Price elasticity of the demand for soft drinks in the Netherlands The effectiveness of a sugar tax
Abel Grünfeld, 10747036 BSc Economics and Business Bachelor’s Thesis Economics Supervisor: Andro Rilović
Date: June 27, 2017
Abstract
Growing empirical evidence shows that the consumption of sugary drinks increases the risk of obesity and type 2 diabetes. The detrimental effects to health have resulted in
recommendations for public policy makers. However, there is need for scientific evidence that focuses on the price sensitivity of the Dutch population. The objective of my research is to estimate price elasticities of the demand for soft drinks in the Netherlands, and thus present useful evidence for policy makers.
This paper used pooled data collected from the Dutch National Food Consumption Survey conducted by the RIVM. A log-log model was formed to estimate price elasticities of demand for soft drinks. My analysis accounted for differential effects regarding gender, age and BMI.
An overall -1.43 price elasticity was calculated for soft drinks. Men are expected to react more strongly to a price change as compared to women. In addition, people aged 8-36 are likely to be more sensitive to price adjustments. No evidence that confirms a relation between BMI and price sensitivity for the demand of soft drinks was found.
The demand for soft drinks in the Netherlands is elastic. Taxation could be an
effective policy to reduce consumption of soft drinks. Assuming a pass-on-rate to consumers of 100%, a tax of 10% results in a demand reduction of 14.3 %.
Statement of Originality
This document is written by Abel Grünfeld who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Table of contents
INTRODUCTION ... 4
SOFT DRINKS AND THE NETHERLANDS ... 5
LITERATURE REVIEW ... 6
METHODOLOGY ... 9
CONSUMPTION AND POPULATION ... 9
PRICE DATA ... 9
EMPIRICAL MODEL: EFFECT OF A TAX RISE ON SOFT DRINK PURCHASES ... 10
RESULTS ... 11
CONCLUSION ... 14
DISCUSSION ... 14
STRENGTHS AND LIMITATIONS ... 14
POLICY IMPLICATIONS ... 15
FURTHER RESEARCH ... 16
BIBLIOGRAPHY ... 17
APPENDIX 1: AGE GROUPS & BMI GROUPS ... 20
APPENDIX 2: REGRESSION OUTPUTS ... 21
APPENDIX 3: EUROMONITOR STATISTICS ... 31
APPENDIX 4: DIFFERENCES AMONG SSBS ... 32
Introduction
Evidence on the correlation between obesity and consumption of sugary drinks is increasing and becoming stronger (Malik, Schulze & Hu, 2006; Vartanian, Schwartz & Brownell, 2007). Globally, an increasing group of experts argue in favor of the introduction of taxation to disincentivize the consumption of unhealthy food and drinks (International Diabetes Federation Europe, 2016; World Health Organization, 2015; Zizzo, Parravano, Nakamura, Forwood & Suhrcke, 2016). In particular, sugar-sweetened beverages1 (SSBs) – drinks sweetened with any form of added sugars, including non-diet soft drinks, flavored juice drinks, sport drinks, sweetened tea, coffee drinks, energy drinks, and electrolyte replacement drinks (Centers for disease control and prevention, 2017; Department of Health, State of Rhode Island, n.d.) – are considered as target group for additional taxation. A group of globally operating diabetes specialists urged the G20-countries to impose a tax on SSBs (Hirschler, 2015). In addition, The World Health Organization (WHO) declared that a 20% sugar tax on SSBs will result in lower consumption and thus decrease the likeliness of obesity (World Health Organization, 2015).
Various countries have introduced taxation on unhealthy consumption to prevent people from ending up with serious health issues. Mexico, in 2014, has been the first country to implement a so-called sugar tax on SSBs, of which the first results are promising (Briggs, 2016). This tax is exclusively levied on SSBs. Recently, also Great Britain has officially agreed to introduce a sugar tax. More and more countries are considering such a tax. The Netherlands is no exception. Although there is not yet a parliamentary majority in favor of a tax on SSBs, it is highly likely that it will be a continuous subject on the political agenda. The third largest political party, CDA2, has included the topic of a sugar tax on SSBs in its
electoral program.
