Research study presentation – University of Banja Luka, BiH Research project: Accelerating progress on tobacco taxES in low-and middle income countries BELGRADE, 10 – 11 December 2018

RESEARCH TEAM INSTITUTION: - University of Banja Luka - Entrepreneurship and Technology Transfer Centre - Center for Project Management and Entrepreneurship, Faculty of Economics TOBACCO TAXATION RESEARCH TEAM: Anđela Pepić, PhD candidate, Country Project Manager Saša Petković, PhD, Research Team Manager Jovo Ateljević, PhD, Senior Researcher Dragan Gligorić, PhD, Researcher Borislav Vukojević, PhD candidate, Researcher

Introduction The aim of this study is to analyze the effects of tobacco taxation and provide research-based evidence for more effective tobacco taxation policies In this report, we describe and analytically ‘consolidate’ the tobacco environment in BiH, covering the most important areas related to tobacco demand and supply, followed by the tobacco use and its health and economic consequences The report provides a brief discussion on tobacco control policies in BiH followed by a more detailed discussion of tobacco taxes, prices and smoker culture and behavior. Based on data on changes in excise taxes, consumption of tobacco products, and the micro-data from the Household Consumption Survey, the UNIBL study confirms that a significant increase in excise taxes can provide additional benefits for society, through a decrease in consumption of tobacco products and faster increases in state excise tax revenues.

Country profile Source: Agency for Statistics BiH (2018) General info Data Note Region Bosnia and Herzegovina RS, FBiH, BD Income Category (2018) upper middle-income country   Population (proj., 000) (year) 3,502.107 Projection Surface area (sq km) (year) 51,209 Capital city (year) Sarajevo Macroeconomic indicators Gross domestic product, in millions EUR (2017) 23,088 current international Gross domestic product, per capita, EUR (2017) 5,989 Gross domestic product, real growth, in % (2017) 4.6 Estimated Inflation (2018) 1.6% Unemployment according to the Survey (in %) (2017) 19.4 Wages (average for the period, in EUR) (June 2018) 450.31 Employment level, average (thousands) (April 30th, 2018) 1,252,860 Declining Source: Agency for Statistics BiH (2018)

Tobacco use trends in Bosnia and Herzegovina Since 2006, consumption of tobacco products has had a tendency to decline steadily, yet it is high among youth as 12% of 15 to 18 years of age are estimated to be daily smokers. The proportion of smokers in Bosnia and Herzegovina is still considerably higher than in most European Union (EU) countries. The study showed that in all of the households surveyed, approximately 41% had an individual that smoked cigarettes. On average, Bosnia and Herzegovina households spent an estimated 5.4% of the monthly budget on cigarettes.

Graph 1. Prevalence of tobacco consumption in Bosnia and Herzegovina, 2000 to 2016 (Source: The World Bank, 2018)

Graph 2. Movement of prices and consumption of cigarettes in Bosnia and Herzegovina - semi-annual data (2008-2017) (Source: Author calculations based on data from ITA, 2018)

Tobacco tax structure The trend of decreasing consumption could be attributed to the harmonization of the regulations of tobacco prices and tobacco products with the regulation of the EU Since prices of tobacco products in the EU are considerably higher when compared to the prices in Bosnia and Herzegovina, there is significant space for tax and price increases. Tax policy of tobacco products is under the responsibility of the Indirect Taxation Authority of Bosnia and Herzegovina The analysis of the institutional framework suggests that the Indirect Taxation Authority of Bosnia and Herzegovina is the only excise policymaker, while the Ministry of Health is completely excluded from this significant process and does not have any ability to influence the creation of an excise policy as a means of controlling tobacco.

The taxation of tobacco products is largely a consequence of the decision of the Board of Directors of the Indirect Taxation Authority of Bosnia and Herzegovina, which is particularly focused on achieving stable and predictable excise revenues. Current excise policy relies on ad valorem excise taxes accounting for 42% of the retail price. The law also includes a gradual increase in specific excise duties (see Graph 3), currently accounting for one third of the most popular brand price.

Graph 3. Structure of tobacco tax in the period 2010-2017 (Source: Authors’ calculations based on data from the Official statistics)

The process of adjusting to the EU regulations should be followed by a significant tax increase in order to meet the minimum prescribed excise duty of € 90 per 1000 cigarettes. The retail price per 1000 cigarettes of the most popular brand in Bosnia and Herzegovina is € 77.5. The share of excise taxes in the retail price is 67.95% in Bosnia and Herzegovina, which is almost in line with the best practices recommended by the World Health Organization. However this includes a significant amount of ad valorem taxes.

