Article Text
PDF
Abstract
Background Brazil has experienced a persistent and substantial reduction in the prevalence of smoking in the population since 2006 due to increased taxes on tobacco and other tobacco control policies. Despite the effectiveness of these measures, however, the socioeconomic costs of smoking are still very high. Tobacco taxation in Brazil plays an important role among the measures adopted to curb tobacco use.
Methods The study combines data from the National Household Sample Survey of 2008 and the National Health Survey of 2013 and applies cross-section, pooled, and probit estimations, to estimate price elasticities of tobacco consumption by distinct population cohorts. The paper presents a comprehensive cost–benefit analysis resulting from a one-time tax increase on manufactured cigarettes using estimated conditional price elasticity of cigarette consumption and probability of smoking by income and age quartiles.
Findings Each 10% price increase (BRL 0.54), due to higher tobacco taxes, reduces cigarette consumption by about 5%, and for poor smokers, it would lead to net income gains by about BRL 39.00 per month (in 2019 values). The highest net income effects were observed for the younger, aged between 15 and 29 years, and for middle-aged individuals, between 40 and 59 years old. Higher tobacco taxes lead to lower medical expenses on tobacco-related diseases and a longer, healthier and more productive life. Most importantly, this policy is progressive, as its economic effects are stronger for the poorer than for the richer according to the income quartiles.
Conclusions A tax increase that rises cigarette prices generates significant social benefits by reducing tobacco spending and medical expenses on tobacco-related diseases and raising future years of life and net income. The total benefits for the individual and the society go way beyond the public finance improvement.
- taxation
- economics
- public policy
- surveillance and monitoring
- tobacco industry
Data availability statement
Data are available in a public, open access repository. The public data used in the paper are: (1) National Household Sample Survey (PNAD) de 2008, available at https://www.ibge.gov.br/estatisticas/sociais/educacao/9127-pesquisa-nacional-por-amostra-de-domicilios.html?edicao=9128&t=downloads (2) National Health Survey (PNS) de 2013, available at https://www.ibge.gov.br/estatisticas/sociais/saude/9160-pesquisa-nacional-de-saude.html?=&t=downloads (3) Risk Factor Surveillance and Protection for Chronic Diseases by Telephone Survey (VIGITEL), which is performed annually, available at http://svs.aids.gov.br/download/Vigitel/All these surveys are anonymized and fully opened to public access.
Statistics from Altmetric.com
Introduction
Brazil is usually referred to as a successful case of application of tobacco control policies, having experienced a persistent and substantial reduction in the prevalence of smoking in both total population and distinct cohorts since 2006. According to data from Risk Factor Surveillance and Protection for Chronic Diseases by Telephone Survey (VIGITEL), the percentage of smoking adults in the population decreased from 15.7% in 2006 to 10.1% in 2017. The country’s adoption of the WHO Framework Convention on Tobacco Control Convention and subsequent ratification by the Brazilian National Congress in 2005 coupled with a rigorous tax policy on cigarette production are considered milestones for this sharp decrease in smoking prevalence.
Two major mechanisms were employed to discourage smoking. First, tax policy was used to increase the price of cigarettes, and second, legal restrictions were imposed to make smoking socially discouraged. This strategy was effective because it raised the monetary costs of smoking by raising taxes on the price of cigarettes and increased the inconvenience of smoking by imposing several legal restrictions on smoking behaviour, such as forbidding smoking in public places, restricting cigarette marketing and sales, advertising against smoking and others.
Despite the effectiveness of these tobacco control policies, the socioeconomic costs of smoking are still very high in Brazil. A recent study1 reports that cigarette consumption accounted for about 150 000 deaths in Brazil in 2015. Premature deaths that reduce the working life of smokers and healthcare costs for treatment of tobacco-related illness represent some of the major economic costs of tobacco consumption. The same study estimates total cost to society amounts to be about 1% of gross domestic product.
