Article Text
Abstract
Background The current tobacco taxation scheme in Turkey, a mix of high ad valorem tax and low specific tax, contains incentives for firms and consumers to change pricing and consumption patterns, respectively. The association between tax structure and price and tax revenue stability has not been studied in detail with micro data containing price segment information.
Objectives In this study, we analyse whether incentives for firms and consumers undermine the effectiveness of tax policy in reducing consumption.
Methods We calculate alternative taxation scheme outcomes using differing ad valorem and specific tax rates through simulation analysis. We also estimate price elasticity of demand using detailed price and volume statistics between segments via regression analysis.
Findings A very high ad valorem rate provides strong incentives to firms to reduce prices. Therefore, this sort of tax strategy may induce even more consumption despite its initial aim of discouraging consumption. While higher prices dramatically reduce consumption of economy and medium price segment cigarettes, demand for premium segment cigarettes is found to be highly price-inelastic.
Conclusions The current tax scheme, based on both ad valorem and specific components, introduces various incentives to firms as well as to consumers which reduce the effectiveness of the tax policy. Therefore, on the basis of our theoretical predictions, an appropriate tax scheme should involve a balanced combination of ad valorem and specific rates, away from extreme (ad valorem or specific dominant) cases to enhance the effectiveness of tax policy for curbing consumption.
- Taxation
- Tobacco industry
- Economics
- Low/Middle income country
- Price
Statistics from Altmetric.com
Introduction
Tobacco consumption is a global threat to public health. Strict regulation of sales and advertising, promotion of public awareness campaigns and protection from secondhand smoke are among the actively used non-price methods. Yet no policy seems more effective than taxation as rational consumers respond to monetary incentives.1 Still, consumers may sometimes act partially rationally, given that the addictive nature of tobacco consumption and the effectiveness of monetary incentives may not be the same for different groups. This responsiveness may also change with income level. Studies from developed countries reveal that the effectiveness of taxation on smoking prevalence may be limited.2–4 Tax increases are effective only if they lead to price increases.5 Further, they are even more effective when combined with non-price policies.6 The negative correlation between price and prevalence is more pronounced for low-income and middle-income countries.7
What have been termed price minimisation strategies also reduce the effectiveness of taxation. Consumers would react more aggressively when faced with price hikes if cheaper cigarette options were not available.8 Product switching is also pursued by consumers.2 Such motivations may also be fuelled by pricing strategies of the tobacco industry and firm behaviour may act to undermine the effectiveness of tax policy.9–11 In addition, illicit trading of tobacco products undermines tax policy and brings about health-related concerns. Such effects are more detrimental for low-income and middle-income countries.12
Determination of the tobacco taxation structure is an important public policy issue. Simpler tax structures exhibit less price variability and reduce opportunities for tax avoidance.13 More complicated tax structures limit the effects of increased taxes on reducing smoking in low-income and middle-income countries.14 On the other hand, the choice of a simpler tax structure is not straightforward as pricing strategies of the tobacco industry through price segments play a key role.9 Assessment of the effectiveness of tax structures and improved understanding of the political economy of the tobacco tax policy have been highlighted as key areas for future research.15 Building on this, our study contributes to the literature by focusing on the political economy of taxation by studying the incentives of firms and consumers and by analysing how these incentives affect tobacco control.
To analyse the tobacco taxation policy, we consider Turkey, where the prevalence of smoking is considerably high. Although aggressive taxation is in parallel with global practice, what makes Turkey special is the considerable reliance of public finance on tobacco tax revenue (accounting for 8% of total taxes in 2014). Tobacco taxation is composed of two interconnected indirect taxes: Value Added Tax (VAT) and Special Consumption Tax (SCT; ad valorem and specific). Furthermore, SCT is calculated from the final consumption price, which differentiates tobacco products from other products subject to SCT in the Turkish tax system.16
The current tax system provides incentives to firms to change prices depending on the combination of ruling ad valorem and specific rates. Consumers also respond to price incentives as they tend to switch between different segments of cigarettes when relative prices change. Hence, firms' pricing strategy and consumer behaviour may substantially trigger opportunities for tax avoidance and affect the power of taxation in tobacco control.
