Are Tobacco Taxes Really Regressive? Evidence from Chile. ARE TOBACCO TAXES REALLY REGRESSIVE? EVIDENCE FROM CHILE Alan Fuchs Francisco Meneses 2 Are Tobacco Taxes Really Regressive? Evidence from Chile Alan Fuchs and Franciso Meneses1 Summary Report1 Abstract Tobacco Taxes are deemed regressive as poorest families tend to allocate larger shares of their budget to purchase tobacco. However, as taxes also discourage tobacco use, some of the most adverse effects, including higher medical expenses, lower life expectancy at birth, added years of disability among smokers, and reductions in the quality of life, among other, would be reduced. This paper describes and simulates the effects of the tobacco tax on incomes in Chile assuming three different price-elasticity scenarios for different income deciles of the population. Results show that although price increase for tobacco through higher taxes generates negative income variations across all groups in a population, under a more comprehensive scenario that includes benefits through lower medical expenses and an increase in working years, the results invert, and the overall monetary effect of the taxation policy becomes positive. Moreover, the reduction in medical expenses seems to be the main driver of the increase in net incomes because of the reduction in tobacco-related problems that require expensive treatments. Lastly, as the distributional effects of tobacco taxes are directly related to the long-term price elasticities of tobacco consumption, it would be advisable a coordination between taxation and behavioral change policies across income groups. (JEL Codes: H23, H31, I18, O15) 1 *Fuchs: The World Bank, Poverty and Equity Global Practice, 1818 H Street NW, I 4-405, Washington, DC 20433 (email: afuchs@worldbank.org); **Meneses, Duke University (email: fjmeneses@gmail.com). Support for the preparation of this report was provided by World Bank’s Global Tobacco Control Program, co-financed by the Bill and Melinda Gates Foundation and Bloomberg Foundation. We are grateful to Oscar Calvo-Gonzalez, Patricio Marquez, Blanca Moreno-Dodson, Gabriela Inchauste, Jasmin Chakeri, Sheila Dutta, Enrique Fanta and Lidia Ceriani for providing comments and support to this paper. The findings, interpretations and conclusions in this research note are entirely those of the authors. They do not necessarily represent the view of the World Bank Group, its Executive Directors, or the countries they represent. 3 4 Are Tobacco Taxes Really Regressive? Evidence from Chile. 1. Introduction population in 1992 to 34.7 percent today3. These consumption reductions could relate to policy In April 2014, the government of Chile proposed measures implemented in Chile during the last a large-scale tax reform, the most significant decade such as education, taxes, warning labels, in 30 years. Among the main changes were and advertising bans. Public anti-tobacco policies tax increases on tobacco, alcohol, and sugared in Chile are in line with the global trend: at least beverages. Because low-income families usually 40 percent of countries have implemented allocate a larger proportion of their budgets to mechanisms to reduce tobacco consumption tobacco, as well as alcohol and sugary beverages, during the last two years (WHO 2015a). at first sight the tax increase would seem to be a regressive policy. However, a closer look shows There is ample and robust evidence linking that the tax increase and resulting reduction in tobacco consumption with health-related consumption would lower the adverse effects problems (Doll and Hill 1956; Wynder and Graham of tobacco consumption, including higher 1950). Diseases associated with tobacco use medical expenses, the negative effect on life range from lung cancer to stroke and even to expectancy at birth, added years of disability congenital malformations in children (HHS 2014.) among smokers, reductions in the quality of life, In 2010, 5 million early deaths were attributed to and a list of negative externalities among first- and tobacco consumption globally, and, in Chile, the secondhand smokers. The increase in tobacco Ministry of Health estimates that 18.5 percent of tax and the subsequent reduction of tobacco annual deaths may be attributable to tobacco consumption could therefore result in potential use (Jha and Peto 2014; Pichón-Riviere et al. measurable benefits for tobacco consumers. This 2014). Scientific evidence linking tobacco and paper describes and quantifies the effects of the health problems triggered important policy shifts tobacco tax on incomes through three factors: among international organizations and policy (1) the tobacco price increase, (2) the reduction makers alike, leading to more rigorous normative in medical expenses, and (3) the rise in revenues frameworks on the sale and use of tobacco. because of the gain in years of employment. To estimate the impact of these effects, the paper Today, more than 80 percent of the world’s makes various assumptions about the price- smokers live in low and middle income countries; elasticity of tobacco and calculates the income harming health, incomes, earning potential, labor gains among various population income groups. productivity, and undermining human capital accumulation – a critical factor for sustainable During the last 15 years, the prevalence of economic growth and social development (WHO tobacco consumption has been falling around 2015). The United Nations (UN) has set the health the world, and Chile is no exception (WHO and wellbeing as one of their priority goals. 2015a).2 According to the Drug Consumption Similarly, the World Health Organization (WHO) Survey in Chile, monthly tobacco consumption has set the reduction of tobacco consumption has decreased from 43.6 percent of the adult as one of its primary goals. It has thus promoted 2 Between 1970 and 2000, tobacco production and consumption showed steady growth around the world. This trend started to change with the reduction of consumption in the developed countries and increases in the developing world, particularly China (WHO 2015b). Tobacco prevalence has decreased, but because of population growth, total consumption has increased (Ng et al. 2014). 3 Prevalence of monthly tobacco consumption. National survey of drugs in the general population 2014. (Estudio nacional de drogas en la población general) SENDA, Ministry of Interior and Public Safety. 5 tobacco monitoring, smoke-free policies, America (WHO 2015a). In 2006, Chile enacted smoking-cessation support programs, relevant the Anti-Tobacco Law, which includes restrictions health advice, advisory deterrents, and taxation on tobacco advertising, a prohibition on selling policies (WHO 2015a). Among strategies, the cigarettes to students or close to schools, tobacco tax seems to be one of the most efficient advertising the risks of tobacco consumption and measures to reduce tobacco consumption and of exposure to tobacco smoke, and designated increase government revenue (World Bank smoking areas in restaurants, casinos, or nightclubs 1999). The inelastic demand of some tobacco (IDB 2011). Some of these policies have been consumers is useful for increasing tax revenues; strengthen further in subsequent normative and the higher price elasticity of younger smokers updates. makes the tax an effective consumption deterrent (Chaloupka et al. 2002; Debrott Sánchez 2006). The government of Chile raised the ad valorem tobacco tax to 62.3 percent and the fixed tax to A recurrent concern about tobacco tax policies US$0.16 per package of 20 cigarettes to finance relates to potential negative income inequality the reconstruction of the areas affected by the effects, as there is a common believe that these earthquake in February 2010 (IDB 2011).4 The tend to affect the poor more disproportionately. change meant that Chile had one of the highest Because low-income households allocate larger tributary loads for tobacco in the Americas shares of their incomes to consumption and, (Pichón-Riviere et al. 2014). The 2014 Tax Reform among expenditures, to purchase tobacco, reduced the ad valorem tax to 30 percent and indirect taxes tend to affect their monthly budgets substantially increased the fixed tax to Ch$681 disproportionately more than the budgets of (US$1.28) per 20-cigarette pack. more well off households. Therefore, tobacco taxes directly affect low-income individuals Beyond short-term changes in tobacco more, compared with other income groups expenditures, the possible benefits of these and other taxes. Nonetheless, if indirect effects policies included lower medical expenditures and are considered, the concern about tobacco tax more years of healthy life, both of which could policies no longer holds. Indeed, it has been translate into private monetary benefits that more shown that the future benefits of nonsmoking than offset the losses generated by tax increases. outweigh the losses attributed to tobacco taxes To test these hypotheses, we use a social welfare among the population in general and, specifically, framework to calculate the effects on various low-income groups (Denisova and Kuznetsova income groups and assume different price- 2014; Verguet et al. 2015). Still, there is no evidence elasticities for tobacco consumption. To establish that these trends are occurring in Chile. a contextual background, section 2 briefly reviews the literature on the health effects of tobacco, During the past two decades, several anti-tobacco tobacco policies, and tobacco price elasticities. measures have been implemented in Chile, Section 3 describes the methodology, parameters, including tax increases, advertisement regulations, and data used to forecast the impact of the and smoke-free public spaces. Although the policy tobacco tax. Section 4 presents the estimation effort has been substantial, Chile still has one of results, and Section 5 concludes with a discussion the highest tobacco consumption rates in Latin on implications and policy implications. 