chris Lovelace and David de Ferranti. ". ': The production sta.ff of the Report includcd.=~ The Report benefited from input from an External Advisory Panel (see acknowledgements) and from many others inside and outside the Bank. Their work is gratefully noted here. External review meetings were held in Bcijing, China in August 1997, Cape Town, South Africa in February 1998, Lausanne, Switzerland in November 1998, and Washington, USA in March 1999. Fred Paccaud hosted the Economists Technical Review Meeting in Lausanne. Support for this report came from the Human Development Network, the Institute for Social and Preventive Medicine, University of Lausanne, and the Office on Smoking and Health at the US Centers for Disease Control Their assistance is warmly acknowledged. Draft: Not for circulation or Citation 321422964

I. Introduction: the purpose of this report Tobacco is expel:ted to become the single biggest cause of death worldwide within the next 20 years, with most of the burden borne by middle and low-income countries. Yet few national governments or international agencies have developed their responses to this global epidemic. Many policy makers have not taken steps to control tobacco because of fears that their action would have undesirable economic consequences. For, whereas there is now little dispute about the damaging impact of tobacco on health, debate about its contribution to national economies is growing. Some policy makers fear that raising taxes on tobacco to reduce demand would simply reduce government revenue or much more tobacco smuggling, lhdeed, the tobacco indus, try itself has increasingly used the~e arguments as it has shifted the substance of its opposition to tobacco control policy into the territory of economics. The World Health Organization, the principle international agency on health issues, has taken a clear lead in responding to the epidemic with its Tobacco Free Initiative. The World Bank aims to work in parmership with the lead agency, offering its particular analytic resources in economics. Since 1991, the World Bank has had a formal policy on tobacco, in recognition of the harm that it does to health. The policy (explained in more detail in Box 10.3) discourages lending on tobacco and encourages control efforts. At the 10th World Conference on Tobacco, in Beijing, China, in August 1997, the World Bank organized a consultation session on the economics of tobacco control. The meeting was part of an ongoing, two-year ~-eview of the Bank's own control policies. There was clear recognition at this meeting that insufficient global attention was being paid to the economics of reducing tobacco consumption in response to the global epidemic of smoking-related deaths. The meeting's participants also agreed that the discipline of economics was not being applied to tobacco control in many countries, and that even where economic approaches were being used, their methodology was of variable quality. At the same time that the World Bank began reviewing its policies on tobacco, economists at the University of Cape Town had begun a project on the economics of tobacco control for Southern Africa. These initiatives were brought together, in partnership with Fred Paccaud at the University of Lausanne in Switzerland and others, to form a wider project for a larger audience. The South Africa work culminated in a conference in Cape Town in February 1998. The proceedings of that conference are published ("The Economics of Tobacco Control: Towards an Optimal Policy Mix". Abedian et al, University of Cape Town, 1998). This report summarizes the policy implications reached as a result of a set of studies on the economics of tobacco control which grew out of the collaboration, and which will be published shortly ("Tobacco Control Policies in Developing Countries'" Jha P, and Chaloupka F J, editors, Oxford University Press, forthcoming late 1999). Draft: Not for circulation or citation 321422965

intended for ministers of economics, finance, trade and commerce, but its other readers may include health ministers and others. Importantly. this report offers striking new evidence of the impact of tobacco on the health and life expectancy of poorer people. The study focuses largely on economic issues It makes no attempt, beyond summarising the latest research, to add to the already extensive literature on the health consequences of tobacco, which are well established in most developed countries Nor does it at'tempt to provide psychosocial analyses of smoking behaviour, on which economic theory has little to say, leaving this important subject for others to discuss else,,~here. The purpose of this report is to provide an informed discussion of the modest, simple and effective economic measures that are possible to stem a growing ancl unprecedented epidemic. Draft: Not for ¢U'cuiauon or citation 321422966 BATCo t. S DOJ v Philip Morris

2. Executive Summary Tobacco is already killing some 4 million people a year. By the time today's children reach early middle age, it is expected to account for one in three of all adult deaths, more than any single disease. By 2030, perhaps a little sooner, 10 million people will die of tobacco-related disease each year-roughly equal to the current combined mortality from diarrheal disease, malaria, pneumonia and tuberculosis. The accelerating epidemic of tobacco-related disease and premature death will not, in the main, be borne by the rich countries. By 2020, 70 per cent of each year's deaths will be in low.income and middle-income countries. Half of these deaths will be in productive and socially important middle age (35-69), losing two to three decades of life both. While few people now dispute the damage that tobacco does to health, debate about the economic consequences of smoking continues. What are the costs of tobacco to society? Who bears those costs? What is smoking worth to smokers? Is tobacco a consumer good like any other or is it different because of i~s"addictivc component, nicotine? Are governments justified in acting to reduce tob~,cco consumption? Can their policies succeed, or will they destroy jobs, squeeze government revenues downward and encourage smuggling? The report addresses these economic questions. It concludes that the social costs of tobacco are large and growing and outweigh the value of smoking to smokers and producers. It argues that tobacco is not a typical consumer good and that governments are justified in intervening in the market. Drawing on the evidence from various countries and from economic models developed for this study, the report identifies the most effective policies and their potential impact, globally and in each region, on cigarette consumption and on tobacco-related deaths. Effective policies to reduce the demand for tobacco include raising taxes, imposing bans on advertising and promotion, improving warning labels and other health information for consumers, restricting smoking in public places and the workplace, and widening access to nicotine replacement therapy. Not only could these measures prevent millions of deaths among smokers alive now--an impact on a scale virtually unparalleled by conventional health measures--but they are also highly cost-effective. In contrast, the report warns, there is little to be gained from measures intended to reduce the supply of tobacco, such as crop substitution policies or trade restrictions. The concerns that have prevented governments from acting to reduce tobacco demand for fear of harming the economy are largely unfounded, the report concludes. Most economies would be unharmed overall by a reduction in tobacco consumption, and the handful of countries in which there would be net job losses would be able to adjust over a generation, not overnight. Imposing higher taxes would not erode Cigarette tax revenues; in fact, they would grow in the short-to-medium tenn. Fears that higher tobacco taxes would simply ifi.crease the illegal trade in smuggled cigarettes are also exaggerated. 321422967

The report begins by summarizing briefly what is known about trends in smoking worldwide and the effects of smoking on health, both in terms of addiction and long-term disease. It describes how people's knowledge of the health effects of smoking mediates their demand for tobacco. It discusses the economic rationale for intervening in the tobacco market and then examines various measures for doing so. It weighs the costs of implementing these measures against their potential gains, and then makes recommendations for governments and international agencies. Trends in smoking An estimated 1.1 billion people worldwide smoke, 80 per cent of them in low-income and middle-income countries. Although tobacco consumption is declining in the high-income countries, it has risen elsewhere, on ayerage by 3.4 per cent per year over the last two decades. Freer trade between nations'is contributing to increased tobacco consumption in low-income and middle-income countries, though not in high-income countries. The habit of smoking is almost always acqui.r.cd in youth or early adulthood. In the USA, 80 per cent of smokers begin bcforc the age of l 8. Poor people in industrialized countries are now more likely to smoke than rich people, a reversal of the pattern in thc early decades of the twentieth century. While this association between smoking and poverty may not be established worldwide, it has nevertheless been observed recently in several large low-income and middle-income countries, for example China and Brazil. The health consequences Nicotine addiction. One component of tobacco, nicotine, is an addictive substance as defined by major medical organizations. Nicotine addiction is established rapidly after the start of smoking. In the USA, studies suggest that four out of five teenage smokers believe that they will have quit in five years' time. In reality, almost half of the would-be quitters still smoke five years later. Studies in several industrialized countries show that around g0 per cent of adult smokers regret having started, and would like to stop. However, success rates for quitting arc comparatively low: about 97 per cent of smokers who try to quit unaided will have started again within 6 months. Long-term disease. About half of all smokers will eventually be killed by their habit, according to studies in the industrialized countries and China. Half of those killed will die in productive middle age, losing perhaps 2 or 3 decades of life. Major fatal diseases that are strongly associated with smoking in industrialized Western nations include lung cancer, heart disease, emphysema and various other types of cancer. In China, heart disease is a less significant cause of death, while emphysema and other chronic airway diseases are more important. Among Chinese smokers, deaths from tuberculosis equal that from heart disease. Since poor people are more likely to use tobacco than rich people, they are also more likely to be harmed by it. Analyses for this report show that, in Canada, Poland, the USA, Draft: Not for circulation or cttation 321422968

and the UK., tobacco may be responsible for at least half the widening gap between rich and poor men's risks of premature death between ages 35-69. Tobacco affects the health of non-smokers too. Babies born to smoking mothers arc at greater risk from respiratory diseases, have lower birth weights, and are more likely to dic of sudden infant death syndrome than babies born to non-smokers. Adult non-smokers exposed to tobacco face slight but real increased risks of respiratory disease and cancer. Non-smokers also face nuisance costs from unwanted smoke. Knowledge and tobacco In general, the greater individuals' knowledge and understanding of the health effects of smoking, the less likely they are to smoke. Two types of study provide the evidence for this assertion--those that monitor trehds over time, and those that examine smoking bchaviour at any one time across different socioeconomic groups in a society. For example, time-based studies in the industrialized nations have shown a steep decrease over the past three decades in the number~)f educated people who smoke; whereas less educated people have reduced their tobacco,consumption only slightly. Cros~-scctional studies from high-income and low-income Countries alike, rangifig from the USA to Taiwan to Nigeria, show that tobacco consumption is greater at any one time among those with only basic education than among those with higher education. It might appear that the health effects of smoking are now universally known. This is not the case, however. A study in China in the mid-1990s found that six in ten of smokers questioned there believed that tobacco did them "little or no" harm. There is also widespread ignorance about the types of disease that smoking can cause. Even for those who have access to health information, however, the relationship between knowledge and demand is complex. Studies assessing individuals' perception of the risks they face from smoking reach different conclusions, but the most widely accepted of these find that people tend to underestimate the hazards of tobacco relative to other perceived risks. Whether they estimate the statistical risks accurately or not, they may fail to internalize those risks, believing themselves somehow exempt from the statistics. Perhaps the greatest problem is that teenage smokers and young adult smokers tend to discount the future more heavily than adults, placing a much lower value on life lived in the future than on life lived in the present. Thus, at the time that they are most likely to adopt smoking, they are also least likely to believe that it will harm them. Once they have a more accurate risk perception, they are likely to be addicted to nicotine. The discipline of economics has comparatively little to offer in understanding how to overcome this problems, but this report acknowledges its significance. there an economic rationale for intervening in the tobacco market? From the standpoint of public health, the world would undoubtedly be better off without any tobacco at all. From the standpoint of economics, however, the situation is more Draft: Not for circulation or citation 321422969

complex. Tobacco is a legal good, and consurncrs of tobacco pay for it because they gain something from smoking. A basic principle of modem economics is that individual consumers know what is best for them and that, by choosing what goods and services to pay for, they will maximize their wellbeing or "welfare" within the limits of their preferences and income. Most economists believe that the most effective means of maximizing all consumers' welfare, and hence the allocation of resources, is through free competitive markets. However, it is recognized that some markets fail to achieve such "allocative efficiency". Market failures arise, for example, where the consumption of a consumer good has external costs-those that are not taken into consideration in the transactions between producer, consumer and distributor, and not bomc by these parties but by others. Other market failures include the inadequate provision of information about a product by its manufacturers to consumers, which complicates consumers' ability to assess what they are getting for their money. Where substantial market failures arise, gov~cmments may be justified in intervening to correct these failures, for example by taxing the consumer good, provided that they can do so without creating further problems in the economy. The tobacco market has several failures. One is that information about the health consequences of smoking-both for addiction to nicotine and for long-term risks of fatal disease--is not made fully available. The industry has no motivation to provide such information and many consumers are not fully aware of the risks. In addition, smoking has several types of external cost, including the nuisance and health risks to non-smokers of exposure to environmental tobacco smoke. Tobacco also has features that make it very different from most typical consumer goods. First, the vast majority of new recruits to smoking are in their teens. At this age, most individuals discount the future heavily and perceive themselves less vulnerable to risks than older adults. Second, tobacco contains an addictive substance, nicotine. Most young smokers become addicted within a short time and by early adulthood regret having taken up the habit. In these circumstances, it is difficult to argue that they choose to buy tobacco as they might choose, say, to buy shirts. This particular consumer good combines inadequate knowledge of the risks, addiction, and a high probability of dying prematurely. The report explores the complex question of what tobacco use costs society, health systems and governments. It weighs the economic costs 9fsmoking against the perceived economic benefits, that is, the value that smokers place on smoking, measured in terms of smokers' consumer surplus. It finds that, as with other drugs, the consumer surplus of tobacco is large, but that, once smokers' lack of information about the health hazards of tobacco, and their addiction to nicotine, are taken into account, the size of the surplus falls dramatically. The report shows that the consumer surplus of tobacco in the high- incbme countries has fallen steadily over recent decades as more and more people quit, and suggests that the global surplus may eventually turn downwards too. In contrast, the Draft: Not for circulation or citation I0 321422970

costs of tobacco use are large and growing. The conclusion is that, while there arc still um-esolved issues of how to quantify the costs and benefits of smoking, the costs arc ultimately greater and rising while the benefits are overestimated by traditional economics, and falling. Government intervention is therefore justified, the report concludes, to reduce the risks to smokers and to shift the costs of smoking more squarely onto tobacco's producers and consumers. Within the confines of the discipline of economics, intervention would be justified only to protect children and non-smokers and to increase information for all consumers. For adult smokers, the principle of consumer sovereignty would dictate that smokers should be leR alone. But conventional economics is poorly equipped to analyze the market for consumer goods that have addictive properties. Policy makers must weigh consequences of several choices. They might decide to adopt only those interventions that protect children and non-smokers and still achieve substantial health gains halfway through the 21st century. Alternatively, they might consider that the principle of consumer sov.creignty is not sufficiently clear in the case of tobacco to allow indifference to the expect .¢d deaths of quarter a billion of today's adull smokers in the shorter term, many of which may be preventableby prompt action. In that case, interventions that will affect all smokers would bc justified. Effective interventions to reduce demand Using the evidence from industrialized and developing countries, the report examines the impact of various interventions. It finds that the single most effective policy is to increase taxes on tobacco. Raise taxes In the past, it was thought that smokers would be so dependent on cigarettes that they would pay anything for them. Today, however, a substantial body of research in the high- income countries shows that price increases induce some smokers to quit, prevent others from taking up the habit, and reduce the number of ex-smokers who return to it. Addiction to nicotine means that these changes will be made comparatively slowly, however. Thus the impact of a tax increase is to decrease tobacco consumption and, for the short to medium term, increase tax revenues. Models for this report show that tax increases of just I 0 per cent applied to the real price of cigarettes worldwide could induce 40 million people worldwide to quit, and could prevent l 0 million tobacco-related deaths, among smokers alive in 1995. The assumptions on which the model is based are highly conservative, and these figures must therefore be regarded as minimum estimates. Policy makers traditionally raise several objections to the idea of higher tobacco taxes. One is that higher tobacco taxes will simply reduce government revenues. The report prbvides evidence from industrialized countries that this is not the case, and that, in fact, tax revenues climb in the short to medium term because of consumers' relatively slow Draft: Nnt for circulation or citation I1 321422971

response. In the UK, for example, a tax increase of I per cent increases revenue by between 0.6 and 0.9 per cent. A model developed for this study suggests that a modest ten percent increase in cigarette excise taxes would increase cigarette tax revenues by about seven percent, or about 10 billion US dollars more globally. [suggested graphics to add if these are possible: summary table to show deaths averted, by region, by 10per cent tax increase (improve to impact of IOper cent per year for a decade when done), NRT and non-price measures; and graphic if possible to show impact of tax increase on tobacco tax revenues over a given period.J A second objection is that higher taxes on tobacco would have a disproportionate impact on poor consumers, creating household hardship. The report argues that poorer consumers may not have higher tax burden from. increases, given they are more price sensitive. Moreover, the burdens of higher taxation need to be considered with benefits-which would accrue more to the poor. A third objection is that higher taxes will simply create greater incentives for tobacco smuggling from low-tax states or countries into high-tax states or countries. If this happens, consumption would remain high but government revenues in the high-tax state or country could fall. The report shows, with empirical evidence and econometric models, that while tobacco smuggling is a serious problem, its impact on tax revenues is unlikely to be large, and that the appropriate response to smuggling is to crack down on criminal activity, not to lower tax rates. "Non-price "' measures: more health information, advertising bans and restrictions on indoor smoidng Next, the report discusses a range of"non-price" measures that governments have already employed effectively to reduce the demand for tobacco. These measures include better warning labels, information campaigns, the publication and dissemination of research findings on the health consequences of smoking, and counter-advertising; comprehensive bans on advertising and promotion of tobacco; and restrictions on smoking in workplaces and public places. The report provides evidence that each of these measures can reduce demand for cigarettes. For example, studies of the effects of so-called "information shocks", such as the publication of research studies on the health effects of smoking, show that these consistently reduce demand by a few percent, and that the cumulative effect of several such shocks over time can have a permanent and substantial effect. A comparison of study of more than 100 countries shows that those with a comprehensive ban on advertising and promotion have seen cigarette consumption fall by 8 per cent over the study period, compared with a fall of only 1 per cent in countries without such bans. Recent econometric studies in Europe support such an effect size. In the case of indoor smoking restrictions, the clearest benefit is to non-smokers, but there is also some evidence that such restrictions can reduce the prevalence of smoking. Models developed for' this study suggest that, employed as a package, such "non-price" measures used globally could persuade some 23 million smokers alive in 1995 to quit and avert the Draft: Not for circulation or citation 12 321422972

deaths of 5 million of them. As with the estimates for tax increases, these are minimum estimates. Wider access to nicotine replacement therapy A third intervention would be to aid those who wish to quit by making it easier for them to obtain nicotine replacement therapy (NRT). NRT is usually self-administered by individuals and depends on buying products such as gums and patches that deliver nicotine without the requirement to smoke tobacco. Evidence from the high-income countries indicates that NRT is a relatively cost-effective and efficient means of enabling people to quit, Compared with other interventions. Yet in many countries NRT is difficult to obtain. Models for this study show that, with extremely conservative assumptions about its effectiveness, NRT made widely available could reduce tobacco-related deaths among today's smokers by 14 milli6h. The combined effect of all these measures is not known, since smokers in most countries are exposed to a mixture of them and none can be studied strictly in isolation. However, even if there were no additive effect, the number of deaths that could be averted by modest demand-reducing measures is clcariy greater than for most traditional health interventions. Most measures to reduce the supply of tobacco will fail While interventions to reduce demand for tobacco are likely to have a substantial effect, measures to reduce its supply are less promising. A useful comparison may be made with the illegal drug trade. The experience of the United States in attempting to reduce the supply of narcotic drugs shows that measures such as crop destruction and domestic law- en£orcing crackdowns had substantially less effect than measures to reduce demand. Action to prevent smuggling, such as better tracking of cigarette consignments and tough penalties to deter smugglers is likely to have some impact, however. Other measures have less chance of success. For example, there is little evidence from around the world that crop substitution can succeed as a means of reducing the tobacco supply, since the incentives to farmers to grow tobacco ate currently much greater than for other crops. Similarly, the evidence so far suggests that trade restrictions, such as import bans, will fail to stop cigarette consumption from rising worldwide. Instead, countries are more likely to succeed in curbing tobacco consumption by imposing the measures that are known to reduce demand (higher taxes, better information, and so on) symmetrically on imported and domestically-produced cigarettes. ls tobacco control worth paying for? The. report asks what it would cost to introduce tobacco controls and what the gains could be, in terms of lives saved and tax revenue generated. Draft: Not for circulation or citation 13 321422973

The costs to economies of reducing tobacco consumption are also explored. The tobacco industry is often.described as playing a major role in national economies. While this is true for a handful of countries, the report finds that most economies would not lose jobs overall if tobacco consumption fell. This is because money not spent on tobacco would be spent elsewhere in the economy, generating further jobs. In the handful of countries that depend heavily on the tobacco industry, there would be net job losses, but these would occur slowly and adjustment, though possibly difficult, would certainly be no worse than for many other declining markets. The report also assesses the likely impact of the various control interventions--tax increases, non-price measures and wider access to nicotine replacement therapy--on welfare. It concludes that tax increases would cause some welfare loss to consumers but that this loss would be offset by any .reasonable willingness to pay for averted premature deaths and averted disability. '~ The report then examines the likely cost-effectiveness of specific interventions, both in terms of deaths averted for a given amount spent and in terms of the amount of premature death and disability that can be avoided ovc~, time for each dollar spent. It finds that raising taxes would be highly cost-effective, costing just cents for each year of healthy life saved in low-income and middle-income countries, less than many basic health interventions that arc assumed to be essential. Nicotine replacement therapy would also be cost-effective in most settings, costing between $15 and $450 per year of healthy life saved in low- and middle-income countries. The package of non-price measures, although slightly less cost-effective than the other demand-reducing measures, would still represent an excellent "buy" compared with many health interventions. The findings are consistent with earlier studies that suggest that tobacco control is highly cost-effective as part of a basic health package in low-income and middle-income countries. [more specific numbers pending]. The report ends by making specific recommendations to governments and international organizations. These are shown in full on pages xx and xx. In sum, the recommendations to governments include raising taxes by at ]east l 0 per cent per year for l 0 years, imposing comprehensive bans on advertising and promotion, restricting smoking in public places and workplaces, and widening access to nicotine replacement therapy. The report stresses that effective efforts by governments are likely to involve using all the measures suggested, not just some, although the mix will vary from country to country depending on the type of economy. Careful political mapping and stakeholder analysis will be needed for each government to assess its needs and the optimal policy mix. For international organizations, the report recommends a careful review of existing programmes and policies, to ensure that tobacco control is being given due prominence and applied in the most efficient ways possible. It recommends that agencies with a remit to sponsor and/or conduct research do so to provide more information about the causes Draft: Not for circulation or citation 14 321422974

and consequences of tobacco use, including its costs, and the cost-effectiveness of interventions at local level. Finally, the international agencies should address tobacco control issues that cross borders and facilitate international agreements. These might include controls on smuggling, discussions on tax harmonization to reduce the incentives for smuggling between neighbourlng states with large differentials in tobacco tax, and means to prevent the industry from avoiding bans on advertising and promotion through the global communications media. The threat posed by tobacco to global health is unprecedented, but equally so the potential for reducing tobacco-related mortality with cost.effective economic policies. This report shows the scale of what might be achieved: modest action could ensure substantial health gains for the 21 st century. Draft: Not for circulation or citation 321422975

3. Global trends in tobacco use 3.1 Rising consumption in low-income and middle-income countries By the mid-1990s, there were approximately 1.1 billion smokers-one in three of the adult population worldwide, consuming about 6 million million cigarettes per year. More than 80 per cent of smokers today live in low-income and middle-income countries, and the prevalence of smoking in these countries rose on average by 3.4 per cent per year between 1970 and 1990. The nature of tobacco use is also changing, from the traditibnal habit of chewing to an increasing reliance on smoking manufactured cigarettes. The upward trend, expressed in ter/ns of cigarette consumption per head, is shown in Figure 3.1. Because of growth in the adult population worldwide, the number of smokers is expected to rise in absolute terms in coming decades to, leach 1.5 billion within the next two decades. Figure 3. I Trends in per capita adult (15 years old and above) cigarette consumption in developed and developing countries 3000 2500 2000 1500 1000 500 0 1970-72 1980-82 1990-92 Source: WHO 1997Year • Develo;~ing J D Wo~ J In contrast, smoking is becoming less common in most high-income countries such as the USA, Canada, the northern European nations and Australasia. In the USA, for example, the prevalence of smoking fell from 40 per cent at its peak in 1964 to 25 per cent by the mid-1990s. In the United Kingdom, cigarette sales have fallen by almost half over the past three decades. The overall decline in smoking in these countries' populations is not a uniform trend, however. Smoking prevalence has risen among certain groups, such as teenagers, since the early 1990s (see, for example, Institute of Medicine 1998. 3.2 The impact of trade liberalization International trade in tobacco, like other goods, was until recently subject to a large number of tariff and non-tariff barriers. Countries implement such barriers to protect their own tobacco industries, and to gain foreign exchange reserves through the domestic production of goods that would otherwise be imported. Draft: Not for circulation or cRation 16 321422976

However, in the past two decades, and particularly in the late 1950s and 1990s, trade in goods and services between nations has been increasingly liberalized as a result of a number of bilateral and multilateral agreements, such as the 1974 Trade Act and its subsequent amendments, and the General Agreement on Tariffs and Trade (GATT). The GATT (1994) agreement .sets limits on tariffand non-tariffbarriers to trade and will eventually eliminate such tariffs. It requires all trading partners to be treated equally and without discrimination, and requires that products be treated equally whether they are produced domestically or abroad. Tobacco is no exception. The removal of trade barriers tends to tends to introduce greater competition that results in lower prices, greater advertising and promotion, and changes in other activities that stimulate demand. Several studies show that people on low incomes are in general more responsive to price changes than people on higher incomes, and therefore countries where, on average, incomes are lowzhay be expected to see the biggest increases in cigarette purchasing wherever prices fall. In addition, companies newly entering each market will spend more than before on advertising and promotion. Studies of a number of Asian countries indi~te that low-income and middle-income countries have seen large and significant inc'-eases in cigarette consumption through the 1980s as a result of being forced to open their markets to US imports. High-income countries, in contrast, have seen little impact. A study of 10 Asian countries, using pooled time-series data, estimated the impact of trade liberalization in the region in response to section 301 of the 1974 Trade Act (Chaloupka and Laixuthai, 1996). It found that in four countries that were forced to open their markets to US cigarette imports during the 1980s- -.lapan, Taiwan, South Korea and Thailand--cigarette consumption per capita was almost 10 per cent higher on average in 1991 than it would have been if the markets had remained dosed. Simulations suggest that the same trends would emerge in other Asian countries if these were opened up to US imports. These suggest that in 1991, average per- capita cigarette consumption would be about 7.5 per cent higher than it would have been had the markets remained closed An econometric model of trade developed for this study (Chaloupka et al, 1998) suggests that a doubling of total imports as percentage of GDP would result in per-pack consumption rising by 13 per cent in low-income countries, 6 per cent in middle-income countries, but less than 1 percent in high-income countries. Draft: Not for circulation or citation 17 321422977

Figure 3.2 cigarette (in million) World tobacco consumption and US exports WORLD CONSUMPTION 6°000°000 5,000,000 4.000,000 3,000,000 2,000,000 1,000,000 0 years cigarette (in million} 250.000 200,000 150,000 100.000 50,000 US exports 3.3 Tobacco use is increasingly associated with poverty As the popularity of smoking shifts from the rich countries to the middle-income and low-income countries, it is also shifting within the rich countries to become increasingly concentrated among the poor. This trend represents a reversal of the pattern found earlier in the smoking epidemic. In the early and middle decades of the century, smokers were more likely to be wealthy than poor. But in the past three decades, more affluent households have increasingly aband6ned tobacco. In Nor~vay, for example, the percentage of men with high incomes who smoked fell from 75 per cent in 1955 to 28 per cent in 1990. Among men with low incomes, the proportion who smoked declined much less, from 60 per cent in 1955 to 48 per cent in 1990, and with little overall reduction since 1970 (Figure 3.3). Similar trends have emerged in many industrialized countries. In the 321422978

USA, for example, more than one-third of people living below the poverty level in 1995 smoked, compared with one-quarter of those at or above the poverty level. Figure 3.3 Tobacco Use is Most Prevalent in the Poor, Smoking Trends in Norwegian Males by Income group, 1955-90 75% ~4:5% 3:5% 2.5% 1935 1960 1965 1970 1975 1980 14|$ 199(I Although smokers in the middle-income countries and low-income countries have tended to be more affluent than non-smokers, there is now growing evidence that this pattern is changing to resemble the situation in high-income countries. In recent years, studies in China, Brazil and India, together home to nearly two-fifth of the world's smokers, have found that tobacco use is more common among people with low incomes than those with high incomes. Similar patterns have been found in South Africa, Vietnam and a number of Central American countries. As we shall see in Chapter 5, education appears to account for much of these differences, although income effects persist in some studies (Novotny, 1995). 3.4 Patterns of tobacco use by region, gender and age Data on the number of smokers in each region have been compiled by the World Health Organization using more than 80 separate studies. For the purpose of this report, these data have been used to estimate the prevalence of smoking in each of the seven World Ba.,~ country groupings.O~ The data are shown in Table 3.1. {~> These groupings are shown in Annex I. In sum they are as follows: (1) East Asia and the Pacific; (2)Eastern Europe and Central Asia (a group that includes most of the former socialist economies); (3) the Middle East and North Africa; (4) Latin America and the Caribbean; (5) South Asia; (6) Sub-Saharan Africa; and (7) the high-income countries, such as North American. Australasian, northern European nations and Japan. Draft: Not for circulation or citation ! 9 321422979