While many parties are concerned with tobacco tax policy, the tax authority determines the rates. When deciding on the rates, the tax authority aims to secure a certain amount of revenue. The authority is also interested in the stability of the revenue and wants to benefit from price increases in the tobacco industry. These objectives introduce a trade-off for the authority. For instance, setting an all-specific SCT ensures the stability of revenue, but the authority cannot benefit from firms' price increases as firms have a strong incentive to do so in this case. Meanwhile, an all-ad valorem SCT guarantees that the tax authority fully benefits from price increases, but the revenues may not be stable (and perhaps decrease) as firms have a strong incentive to reduce prices in order to increase market share in this case.
In this background, our findings first suggest that a balanced SCT framework making use of both specific and ad valorem SCT at reasonable rates would ensure that the tax authority minimises firms' price increase and price decrease incentives and thus fulfils the above-cited objectives. Second, given that the demand for high-price cigarettes is highly inelastic, the tax policy per se may not be enough to discourage a certain proportion of the population from smoking.
Background
Consumption and governmental control of tobacco in Turkey
The prevalence of tobacco consumption has been high in Turkey.17 Until the 1980s, the tobacco market was controlled by a state-owned monopoly (TEKEL). Along with privatisation and entrance of the multinational firms in the market, an era of intense advertising and promotion increased consumption significantly, which necessitated control measures to be taken.18
In 1996, with the introduction of the Tobacco Control Law, the upward trend in consumption had eased. Smoking prevalence in Turkey, which was 44% in 1988, dropped to 31.2% in 2008 and to 27.1% in 2012.19 This was mainly accomplished by effective tax and price increases.17 The effectiveness of the tax policy was further strengthened with the implementation of comprehensive tobacco control policies including prohibition of indoor smoking, public education and pictorial warning labels on packs. The prohibition of indoor smoking has already had immediate positive health effects.20
Current tobacco tax scheme in Turkey
Starting from late 2002, SCT has been levied on tobacco products in Turkey in addition to the VAT. From 2002 to 2005, the SCT framework included ad valorem rates, but the dominant part of the SCT was the specific amount, which was determined either on the basis of the price range or on the basis of the share of oriental tobacco in a given cigarette. In this period, the main concern of the tax authority was the loss of potential tax revenue due to the dominance of the specific amount and the price increase tendency of firms. Therefore, as of August 2005, the tax authority switched to an ad valorem-dominant strategy, setting the ad valorem rate at 58% (on condition that it is not less than the minimum specific amount) with the specific rate abolished. The main issue with this strategy is that it provides price decrease incentives to firms which threaten the level and the stability of tax revenue. From late 2005 to 2013, this strategy continued with ad valorem rate adjustments and the ad valorem rate peaked at 69% in October 2011. After 2013, specific SCT was reintroduced, albeit at a very low amount of 0.09 Turkish Lira (TL) per pack (same for all cigarettes). Finally, the total SCT paid per pack is the sum of ad valorem and specific amounts. Valid as of 1 January 2016, the current ad valorem rate is 65.25%, while the specific amount is 0.2468 TL per pack. Using the information mandated in the law, table 1 shows price formation for cigarettes along with an illustrative example.16 ,21 ,22
Here, vat stands for the VAT rate (%), M stands for the specific SCT amount, sct denotes the ad valorem SCT rate and p is the distributor's share (%). As seen from equation (1), the effect of changes in specific SCT (M) is linearly correlated with the final consumer price. On the other hand, changes in the ad valorem SCT (sct) non-linearly affect the final consumer price.
Data
In the study, monthly price and volume information on all cigarette products (around 150–250 each month) in the 2006–2014 period was obtained from the Turkish Tobacco and Alcohol Market Regulation Board (TAMRB). The product segment information on all the brands is supplied by TAMRB as well. That is, the cigarettes are grouped into three categories by price: economy, medium and premium. The price range in each category is adjusted over time to cover the same brands within each category, unless a change in quality is observed. For example, in 2012, following a significant price increase across all categories, the lower bound of the premium segment increased from 5.5 TL to 7 TL. The evolution of the sales volume and sales share of each category across years is given in figure 1. While the share of the premium segment follows a smooth pattern over the years, dramatic changes are witnessed in the share of the economy and medium segment cigarettes.