4 This represents 0.0000675 unidad tributaria mensual (monthly tax units), a tax unit that is monthly defined. The exchange rate was US$1.00 = Ch$670. 6 Are Tobacco Taxes Really Regressive? Evidence from Chile. 2. Literature malformations in babies, male sexual dysfunction, low birth weights and complications in pregnancy There is extensive literature analyzing the health (HHS 2004, 2014). The exposure to secondhand effects of tobacco, public policies, and price smoke has a causal relationship with many elasticities on the international level, but there is respiratory diseases in children. There are more surprisingly little evidence on Chile. than 4 thousand chemicals in tobacco smoke (of which at least 250 are harmful and more than a. Tobacco and health 50 cause cancer). Although the evidence to infer In the 20th century, about 100 million deaths a relation between secondhand smoke and were potentially related to tobacco use (Peto cancer or its impact on reproduction is not clear, and Lopez 2004). If current trends were to some research points in this direction (HHS 2004, remain constant, about 1 billion people could 2014). Moreover, according to the World Health die from tobacco-related diseases during this Organization, second hand smoke is responsible century (Jha and Peto 2014). In Chile, over for over 600 thousand premature deaths. 92,000 children and more than 4 million adults consume tobacco every day (Tobacco Atlas 2010). b. Tobacco policies Moreover, according to the Chilean Ministry of Globally, anti-tobacco policies range from Health tobacco caused more than 16,000 deaths separating smokers in particular locations in 2013 (nearly 1 in 5 annual deaths). The same from completely smoke-free environments to year, almost 8,000 Chileans were diagnosed with advertising to deter tobacco use, prohibitions cancers related to tobacco consumption. on tobacco sales close to schools, and taxation and smoking cessation programs. Although The relation between tobacco consumption and various approaches have been assessed, they health problems has been extensively researched show diverse effects in tobacco use and exposure since the 1950s (Doll and Hill 1956; Wynder and among the population. Graham 1950). According to the U.S. Department of Health and Human Services, the conclusion WHO (2015a) argues that entirely smoke-free that tobacco consumption is responsible has environments, rather than separate smoking been reached in many studies of cancer, including rooms or good ventilation systems, are the only lung cancer, oral cancer, laryngeal cancer, way to prevent the harmful consequences of pancreatic cancer, kidney cancer, cervical cancer, secondhand tobacco smoke. Smoke-free laws and acute myeloid leukemia (HHS 2004). are popular because there is evidence showing they not only improve health outcomes, but do Active smoking is associated with respiratory not affect business. In 2014, these laws benefited problems such as tuberculosis, chronic respiratory 18 percent of the world's population, and Chile symptoms, influenza, pneumonia, infections, was one of five countries that had implemented a chronic bronchitis, emphysema, and asthma. It smoke-free law covering all public places and all is also associated with cardiovascular diseases, workplaces (WHO 2015a).5 The benefits depended such as aneurysms, strokes, and coronary heart on the breadth of the legislation. For example, disease, as well as adverse reproductive and prohibiting smoking in all indoor workplaces developmental effects, such as congenital reduced the exposure to secondhand smoke by 5 In 2014, 1.3 billion people (18 percent of the world's population in 49 countries) were covered at the most comprehensive level of smoke-free policies, an increase of about 200 million people since 2012 (WHO 2015a). 7 80–90 percent and decreased the incidence of Mass media campaigns that reach large acute respiratory illness (IARC 2009). However, the populations represent the most massive way to impact of smoke-free policies on consumption is combat tobacco use. In 2014, such campaigns small compared with taxation. Levy et al. (2012) covered 55 percent of the world's population. estimate that the vast reduction in tobacco use People in low-income countries are less likely in Brazil was mostly caused by higher tobacco to be exposed to campaigns, and there is also prices (46 percent of the impact) and smoke-free limited information about the cost-effectiveness policies (14 percent). of this approach (WHO 2015a). Durkin, Brennan, and Wakefield (2011) conclude that mass media In 2014, about 15 percent of the world's prevention programs could promote quitting. population had access to smoking cessation The impact depends on the duration of the support programs, 2 percent more than in campaigns, especially among low-income 2012.6 These programs represent the fifth most smokers. It also depends on the message; widespread policy in the world (WHO 2015a). information about the adverse health risks of They significantly raise quit rates among smokers smoking are the most efficient means to reach who want to quit (Fiore and the Guideline Panel users. 2008). They are more cost-effective relative to with other health care programs (Cromwell c. Tobacco taxes et al. 1997). Even though they are an effective Although media campaigns are cost-effective, way to quit, they consider only the group of tobacco taxation is considered one of the addicted individuals who want to be treated. Their most efficient measures for reducing tobacco provision is also associated with country income, consumption; as a second-level benefit, they and they are widely concentrated in high-income also increase government revenue (World Bank countries. 1999)7. Because both effects are desirable from a policy standpoint, the use of taxes is economically Another way to discourage tobacco consumption justified on the externalities caused by tobacco. is through health warning labeling on tobacco The higher price elasticity of young people makes packages. This is the third most common policy taxes a good way to fight tobacco. Chile is one against cigarettes in the world. In 2014, almost of the countries covered by higher tobacco taxes 20 percent of the world’s population was being (WHO 2015a). exposed to such labeling. Warning labels are widely supported by the public and must be Institutions such as IDB (2010), WHO (2008), The redesigned regularly to maintain their impact. International Agency for Cancer Research (IARC, They may not represent a cost to governments 2011), and the World Bank (1999) and authors (WHO 2015a). The use of warnings may influence such as Levy et al. (2014) have associated price people against tobacco consumption; however, increases with significant declines in tobacco their use would account for only a marginal consumption. They estimate that higher taxes decrease (Borland 1997; Fathelrahman et al. 2009; are responsible for almost 50 percent of the Levy et al. 2012). decline in smoking. The effects of these policies 6 Smoking cessation support programs are implemented in 24 countries and cover 1.1 billion people (WHO 2015a). 