Thirty-six per cent of all smokers aged 15 and over worldwide in 1995 lived in East Asia and the Pacific nations. The high-income countries accounted for a further 18 per cent. Eastern Europe and Central Asia had the third highest prevalence of smoking with 13 per cent.of the total. Together, therefore, these tba'ee regions account for two-thirds of all adult smokers. South Asia accounts for approximately half of the remainder. In this region, hand-rolled cigarettes or bidis are as currently about as popular as manufactured cigarettes. Table 3.1 Estimated smoking prevalence (by gender) and number of smokers, 15 years of age and over, b~v World Bank region, 1995 World Bank Smokinl~ Prevalence Total smokers Region Males Females Overall (millions) (% of all smokers) East Asia and Pacific 59% 7% 33% 415 Europe and Central Asia 59% 26% 41% 148 Latin America and 40% ' ,. 21% 30% 95 Caribbean ~ Middle East and North 44% 5% 25% " 40 Africa South Asia (cigarettes) 20% 0.83% 11% 86 South Asia (bidis) 20% 3% 12% 96 Sub-Saharan Africa 33% 10% 21% 67 36% 13% 8% 3% 7% 8% 6% Low & Middle Income 49% 10% 30% 948 82% High Income 39% 22% 30% 209 18% World 47% 12% 30% 1,157 100% Source: Ranson, K. et al. Male-female differences in smoking prevalence In many middle-income countries and some low-income countries, more than half of all men are smokers. In Indonesia in 1995, eight out of every ten adult males smoked, and in Cuba, Colombia, Bangladesh and the Philippines, the proportion was seven out of ten. In China, where average cigarette consumption per man per day has climbed eleven-fold since the early 1950s, 63 per cent of men smoked in 1996. Overall in East Asia (including the Pacific nations), 59 per cent of men smoked in 1995--a similar proportion to that . found in Eastern Europe and Central Asia. This compares with 39 per cent of men in the high-income countries. Worldwide, fewer women than men smoke. The prevalence of smoking among all adult males is 47 per cent compared with 12 per cent in women. However, patterns of smoking among women differ sharply between regions and do not predictably correspond to male smoking rates. More women smoke in Eastern Europe and Central Asia (mainly in the former socialist economies of Europe) than in any other region: 26 per cent in 1995, Draft: Not for circulation or citation 20 321422980

compared with 22 per cent in the high-income countries. However, in East Asia in the same year, only 7 per cent of women smoked. And in Latin America and the Caribbean, 21 per cent of women smoked compared with 40 per cent of men. Thus while there are eight male smokers for every female smoker in Fast Asia, the ratio in Latin America and the Caribbean is' only two to one. Even within regions, there are sharp variations. For example within East Asia, only 7 per cent of Chinese women smoke whereas in Indonesia the proportion is much higher. Women's smoking habits have been studied less extensively than men's. While Western countries have experienced increases in the uptake of smoking among young women, recent data from China indicate, unexpectedly, that the prevalence of smoking among younger women there is declining. There are also fewer data on the relationship between income and smoking in women. , ~. Age and smoking The habit of smoking is usually acquired ih outh. In the USA, for example, approximately 80 per cent of smokers begin before the age of 18 .(CDC 1997). In 1996, 10 per cent of children aged 12-13 were already smoking cigarettes daily. By age 16-17 the proportion smoking cigarettes daily had more than doubled to 22 per cent (Institute of Medicine 1998). In middle-income and low-income countries for which data are available, the habit is usually acquired a few years later. In China, for example, two-thirds of men smoke by the age of 25 (Liu et al 1998). In India, 70 percent of cigarette smokers start before age 25 (Gupta, 1995). WHO data analyzed for this report show that the prevalence of smoking in males worldwide is already 33 per cent by the age of 19. After this age, fewer men take up smoking, with prevalence reaching a peak at 58 per cent in the age group 40-49. Among women, worldwide, the uptake of smoking is most rapid in the twenties, peaking at 15 per cent in the age group 30-39. Quitting is less common in developing countries In many high-income countries and in some middle-income countries, millions of smokers have quit, often before the onset of disease. This is best measured by the prevalence of"ex-smokers". For example, in Sweden, the prevalence of male ex-smokers doubled from I963 to 1994 to reach 41 per cent of all men (Table 3.2). In contrast, ex smokers in China and in India are relatively rare, and prevalence figures are generally below 10 per cent. This is partly due to the fact that smoking has become a widespread habit more recently in these countries than in the high-income countries, and partly due to a lack of information and tobacco control policies that can help people to quit. 321422981

Table 3.2. Percentage of males that were forn~r smokers in selected countries, ranked by per capita GDP over lime periods Country High Income United States Sweden Australia Italy Middle Income Hungary Poland Low Income China India Vietnam Time 1 [ Time 2 20 (1965) . 30 (1991) 20 (1963) 41 (1994) 28 (1986) 32 (1992) 22 (1990) 28 (1995) 15 (1986) 14 (1994) 18.(1974) 21 (1997) 2(1993) 5 (I 992-1994) 10(1997) Dra/rt: Not for circulation or citation 22 321422982

4. The Health Consequences of Smoking The damage caused by tobacco to health has been extensively documented elsewhere. This report does not seek to repeat what is known in full detail, but sim#y to summarise the evidence. This section is divided into two parts: first, a brief summary of the evidence that nicotine, a constituent of tobacco, is addictive; and second, a brief discussion of the disease burden attributable to tobacco. 4.1 The addictive nature of tobacco use Tobacco contains nicotine, a substance that meets the key criteria for addiction or dependence used by major medical international organizations (Kessler, 1995). These include regular and compulsive use (despite a desire, or repeated attempts, to quit), psychoactive effects produced by the action of the substance on the brain, and behaviour motivated by the "reinforcing" effects of the psychoactive substance. Debate about the nature of addiction and drug dependence has continued for decades (Kalant, 1989). However, there is general consensus with the, conclusion of the US Surgeon General's report for 1988, which found that "Nicotine is the drug in tobacco that causes addiction." Internal documents of the tobacco industry show that its researchers and executives are aware of nicotine's addictive properties, and have bccn for decades (Kessler, [995). Cigarettes, unlike chewed tobacco, enable the user to obtain nicotine directly into the lungs rather than through blood or digestive systems. Thus, nicotine from a cigarette reaches the brain in about I0 seconds, twice as fast as if it had been given intravenously. Cigarette smoking enables the user to regulate the dose of nicotine puff by puff. Compared with other addictive drugs, nicotine dependence is achieved relatively rapidly. Most new smokers gradually increase their cigarette consumption to reach a stable level, which in 90 per cent of cases is at least five cigarettes per day. Like other addictive substances such as heroin (Robins 1993), it is undoubtedly possible to abstain permanently from tobacco. Half of smokers currently alive in the US have quit, usually without clinical assistance. However, most of the remaining half want to quit but find it impossible (US Department of Health and Human Services, 1990). Since, as we have seen in the preceding section, most smokers start young, nicotine addiction is usually established at an early age. The plasma concentrations of cotinine, a metabolite of nicotine, in regular smokers aged between I l and 16 have been found to be similar to the levels found in adults who smoke regularly for any given daily number of cigarettes smoked. Once individuals realist that they are dependent on cigarettes, most appear to regret the fact: in the US, seven out of ten young people who smoke report that they regret having started. Between half and three-quarters of young smokers there report having tried to quit at least once and failed (Poulin, 1996; Kessler, 1995). Industry- sponsored studies indicate that teenagers as young as 16 or 17 already regret their reliance on cigarettes (Kwechansky Marketing Research 1977). 321422983

Among high school seniors in the USA who smoke less than one cigarette per day, 85 per cent believe that they will not be smoking five years later. However, 42 per cent of these hopeful students actually arc smoking at least one cigarette per day five years later (Surgeon General 1994). 4.2 Long-term disease caused by tobacco In 1998, tobacco was estimated to cause almost 4 million deaths worldwide. Within the next three decades, when the children of today become adults, smoking is expected to cause about 10 million deaths a year. This is more than any other single disease, injury, or risk factor, and more than the current combined total death toll from pneumonia, diarrheal disease, tuberculosis and malaria. So~e 70 per cent of these deaths will be in developing countries, with China alone accounting for about 2 million deaths per year. Projections suggest that by the earlier date of 2020, tobacco alone will account for approximately 8.4 million deaths worldwide per year--one in eight deaths--compared with 2.3 million deaths from tuberculosis and 2.5 million deaths frbm pneumonia. Put differently, about half a billion people alive today will be killed by tobacco--half of them during their most productive decades of middle life. Figure 4.1, Percentage of all deaths attributable to tobacco by regiont, estimates for 1990 and projections for 2020 25% . r 23% 20% o 15% 15% - 14~, 13% 16% 10% - 9% 9% 12% 12% 3% 0% ,~ . z~ o 5 '" -- "~" ~ o Source: Murray and Lopez 1996 Note. For an explanation of the regional terminology used in this figure, see foomote (I) below. 4.2.1 Damage to smokers thentselves ~ EME: Established Market Economies, FSE: Formerly Socialist Economies of Europe, OAI: Other Asia and Islands, SSA: Sub-Saharan Africa, LAC: Latin America and the Caribbean, MEC: Middle Eastern Crescent 321422984

There is no serious dispute that tobacco causes chronic disease and premature death. However, many individuals may underestimate the scale of the hazard. In particular, the strength of the relationship between tobacco and death is obscured by the long delay between the uptake of smoking and the onset of disease, which may be as much as 30 years. As a result, populations in which smoking has become a mass habit only recently have yet to witness its impact on health. Figure 4. 2 Time lag between tobacco consumption and disease, Brazil Source: Worm Bank, 1989 4. 2.1.1 The evidence from industrialized countries Most of the data that demonstrate tobacco's damaging effects on health come from the developed countries, where smoking prevalence was at its height some three to four decades ago. In the USA, for example, tobacco consumption per capita increased by 44 per cent between 1920 and 1950, but lung cancer rates showed their most marked increase after 1950, by which time tobacco use was already beginning to decline. Thus, in any given age group, lung cancer rates in the USA in the early 1980s were eleven times higher than in the early 1930s, compared with just three times higher in the early 1950s. Data such as these are not yet available from most developing nations because most of their populations have only recently taken up smoking in large numbers. Long-term studies from the USA and Britain indicate that, for middle-aged smokers, defined as those aged 35-69, death rates are about three times greater than for non- smokers. About half of all smokers are eventually killed by their habit. The causes of death in smokers are varied, but include cardiovascular disease--most frequently ischemic heart disease--stroke and obstructive respiratory disease as well as lung cancer and other cancers. The strength of the association between smoking and each of these diseases Draft: Not for circulation or citation 25 321422985

differs, but for lung cancer, where the association is strongest, the excess risk to smokers is about twenty times that of non-smokers. In order to estimate the proportion of deaths that can be attributed to smoking in Western countries, researchers have developed a method that compares the rates of lung cancer in any given country's smokers with the rates of lung cancer in non-smokers in the US. This comparison gives a guide to the extent to which the population is being damaged by smoking. The excess number of deaths from lung cancer in smokers is then used as a guide to indicate what fraction of the deaths from each other cause, such as ischemic heart disease, can be attributed to tobacco. Using this method, researchers estimate that tobacco is already responsible for 28 per cent of all deaths among middle-aged individuals in developed countries (Peto et al, 1994). The damage caused by tobacco appears to be related to the length of exposure. Evidence from countries where the tobacco epidemic is mature suggest that the longer the duration of smoking, the greater the risk of disease. A person's risk of developing lung cancer is more closely related to the duration ofsmoking than by the number of cigarettes smoked daily. Put differently, a threefold increase in the duration of smoking is associated with a 100-fold risk of lung cancer, whereas a threefold increase in the number of cigarettes smoked each day, is associated with only a threefold risk of lung cancer. By the same token, there are significant health benefits to smokers who quit. The benefits are greatest in young middle age, defined as 35 to 44 years, but substantial even at ages 45 to 54. Data from industrialized countries suggest that, ten years after they stop smoking, former smokers face either a small or an insignificant excess risk of heart disease compared with non-smokers. Lung cancer rates in former smokers remain higher than those of non-smokers, even after 15 years, but the risk is much lower than for smokers. 4.2.1.2 Tobacco accounts for at least half of the health disadvantage of the poor As tobacco use is associated with poverty, so arc its damaging effects on health. Analyses commissioned for this study show this relationship clearly. In developed countries for which long-term data are available, tobacco is responsible for half to two-thirds of the increased risk of premature death experienced by men in the poorest income groups and those with the lowest levels of education. The available evidence suggests a similar, though less pronounced, relationship in women, although the shorter nature of the smoking epidemic in women means that more data are needed. In England and Wales in 1991, a man of 35 with a professional or managerial occupation had a 22 per cent risk of dying before he reached the age of 70. For a man of the same age in unskilled work, the risk was 48 per cent--more than double that for his richer peer. Tobacco appears to be responsible for a very large proportion of the unskilled man's increased risk: in the highest income group, the risk of dying from a cause that can be attributed to tobacco, such as ischemic heart disease or lung cancer, is only 5 per cent. For Draft: Not/'or circulation or t'itation 321422986

men with low earnings, the corresponding risk is 23 per cent, almost five times greater (Figure 4.3). Figure 4.3, The contribution of smoking to the risk of premature death among males at ages 35-69, by social class, England and Wales 1996 (to be replaced by 1991figure) 5O% 45% 40% 35% 30% 25% 20% 15% -1 10% ~ 5% -' 0% [~ Z0% I1% 1/11 IIIN Z0% . IIIM/IV V IIS1 Other causes IBAttdbut~l to SMOKING but would haw died ! anyway at ages 35-69 i~'lAttributed to SMOKING The impact of tobacco on the health of different income groups can best be shown by analysing how their health might change if it were possible to eliminate smoking. The effects on survival would be dramatic for all groups, but most striking in those on low incomes. Without tobacco, the risk of dying before age 70 for a 35-year-old English male falls to 17 per cent for the highest earners and 25 per cent for the lowest. So, as well as reducing the overall number of premature deaths, avoiding tobacco would virtually halve the poor man's risk and narrow the health gap between poor and rich to a risk ratio of 1.5 to 1. A similar pattern emerges from Canada. In 1991, the richest urban men faced a one-in- five risk of dying in middle age, compared with a risk of more than one-third for their peers in the lowest income group (Figure 4.4). In the absence of tobacco, the richest group's risk of premature death would fall to about one in seven and the poorest group's risk to one in five. Figure 4. 4 The contribution of smoking to the risk of premature death among males at ages 35-69, by income quintile, Canada 1996 Draft: Not for circu/ation or citation 27 321422987

40% 35% 1 3O% 25% 20% , [----] . 15%~ 16~ 10% 0% 7~ IE3 Other causes anyway at ages 35-69 E]Attributed to SMOKING Comparable results emerge from studies ~ which men are classified by their level of education, rather than their income or occ@ation. Educational level broadly reflects socio-cconomic status, suggesting that the relationship between poverty and tobacco- related disease is a robust one, at least for those natio~ where evidence is available. In Poland, the overall risk of death for men aged between 35 and 69 in 1996 was 26 per cent for those with a university education, and 52 per cent for men with only a primary education. Tobacco-attributable causes account for 5 per cent of the risk in the most educated group and 23 per cent in the least educated group--a relative risk of 4.6, dose to that found between social classes I and V in England and Wales. If tobacco could be eliminated, the risk of premature death for the most educated group would fall to 21 per cent, while the risks for the least educated group would fall to 29 per cent (Figure 4.5). Figure 4. 5 The contribution of smoking to the risk of premature death among males at ages 35-69, by educational level, Poland 1996 60% • 50% . 4O% - 30% . 20% - 121%1 10% • I 1~ 0% ...... Higher Secondary I| P~mary 1-1 Other causes BAttdbuted to SMOKING but would have died anyway at ages 35-69 I~Attributed to SMOKING Draft: Not for circulation or citation 28 321422988

In the USA, the pattern is similar to that found in Poland. Highly educated men in 1996 faced a 20-per=cent chance of dying between the ages of 35 and 69. In men with less than 12 years of education the risk was 37 per cent. Tobacco accounts for 4 per cent ofthe risk of dying in the most highly educated men and IS per cent of that risk in the least educated men (Figure 4.6). Figure 4. 6 The contribution of smoking to the risk of premature death among males at ages 35-69, by years of education, USA 1996 4O% 35% t 30% -~ 25% ~ 20% • 15% , 10% • 0%, 22% 15% ."13 ,~, ? 12%! 0% ) ........ .~ , ~ <12 12 13+ El Other causes • Attributed to SMOKING but would ha~e died anyway at ages 35-69 [3Attributed to SMOKING In recent years, the health gap between rich and poor households has grown in many developed countries. For example, in Canada in 1971, the risk that the poorest would die prematurely was 46 per cent, compared with 32 per cent for the richest, a ratio of 1.4. Twenty years later, the risk for the poorest was 35 per cent, and for the richest, just 20 per cent, widening the ratio to 1.8. Tobacco appears to have contributed substantially to this widening gap: put differently, it appears that the gap would have narrowed substantially in the absence of tobacco. In 1971 in Canadian men, the risk of premature death due to causes specifically attributable to tobacco was 9 per cent for the richest and 17 per cent for the poorest. By 1991, the tobacco-attributable share of the risk had fallen to 6 per cent for the richest, but by a much smaller fraction in the poorest, to 15 per cent. Thus, while tobacco accounted for almost half of the risk for rich men in 1971, it accounted for only just over one-quarter of the risk twenty years later. By contrast, for poor men, tobacco remained responsible for about one-third of the risk throughout the period. 4.5. I. 3 The impact of smoking in developing countries Draft: Not for circulation or citation 29 321422989

Because the smoking epidemic is at an earlier stage in developing countries, its impact on health is more difficult to assess. However, some data have emerged from China, which has 27 per cent of the world's smokers. A new analysis of more than one million deaths in some 100 rural and urban districts .suggests that by 1990, tobacco was already responsible for about 12 per cent of adult male deaths and 3 per cent of adult female deaths. Since there are 300 million males aged 29 or under in China, and two-thirds of young men smoke, assuming that current consumption trends persist, about 100 million young males now alive will eventually be killed because of their habit. Half of these men will die in middle age, losing up to 25 years of life (Liu et al 1998). Importantly, the pattern of diseases that kill smokers may bc different in developing countries from the industrialized countries. The data from China suggest that deaths from ischemic heart disease make up amu, ch smaller proportion of the total deaths among smokers than in the West, while rcsj~iratory diseases and cancers account for most of the deaths. Strikingly, a significant minority arc duc to tuberculosis. Other differences may emerge in other populations; for instance, in South Asia, the pattern may bc affected by a high underlying prevalence of cardiovascdlar disease. So far, there have been too few studies to reach clear conclusions on these qhestions. 4.2.2 Damage Io non-smokers Smoking affects not only the health of smokers but also the health of those around them. Non-smokers who spend considerable time in the company of smokers, such as spouses, children and work colleagues, are at greater risk than those whose exposure to tobacco smoke is only occasional. Passive smoking is associated with a small increased risk of cardiovascular disease and of lung cancer. Mothers who smoke during pregnancy risk the survival and health of their child. Maternal smoking is linked with a higher risk of spontaneous abortion and other complications of pregnancy that threaten the fctus's survival. Newborn babies are up to 35 per cent more likely to die if their mothers smoke than if their mothers arc non-smokers. Babies born to smoking mothers arc smaller, have reduced growth, and may suffer behavioural and cognitive problems. Sudden Infant Death Syndrome is also significantly more common in babies born to smokers than those born to non-smokers. Recent research has shown that a carcinogen found only in tobacco smoke is present in the urine of newborn babies. Despite this body of evidence, however, many women continue to smoke during pregnancy, possibly because of inadequately understood information about the hazards. Even in the USA, where information about the hazards is widely available, the proportion remains as high as 14 per cent. Draft: Not for circulation or citation 321422990

5. Do smokers know their risks? Individuals learn about the hazards of smoking through various information sources, including: warnings on cigarette packs and cigarcttc advertisements; official reports that receive media coverage; specific anti-smoking advertising campaigns (or "countcradvcrtising") by governments and others; and school health education programmcs. This section is concerned with the extent to which such information affects consumption, and summarizes what is known about the way in which different social groups respond to that inforn~ation. 5.1 Awareness of the health hazards is a deterrent to tobacco use Studies have shown that, in general¢ the more people know about the effects of smoking, the less likely they arc to smoke (Kenkcl et al, 1998). Most of the studies fall into one of two categories--those that monitor trends in smoking bchaviour over time and those that examine the smoking habits of different grqups within society at any one time. We shall briefly summarize these studies. :~ Time-based studies The first category of studies consistently shows that, whenever there have been major "information shocks" about the health effects of smoking, such as the publication of widely-publicised official reports on the subject, consumer demand for cigarettes has fallen. The impact of such information shocks has been studied in diverse countries including the US, UK, Switzerland, Finland, Greece, Turkey and South Africa. Most studies find that demand falls by between 5 per cent and I0 per cent with each shock, more when several shocks come close together. In some cases, the effects are only temporary, and, after a period, demand starts to creep back up. In the US, for example, a study based on data from 1930 to 1978 suggests that three so- called "information shocks" between them cut consumer demand by about 30 per cent over the period. These included the publication of a report by the American Cancer Society in 1953 and the Surgeon-General's report in 1964. In more recent decades, studies from several established market economies appear to show a more permanent decline in consumption linked to information. For example, a study based on data between 1960 and 1994 in the USA suggests that parents' consumption of tobacco has decreased much more rapidly than that of single adults living without children, as parents have become increasingly aware of the effects of second-hand smoke on their children's health. The more highly educated adults have abandoned tobacco much faster than their less highly-educated peers: in the USA during the 1970s and 1980s, for example, the prevalence of smoking declined nine times as rapidly in the most-educated group than in the least-educated group. These findings are consistent with earlier research that suggests that, wherever data are available for a population, the more highly educated respond more 321422991

rapidly than the less educated to new information affecting their health. This pattern was seen in the USA at the turn of the century, when richer households--whose access to education was greater than poorer households- were first to respond to recent discoveries linking microbes with infectious diseases by adopting healthy behaviours such as handwashing and sterilising feeds for babies. Clearly, such studies are fraught with difficulty because they cannot control for other factors that may interfere with behavioural trends. Estimates of the extent to which tobacco consumption is affected by knowledge vary widely (Warner 1989; Kenkel 1991). However, it appears that there is a clear association between increased information and decreased demand. The long-term downward trend in smoking prevalence in most industrialized countries over the past three decades has coincided with a long-term upward trend in people's knowledge about the harmful effects of tobacco. For example, in the 1950s, in the US, only 45 per dent of adults identified smoking as a cause of lung cancer. By 1990, 95 per cent did so. Knowledge about the links between smoking and emphysema and heart disease has also increased. Over the same period, smoking prevalence has fallen steadily, as Chapter 3 showed. Cross-sectional studies The second category of studies examines the smoking habit of different groups in society at any given time. Such "cross-sectional" studies tend to find that the least educated groups are most likely to smoke regularly, while the most highly-educated people are least likely to do so. This is the ease not just in the US and other industrialized countries where the tobacco epidemic is well advanced (Kenkel, 1991; Viscusi (1991), but also outside the West, in countries as diverse as Taiwan (Hsieh et al, 1997) and Nigeria (Obet [ck], 1990), where almost 30 per cent of respondents with only primary education were regular smokers, compared with 17 per cent of individuals with higher education. These findings reinforce the basic conclusion that the more individuals know about the hazards of smoking, the less likely they are to smoke. Nevertheless, it is clear that tobacco consumption continues to be a widespread habit, both worldwide and in the industrialized nations where health warnings about tobacco are ubiquitous. The following section discusses some specific factors or "information failures" that contribute to the continued popularity of smoking despite widespread knowledge of its risks. 5.2 Factors that complicate the relationship bet~veen knowledge and demand Inadequate knowledge For educated adults in the industrialized world, it may be difficult to accept that there is still widespread ignorance about the health hazards of smoking elsewhere. However. evidence suggests that a significant proportion of smokers in low-income and middle- income countries simply do not know that their behaviour is risky. In China, for example, Draft: Not for circulation or citation 32 321422992

where one-third of the world's cigarettes are smoked, the 1996 National Prevalence Survey on Smoking revealed that 61 per cent of smokers surveyed believed that their habit did them "little harm". Fewer than one in four smokers believed that the habit did them "serious harm". Most adults thought that smoking could increase their risk of developing bronchitis, but did not recognize any link between tobacco and other discases. The researchers who conducted this study concluded that" most of the general population in China lacks a thorough understanding of the health risks from smoking" (China Science and Technology Press, 1996). In South Africa, a survey published in 1996 found somewhat greater overall knowledge that smoking harms health. Among the population as a whole, 87 per cent believed the health effects of smoking to be "serious" or "very serious". However, when questioned about the specific diseases caused b,y tobacco, many respondents revealed their lack of knowledge. Among black respondents, only about one-fifth realized that smoking is associated with heart disease, and little more than half knew that smoking is associated with cancer. Among so-called coloured respondents, the proportions were 40 per cent and 65 per cent respectively. For the populatiSti as a whole, fewer than one-third knew of the link with heart disease and little more than.A half knew of the link with cancer. Failure to perceive the magnitude of the tobacco risk relative to others Even where individuals have access to information about the hazards of tobacco, they may fail to appreciate the scale of those hazards relative to other health risks. In Poland, for example, only 27 per cent of people surveyed in 1995 thought that smoking was among "'the most important factors influencing human health", compared with 62 per cent who listed "the environment" and 45 per cent "stress and a hectic lifestyle" (Table 5.1 ). Similarly, in the USA, for example, a 1993 Harris poll asked people to rank activities that "help people in general to live a long and healthy life". Avoidance of smoking was ranked tenth, far behind good air quality, good water quality, domestic fire detectors and other factors with a statistically marginal effect on survival. Interestingly, smoking is the leading causes of fires in the US. Table 5.1 Adults 'perception of "the most important factors influencing human health ", Poland, 1995 Percentage Factor 62 Environment 55 Dietary habits 45 Stress, hectic lifestyle 27 Tobacco smoking 25 Genetics 24 Sport, phE.sical activity' Source: Zatonski, 1996 Draft: Not for circulation or citation 321422993

Some economists argue that smokers do indeed know their risks. For example, Viscusi (1990, 1991, 1992, 1998) concludes from analyses of national survey data from the 1980s and I990s that smokers actually overestimate their risks. In these studies, perceived risk of lung cancer is measured on the basis of answers to the question "Among 100 cigarette smokers, how many of them do you think will get lung cancer because they smoke?". Similarly, smokers' perceptions of their total mortality risks are judged on the basis of answers to the question: "Out of every 100 cigarette smokers, how many of them do you think will die from lung cancer, heart disease, fin'oat cancer, or any other illness because they smoke cigarettes?" The perceived risks are calculated by dividing the answers by 100. The resulting figures arc much greater than the reasonable scientific risk estimates, based on available data at the time of the surveys. However, other economists question, these conclusions, arguing that the measures of risk perception arc inadequate (Kcnkel '1999). Some recent studies comparing actual survival probabilities of middle-age smokers in the US find that heavy smokers significantly underestimate their probability of survival to age 75 (Schoenbaum 1997). Schocnbaum's study is based on data used widely by health and retirement economists, the Health and Retirement Survey. Respondents were askett to assign a value between 0 and 10 on the odds that they would live to be 75 or more. The resulting figures wcrc compared with actual survival probabilities. While moderate and light smokers have reasonably accurate perceptions of their risks as assessed by this measures, heavy smokers overestimate their survival probability, and hence underestimate the risks of smoking. Those individuals who had a low expectation of surviving to 75 also tended to overestimate all health risks, not just smoking risks. It is important to note that the survival probabilities on which these studies were based may themselves underestimate the risks of smoking, as they were based on actuarial data from an earlier stage in the tobacco epidemic, which suggested that the risk of smoking was only half as great as it has since been revealed to be. Failure to internalize the risks at personal level Even among those whose perception of the statistical risks of tobacco use is reasonably accurate, risks may not be internalized. In other words, many smokers know that cigarettes cause disease but do not believe that they, personally, will become sick. For example, surveys in the USA have suggested that some 90 per cent of smokers believe the habit is harmful to health but only 75 per cent are concerned about the effects on their own health (Kenkel 1999). This is clearly a critical factor in determining whether individuals start to smoke in the first place or in determining whether existing smokers try to quit. To date, however, research on the process by which knowledge is absorbed by individuals, and either acted upon or not acted upon, has received little attention within the field of economics. The focus within this discipline has been more on econometric studies of the ways in which consumer information about tobacco shifts the demand curve. Clearly, such studies can provide only a partial picture. Youth and the heavy discount placed on the future 34 321422994