The objective of this study is to provide a comprehensive cost–benefit analysis resulting from tobacco taxation in Brazil using estimated price elasticities of cigarette consumption by different cohorts of the population. This analysis is essential to understanding the smoking behaviour of individuals and to evaluate the associated economic costs of smoking to the Brazilian society.
There are at least three channels by which a tobacco tax increase could affect social welfare.2 The first channel is that higher cigarette prices due to higher tobacco taxation reduces cigarette consumption and prevents smoking initiation. The second channel comes from the reduction in healthcare expenses associated with the averted treatment costs of tobacco-related diseases and the third one is the increase in income due to gains in years of employment derived from an extension in life expectancy. (One may consider other effects, such as reducing secondhand smoking and reducing adverse birth outcomes for instance. Therefore, the methodological choice might underestimate the benefits of taxation as it will not be able to capture all these channels.) Based on two 2017 and 2018 studies,2 3 we estimate the impact of these channels by estimating price elasticity of tobacco and calculating the welfare gains among various income groups resulting from a tobacco tax increase that raises cigarette prices and lowers tobacco consumption.
We apply a similar methodology to access the welfare gains resulting from a tax increase in manufactured cigarettes for Brazil by using household survey data from 2008 and 2013. We estimate price elasticity of cigarette consumption by gender, income quartiles and age groups. The estimated elasticities are used to implement an extended cost–benefit analysis of increasing tobacco taxes on aggregate welfare by income quartiles and age groups.
The extended cost–benefit analysis considers an increase of 10% in the price of cigarettes resulting from higher taxes on manufactured cigarettes. The analysis uses the estimated price elasticities of demand to simulate the effect of the price increase on tobacco spending, medical expenses and wage income due to the gain in future years of employment. The positive net income effect of the decrease in cigarette consumption due to a 10% increase in cigarette taxes reaches 4.24% and 5.13% in the first and second income quartiles, respectively. These gains in income come from combining the increase in cigarette expenses due to the higher price, reduction in medial expense with the decrease in smoking and gain in future years of working life due to health improvement. These gains decrease as the quartiles of income increase, indicating the progressiveness of this tax policy.
Considering the net income effect by age group, there are similar results. The highest income effects are for the young (15–29 years old) and middle aged (40–59 years old). This is because younger people have lower income and thus a higher benefit from future years of working life by reducing or quitting smoking. Middle-aged people also have high net income effects because smoking-related illnesses and disease usually appear at this age and result in elevated medical expenses for treatment.
The paper is organised as follows. The next section discusses the data set and smoking behaviour of the Brazilian population. The third section describes the methodology used in the price elasticity estimation and cost–benefit analysis. The fourth section reports and discusses the extended cost–benefit analysis resulting from an increase in cigarette taxes. Finally, the fifth section is dedicated to the concluding remarks and policy recommendations.
Data and smoking behaviour
In order to describe the smoking behaviour of the Brazilian population and to estimate the sensitivity of cigarette consumption with respect to cigarette prices, this study combined two exclusive individual surveys: the National Household Sample Survey (PNAD) of 2008 and the National Health Survey (PNS) of 2013. Both are repeated surveys for distinct purposes, but in the two selected years, their questionnaires include a special section on smoking behaviour. We select those relevant items that are identical in both questionnaires to guarantee comparability between the 2 years and uniformity of analysis. The PNAD and the PNS are representative surveys, which are organised by the Brazilian Institute for Geography and Statistics under the same sampling scheme. Both have a household and an individual component. This study focuses on the latter questionnaire because smoking behaviour is essentially individual. The use of the provided sample weights makes the statistics representative of the entire population.4 5 Monetary variables refer to the base year of 2013.
Table 1 reports the proportion of regularly smoking individuals across income quartiles and five age groups in 2008 and 2013. (Note that, according to the nature of our data, income and the income quartiles refer to the total income that the interviewed individual received per month.) The data confirm that, independent of the characteristics, there is an overall tendency to reduce smoking. Moreover, one observes that the propensity of smoking declines monotonically with income and increases with age, except for individuals aged 60 years and above.