7 Tobacco tax increases have also been associated with the increase of contraband and illegal tobacco sales, reducing the expected increase in government revenue. (Jha, P., & Chaloupka, F. J. (2000)) 8 Are Tobacco Taxes Really Regressive? Evidence from Chile. mainly depend on the type of taxes. Ad valorem (1998) and Iglesias and Nicolau (2006) calculate taxes consider a portion of the price; so tobacco the short-run price elasticity in Brazil, and Ramos companies have the opportunity to avoid higher and Curti (2006) do same in Uruguay. The results taxes by setting lower prices. For this reason, in both countries range between −0.2 and −0.6. In the levels of consumption and the amount of Argentina and Mexico, the estimations are slightly government tax revenue depend on the industry lower. Gonzáles Rozada and Rodríguez-Iglesias pricing strategy. Alternatively, specific excise (2013) and Olivera-Chávez et al. (2010) calculate a taxes establish a fixed tax amount, although they short-term price elasticity at between −0.10 and must be adjusted periodically for inflation to −0.30. Debrott Sánchez (2006) estimates a general accomplish their mission, and they are associated price elasticity in Chile of −0.21 and calculates with the risk of encouraging contraband sales that the price elasticity in the short and long term (WHO 2015a). The taxation system in Chile is −0.22 and −0.45, respectively. considers both types of tobacco taxes: the recent tax increase on tobacco included a reduction in Thus, some studies find lower-bound elasticities the ad valorem tax and a significant increase in between −0.1 and −0.3, while other researchers the specific excise tax. These measures are aligned find medium-bound elasticities between −0.3 with WHO (2010) best practice recommendations, and −0.6. A third set of studies finds higher-bound making almost every cigarette brand in Chile elasticities, from −0.6 to −1.5, in particular among equally costly. younger population groups and low-income countries. A similar classification can be found in d. The price elasticity of tobacco China (Hu et al. 2010). consumption The measurement of price elasticity is crucial in There are two important relationships between defining and calibrating taxation systems because tobacco price elasticities, income and age. it determines the sensitivity of demand to a People from low income groups have shown to change in tobacco prices. In general, tax increases have more elastic demands than medium and generate more impact on tobacco consumption higher income groups of the population (Jha & in low- and middle-income countries than in Chaloupka 1999). At the same time, young groups high-income countries (WHO 2015a). of the population are also more responsive to tax increases because they tend to be less affected There is an extensive literature defining the by addiction, more affected by peer effects and relationship between tobacco prices and at the same time have less disposable income consumption. Guindon (2013) provides a broad (Jha & Chaloupka 1999). Studies in United States review of 26 international studies. Chaloupka have consistently shown that younger groups and Grossman (1996) and Lewit and Coate (1981) of the population have higher elasticities than estimate the elasticity among under-18-year- older groups (for example, U.S. Centers for Disease olds in the United States at values of −1.44 and Control and Prevention,1998). Chaloupka and −1.31, respectively. Among adults ages 18 years Grossman (1996) and Lewit and Coate (1981). and older, Chaloupka (1991) and Lewit and Coate (1981) estimate the elasticity in the United States e. Further costs of tobacco: life, work, and at between −0.27 and −0.42. For Latin America, medical expenses Guindon, Paraje, and Chaloupka (2015) analyze The major costs of tobacco consumption beyond 32 studies and conclude that the price elasticity the direct price effects are associated with public is likely to be below −0.5. Carvalho and Lobão and private health costs. Tobacco-related health 9 costs are usually considered to be either direct or indirect. Direct costs include the monetary value of the consumption of goods and services motivated and, in many cases, compelled because of tobacco use. These are divided into health care costs (hospitalization, medication, medical supplies, equipment, and so on) and non–health care costs (job replacements for sick smokers, insurance, cleaning up the cigarette ash, stubs, packaging, and smoke residue of smokers, and so on). The indirect costs include the loss of productivity because of lost working days related to smoking illnesses and the value of the lives prematurely lost. Both effects are incorporated in the disability-adjusted life years indicator (WHO 2011). Focusing on the costs of health systems, Lightwood et al. (2000) estimate the cost of tobacco use. They suggest that the gross health cost in high-income countries fluctuates between 0.1 and 1.0 percent of gross domestic product (GDP). Limited data inhibit accurate estimates on low- and middle-countries, but the authors argue that the price elasticity could be as high as the elasticities in high-income countries. Meanwhile, Verguet et al. (2015) analyze the health effects of a price increase in China. Their research concludes that a 50 percent rise in prices would result in 231 million years of life gained over 50 years, with a significant impact in the lowest income quintile. Pichón-Riviere et al. (2014) estimate that tobacco use in Chile would reduce life expectancy by nearly 4.0 years among women and 4.3 years among men. There would also be about 379 thousand life days lost, which is over a thousand years.8 They estimate the annual direct cost of tobacco-related disease in the Chilean health system is about 0.6 percent of GDP. 8 Days lost refer to years of life lost (YLL) because of premature mortality. Another indicator is years lost because of disability (YLD) among people living with poor health and its consequences. Usually, disability- adjusted life years = YLL + YLD. 10 Are Tobacco Taxes Really Regressive? Evidence from Chile. 3. Model A partial equilibrium model allows the distributional effects of the tobacco tax to be We estimate the impact of the tobacco tax in assessed, resulting in an estimation of the first- Chile using a social welfare framework. This is order effects of these policies. We proceed using common in the literature (Pichón-Riviere et al. a partial equilibrium approach and evaluating 2014, Verguet et al. 2015). We estimate the way in the change in prices, which relies mainly on which tobacco taxes would change household household expenditure patterns. This decision income, analyzing three factors: (1) the rise in implies that we are assessing only the first-order tobacco expenditure because of the tax increase, effects, but does not include further behavioral (2) the reduction in medical expenses because of changes of economic agents. In consequence, less tobacco consumption, and (3) the change in we are observing the upper-bound effects of incomes because of the additional years of labor the short-term response. These assumptions recovered through the reduction in mortality.9 The imply that the model uses the share of tobacco aggregated effect of the tax policy is estimated as consumption in household budgets according follows: to price increases. The loss of real income arising from price increases in products i = 1, …, n is Income effect = change in tobacco expenditure (A) obtained by + lower medical expenses (B) + rise in income (C) (1) ∑n (ω + ∆ω ) * ∆Pi , (2) Our baseline scenario is derived from the Chilean i i i Pi,o Household Expenditure and Consumption Survey (Encuesta de Presupuestos Familiares). The latest where ωi is the share of product i in total available survey is from 2011–12, and we use it household expenditure, and ∆pi is the percent to measure general consumption and tobacco price increase.11 Therefore, if 10 percent of consumption, as well as medical expenditures.10 the total budget is destined for cigarettes, for Data limitations do not allow us to simulate the example, and the price of cigarettes increases by exact tobacco price increases by brand, but we 10 percent, the real loss of income amounts to can accomplish this through a relatively good 1 percent; ∆ωi is the change in consumption of proxy based on aggregate sales volumes and the taxed good, and it will depend on the price prices faced by all households. elasticity of the product. 9 Other studies have also evaluated productivity loss, disability costs, externalities, and so on. Because of the availability of relevant data, we focus solely on medical expenses and income changes associated with shifts in mortality. 10 The survey was collected every 10 years between 1956–57 and 2006–07; the latest round is the first one following the decision to undertake the survey every five years. The survey is representative nationally, as well as in the metropolitan areas of Santiago and other capitals of administrative divisions. The final sample of the 2011–12 survey includes 13,056 households. 11 For a detailed discussion of the methodology, see Coady et al. (2006) and Kpodar (2006). 