Smokers usually begin smoking during their teens, when they arc relatively uninterested in their long-term health and where rebellion plays an important role in behaviour. Widespread studies have shown that teenagers and young adults discount the future much more heavily than older adults: that is, they place a much lower value on years to bc lived 30 years ahead than the year being lived now (Kenkcl, 1999). Since the health damage caused by tobacco may take decades to emerge, most young people arc therefore unmoved by being told that they will one day develop disease. By the time they develop a more realistic perception of the harm they arc risking the.msclvcs thxough smoking, they are likely to be addicted to nicotine. 5.3 Predicting ,the impact of information in developing countries Knowledge about the health impact of smoking is less widespread and less detailed in developing countries than in the indus~xialized countries. The implication is that a significant increase in the availability of a.ccurate information in developing countries would be expected to lead to a significant decrease in global tobacco consumption. In t Chapter 7, where we turn to a discussion oPmcasurcs to reduce the demand for tobacco, we shall discuss the evidence for the effectiveness of each different type of information mcasuzc in curing cigaxctte consumption, based mainly on the experience of industrialized countries. It is difficult, with the present levels of knowledge, to predict the extent to which increased information would affect consumption in practice in the lower- income countries. Since educational levels and literacy are generally lower in developing countries, the effect of providing that information may--conceivably--be less marked than in generally well educated populations. However, there are important caveats to this statement: evidence from other fields of research suggests that people in low-income countries have responded more rapidly to new health information than people in higher- income countries. Note, for instance, that cigarette consumption is now falling in Thailand, despite the opening of the market to many imported brands, following the adoption of aggressive warning labels and advertising controls by the government. Bibliographic note. The section on tobacco consumption ~rends draws on diverse sources including Gajalakshmi CK et al. 1998: Ranson K et al, 1998; and Bobak, Mct al, 1995. The section on trade liberalization draws on Chaloupka FJ et al, 1998. Other sources include Institute of Medicine/National Research Council |998; and Pcto, R el al i 994. The section on addiction draws on Kalant H. 1989; Kessler D. 1995; Report of the Surgeon General. 1988, and others. The section on long-term health effects draws on Gajalakshmi CK. Jha P, eta[. "The health consequences of tobacco", a background paper to this study. It also draws on Peto. R. ct al. British Medical Bulletin 1996: 52, 12-2 i; Murray CJL and Lopez AD. The Global Burden of Disease. Harvard School of Public Health/WHO/World Bank, distributed by Harvard University Press. Boston, 1996. Investing in Health Research and Development, report of the Ad Hoc Committee on Health Research Relating to Future Intervention Options (Annex i) WHO. Geneva 1996; Liu, Bo-Q[, et al, British Medical Journal, 1998: 317: 14 ! !). The information on the detection of a tobacco-related carcinogen in newborn infants' urine is from 321422995

Hecht eZ al, paper presented to the annual conference of'the American Chemical SocieW, 1998. The discussions of'the relationship between tobacco and poverty draw on Bobak M et al, PoverW and Tobacco, a back~ound paper for this study. The poverty discussion also draws on Peto eta], The Lancet, :~39: ]265- 78; Anonymous, Morbidity mzd MorwJity Weekly Report, December 27 ! 99?, vol 46, no 5 ! pp 12 i 7-20. The discussion on inf.onnation draws on Kenke], eta]. Consumer information and tobacco use. Background paper m this study. Discussions of'the relationship bez~veen education and behaviour draw on Preston, S. H. • nd Haines, M. R. Fmol Yem's: Child Monali~' in Late Nineteenth-Century America. P~nceton Unive~it7 Press. 199]. The section also draws on The Economics of.Smoking. Chaloupka F .~ and Warner K. E. From The Handbook of" Health Economics. Newhous¢ J and Culyer A. Editors. ]999 [publ?] Draft: Not for circulation or citation 321422996

6. Is there an economic rationale for intervening in the tobacco market? We have described trends in tobacco consumption and summarized the evidence that tobacco use is both addictive and a major cause of long-term fatal disease. We have seen that in general, the more people know about the health damage caused by cigarettes, the less willing they arc to smoke, while studies in developing countries indicate that many smokers simply do not know the risks. In thi~ chapter, we focus on the economic impact of tobacco use. We ask what its costs and benefits are, and discuss the economic arguments for and against tobacco control. 6.1 Counting the costs of tobacco use While the health effccLs of tobacco,are well documented, studies of its costs and benefits to society are still at a much earlier stage of evolution. Yet there is a clear need for governments, health service providers and others to find out how tobacco affects their economic health. In high-income countrie.s, the available information about the costs of tobacco use not only helps to plan for futurb health service costs but has also shaped t tobacco control policies, raised tobacco high on the public policy agenda, and has been used as evidence in lawsuits against the industry (Lightwood et al 1999). Equally, the industry uses information on the costs and benefits of smoking to support its case against tobacco control. There is no single answer to the simple question: "what does tobacco cost ?" in any given society. Clearly, tobacco use has both costs and benefits in the economy. Consumers derive some perceived benefit from tobacco in that they pay to smoke; in economic terms, this may be measured in terms of the consumer surplus, that is, what they would be prepared to pay over and above what they already pay to continue smoking. Equally. tobacco use has obvious gross costs to society: the health care of smokers, for example. However, there is much debate about how to measure these costs and benefits. The available studies have different purposes and methodologies, some of them complex and controversial. In this section, we summarize some approaches to analysing costs, and identify a range of different types of cost and benefit associated with tobacco in order to clarify the terminology and avoid confusion and double counting. Next, we summarize some key findings of the existing studies, and focus on some of the main controversies in the debate on this issue. For example, does the price of cigarettes reflect the cost of life lost to smokers who did not know or understand their risk when they became addicted to nicotine? Do smokers' net health care costs over their lifetimes add up to any more than those of nonsmokers? Can the costs of consuming an addictive good be counted in the same way as those of consuming a nonaddictive good? The discussion points out important differences between high-income countries and low- and middle-income countries whose tobacco costs have yet fully to emerge. It ends with the broad conclusions of what is known now and identifies some important research needs for the future. Draft: Not for cirrtdntion or citation 321422997

6.I.1 Approaches to cost analysis Because there are several different ways of analysing costs and benefits, it is vital to avoid confusion when describing and comparing the outcomes of different studies. It is also important to identify precisely what is being included under different cost headings so as not to risk double counting of any cost or benefit. Three types of cost analysis are described here. (i) Econondc cost analysis This type of analysis takes the point of view of society and considers all the economic costs and benefits borne by all individuals. Economic costs are defined as the actual amount of real, consumption that individuals must give up in order to engage in some activity. In the case of tobacco use, ~is would include the food, leisure and other goods and services that the smoker foreg6es in order to smoke. Similarly, the economic benefits of tobacco for the consumer are defined as the amount of other real goods and services that are forfeited for the pleasure of smoking (or, possibly, the goods and services the addict is prepared to give up to maintain the addiction). Economic cost analyses focus on opportunity costs, that is, the value of whatever was sacrificed to buy the good, or the value of the next best good ($amuelson and Nordhaus 1985). This type of analysis is very different from those that analyse money costs. Unlike analyses of money costs, economic costs analyses avoid counting any cost that is in fact only a transfer of money from one part of the economy to another. For example, revenue raised by tobacco taxes is a money cost to the smoker but not a cost to society as a whole because it simply requires a shift of resources from smokers to the whole population. Taxation only results in economic costs if it costs money to raise taxes, and this reduces economic resources as a whole. Economic costs analyses arc complex since all costs and benefits to society can be included. The economic benefits include, for example, the consumer surplus, which is the amount that smokers are prepared to pay over and above what they already pay to continue smoking, and the producer surplus--the net gain to producers of tobacco and tobacco products, taking account of what they could have earned from an alternative product. The economic costs can be divided in terms of tangible costs (those that can be measured in money terms), and intangible costs, (such as the length and quality of life). Another important distinction between types of economic cost is whether they are external or internal. We discuss this next. Internal and external costs and benefits When consumers buy a good or service from a producer, they are assumed, in economic cost analyses, to have taken all the costs of that good or service into account, weighed them up and chosen this good in preference to anything else that they could have bought with the same money, because the perceived benefits from the purchase outweigh the costs. The costs and benefits of the transaction are said to be internalmthat is, they are all Draft: Not for circulation or citation 38 321422998

taken into account and all borac by the direct parties in the exchange. Internal costs are excluded from analyses of the social costs and benefits of a good or service. However, some goods and services also carry external costs--that is, costs that are borne not by the producer, consumer or distributor, but by others. They are not usually taken into account by those involved in a transaction and neither buyer nor seller has any motivation to pay them. For example, car use carries the external costs of pollution and traffic congestion that must be borne by nonusers of cars. Smoking involves some obvious external costs, such as the nuisance of environmental tobacco smoke to nonsmokers, and the risk to their health, particularly to the fetuses and infants of smoking parents. A k~y question for analysts is whether the health damage to smokers and the years of life that may bc lost through premature d~ath should be considered as internal or external costs. For these costs to be internal, the smoker must have taken them into account in full. If smokers know that they arc likely to lose a certain number of years of healthy life and decide to smoke anyway, then their costs may be regarded as internal and already accounted for in the price of cigarettes. In thi~ case, it would be double-counting to assess them as costs to society, If, however, these costs arc not fully takdn into account because the smoker is uninformed about the scale of the risks, then, in proporlion to the extent by which the smoker has underestimated the risks, there will be external costs. Wc shall argue in the next section that there arc major external costs of smoking duc to consumers' lack of information about the health consequences. These cos.is may bc measured in terms of what they would bc willing to pay to buy back the years of healthy life that they forfeit through smoking. (ii) Budgetary cost analysis This second type of cost analysis is used to enable governments and other agencies, such as health care providers, to estimate and plan ahead for both the revenue and expenditure that are associated with specific goods or services. Budgetary cost analysis relies on assessment of the money costs and benefits of tobacco, and can therefore produce very different results from an economic cost analysis. For example, in an economic cost analysis, tobacco taxation would be seen, as we have shown, simply as a transfer of resources; in this case, however, it would be counted as a cost to smokers and a benefit to the exchequer. On the consumption side, the types of benefit that would be counted under this type of analysis include tax revenue, and any changes in revenue due to changes in life expectancy. Costs under this type of analysis would include expenditures for treatment and loss of income while sick due to smoking, all revenue changes due to caring for the di,seased smoker, and revenue changes due to the hardship of survivors, the costs of fires and environmenta[ pollution due to tobacco. (iii) Cost effectiveness analysis Draft: Not for circulation or citation 39 321422999

In this type of anaiysis, there is no attempt to assess the total costs and benefits of the activity of smoking. Instead, the aim is to compare different interventions in tobacco control, such as the provision of counselling therapy to help smokers quit versus the provision of nicotine replacement therapy, and see which buys the biggest health gain for the smallest investment. Thus, while the costs of the interventions can be measured in terms either of economic or budgetary costs, the health gain is measured in a different currency, usually either years of life or years of healthy life saved. Cost-effectiveness analyses can also compare the use oftbe same intervention by different agencies, to show, for example, whether smoking cessation therapy is more cost-effective when provided as an in=patient service or as an out=patient service. Important[y, cost malyses can b¢ performed on sections of the economy or categorised by those groups on whom they impact; ~These are often used for policy purposes. 6.1.2 Types of cost Just as it is important to understand the different approaches to cost and benefit analysis, it is essential to have a clear taxonomy ofc, Osts. Table 6. I summarizes one such taxonomy in the context of the kinds of analysis that are frequenfly used by policy- makers, recognizing that others may be equally valid. This table shows how cdsts and benefits might be defined, first, when analysing the aggregate economic costs of tobacco consumption, second by assessing the budgetary costs of tobacco from the point of view era national government, and third, by an incidence analysis of the money transferred from nonsmokers to smokers. Draft: Not for circulation or citation 4 0 321423000

Table 6.1.-Taxonomy of Costs for Tobacco Polic,. COST CATEGORY ECONOM IC COST I I.I 1.1.1 1.1.2 i.!.3 1.2 1.2.1 1.2.2 1.2.2. .2.2. 2 1.2.2. 3 1.3 |.3.1 1.3.2 1.3.3 1.3.4 2 2.1 2.1.I 2.1,2 2.2 2.2.1 2.2.2 AI~AL,ySIS CO SUMPTIO I Benefits Consum¢r surplus internal to rr~rket, to smoker BUDGETARY INCIDENCE ANALYSIS ANALYSIS 2.2.3 internal to market, to Efficiency cost of taxation Tax revenue Transfers of tax revenues others from smokers to others Economic cosVbe~efit of external to market, to others Costs unexpected changes life expectancy Internal. to smokers Ffficien,c~ cost of ,taxation All revenue changes due to changes in life expectancy ETS and parental smoking Transfers arising from changes in life expectancy from smokers External costs to smoker Addictive consumption Cost of exce.~ Iobacco - - consumption from unexpected addiction Morbidity Cost of unexpected illness Expenditures for 'Subsidies for treatment ( treatment, loss of and loss of income while ~ income while sick sick Mortality Cost of unexpected death - - External costs to others' All revenue changes Morbidity of smokers Mortality of smokers Other extcmaJitics Costs of disability and death in children and others Costs a~d caring externalities Loss of income and companionship to survivors Property losses from fires and environmental pollution Producer surplus PRODUCTION Benefits. Internal to Market due to changes in life expectan~ and disease All revenue changes due to caring Ibr diseased smoker All revenue changes due to hardship of survivors All revenue changes due to fires and environmental pollu.tion Corporate and business tax revenues to producers to others Transfers arising from disease and death of children and others All revenue changes due to changes in life expectancy Transfers arising from cost ol'discascd smokers Transfers arising fi-om societal costs imposed by survivors Transfers arising from fires and environmental pollution Transfers arising from corporalc and business taxes Costs, External to Market Morbidity Unexpected costs to workers All revenue changes Transfers due to changes due to changes in in disease disease Mortality Uncxl~ected costs to workers Transfers due to changes in life expectancy Environmental Cost of environmental damage All revenue changcs due to environmental Transfers duc It) environmental damage damase Types of cost and benefit are briefly summarized here. Consumption costs and benefits are discussed first, production costs and benefits second. dt 321423001

Consumption Consumption costs and benefits are all those that arise after a smoker buys tobacco: the consumer surplus, tax revenues, and the costs of addictive consumption. We briefly discuss some of these below. Assessing costs and benefits in the context of addiction As we have seen, smokers arc prepared to pay to smoke and may perceive some benefit from their action. However, a smoker who is addicted to nicotine may not perceive that she or he gains benefit from the bchaviour, whereas in the initial stages of taking up smoking, there may be greater perceived benefits such as enhanced self-image. The change from willing choice to regretful addiction over time has important implications for deciding whether the consumer ~urplus of tobacco should be counted entirely as a benefit or not (Lightwood et al 1999). The smoker who is addicted cannot truly be said to be choosing to smoke. If there is no real choice in the consumption of tobacco, nor is there an opportunity cost involved, because,it is unlikely that the money spent on smoking could be spent on anything else instead." ,However, if the addict reduces his or her consumption, there will be a loss of welfare. Thus, for the purposes of economic cost • analysis, the consumer sttrplus of smoking is counted as a benefit to the smoker. But if the analysis takes account of a smoker's behaviour over a long time frame, starting when the smoker was not addicted and continuing through a period of addiction, the total measured consumer surplus may not represent a benefit. In these terms, the consumer surplus of tobacco can be interpreted as the cost of enforced tobacco consumption arising from a youthful underestimate of the probability of becoming addicted (Lightwood et al 1999). This would mean that enforced consumption should be counted as an economic cost. We return to this issue later in the chapter. When a budgetary analysis is performed, the consumer surplus makes no impact either as benefit or cost. l.n this type of analysis, the purchase of cigarettes produces tax revenue, which counts as a pecuniary benefit, as shown in row 1.1.2 of the table. In terms of an incidence analysis, tax revenue raised through smoking represents a transfer of resources from nonsmokers to smokers. Costs to the health of smokers The health costs to smokers are shown in the table under rows 1.2.2.2, 1.3.1 and 1.3.2. In terms of economic costs analysis, these should be counted as costs to smokers themselves only when they are unexpected by the smoker, and therefore not taken into account on buying cigarettes, as we discussed in the previous section. The budgetary costs include expenditures for treatment and loss of income while sick, and all revenue changes due to changes in life expectancy and disease. Healthcare costs fall into two types. The first are direct costs, which involve all professional health services provided to treat ill health that results from smoking. They Draft: Not for circulation or citation 42 321423002 BATCo US DOJ v Philip

include hospital services, medical services, prescription drugs, nursing homes, and allied health services provided by professionals such as physiotherapists. Indirect costs are not related to treatment but arise as a result of the income lost by the sick or dying person. They affect both the smoker and others and include the costs to the family of caring for the sick smoker, as well as any knock-on effects of the presence of a sick relative in the household, such as the removal of children from school to provide care or replacement work. It is extremely difficult to identify all these costs, and their size is likely to differ substantially depending on the structure of the economy. For example, if a given society has a social security structure that enables the family of a sick smoker to claim benefits to counteract the loss of the smoker's income, this becomes a cost to the government. If no such payments are available, the family obviously bears that cost. Another example is insurance. Extra insurance payments may be made because of tobacco use. In developed countries, these costs can be estimated because insurance markets exist and the price of the insurance available can be m, onitored and related to the incidence of disease. Premiums will take into account the expected costs of illness for each insured individual and can therefore be used to identify the excess costs to whichever agency pays for the increase in premiums. The true economic cost is, of course, only that part of the increased insurance payments that are due to unexp, ecled increases for smokers, while for nonsmokers all increased payments count as costs. In contrast, where insurance markets do not exist, the sick individual and his or her family cannot recover the resources they lose. The costs--and benefits--of early death due to tobacco are an equally important consideration and one that has attracted controversy. The cost to the individual of dying earlier than expected is clearly large, as it is for the person's family and friends. But in some types of budgetary cost analysis, there may be benefits from a smoker's death if that death comes after he or she has paid pension contributions, but before he or she has drawn that pension. The saver is the organization that pays the pension. Again, the picture will differ sharply between high-income and low-income countries. Nations that have no sophisticated pension structure will not see these savings. If costs and benefits are analysed in incidence terms, premature death shows a flow of resources from one group to another. If sales taxes on tobacco produce a net transfer from smokers to nonsmokers, the premature death of a smoker is a cost to nonsmokers. But if most smokers die after paying their pension contributions and before buying them, then premature death is a net benefit, producing a net increase in transfers from smokers to nonsmokers. It is, however, controversial to categodse premature death as a benefit because this takes no account of the value of life. We return briefly to this issue in the next section. Another controversial issue is how to calculate the costs of the smoker's deafla to others in society. The indirect costs include lost production to the rest of the household, and lost output for society as a whole. These losses camaot be regarded as economic costs if the family, and individuals in society as a whole, are judged to know that smoking will cause premature death, because the costs of goods and services that compensate for the loss, Draft: Not for circulation or citation 43 321423003

such as insurances, would be expected to take account of them. But if the compensation that others must make for the smoker's death is unexpected, these losses should be considered as economic costs. This is particularly true for a developing country with few insurance markets and low general levels of awareness of the probability of being killed by tobacco. As we have seen, some of the costs of smoking are borne not by smokers but by nonsmokers, or by a combination of nonsmokers and smokers. These are defined as external costs. They include the health hazards of tobacco smoke for nonsmokers, the property damage from fires and environmental pollution caused by tobacco and, the costs to relatives and other members of society, as discussed above, and, more controversially, the loss of life from uninformed smokers. Table 6.2 summarizes some external costs of tobacco use, making a distinction between direct and indirect costs. Table 6.2 Direct and indirect external costs of tobacco consumption: some examples and estimated size (ma~, be separated later into budgetary and economic costs) Direct external costs Indirect external costs Health hazards to non-smokers of passive smoking Size: minimal but significant Physical nuisance to non-smokers of exposure to environmental tobacco smoke (eye irritation, nasal symptoms, coughs etc)/ Size: considerable but difficult to measure Financial nuisances to non-smokers (fire damage, cleaning) Size: not well studied but considerable (such as the costs of fires) Lbss of life from uninformed smokers (i.e. internalized external costs) Size: willingness to pay studies suggest this to be considerable. This is by far the largest cost, but also the hardest to measure. Additional life insurance premiums shared by non-smokers as a result of tobacco-related disease in society Size: unkmown and variable across societies Additional health care costs shared by non-smokers as a result of tobacco- related disease in society Size: gross costs of 0.7 to 2.0% of GDP in some countries, unclear net COSLS Existence value (non-smokers' willingness to pay for initiatives that may reduce smoking prevalence: non-smokers' emotional distress at ill health loss of life among smokers) Size: unclear but a percentage of public spending on health is one indicator Having described some costs and benefits of tobacco consumption, we now turn to a brief assessment of types of costs and benefits of its production. Production Everything from the farming of tobacco to the production of cigarettes and other tobacco products is included in this analysis. A key benefit is the producer surplus, which is the net gain that producers could expect over and above the amount they could earn from selling other goods using the assets they currently invest in tobacco. The size of the producer surplus is determined in part by the competitiveness of production conditions and the elasticity of the tobacco supply, which itself reflects how much tobacco farmers Draft: Nor for circulation or citation 44 321423004 BATCo US DOJ v Philip Morns

can earn for their crop at any one time relative to what they could earn from alternative crops. These benefits are summarized in rows 2.1.1 and 2.2.2 of Table 6.1. Production costs include the adverse health effects to workers of farming tobacco, such as handling large amounts of pesticides. They also include external costs such as the impact on the environment of tobacco agriculture, including deforestation, wood shortages and pesticide pollution. These are quantitatively small effects at global level, but may be significant in certain countries where a high percentage of land is given over to tobacco farming (Lightwood et al 1999). 6.1.3 Keyfindings and controversies of cost estimates Most of the studies available to date focus on high-income countries. Their applicability to low-income countries and middle-income countries, where tobacco use is increasing, is not yet clear. There are significapt differences in low-income countries' patterns of health spending, and in the burdens of' ~moking-attributable diseases on families. Health care costs: do smokers pay their way? Researchers have made both budgetary. ~ost analyses and incidence cost analyses of the healthcare costs of smokers. In budgetary terms, the gross healthcare costs of smoking range from about 0.7 to 2 per cent ofGDP (Jha et al 1998; Lightwood et ai, 1999). In the USA, studies have consistently found the gross costs of smoking to be responsible tbr about 6 per cent of all the nation's medical care expenditures. However, recent research suggests that these studies may have seriously underestimated these costs and that a figure closer to 11 per cent of all medical expenditures may be more realistic (Warner, xx ck reference). Clearly, the size of the gross healthcare costs bill will differ sharply from one country to another depending upon the sophistication of the health system, the stage of the tobacco epidemic, and so on. Gross costs also vary over time. Thus, the figures cannot be extrapolated from one country to another. A key issue in recent debates over the true costs of smoking is whether smokers pay their way in healthcare by dying younger than nonsmokers. All cross-sectional studies of the health care costs of smoking, that is, analyses in any one year, show that smokers' costs outweigh those of nonsmokers (Lightwood et al 1999). However, there is no consensus at present as to whether their lifetime costs are greater. Some economists, such as Leu and Sehwaub, have demonstrated that nonsmokers' longer lives accrue healthcare costs that compensate wholly or partly for smokers' higher annual costs during their shorter lives (Leu and Schwaub 1983). Another study showed that a I per cent decrease in cigarette sales in the US would increase the net present value of health care costs by US$405 million for a population between the ages of 25 and 79 (Lippiatt, 1990). This sum outweighed the costs savings of smoking cessation. However, the figure fell to $91 million when the cost of future health care [ck] was valued at an increased discount rate. The money saved by ex-smokers not buying cigarettes was excluded from the analysis. Draft: Not for circulation or citation 45 321423005 I~TCo US DOJ v Philip

Another study, by Viscusi (Viseusi 1995) found that smokers impose no external burden on society in terms of budgetary costs. Here, the external health care cost of cigarettes, defined as the sum of medical and nursing home costs, was found to be between 8 cents and 43 cents per pack, depending on assumptions about cigarette tar levels and discount rates. Viscusi suggests that income is transferred from smokers to nonsmokers. While some studies therefore show that smokers' lifetime healthcare costs are either equal to or smaller than nonsmokers', other studies have reached conflicting conclusions, finding that smokers' higher annual costs eventually do outweigh the costs of longer-lived nonsmokers (Hodgson 1992). Hodgson estimates that the smoking population aged 25 years and over increases the net present value of lifetime health expenditures by US$501 million, or just over $6200 per smoker. Another study in the UK found that if the number of smokers fell as a result of increased taxes on cigarettes, the budgetary costs of smoking would fall too (Atkinson and Townsend 1977). In general, those life-cycle cost analyses that use a more complete list of tobacco-related diseases, and include more health services, find that smoking increases aggregate health system costs. There are methodological difficulties With both lifetime analyses and cross-sectional analyses. For the lifetime analysis, chan~es in the demographic structure of the population, such as ageing, will affect the size of future costs. In populations where patterns of disease and life expectancy are changing rapidly, forecasts of future demographic change can be made only by making strong assumptions. Also, in both types of estimate, there are uncertainties about the type of healtheare that will be needed. Most of the healthcare needs of smokers are concentrated into the period before death, and these may not necessarily be highly sophisticated medical services, but palliative care, which in developing countries at least, and possibly also high-income countries, may be provided by relatives or nursing homes and is less expensive than hospital care (Scitovsky 1994). Another often neglected issue is that older relatives may care for or otherwise support younger, sicker relatives. There are also a number of other methodological issues that require careful attention, including the types of cost included, the estimates made for these costs, and the number of disease attributed to tobacco in each study (Lightwood et al 1999). Whether a consensus is eventually reached or not, the underlying notion that smokers somehow benefit society by dying young reveals some worrying assumptions. Certainly, no economic analyses before these have regarded premature death as an economic benefit (Warner and Chaloupka 1999). If, in order to save healthcare costs, adults should ideally die around the age of 65, then society must, logically, have no regard tbr human 1i1~. Extending the logic of this argument would mean that governments should encourage older adults to end their lives to save public money, which is clearly absurd. It should be stressed, once again, that most of the studies to date refer to high-income countries where'different cost structures, health system expenditures and pension benefits would be expected to produce different patterns of expenditure for smokers relative to nonsmokers. In particular, the premature deaths of smokers in most developing countries Draft: Not for circulation or citation 46 321423006

would not amount to an economic benefit because old age expenses are largely privately financed. We briefly outline some studies that refer to developing countries below. Health care costs in developing countries Analyses made as part of the background work to this study have attempted to project budgetary heal[hcare costs for two large developing countries for which dam are available (Lightwood et al 1999). Using existing projections of the health effects of tobacco use (Murray and Lopez 1996), and indicators such as income per capita and life expectancy, the studies find that China's GDP will have to grow at least 7 per cent per year for the next two decades to achieve a substantial reduction in health care costs related to tobacco, expressed as a percentage of GDP. While China's growth rate has been between 9 and I 0 per cent per annum for the past 20 years, it is extremely optimistic to assume that it can sustain these rates of growth i~.future. India's prospects, according to the analysis, are worse. The nation will not be able to "grow itself out" of an increasing tobacco burden unless it can sustain an average growth rate of 7 per ccnt each year for the n~Xt two decades. On average, over the past 20 years, India's economy has grown at 5.8 per ~:cnt. Life expectancy in China is already relatively long, so that the diseases suffered by smokers are already prevalent in its population. By contrast, Indian life expectancy is lower, at around 60 years, and as its population ages in coming years its tobacco-attributable burden is likely to grow considerably. An important finding of another study is that the current healthcare costs of tobacco use in developing countries are no lower than those in developed countries, when assessed in terms of the proportion of GDP devoted to treating tobacco-related disease [ck typo. in source, Lightwood et al, conclusions section]. This may be explained by the fact that life expectancy in many middle-income countries is already well above 60, and by the possibility that some smoking-related diseases occur in younger adults particularly in the presence of co-factors such as infection with Mycobacterium tuberculosis, the cause of TB. Also, it is important to note that even if the total current and future discounted costs of smoking to the economy arc zero or negative, a reduction in the number of people who smoke will result in cost savings for up to ten years. This reduction may be particularly valuable in low-income countries. How much do smokers" lack of h;formafion and addiction affect t/be estimates of tobacco costs? In section 6. I we discussed the question of whether smokers' loss of health and life should be considered as an internal cost, and therefore taken into account in the price of cigarettes, or whether it should be considered in part as an external cost because smokers arc not fully aware of the risks they face in consuming tobacco. This is the subject of considerable controversy, not least because the cost of lost life for uniformed smokers would constitute by far the biggest component of the external costs of smoking. In this Draft: Not for circulation or citation 47 321423007

report, we conclude from the evidence available that the costs of uninformed or inadequately informed smoking should be regarded as external. If all smokers fully understood the risks to their health when buying cigarettes, and if all cigarette purchases were made on the basis of free choice, then the r~sulting health damage and loss of life could be regarded as internal. But, while the evidence is far from complete, it appears that a substantial proportion of smokers worldwide either do not know their risks or fail to assess them accurately. As we saw in Chapter 5, there is considerable disagreement between researchers as to smokers' true estimation of their health risks. However, some of the most recent studies based on widely used survey data (Schoenbaum 1997) indicate that heavy smokers underestimate their chances of survival, while a range of other studies shows that the risks of smoking relative to other influences on health are consistently underestimated. Equally important, the majority of newer recruits to smoking now live in,low-income and middle-income countries and, if a study of Chinese smokers is representative, more than half believe that tobacco will do them "little or no harm". If, as we suggest, an unknown but siz6able number of smokers are inadequately informed about their health risks, it follows that ~e damage to their health and loss of life should be counted as an external cost. This would mean that the price of tobacco failed to capture its true costs, amounting to a market failure. Of course, the actual number of uninformed smokers is not known. As part of the studies underlying this report, researchers performed a cost-benefit analysis of smoking using innovative methods that may help to shed new light on the debate, gather than try to estimate the total economic costs of the smoking, the analysis estimates the size of the consumer surplus and the producer surplus of tobacco, and then, using estimates of the value of life foregone and the willingness of individuals to pay to live, estimates what proportion of the total number of smokers would have to be uninformed for the costs to equal the benefits (Peck at al 1999). The study concludes that the social costs of smoking would equal its benefits even if the percentage of uninformed smokers was very small. Depending on various assumptions, tobacco has no economic benefits for the world as whole if as few as 3 to 25 per cent of smokers do not know their risks (see Box 6. I, Is the world worse off or better off with tobacco?). Observations of trends over time support the view that the consumer surplus of tobacco may be smaller than expected once consumers' access to information is taken into account. The surplus due to tobacco consumption is falling in the US, the UK and other countries where tobacco consumption is falling in an increasingly well-informed population. In the UK, for example, consumption of cigarettes has almost halved since the mid-1960s, resulting in a similar fall in consumer surplus. It is possible that tobacco consumption will eventually fall in other countries where it is now rising. If this were to happen, the surplus would shrink again. The scale of the consumer surplus is further diminished by the fact that tobacco is an addictive good. Until recently, economists have either ignored the addictive nature of Draft: Not for circulation or citation 48 321423008

goods such as cigarettes when estimating demand, or have assumed that smoking was such an irrational behaviour that it could not be examined in the framework of economics. However, economic models of addictive consumption are now being developed. One that gained wide acceptance was the rational addiction model of Becket and Murphy (1988). This held that individuals constantly maximize utility over their lifetime, taking into account the future consequences of their choices. In this model, current consumption is determined by past consumption as well as current price, and predicts that the long-run effect of a change in price will exceed its short-run effect. Thus, consumers of an addictive good will respond much more slowly to price changes than consumers of a non-addictive good, because their addiction must be suslained. A problem with this model is that it fails to take account of the fact that addicts regret their addiction, as is apparently the case with smokers. The model assumes that individuals ar~ fully aware of thd consequences of consuming the addictive good when they decide to consume it, and implies that there are "happy addicts" (Winston 1980). More recently, economists have b~gun to take account of the idea that inexperienced users of an addictive substance are not ~..fully aware of its potential harm (Orphanides and Zervos, 1995). Their model allows for the possibility that individual consumers have subjective beliefs about their potential td become addicted, and that these subjective beliefs change over time towards a state of regret (Chaloupka and Warner t999). In this case, they are ndt deriving benefit from buying tobacco but are simply spending the money on cigarettes because they have to maintain their addiction. Thus, as we set out in the discussion of costs in the first pan of this chapter, some of the consumer surplus of tobacco turns into a cost once addiction is taken into account. Draft: Not for circulation or citation 49 321423009