The evolution of average prices by segment and price gaps between segments are shown in figure 2. Overall, prices in each segment have increased considerably over the sample period, almost at the same pace. However, the pace of the economy segment has slowed down compared with others in the second half of the sample. In the early (late) sample, we observe episodes of price increases (decreases) despite tax rates being unchanged. Meanwhile, the price gap between segments followed different paths. While the gap between the premium segment and others has steadily increased, the gap between the medium and economy segments has widened after 2012. Overall, we may conclude that the premium segment is certainly a separate category and the transition between the medium and economy segments is mostly attributed to lower prices and slower price changes.
The contribution of each segment to taxes also differs. For instance, as of 2016, while each segment brings in 0.2468 TL per pack as specific SCT, the ad valorem SCT received from each segment is on average 4.8, 5.6 and 7.4 TL for the economy, medium and premium segments, respectively.
Analysis
Analysis of firm strategy under the current tax scheme
Tax structure drives the price formulation in the cigarette market, but the final consumer prices are set by the firms. Thus, understanding the incentives that the tax structure provides to firms is important. For this, we rely on simulation analysis. Simulations are generated for discrete increments in ad valorem SCT (from 0% to 100%, with increments of 0.5%) and in specific SCT (from 0 to 4 TL, with increments of 0.02 TL). So, for 200 distinct ad valorem rates and 200 distinct specific taxes, we simulate average price and average tax revenue per pack for 40 000 different SCT combinations under the assumption of constant producer prices, that is, tobacco manufacturers do not adjust their prices in order to increase profitability in response to a tax change (which may not be the case in the real world) using equation (1). In simulation analysis, we assume that minimum excise is not binding for cigarettes, since as of end-2013, minimum excise covered <10% of the market.
Firms' incentive for price increase
Consider a tax scheme which induces final consumer prices in a way that the price gap between different segments is low. Such an outcome is possible if the specific component of SCT is dominant, where the same amount of tax is collected from each price segment. Such an environment provides incentives to firms to increase prices in the premium and medium segments. Since the price gap between segments is low, consumers will be willing to switch to a higher segment if the extra utility received from consuming a higher segment product is more than the price differential between segments.
In this framework, the ‘price gap between the premium and economy segments’ will be a good measure to look at. We may track this price gap for various specific and ad valorem SCT combinations that would give the same average tax revenue per pack.i For illustration purposes, we consider all combinations that would provide 4 TL of average tax revenue (sum of VAT and SCT) per pack among 40 000 simulated observations.
Figure 3A shows all combinations of SCT that yield 4 TL tax revenue (sum of VAT and SCT) per pack on average. Figure 3B shows the premium economy segment price gap that emerges from the corresponding tax combinations in panel (A). The figure shows that 0% ad valorem combined with 3.18 TL-specific SCT (leftmost dot) and 62% ad valorem combined with 0 TL-specific SCT (rightmost dot) both generate the same tax revenue. However, the price gap ranges between 0.6 TL and 3.2 TL.
For the sake of our argument, combinations with higher (lower) ad valorem (specific) tax induce larger price gaps and thus reduce the incentives for price increase in higher price segments. To raise the same revenue from a pack of cigarettes, the tax authority faces several alternatives, from all specific to all ad valorem to combinations in between. However, as figure 3 suggests, the corresponding gap between the premium and economy segments is minimised (ie, where price increase incentives of firms are maximum) with a purely specific tax and maximised with a purely ad valorem tax. This has consequences for what sort of incentives smokers face, and whether they substitute between the segments, which in turn influence the total tax revenue.
Firms' incentive for price decrease
Consider a tax scheme where the sensitivity of final consumer price to changes in producer price is high. Such an outcome is possible if the ad valorem component of SCT is dominant. This environment provides incentives to firms to decrease prices in any segment in order to capture a higher market share. First, firms can reduce the sales price substantially by reducing the producer price marginally. For instance, since the current multiplier (the coefficient affecting X in equation (1)) is about 8, a firm can bring down the sales price by 0.8 TL, foregoing its revenue per pack by only 0.1 TL. Second, if the market share of the firm increases sufficiently, the firm may enjoy a higher profit regardless of the declining sales price. Meanwhile, tax revenues also decline.