11 Tobacco expenditure: We estimate the variation in tobacco consumption after the tax increase, considering the change in prices (∆P), the tobacco price elasticity ε, and the tobacco expenditure of decile i in period 0 (Expenditurei0). (3) ∆Tobacco Expenditurei=((1+∆P)(1+ε * ∆P)-1) * Expenditurei012 We divide the change in tobacco expenditure by the total expenditure for each decile group i, thereby obtaining a comparable per household measure of the change in tobacco expenditure relative to the total expenditure of each decile group. (4) ∆ Prop.Tobacco Expenditure = ((1+∆P)(1+ε * ∆P)-1) * Expenditurei0 Total Expenditurei Medical expenses: We estimate the change in medical expenses of tobacco-related diseases in equation (5), where the cost of the treatment of tobacco-related diseases for income decile i is obtained from Pichón-Riviere et al. (2014) and adjusted according to the expenditure survey. (5) ∆ Prop. Medical Exp. = ((1+ε * ∆P)-1) * Cost Treat.Tobacco Related Diseasesi Total Expenditurei Equation 5 shows the income gains related to the reduction of medical expenses because of the reduction of tobacco consumption in the long term. Although the calculation is not realistic in the short-term, we assume that the effects of tobacco-related diseases will immediately diminish with the reduction in tobacco consumption.13 12 Another expression might be ∆ Expenditure=∆C∆P + ∆CP + ∆PC . 0 0 13 Other studies have forecast the pass-through between the decline in tobacco consumption and the effect on medical expenditures. These estimates may also differentiate the effect associated with people who stop consuming tobacco versus people who do not start because of the tax policies. Because of data restrictions, we cannot use these assumptions in this paper. 12 Are Tobacco Taxes Really Regressive? Evidence from Chile. Increase in working life: We estimate the impact on incomes arising from the increase in labor years. In our base line, we estimate the income lost because of lost working years associated with tobacco consumption (equation 6)14. The years lost are distributed across each decile proportionately to the number of households that consume tobacco, and the income lost is estimated as the average income per household. We then estimate the effect of the tax increase in relation to the income gains because of increases in working years. We expect that incomes will increase as the number of years lost because of premature deaths from tobacco consumption decrease. ∆ Proportional Income = ((1+ε * ∆P)-1) * Years lost Per Decile*IncomeLossi (6) Total Expenditurei Lastly, we estimate the total income gains for each income group by adding the results of the reduction of medical treatments, the gains in working years, and the increase in tobacco expenditures (see equation 1). a. Elasticity parameters After defining the model to calculate the income impact of tobacco taxes, we examine the estimates in the literature on elasticities, disease prevalence, the cost of medical treatments, and mortality patterns. Several studies have estimated tobacco price elasticity in Chile and Latin America. Within this research, the work of Debrott Sánchez (2006) stands out for the use of prices and quantities among tobacco firms in Chile. The parameters estimated by Debrott Sánchez are −0.21 for the price elasticity and −0.23 for the income elasticity. Using a myopic consumption model, the author estimates a short-term elasticity of −0.22 and a long-term elasticity of −0.45. For China Verguet et al. (2015) consider an average price elasticity of −0.38, which varies from −0.64 in the poorest quintile to −0.12 in the richest quintile. 14 We assume that income is equal to the average consumption of each household per decile. 13 As our initial elasticity assumption, we use a After obtaining the changes in tobacco lower-bound price elasticity of −0.21, estimated consumption because of the price increase, we by Debrott Sánchez (2006). For a second level, estimate the incidence of tobacco consumption we use a medium-bound elasticity of −0.38, in medical treatments, lost years of work, and the considering variations for different income groups cost of these factors. The incidence of tobacco and extrapolating the price elasticities used by in medical spending was estimated by the Verguet et al. (2015) into 10 income deciles. We Ministry of Health of Chile and the Instituto de found that the average elasticity of Verguet et al. Efectividad Clínica y Sanitaria in Argentina, as (2015) is similar to other estimates for Chile and reported in Pichón-Riviere et al. (2014). Disease Latin American countries, but the innovation prevalence per age-group and the monetary is use a different elasticity profile for different cost of treatment have also been estimated. This income deciles15. We also consider an upper- information allows us to calculate the average bound elasticity of −0.75. This upper-bound cost of medical treatment for each disease. The elasticity tends to reflect a longer-term scenario, cost of lost working years can be estimated using echoing the effect the tobacco tax would have the Household Consumption Survey described in on the young population. After a few decades the next section. only these groups of the population would still be alive, therefore the total average effect of the price increase would be better approximated by the upper-bound price elasticity. Table 1 describes the assumed elasticities. Table 1: Tobacco price elasticities Decile Decile Decile Decile Decile Decile Decile Decile Decile Decile Elasticity 1 2 3 4 5 6 7 8 9 10 Lower-bound −0.21 −0.21 −0.21 −0.21 −0.21 −0.21 −0.21 −0.21 −0.21 −0.21 Medium-bound −0.64 −0.58 −0.52 −0.47 −0.41 −0.35 −0.29 −0.24 −0.18 −0.12 Upper-bound −0.75 −0.75 −0.75 −0.75 −0.75 −0.75 −0.75 −0.75 −0.75 −0.75 Sources: Lower-bound: Debrott Sánchez 2006; medium-bound: Verguet et al. 2015. 15 To our knowledge there are no papers that estimate different elasticity profiles per income deciles in Chile or other Latin American countries. 14 Are Tobacco Taxes Really Regressive? Evidence from Chile. b. Tobacco mortality and morbidity To estimate the medical costs of tobacco consumption, we examine tobacco-related medical treatments and deaths. The Department of Health Statistics and Information of the ministry provides information about mortality related to tobacco consumption in Chile during 2009. As a second source of information, the total number of tobacco-related events are obtained from Pichón- Riviere et al. (2014) (table 2). Table 2: Tobacco mortality and morbidity Deaths Events Tobacco Tobacco Total Total related related Mouth and throat cancer 200 163 349 255 Esophagus cancer 686 527 1,089 793 Stomach cancer 4,106 1,084 6,009 1,618 Pancreatic cancer 1,069 339 1,412 472 Laryngeal cancer 152 124 296 252 Lung cancer 2,532 2,175 3,517 3,076 Uterus cancer 868 153 2,176 389 Kidney cancer 617 225 1,374 461 Bladder cancer 394 172 999 467 Leukemia 396 82 457 98 Myocardial stroke 5,438 1,379 53,792 18,944 Other cardiovascular events 7,394 1,185 346 62 Cerebrovascular accident 8,130 4,305 55,314 12,050 Pneumonia and influenza 3,407 722 19,812 5,269 Bronchitis, emphysema and others 2,995 2,396 67,426 51,450 Total 38,384 15,031 214,368 95,656 Sources: Ministry of Health 2009; Pichón-Riviere et al. 2014. Note: Events refer to detected and treated diseases or affections. 15 c. Cost of treatment treatment, but the out-of pocket expenses that After producing estimates of the deaths and families face could be reduced by the health other events related to tobacco, we look for the insurance scheme currently existing in Chile. medical cost of treatments of tobacco-related Table 4 shows the annual medical costs of diseases. These costs are contained in Pichón- tobacco-related diseases. The results are taken Riviere et al. (2014). Table 3 shows estimates of the from calculations based on the data in tables 2 average cost of medical treatment for tobacco- and 3. The total medical costs of tobacco-related related diseases in the public and private sectors diseases are US$496 million. in Chile. These expenses refer to the total cost of Table 3: Cost of Medical Treatments (US$) First Year Second Year + Total Cost Mouth and throat cancer 15,651 10,681 26,332 Esophagus cancer 18,260 12,367 30,627 Stomach cancer 17,825 13,491 31,316 Pancreatic cancer 13,491 10,119 23,610 Laryngeal cancer 17,825 12,929 30,754 Lung cancer 21,738 28,107 49,845 Uterus cancer 13,477 7,299 20,776 Kidney cancer 15,651 10,962 26,613 Bladder cancer 14,782 13,491 28,273 Leukemia 23,477 26,983 50,459 Myocardial stroke 3,946 1,445 5,391 Other cardiovascular events 2,703 2,703 Cerebrovascular accident 4,433 1,522 5,956 Pneumonia and influenza 235 235 Bronchitis, emphysema, and others 552 552 Source: Pichón-Riviere et al. 2014. Note: Exchange rate US$1.00 = Ch$670. 