Box 6. l Is the world economically better or worse off with tobacco? The model developed for this study (Peck et al 1999) aims to set some value on the costs and benefits o~" tobacco to society. Given that many of the required data for a cost analysis would be difficult or impossible to obtain, for the reasons outlined at the beginning of the chapter, the study approaches the question from a different angle. First, it assesses the combined comumer surplus and producer surplus from tobacco in order to estimate the economic benefits of smoking. Then, rather than try to estimate costs, it asks how many smokers would have had to fail to take the full costs of smoking into account for their costs, in terms of lost life and health, to cancel out the global benefits. The consumer surplus of tobacco is assessed as the amount that smokers are prepared to pay over and above what they already pay for the opportunity to continue smoking. This sum is assessed on the basis of existing data on cigarette consumption by region, which are described in Chapter 3, price of cigarettes, and price elasticily. The price eta pack of cigarettes is averaged by region, on the basis of available data, at $I .80 for high-income countries, $0.40 for India, :$0.20 for China and S0.60 for the rest of the world. Price elasticity is set at levels indicated by various studies of high-income and low-income countries; a range of different estimates produces widely differing results, as expected: if people are prepared to pay much more for cigarettes than they currently pay, the size of the consumer surplus will be found to be much greater. The benefits are assumed to be discounted into the future at rate consistent with existing studies. The calculations show that the annual consun~e'r..surplus differs sharply fi'om region to region, reflecting the different expenditures of different populations o,n cigarettes. The bulk of the consumer surplus accrues to the high-income countries, with an annual per ~pita surplus of $173. In Sub Saharan Afi-ica, by contrast, the consumer surplus is jusl $7 per person per year and even in China. because of the low. price of cigarettes there, the surplus per person is just $9 per year. For low-income and middle-income countries as a group, the average consumer surplus is $16.50 per person per year, less than a tenth of its value in the rich countries. In aggregate, the consumer surplus is somewhat over $126 billion a year (figm'e to be verified). Next, the producer surplus is estimated. This is simply the payment that producers receive for producing tobacco, over and ~,bove their opportunity costs. The estimate is based on existing data on the price of tobacco, assumptions about the degree of competitiveness in the indust~ and estimates of the elasticity of supply. In comparison with the consumer surplus, the producer surplus of tobacco is small, at about $40 billion. Notably, however, the relative shares of this surplus between regions varies sharply. Sub-Saharan Africa has a large proportion of the producer surplus because of the scale of tobacco farming in Zimbabwe, Malawi and certain other countries. This has implications for tobacco control policies and economic adjustment, which we discuss elsewhere in this report. In assessing the costs of tobacco consumption, the key issue is to determine which costs are already accounted for by consumers, and which not. Those that are accounted for, or internal, should be excluded fi'om the analysis. Rather than attempt to decide what is an external cost, and how big these are, the analysis estimates the uncounted value of life lost from uninformed smoking and, by setting a value on each year of life lost, calculates how many smokers would have to be uninformed for the costs to match the size of the surplus. First, the total number of years of healthy life that are projected to be lost through tobacco between 1990 and 2020 are estimated. These estimates are available already (WHO Ad Hoc Committee on Health Research Relating to Future Intervention Options, 1996). As with other assessments of disease burden due to a risk factor, the years of life lost through premature death and years lived with a disability of given severity due to tobacco are combined into a single time-based measure, the disability-adjusted life year (DALYs2). 2 The disability-adjusted life year is a time-based measure that enables epidemiologists to combine in a single indicator time lost due to premature death (defined as death before the current age of life expectancy at birth in the longest-surviving population, Japan, which is approximately age 80), and time lived with a Draft:/Votfor circulation or citation 50 321423010

Next, a value is set on each tobacco-related DALY in an attempt to value the lost life. On extremely conservative assumptions, this is set at US$7750. This is average per capita GDP weighted for tobacco consumption, in the USA, for instance, a survey of 23 studies suggested that the value of a whole life is $5.8 million, while average income is $22, 600 (Viscusi, ~993). It', allowing for age-weighting and discount rates, one life is composed of 40 DALYs, then the value ofa DALY in this case would be worth about $320 000, or about 14 times income. The figure of $7750 is used to gauge what individuals who did not fully understand their health risks on starting to smoke, and who now regret their habit, would be willing to pay to buy back each DALY, or year of healthy life they have lost to smoking. If, as suggested by the studies in the USA, the value ofa DALY is 14 times per capita GNP, then we can calculate that only 3 per cent of smokers would need to be uninformed for the total cost of the lost life to equal the size of the combined producer and consumer surplus, that is around Sxxx billion [final figure pending-]. If still more conservative assumptions about the value ofa DALY are taken, the proportion of smokers who would have to be uninformed for their costs to equal the benefits of tobacco rises to about a quarter. Only further research will provide a greater degree of certainty about the percentage of smokers who are uninformed, but this range [or a range between 1/3 and 2/3, with figures, as discussed] seems ~'easonable on present information. Therefore, it is fair to conclude from this analysis that even if only small numbers of smokers are unaware of their risks, the world is economically worse off with tobacco than without it. BOX NOTE: It is worth noting that these estima.tes produce a range of values for the costs of tobacco use [figures] that differ from previous estimates (Ba/~num 1993). The assumptions underlying the earlier estimate were different, particularly in that marginal costs for an incremental production of tobacco was used. The earlier estimate also considered as external some costs that many economists would consider internal. END OF BOX disability. One DALY is thus one lost year of healthy life. Thus to measure the disease burden in DALYs of, say, road traffic accidents in India in 1998, add the number of years lost to premature deaths from accidents and the number of years lived with nonfatal injuries of a given, weighted severity, in the population. Draft: Not for circulation or citation 51 321423011

In conclusion, then, the costs and benefits of tobacco use are difficult to measure and the existing studies are subject to much debate over methodology and interpretation. However, on ba!ance, it appears that: • The costs of smoking to society are at least equal to its economic benefits, and probably exceed its benefits, when smokers' lack of awareness of their tru¢ health risks and their addiction to nicotine arc taken into account. If addiction is allowed for, then part of the consumer surplus of tobacco, traditionally counted as a benefit, becomes a cost. * The health care costs of smokers arc certainly greater than those of nonsmokers in any given year. Over their lifetime, smokers' costs in high-income countries may be no greater than nonsmokers' lifetime health care costs, but different studies produce contradictory evidence. In low-income countries, there is unlikely to be a flow of resources from smokers to nonsmokers because of the absence of pension schemes and the relative lack of an infrastructure for the formal care of chronically diseased elderly people. • Low-income countries' overall health care costs, as a percentage of GDP, are found to be as high as those in high-income ~ountries. • Low-income countries where the full lost of the smoking epidemichas yet to emerge will have to maintain optimistically high economic growth rates to contain their burgeoning tobacco-related health care costs. 6.2 Market failures and the grounds for intervention From the standpoint of public health, the world would be better off entirely free of tobacco. A substance whose annual death toll is expected to reach about 10 million by 2030 should, in the eyes of the public health community, be eliminated, and the case for doing so has been extensively discussed elsewhere. Economists would naturally share the public health community's human and moral concerns that disease on such a large scale warrants immediate and far-reaching action. However, from the standpoint of their discipline, the arguments for intervention differ. Within modem economic theory, there is no justification for eliminating tobacco, but only for protecting children and non-smokers from it, and ensuring that adult smokers know their true risks and bear all the costs of smoking. This section summarizes these economic arguments, then examines their validity and their applicability in the context of the tobacco epidemic. The chapter concludes that intervention is justified for some groups. 6.2.1 Consumer sovereignty, market efficiency and the optimal price for cigarettes A basic starting point in modem economic theory is to respect individual consumers' choices and to assume that they know what is best for them--a concept "known as consumer sovereignty. By exercising this sovereignty in choosing what goods and services to pay for, consumers are assumed to maximise their wellbeing, or welfare, within the bounds of their preferences and what they can afford. Maximal welfare is said Draft: Not for circulation or citation 52 321423012

to be achieved when the individual consumer cannot improve that welfare by changing the pattern of his or her spending, say, by buying more shirts than shoes. In these conditions, consumers exercise informed choice and their marginal costs and benefits are balanced equally. A second assumption of most modem economists is that, under specific conditions, free competitive markets provide the most efficient way of allocating a society's resources. According to this assumption, the best way to allocate resources is according to the Pareto principle, which holds that maximum efficiency is achieved when no single individual's welfare can be improved by reaIlocating resources without worsening the welfare of another individual (see, for example, Layard and Waiters 1978). If it is possible to reallocate resources and improve someone's welfare, this reflects underlying inefficiency in the existing allocation of resources. According to this approach, then free competitive markets ensure a more efficient'~listribution of society's scarce resources than anything that governments can achieve. However, it is accepted that some markets fail to achieve the most efficient allocation of resources--a situation known as market f.ai[ure. Most economists agree that a case can be made for government intervention if the following conditions apply: a) there is an identified market failure; b) the market failure is quantitatively significant; and c) government intervention can help to correct the failure without generating new problems elsewhere in the economy. In principle, government intervention would be justified only if it did not violate the Pareto principle: In practice, however, few situations meet these requirements. Any public intervention that affects the distribution of income, for example, such as income tax, will reduce some individuals' welfare. Markets achieve complete efficiency only where, other things being equal, the consumption of a commodity has no external costs (or where external costs equal external benefits). As section 6.1 makes clear, economists are still debating both the nature and the size of the external costs of smoking, but virtually all agree that those costs exist and that they are considerable. If the price of cigarettes (incorporating taxes) fails to take the external costs of smoking into account, it will be artificially low, because smokers will not be bearing all the costs. If prices are artificially low, demand for cigarettes will be driven artificially high. The result will be higher external costs. Whether current tax levels take account of all the external costs of tobacco is a subject of ongoing debate among economists, largely revolving around varying definitions of externalities and consequent wide variations in cost estimates. According to some analysts (Viscusi, 1995; Manning 199 I), current taxes cover the external costs of smoking. These analysts, as we have seen, assume that smokers and their families know their risks and factor them into their calculations about what they are prepared to pay to Draft: Not for circulation or citation 53 321423013

smoke, and that the extemal costs of smoking are confined largely to the impact of environmental tobacco smoke. Economists from a broad range of political traditions have argued that the correct basis for estimating the optimal tax on tobacco is to define social costs by the traditional economist's measure of external costs--that is, costs imposed by smokers on others, including their family members (Chaloupka and Warner 1999). However, most accept that the size of these costs is significant and that better estimates of their size will be instrumental in determining optimal tax levels. 6.2.2 Grounds far intervention Wc argue that, in light of the identified market inefficiencies of inadequate information, addiction and the presence of external costs, however defined, governments arc justified in intervening to control tobacco on the following grounds: (i) to protect children and teenagers who do not know or accurately appreciate the risks of becoming addicted to nicotine; (ii) to protect non-smokers; and (iii) to provide adult smokers with'i~ore information so that their choices can be informed. ~ The rationales for protecting children and nonsmokers are obvious. For adults, the picture is more complex. Smokers obtain very little reliable information from the tobacco industry, and governments should provide full information on the hazards of smoking to all, as a public good. But whether, in addition to providing information for all, governments should actively discourage adults from smoking is, from the viewpoint of the economist, much less certain. Adult smokers fall into sevcral (probably overlapping) categories: -those who genuinely do not know their risk, as the study from China cited in the previous chapter made clear; -those who, at some level, know their risks, and would like to stop, but are addicted to nicotine; -those who, at some level, know their risks and choose to smoke anyway, but allow the costs of their action to be imposed on others; and, finally -those who, at some level, know their risks, choose to smoke anyway, and ensure that they themselves bear all costs. Some of the adult smokers listed above would presumably claim consumer sovereignty in their decision to smoke. In any case, governments would be unable to persuade all adult smokers to quit; economists would argue that they would have no business doing so anyway. Some adult smokers undoubtedly consider that the satisfaction or "benefit" that they gain from smoking exceeds the costs, and, provided they impose no external costs on others, govemments have no justification whatever for stopping them. From the standpoint of public economics, then, there is no clear rationale for intervening to actively discourage or prevent adults from smoking. Draft: Not for circulation or citation 54 321423014

Limitations in the theory of public economics as an analytic tool for lhe tobacco market However, this conclusion raises as many questions as it answers. Tobacco, as we have seen, is not a typical consumer good because of its addictive component. Just as important, the crucial first few weeks and months of smoking, in which addiction becomes established, usually occur when the smoker is young and fails to assess future health risks accurately, an attitude known as myopia, or short-sightedness. For sure, there are other activities popular with myopic young adults that carry a high risk of dying. Young people may also risk their lives in dangerous sports, fast ear driving, or reckless use of alcohol. The difference is that, unlike dangerous sports, fast driving and alcoholic binge-drinking, cigarette smoking is rapidly addictive. There is thus a high probability of becoming addicted to nicotine at precisely the age at which the individual's behavioural inclination to take risks is great~st (Kessler 1995). Because cigarettes combine the high risk of addiction in youth and the unusually high probability of dying--one in two, or one in four in middle age--they are unlike any other legal consumer good. Yet, once the young smoker reaches adulthood and has a c.!e..arer sense of his or her health risks, he or she may be unable to stop consuming them. It is therefore reasonable to ask whether, within its current limitations, the .conventional theory of modem economics is capable of providing a meaningful framework for analyzing the market for a consumer good that has addictive properties. Greater development of economic theory to take account of addictive behaviours in the market for consumer goods is a long-term research priority. In the shorter term, however, there is a more practical decision facing policy makers. They may consider that, by acting purely on the justification of public economics and intervening only to protect'children and non-smokers, they will still achieve substantial victories and save a large number of lives in 50 years' time. But by leaving adult smokers alone, they must expect that tobacco-related deaths will continue to rise well into the next century, because the 10 million deaths that tobacco is expected to cause annually by 2030 will almost entirely concern adults who are already smokers (Figure 6.1). Equally, policy makers may consider that the principle of consumer sovereignty in the tobacco market is not sufficiently robust to justify government indifference to adult smokers' mortality on the scale expected in the next few decades. It is expected that 0.5 billion people alive today will eventually be killed by tobacco. Some policy makers might judge that further interventions are justified, if not on public economics grounds, on public health grounds instead. Draft: Not for circulation or citation 55 321423015

Figure 6. I. Tobacco deaths in the next century, and the expected impact of interventions 400 - ¢~ 300 ~" 200 O ._~ 100 2000- 202;S- 20SO* 2024 2045 Year Ir-'-tB a s elin e ] ~ N o t s ta rtin g := '1 r~lQ u ittin g ..i Notes: A not-starting strategy assumes that that proportion of young adults who become smokers is halved by 2020. A quitting strategy assumes that global cigarette consumption per adult can be halved by 2020: Source: Peto et al, 1998. Draft: Not.for circulation or citation 56 321423016 B.~TCo us DOJ v Philip Mom~

6.3 What types of intervention should be considered? Having concluded that intervention is justified at least for the protection of children and non-smokers, and to increase information for all smokers and potential smokers, we now turn to the types of interventioh that may be appropriate. The vast majority of interventions are intended to reduce demand for tobacco rather than reduce its supply; as later sections will show, the available evidence suggests that efforts to reduce the tobacco supply are unlikely to succeed. We discuss measures to reduce demand in the next chapter, turning subsequently to a brief discussion of the. supply-side issues. Most interventions aimed at reducing demand, such as raising tobacco taxes, banning tobacco advertising and increasing the amount of health information available, have already been widely used in many developed and developing countries--often with considerable success--to reduce the demand for tobacco. Before discussing the evidence for the effectiveness of each intervention, we briefly explore the relationships between each intervention, the specific problems in the tobacco market that each would be expected to address, and the groups most ..affected by each (see Table 6.3). It is vital to stress that there is no neat one'-to-one correspondence between each intervention and any particular subset of smokers: indeed, most interventions will have some impact on all smokers. The implication is that, by intervening to protect children and non-smokers, adults will also be affected. Table 6.3. Interventions, Problems and Affected Groups Type of Intervention Tax increases Information campaign~: -about health risks -about addiction risk -about non-smokers' rights Advertising and promotion restrictions Research on: -consumer behaviour -safer cigarettes Bans on smoking: -in public places -in workplaces Restrictions on access (sales to youths) Risk-adjusted insurance or pension contributions Cessation programs Promote nicotine replacement oroducts Source: Jha P, Musgrove P, et a| Pro ,blem(s) Addressed Non-specific: reduces total consumption Incomplete information, myopia Addiction Direct (physical) externalities Non-specific: reduces total consumption Group(s) Most Affected All: greatest impact on the more myopic ( young, poor and less educated) Adult smokers, especially long-term Youths, beginning smokers Mothers and fetuses; adult non-smokers All smokers, especially youth Incomplete information; addiction Non-specific: reduces health loss Direct (physical) externalities Direct (physical) externalities Non-specific: reduces total youth consumption Financial externalities Addiction Addiction All smokers, especially youth All smokers All non-smokers (including children) Adult non-smokers Youth only All non-smokers Smokers wanting to quit Smokers wanting to quit 1998. Draft: Not for circulation or citation 57 321423017

These control measures are likely to affect adults with low incomes differently from adults with high incomes. Poorer adults are more likely to smoke, and more likely to be harmed by tobacco, than rich adults. They are also more likely to be sensitive to price measures, and less sensitive to information. While higher disease burdens in the poor do not, in themselves, justify stx:cific health interventions on their behalf by governments, interventions that particularly benefit the poor will be efficient in the sense that they will yield large health gains in a population. Several earlier reviews have concluded that governments have a responsibility to improve the health outcomes of the poor (scc, for example, World Development Report 1993). Policy makers are therefore obliged to ensure that tobacco control interventions reach poor consumers. Table 6.3 identifies the market failures that each intervention is intended to address, but also clearly demonstrates that most interventions will have an impact on most smokers. As the following sections will also make clear, individual interventions are unlikely to be used in isolation from each other, but often are applied as a mix. The challenge for governments is to assess the appropriate mix for their own population, bearing in mind that a very large degree of variation between one country's needs and another is unlikely. Draft: Not far circulation or citation 58 321423018

7. Measures to reduce demand can succeed--and raise dollars 7.1 Reducing demand by raising the price: tobacco taxes For centuries, tobacco has been considered an ideal consumer good for taxation: it is not a necessity, it is consumed widely, and demand for it is relatively inelastic, so it is likely to be a stable and easily-administered source of government revenue. Adam Smith, writing in 1776, suggested that, through such a tax--as well as one on rum and sugar-the poor "might be relieved from some of the most burdensome taxes; from those which are imposed either upon the necessaries of life, or upon the materials of manufacture". A tobacco tax, Smith argued, would allow poor people to "live better, work cheaper, and to send their goods cheaper to market". Demand for their work would increase, in turn raising the incomes of poor people and benefiting the entire economy. Two centuries later, almost all governments continue to tax tobacco, sometimes heavily, by a variety of different methods. Their motives have always been to generate revenue, but in more recent years taxes have also reflected an increasing concern with the need to minimize the health damage of smoking~,. This section reviews the evidence on how increased taxation affects the demand for cigarettes and other tobacco products. It concludes that higher taxes do significantly reduce the demand for tobacco. Importantly, given that most people start smoking in youth, the impact is likely to be greatest on young people, who are more responsive to price rises than older people. Equally important, the discussion concludes that higher taxes will reduce the demand for tobacco most sharply in lower-income and middle- income countries where tobacco users are more responsive to price increases than in the high-income countries. Even with this sharply reduced demand, however, governments" revenues need not be harmed. Indeed-in contrast to the claims of the industry and others- -higher taxes may bring substantially higher revenues in the short to medium term. Here, we briefly summarize the types of tobacco tax used by most governments. We assess how changes in these taxes affect the prices set by the industry and, critically, how price increases affect demand. The evidence from middle-income and low-income countries is compared with that from high-income countries. The implications for policy are discussed. 7.1.1 Types of tobacco tax Tobacco taxes can take several forms. Excise taxes are fixed amounts added to the base price of tobacco. Specific tobacco taxes, added as a fixed amount to the price of cigarettes, allow the greatest flexibility and allow governments to raise the tax with less risk that the industry will respond with manoeuvres that keep the real amount charged low (see Table '7. I ). Ad valorem taxes, such as value-added taxes or sales taxes, are a percentage of the base price, are imposed by virtually all countries--often on top of the excise tax. Ad valorem taxes may be imposed at the point of sale or, as in many African Draft: Not for circulation or citation 59 321423019

countries, on the wholesale price. Taxes may be levied per cigarette, by weight of tobacco, or by the place of manufacture--for example, some governments impose higher taxes on cigarettes produced abroad than on domestically-produced ones. Governments may impose different taxes on different types of tobacco product and on different types of cigarette: for example, high-tar cigarettes may be taxed more heavily than low-tar brands. In addition to domestic taxes, many countries impose import duties to protect their domestic tobacco industry and to generate more revenue. The structure of tobacco tax is an important consideration in any strategy of tobacco control, because it is likely to affect the level of the tax and the outcomes for public health. Table 7.1 Some effects of tax Tobacco tax Specific: By weight of tobacco By cigarette Ad valorem: Low for non- cigarette tobacco products High tax for high tar structure on smokers, manufacturers and public health Kffect Public health inwlications Manufacturers' influence Beneficial if high limited,. Beneficial Manufacturers reduce size of cigarette Manufacturers increase size of cil]arette Manufacturers keep base level low Smokers switch to non- cigarette tobacco products Smokers switch to low~ tar Mostly beneficial Limited benefit Limits benefit of tax Slightly beneficial Source: Townsend 1998 Each type of tax has its advantages and disadvantages. Excise taxes are simple and easy to administer, but they may fall behind inflation if they are not regularly updated. Ad valorem taxes keep pace with inflation, but, because they are proportionate to the price of the product, the industry itself has some control over them, and may keep them low by keeping its prices artificially low. An increasing number of countries now earmark taxes raised on tobacco for anti-smoking activities or other specific social benefits. For example, one of China's largest cities, Chongquing, and several US states earmark part of the revenue from tobacco taxes for education about tobacco's effects, counteradvertising and other control activities. Other countries use earmarked tobacco taxes to support health services. Earmarked taxes have been unpopular with many public finance economists, who fear that they reduce governments' freedom to make the most efficient use of revenues for competing needs. However, others argue that the use of earmarked taxes for health purposes is consistent with the so-called "benefit principle" of taxation--that is, the idea that individuals should Draft: Not for circulation or citation 60 321423020

pay for their use of governments services according to their level of use. Since smokers are likely to need more health care than non-smokers in any given year of their life, earmarked taxes may be seen as a form of"user fee" for health care, just as earmarked highway taxes are a means of charging heavy road users more than light users. Because the tobacco industry is oligopolistic in many countries--that is, it has a small number of highly powerful manufacturers--there is evidence that its members are likely to collude to raise their prices when taxes rise, rather than compete for the lowest price and try to avoid passing on the costs to their consumers. Most research on the effect of tax increases on the industry's pricing policy is drawn from the USA, and within this country different analyses reach slightly different conclusions. However, the general finding is that tax rises are followed by prices rises of about the same amount as the tax itself, and sometimes by more than this amount. ,i The amount of tax charged varies sharply from country to country (Table 7.2). In the high-income countries, as a rule, taxes are high, amounting to two-thirds or more of the price of a pack of cigarettes. In general,,the lower-income countries raise lower taxes, amounting to not more than half the price of a pack of cigarettes. Thus, globally, prices vary sharply. At the end of 1996, a pack;of 20 cigarettes cost just 47 cents in South Africa, for example, compared with $5.23 in Norway. Table 7.2 Tobacco taxation in various countries, !99..3-97~ Pack of Cigarette Prices (US $) by income groups. Tax as % of price High Income Countries Australia 4.85 65 Belgium 3.32 75 Denmark 5.21 84 Finland 4.49 73 Italy 2.19 73 United Kingdom 4.16 78 United States 1.94 30 Middle Income Countries Upper Argentina 1.3 g 70 Chile 0.88 70 Slovenia1.08 63 Lower Albania 0.29 70 Bolivia 0.32 61 Thailand0.60 62 Turkey 0.51 42 Low Income Countries Armenia 0.20 50 Bangladesh 0.09 30 Cambodia0.05 20 China 0.20 38 India (white sticks) 0.37 75 Pakistan0.28 73 Zambia 0.65 30 Zimbabwe0.43 80 Draft: Not for circulation or citation 61 321423021

Tobacco tax is a significant source of revenue in some countries. In others, the proportion of government revenue raised through tobacco tax is well below its potential. The illustrative table below indicates the range in low-income and middle-income countries (Table 7.3). Table 7.3 Tobacco tax as a percentage of total government revenue I Country % of total government reven ue IHigh Income Countries [Australia 3.04 IDenrnark !.73 IFinland ! .73 Spain 2.20 United Kingdom ,i.98 United States 0.4 I Middle Income Countries Upper , Argentina 4.00 ' Brazil 4.88 ~ Chile 3.38 Greece 7.72 Lower Bulgaria 2.80 Colombia 0.73 Costa Rica ! .35 Egypt, Arab Rep. 0.78 Estonia 1.15 Lower Income Countries China 2.57 India 1.8 I Nepal 5.40 Zimbabwe 1.04 7.1.2 The bnpact of tax increases on demand A basic law of economics is that, as the price era commodity rises, demand for it should fall. In the past, researchers have argued that tobacco's addictive nature would make it an exception to this rule: smokers, according to this argument, are so addicted to smoking that they will pay anything for a cigarette and continue to smoke the same number of cigarettes to satisfy their needs, whatever the price. However, a growing volume of research now shows that this argument is wrong and that smokers' demand for tobacco is strongly affected by its price (Townsend, 1998). Studies have repeatedly found that price increases lead to some people stopping smoking, that they prevent others from starting in the first place, and that they reduce the numbers ofex-smokers who resume the habit. Draft: Net for circulation or citation 62 321423022