Smoking behaviour by cohort and income quartiles 2008 and 2013
It also indicates how many cigarettes the respondent smoke per day and which price the smokers paid for a pack of cigarettes in their last purchase. In contrast to the extensive margin (or smoking prevalence), individuals with high incomes tend to smoke more cigarettes per day and tend to consume more expensive brands. While individuals below the age of 30 years and above 60 years smoke less than 13 cigarettes per day, the number is as high as 16 in group of people aged between 50 and 59 years. Notwithstanding, the intensive margin of smoking (smoking intensity) also clearly is decreasing over time in line with the rising consumer price of cigarettes.
Notice that cigarette prices in the different population groups increased by 50% and more, while the general price level increased by 28.5% as measured by the official consumer price index during the same period. The differences between cigarette prices are reasonable but still lower than those observed in a study on Moldova,6 where the average prices between the first and last income decile differ by a factor of more than two.
Methodology
Price elasticity of cigarette consumption
One key parameter in this cost–benefit analysis is the price elasticity of cigarette consumption. Its value indicates how individuals adjust their consumption to price changes. Specifically, the price elasticity (ε) measures how many percentage points the amount of cigarette consumption will decrease if the final price of cigarettes is increased by 1%. To account for the potentially different effects of a tobacco tax increase over the income distribution, price elasticities are estimated for each income quartile. Therefore, the results can capture whether the tax change is progressive or regressive, that is, whether rich individuals are relatively more or less affected. The conditional price elasticities are derived from the following estimation:
(1)
where Qidt is the number of cigarettes smoked per day by individual i in income quartile d and year t, Pidt is the price that individual i actually paid per cigarette, Idt is a binary variable that indicates to which income quartile the individual belongs and the vector Xidt includes control variables for age, education, years of smoking, income, gender and federal state fixed effects. eidt is the random error term of the regression. In addition to equation (1), we also estimate a modified model where the interaction with income quartiles is substituted by age cohorts.
Following WHO Handbook,7 we estimate the unconditional price elasticity related to the quantity of smokers from the following probit model:
(2)
The dependent variable is an indicator whether individual i is a smoker or not, and the other variables are the same as in equation (1). Combining both equations (1) and (2) to a two-part model gives the overall effect of how a price increase would affect total cigarette consumption. The total price elasticity thus reflects adjustments along two dimensions: (1) the consumption quantity (smoking intensity), that is, the intensive margin; and (2) the smoking prevalence, that is, extensive margin.
Since the product is highly similar but prices between different brands vary quite substantially, consumers may adjust to price changes by switching to a cheaper brand. To deal with a possible endogeneity bias and the problem of misreporting the price of the individuals’ last purchase, leading to measurement error and the well-known attenuation bias in the coefficients towards zero, reported prices are substituted with average prices in each federal state. These average prices are calculated as the average of self-reported prices across smokers within each federal state. Prices differ substantially between federal states because a part of the tobacco tax rate is state specific and because distribution and transport costs to the interior of the country are quite high due to poor infrastructure.8 These average prices can be seen as exogenous to the individual consumer, and this procedure can be understood as an adaptation of the Deaton method to the context of individual-level data.9
Extended cost–benefit analysis
The cost–benefit analysis presented here is based on tobacco literature,2 6 the main hypothesis being that the change in tobacco consumption has a direct impact on household tobacco expenditures and on medical expenditures and productivity costs. The net income effect due to a tobacco price change can be decomposed into three effects: (1) change in tobacco expenditure; (2) change in medical expenses and (3) change in income related to years of productive life lost.
The change in cigarette expenditures (
) is given by
(3)
where 
is the change in cigarette price, 
is the cigarette price elasticity for quartile
q
and 
is the cigarette expenditure proportional to total expenditures (per cent) for quartile q.
The change in medical expenditures (
) can be obtained by
(4)
where 
is the medical treatment expenditures (with tobacco related diseases) to total expenditures ratio for quartile
q
. That is, the weight of medical costs added to the total household expenditures in a given period of time.