16 Are Tobacco Taxes Really Regressive? Evidence from Chile. Table 4: Cost of Medical Treatments (US$) Total medical cost Total national cost Events Tobacco-related (US$) (US$) Mouth and throat cancer 255 26,332 6,714,605 Esophagus cancer 793 30,627 2,4286,992 Stomach cancer 1,618 31,316 50,669,703 Pancreatic cancer 472 23,610 11,143,912 Laryngeal cancer 252 30,754 7,750,037 Lung cancer 3,076 49,845 153,322,389 Uterus cancer 389 20,776 8,081,869 Kidney cancer 461 26,613 12,268,526 Bladder cancer 467 28,273 13,203,488 Leukemia 98 50,459 4,945,027 Myocardial stroke 18,944 5,391 102,131,232 Other cardiovascular events 62 2,703 167,595 Cerebrovascular accident 12,050 5,956 71,768,811 Pneumonia and influenza 5,269 235 1,238,341 Bronchitis, emphysema, 51,450 552 28,412,687 and others Total 95,656 333,442 496,105,212 Source: Calculations using data in Pichón-Riviere et al. 2014. Note: Exchange rate US$1.00 = Ch$670. We need to adjust national data from the Ministry Table 5: Medical Expenses Arising from of Health to the Household Consumption Survey. Tobacco Taxation (US$, million) The survey is representative of 60 percent of the Chilean population, particularly urban Expenditures Baseline households in the regional capitals of Chile. Therefore, it represents 10,516,229 individuals of Previous expenditures, total 496 the total population of 17.6 million in 2013. Thus, population the equivalent cost would be 60 percent of the Analyzed households 297 estimated US$496 million (table 5). The costs of Source: Calculations using data in Pichón-Riviere et al. 2014 tobacco-related medical expenses are distributed and the 2011 Household Consumption Survey. across the income deciles of the consumption Note: Exchange rate: US$1.00 = Ch$670. survey according to the proportion of households that consume tobacco in each income decile. 17 d. Age pattern of mortality To obtain the working years lost, we need to look at early mortality related to tobacco events. The medical events associated with tobacco have a strong correlation with age and years of smoking. Table 6 shows age patterns in tobacco-related deaths. These years lost will be distributed among the population in analysis, and the income (or wages) of this population segment is going to be used to estimate the working years lost. Table 6: Age Pattern of Mortality Age-group Disease Total 5–10 10–14 15–19 20–24 25–29 30–34 35–39 40–44 Stomach cancer 3,350 0 0 0 1 13 12 28 61 Lung cancer 2,532 1 1 4 3 2 7 15 20 Myocardial stroke 5,438 0 0 4 9 6 27 37 84 Cerebrovascular 8,130 4 4 10 16 20 28 51 114 accident Pneumonia and 2,407 7 3 10 7 16 17 37 65 influenza Others 6,466 8 4 18 9 18 22 45 76 Total 28,323 20 12 46 45 75 113 213 420 Disease Total 50–54 55–59 60–64 65–69 70–74 75–79 80–84 85+ Stomach cancer 3,350 176 237 341 459 509 539 462 399 Lung cancer 2,532 125 224 327 420 418 413 262 222 Myocardial stroke 5,438 299 400 535 609 637 733 780 1,072 Cerebrovascular 8,130 286 350 466 673 880 1,332 1,499 2,210 accident Pneumonia and 2,407 75 94 108 154 231 368 618 1,634 influenza Others 6,466 12 46 45 75 113 213 420 825 Total 28,323 973 1,351 1,822 2,390 2,788 3,598 4,041 6,362 Source: Ministry of Health 2009. 18 Are Tobacco Taxes Really Regressive? Evidence from Chile. e. Summary of descriptive statistics Table 7 summarizes the most important indicators, which shows total monthly expenditure from the Household Consumption Survey, as well as the proportion of cigarette expenditure. We estimate the cost of medical treatment of tobacco-related diseases as a proportion of monthly income for each income decile. We then estimate the share of income lost because of the working years lost because of tobacco-related mortality. Table 7: Baseline Descriptive Results Price shock Decile Decile Decile Decile Decile Decile Decile Decile Decile Decile under 1 2 3 4 5 6 7 8 9 10 Total monthly 227 399 484 583 731 851 1,027 1,272 1,788 3,467 expenditure (US$) Households that 21% 25% 29% 29% 35% 32% 35% 37% 32% 29% smoke* Cigarette expenditure/total 1.42% 1.55% 1.42% 1.31% 1.33% 1.05% 1.01% 0.95% 0.65% 0.35% expenditure* Share of tobacco health 2.94% 2.08% 1.81% 1.46% 1.41% 1.04% 0.88% 0.70% 0.41% 0.18% expenditure* Income loss: 0.60% 0.74% 0.79% 0.76% 0.93% 0.79% 0.81% 0.80% 0.66% 0.55% working years* * Proportional to total consumption in each decile. 19 4. Results in table 7, and a price increase of 25 percent would give an increase in expenditure of 0.26 percent After obtaining the baseline results, described in (equation 7). This represents a loss in welfare among table 7, we estimate the effect of the tax increase consumers. The results for all income deciles and on prices, medical expenditures, and income elasticity scenarios are shown table 8. and aggregate these three effects into a single measure. We use the three scenarios in the Across the three elasticities, the direct effect tobacco price elasticity: lower-bound, medium- of the tobacco tax is a welfare loss, but, in the bound, and higher-bound, presented in table 1. lower-bound elasticity, the effect is clearly These three scenarios allow us to have a notion of regressive because the lower-income groups how results change under different assumptions. lose proportionally more of their incomes (figure 1). In the medium-bound elasticity, the effect of a. Tobacco price increase elasticities varies across different income groups, As a first step, we estimate the income changes making the tax effect more flat. To show the effect for each income decile arising from the increase in of the elasticities on prices, table 8 includes the tobacco prices based on low-, medium-, and upper- estimates on a complete pass-through scenario, bound elasticity. Using equation (4) and tables 1 and whereby the increase in prices is passed on 7, we can calculate the effects of the tobacco price completely to consumers without a reduction in increase. For example, the lower-bound elasticity consumption. (−0.21) in table 1, the proportion of tobacco expenditure among the first decile (1.42 percent) Table 8: Direct Effect of Price Increase of Taxes Price shock Decile Decile Decile Decile Decile Decile Decile Decile Decile Decile under 1 2 3 4 5 6 7 8 9 10 Complete pass- −0.35% −0.39% −0.36% −0.33% −0.33% −0.26% −0.25% −0.24% −0.16% −0.09% through Low-bound −0.26% −0.29% −0.26% −0.24% −0.25% −0.19% −0.19% −0.18% −0.12% −0.06% elasticity Medium −0.07% −0.11% −0.12% −0.14% −0.16% −0.15% −0.16% −0.17% −0.13% −0.07% elasticity Upper- bound −0.02% −0.02% −0.02% −0.02% −0.02% −0.02% −0.02% −0.01% −0.01% −0.01% elasticity Income loss: working 0.60% 0.74% 0.79% 0.76% 0.93% 0.79% 0.81% 0.80% 0.66% 0.55% years* Source: Based on data of the 2011 Household Consumption Survey. Note: The table shows the share of total consumption of each decile. Complete pass-through refers to elasticity equal to zero, consumers pay all the increase in prices. 20 Are Tobacco Taxes Really Regressive? Evidence from Chile. Figure 1: Income Gains: Direct Effect of Tobacco Taxes Increase of expenditure due to tobacco tax 0 Upper Bound Elasticity Income Gains (%) Medium Elasticity - -0.1 Decile Variations Lower Bound Elasticity -0.2 -0.3 0 2 4 6 8 10 Income Decile Source: Author's estimation using a price shock of 25% b. Medical expenses whereby a reduction in tobacco consumption We estimate the cost of yearly medical expenses would be strongly related to a reduction in associated with tobacco consumption, assuming tobacco-related diseases. Health expenditures are a direct medical impact on health. Although this estimated using equation (5) and tables 1 and 7. assumption is unrealistic in the short run, the For example, using the medium-bound elasticity long-run reduction of tobacco consumption assumption, the result for the first decile would be would tend to behave according to this pattern, as follows: ∆ Medical Exp1=((1-0.64 * 25%) -1) * 2.94%=0.469 (8) Table 9: Reduction in Medical Costs (%) Price shock Decile Decile Decile Decile Decile Decile Decile Decile Decile Decile under 1 2 3 4 5 6 7 8 9 10 Lower- bound 0.15 0.11 0.10 0.08 0.07 0.05 0.05 0.04 0.02 0.01 elasticity Medium 0.47 0.30 0.24 0.17 0.14 0.09 0.06 0.04 0.02 0.01 elasticity Upper- bound 0.55 0.39 0.34 0.27 0.26 0.19 0.17 0.13 0.08 0.03 elasticity Source: Based on data of the 2011 Household Consumption Survey. Note: The table shows the share of total consumption of each decile. 