Figure 7. Canada Figure Z 1, The effects of price on consumption: evidence from the UK, South AJi'ica and l.a Real price and consumption of cigarettes in the UK, 1971- 96] Souse: To~md J 1998, C~i S~a~l O~cc (UK) (1965-93) Figure 7.1. b Daily consumptibh of c~arettes per capita among people 15 and over, and real price of tobacco, Canada 1950-91 (Son- we need to remove smuggling reference) 12 . ~ 20O 10 9 ~ 8 120 6 ~ ~ ~ Smuggling 80 5 I I t I I t 60 1950 1956 1~2 1~8 1974 1980 1986 Year Souse: To~s~d (i~8), Sweanor D ( 1985-91) Figure Z 1. c Cigarette consumption and real price of tobacco in South Africa 1970-89 Draft: Not J'or circulation or citation 63 321423023

Sours,': Towns~d J (1998)o Saloojee (1995) How addiction affects the response to higher prices Economic models that attempt to assess the impact of nicotine addiction on the effects of price increases make varying assumptions about whether smokers look ahead at the consequences of their actions or not. However, all models agree that, for an addictive substance such as nicotine, an individual's current consumption levels will be determined by their past consumption levels as well as by the current price of tobacco. This relationship between past consumption and current consumption has important implications for modelling the impact of price rises on demand. It implies that smokers will respond relatively slowly to l~rice increases, but that their response will be significant in the long term. The economic literature suggests that a real and permanent price increase will have twice as great an impact on demand in the long run as in the short run. Differing responses to price increases in low-income and high-income countries Poor households are generally more responsive to price increases in consumer goods than rich households, according to several studies. The proportion to which demand for a good is altered in response to a change in price is known as the elasticity of demand. Estimates of elasticity vary from study to study; however, in the USA, for example, researchers have found that a price rise of I per cent in the price of cigarettes decreases demand by about 0.4 per cent (an elasticity of-0.4). In lower-income countries, elasticity is probably greater than in higher-income countries. For example, a study in Sichuan, China, found a price rise of 1 per cent would be expected to reduce demand by between 0.65 per cent and 0.8 per cent (elasticity of-0.65 to -0.8). Another study in Taiwan suggested that a price rise of I per cent would reduce demand by between 0.5 and 0.7 per cent; a third study, based on national Chinese data, found even higher elasticity approaching I per cent. Studies in Brazil and South Africa have produced results in the same range (Abedian et al 1998). There are further reasons why economists would expect that people in lower-income countries would respond more to price increases than people in high-income countries. The age structure of the developing countries' populations is generally younger. Research from the high-income countries suggests, on the whole, that young people are more price- responsive than older people--partly because they have lower disposable incomes, partly because they may be less heavily addicted to nicotine, and partly because they are more susceptible to peer influences. Thus if one youiag person stops smoking because he or she can no longer afford to do it, friends are more likely to follow suit than amongst older age groups. Studies from the USA indicate that price elasticity of demand in young adults is higher than for smokers as a whole. A study by the Centers for Disease Control and Prevention found that demand elasticity among young adults was -0.58, higher than the figure for all smokers. Researchers conclude that not only are existing young smokers more likely to quit, but that fewer potential young smokers take up the habit, when prices Draft: Not for circulation or citation 64 321423024

are high. Finally, less educated groups tend to be more price-responsive than those with a high level of education. Since educational levels are generally lower in developing countries than in developed countries, price rises would appear to be an effective way of reducing demand. Based on the evidence cm'rently available, it is therefore reasonable to draw two clear conclusions: first, that tax increases are a highly attractive way to reduce tobacco consumption in low and middle-income countries, where most smokers now live. Second, that the effect of such tax increases will be more marked in these countries than in high- income countries. 7.1.3 The potential impact of tax b~creases on global demand for tobacco For the purposes of this report, re;searchers have modelled the potential impact of a range of tax increases on demand for t~igarettes worldwide. The design of the model and its inputs arc described in Box 7. I. The asstunptions on which the model is based, concerning price elasticity, health impact and other variables, axe highly conservative. Thus the results are likely to be underestimates of the potential impact of each intervention. The model reveals that eveh modest price increases could have a striking impact on the prevalence of smoking and on the number of tobacco-related premature deaths among those alive in 1995. The researchers calculate that a sustained real rise in price of cigarettes, by as little as 10 per cent over the average estimated price in each region, 40 million people worldwide would quit smoking. Given that not all of these quitters would avoid death, the number of premature deaths avoided is still extraordinary by any standards--I 0 million, or 3 per cent of all tobacco-related deaths, from this price increase alone. Nine million of the premature deaths avoided would be in developing countries, of which 4 million would be in East Asia and the Pacific (Table 7.4). If prices were increased by I 0 per c'cnt per year for a decade, the impact would be xxx xxx xxx x.[to come]. Table 7.4, Change in the number of smokers and premature deaths resuitbtgfrom cigarette price btcreases of lO attd lOOper cent, for smokers alive b~ 1995, by Worhl Bank region Region Change in # of smokers (millions) with price increase of." Change in # deaths (millions) with price increase of: 10% I00% % all smokers 10% 100% % all deaths EAP - 17 - i 66 41% .4 -42 42% ECA -6 -59 15% -I.5 -15 15% LAC -4 -38 9% -1.0 -I0 10% MNA -I.6 -16 4% -0.4 -4 4% SA (cigarettes) -3 -35 9% -0.7 -7 7% SA (bidis) -2 - 19 5°/~ -0.4 -4 4% SSA -3 -27 7% -0.7 -7 7% Low/Middle -36 -360 90% -9 -90 90% High -4 -42 10% - 1 - I 0 10% Draft: Not for circulation or citation 65 321423025

World -40 -402 100% -I0 -|00 100% Source: Ranson, K 1998 A bigger increase in price would bring further proportionate reductions. If the real, inflation-linked, price of a pack of cigarettes doubled, the model suggests that some 400 million smokers alive in 1995, or just over one-third of the total, would quit and I00 million deaths would be prevented, 90 million of them in developing countries. Again, it is stressed that d/ese are the minimum impacts of such. In either scenario, the biggest proportionate reductions in smoking prevalence would be among young adults. More than a quarter of the quitters would be in their twenties and about a fifth would be in their teens. With a price increase of l O0 per cent, it is expected that 80 per cent of smokers aged 15 to 19 would quit, compared with 40 per cent of smokers aged 20-29 and about a quarter of adults aged 30 or more. Clearly, again, because people in low-income and middle-income countries are exI~ted to be more responsive to price rises, the impact of the increases would be greater there than in high-income countries. Draft: Not for circulation or citation 66 321423026

Box 7.1 Estimating the impact of control measures on global tobacco consumption: the inputs to the model First, the researchers took estimates of the population in each region, with breakdown by age groups and gender, using standard World Bank population projections for the seven World-Bank regions, as before. (see table at the back of this report). Second, they estimated the prevalence of smoking, by gender, for each of the seven regions, using a compiled set of 87 studies from individual countries used by the World Health Organization (the data are shown in Chapter 3, Table 3.1). In the case of India, where bidis are a widespread alternative to cigarettes, the prevalence of both types of smoking was estimated. Third, using the available data, the team estimated the age profile of smokers in each region, extrapolating from large-scale individual country studies, and estimated the ratio of adult smokers to youth smokers. Fourth, the total number of smokers and the predicted number of deaths attributable to tobacco were estimated by region, gender and age. In this step, the researchers assumed that only one in three of smokers in developed countrieseventually die of their, habit. This assumption is conservative, given studies from the UK, the USA and elsewhere suggesting that the actual figure is one in two. For developing countries, where infectious diseases, injuries and other causes still account for a higher proportion of deaths than in developed countries, the researchers also made the conservative assumption that one-third ors .m. okers in developing countries will be killed by tobacco. This is likely to be an underestimat¢~, as recent research from China indicates that the proportion of smokers killed by tobacco wiil'soon equal that found in the West (Liu et al 1998). Next, the researchers estimated the number of cigarettes or bidis smoked each da)~ by each smoker in every region, using WHO figures and various published epidemiological studies. They also made estimates of the number smoked by adults and by youths in each region to arrive at a ratio of adult-to-youth daily smoking rate. The researchers then attempted to gauge the price elasticity of demand for cigarettes in each region, using data from more than 60 studies. Where more than one study had been done in any given country the resulting figures were averaged. The researchers combined the figures to arrive at averages t'or low-income and high-income regions. These figures were also weighted hy age, since young people are more price-responsive than older people. The short-run price elasticity for high-income countries was calculated to be relatively low, that is -0.4 ( ! ) (meaning that a I per cent increase in the price of a pack of cigarettes would decrease demand by 0.4 per cent), wbereas for low-income countries it was calculated to be -0.8 (meaning that a ! per cent price increase would reduce demand by 0.8 per cent). To take account of uncertainty about these values, the researchers entered a range of values for price elasticity on either side of these selected figures, from 75 per cent of the value to 125 per cent of it in each case. Next, the model took account of variation in price elasticity by age. The researchers assumed that, in line with one major study, half of the effect of a price increase would be on the number of people who smoke, and half would be on the number of cigarettes smoked by those who continued. Also in line with research evidence, they assumed that younger quitters would be more likely to avoid tobacco-related deaths than older quitters, and that the risks of tobacco-related death would persist for all continuing smokers, despite a reduction in the number of cigarettes smoked. All of the variables in the model were subjected to a sensitivity analysis to allow for uncertainty, with ranges of 75 per cent to 125 per cent of the baseline values used in the calculations. It Draft: Not for circulation or citaIion 67 321423027

should be stressed that the assumptions on which the model has been based are all very conservative ones, so that the results are likely to err on the low rather than the high side. [to be added: summary of me~hods for cost-eff~;tiveness estimates] for a fuller account of lhe model and it~ method~, see ~ page y. [box 7.1 ends] 7.1.4 Wili higher tobacco taxes have any adverse effects on the economy? A number of policy makers and the tobacco industry have raised objections to the idea of raising tobacco taxes. Each of these is considered in turn. Objection 1. Higher tobacco taxes will result in greater rates of srauggling between low- tax countries and high-tax countries. It has been argued, notably by tile tobacco industry itself, that higher taxes will contribute to increased cigarette smuggling and associated criminal activity, without any accompanying decrease in cigarette consumption (Joossens 1998). Tobacco smuggling is a serious problem, accounting for an estimated 6 per cent of cigarettes traded worldwide. Policy makers fear that cigarettes smuggled into countries where the price is relatively high and sold at.lower rates than those~vailable legally are likely, in theory, to force down the price of legal brands. However, while any increase in smuggling and the associated criminal activity is to be vigorously combated, the scale of the problem should not be overstated. The experience of several countries is illuminating. In Belgium, for example, the authorities increased taxes by 20 US cents per pack in 1993. The industry immediately put an advertisement in Belgian newspapers claiming that it would cause more smuggling and reduce revenues. In fa.et, tobacco tax revenue increased and consumption decreased. In the USA, the tobacco industry campaigned against a legislative proposal for a price increase of $1.10 per pack spread over five years. The increase would result in a black market for cigarettes, the industry argued. Yet, in under two years, the industry.had itself raised prices six times, by a total of at least 70 cents. In France, taxes increased nine times between 1991 and 1996, almost doubling the price of cigarettes. Sales fell by more than 10 per cent and the prevalence of adult smokers dropped by 6 per cent. But smuggling remained at a low level, accounting for about 2 per cent of the market. In Canada, taxes were increased sharply during the 1980s and early 1990s but reduced again in 1994 in response to a perceived increase in smuggling. Between 1979 and 1991, the price of cigarettes climbed by 159 per cent and teenagers largely abandoned smoking, with prevalence falling from 42 per cent to 16 per cent. Revenue rose. However, in 1994, the price of a pack of cigarettes was cut by a third as the government reduced taxes in the belief that this would reduce smuggling. The result was to increase teenage smoking again to 20 per cent, and for overall smoking prevalence to grow too. Federal revenues, Draft: Not for circulation or citation 68 321423028

meanwhile, fell by more than twice as much as predicted. The implication is that Canada's action had undesirable consequences, and that a better response might have been to crack down on criminal activity to control smuggling. In South Africa, excise taxes on cigarettes were increased rapidly during the 1990s, by 351 per cent. Smuggling rose too, from virtually nothing to about 6 per cent, the global average. However, importantly, total tax revenues climbed by 177 per cent, and sales dropped by more than a fifth. The example of South Africa reinforces the point that smuggling is a serious problem and that the relatively reassuring experiences of affluent northern economies should not be a reason for complacency about the future. But it also shows that tax revenues need not necessarily fall even if smuggling is at significant levels. Once again, the most effective response may be to enforce anti-smuggling laws aggressively. South Africa's experience also highlights the need for better data and monitoring in developing countries. So far, most econometric models of the impact of tax increases on smuggling have been confined to north America. Analyses from other countries are scarce. Limited data from a few small studies in developing countrieS, suggest that smuggling may increase with increased levels of tariffs and taxes. For the research underpinning this report, an econometric analysis was made to determine whether the methods used to study smuggling in the USA and Canada can be applied to other countries (Mcrriman ct al 1998). The study focused on European countries and, using a number of variables such as cigarette price, income per capita, and the incentive to import or export cigarettes in each country, it tested the outcomes of various different policies to give some guide to whether the incentives to smuggle would bc significant or not. If, for example, the price rises by a uniform 10 per cent across Europe, overall consumption would fall by 4.6 percent. The impact on trade between European nations would be somewhat more complex, but probably not significant. In a country whose cigarettes are already expensive, the increase of 10 per cent will bring a bigger absolute price rise than in a country whose cigarettes are cheap, and the high-priced country would probably see its exports fall slightly, while the low-priced country might see its exports rise. But because the price rise has been applied uniformly, no single country would see its exports drop very far. Thus, with uniform pricing and taxing structures in neighbouring countries, the incentives to smuggle are clearly reduced. In a second, contrasting scenario, one country increases its prices unilaterally by 10 per cent. Neighbouring countries do not. Domestic sales in the high-price country would drop by about 6 packs per person per year. But consumption would only fall by half as much, because smokers in the high-price country would partially compensate by buying cheaper cigarettes from neighbouring countries. In such circumstances, there would clearly be some incentive to smuggle into the high-price country. Draft: Not for circulation or citation 69 321423029

The analysis concludes that smuggling induced by price rises is likely to be a more significant problem in countries whose cigarettes are already priced high. Countries with relatively cheap cigarettes, including Bulgaria, Hungary, Italy, Poland, Slovakia and Spain would be relatively unaffected by price increases. However, the model takes no account of other factors that influence smuggling, such as the degree of tolerance of street sales. Importantly, the study found that, even with rates of smuggling several times higher than those actually reported in Europe, higher taxes would still result in higher overall prices and in increased tax revenues. In a separate study in the USA, researchers have examined the impact of interstate smuggling on tax revenues (Yurekli 1998). They find that the percentage of tobacco tax revenue lost due to smuggling declined during the 1980s but has climbed again to reach a level of 5.5 per cent in 1995. The increased smuggling activity is thought to be duc mainly to large variations in excise tax between neighbouring states, which create strong incentives to smuggle. '; In the f'mal chapter of this report, we examine some policy responses to the smuggling problem. ". : Objection 2. Tobacco tax increases dispr'oportionately affect the poor A second common objection to the policy of raising tobacco taxes is that they are regressive-that is, that they disproportionately affect people on lower incomes. Critics have suggested that families on low incomes might be caused significant hardship by higher cigarette prices. The first point response to this objection is that taxation systems as a whole are regressive, not individual taxes. Second, an increased tax on tobacco would not be expected to overburden the poor disproportionately, because the poor are more responsive to price ris¢s than the rich, and therefore their consumption of cigarettes is likely to fall more sharply. Thus, overall, even if tobacco taxes in themselves are regressive, increases in tobacco taxes may not be. Finally, since people on lower incomes suffer more tobacco-related illness than rich people, higher taxes may be argued to benefit poor people more by reducing their health disadvantage. Indeed, it has been suggested that the downward drift in the real price of cigarettes in the UK in recent decades may have effectively increased smoking levels in men in lower socio-economic groups relative to those in higher socio-economic groups, hence worsening their health (Townsend 1998). Further, it may be argued that tax increases are a more efficient way to benefit poor consumers than other control measures, such as counter-advertising or the publication of new health information about tobacco. As we have seen. richer and better- educated consumers are more likely to act on new health information than poor consumers. Studies in the UK and the USA in the 1990s have suggested that increased cigarette taxes would reduce the observed differences in the prevalence of smoking between rich and poor, because they would reduce consumption more among poor consumers than rich consumers. Draft: Not for circuh~tion or citation 70 321423030

Objection 3. Higher tobacco taxes will simply reduce government revenues Policy makers frequently argue against raising tobacco taxes on the basis that the resulting reductionin.demand will cost governments vital revenue. In fact, the reverse is true in the short to medium term: raised tobacco taxes generally result in higher government revenues. For example, studies in the UK suggest that a tobacco tax increase of 1 per cent provides an increase in government revenues of between 0.6 and 0.9 per cent (see Figure 7.2). What is equally clear is that, as the real price of cigarettes tends to become eroded over time, government revenues from tobacco tax tend to fall well below their potential, as was the case in South Africa in recent years (see Box 7.2). Fig 7.2, Real price and tobacco taxation revenue in the UK, 1971-1995 t TAX REVENUE ,! 8500 " :. ,_~. ,-~a~..I ~ 7500 ~ 7000 6000 I I I " I I I I 1971 1974 1977 1980 1983 1986 1989 1992 1995 Year £ 3.00 £ 2.80 £2.60 £ 2.40 £ 2.20 £ 1.60 £ 1.60 £ 1.40 Source: Townsend J A model developed for this study suggests that a modest ten percent increase in cigarette excise taxes would increase cigarette tax revenues by about seven percent, or about l 0 billion US dollars more globally in the 70 countries included in the analyses (Yurekli 1999). Since a price rise of I 0 per cent is modest, more ambitious price rises might be preferred by some policymakers, such as I0 per cent per year for I0 years. The impact in this case would be xxx xxx xxx xxx. Bibliographic no!e This section draws on Cha]oupka, FJ eta]. The taxation of tobacco produc~s. Background paper to the study; and Townsend, J. The role of~axation policy in tobacco conu'oi. Abedian et ai, 1998. Chaloupka, F J and Warner K. The Economics of Smoking. Prepared for the Handbook of Health Economics, eds Newhouse J and Culyer A, in press |chk]. Pekkurinen M. Economic Aspects of Smoking. National Agency for Kesearch and Health. P, csearch P,,epons 16/9. VAPK Publishing. Helsinki 1992.; Yurekli. A. [title to come] 1999. Draft: Not for circulation or citation 7] 321423031

Box 7.2 How declining tobacco taxes eroded optimal tax revenues in South Africa. Rotvena van tier Merwe The real price of cigarettes declined noticeably from the late 1960s to the late ]980s in South Africa. Concurrently, excise taxes also declined markedly as a proportion of the retail price of cigarettes, from around 45 per cent in the 1970s to 20 per cent in the early 1990st, showing that the Finance Ministry over that period did not let the tax increases keep pace with inflation. Not only did government not optimize revenue from this source, but by allowing a steady decline in tobacco taxes it in effect encouraged consumption. Not suq3risingly, cigarette consumption reached an all time high in the early 1990s. The price elasticity of demand estimate is particularly important because it helps determine the optimal level of cigarette tax. Apart from its role as a source of government revenue, excise tax is the single most important constraining influence on cigarette consumption in South Africa. The government therefore has an incentive to find a level of cigarette tax which reduces consumption (and so benefits public health and saves health care resources), without damaging the industry to the extent.that consumer taxes can no longer be levied on this product. A simulation model was established to show an optimal tobacco tax revenue for South Africa using calculated estimates of price elasticity of demand and income elasticity of demand. It revealed the revenue foregone by gov.emment due to its failure to maximize the excise tax opportunities available from cigarettes. The simulation involved maximization of govermnent revenue from excise tax on cigarettes, sul~j¢ct to the constraint that increasing cigarette tax increases the retail price of cigarettes, which in turn reduces quantity demanded. The simulation indicates that excise tax revenue could have been 157 percent higher in the first half of the 1990s. This translates into approximately 1.2 billion (1990 Rand) of additional receipts or 2 percent of total government revenue. [f the tax revenue was used specifically for health expenditure, approximately 10 percent more funds on average could have been allocated to the health budgett. The actual and simulated maximum real excise revenue in ! 990 Rands, is shown in the figure. The figure shows that excise taxes were near their optimal level in the 1970s, after which tax rates did not maintain their real value, causing increasingly significant levels of potential excise tax revenue to be foregone reaching its maximum in 1990 and 1991. This suggests that the taxation objectives of revenue and deterrence were not accomplished in South Africa over this period, because excise tax rates did not keep pace with inflation, thus allowing real cigarette prices to decline and consumption to increase. Since 1995, however, the South African government has committed itself to a stronger taxation policy, and has introduced increases in duty that have brought the effective cigarette tax rate to 50 per cent of the retail price in 1998t. box 7.2 ends i Va~ dcr Mcrwe. R. {1~98), "The Economics of Tobacco Control in So~lh Africa" in Abedian. I.. van der M¢rwe. R.. Jha, P. & Wilkins, N. (Eds.), The Economtca o/Tobacco Control: Tmrard~ m~ Optimal Policy Mix. Cape Town: Medical Association of South Africa Press. Draft: Not for circulation or citation 72 321423032

Actual and foregone real government exci~e revenue from cigarettes in South Africa: 1970 - 1995 20O0 1800 1600 1400 1200 Rands Million 19901000 8OO 6OO ' 400 2OO 0 yelr 7.2 Reducing demand with improved h, ealth information As we saw in Chapter 5, there is extensive evidence from the industrializedcountries that information about the addictive nature and long-term health consequences of tobacco use can help to reduce demand for it. The chapter concluded that knowledge mediates consumers' and potential consumers' demand for tobacco, but imperfectly, because: a) their knowledge is inadequate, b) they may fail to perceive the risks accurately, c) they may fail to internalize those risks, and d) most consumers start smoking when they are young and discount the future so heavily that their long-term health is of little concern to them. in this section, we shall look at the major types of health information: publicised research into the health consequences of smoking; warnings on cigarettes and cigarette advertisements; counter-advertising, and school education programs. We shall also look at the advertising and promotional activities of the tobacco industry. We shall summarize briefly what is known about the effectiveness of each of these activities in altering demand for tobacco. It is difficult to disaggregate the effects of different health information strategies, because most countries employ several at once, and it would be spurious to attempt to state their relative efficacy in any precise quantitative fashion. However, the findings summarized below convey the consensus of econometric research into the effects of each type of information on demand. 7.2.1 Health bformation Most of the information measures discussed here are introduced by, or financed by, governments. However, nongovernmental organizations and international organizations such as the World Health Organization are also major players. Each type of information measure is discussed briefly. Draft: Not for circalalion or citation 73 321423033

Publicised findings of research on the health effects Analyses of the impact of official reports on the effects of tobacco on health, such as the Surgeon General's report discussed above, consistently find a substantial effect on consumption. In the UIC, a report on the damaging effects of smoking by the Royal College of Physicians published in 1962 cut ~onsumption immediately by almost 5 per cent, with a subsequent slow recovery at around I per cent per year. Another report by the same organization in 1971 caused another drop in consumption of about 5 per cent, which again rose slowly back. Leu (1984) found that the publication of the US Surgeon General's Report affected consumption in Switzerland while Pekkurinen found a similar effect in Finland in response to two reports in the 1960s. It is also important to note the value to policy makers, both nationally and internationally, of studies that "count the tobacco dead". Without data for each region and nation on the health impact of tobacco, it is difficult for governments to act knowledgeably to control its consumption or plan health service needs. Increased efforts using simple data collection at national level are proving valuable, as the example of South Africa, China and India illustrates (see Box 7.3). Box 7.3 Counting tobacco deaths: Smoking on death certificates and proportional mortality analyses To be modified later with input from Alan an.d Richard. KEY POINTS 1. low-cost reliable methods are key to monitoring the epidemic. 2. proportion mortality studies require ~a) reliable cause of death by age and sex; (b) a question on smoking or other tobacco use about 5 years ago; (c) large sample size 3. Taken together these permit: odds ratio calculations (A- tobacco-death, smoker B:non tobacco-death smoker;C, tobacco death, non-smoker, D. non-tobacco death, non-smoker) with Odds ratio is made by multiplying A*D and dividing by B*C. 4. Such efforts are ongoing in :South Africa: smoking on death certificates since 1990, China: India: pilot efforts with Registrar General of India. 5. Questions as simple as . Was the deceased a smoker five years ago? [Yes / No / Do not know / Not applicable (minor) / Refused] . Was the informant a smokerfive years ago? [Yes / No / Refused]. The reason for choosing smoking status 'five years ago' is to elicit the smoking patterns of individuals prior to the onset of any disease that may have caused them to stop smoking. Furthermore, to study the effects of prolonged smoking, persons who started smoking less than five years ago would 'correctly' be classified, as non-smokers.. [box 7.3 ends] Draft: Not for circulation or citation 74 321423034

Warning labels Since the early i 960s a growing number of governments have required cigarette manufacturers to print health warnings on their products. By 199l, 77 countries required such warnings, although only a quarter of these countries insist on strong warnings with rotating messages (see Figure 7.2 for a prototype). A study from Turkey suggests that health warnings caused a decrease in consumption there of about 8 per cent over six years. A study from South Africa (Abedian 1998) found that warning labels contributed to declining per-eapila consumption. A second study found that, in the three years after 1994, when serious warning labels were introduced in South Africa, cigarette consumption fell there by 15 per ce,n/(Public Citizen 1998). According to the survey, 58 per cent of smokers were motivated by the warning labels to quit or reduce their tobacco consumption. This finding is taken as evidence that such warnings can produce a significant and rapid effect in developing .COuntries. Although it is argued that, in the more infor~ned populations of the industrialized nations, smoking prevalence is unlikely to fall much farther than it has already as a result of cigarette pack warning labels, evidence from Canada and Australia suggests that such labels can still be effective, provided that they are very large, very prominent, and contain hard-hitting and specific information.. In Australia, warning labels were strengthened in 1995. The impact appears to have been greater in inducing smokers to quit than when the older, less strongly worded, labels were used. In Canada, a survey in 1996 suggested that half of smokers intending to quit or cut back their consumption were motivated by what they had read on their cigarette packs. Figure 7.2 pendin~ (warning labels) Government counter-advertising and industry advertising Counteradvertising.There have been a number of studies to analyse the impact of negative messages about smoking on cigarette consumption. These negative messages, or counteradvertising, are disseminated by governments and health-promotion agencies. The impact ofcounteradvertising would be expected to reduce consumption, but econometric theory would expect its effect to diminish after a certain level of investment so that the rate at which consttmption falls flattens out. In fact, eounteradvertising does reduce consumption, according to studies at both national and local levels from North America, Australia, Europe and Israel (Saffer, 1998). In Switzerland, researchers conclude from a study of adult tobacco consumption between 1954 and 1981 that mass media anti- smoking publicity permanently reduced consumption by 11 per cent over the period. In Finland and Turkey, anti-smoking campaigns are also judged to have contributed to declines in consumption. 321423035

Research from the established market economics suggests that the content of counteradvcnising may affect its impact: for example, messages describing the health risks of smoking arc rarely effective when aimed at young people, who are typically unconcerned about the far future, whereas messages that depict the industry as manipulating its consumers arc more effective. There is some evidence that the ~axation of advertising itself can be an effective contr61 measure, since advertisers are responsive to price. A tax on tobacco advertising that is applied evenly to all media has an added advantage, in that it raises revenues that can be used to fund countcradvenising, while simultaneously preventing the industry from moving its advertising to other media. School anti-smoking educational programs School antismoking programs arc widespread, particularly in developed countries. However, they appear to be less effective than many other types of information dissemination. Even programs that have initially reduce the uptake of smoking appear to have only a temporary effect: they can delaythe recruitment of adolescent smokers by several years, but not prevent it. The apparen.t weakness of school-based programs may be less to do with their nature than the audience at which they are targeted. As we have seen, adolescent responses to information about the long-term consequences of an action on their health are not the same as adult responses due to rebellion, myopia and the early addictive properties of nicotine. 7.2.2 The role of the industry Research The tobacco industry has played an active role in funding and disseminating research that casts doubt on the links between tobacco and death. In the words of the tobacco company Brown and Williamson: "'Doubt is our product....if we are successful in establishing a controversy at the public health level then there is an opportunity to put across the real facts about smoking and health". Such research is actively promoted and disseminated and finds its way to legislators, decision makers and journalists. Frequently used tactics include review articles that cite only some of the relevant studies, mainly those that favour the industry, and a policy of writing letters to the editors of medical journals to rebut the conclusions of papers they have published providing evidence of the damaging health effects of tobacco. Letters to editors are not always subjected to peer review, but are cited in literature and logged by electronic literature databases, thereby giving them wide currency and apparent respectability. Draft: Not l'or cir'culation or citation 76 321423036