Interesting remarks based on qualitative study (6 focus groups) Citizens are aware of the negative effects of consuming tobacco products, but this is not sufficient for smokers to stop smoking; The most common reasons for consuming tobacco products are perceived satisfaction, habits and social interaction with others; The prohibition of smoking in public places is supported by arguments that there are countries where it is forbidden, and people still live without problems - it's all a matter of habit, it takes time to adjust; Increasing excise taxes on tobacco products is justified by the character of cigarettes; this is not a basic human need, and therefore the consequences must be dealt with; Arguments against increasing excise taxes on tobacco products are the following: (1) lack of transparency in tax administration; (2) beliefs that price increases do not reduce tobacco consumption

Elasticity of demand for cigarettes In order to estimate elasticity of demand cigarettes in BiH we develop the following econometics models: econometrics model of cigarettes demand based on macro level data, for the period 2008-2017, on semi-annual data – „macro model” econometrics model of cigarettes demand based on micro level data (Deaton model), using HBS for 2011 and 2015 – „micro model”

Macro model Data and Methods The data related to cigarettes consumption and prices in BiH are available for the very short period (2008-2017) To have sufficient data for implementing time-series analysis, we use semi-annual data fom 2008p1 to 2017p2 We estimatd the following model const = α + β0 rtcpit + β1 rincomt + β2 tcontrolt + εt , where const - consumprion of cigarettes per capita (per adult) rtcpit - real tobacco CPI index rincomt – Real income, (proxied by three different indicators – real GDP pc, real disposable income and real wages) tcontrolt – we use three different indicators related to the tobacco control envirnoment tlow – dummy variable which represent introduction of Law on tobacco in 2010 aban – dummy variable which represent the adoption of Comercial code of Comunication in December in 2015 tcindex – sum of two dichotomus indicators tlaw and aban Inability of official population statistic in BiH prevented from using othes control variables such as m/f ratio, education...

Macro model Data and Methods We use all variables in logarithm, exept, of course tobacco control variables which is dummy variables We detect that three variables, lrtcpi, lrincome and lrdincome have seasionality, therefore, we use Tramo/Seats statistical tools to eliminate seasionality from this variables. Dickey-Fuller unit root test indicates that only real income and real disposable income have unit roots in levels, but not in the first differences According to the results of unit root test, we can use the following econometric methods: OLS, but if variables is I(1), such is Real GDP pc and Real GNI pc, we use them in first differences - this is correct because dlrincom and dlrdicom represent real growth rates We can use ARDL (risk for reliability of results is very short time series) We opt for OLS and for use ARDL as the second robustness check

Macro model Results (with real GDP as a proxy for income)   (1) (2) (3) (4) VARIABLES lcons lrtcpi -0.827*** -0.778*** -0.839*** -0.790*** [0.077] [0.133] [0.101] [0.097] dlrincome 0.250 0.342 0.189 0.215 [1.404] [1.454] [1.479] [1.431] tcindex -0.029 [0.064] tlaw 0.015 [0.076] aban -0.043 [0.066] Constant 11.355*** 11.131*** 11.408*** 11.173*** [0.392] [0.635] [0.482] [0.487] Observations 19 R-squared 0.913 0.914 0.916 F statistic (prob.) 0.000 D-W d-statistic 2.061 1.928 2.120 2.002 Breusch-Pagan / Cook-Weis. test for heter. (Prob > chi2) 0.809 0.832 0.763 0.769 JB test (Prob.) 0.890 0.921 0.876 Ramsey RESET test (Prob.) 0.721 0.741 0.823 Multicolinearity test (vif) 1.50 3.28 2.06 1.83

Macro model Results (with real disposable as a proxy for income)   (1) (2) (3) (4) VARIABLES lcons lrtcpi -0.776*** -0.756*** -0.799*** -0.736*** [0.094] [0.133] [0.103] [0.112] dlrdincome -1.200 -1.010 -1.908 -1.279 [2.001] [2.223] [2.350] [2.037] tcindex -0.016 [0.068] tlaw 0.051 [0.083] aban -0.046 [0.065] Constant 11.106*** 11.012*** 11.189*** 10.905*** [0.478] [0.641] [0.506] [0.564] Observations 19 R-squared 0.915 0.917 0.918 F statistic (prob.) 0.000 D-W d-statistic 2.039 1.970 2.244 1.993 Breusch-Pagan / Cook-Weis. test for heter. (Prob > chi2) 0.745 0.767 0.615 0.757 JB test (Prob.) 0.868 0.888 0.761 0.934 Ramsey RESET test (Prob.) 0.458 0.477 0.528 0.571 Multicolinearity test (vif) 2.27 3.86 2.82 2.37

Macro model Results (with real wages as a proxy for income)   (1) (2) (3) (4) VARIABLES lcons lrtcpi -0.786*** -0.730*** -0.846*** -0.747*** [0.076] [0.138] [0.127] [0.083] lrwage 0.194 0.207 0.450 0.637 [1.012] [1.036] [1.121] [1.086] tcindex -0.032 [0.066] tlaw 0.048 [0.082] aban -0.077 [0.071] Constant 9.980 9.642 8.714 7.137 [5.849] [6.024] [6.344] [6.376] Observations 20 R-squared 0.906 0.907 0.908 0.912 F statistic (prob.) 0.000 D-W d-statistic 1.985 1.858 2.171 1.947 Breusch-Pagan / Cook-Weis. test for heter. (Prob > chi2 0.675 0.694 0.619 0.746 JB test (Prob.) 0.786 0.826 0.7692 0.939 Ramsey RESET test (Prob.) 0.455 0.444 0.486 0.644 Multicolinearity test (vif) 1.54 3.66 2.98 1.88

Macro model Results The only variable, which has significant impact on demand for cigarettes, was real tobacco CPI. Value of estimated coefficients was in the range of -0.73 to -0.85, and the estimated coefficient was stable across different models and different versions of a particular model. This implies that an increase in prices of cigarettes for 1% led to a decrease in the consumption of cigarettes in the range of 0.73% to 0.85%. ARDL model also confirms statistically significance of the increase of cigarettes prices on the decrease ion cigarettes consumption.