21 Table 9 reports the income gains derived from the c. Income gains arising because of longer reduction in medical expenses for each income years of working life decile. The reduction in tobacco consumption We estimate the cost of working life lost because of would have a positive effect on income through tobacco consumption, assuming that there is a direct the reduction in medical treatments. The income impact of lower tobacco use on health and work- gains would vary between 0.15 and 0.00 percentage generated income. We calculate the impact for each points in the case of the lower-bound elasticity income decile, using the age pattern of mortality and assumption, between 0.49 and 0.00 percentage estimating the years of life lost. We then estimate the points in the case of the medium-bound elasticity, welfare effect using the lower-bound price elasticity, and between 0.55 and 0.03 percentage points in medium-bound price elasticity, and upper-bound the case of the upper-bound elasticity (figure 2). elasticity with decile variations. These results show the importance of the elasticity assumptions; they also stress the relevance to the possible elasticity variations across income groups. Figure 2: Income Gains: Medical Costs of Tobacco Taxes Reduction of expenditure expenditures 0.5 Upper Bound Elasticity 0.4 Income Gains (%) Medium Elasticity - Decile Variations 0.3 Lower Bound Elasticity 0.2 0.1 0 0 2 4 6 8 10 Income Decile Source: Author's estimation using a price shock of 25% 22 Are Tobacco Taxes Really Regressive? Evidence from Chile. The 15,031 deaths attributed to tobacco The results show that the reduction in tobacco consumption are distributed using the occurrence consumption and the expected reduction in work of mortality profile. For each death, the number of years lost have positive impacts on welfare. In working days left are calculated, and the working the first scenario, the gains are evenly distributed years lost are divided across the deciles according across income groups. However, elasticities vary to decile tobacco consumption. Using equation across deciles, generating an important impact on (6) and tables 1 and 7, we can calculate the results lower-income groups (figure 3). of the tax increase. For example, for the first decile assuming the higher-bound elasticity, the income increase arising because of gains in working life would be as follows: ∆ Income1=((1-0.75 * .25) -1) * 0.60%=0.11% (9) Table 10 shows the results for all decile groups using the three elasticity scenarios. Table 10: Years of Working Life Lost (%) Price shock Decile Decile Decile Decile Decile Decile Decile Decile Decile Decile under 1 2 3 4 5 6 7 8 9 10 Low-bound 0.03 0.04 0.04 0.04 0.05 0.04 0.04 0.04 0.03 0.03 elasticity Medium 0.10 0.11 0.10 0.09 0.09 0.07 0.06 0.05 0.03 0.02 elasticity Upper-bound 0.11 0.14 0.15 0.14 0.17 0.15 0.15 0.15 0.12 0.10 elasticity Source: Based on data of the 2011 Household Consumption Survey. Note: The table shows the share of total consumption of each decile. Figure 3: Income Gains: Production During Years Lost Production during years lost by income decile .15 Upper Bound Elasticity Income Gains (%) Medium Elasticity - .10 Decile Variations Lower Bound Elasticity .05 0 0 2 4 6 8 10 Income Decile Source: Author's estimation using a price shock of 25% 23 d. Out-of-Pocket Medical expenses associated with tobacco consumption. The For a country like Chile, where 90% of the household survey has a few limitations regarding population is covered by either public or private the identification of tobacco out of pocket related health insurance systems the total medical cost medical expenses. First, descriptive results at the of tobacco related diseases reflects the total households’ level show a positive relation between average societal cost of the tobacco burden17. On health expenses and income level. Nonetheless, the other hand, from a household perspective, households that consume tobacco spend less the out-of-pocket expenditure could be the most money on health than those where nobody relevant one. According to the OECD Health Data, smokes. These unexpected results could be due out-of-pocket expenditures represent 33% of total to several factors. First, the use of a household medical costs. For tobacco-related diseases, cross- consumption survey and not an individual survey, subsidies on health treatment, selectivity of public leads us to inexact statistics. It may be necessary versus private hospitals or governmental subsidies to control for number of household members, of health care could provide relevant changes in their age, and tobacco consumption. In particular out-of-pocket expenditures. kids and old adults that do not smoke could be increasing medical expenses. Second, people that Using the household consumption survey, we smoke may be less inclined to look for medical try to identify out-of-pocket medical expenses attention and preventive care. Figure 4: Out of Pocket Medical Expenses and Tobacco Consumption Household Survey: weighted sample .1 No Tobacco Consumption Income Gains (%) .08 Tobacco Consumption .06 95% CI .04 .02 0 0 2 4 6 8 10 Income Decile Source: Author's estimation using a price shock of 25% 17 The private insurance companies in Chile (ISAPRES) can differentiate via income, age, gender and pre- existing conditions. From an aggregate point of view, the amount paid by individuals would tend to reflect the average cost of their of health expenditures. On the other hand, the public health insurance system does not provide these types of differentiations, but only due to income. ISAPRES insure 16.5% of the country population, while the public insurance (FONASA) insures 74.1%. The private insurance convers a national average of 60% of medical costs (Superintendencia de Salud, 2010). From a national level perspective, out- of-pocket medical expenditures represent 33% of medical expenditures in the country, representing 4.6% of the households incomes, one of the highest in the OECD. (Health Data, OECD 2012) 24 Are Tobacco Taxes Really Regressive? Evidence from Chile. To approximate the first problem, we analyze only single member households, with members over 40 years old. For these households we can see a patter between Medical expenses and tobacco consumption. In particular, individuals that with tobacco monthly consumption of over US$ 100 show high medical expenses. In the appendix we add estimations for single households with members over 20, 30, 40 and 50 years old. These results provide us with hints regarding the limitation of a household survey, and the importance of medical expenses for heavy smokers. Figure 5: Out of Pocket Medical Expenses and Tobacco Consumption Household Survey: weighted sample .3 Monthly Medical Expenses Income Proportion Fitted values .2 95% CI .1 0 0 50 100 150 200 Income Decile Source: Author's estimation Nevertheless, we can forecast results for this group As out-of-pocket health expenditures account using a linear estimation of medical expenses for 5.16% of all expenditures, and households for this group of the population we can obtain that have smokers account for 30.5% of the an elasticity of out-of-pocket medical expenses sample, we could obtain a rough estimation for and tobacco consumption. A one percent the reductions on tobacco consumption. Table increase in tobacco consumption is associated 11 shows the estimations for reduction in out of with a 0.12% increase in medical expenditures. pocket medical expenses. ∆ Out of Pocket Medical Exp. = ((1+ε * ∆P)-1) * 0.12 * Out of Pocketi Total Expenditurei 25 As an example, for the lower-bound elasticity, the e. Net effects: Total distributional impacts change in out-of pocket expenditure would be: After separately calculating the effects of tobacco ((1-0.21*0.25)-1)*0.0516*0.305*0.12=-0.01% tax policy on prices, medical expenditures, and working years’ gains, in this subsection we add Table 11: Reduction of out-of pocket the results. Using a lower-bound elasticity the Medical Expensese (%) results show the net negative effect of tobacco tax policy; it is regressive, affecting the lower- Change Income income groups of the population more severely Out-of Pocket Medical Prop (table 12; figure 5). Obviously, a population that Low-bound elasticity -0.01% is not sensitive to tobacco price changes will not reduce consumption sufficiently to allow health Medium elasticity -0.02% and work benefits to offset cost increases. For medium-bound elasticity, the effects are not Upper-bound elasticity -0.04% so clear. Among lower-income groups of the Source: Based on data of the 2011 Household Consumption population, the effects of the tobacco tax would Survey. Assumptions: out-of-pocket tobacco elasticity =0.12%, Out-of-pocket medical expenses 5.16%, proportion be positive, while, among higher-income groups, of smokers 30.5% Note: The table shows the share of total the tax would have a negative effect. Under these consumption circumstances, the tax would have a progressive distributional effect, benefiting lower-income These results give us a dimension of the groups in larger proportion. Although this effect magnitude of out-of-pocket expenses and the is driven mostly by the elasticity variance among reduction of tobacco consumption, for a very income deciles, the elasticity level is also relevant. limited group of the population, and without For upper-bound elasticity, the tax would have estimating the effect of second hand smoking. a positive effect on all income groups and also show a progressive pattern, that is, greater benefits among lower-income groups. Table 12: Total Net Effect (%) Price shock Decile Decile Decile Decile Decile Decile Decile Decile Decile Decile under 1 2 3 4 5 6 7 8 9 10 Low-bound −0.08 −0.14 −0.13 −0.12 −0.12 −0.10 −0.10 −0.10 −0.06 −0.03 elasticity Medium 0.49 0.30 0.22 0.12 0.08 0.01 −0.03 −0.08 −0.08 −0.05 elasticity Upper-bound 0.64 0.50 0.46 0.40 0.42 0.33 0.30 0.27 0.19 0.13 elasticity Source: Source: Based on data of the 2011 Household Consumption Survey. Note: The table shows the share of total consumption of each decile. 