It is difficult to assess the impact of such information on overall consumer knowledge. However, it is likely that it has impeded individual assessments of the true risks of smoking and, because of its disproportionate impact on policymakers, it may have slowed the spread of government-initiated anti-smoking information campaigns. Advertising and promotion The tobacco industry spends an estimated 6 per cent of its sales revenue on advertising-- about 50 per cent more than most industries. Yet there is a large body of econometric research that finds that advertising has little or no effect on aggregate cigarette consumption. Here we examine this research and its methodology more closely and reassess the conclusions that are drawn from it. We show that more sensitive, non- aggregate measures, unlike aggrega, t~ measures, do find an effect. Thus, in contrast to the arguments advanced by the industry, advertising almost certainly does have a positive effect on cigarette consumption when methodological complexities are taken into account. The discussion also shows that a, dvenising bans do reduce consumption, but only if they are comprehensive and applied".to all advertising media; and that young people are a principal target of advertisers., ~Ve summarise the current state of knowledge on the impact of tobacco advertising, and discuss the implications for government policy. First, it is necessary to define what we mean by "advertising and promotion" in the context of the tobacco industry. Advertising is the use of media such as billboards, newspapers, television or radio to create positive imagery around a product, to create positive associations with that product, and to eormeet the product with features or events that are desirable to individuals. Thus cigarette smoking may be portrayed as a sophisticated and glamorous habit of attractive people who live desirable lifestyles. Promotion is a wide range of activities that increase sales. For example, cigarette manufacturers may pay shopkeepers to display their product in a prominent position, or they may sponsor sports or cultural events, or pay for consumer accessories such as baseball caps bearing their brand name. In the USA, where spending on cigarette advertising and promotion is monitored, the industry has decreased its advertising expenditure by 40 per cent in the past decade but increased its promotional activities by a slightly larger percentage, amounting to an overall increase in activities. Clearly, although there is much less research on the effects of promotion, it will become increasingly important to understand its impact. Measurh~g the relationship between advertising and consumption The idea that tobacco advert/sing does not recruit new smokers but merely moves them from one brand of cigarette to another is regularly used by the industry to justify its advertising activities. However, in practice, it is difficult to measure exactly how advertising does affect consumption, and the methods used will affect the results. Some of the measurement methods that are widespread are summarized here, and their effects on ihe overall conclusions of studies is outlined. 321423037

Aggregate data First, data on cigarette advertising expenditures may be measured as annual or quarterly aggregate national figures for all advertisers, in all media, in all regions. Yet such "lumping together" of data. may mask subtle variations in sales over time and across areas. This insensitivity may be particularly misleading in an industry such as the tobacco industry, which advertises heavily, for the following reason. According to economic theory, advertisers who continue to invest heavily and incrementally in their product should eventually expect to see ever smaller increments in their output. Therefore increases in sales that are the result of advertising might be expected to be small. Data that are aggregated by all advertisers, all media, and for all areas, may not show changes that would be significant at a more dis,aggregated level of analysis. This is indeed what happens. Fifteen' studies from the established market economies using aggregate annual or quarterly data show either no effect or a small positive effect of advertising on consumption. Reviews of this literature have concluded that advertising does not significantly affect consumption. However, it might be equally valid to conclude that aggregate annual data are an inappropriate measure for such .analyses. Data that show expenditure over time and place Second, data on expenditures may be available in cross-sectional formats--for example, within a state or a metropolitan area. These formats disaggregate data by place or time. Such data are more likely to show changes in both advertising expenditure and sales, because they will reflect factors such as local variations in the cost of advertising, and periodic short-term increases in the advertisers' expenditure for so-called "pulses" of advertising. Researchers analysing data held in these cross-sectional formats are more likely to find a positive relationship between advertising expenditure and cigarette consumption. Only three such studies have been performed, because they are obviously more expensive and time-consuming. These three show that when advertising is measured over a wide range, a positive effect of advertising emerges. The impact of advertising bans Third, researchers may use data on an advertising ban--for example, the removal of TV ads--to gauge indirectly the impact of the lost advertising on tobacco consumption. Yet this method is also problematic. If advertisers are prevented from using one medium they are likely to compensate by shifting more resources into the remaining media, such as radio, newspaper or billboards. Thus partial bans are unlikely to affect measures of overall consumption, and studies which use partial bans to conclude that all bans are ineffective are misleading. Draft: Not for circulation or citation 78 321423038

Three analyses of international data on bans have been published. They show that, betbre 1973, bans had no effect on consumption. But this finding fails to take account of the advertisers' immediate shift of resources from the banned medium, usually TV, into others. After 1972, the legislation on cigarette advertising became more restrictive and there were fewer alternative media available. The sole study that separates pre-1972 and post-1972 data finds that bans on television advertising reduced consumption. By contrast, a study that covers the period 1964 to 1990 but which does not control for the industry's shift into other media, and which makes no distinction between the pre-1972 and post-1972 data, finds no effect. These studies are interesting not only because they reveal the complexity of methods required to analyse the impact of advertising on trends in tobacco consumption, but because they have implications for po,!icy. It is clear that bans on cigarette advertising are ineffective if they are restricted to a single medium: to reduce consumption significantly, a ban must cover all or most media. A recent econometric study on 22 OECD c~tintries that made allowance for some of the methodological cottstraiats noted above found that a comprehensive set of tobacco advertising bans would reduce consumption by six percent, but a limited set of bans would have no effect. This study predicts that the European Union ban on advertising (see Box 7.4) would reduce consumption by nearly 7 percent. Indeed, a comparative study of cigarette consumption from more than 100 countries confirms the importance of comprehensive bans. In countries with comprehensive bans, cigarette consumption per capita has fallen by about 8 per cent whereas countries without comprehensive bans have decreased consumption by only about 1 per cent. When countries with a comprehensive ban arc compared with countries with no ban, there is a clear effect (see Figure 7.3). The methods discussed here show that it is a difficult and complex process to make accurate measurements of the impact of cigarette advertising on consumption. The difficulties are compounded by the fact that countries may implement several anti- smoking measures at once, making it virtually impossible to tease out the relative contributions of each to a decline in smoking prevalence. However, where methods are sufficiently sensitive to detect the kind of changes in consumption that could be expected in a heavily-advertised industry of this nature, we find that--in contrast to the industry's preferred argument--advertising does indeed have a positive effect on consumption, while counteradvertisiag has a negative effect. Where advertising bans are comprehensive, they significantly cut cigarette consumption. 321423039

Figure. 7.3, The impact of comprehensive advertising bans on cig~w¢{l¢ ¢o/~sttmptio~ (All countries wilh a comprehensive ban compared with countries with no ban) How tobacco markets are enlarged Despite the tobacco companies' claims that their advertising is not intended to recruit new smokers, their actions suggest that they obtain substantial return from investing in new markets. Tobacco companies are listed among the top 10 advertisers in 21 out of 50 countries in Europe, Asia and the Middle East. And in four Asian countries where American tobacco companies have invested in significant advertising in recent years, total tobacco use has increased by 10 per cent. Conventional advertising is not the only means that tobacco companies use to increase their market. One popular method is to increase the number of brands of cigarette that a company sells--a strategy that requires segmenting the market into subgroups defined by, for example, age, sex, attitudes or lifestyles, and then targeting brands, each with their specific advertisements, at those subgroups. For example, Marlboro cigarettes are portrayed to represent a rugged, wholesome outdoor male lifestyle. Virginia Slims are supposed to represent slim, sassy independence. This so-called brand proliferation can increase a company's sales and, theoretically, the total size of the market may be increased by increasing the number of brands, depending on the costs of proliferating and the available potential customers. Studies from the USA suggest that companies do indeed increase their market size when they proliferate brands, although some studies suggest that their market may increase at the expense of a competitor's. The tqrget audiences of advertisers 321423040

Researchers have studied the content of advertisements to identil) whether they are designed to appeal to specific groups, and to measure whether these groups respond. These suggest that young people are a principal target of advertisers-unsurprisingly, since the vast majority of smokers take up the habit in adolescence. A study of advertisements directed at women in the USA [?ck] found that adolescent girls started to smoke at an increased rate after these advertisements had appeared. A study of tobacco advertisements in magazines for young people found that they emphasized themes of risk and adventure, and a study of promotional activities by tobacco companies in California showed that young people were more likely to start smoking after exposure to these activities. However, there are methodological flaws in some of these studies, and more research is clearly needed, particularly in developing countries where tobacco companies are currently seeking new markets. The research discussed in this section suggests that various controls on advertising may be worthy of further consideration in specific settings. We have seen that, when more subtle variations in expenditure and sales are taken into account, advertising increases consumptioa, while couateradvertising anal'comprehensive--but not partial--bans decrease it. In addition, it is clear that the tobacco cor~panies are willing t? invest in advertising in new markets and that sales have increased in those markets following advertising campaigns. Whereas there i;~ a reasonable body of research on advertising, there is relatively little published information about tobacco promotion. Since promotion makes up an increasing proportion of the industry's activity, it is imperative that more research is undertaken, particularly in the new markets for tobacco outside the established market economies. BiblioRraphic note This section draws on Kenkel D. Chen L. Hu T-W. Bero L. Consumer information and tobacco use. 1998 [1999??ck]. Background paper to the study; and Saffer, H. The Control of Tobacco Advertising and Promotion. 1998. Background paper to this study. It also draws on Jha, P. The rationale for government involvement in tobacco control, and Jha, P. The role of governments in tobacco control, op.cit. Box 7.4 The European Union (EU) Tobacco Advertising Directive Luk Joossens In 1989, as part of a wider initiative against cancer, the European Commission proposed a Directive to restrict the advertising of tobacco products in the press and by means of billboards and posters. The European Parliament amended in 1990 the Commission's proposal and voted for an advertising ban. The Commission observed that they could only secure agreement for a partial ban at the time, but added that a new proposal for a total ban ,night be made, depending on progress achieved by individual Member States. In June I991 the Commission introduced a modified proposal for a Directive on tobacco despite massive lobbying against it by the industry. In the period 1992-1996 no progress was made in implementing the proposal because of opposition from at least three member states, Germany, the Netherlands and the UK. However, this opposition collapsed in 1997, when the Labour Party won the general election in the UK, with a manifesto commitment to introduce a tobacco advertising ban. The text of the proposed 321423041

Directive was finally adopted by the Council in June 1998. Directive 98/43/EC stipulates that all direct and indirect advertising (including sponsorship) of tobacco products will be banned within the EU, with full' and final enforcement of all provisions by 1st October 2006. Key points are: 1. All Member States of the European Union must introduce national legislation not later than 30 July 2001. 2. All advertisements in the prin~ media have to cease within one further year. 3. Sponsorship (with the exception of events or activities organised at world level) has to cease within two further years. Tobacco sponsorship of world events - such as Formula One motor racing - may continue Ibr a maximum further three years; but has to finish by I st October 2006. There must be a reduction in overall sponsorship support, and also voluntary restraint on tobacco publicity surrounding these events, during the phasing out period. Product information is allowed at points of sale. Tobacco trade publications may car~ tobacco advertising. Third country publications, not intended specifically for the EU market, are not affected by the ban. [box ends] Draft: Not for circulation or citation 82 321423042

7.3 Reducing demand by restricting smoking in workplaces and public places Clean indoor-air laws in the USA have been credited with reducing tobacco consumption there by between 4 and 10 per cent.(Yurekli 1998; Chaloupka and Warner 1998). However, studies of the impact of shch restrictions are complicated by the possibility that states that implement such bans do so because they already have a general social environment opposed to public smoking (Warner 1998). Data on the effectiveness of such bans outside the USA is scarce; however, there are reports that bm~s on public smoking are flouted in environments where tobacco is socially acceptable. For the time being, therefore, robust conclusions about the effectiveness of bans on reducing total consumption cannot be reached without further research. What is clear, however, is that where, vcr restrictions on smoking in workplaces and public places are observed, they have a distinct benefit for non-smokers, who are spared the nuisance, discomfort and laundering costs of exposure to environmental tobacco smoke, as well as reduced risks of any damage to their health. For this reason alone, smoking restrictions are a valuable component of gbvernments' responses to tobacco. The potential impact of non-price measures on global tobacco consumption The model described in section 7.1 (Box 7.1 ) was also used to assess the potential impact of a package of measures other than price rises on global demand for tobacco. This package of"non-price" measures includes: complete bans on advertising and promotion of tobacco products, related logos and trademarks; dissemination of the findings of research and other new information on the health consequences of smoking; and restrictions on smoking in public and in the workplace. Because there have been few attempts hitherto to estimate the aggregate impact of such non-price measures, the researchers made an extremely conservative assumption on the basis of evidence for each separate measure that together they might reduce demand by between 2 and 10 per cent--that is, that they would together persuade between 2 and 10 per cent of smokers to quit. The model assumes that such a package would have no impact on the number of cigarettes smoked daily by those who do not stop, and it also assumes that the package would be twice as effective among young people aged 15-19 as among older adults. This is due to econometric studies that show that young people are most responsive to advertising and promotion (Saffer 1998; Chaioupka and Warner 1998). The results indicate that if the package reduces demand by just 2 per cent, this would cause 23 million people who smoked in 1995 to quit, and would avert the deaths of 5 million smokers. If the package were to reduce demand by 10 per cent, the impact would clearly be five times greater (Table 7.5). Draft: Not for circulation or citation 83 321423043

Table 7.5. Worldwide change in number of smoking-attributable deaths, with non- price intervention~ of i 0% effectiveness, by age and gender. Age Males Females Males & Females Categories # (millions) % Total # (millions) % Total # (millions) % Total 15-19 -4.8 22% -0.7 14% -6 20% -(i 7%) -(i 8%°) .(i 7°/.) 20-29 -6.0 28% - 1.7 32% -8 28% -(9%) -(9%) -(9%) 30-39 -5.3 24% - 1.4 26% -7 25% .(7°/°) .(7%) .(7%) 40~9 -3.9 18% -1.0 19% -5 18%0 -(6%0 ~W0 ~/~) 50-59 -~ .6 7% ~0.4 7% -2 .(5%) ~5%) 60+ -0.3 1% -0. ~ 2% 0 2% -(~ %) -( ~ %0 ' ' ~ ~ %) TOT~ -22 100% -5 . ~I00% -27 100% -(~%) ~7%) -(~%) % Total 80% 20% 100% Source: Ranson, K. 1998 7.4 Nicotine replacement therapy Beyond the price and non-price measures discussed above, there is a third set of measures aimed at helping smokers to quit: these are various types of cessation aids and cessation programs, including individual training, hospital treatment, counselling programs and the growing market for nicotine replacement therapy (NRT) products. These products, such as gums and patches, enable smokers to obtain the nicotine they crave without smoking tobacco. Because of evidence suggesting that nicotine replacement therapy is more cost- effective than many of the alternatives, this section focuses most of its attention to it. Possible roles for governments in liberalizing access to NRT are discussed. Quitting benefits health, but only some methods work There is substantial evidence that quitting smoking brings major health benefits to smokers, especially at younger ages. Equally, there is convincing evidence that most smokers would like to give up their habit, at least in countries such as the USA, where knowledge of tobacco's health dangers is widespread. However, quitting is very difficult. Even in countries where most people know that tobacco is harmful, at best one in twenty smokers succeeds at quitting unaided on any given attempt. Equally discouragingly, the evidence suggests that various types of cessation program aimed at the smoking population as a whole, including public information, interventions by health professionals, and anti-smoking activities in the workplace, make only a marginal difference to the success rates of would-be quitters. Draft: Not for circulatiott or citation 84 321423044

In contrast, there is more encouraging evidence that intensive help for individuals, including nicotine replacement therapy, behavioural modification techniques, biofeedback, the involvement of{he smoker's family, and relapse prevention training, can be effective. One trial showed that 35 per cent of smokers offered a mix of these individual interventions managed to quit, compared with 9 per cent of those who had no help. Within these intensive individual approaches, however, NRT emerges as perhaps the most cost-effective and efficient aids to quitting. Smokers who use a pharmacological aid have a higher overall chance of quitting than other would-be quitters: their success rates range from 10 to 30 per cent. Equally significant, NRT has been found consistently to double a smoker's chance of success in quitting, with or without a parallel course of behaviour therapy. This suggests that nicotine replacement therapy alone may offer would-be quitters significant benefits even when no other help is available (Table 7.6 ). Table 7.6 Success rates of various cessation methods in the USA (N=50 million smokers) "Intervention Efficacy (% quit at Utilization (# using Impact (# quitters)* None NRT Prescript{on, 1995 6 months) 14 method annually) 122,800.000 2,500,000 ~84,006' 280,000 NRT Over the 14 6,300,000 560,000 counter, 1996 395,000 9'4,800 '500 24 Behavioural counselling Inpatient treatment 160 * sustained quitting (after six months)Source: Shiffman 1998 Clearly, the public resources'available to invest in anti-smoking treatments are not large in any country, and particularly not in the low-income and middle-income countries where 800 million of the world's 1.1 billion smokers live. For these reasons, NRT may be worth consideration by governments as a potentially more cost-effective intervention than those requiring intensive human input. In the remainder of this section, we assess the current availability and affordability of NRT to smokers worldwide. We discuss the implications of this growing market for public health and set out the issues facing governments in all regions. Nicotine replacement therapy products Nicotine addiction is the principal reason why most adult smokers continue to smoke. Yet nicotine itself is almost certainly less harmful to health than the tobacco that is its vehicle. The lung cancer, heart disease, and many other chronic and fatal diseases that affect smokc~rs are attributable to tobacco and its combustion products, not nicotine. Draft: Not for circulation or citation 85 321423045

NRT products been shown to be effective in dozens of studies. They have not been [inked with any cardiovascular or respiratory disease. They are prescribed solely for treating the symptoms of nicotine withdrawal in smokers attempting to quit. However, they are available over the counter in many countries and different smokers are likely to put them to different uses. Some people will manage to quit with only relatively short-term use of NRT. Others will never quit, but will use NRT to reduce the number of cigarettes they smoke and to cope with extended periods in Which they may not smoke, such as during air travel or in workplaces. Major players in the NRT market Until the spread of NRT, tobacco companies had an effective monopoly on the marketing of nicotine--a market that they continue to dominate, with sales of $300 billion per year for cigarettes alone compared with arbund $725 million for NRT in 1998. However, the emergence of nigotine replacement therapy products, which are manufactured by the pharmaceutical industry, has altered the market situation considerably. There are important differences between NR.T products and cigarettes. First, smoking tobacco produces a rapid nicotine "fix" whereas NRT products release smaller amounts of nicotine more slowly. Second, NRT products contain only nicotine, whereascigarettes contain a number of other substances that are known to be much more harmful to health. Third, partly because of their different industrial sources, cigarettes and NRT products are subject to very different market regulations. The market for these products is highly regulated because smokers must visit a pharmacist to buy them. By contrast, the cigarette market is virtually unregulated throughout the world. Fourth, and importantly, as stated earlier, NRT products are designed for short-term use to aid quitting, whereas cigarettes tend to lead to long-term nicotine addiction. Availability and affordability Nicotine replacement therapy products are widely available in the high-income countries where, in countries such as the USA and the UK, they are sold over the counter. Availability is variable in middle-income countries: for example, NRT products are sold in Brazil, Argentina, Mexico, Malaysia the Philippines, Indonesia, Thailand and South Africa, but in widely varying quantities, whereas in some middle-income countries they are not available at all. The very limited data currently available suggest that they remain a very small part of the nicotine market, and that sales are highly variable. However, much firmer monitoring is needed before general conclusions can be drawn. The price of NRT products is currently roughly equivalent to average daily doses of tobacco. However, because they are sold in large quantities, aimed at quitters, individuals must find a relatively large sum to spend on them at any one time. For example, in the USA, a full course costs between $180 and $275 retail. This comparatively large outlay might deter their use by smokers in low-income and middle-income countries. However, the current price may be artificially high, driven upwards by the requirement for Draft: POor for circulation or citation 86 321423046

specialised sales outlets, such as pharmacies, for these products. Since in many low- income and middle-income countries the norm is for people to buy their own medication over the counter and approach pharmacists, rather than doctors, for medical advice, it is reasonable to believe that the market will grow in these countries. The cost-effectiveness of nicotine replacement therapy has not been studied widely, especially in the low-income and middle-income countries where most smokers live. It is clear that more information on cost-effectiveness would be useful, both in determining whether these devices should have a claim on limited public funds, and in giving policy makers a firmer basis on which to act. Regulating the market: options for governments While the existing evidence indicates that NRT used properly is safe, nicotine is a psychoactive substance and most societies traditionally limit the sale of psychoactive substances as so-called "demerit products". The usual forms of regulation fall into four categories. First, it is possible to regulate the product, specifying standards on its purity, safety, packaging, strength and so on. Second, it is possible to regulate the seller or provider, for example by making the produ~:t available only thrbugh government outlets or on prescription. Third, the conditions of sale can be regulated, for example by limiting the hours of sale and imposing sanctions on a seller whose consumers come to harm. Fourth, the consumer may be regulated, for example by demanding the buyer is over a certain age. For nicotine, it is clear that these conditions are followed imperfectly and inconsistently, depending on whether the nicotine is delivered by NRT products or by tobacco products. Tobacco products are rarely subject to regulation by the first approach, in terms of health or manufacturing standards. Only 36 countries demand that cigarettes are labelled for their tar and nicotine content. By contrast, pharmaceutical products are subject to a long approvals and licensing process and to packaging and labelling restrictions. Where the second approach--the regulation of the seller--is concerned, there are also obvious differences. Cigarettes are sold widely and only in a few countries are sellers subject to licensing. NRT products are usually available only through pharmacists. Under the third type of regulation, concerning conditions of sale, once again tobacco is available widely, while NRT product sales are more heavily restricted. Also, because these products are classified as drugs, any minor modifications must be subject to the same approval process that all drugs follow. Under the fourth type of regulation, concerning consumers, there are further inconsistencies. Tobacco buyers a.r.e increasingly subject to regulation; for example, tobacco products could not be bought by minors in 43 of 134 countries surveyed by the WHO in 1995. However, in the countries where smoking is on the rise, regulations on tobacco buyers are relatively uncommon while regulations for the sale of NgT products are comparatively tight. This imbalance may be artificially reducing smokers' access to a relatively effective aid to quitting. Draft: Not for circulation~ or citation 8 7 321423047

Given this situation, there are several broad options for governments. At extremes, policy makers might decide on the one hand to regulate nicotine in all delivery forms in an integrated fashion. Alternatively, they might open up the nicotine market entirely, de- regulating the sales of both tobacco and NRT products. It is important to be aware that an entirely open market for nicotine could have both benefits and risks. The external costs of smoking would be driven downwards and there would a reduction in the harm to public health. However, there could be disadvantages if, for example, NRT became a gateway to tobacco use among the young; if existing smokers thought that NRT would allow them to go on smoking for longer, or if lapsed smokers were seduced back into the tobacco market through it. In reality, therefore, the preferred strategies of most policy makers would avoid the extreme positions. Realistic strategies would probably include discouraging the use of the most hazardous nicotine-delivery vefiicles, while making less hazardous products more widely available to adults. Thus rational regulations should be applied both to tobacco products and NR.T products. Practical options that governments could c0.asider would include, for example, removing barriers on NRT products that limit their availability, their advertising and promotion, and their package size. If nicotine replacement therapy is shown to be cost-effective in certain situations and for certain users, governments might consider providing finance to support their use, or even direct provision of these devices to poor smokers. At the same time, governments of countries whose healthcare is financed by insurance or taxation systems might consider extending cover to NRT in certain limited groups of low-income smokers, perhaps for limited periods. Finally, international agreements to harmonize registration procedures for the pharmaceutical industry could be beneficial. The current plethora of registration procedures slows down the approval and marketing of drugs, and may be partly responsible for the relatively slow growth of the NRT market to date in developing countries. The potential impact of nicotine replacement therapy on global tobacco consumption The model used to estimate the potential impact of higher taxes and non-price measures on tobacco consumption was also used to assess the potential of nicotine replacement therapy. The methods used to design the model and the inputs to it are described in Box 7.1. Making the conservative assumption that widely available nicotine replacement therapy would reduce demand by between 5 and 10 per cent, that is, enable 5-10 per cent of smokers to quit, the impact would be considerable. If the effectiveness of the therapy were 5 per cent, 14 million smoking-attributable deaths would be averted worldwide. With 10 per cent effectiveness the number would be twice as high (Table 7.7). The relative impact of nicotine replacement therapy would be expected to be greatest in individuals aged 30 to 49 years, among whom 14 per cent will quit, compared with 7 per cent in other groups (Ranson 1998). This age group would be expected to account for almost two-thirds of the averted deaths. About 80 per cent of the Draft:/Voffor circulatiott or citation 88 321423048

quitters and averted deaths would be expected to be in low-income and middle-income countries. Table 7.7. Worldwide change in number of smoking-attributable deaths, with liberalized access to nicotine replacement therapy of 10% effectiveness, by age and gender. Age Males Females Mates & Females Categories #(millions) %Total # (millions) %Total # (millions) %Total I5-19 -2 9% -0.3 5% -2 8% -(7%) -(6%) -(7%) 20-29 -5 21% - 1.3 23% -6 22% -(7°/°) -(?'/o) -(7O/o) 30-39 -8 37% ,~2.0 38% -I0 37% -00%) -00%) -00%) 40-49 -6 27% -1.5 27% -7 27% -( t O'/o) -( ~ 0%) -( t 0%) 50.59 -! .2 6% -0.3 ~ ", 5% -1.5 6% -C3%) -(3%) ~ -(3%) 60+ -0.2 !% -0. ! 2% -0.3 1% -(1%) -(1%) -C i %) TOTAL -22 100% -5 100% -27 100% -0%) -(7%) -(7%) % Total 80% 20% 100% Source: Ranson, 1998. Note that additional analyses will incorporate drug costs and not policy changes alone. 7.5 Conclusion The discussion has reviewed the evidence for the effectiveness of three measures to reduce the demand for tobacco: first, by imposing higher prices on tobacco and tobacco products through raised taxes; second, by implementing a package of"non-price" measures comprising better and more widely available information about the health effects of using tobacco, the banning of industry advertising and promotion, and the implementation of restrictions on smoking in public places and workplaces; and, third, by widening access to nicotine replacement therapy. In line with the experience of the industrialized counu'ies, the report concludes that a highly effective means of reducing demand is to raise taxes on tobacco products so that their real price rises. The impact of this intervention is likely to be significant on most consumers, but especially among the young and on others who are particularly responsive to price rises, notably people on lower incomes. A very modest price increase of 10 per cent could, according to the model described here, persuade 40 million people to quit smoking worldwide and prevent l 0 million smoking-attributable deaths among smokers alive, in 1995. These estimates are conservative. A more ambitious price rise of 100 per cent would clearly have an even greater impact. Draft: Not for circulation or citation 89 321423049

By implementing a package of measures to inform consumers and regulate smoking in certain places, even with a highly conservative set of assumptions about their effectiveness, 23 million people could be persuaded to quit and 5 million deaths could be averted. By widening access to nicotine replacement therapy, 58 million adults might be persuaded to quit and 14 million smoking-attributable deaths among smokers alive in 1995 might be prevented. It is important to point out that, on the basis of currently available evidence, it is not possible for researchers to assess whether the measures used together have an additive effect or whether they may overlap. In other words, it cannot be concluded from this model whether the total number of deaths averted would be as many as 29 million (I0 million plus 14 million plus 5 million) or fewer. Clearly, as the discussion shows, different measures will have greate~;¢r lesser effects on different age groups and regions, and much more evaluation reseaxch would be required before more detailed conclusions can be drawn. The discussion shows, however, that several, different measures to reduce demand can be highly effective. In the following chapters We shall assess tl~eir cost-eflbctiveness and discuss their implications for policy-makers. B ibliogra~hic This section draws on Novotny T, Cohen J and Sweanor D. Cessation programs and nicotine replacement markets. Background paper to this study; and Ranson K. Potential impact of price increases and other tobacco control policy interventions. Background paper to this study. 321423050