Finally, the change in income due to 
− years of productivity life lost − (
) is given by
(5)
where the working years, 
, is the 
distributed across quantile
q
proportionally to the number of smokers 
; 
is the monetary cost of 1 year of life cost, thus 
is the monetary cost of 
and 
is the working years cost to total expenditures ratio for quartile
q
. The rationale of equation (5) is that a reduction in tobacco consumption, due to a price increase, diminishes the years of productive life lost (ie, less premature death and less people living with poor quality of life). Since people potentially will live more years (or live years with a better quality of life), they will also have an income gain.
Results
The overall price elasticities by cohort and income quartiles used to calculate the components of the net income effects are reported in table 2. The estimated elasticity values are highly significant according to the 95% CI, although the statistical significance cannot be confirmed for the differences between cohorts within a given population group.
Smoking overall price elasticities by cohort and income quartile
From table 2, one observes little differences across age cohorts and income quartiles. Specifically, for the majority of Brazilians, a 10% cigarette price increase should reduce the number of smoking individuals by about 2.6% when considering the only the unconditional price elasticity. The largest deviation from this number is observed for the group of people aged 60 years and above. (This increase of the price elasticity can be explained using descriptive statistics. Among this group of individuals, the prevalence of smoking drops to 13%, as compared with 22% among the individuals aged 50–59 years. Dictated by demographics, the oldest cohort also contains a considerably higher share of women who were shown to smoke less and tend to have a more elastic price elasticity.) Consequently, the observed differences in the total price elasticities stem from its second component, the conditional price elasticity. Age is thus apparently a factor that makes smokers less sensitive to price changes while the contrary is observed for income. While this evidence is in line with the findings from Egypt, Bulgaria and Turkey,10 other related studies2 6 from Eastern Europe report that the magnitude of the price elasticity declines with income. However, as in our case, these differences are relatively small and not statistically different across the groups. Our estimates of the price elasticity are in the range of –0.4 to –0.8, which were also found by11 in a variety of studies for developing countries using alternative methodologies.
Tables 3 and 4 assemble the medical costs and deaths of several chronic conditions that are attributable to smoking such as cardiovascular diseases, strokes, pneumonia, cancers and other pulmonary diseases that occurred in Brazil during 1 year. From the number of premature deaths and the average life expectancy, it can be inferred the total number of years of life lost due to smoking.1 In a second step, costs are attributed to the premature deaths and to living with poor quality of life. These costs sum up to BRL 12 308 billion.
Medical costs and deaths attributable to smoking
Productivity related costs attributable to smoking
We consider the effects of a 10% cigarette price increase. This is equivalent to raising PIS/COFINS (a federal consumption tax levied on the company’s turnover) from 11% to approximately 16%, that is, a five basis points increase. This new tax rate would lead to, approximately, a 2.5 basis points increasing in the total cigarette tax burden. The increase in the PIS/COFINS tax rate is relatively easy to implement, and most importantly, the tax revenue is earmarked to finance social programmes and social security expenses.
Table 5 reports the effects of a 10% increase in the price of cigarettes by income quartile, resulting from a tax increase of 5 basis points in the PIS/COFINS, for instance. According to Panel A, the 25% poorest individuals experience the largest increase in cigarette expenditure (0.77%). Following the literature, we can investigate what would happen if individuals had no sensitivity at all to cigarette price changes (smoking price elasticity equal to zero). This complete pass-through case can be considered as a baseline scenario where all individuals experience a net income loss since they have an increase in their cigarette expenses with no other income benefit (in terms of reduction in medical expenses and years of life lost costs, for instance). It is worth highlighting that the poorer would be in disadvantage in this scenario since they face the highest net income loss (1.6%). However, they benefit most from reduction in medical expenses (2.44%) and decrease in years of life lost (2.57%). There is a net income gain of 4.24%. Similar results could be observed for quartile 2, where individuals obtain a net income gain of 5.13% as a result of the 10% increase in cigarette prices. Thus, this is a very effective progressive tax policy as it benefits the poorer most.