18 The results for the Medium-bound elasticity are highly dependent on the decile profile chosen. This particular profile follows the paper from Verguet et al. (2015) in China. 26 Are Tobacco Taxes Really Regressive? Evidence from Chile. Figure 6: Total Income Effect: Direct and Indirect Effect of Taxes (tobacco price increase, medical expenditure and working years gained) .6 Upper Bound Elasticity Income Gains (%) .4 Medium Elasticity - Decile Variations .2 Lower Bound Elasticity 0 -.2 0 2 4 6 8 10 Income Decile Source: Author's estimation using a price shock of 25% In conclusion, the tobacco tax increase has a small but negative effect in the presence of a low tobacco price elasticity. This assumption does not consider other policies, such as smoking cessation and advertising programs that may be implemented in parallel. On the assumption of medium-bound elasticity, with variations across income deciles, we find that the tobacco tax generates welfare gains among lower-income groups. In a scenario of upper-bound elasticity, there would be income gains across all groups of the population. Moreover, the impact would be particularly important in the case of lower-income deciles. These results are in line with the literature, showing the important benefits of the reduction in tobacco consumption that may occur through diverse policy mechanisms. 27 5. Discussion increase in working years, the scenario reverses the negative effects, and the overall monetary The tobacco literature has shown the negative effect of the taxation policy becomes positive effects of tobacco consumption on health and in all cases, with the exception of lower-bound well-being, as well as the benefits of various elasticity. The reduction in medical expenses is public policy mechanisms aimed at reducing the main driver of the increase in net incomes tobacco use. One of the most efficient ways to because of the reduction in tobacco-related deter tobacco use is the application of taxes, both problems that require expensive treatments. ad valorem or excise taxes. Yet, questions remain In the three variations based on elasticity, the regarding the net positive effect of these policies benefits of medical expenses are greater than and the progressivity or regressivity that they the benefit of the increase in working years, entail, particularly because the welfare effects particularly among lower-income groups. This is derived from the increase in the taxes on tobacco because of the reduced salaries among the lower- heavily depend on the price elasticity of this item income groups; while medical expenses remain across different sectors of the population. The constant across all segments of the population, price elasticity will determine the magnitude of the income increases are lower among the lower- the income shock, as well as the increase in the income groups. benefits, because of the reduction in tobacco consumption. The different elasticity assumptions generated three particular results. The lower-bound elasticity The net welfare gain may be supported through assumption generated an income loss across all the reduction in medical costs and the expansion decile groups in the population. The medium- in working years, which are an effect of the lower bound elasticity, with decile variations, generated levels of tobacco consumption promoted by income gains among lower-income groups and the tax increase. Thus, it is critical to justify the small losses among the upper-income groups. maintenance or intensification of the use of these The upper-bound elasticity generated income kinds of policies by means of a demonstration gains across all groups of the population, but of the aggregated monetary gains or losses particularly the lower-income groups. This means generated. At the same time, a responsible and that the effect of taxes on the different income comprehensive policy analysis should address groups is exacerbated if the variation of price- the differentiated effects among particular elasticities among income deciles is higher. socioeconomic groups of the population, which is one of the main motivations of this research. The three price elasticity scenarios mimic the short- versus the long-term effects of the tobacco Results show that, considered by itself, a price tax. There is evidence that adult smokers will increase for tobacco through higher taxes only present small changes in their behavior if generates negative income variations across all price increases, the lower-bound elasticity would groups in a population, because the overall prices tend to represent this behavior more closely. go up. These negative effects are particularly In contrast, younger groups of the population acute in the scenario of lower-bound elasticity, would show more elastic demand, similar to the and they are more moderate as elasticity expands upper-bound elasticity. After a few decades, we in absolute terms. Based on an assumption of expect the impact of the tax policy to resemble a more comprehensive approach, including the upper-bound elasticity scenario. benefits through lower medical expenses and an 28 Are Tobacco Taxes Really Regressive? Evidence from Chile. The results provide evidence to support the maintenance or increase in tobacco taxes. The analysis also shows the importance of the price elasticity of tobacco, which indicates the relevance of the interaction of taxation and other policies. In particular, this research suggests that taxation should always be supported by actions that could change or alter the behavior of consumers, increasing elasticity levels. Because the reaction to a price change is different in each income group, specific programs should be implemented to target consumption among lower-income groups by supporting smoking cessation and through advertising adapted to the particular sociocultural context. There is thus a need for an integrative approach to anti-tobacco policies that favor coordination between taxation and behavioral change policies to generate higher social returns. Further research could investigate which combination of programs is more efficient in deterring smoking among each socioeconomic group. 29 6. References Fathelrahman, Ahmed I., Maizurah Omar, Rahmat Awang, Ron Borland, Geoffrey T. Fong, David Hammond, and Zarihah Zain. Borland, Ron. 1997. “Tobacco Health Warnings and Smok- 2009. “Smokers’ Responses toward Cigarette Pack Warning ing-Related Cognitions and Behaviours.” Addiction 92 (11): Labels in Predicting Quit Intention, Stage of Change, and 1427–35. Self-Efficacy.” Nicotine and Tobacco Research 11 (3): 248–53. Carvalho José L., and Waldir Lobão. 1998. “Vício privado e Fiore, Michael C., and the Guideline Panel. 2008. “A Clinical políticas públicas: a demanda por cigarros no Brasil.” Revista Practice Guideline for Treating Tobacco Use and Dependence: Brasileira de Economia 52: 67–104. 2008 Update.” American Journal of Preventive Medicine 35 (2): 158–76. Centers for Disease Control and Prevention (CDC. (1998). Response to increases in cigarette prices by race/ethnicity, Gonzáles Rozada, Martín, and Germán Rodríguez-Iglesias. income, and age groups--United States, 1976-1993. MMWR. 2013. “Analysis of the Cigarette Tax Structure and Cigarette Morbidity and mortality weekly report, 47(29), 605. Demand in Argentina.” Fundación Interamericana del Corazón, Buenos Aires. Chaloupka, Frank J. 1991. “Rational Addictive Behavior and Cigarette Smoking.” Journal of Political Economy 99 (4): 722–42. Guindon, G. Emmanuel. 2013. “The Impact of Tobacco Prices on Smoking Onset: A Methodological Review.” Tobacco Control Chaloupka, Frank J., K. Michael Cummings, Chris P. Morley, 23 (2): e5. and Janet K. Horan. 2002. “Tax, Price, and Cigarette Smoking: Evidence from the Tobacco Documents and Implications for Guindon, G. Emmanuel, Guillermo R. Paraje, and Frank J. Cha- Tobacco Company Marketing Strategies.” Tobacco Control 11 loupka. 2015. “The Impact of Prices and Taxes on the Use of To- (1): i62–i72. bacco Products in Latin America and the Caribbean.” American Journal of Public Health 105 (3): e9–e19. Chaloupka, Frank J., and Michael Grossman. 1996. “Price, Tobacco Control Policies, and Youth Smoking.” NBER Work- HHS (U.S. Department of Health and Human Services). 2004. ing Paper 5740 (September), National Bureau of Economic The Health Consequences of Smoking: A Report of the Research, Cambridge, MA. Surgeon General. Atlanta: Office on Smoking and Health, National Center for Chronic Disease Prevention and Health Coady, David, Moataz El-Said, Robert Gillingham, Kangni Kpo- Promotion, Centers for Disease Control and Prevention, U.S. dar, Paulo Medas, and David Newhouse. 