In this framework, the ‘elasticity of sales price to producer prices’ will be a good measure to look at. Figure 3A shows all the combinations of SCT that yield 4 TL tax revenue per pack on average under current producer prices. While figure 3C shows the per cent reduction in sales price given a 10% cut in producer prices under the corresponding tax combinations in panel (A), which ranges between 0.5% and 10%.
For the sake of our argument, tax combinations with a lower (higher) ad valorem rate (specific) induce a smaller elasticity and thus reduce the incentives for price decreases.
Implications of firm strategy
The Turkish case provides evidence for both price increase and decrease incentive episodes as the tax scheme moved between two extreme cases over the past decade.
From January to July 2005, the specific SCT was around 1 TL while the ad valorem SCT rate was 28%; thus, the specific SCT was dominant (accounted for around 70% of SCT). In fact, in that period, the average price in the medium and premium segments increased by 7% and 6%, respectively, while tax rates remained unchanged. This meant that the tax authority did not benefit from price increases in terms of an increase in tax revenue. Considering this, the tax scheme has been moved to the other extreme where ad valorem SCT is dominant (accounts for around 95% of SCT).The tax authority guarantees that any increase in consumer prices will be reflected in tax revenue as well. However, given the non-linear relation between the ad valorem rate and price, firms are given an incentive to enter into price cut wars, as it may be profitable for them. Indeed, for instance, over the January to December 2013 period, the ad valorem rate was 65.25% while the specific amount was 0.09 TL. In that period, average prices came down by 8% and 4%, respectively, in the economy and medium segments despite any change in the tax rates. Such an outcome leads to a reduction in tax revenue and it also undermines the effectiveness of tax policy in reducing consumption.
Overall, price change incentives in any direction may be harmful both for the level and the stability of tax revenue. Hence, the prediction of firm strategy analysis suggests that eliminating both incentives will be appropriate in this sense. As discussed, eliminating both incentives requires opposite actions. The tax combination minimising one incentive maximises the other. Therefore, the tax authority should assign probabilities to both incentives when deciding on the tax combination. Given the Turkish experience, we argue that both incentives are important. Therefore, an appropriate tax scheme should involve a balanced combination of ad valorem and specific SCT, away from corner solutions to enhance the effectiveness of the tax policy for curbing consumption.
Analysis of consumer choice under the current tax scheme
The general tendency of consumers is to consume tobacco products in the highest price segment possible. However, if prices increase, consumers may choose to shift their consumption across price segments. Likewise, if a product of the upper segment sees a price cut, consumers are willing to switch to that relatively cheaper product. A survey study among a sample of university students in Turkey verifies this assertion as a considerable number of participants consuming a medium price segment cigarette previously switched to the economy segment following a tax rate hike.23
As seen in figure 1, the share of the premium segment follows a relatively stable course over the sample period. On the other hand, the transition between the medium and economy segments is much more pronounced. The gap between the two segments, especially after 2009, increased dramatically. The main reason for such a shift is the rapid increase in prices due to aggressive tax adjustment backed by public health and fiscal concerns.
Regarding consumer behaviour, the price elasticity of demand for tobacco products deserves further discussion. The main argument for increasing tobacco tax is that a given price increase will reduce overall consumption. In fact, this has been the case for Turkey as well, where the amount of cigarettes consumed dropped in 2000s.17 Using the 1994 Household Expenditure Survey, the price elasticity is estimated as −0.41.24 In a replication of the micro study with the 2003 Household Expenditure Survey, the elasticity is estimated as −0.67.25 The evidence from the use of annual time series data also points to an estimate of −0.39.17
Here, we use monthly data of price and quantity over the January 2006 to July 2014 period for all 81 brands of cigarettes. We estimate the price elasticity within a range of −0.55 and −0.90 (table 2). The figures estimated for all cigarettes are higher than the elasticity estimates found in the aforementioned studies in absolute terms.
We may further analyse the elasticity based on the price ranges of cigarettes. For this analysis, the price segments are defined as follows: prices less than the median price, prices between the median and 75th centile and prices higher than the 75th centile at each period. The estimation results reveal that price elasticities of demand for cheaper products are much higher than the figures reported previously, when brand-specific factors are controlled for (table 3, columns 1 and 2). Meanwhile, results show that the demand for cigarettes in the relatively expensive price segment is highly inelastic.