8. Most measures to reduce supply will fail The previous chapter examined the effectiveness of measures to reduce the demand for tobacco and concluded that tax rises, wider access to nicotine replacement therapy and better availability of health information can each play a substantial role. We turn now to a brief discussion of measures intended to reduce the supply of tobacco, once again based on the evidence available from a number of countries. As the discussion makes clear, most supply-side measures are unlikely to reduce tobacco consumption. Before discussing different measures in turn, it is worth comparing efforts to reduce the supply of tobacco with the war against illicit drug use in the USA. Supply-side measures against narcotics, such as domestic law enforcement and efforts to control crop production in source countries, have been less effective than originally hoped. One study shows, for example, that variations in the level and intensity of law-enforcement activities have no link with fluctuations in either the price or use of cocaine in the USA (Dinardo, 1993). Similarly, crop eradication, crop substitution and refinery destruction in major source countries have failed to reduce the t~ow.of cocaine into the USA. In contrast, efforts to reduce the.demand for cocaine through medical approaches such as treatments for drug users have proved more successful and more cost effective. For example, also in the USA, good treatment regimes can reduce cocaine consumption 7 times more effectively than domestic law enforcement. Sensitivity analyses show that these results are robust (Rydell et al, 1994, 1996). The history of supply reduction or control in the International Coffee Agreement, and for other commodities is of limited or no success (Economist, June 9, 1984; Bohman et al, 1996). While there are clearly differences between illegal drugs and tobacco, the lessons for tobacco control are significant. Various specific measures are now evaluated here. 8.1 Policies with a low probability of success Crop substilution and diversification Worldwide, more than 100 countries grow tobacco, of which about 80 are in developing countries. Four countries between them account for two-thirds of the total production: in 1997, China was responsible for 42 per cent of all tobacco grown, with the USA, India and Brazil producing about 24 per cent between them. The top 20 countries between them produce more than 90 per cent of the total (see Table 8.1) Draft: Not for circulation or citation Q| 321423051

Table 8. I The lop 30 ~obacco producing countries 1997 Country Area Share of Produc Produc Share of (1000 World tion tion World hectares) Total (1000 Chemge Total (Percent) n|etri¢ (PercenO 1997 1997 1997 1994-97 1997 Export Import RalioI Ratios (Purc'en0 Tobacco Export czs percent of Total Exports US$ I000 (Percent) 1995 China !,880.0 38.4 3,390.0 51.5 42.12 United States 328.4 6.7 746.4 4.0 9.27 India 420.2 8.6 623.7 18. i 7.75 Brazil 329.5 6.7 576.6 30.5 7.16 Turkey 323.0 6.6 ,; 296.0 57.7 3.68 Zimbabwe 99.3 2.0 192.1 g.0 2.39 Indonesia 217.5 4.4 184.3 15.2 2.29 Malawi 122.3 2.5 158.6 61.7 i.97 Greece 67.3 !.4 132.5,, ,. -2.2 1.65 Italy 47.5 1.0 131.4 0,3 i.63 Argentina 71,0 1.5 123,2 . ~ 50.3 1.53 Pakistan 45.9 0.9 86.3 - i 4.0 1.07 Bulgaria 48.5 1.0 78.2 124.3 0.97 Canada 28.5 0.6 71. I -0.5 0 ,gg Thailand 47.0 1.0 69.3 17.4 0.86 Japan 25.6 0.5 68.5 -13.8 0,85 Philippines 29.4 0.6 60.9 8.7 0.76 South Korea 27.2 0.6 54.4 -44.8 0.68 Mexico 25.4 0.5 44.3 -35. I 0.55 Bangladesh 50.3 1.0 44.0 -26.7 0.55 Spain i 3.3 0.3 42.3 0. I 0.53 Poland 19.0 0.4 41.7 -3.3 0.52 Cuba 59.0 1.2 37.0 I 17.6 0.46 Moldova 17.2 0.4 35.8 - 15.8 0.45 Vietnam 36.0 0.7 32.0 N/A 0.40 Dominican Rep 21.2 0.4 30.3 4 ! .7 0.38 Macedonia 22.0 0.4 30.0 N/A 0.37 Kyrgyzstan 12.0 0.2 30.0 -33.3 0,37 South Afriea 14.9 0.3 29.0 - 1.4 0,34 Tanzania N/A N/A 25. I 15. I 0.3 I World Total 4,893.8 100.0 8,048.4 25.9 100.0 Not=: I. Ratio of exports to domestic production. 2. Ratio of imports to domestic production. 3. Less than 0.1 percent. N/A = not available USDA (1998); FAO (1998); [EC (1998) 2.9 4.7 0.68 35.5 7.4 0.55 23.2 s 0.44 77.0 0.2 2.55 89.3 0.5 I. 17 109.7 ~ 23.05 10.2 27.6 0.42 74.2 J 60.64 74,5 12.8 2.05 78.7 18.3 0.04 60.6 5. I 0.59 1.6 J 0.08 53.5 58.3 5.40 24.0 12.6 0.04 48.5 15.3 0.I I 0.5 145.4 0.04 17.2 18.3 0.17 8.4 26.2 0.02 31.8 8.3 0.11 ~ 16. I 0.03 53.9 126.7 0,06 6.9 66.4 0.12 13.5 0.8 N/A 61.4 6.7 6.90 N/A N/A 0.04 58. I 2.2 5.26 N/A N/A 5.44 76.7 3.3 6.96 41.5 55.5 0.31 55.8 ~ 4.53 25.3 24.4 Whereas China uses most of its tobacco crop for its domestic market, other major producers export large proportions of theirs. Brazil, Turkey, Zimbabwe, Malawi, Greece and Italy all export more than three-quarters of what they grow. Only two countries worldwide are dependent on tobacco for their export earnings--Zimbabwe, with 23 per cen~ of export earnings, and Malawi, with 61 per cent. A few other countries--Bulgaria, Moldova, the Dominican Republic, Macdeonia, Kyrgyzstan and Tanzania--rely heavily Draft: Not for circulation or citation 92 321423052

on tobacco as a source of foreign exchange. None, however, has marc than I per cent of the global tobacco-growing market and most have less than 0.5 per cent. Tobacco is clearly a major eamer for a few countries with heavily agrarian economies, including MaJawi, Zimbabwe, India and Turkey. Regardless of the health arguments against tobacco, it remains a highly attractive crop to farmers, providing a higher net income yield per unit of land than most cash crops and substantially more than food crops. In the best tobacco areas of Zimbabwe, [br example, tobacco is approximately 6.5 times more profitable than the next-best altemative crop (Maravanyika, 1997). Farmers also find tobacco an attractive crop for more practical reasons. First, obviously, with price supports and subsidies, the price of tobacco is relatively stable compared with other crops. This allows them to plan ahead and obtain credit for other enterprises as well as tobacco farming. Second, the tobacco industry generally supplies farmers with strohg in-kind support, including materials and advice. Third, the industry often gives farmers loans that they are unlikely to be able to repay, thus preventing them from diversifying into other crops. Fourth, other crops may cause farmers problems related to storage and markets. Tobacco is non-perishable and the industry may assist delivery or collection of the crop; by contrast, other crops are perishable and may be blighted by late collection, [ate payment,and price fluctuations. There have been a number of experimental schemes to substitute other crops for tobacco (Van dcr Merwe 1998). However, with the arguable exception of Canada (Crescenti 1992), there is no hard evidence that these schemes succeed as a means of reducing tobacco consumption, because of the lack of motivation for farmers to participate while current tobacco prices persist. What about subsidies on tobacco production? While developing countries tend to tax export earnings from tobacco, many industrialized countries and China have a tradition of providing price supports and other subsidies to farmers who grow it. The motives for subsidizing tobacco production include keeping prices high and stable, supporting small family farms, controlling the importing of tobacco from abroad to conserve foreign exchange and maintaining political support. Often these subsidies go hand in hand with import restrictions. With these producer price support polices, governments artificially raise domestic tobacco price and that of tobacco products. Economists have argued that, whenever the price is raised in this way, smokers may respond by reducing their consumption. However, this beneficial effect is a small one. In countries such as the USA, where the producer price of tobacco leaf accounts for only a small pan of the price of cigarettes and where imports of lower-priced tobacco are rising (Ginsberg 1999) such subsidies will make only a negligible difference to the price of a pack, raising it by 1 cent. An increase of this order will reduce consumption by only 0.23 per cent (Zang 1998). Draft: Nat for circulution or citation 93 321423053

The impact of removing these producer supporls on global production is unclear. Higher domestic prices in the US may help to raise the global price of raw tobacco leaf, and thus offer better returns to farmers in low-income countrics (Lcwitt, 1987). On the other hand, lowering the gap between domestic and global price could permit more imports of raw tobacco from low-income countries, benefiting farmers in countries such as Malawi and Zimbabwe. Morcover, there are sound agricultural reasons for removing subsidies and import barriers and letting all farmers worldwide compete fairly. Help for farmers Farmers in a selected number of low income countries would stand to lose most from any supply-side interventions. However, ,with supply reduction responding to demand reduction, the transition will be a slow process stretching over many decades. This means that the costs of adjusting supply as demand diminishes would also be stretched out, and much less difficult to deal with than would be the ease if there were drastic reductions in supply (quote from EIU pending). Studies in,the US suggests that the biggest impact would be that the children of today's tobacco 'farmers would be less likely to stay in farming from demand reduction. Even if per-capita demand for tobacco were to fall quite steeply in coming decades, absolute tobacco consumption is likely to stay significant because of the growing number of adults in the global population. Estimates done. for this report suggest that even if cigarette prices rise 25 percent, and cigarettes per smoker fall 14 percent by 2020, the absolute market would remain about i billion smokers, or about the same as today, although total consumption would be lower. Finally, governments and developmental agencies may find may consider it prudent to ease the gradual, decades-long process of transition for farmers and other tobacco workers by providing broad rural development support programmes, assistance with crop diversification, and other "safety net" systems. Such support could be financed out of earmarked tobacco taxes (see Box 10.1 ). Restrictions on international trade will not work, but symmetrical regulation can work The principles of free trade arc based on certain assumptions: that free trade results in gains, both by improving consumers' options and therefore their welfare, and by making production as efficient as possible. These gains are desirable provided that there are net benefits to society. In general, while free trade appears to have had little impact on job numbers in the high-income countries, there is evidence from a study of 30 low-income and middle-income countries that free trade in goods and services overall has increased their growth rates. In principle, increased growth, economic development and higher incomes are associated with better overall health. While the arguments in favour of free trade'in general, then, are robust, tobacco is clearly specifically harmful to health on a scale unprecedented among traded consumer goods, and merits specific controls that do not undermine the overall benefits of free trade. 321423054

As we saw in Chapter 3, trade liberalization has contributed to an increase in the consumption of tobacco in certain countries, particularly in Asia, in recent years. Econometric models suggest that the removal of trade barriers will bring significant increases in consumption in low-income and middle-income countries, but not in high- income countries. However, trade liberalization has also resulted, through the GATT agreement, in an international response that gives countries the fight to adopt and enforce measures to protect public health-provided those measures are applied equally to domestic and imported products. Article XX of the General Agreement explicitly states that measures necessary to protect human health shall not be prevented by the requirements for free trade. , ~ In 1990, Thailand attempted to ban cigarette imports and advertising, a move that prompted a challenge from US tobacco companies. A GATT panel investigated the situation and ruled that Thailand could not" ban imports of cigarettes, but that it could impose ad valorem taxes, advertising bans ahd price restrictions, and that it could demand that all manufacturers whose products were available in Thailand should label their products with strong warning labels and descriptions of the ingredients. The GATT panel even said that Thailand could ban sales of all tobacco products in the country, provided the restriction was applied symmetrically to domestically produced and foreign-produced cigarettes. Thailand implemented strong demand-reduction measures and, in the late 1990s, its national tobacco consumption has indeed begun to decline again (ref). The This landmark decision, and Thailand's prompt and aggressive controls has been seen as setting a precedent that allows countries to intervene to control tobacco on public health grounds, while remaining within the principles of free trade. Since 1997, litigation in the USA against the tobacco companies has made the domestic market less favourable, raising questions as to whether the industry will simply adopt even more aggressive policies of opening up new markets in developing countries. As part of the studies underlying this report, economists have investigated the likely impact on low-income and middle-income countries, with some surprising and somewhat reassuring results (see Box 8.2) [to come]. [[Box 8.2 To come. Will cover the "economics of litigation in the US- and overseas exoansion ]] Consider box on privitatization of tobacco companies- impact on consumption and control? Consider a box on why attempts to ban Gutka in India failed? Rest~'ictions on youth access to tobacco F)rttft: Nnt fr~r rirculatlnn or citntinn O~ 321423055

There have been a number of attempts to impose restrictions on the sales of cigarettes to teenagers. Unfortunately, the balance of the evidence from industrialized countries is that these restrictions are ineffective and that enforcing them is not cost-effective. In low- income countries where tobacco consumption is rising, the necessary systems, infrastructure and resources for implementing such restrictions and enforcing tl~em are even less likely to be available, further casting doubt on the validity of this approach to reducing the tobacco supply on a global scale: [to be expanded] 8.2 Control of smuggling remains a priority As we saw in Chapter 7, smuggling is a serious problem, especially where there are large variations in tax between neighbouri,n.g states or countries and where there is cultural tolerance of contraband sales. But while it is undoubtedly important to deal with this criminal activity with appropriate vigour, its impact on economies should not be overstated. Nevertheless, smuggling controls represent one clear way in which governments may reduce the supply of tobacco. t Large-scale tobacco smuggling relies on criminal organizations, ~omparatively. sophisticated systems for distributing smuggled cigarettes in the destination country, and a lack of control on the international movement of cigarettes. Significant sums of money are involved: organized smugglers can buy a container of 10 million cigarettes, on which they pay no taxes, for $200 000. The fiscal value of this quantity of cigarettes in the European Union is at least $1 million, taking account of excise duties, value-added tax and import taxes. The profits to smugglers are thus so high that they can absorb long- distance travel costs. Most smuggled cigarettes are well-known international brands. Cigarettes are usually smuggled in transit between their country of origin and their official destination. To encourage trade between countries, a so-called transit system operates that temporarily suspends custom duties, excise and VAT payable on goods originating in country A and bound for country B while they are in transit through countries C, D, and so on. However, many cigarettes simply fail to arrive at their destination, having been bought by fraudsters. Similarly, large-scale operations can take place through so-called "round-tripping" where there are relatively large price differentials between neighbouring countries. Exported cigarettes from the USA, Brazil and South Africa, for example, have been documented entering neighbouring countries and then reappearing in their country of origin at cut-rate prices, untaxed. The success of smuggling relies on the cigarettes passing through a large number of owners in a short time, making it virtually impossible to track their movements. Additionally, poor enforcement of illegal sales and difficulty in separating legal and illeg~il sales may reduce the risks to smugglers. In Russia. and in many developing countries, the majority of cigarettes are sold on the streets. 321423056

The role of the industry The tobacco industry formally denies involvement in smuggling. However, there have been a number of incidents where officials employed by the industry have faced charges for crimes that have allowed others to smuggle cigarettes. In December 1998, a marketing subsidiary of the company R. J. Reynolds pleaded guilty to criminal charges and paid $15 million for actions that allowed others to smuggle cigarettes into Canada. The subsidiary admitted only to customs violations, not smuggling, but prosecutors are reported to have said that the subsidiary knew that the cigarettes were destined to be smuggled to Canada. In another incident in 1997, two sales managers for Brown & Williamson pleaded guilty to the charge of aiding smugglers illegally bringing cigarettes into Canada by supplying them with untaxed cigarettes from a ~bondcd warehouse belonging to the company in Alabama. In Hong Kong, an executive of British American Tobacco was convicted of accepting bribes from a smuggling syndicate. The judge in the case was highly critical of the tobacco company for becoming involved in smuggling tobacco into China. Other incidents suggest that the industry is a't least aware of the problem. For example, the South African tobacco conglomerate Rembrandt Group, Ltd, sued Philip Morris, charging that Philip Morris was supporting the illegal smuggling of cigarettes into South Africa and thereby violating a licensing agreement between the two companies. Other' indications suggest that some within the industry are turning a blind eye to smuggling. Few manufacturing industries would "lose" 355 billion units a year--the number of cigarettes not arriving at their intended destination--without taking action. Theoretically, also, the incentives to smuggle are substantial. When cigarettes arc smuggled into a new market they help to gain market share for foreign brands and they also pressurise governments to keep taxes on legal sales low. Researchers who have studied the entry of transnational tobacco companies into new markets conclude that smuggling has been used as a means of getting a foot in the door, for example in Latin American countries during the 1960s and 1970s. Raise the costs of smuggling to smugglers There is unfortunately very little experience and research on smuggling, including the effectiveness of anti-smuggling efforts. Various practical options are available. First, the penalties for smuggling could be made sufficiently severe to deter those who currently perceive the risks as low. Second, the legality or otherwise of cigarette packs could be made more immediately visible to consumers and law enforcers, for example by the addition of prominent tax stamps--which must be difficult to forge--on duty-paid packs, mad special packaging on duty-free packs. Strong and varied warning labels in local languages also help to distinguish legal from illegal sales. Third, all parties in the chain between manufacturer and consumer should be licensed. This is already the case in Fraiace, and as of September 1998, in Singapore. Fourth, manufacturers should be required to stamp each pack of cigarettes with a serial number to enable tracking. With 321423057

increasingly sophisticated technology, pack marking could provide inl'ormation about the distributor, wholesaler and exporter too. Fifth, manufacturers should be required to take responsibility for better record-keeping to ensure the final destination of their products is as officially intended. Computerised control systems would enable governments to track individual consignments and inspect their progress at any time. Such a system has already contributed to a fall in cigarette consumption in Hong Kong. Sixth, exporters should be required to labe! packs with the name of the country of final destination, and print health warnings in the language of that country. Where international companies produce their cigarettes locally, this could also be stated on the pack, to aid detection and increase awareness of smuggled cigarettes. 8.4 Conclusion This chapter has summarized some a'tguments about the effectiveness of interventions designed to reduce the supply of tobacco. It concludes that most such measures are unlikely to reduce the demand. However, con~'olling smuggling could prove effective in many countries. Demand reduction would lead to only gradual transition in supply for most countries, and the absolute market siz~ will remain considerable for the foreseeable future. However, governments in countries heavily dependent on tobacco for jobs and export earnings may prefer to put in place safety nets to farmers, perhaps financing such support schemes out of earmarked tobacco taxes. Draft: Net for circulation or citation 98 321423058

9. Is tobacco control worth paying for?. We have analysed whether there is a rationale for intervention in the tobacco market on economic grounds, and concluded that control measures are justified. We have discussed a number of measures to reduce the demand for tobacco, concluding that raised taxes, wider access to nicotine replacement therapy and better health infom~ation are all valuable and effective measures with the potential to prevent millions of deaths in the coming decades. Here, we assess what tobacco control will cost economies, both as a result of falling demand and in terms of the costs of the interventions themselves. Wc also assess the cost- effectiveness of various tobacco control interventions, relative to each other and in the context of other public health interventions. 9.1 The impact of falling demand on consumer welfare and on jobs Welfare losses would be offset by health gains Some critics suggest that national economies as a whole will be worse off as a result of higher tobacco taxes. They argue that such taxes will create a significant "dead-weight" loss--sometimes called the "hidden cost" of taxes. Dead-weight losses are incurred when, as a result of tax raised on some good, prices paid by consumers rise, while prices received by suppliers fall. This shifts downwards the point at which the supply curve and the demand curve meet, and the surpluses to both consumer and producer are reduced. Yet government revenues arc not raised either: in other words, the money is effectively lost from the economy altogether. Consider some hypothetical country where tax on tobacco is raised by a fixed and certain amount, raising the cost to smokers and reducing consumption accordingly. Because they are paying higher tax, consumers lose part of their surplus and suffer a loss of welfare. In addition, by reducing their consumption, consumers would suffer a further loss of surplus. However, at the same time, smokers would gain welfare from the perceived benefits of reducing their risks of ill health thxough reduced consumption (Pekkurinen 1992). Income is not lost to society but transferred from producers and distributors of tobacco to producers and distributors in other commodities. Thus, the net welfare loss is far smaller than suggested. As part of the studies underpinning this report, a model was designed to assess the impact on welfare of three interventions: (I) raising taxes on tobacco by 10 per cent, (1I) implementing non-price measures such as information campaigns, advertising bans and smoking restrictions; and 0It) widening access to nicotine replacement therapy (Yurckli 1998). The model indicates that the gains to smokers, in terms of the value of life saved by quitting, outweigh the loss of consumer surplus. Box 9.2 provides these results in more detail. 321423059

Box 9.2. Welfare losses from increased taxes are offset by health gains Fear of the impact of tax increases on economic welfare may be a significant impediment to their implementation by governments. The conclusions reached in studies perforated for this report suggest that such fears should not be overstated. A model was developed to assess the impact on welfare of a price increase of 10 per cent in the price of a pack of cigarettes. The study took account of the producer and consumer surplus figures estimated in the cost-benefit analysis described in Chapter 6, Box 6. I. With the baseline aggregate consumer surplus set at $126 billion per annum, and the producer surplus at $41 billion, assuming a discount rate of 5 per cent, a 10 per cent price increase would reduce the annual surplus to consumers to $104.3 billion, and the annual producer surplus would fall to $31.54 billion. Over a lifetime, the benefits would fall from about $3300 billion to around $2700 billion. ,; While the losses arc not trivial, it should be remembered that the consumer surplus of tobacco may in itself be overestimated, when addiction is taken into account. And, as the discussion in Ch, apter 6 made clear, many sn~okcrs arc not fully aware of their health risks, so that the external costs of smoking may be much larger than conventional economic theory has allowed. Taking the value of unintentionally lost years ofhealthy life into account, measured in terms of what the smoker would be willing to pay to buy back those years, the social benefits of smoking fall to zero if even just 3 per cent of smokers arc uninformed about their risks (Peck ct al 1999). The model has been used to estimate the value of life saved by a price increase of 10 per cent on a pack of cigarettes. Using the data from the impact model described in Chapter 7 (Ranson 1998) on the potential number of deaths averted by the price increase, and assigning a value of $7750 to each lost year of healthy life (or DALY) due to uninformed smoking, the model shows that the total benefit of saving these lives is $3,896 billion (To be rechecked once revised DALY estimate is available). This more than offsets the lifetime loss of surplus, indicating that this form of tobacco control is an excellent "buy" for the economy as well as for health. Source: Yurekli, 1999 Jobs, economies and the tobaccoindustry A major reason for governments' inaction over tobacco is their fear of creating unemployment. This fear is derived mainly from the arguments of the tobacco industry, which says that control measures will result in millions of job losses across the world. Yet a closer inspection of the arguments, and the data on which they are based, suggests that the negative effects on employment have been greatly overstated for most countries of the world, even though they must be carefully considered and minimized by judicious policymaking. Independent research summarized below suggests that, for all but a very few agrarian countries heavily dependent on tobacco farming, the effects of tobacco controls on the economy would be negligible and in some cases positive. 321423060

The tobacco industry estimates that 33 million people, including seasonal workers, family members and other labourers, are engaged in tobacco farming world-wide. Independent estimates have not been made recently. According to the industry, some 1S million of the total are in China, and another 3.5 million in India. Zimbabwe has some 100 000 tobacco farm workers. Small but significant numbers are employed in the industrialized world: the USA, for example, has 120 000 farms and the European Union has 135 000 farms, mostly small, in Greece, Italy, Spain and France. However, it should b¢ noted that most of these jobs are not full-time, but, in common with many other rural jobs, part-time or seasonal. Their contribution to national economies should therefore not be overestimated. The manufacturing side of the tobacco industry is only a small source of jobs as it is highly mcchaniscd. In most countries, tobacco manufacturing jobs account for well below l per cent of total manufacturing employment. There are a few importhnt exceptions to this pattern, with Indonesia relying on tobacco manufacturing for 8 per cent of its total manufacturing output, and Turkey. Bangladesh, Egypt, the Philippines and Thailand relying on it for between 2.5 and 5 per cent. '. Studies financed by the industry have repeatedly argued that tobacco control policies would create unemployment and economic downturn. These studies take account of jobs attributable to tobacco in each sector, incomes associated with these jobs, tax revenues generated by tobacco sales, and the contribution of tobacco to the country's trade balance wherever this is relevant. They then estimate the extent to which this tobacco-related economic activity stimulates further economic activity, for example, in the advertising industry, the sports industry and elsewhere. However, the industry-funded studies may be regarded as flawed for several reasons. Typically, they assess the gross contribution of tobacco to employment and the economy. garely, if ever, do they take account of the fact that people who stop spending money on tobacco usually spend it on other things instead, thus generating alternative jobs to compensate. Also, their methods overstate the impact of any intervention that reduces demand because their estimates of certain variables, such as trends in smoking and trends in the mechanization of cigarette production, tend to be artificially static. Independent researchers have recently done their own studies of the impact of tobacco controls. Rather than consider the gross economic contribution of tobacco to the economy, the independent studies estimate its net contribution--that is, the benefit to the economy of all tobacco-related activity after taking into account the compensating effect of alternative jobs that would be generated by the money not spent on tobacco. The conclusions of these studies are that tobacco control policies would have little or no negative effects on total employment, except in a very few tobacco-producing countries. The studies find that there would be job losses in the production, manufacture and sales of tobacco and, in some cases, in government posts due to an eventual loss of government revenue. But in most economies, these losses would be outweighed by gains in service industries, which are labour-intensive. Indeed, the tobacco manufacturing is above Draft: Not for circulatiott or citatiott 101 321423061

average in the amount of capital intensity needed, suggesting other sectors would create more jobs. The independent models make certain assumptions, which may vary from model to model, and which clearly will affect the outcomes. For example, most assume that money not spent on tobacco will be spent on other goods and services according to the consumer's average existing spending patterns. On this assumption, there are clear net gains in economic activity. Others assume that individuals who have recently stopped smoking spend their new-found extra money on luxury items and services rather than on essentials. On this assumption, there will bca greater need for adjustment in the economy, but also larger net gains. Some assume that governments will respond to a long-term reduction in tobacco tax revenues by raising new taxes on alternative goods and services; others assume govemn3,1ents will respond by reducing expenditure, and hence employment. On the manufacturing side, the adjustmen ,ts.required by a diminishing tobacco industry would no doubt extend over decades or even longer. They will also most probably occur at a far slower rate than recent mecahniszatidn in tobacco manufacturing which has reduced employment in the industry by more than a quarter in ~e US and Western Europe and in India. The economies involved were insignificantly affected by these changes, an experience that bodes well for dealing with the impact of a slow decrease in demand. Figure 9.2 (pending) If tobacco control measures take the form of advertising bans or other non-tax measures, money not spent on tobacco should be available to be spent on other uses, generating new jobs. If, on the other hand, tobacco controls are in the form of higher taxes, then governments will have increased revenues in the short to medium term, with which to create new jobs. Provided the money is spent, either by individuals or by government, alternative jobs are created, with the number depending on how labour-intensive is the alternative spending. For example, a study in the United Kingdom found that jobs would increase by more than 100 000 full-time equivalents in 1990 if former smokers spent their money on luxury items, and if any decline in tax revenues brought about by non-tax measures to reduce demand were offset by taxing other goods and services. A study in the USA found that jobs would initially increase in modest numbers but that, between 1993 and 2000, the number of jobs would rise by 20 000. While in the tobacco-growing region of the USA, there would be net job losses, the national total would rise because of the money freed from tobacco and injected into other areas of the economy. Of course, the transition may be difficult and may create social and political problems in the short term. But economies go through many such transitions, and this one is not exceptional. 321423062

The findings are not restricted to the high-income countries, indeed, there are some developing countries that might experience striking benefits. For example, Bangladesh would benefit markedly, according to similar models, with a net job gain of 18 per cent across the whole economy. This is because in a poor country such as Bangladesh, most of • household income is spent on food. If the money spent on tobacco were freed up for other uses, it would generate new employment. However, the small number of agrarian countries that are heavily dependent on tobacco could experience net losses. One model suggests that in Zimbabwe, if all tobacco production stopped tomorrow, there would be a net loss of 12 per cent of jobs. If all slack in the industry were taken up by other agricultural jobs, the net job loss would still be 7 per cent. It should be stressed, however, that this scenario is implausibly extreme. Tobacco production is not going to stop overnight, and the adjustment would be made over a generation. ' ~ Identifying the needs of different types of economies As the discussion has shown, the economic ~npact of tobacco controls, are likely to be much less damaging for most countries than the industry argues. However, the impact will be different in different types of economy. Five broad types can be identified: 1. countries that produce, but do not consume, tobacco--that is, full exporters. Such countries are a theoretical possibility but a practical rarity. 2. countries that produce more tobacco than they consume--that is, net exporters; 3. countries that consume what they produce--that is, so-called "closed" tobacco economies; 4. countries that consume more than they produce--that is, net importers; and 5. countries that produce no tobacco but consume it--that is, full importers. Not surprisingly, countries in the first two categories will face the greatest challenges in adjusting to tobacco controls. For tobacco-consuming countries, the impact of controls is borne by consumers and more jobs are likely to be created than lost (Figure 9.2). The main concern for policymakers is to maintain tax revenues in the long term. In contrast, tobacco-producing countries face initial job losses, although other jobs may gradually replace them. At the level of households and small rural communities, such adjustment means loss of income, upheaval and possibly relocation. The worst-affected producer countries are those that export most of their crop such as Malawi and Zimbabwe. However, policy maker in even these countries may consider that they stand to gain from tobacco controls in the longer term. While many resource-poor countries may be forced o put a higher value on short-term economic gains than on long-term public health, it should be stressed that the long-term benefits of improved health include economic development (WDK 1993). As the discussion has shown, then, tobacco control policies are unlikely to cause either significant welfare losses to the economy or cause significant net job losses for the vast Draft: Not for circulation or citation 103 321423063

majority of countries. A very small number of countries with a reliance on tobacco production would have adjustment costs. In these countries the adjustment would be slow and governments would be entirely justified in providing support systems for poor farmers and others directly affected by falling demand. The losses expected to economies from tobacco control are therefore relatively small. Against this, the cost of doing nothing must be compared, as both the loss of life and the attendant gross healthcare costs of rising numbers of smokers are expected to climb significantly for the next few decades. 321423064