2006. “The Magnitude Department of Health and Human Services. and Distribution of Fuel Subsidies: Evidence from Bolivia, Gha- na, Jordan, Mali, and Sri Lanka.” IMF Working Paper WP/06/247, ———. 2014. The Health Consequences of Smoking, 50 Years International Monetary Fund, Washington, DC. of Progress: A Report of the Surgeon General. January. Atlanta: Office on Smoking and Health, National Center for Chronic Cromwell, Jerry, William J. Bartosch, Michael C. Fiore, Victor Disease Prevention and Health Promotion, Centers for Disease Hasselblad, and Timothy Baker. 1997. “Cost-Effectiveness of the Control and Prevention, U.S. Department of Health and Clinical Practice Recommendations in the AHCPR Guideline Human Services. http://www.surgeongeneral.gov/library/ for Smoking Cessation.” Journal of the American Medical Associ- reports/50-years-of-progress/full-report.pdf. ation 278 (21): 1759–66. Hu, Teh-wei, Zhengzhong Mao, Jian Shi, Wendong Chen. 2010. Debrott Sánchez, David. 2006. “Economía del Control del “The Role of Taxation in Tobacco Control and Its Potential Tabaco en los países del Mercosur y Estados Asociados: Chile.” Economic Impact in China.” Tobacco Control 19 (1): 58–64. Pan American Health Organization, Washington, DC. http://iris. paho.org/xmlui/handle/123456789/2811. IARC (International Agency for Research on Cancer). 2009. Evaluating the Effectiveness of Smoke-Free Policies. Vol. 13 of Denisova, Irina, and Polina Kuznetsova. 2014. “The Effects of Handbooks of Cancer Prevention, Tobacco Control. Lyon, France: Tobacco Taxes on Health: An Analysis of the Effects by Income IARC, World Health Organization. https://www.iarc.fr/en/publi- Quintile and Gender in Kazakhstan, the Russian Federation, cations/pdfs-online/prev/handbook13/handbook13-0.pdf. and Ukraine.” Discussion Paper 92765 (October), World Bank, Washington, DC. IDB (Inter-American Development Bank) Arpón, P., Bitrán, R., & Escobar, L. (2011). Consecuencias del aumento del impuesto Doll, Richard, and A. Bradford Hill. 1956. “Lung Cancer and Oth- al tabaco en Chile en el año 2010. er Causes of Death in Relation to Smoking: A Second Report on the Mortality of British Doctors.” British Medical Journal 2 Iglesias Roberto, and Júlia Nicolau. 2006. “A economia do (5001): 1071–81. controle do tobaco nos países Mercosul e associados: Brasil.” Pan American Health Organization, Washington, DC. Durkin, Sarah, Emily Brennan, and Melanie Wakefield. 2012. “Mass Media Campaigns to Promote Smoking Cessation Jha, P., & Chaloupka, F. J. (2000). The economics of global among Adults: An Integrative Review.” Tobacco Control 21 (2): tobacco control. British Medical Journal, 321(7257), 358. 127–38. http://tobaccocontrol.bmj.com/content/21/2/127.full. 30 Are Tobacco Taxes Really Regressive? Evidence from Chile. Jha, Prabhat, and Richard Peto. 2014. “Global Effects of Smok- Pichón-Riviere, Andrés, Ariel Bardach, Joaquín Caporale, An- ing, of Quitting, and of Taxing Tobacco.” New England Journal of drea Alcaraz, Federico Augustovski, Francisco Caccavo, Carlos Medicine 370 (1): 60–68. Vallejos, et al. 2014. “Carga de enfermedad atribuible al tab- aquismo en Chile.” Documento Técnico IECS 8 (April), Instituto Kpodar, Kangni. 2006. “Distributional Effects of Oil Price de Efectividad Clínica y Sanitaria, Buenos Aires. Changes on Household Expenditures: Evidence from Mali.” IMF Working Paper WP/06/91, International Monetary Fund, Ramos, Alejandro, and Dardo Curti. 2006. “Economía del con- Washington, DC. trol del tabaco en los países del Mercosur y Estados asociados: Uruguay.” Pan American Health Organization, Washington, DC. Levy, D., de Almeida, L. M., & Szklo, A. (2012). The Brazil Sim- Smoke policy simulation model: the effect of strong tobacco Verguet, Stéphane, Cindy L. Gauvreau, Sujata Mishra, Mary control policies on smoking prevalence and smoking-attrib- MacLennan, Shane M. Murphy, Elizabeth D. Brouwer, Rachel utable deaths in a middle income nation. PLoS Med, 9(11), A. Nugent, Kun Zhao, Prabhat Jha, and Dean T Jamison. 2015. e1001336. “The Consequences of Tobacco Tax on Household Health and Finances in Rich and Poor Smokers in China: An Extend- Levy, D., Rodríguez-Buño, R. L., Hu, T. W., & Moran, A. E. (2014). ed Cost-Effectiveness Analysis.” Lancet Global Health 3 (4): The potential effects of tobacco control in China: projections e206–16. from the China SimSmoke simulation model. WHO (World Health Organization). Lewit, Eugene M., and Douglas Coate. 1981. “The Potential for ———. 2008. WHO report on the global tobacco epidemic, Using Excise Taxes to Reduce Smoking.” NBER Working Paper 2008: the MPOWER package. 764 (September), National Bureau of Economic Research, Cambridge, MA. ———. 2010. WHO Technical Manual on Tobacco Tax Adminis- tration. Geneva: WHO. Lightwood, James, David Collins, Helen Lapsley, and Thomas E. Novotny. 2000. “Estimating the Costs of Tobacco Use.” In Tobac- ———. 2011. “Economics of Tobacco Toolkit: Assessment of co Control in Developing Countries, edited by Prabhat Jha and the Economic Costs of Smoking.” WHO, Geneva. http://apps. Frank J. Chaloupka, 63–103. Oxford: Oxford University Press. who.int/iris/bitstream/10665/44596/1/9789241501576_eng. pdf. Ministry of Health, Chile. 2009. Estadísticas de Natalidad y Mor- talidad Chile, Año 2009. Santiago, Chile: Department of Health ———. 2011. Effectiveness of Tax and Price Policies for Tobac- Statistics and Information, Ministry of Health. co Control. IARC Handbooks of Cancer Prevention in Tobacco Control, 14. Ng, Marie, Michael K. Freeman, Thomas D. Fleming, Margaret Robinson, Laura Dwyer-Lindgren, Blake Thomson, Alexandra ———. 2015a. Global Report on Trends in Prevalence of Tobacco Wollum, et al. 2014. “Smoking Prevalence and Cigarette Con- Smoking. Geneva: WHO. http://apps.who.int/iris/bitstre sumption in 187 Countries, 1980–2012.” Journal of the American am/10665/156262/1/9789241564922_eng.pdf?ua=1http:// Medical Association 311 (2): 183–92. www.who.int/mediacentre/news/releases/2015/trends-to- bacco-use/en/. Olivera-Chávez, Rosa Itandehui, Rodolfo Cermeño-Bazán, Belén Sáenz de Miera-Juárez, Jorge Alberto Jiménez-Ruiz, and ———. 2015b. “WHO Report on the Global Tobacco Epidem- Luz Myriam Reynales-Shigematsu. 2010. “The Effect of Tobacco ic, 2015: Raising Taxes on Tobacco.” WHO, Geneva. http://apps. Prices on Consumption: A Time Series Data Analysis for Mexi- who.int/iris/bitstream/10665/178574/1/9789240694606_eng. co.” Salud Pública de México 52 (2): S197–S205. pdf?ua=1&ua=1. OECD Health Statistics 2012. World Bank. 1999. Curbing the Epidemic: Governments and the Economics of Tobacco Control. Report 19638 (May). Develop- OECD Health at a Glance 2013. ment in Practice Series. Washington, DC: World Bank. http:// documents.worldbank.org/curated/en/1999/05/437174/ Peto, Richard, and Alan D. Lopez. 2004. “The Future Worldwide curbing-epidemic-governments-economics-tobacco-control. Health Effects of Current Smoking Patterns.” In Tobacco and Public Health: Science and Policy, edited by Peter Boyle, Nigel Wynder, Earnest L., and Evarts A. Graham. 1950. “Tobacco Gray, Jack Henningfield, John Seffrin, and Witold Zatonski, Smoking as a Possible Etiologic Factor in Bronchiogenic Carci- 281–86. New York: Oxford University Press. noma: A Study of Six Hundred and Eighty-Four Proved Cases.” Journal of the American Medical Association 143 (4): 329–36. Pichón-Riviere, Andrés, Federico Augustovski, Ariel Bardach, Lisandro Damian Colantonio, and Latinclen Tobacco Research Group. 2011. “Development and Validation of a Microsimula- tion Economic Model to Evaluate the Disease Burden Asso- ciated with Smoking and the Cost-Effectiveness of Tobacco Control Interventions in Latin America.” Value in Health 14 (5): S51–S59. 31 32 Are Tobacco Taxes Really Regressive? Evidence from Chile. Appendix Income Gains: Direct and Indirect Effect of Tobacco Taxes Lower Bound Elasticity = -0.21 .1 Reduction Medical Costs 0 Years Lost Income Gains (%) Net-Effect Tobacco Taxes -.1 Tax Increase -.2 -.3 0 2 4 6 8 10 Income Decile Source: Author's estimation using a price shock of 25% Income Gains: Direct and Indirect Effect of Tobacco Taxes Medium Elasticity = Decile Variations .4 Reduction Medical Costs Years Lost Income Gains (%) .2 Net-Effect Tobacco Taxes Tax Increase 0 -.2 0 2 4 6 8 10 Income Decile Source: Author's estimation using a price shock of 25% 33 Income Gains: Direct and Indirect Effect of Tobacco Taxes Upper Bound Elasticity = 0.75 .6 Reduction Medical Costs Years Lost Income Gains (%) .4 Net-Effect Tobacco Taxes Tax Increase .2 0 0 2 4 6 8 10 Income Decile Source: Author's estimation using a price shock of 25% Elasticity of out-of pocket medical expenses for single member households. Table 13: Elasticity Tobacco and out-of pocket Medical Expenses Age Household Elasticity P Value Over 20 .0096678 0.027 Over 30 .0108653 0.062 Over 40 .0125843 0.058 Over 50 .0218003 0.002 Source: Based on data of the 2011 Household Consumption Survey. Note: Total health expenses estimated as share of total consumption 34 35