Discussion
In this study, we aim to shed light on the nature of tobacco taxation in Turkey by considering the reactions of economic agents: firms and consumers. A tax scheme based on ad valorem and specific components introduces incentives for firms to change prices. Consumers also react to price changes induced by tax hikes, partly by quitting, partly by switching to lower priced cigarettes and partly by switching to non-duty paid cigarettes. Thus, firm and consumer behaviour may undermine the effectiveness of taxation as a public policy tool for reducing tobacco consumption.
The findings of our study have important policy implications. First, given the nature of taxation in Turkey, firms' strategy may deter the expected effects of a tax hike. For instance, when faced with tax hike, firms may choose not to transfer the entire tax rise to final prices. That is, prices will not go up as much as intended by the tax policy, and thus the expected reduction in consumption will be limited. Furthermore, a very high ad valorem rate provides strong incentives to firms to cut prices. In this way, the firm can enjoy a much higher demand for its products. Therefore, this sort of tax strategy may induce even more consumption despite its initial aim of discouraging consumption.
Second, price elasticity of demand estimations show that higher prices induced by tax hikes actually lower overall consumption. However, the overall finding of inelastic demand for cigarettes in Turkey hides a good amount of elastic demand/substitution behaviour in the non-premium category. In fact, our results reveal that the demand is actually elastic and higher prices dramatically reduce consumption of cigarettes with prices lower than the 75th centile. Meanwhile, the demand for cigarettes with prices higher than the 75th centile (ie, premium segment) is not significantly affected by price changes. This finding has strong implications for public policy. High price policy is still an effective tool for reducing tobacco consumption for the majority of consumers smoking relatively cheaper cigarettes. On the other hand, in an ad valorem-dominant regime, while increasing the tax revenue, a tax hike is ineffective for inducing consumers smoking relatively expensive cigarettes to reduce consumption. Therefore, non-price policies such as more aggressive public awareness campaigns could be directed more towards these consumers.
It is true that Turkey has a responsive and aggressive tax policy. However, the tax mix, which historically stayed on extreme cases, imposes some risks on the level and stability of tax revenues and induces switching behaviour as well. Therefore, the main conclusion of our study is that securing high tax rates is not enough. What needs to be done is to secure high rates by considering the composition of tax in terms of specific and ad valorem. To conclude, the current tax scheme, based on both ad valorem and specific components, introduces various incentives to firms as well as to consumers, which reduce the effectiveness of the tax policy. Thus, moving to a more balanced (in terms of ad valorem and specific) tax scheme would be the first step in curbing such incentives and in building a more effective policy tool for reducing tobacco consumption. In this regard, our findings provide a real-life case for the theoretical predictions of previous studies on the balance between specific and ad valorem rates, where the main conclusion is that the optimal balance between the two taxes is quite sensitive to the characteristics of the markets.26 ,27
What this paper adds
The firm behaviour and consumer choice may lessen the effectiveness of tobacco taxation in reducing consumption, as previously touched on to some extent in the literature. However, these issues are not extensively discussed with reference to the nature of the tax system in terms of specific and ad valorem rates.
An all-ad valorem (all-specific) tax maximises (minimises) price decrease incentives, while an all-ad valorem (all-specific) tax minimises (maximises) price increase incentives. Thus, firms' incentives postulate a trade-off for the tax authority when setting the tax mix.
Looking at the aggregate figures, one can conclude that the demand for cigarettes in Turkey is inelastic. However, this finding does not hold for all price segments. For the economy and medium segments, demand is found to be elastic. On the other hand, the demand for the premium segment is not significantly affected by price changes.
Therefore, for reducing tobacco consumption, tax policy is still an effective tool for the majority of consumers smoking relatively cheaper cigarettes, but for consumers smoking premium cigarettes non-price policies could be more effective.
Acknowledgments
OA and MUÖ have both contributed to the paper as authors.
References
Footnotes
Disclaimer The views and opinions presented in this study belong to the authors and do not necessarily represent those of the Central Bank of the Republic of Turkey or its staff.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
↵i Based on interviews with the biggest producers in the market in 2014, distributor's share is taken as 7%. Knowing the distributor's share, the VAT and SCT rates, and the total tax revenue (4 TL in this case) final consumption and producer's price can be calculated using equation (1).