~ge ,yment ,ntage ne~t )my year I 25% 20% 15% 10% 5% 0% -5% -10% -15% NET ~.: EXPORTERS ' ' " • L i - ' :~ ' " " " ":: ....... ~: : • : :i;. Reduction in :: Elimination of ' i tobacco :. " domestic - : " consumption • tobacco Elim Ination of expenditures of consumption domestic . 40%, spending expenditures, tobacco " • according to • , .. ~:~ii'. ~. spending consumption ' recent, ..!: L.. • according'to • and production, stopper" .~i.~. "average" ..:. redistrib~Jted " " .expenditure expenditure .-" .accord~ngto ": :pattern;s., patterns i,.:.i: .'average" input- • ..-.] outpul pallerns :;~, " .... : " !,. ; :: .;; . " i i • USA (1993) Z mbabv UK (1990) : CLOSED .~ECONOMY Elimination of :;"i •domestic domestic . ..i.!i:.tobacco . tobacco~ : :ii;~ consumption consum ptlon . i:~ilexpenditures, : .... expend tures,~:~~pend ng " spending i~ i according to according to i :'average" .. "recent .... .' :: ;ix expenditure stopper .; :. patterns " ~i L. expenditure .:.: ; ..:. patterns i. ' .. ' " ' ' I ~,,r " , South Africa Scotland [1989] (1995) • ' NET i.i: ..i:: " :..., ' ';: Elimination of :i: .i.IBI i domestic.:..:. :., :.:.i .;. :if: I:,~1 , iobacco • " : .: ..": .; I~ consumption ~ :~:1~1 Spending ' :; ;":, '. :"1~ average :~ expenditure : :'.~ ...:.~.. ;~"1~ t • Michigan 11992) Banglades~ ::: . ..:~::~".~ ' domestic '. • ~ ': " "~: :: tobacco . : ...~._..~ consumption . .:. expenditures, spending : according to "average" expendilure patterns Studies on the employment effects of tobacco control policies

9.2 The cost-effectivene.~ of tobacco control interventions Tobacco control is a highly cost-effective way of averting death and disability Another apl~roach to assessing the costs and benefits of intervening is to examine the cost of each intervention per life saved, or per year of healthy life saved. According to the World Bank's 1993 World Development Report, Investing in Health Research and Development, tobacco control policies are considered cost-effective and worthy of inclusion in a minimal package of health care (WDR 1993). Existing studies suggest that policy-based programmes cost about US$20 to $80 per discounted year of life saved (Jha 1998), a level that would be regarded as a "8ood buy" in middle income countries and in some low-income countries. For this study, estimates were made of the cost-effectiveness of each of the demand- reducing interventions discussed in Chapter 7: tax rises, a package of non-price measures including advertising and promotion bans, wider health information and indoor smoking restrictions; and nicotine replacement therapy. The findings may be of particular value to low-income and middle-income countries in'assessing the appropriate mix of intervention strategies for their own needs. The estimates were made within the model used to estimate the potential impact of these interventions in saving lives and reducing the prevalence of smoking. The model's design and inputs are described in Box 7.1. Years of healthy life saved are measured in terms of DALYs (Murray and Lopez 1996). The results suggest that tax increases are by far the most cost-effective intervention, and one that compares extremely favourably with many health interventions. Depending on the assumptions made about the administrative costs of raising and monitoring higher tobacco taxes, the global estimate ranges between US$1.88 per DALY at the low end to $247 per DALY at the high end. For low-and middle-income countries, the figure is between 42 cents per DALY and $56 per DALY, while in the high income countries, the figures are, as expected, higher, ranging from $16 to over $2000 per DALY (see Table 9.1) 321423066

Table 9.1. Range of cost-effectiveness values for price increases (USD/DALY Saved), by region. Region Low End High End Estimate* Estimate** EAP 0.31 41 ECA 0.38 5 I LAC 1.04 i 37 MNA 0.71 93 SA 0.26 34 SSA 0.20 27 Low/Middle 0.42 56 High 16.11 2,124 World i.88 ~, 247 * Calculations based on: price increase of 100%, intervention cost of 0.005% of GNP, benefits (DALYs saved) distribuied over 30 years and discounted at 3%. ** Ca]culations based on: price increase of ] 0%, intervention cost of 0..02% of GNP, benefits (DALYs saved) distributed over 50 years and discounted at 10%. Source: Ranson 1998 The second most cost-effective intervention is nicotine replacement therapy, with global figures ranging from $119 to $9 133 per DALY. The high and low estimates for this intervention are closer together than for the other interventions (Table 9.2). We may add the cost-effectiveness of actual provision of the drugs as a third and fourth column.]] Table 9.2. Range of cost-effectiveness values for NRT liberalization policies(USD/DALY Saved), by region. Region Low End High End Estimate* Estimate** EAP 12 353 ECA14 4.22 LAC38 I, 175 MNA26 788 SA 6 198 SSA8 231 Low/Middle 15 447 High 290 8,877 Drnft: Nnr for circulation or citation 107 321423067

World 60 1,827 * Calculations based on: effectiveness of 10%, intervention cost of 0.005% of GNP and repeated annually over 10 years and discounted at 3%, ~nefits (DALYs saved) dis~'ibutecl over 30 years and discounted at 3%. ** Calculations based on: effectiveness of'5%, intervention cost of 0.02% of GNP and repeate~ annually over 50 years aed discounted at I 0%, benefits (DALYs saved) distributed over 50 years and discounted at I0%. Source: Ranson 1995 The least cost-effective of the demand-reducing interventions is the non-price package of measures, which is estimated to cost between 1;118 and 1;22 600 per DALY (Table 9.3). Table 9,3. Range of cost-effectiveness values for non-price interventions other than NRT (USD/DALY Saved), by region. Region Low End ,; High End Estimate* Estimate*" EAP I 1 858 ECA14 :, 1,050 LAC 37 : 2,823 MNA 25 1,948 SA7 524 SSA7 550 Low/Middle 14 1,105 High 288 22,075 World 59 4,521 * Calculations based on: effectiveness of 10%, intervention cost of 0.005% of GNP and repeated annually over I0 years and discounted at 3%, benefits (DALYs saved) distributed over 30 years and discounted at 3%. ** Calculations based on: effectiveness of 2%, intervention cost o1'0.02% of GNP and repeated annually over 50 years and discounted at I 0%, benefits (DALYs saved) distributed over 50 years and discounted at 10%. Source: Ranson 1998. The implications are that it would pay virtually all countries to raise tobacco taxes as a higl'dy cost-effective means of reducing premature death and disability due to tobacco. Nicotine replacement therapies also perform surprisingly well, while non-price measures arc clearly valid in many contexts. For the latter two measures, it is likely that countries will need to make their own cost-effectiveness estimates before implementation. A summary of the likely relative value of each intervention in different settings is shown below in Table 9.4. 321423068

Table 9.4 The relative cost-effectiveness of policies in countries of different income levels Policy instrument Price measures Non-price measures (information, ban on advertising and promotion, regulate smoking indoors~) Widen access to nicotine replacement Low-income +4-+ Middle-income 4-+ -4- High-income ++ therapy Note: +++highly cost effective; ++ moderately cost effective + acceptably cost-effective: Source: based on data from Ranson !~)99. We turn in the final chapter to a discussion of the policy implications of this report's findings, and make recommendations fox" action. Draft: Not for circulation or citation 109 321423069

10. An Agehda For Action Tobacco is one of only two causes of death that arc large and growing world-wide: the other is HIV. While most countries have begun, at least, to respond to the HIV epidemic, the response to the global tobacco epidemic has so far bccn limited. Yet, faced with projections of 100 million tobacco-related deaths over the next 20 years alone, governments must plan how to act. This report has argued that intervention in the tobacco market is justified on economic grounds. Using evidence from a number of countries, and a model based on highly conservative assumptions, we have shown that cost-effective measures such as higher taxes, the provision of better and more widespread health information, and greater access to nicotine replacement therapy, could cut the number of tobacco-related deaths among today's smokers by millions. For example, as we saw in Chapter 7, a 10 per cent increase world-wide in the real price of cigarettes could prevent I0 million tobacco-related deaths among smokers alive in 1995 and signific .antly increase government revenues at the same time. Widely improved access to nicotine re~,lacemcnt therapy could cut the number of these deaths by 14 million. Already, the evidence from the industrialized world, where demand for tobacco has been falling for decades, suggests that tobacco control measures are feasible. Yet, partly because of misguided fears of economic harm, and partly because of political pressures, many governments have yet to adopt them. In this chapter, we examine some of the barriers to change and offer specific recommendations for action to governments and international agencies. 10. 1 Overcoming political barriers to change Policy makers who decide to implement tobacco controls will obviously wish to maximise their positive impact and minimize any difficulties. By analysing the key stakeholders on both the supply and demand sides in each country, policy makers can identify the size of each constituency, whether it is dispersed or concentrated, and so on. By noting the likely winners and losers from each intervention, policy makers can attempt to design approaches to adjustment that are tailored most closely to reflect national needs, constitutional issues, and so on. Judicious choices should ensure that the number of winners exceeds the number of losers, and that market efficiency is maximized so that alternative work emerges quickly for losers. However, the nature of the transition may be difficult. Winners, such as non-smokers, may be a scattered and dispersed group, while losers, such as tobacco farmers, may have a powerful political and emotional voice. An illustration of a stakeholder analysis is given hcrc to show the kinds of issues that must bc considered (Table 10.1). Careful planning and political mapping is essential to achieve a smooth transition from reliance on tobacco to independence from it, whatever the nature of the economy and the national political framework. 321423070

Table 10.1 Theoretical analysis of key stakeholders affected by tobacco control policies Key Stal~holderJ Effect of tobacco Degree of influence of Straleg), of .~takehol~rs an~b'or policy opuon~ co~u'ol policies on stakeholder~ over stakehol~r$ ÷ 0 Supply Side 1. Manufacmrm 2. Agricultur~l produc=ts I famlm Distribution and sales 4. Ma|keting and advenisin[ indus~ 5. Bu~inoss Community "Demtnd Side 0 in the long run High. IvflqCs give suong economic and political backing to local compani¢s High. Strong support from indust~/. Impact mainl~ on profile. Can shift tax borden increasingly to coflsunlcI1. . " Most ~ract~d m'¢ pr~uc~r-cxp~rt~r countries. Us~ supply-side meas.ures. Retail ma~ins will bc [aiqcd asain on othcr products. Exponditm~ sw~chcs will gencratc new cli=nt has=. Benefician,'~ (productivity improveme.n~, increased demand). Use their support. 6. Smokes 7. Non-smokers High in most countri~ Low in most counuics 8. Families + Low rcsomcc utilisation possible. Health bcneftts from cessation. Reduced disabilit,v of breadwinners. Government 9. Public health + High Rccoupm~nt of social and private costs and cxtcmahty addressed. I O. Health sector 0 Reduced tobacco health ~pending will be replaced by other Continuing poor smokers have reduced sl~nding ability from hiFhcr taxes. More likcl,v to quit thnuFh. Reduced burden of external c~u. ~=du~d long ~ subsidisation of h¢~ insur~c, t~ ~lf~¢ ~fi~ ~d pensiom. Rcdu~d need to pmvidc'~e f~ ill s~k~. Alte~tive Medium I I. Government tsxes ÷ High (Finance) depends Country - dcl~ndcnt 12. Balance of payments and uad¢ 13. Smueeling 14. Government owned companies 15. Education 16. Subsidies 17. Pensions and wetfa~ provision 18. The legislative process High High Hi.eh Medium - counu~ dependent High Medium health spcndin.= as dcmo[raphics chan~e from l~-s.~ smokinF. Increased tobac~ ~ will incre=c revenue in shun ran. Can use txx ~o suF~ loF~. replace consumers" and producers" su[plus For importing coun~, fall in consumpt=on will lead to bal=ce or paym~u improv~cnt, ~clin= in profit ~patrimi~. Fm major cx~crs, t c~c~. Supply-side mc~ur~ must considc~ ~ ~ing ~pacity. ~clinc in local consumption may lead to attempt to incrc~c cxpo~ b~ produce~. Switch from tobacco to ~hcr ~ds. Law enforcement rcquircd Conflict of interest. Could consider privntisation, profits arc replaced with tox revenue, no nccessar.v income loss. Use to influcncc continuin~ smokers. Subsidies do little to atTect prices of cigarettes, but help farmers. Can distort m~kets, prevent crop substituUon. Quota systems prcvenl efficient farmin~ techniques. Business and government may have to increasingly cart7 I'=turc smokine costs (with laeecd ons¢l of illness). Legislative process osual)y uncomplicated and inexpensive. Political acccptabilit)" more intcn~ and complex. Other 19. Environmcntalis= ÷ Medium to low Reduced costs from fires, pollution, litter collectio~. deforestation, d¢[radation, accidents Positive externalities. 20. The legal profession + 2 I. Investors a~d sha~cholder~ dcpcnds Medium - counU7 dcpcndcnt Low - Legal channels increasingly used to claim health cosls attributable to tobacco. Not common yet in dcvcto~i.~ ctqmtrics. Tobacco companies incrcasingly divesting into other lines of business. Individual ponl'olio shifts likel,v. Source: van dcr Mer~ve 1998 Draft: Not for circulafio, or citation Ill 321423071

Providing evidence of the cost-effecliveness of policies at national level can increase political support for implemenling lhem We saw in Chapter 9 that a combination of (I) price increases, (2) non-price measures including better health information, comprehensive bans on advertising and promotion, and restrictions on indoor smoking, and (3) wider access to nicotine replacement therapy could prevent millions of deaths among today's smokers. The scale of the impact of these measures is unprecedented compared with other health interventions such as the Expanded Programme on Immunization, treatment for tuberculosis, and other vital activities against major health threats. Producing evidence at national level that tobacco control represents a "good buy" for public health is likely to be an important way of overcoming political barriers to action. i Table 10.2 Actual and proposed allocation of annual public expenditure on health~ AIDS and tobacco control programs (in US dollars per c~tpita) Proposed spending Actual spending Developing LI MI Dev,,eloping HI A) World Development Report, 1993 Public heath 4 Essential clinical services 8 (minimum packages) Discretionary clinical -6 services Total 6 HI 7 15 40 62 o o 5-'/ B) AIDS control programs Actual spending (I 993) C) Tobacco control CDC recommendation for the US California Massachusetts US Canada This report "s recommendation I 4-6 13-15 21 0.04 (LI) 0.05 0.15 NA 8.65* 20.56 (US) 2.5 13.0 0.46 0.65 * Weighted average for 19 countries, including the US. LI is low income, MI is middle income, and H! is high income. Sources: CDC Program and Funding Guidelines for Comprehensive Tobacco Control Programs. 1999; Pechmann et el, Am J Public 1998; $8; 1362-7. California and Massachussets are averages for 1994-199"/; US and Canada is the average for 1989-1997. It should be stressed that the tobacco control interventions discussed in this report are all policy instruments, and implemented essentially with a "stroke of the pen" rather than through expensive technical interventions at grassroots level. This not only makes them affordable; it also means that they are financed out of budgets other than the health 321423072

budget, and therefore not competing with other much-needed health interventions such as those discussed above. As the previous chapter showed, models of the cost-effectiveness of the three types of intervention show varying cost-effectiveness between each intervention and between high-income, middle-income and low-income countries. The proper policy mix for each country is likely to vary somewhat, but for an effective programme, all three components are likely to be present. As part of the studies underpinning this report, a panel of economists was invited to identify ideal policy mixes for a range of needs (Table I0.3) details of sample size etc to come, based on the input from the Lausanne meeting: ]] Table 10. 3." Relevance of Tobacco Control Policies to Different Countries (o.g) 2.0 (0.8) Mass reform o .:.~z~ .... ,.,.....-_.:. ..... .:...: . 2.7 ..... '. .... • ' =~ .....::.,'.:':-:::~- (0.5) (0.5) (0.8) Ban advertising and promotion.' 2.8 2.8 2.5 • .~. ~ .!.: '.::~,.~-.':::: ..,(0.5) (0.5) (0.6) " ' " '~'~" " :"~i ". ~' r:.!::" , Restrict pubfic and wo,r'kp~ce;.:-.-:.;': :- .i 2.0 2.2 2.6 smokingi".:ii!,i~:~',.~i:'~i~i:~i~!~!:iii:~~ ' )". (0.9) (0.8) (0.7) Deregulate nicotine replacement: .... 1.2 1.6 2.3 Pr°~Uets. .... " . .-:i .., !5:: :"~i (0.8) (i.0) (0.9) Control smuggling .:../;:.:~...i!!~,.._.:.: :.: ::. ~ ' ' 2.0 1.7 !.4 :.. :.-'.7!". :~':: ';'-'~';: ,.7~ " "" • • .... .- . - . .... (I.0) (0.9) (0.7) 3 = Highly relevant; 2 = relevant; 1 = somewhat relevant; 0 = not relevant. Ratings are based on reviews by a group of 30 economists and epidemiologists. The mean score and standard deviation (in brackets) are shown. 10.2 Research priorities For tobacco control efforts to succeed in reducing deaths among today's smokers, immediate action is necessary. Demand-reducing measures such as higher taxes and bans on advertising and promotion have already been seen to work well in industrialized countries, and enough is known now to implement these measures now. At the same time, ho,,~ever, a concurrent research agenda, both in epidemioiogy and economics, will be 321423073 BATCo US DOJ ~ Philip ~,l~rl~

needed to help governments to fine-tune the mix of policies that has the greatest chance of success for their needs. I. Research into the causes and consequences of smoking at national and regional levels Until the late 1980s, epidemiological research into the long-term health effects of smoking was almost entirely restricted to theindustrialized world (Peto et al 1994). However, studies now emerging from China suggest that the pattern of diseases caused by smoking are significantly different there from in the West, even though the overall proportion of smokers who are killed by tobacco is strikingly similar (Liu et al, 1998). Yet at present, some of the regions in which tobacco-related deaths are projected to be greatest lack direct evidence of the absolute and relative health risks to their populations. For example, India is expected to ac,qount for about one-fifth of worldwide tobacco deaths by 2020, but so far there are no complete vital registration data or cause-of-death records for 99..5 per cent of the population. During the early 1990s, the most recent ti~¢ period for which data are available, investment in research and development in t6bacco control amounted to US$ 50 per 1990 death (a total of US$148-164 million). In contrast, HIV research'and development received about US$ 3 000 per 1990 death (a total of US$ 919-985 million). Spending on both diseases is primarily in developed countries (World Health Organisation, 1996). There is therefol'e a need for research at national or regional level to "count the tobacco dead" and classify deaths by cause. The benefits of such research go farther than their practical value of informing governments of the status of their tobacco epidemic or a baselin6 against which to monitor the impact of control efforts. They stimulate policy responses and may have a significant impact on tobacco consumption. 2. While epidemiological research into the consequences of smoking has at least begun to spread outside the industrialized countries, research into the causes of smoking, the addictive nature of tobacco use and the behavioural factors associated with smoking uptake--such as young people's tendency to discount the future--remains heavily biased towards north America and northern Europe. While control interventions are being implemented, parallel research activities into these issues may help to refine the targeting of interventions, such as those designed to improve health information~ for maximum effect. 3. For economists, research into the cost-effectiveness of each intervention at national level is also a priority. Further data on price elasticity in low-income and middle-income countries would be valuable, as would estimates of the social and healthcare costs of tobacco use in these countries, and the optimal tax level required to reduce demand by the desired amount and shift the costs of tobacco use away from non-smokers and onto smokers and tobacco producers. I)rtt~'t. /Vt)e D)r r;e,'rlh)tl,))) ,). 321423074

I0. 3 Recommendations for action We make two sets of recommendations, the first for governments, the second for international agencies. The recommendations are as follows: A. Government action 1. Governments should immediately implement a programme of measures to reduce the demand for tobacco. The programme should have three basic components, although the relative importance 6f each may vary slightly from country to country: la. PRICE MEASUILES. Taxes on tobacco should be increased in real terms by not less than 10 per cent per year for the next ten years. A regularly updated specific tax is least vulnerable to erosion. "... lb. NON PRICE MEASURES. A package of measures to increase awareness of the health effects of smoking (addiction and long-term disease) should be implemented. This package will include a comprehensive ban on all advertising and promotion; prominent health warnings in local languages on cigarette packs; counteradvertisiug programmes; and increased support for research and information about the consequences of tobacco on health, all of which have been shown effective in reducing the prevalence of smoking. While school health programmes on tobacco may have some advantages, their effectiveness has not been proven and countries should assess their resources before investing in them. Governments should also introduce bans on smoking in workplaces, public places and other indoor environments. These will definitely benefit non-smokers and may help to cut smoking prevalence as well. lc. NICOTINE REPLACEMENT THERAPY. Governments should widen access to nicotine replacement therapy, a cost-effective means of quitting. This should be done by careful overhaul of the regulatory framework for sales of NRT products to reduce unnecessary restrictions. The impact of the overhaul on smoking prevalence and nicotine addiction should be monitored. 2. Governments should implement aggressive controls on tobacco smuggling, including better tracking of cigarette consignments, requirements on the industry to provide fuller documentation of the destination and final selling point of its exports; tough penalties for crime, local warning labels, etc. B. International action: 321423075

International agencies such as the World Health Organization, the World Trade Organization, UNICEF, the World Bank, the International Monetary Fund and others should act where individual governments cannot to control tobacco. Their priorities are likely to include issues that cross-national borders and the dissemination of international public goods such as research results. The World Health Organization has already taken a strong lead in this respect (see Box I0.2). Other agencies, including the World Bank, (Box I0.3), are actively reviewing their roles to complement that of the lead health agency. The International Monetary Fund has identified tobacco taxation to be an efficient source of government revenues (see Box 10.4). 1) International agencies should review their own programmes, to ensure that tobacco control is given due priori.ty in accordance with its importance for global health and global expenditure. 2) International agencies should both sponsor and, where appropriate, conduct research into the causes and consequences (including costs) of smoking in all regions and the cost-effectiveness of a package of tobacco control measures. For the f~tur¢, possible medical interventions to controlling nicotine addiction should be investigated. The results of such research should be widely disseminated a~ an international public good to help to reduce smoking prevalence in the most efficient way possible. 3. International agencies should engage in multilateral action to deal with tobacco problems that cross borders. In particular, a) they should work with groups of neighbouring countries to help them achieve low or zero differentials in tobacco taxes as a means to reduce the incentive for international tobacco smuggling; b) they should provide legal and technical advice to enable more efficient international smuggling control activities; and c) they should collaborate to develop policies to prevent circumventing national bans on tobacco advertising and promotion via global media and communications technology. Advertising taxes would be one efficient means of achieving this. [[BOX 10.1 pending: on EARMARKING; to cover the pros and cons, efficiency, political usefulness, ere, sourcing partly from Hu, T-W. Cape Town Paper ]] Box 10.2 WHO and the Framework Convention on Tobacco Control [to be shortened somewhat.] At the World Health Assembly in May 1996, WHO's Member States adopted a resolution calling upon the Director-General of WHO to initiate the development of a framework convention on tobacco control. WHO, under the leadership of Director- General Gro Harlem Brundtland, has assigned priority to reinvigorated work on tobacco control, and has established a new project, the Tobacco Free Initiative (TFI). A 321423076

cornerstone of TFI's work is the WHO Framework Convention on Tobacco Control (FC'rC). The WHO FCTC will be an international legal instrument designed to circumscribe the growth of the global tobacco pandemic, especially in developing countries. If entered into force, the Convention will be a first for WHO and a first for the world. This will be the first time that the 191 WHO Member States exercise WHO's constitutional authority to serve as a platform for the development of a convention. In addition, this will be the first multilateral convention focusing specifically on a public health issue. The development of the WHO FCTC will be helped by knowledge of the addictive and lethal qualities of tobacco use, combined with many countries' interest to improve tobacco regulation through international instruments. The international regulatory strategy being used to promote multilateral agreement and action on tobacco control is the framework convention-protocol approach. This strategy promotes global consensus in incremental stages by dividing the negotiation of separate issues into individual agreemefits:, States first adopt a framework convention that calls for c0-operation in achieving broadly stated goals and establishes the basic institutions of a multilateral legal structure; Separate protocol agreements containing specific measures designed to implement the broad goals called for by the framework convention. The framework convention-protocol approach has been used to address other global problems, for example, the Vienna Convention for the Protection of the Ozone Layer and the Montreal Protocol. The negotiation and implementation of the WHO FCTC will help to curb tobacco use by mobilizing national and international awareness as well as technical and financial resources for effective national tobacco control measures. The Convention would also strengthen global co-operation on aspects of tobacco control that transcend national boundaries, including global marketing/promotion of tobacco products and smuggling. Though the negotiation of each treaty is unique and depends upon the political will of states, the WHO FCTC Accelerated Work Plan foresees adoption of the Convention no later than May 2003. [box 10.2 endsl Box 10.3 The World Bank's Policy on Tobacco The World Bank has since 1991 had a policy on tobacco, in recognition of its harmful effects on health. The policy contains five main points. First, the Bank's activities in the health sector, such as policy dialogue and lending, discourage the use of tobacco products. Second, the Bank does not lend directly for, invest in, or guarantee investment or loans for tobacco production, processing or marketing. However, in a few 321423077

agrarian countries that are heavily dependent on tobacco as a source of income and of foreign exchange earnings, the Bank aims to deal with the issue by responding most effectively to these countries' development requirements. The Bank aims to help these countries diversify away from tobacco. Third, the Bank does not lend indirectly to tobacco production activities, to the extent that this is practicable. Fourth, tobacco and its related processing machinery and equipment cannot be included among imports financed under loans. Fifth, tobacco and tobacco-related imports may be exempt from borrowers' agreements with the Bank to liberalize trade and reduce tariffs. The Bank's policy is consistent with the arguments for ending subsidies made in this report. However, the emphasis on supply-side measures has not reduced tobacco consumption in any measurable way from 1991 to today. In the interim, the Bank's work on tobacco, comprising about 14 co ,unities with total project costs over $ I00 million, has largely been on health promotion and information. Extending this work to focus on pricing and regulation was supported in principle by the Bank's 1997 Sector Strategy Paper. This report confirms the importanqe of focusing on price as an effective means of reducing demand. ' ", [box 10.3 ends] Box 10.4 The Design and Administration of Tobacco Excises: a view from the International Monetary Fund Increases in tobacco excise rates are oRen included as a comPonent of Fund-supported stabilization programs for countries that need to mobilize additional tax revenue to reduce the fiscal deficit. While excise rates on tobacco products may be increased primarily to raise revenue, there are also health benefits from reduced tobacco consumption. In setting tobacco tax rates, governments need to take into account several factors, including the impact of smuggling, cross-border shopping, and duty-free purchases on ferries and planes. It is in the interest of governments to reduce tobacco smuggling not only to increase excise revenues but also limit the loss of revenues from other taxes including income and value-added taxes, as underground transactions replace legal ones. Ultimately tobacco excise tax rates must reflect the purchasing power of the local consumers, rates in ncighbouring countries, and - most importantly -- the ability and willingness of the tax authority to enforce compliance. With respect to the structure of tobacco excises, countries should tax all types of tobacco---cigarettes, cigars, pipe tobacco, snuff or chewing tobacco, and hand-rolling tobacco. The best international practice is to impose excises on the destination basis under which imports are taxed and exports are freed of tax. Excises can be either specific taxes (based on quantity) or ad valorem (based on value). If a primary purpose of the excise is to discourage tobacco consumption, a strong case can be made for specific excises that would impose the same tax per stick. Specific 110 321423078

taxes also are easier to administer because it is only necessary to determine the physical quantity of the product taxed, and not necessary to determine i~s value. Ad valorem taxes, however, may keep pace with inflation better than specific taxes, even specific taxes that are adjusted fairly frequently. The administration of domestic tobacco excises requires an integrated strategy for taxpayer registration; filing and payment; collection of overdue taxes; audit; and taxpayer services. Developing and transition countries may ne~ to treat tobacco production facilities as extra-territoriai and administer excises similar to customs duties. The tax authority would control shipments into and out of the production facility. Excise stamps can assist in ensuring the payment of excises and ensuring.that goods which have paid the utx appropriate for one jurisdiction don't get shipped to another. Introduction of stamps, however, involves considerable costs for producers of excised goods• Stamps will serve little purpose in control unless their utilization is monitored at the retail level. [box 10.4 ends] ".'k Draft: Not for circulation or citation 119 321423079