Percentage change scenario for a 10% increase on cigarette prices by income quartile
Table 6 illustrates the effects of a 10% increase of cigarette prices by age cohort. It is noticeable that cigarette expenditures increase by 0.48% on average, with small differences across age cohorts. However, individuals aged between 30 and 49 years old enjoy larger benefits in terms of reductions in medical expenses than individuals aged above 50 years. Those between 50 and 60 years old benefit most in terms of reduction in years of life lost cost. This result is a little surprising because this cohort has a smaller probability of reducing, or even quitting smoking and the advanced age mechanically reduces the years left in life. However, the high average income dominates and explains the observed outcome. The younger group experiences the largest reduction in medical expenses and productivity costs. However, the probability they suffer from any tobacco-related disease described in table 6 is quite small due to their lower age.
Percentage change scenario for a 10% increase on cigarettes price by age cohort
Table 7 reports the changes in death rates for quartiles of income and age cohorts resulting from the same previous 10% increase on cigarette prices. It can be observed that individuals between 40 and 60 years old benefit the most. The reduction is between 10 and 15 deaths per 10 000 people in these groups. For the sake of comparison, the overall death rate in Brazil is stable: around 6.4 deaths per 1000 people for the last 5 years. The reduction in deaths due to cigarette price increases represents a decrease between 15% and 23% of the overall death rate in the country. This is a significant result since Pinto et al 1 estimates that the deaths attributable to smoking represent nearly 28% of the total number of deaths.
Change in death due to a 10% increase on cigarette prices by cohort and income quartile (per 10 000 people)
Conclusion and policy recommendations
The results of this research suggest that a tobacco tax increase would benefit the entire Brazilian population. The positive effects emerge because higher tobacco taxes reduce cigarette consumption and lead to lower medical costs and longer, healthier and more productive lives, which ultimately results in higher incomes for everyone.
The extended cost–benefit analysis considered a 10% price increase resulting from a raise in taxes on manufactured cigarettes. The average net income effects were 4.24% and 5.13% increases in the first and second quartiles, respectively. These gains in income result from combining the effects of higher cigarette expenses due to the higher price, reduction in medical expenses with the decrease in smoking and gain in future years of employment due to the health improvement. More importantly, the effects of such a policy are progressive in the sense that they are more beneficial for those individuals below the median income. Thus, the gains increase as the income quartile decreases.
The findings are similar by age cohorts. The highest net income effects were observed for the youngest, aged between 15 tand 29 years, and for middle aged people, between 40 and 59 years old. This is the case because the youngest have low income and thus a higher benefit on future years of employment by reducing or quitting smoking. In case of middle-aged people, the high income is driving the effect.
We recommended a raise in PIS/COFINS be used to generate the 10% increase on manufactured cigarettes price used in the simulation exercises. The advantages are that the tax change could be more easily implemented by the government because the additional tax revenue is earmarked for social expenses. For example, a 10% price growth is roughly reached by a tax increase of about five basis points in the PIS/COFINS tax rate. Tobacco tax increases have an important advantage, meaning that the poorest individuals represented by the first quartile of income experience the largest increase in cigarette expenditures and biggest reductions in medical expenses. Additionally, those in the second quartile of income shows the biggest reduction in the years of life lost. As a result, both of them (individuals with income below the median) have the largest gain in net income.
To broaden the social reach of the tobacco control policy, it is advisable that the tax increase policy on manufactured cigarettes be accompanied by other measures that increase the economic and social costs of smoking and curb illicit trade through better policy coordination, higher levels of monitoring and penalties. A good example comes from the state of São Paulo, which has the highest tax burden on cigarette prices in the country and recently prohibited smoking in all municipal parks across the city. Coupled with the highest cigarette tax burden, this measure adds to several others adopted by the state focused on raising the social inconvenience of smoking by imposing several legal restrictions on the smoking behaviour.