Micro model Data and methods Due to the very short time series data on macro level, micro data from Household Budget Surveys (HBS) is more reliable source for estimation of prices elasticity Methodology, which use micro data from HBS in estimating price elasticity, has developed by Angus Deaton We follow Deaton demand model and apply it on micro data, obtained from the Household Budget Surveys in BiH in 2011 and 2015 HBS in 2011 and 2015 contains 7,048 and 7,250 households respectively. In 2015 compared to 2011 number of smoking households decrease from 48.4% to 33.8%. Unit value of cigarettes, which represents not only thr changes in prices of cigarettes increased from 2.37BAM to 3.65BAM (54%).

Micro model Data and methods Share in expenditures on cigarettes in total expenditures has increased from 5.17% in 2011 to 5.69% in 2015 According to the criteria requested to apply Deaton model, we drop the households which have not expense on cigarettes, households which monthly expenditure is more than five standard deviation higher than the mean monthly expenditure of the sample, and clusters which have only one household So, due to the Deaton model estimation procedure, we reduced the sample and it contained 253 clusters and 5,820 households. The data indicate that about 41% of the households have expenses on cigarettes

Micro model First-stage regression results VARIABLES Unit Value (per pack, ln) Cigarettes budget share (in %) Total expenditure (ln) 0.059*** (0.003) -0.024*** (0.001) Household size (ln) -0.040*** (0.004) -0.005*** Male ratio -0.008 (0.007) 0.019*** (0.002) Adult ratio -0.033*** (0.009) 0.009*** Mean education 0.003*** -0.000 (0.000) Maximum education -0.001*** Household type - Employed Omitted Unemployed -0.023*** (0.008) 0.003 Pensioners (0.006) -0.002 Self-employed -0.006 (0.005) Cluster dummies F(252, 5558)   71.160*** 3.235*** Constant 0.626*** (0.025) 0.230*** Observations 5820 R-squared 0.811 0.282

Micro model Results Results indicate a negative price elasticity of -0.649. If cigarette prices in BIH increase by 10%, the demand for cigarettes, among smokers of cigarettes, will decrease by 6.5%. Standard error of the elasticity, calculated via bootstrapping procedure indicates that the value of the price elasticity is significantly different from and lower than zero (ξ= -0.649; SEξ = 0.092, t = -7.053). The estimated value of total expenditure elasticity is, in line with the expectations positive and estimate at 0.496. In other words, among the households which consume cigarettes, 10% higher total expenditure is associated with 4.96% higher the quantity of cigarettes smoked.

Conclusions Tobacco tax policy in Bosnia and Herzegovina currently neglects public health protection and the reduction of tobacco product consumption. The specificity of Bosnia and Herzegovina and their complex political systems strictly separate competencies. The Indirect Taxation Authority is operating at the state level, while health policies are in the hands of the entities. Tobacco taxation policy has not resulted in a significant decrease in tobacco consumption in Bosnia and Herzegovina, as it is still at a very high level and has resulted in numerous negative economic and health consequences for society. Although prices of tobacco products are increasing slowly, Bosnia and Herzegovina have a lot of room to catch up with the higher prices of tobacco products in the EU.

Measure 1. Adopt an evidence-based system of tobacco taxation. The results of the study show that a 50% increase in price as a result of a significant increase of excises, along with other unchanged factors, would lead to a decrease in quantity demanded for tobacco products between 32.5% and 41.5%. The current approach relying on ad-valorem excise should be reconsidered as it enables strong variation of the retail prices depending on the specific brand of cigarettes. Greater reliance on specific tax than on ad valorem, is needed.

Measure 2. Create a strategic framework for tobacco control that focuses on necessary adjustments needed to align with EU regulations in tobacco taxation It is necessary to increase the excise taxes on tobacco products reaching the level of EU regulations suggesting the overall excise to reach € 90 per 1000 cigarettes. The EU-28 weighted average price of cigarettes is €4.80, which is much higher than the €2.66 price of the most popular brand of cigarettes in Bosnia and Herzegovina in 2017. The creation and implementation of the new strategic framework must take into account the principles of participation and coordination of all stakeholders (Indirect Taxation Authority, Ministry of Health, and Civil Sector) taking into account the overall interests of society.

Thank you for your attention! For further information regarding BiH research study, please contact as via following adresses: Saša Petković (sasa.petkovic@ef.unibl.org) Borislav Vukojević (borislav.vukojevic@fpn.unibl.org) Dragan Gligorić (dragan.gligoric@ef.unibl.org)