Brazil has reached undeniable progress in reducing smoking prevalence trough an adequate combination of public policies. Our key message is that the country can go one step further on the tax increase policy. The present study clearly indicates that such a tax increase not only reduces cigarette consumption but, most surprisingly, higher cigarette prices even imply an increase in net income for smokers.
There are some caveats to the analysis. This is a partial equilibrium analysis focused on the legal market of cigarettes, which does not account for second order effects of price adjustments that would appear in a general equilibrium framework. The medical costs and years of life lost (YLL) are not specific by income quartiles or age cohorts due to the lack of data availability. Due to data limitations, we used average measures of medical costs and productivity. Nonetheless, improved and more disaggregated data are necessary to assess the heterogeneities in medical expenses and YLL across income quartiles and age cohorts, which may influence the magnitudes and distribution of net income effects. Similarly, by using state average prices, we are able to avoid an endogeneity bias but are likely to miss some heterogeneity across consumers. The cost–benefit analysis refers to the net income variation and does not represent a welfare analysis. Consequently, other sources of costs and benefits, including forgone (dis)utility of smokers, are not considered because of the difficulty to measure such effects. Spillover and peer effects from less smoking are also not accounted for. Some of these issues, however, are being addressed in our ongoing research.
What this paper adds
What is already known on this subject
About 150 000 Brazilians die from smoking related diseases every year. Smokers and their families also suffer from forgone income and worse living conditions. Medical costs from smoking amount to approximately BRL 38 billion per year. Progress has been made in recent years to reduce smoking prevalence, through legal restrictions on smoking behaviour and tax policy measures. However, more needs to be done to reduce tobacco consumption and improve Brazilians’ quality of life.
What this paper adds
The PIS/COFINS (a social contribution levied on cigarettes) should be used to generate the 10% increase in manufactured cigarettes price. The advantages are that the tax change could be easily implemented by the government, and the additional tax revenue is earmarked to social expenses.
We found that tobacco tax increases is a very effective progressive policy as it benefits the poorer the most. Each 10% price increase (BRL 0.54), due to higher tobacco taxes, reduces cigarette consumption by about 5%, and for poor smokers, it would lead to net income gains by about BRL 39.00 per month (in 2019 values).
We evaluated the impact of price increases as a result of higher taxes on cigarettes. Higher tobacco taxes lead to lower medical expenses on tobacco-related diseases and a longer, healthier and more productive life.
Data availability statement
Data are available in a public, open access repository. The public data used in the paper are: (1) National Household Sample Survey (PNAD) de 2008, available at https://www.ibge.gov.br/estatisticas/sociais/educacao/9127-pesquisa-nacional-por-amostra-de-domicilios.html?edicao=9128&t=downloads (2) National Health Survey (PNS) de 2013, available at https://www.ibge.gov.br/estatisticas/sociais/saude/9160-pesquisa-nacional-de-saude.html?=&t=downloads (3) Risk Factor Surveillance and Protection for Chronic Diseases by Telephone Survey (VIGITEL), which is performed annually, available at http://svs.aids.gov.br/download/Vigitel/All these surveys are anonymized and fully opened to public access.
Ethics statements
Patient consent for publication
Acknowledgments
We are grateful to Frank Chaloupka, German Rodriguez Iglesias, Erika Siu, Alan Fuchs and seminar participants at the 2019-UIC and LAC partners meeting in Mexico City for their comments and suggestions. We would also like to thank two anonymous referees and the editor for valuable comments and suggestions.
Footnotes
Contributors All coauthors contributed to the research and paper drafting.
Funding This research (grant number 17409) is funded by the University of Illinois at Chicago’s (UIC) Institute for Health Research and Policy to conduct economic research on tobacco taxation in Brazil. UIC is a partner of the Bloomberg Initiative to Reduce Tobacco Use. The views expressed in this document cannot be attributed to, nor do they represent, the views of UIC, the Institute for Health Research and Policy, or Bloomberg Philanthropies.
Disclaimer The views expressed in this document cannot be attributed to, nor do they represent, the views of UIC, the Institute for Health Research and Policy, or Bloomberg Philanthropies.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.