Crucial in the discussion of enforcing a tax will be the effectiveness of the tax. Therefore, this paper focuses on a (micro)economic analysis of a so-called sugar tax. Using available empirical data, I am forced to focus on soft drinks rather than SSBs due to the insufficient availability of appropriate data in the Netherlands. By estimating price elasticities of demand for soft drinks, I hope to add some valuable, empirical results to the existing literature.
1 See appendix 4 for various types of SSBs.
2 CDA is the Christian Democratic Appeal, established in 1980. Currently they possess 19 (out of 150) seats in
Soft drinks and the Netherlands
The topic of a sugar tax on soft drinks to reduce consumption, and thus the risk of obesity and type 2 diabetes, is frequently debated. Historical data from the CBS3 shows that over the last 40 years the consumption of soft drinks has dramatically increased. Average individual consumption increased almost 85% between 1970 and 2012. From 2012 to 2015, average consumption decreased by 10%. Possibly, this decrease can be explained by a consumer tax rise enforced by the Dutch government since 2012. The tax was levied to increase revenue rather than reducing the consumption of soft drinks. More importantly, empirical evidence to verify the correlation between the tax rise and the reduction in soft drink consumption is lacking.
Although in recent years the consumption of soft drinks has decreased, the overall consumption is still high compared to other European countries (FWS, 2015). Another argument in favor of a policy aimed at reducing soft drink intake can be made after studying the data on Body Mass Index (BMI). Data retrieved from Euromonitor shows a constant increase in mean BMI in the Netherlands4. Euromonitor also predicts that this upward trend will not change in the coming years5. From 1977 to 2016, the percentage of the population aged 15+ classified as obese in the Netherlands has increased 189%. Clearly, public policy to tackle this trend is desirable as public health costs will rise to unsustainable and undesirable heights. Fiscal policy aimed at SSBs could be one way to stop this trend. Furthermore, additional government revenue from such a fiscal policy can be used to compensate for the regressive effects of a tax and further promote healthy substitutes for soft drinks.
3 Centraal Bureau voor de Statistiek (Central Agency for Statistics).
4 See appendix 3 5 See appendix 3
Literature review
Relevant empirical research has been conducted. However, no evidence is available that estimates price elasticities in the Netherlands. Colchero, Salgado, Unar-Munguia, Hernandez-Avila & Rivera-Dommarco (2015) have estimated the price elasticity of the demand for SSBs in Mexico by setting up an Almost Ideal Demand System6 (AIDS). They used 2006, 2008, and 2010 Mexican National Income and Household Expenditure Surveys (MNIHES) that included cross-sectional data on the consumption of sugary drinks combined with several factors such as income level and urban vs rural living conditions. Price elasticity estimates for soft drinks were -1.06 versus -1.16 for other SSBs. Also, they concluded that higher
elasticities occur among households living in rural areas rather than urban and with lower compared to higher income.
A frequently used argument by opponents of a sugar tax on SSBs is that such a tax is unfair because it is regressive and thus hurts the poorer population more. Colchero et al. (2015) reject this argument by mentioning that lower-income households have a more elastic demand for soft drinks. As a consequence, the financial burden of a possible tax will mainly hit higher-income households.
In Chile similar research has been conducted to find out whether the demand for SSB is elastic or not. Guerrero-López, Unar-Munguía & Colchero (2017) have used an approach that is very similar to the one of Colchero et al. (2015). They estimated a demand system for beverages using a cross-sectional household survey. They found out that soft drinks have a price elasticity of -1.37 versus -1.67 for other SSB. Also, a correlation regarding income level was found. Lower-income households consume relatively more SSBs and show higher
sensitivity to price changes. Guerrero-López et. al (2017) conclude that only a significant price increase (³ 10%) in the medium to long run will have an economic and health impact. They argue that extra fiscal revenue should be used to reduce the regressive character of the
6 AIDS is a system of equations that describe consumer demand. Introduced by Deaton & Muellbauer (1980),
the system offers researchers the chance to estimate price elasticities. The model is frequently used but criticism remains. When multicollinearity among prices occur, (cross-) price elasticity estimates are often unreliable (Alston, Foster & Green, 1994). Applied to the beverage industry, a price increase of soft drinks could foster a price increase for soft drink substitutes, and thus multicollinearity among prices arises. Also Green & Alston (1990) warn for wrong results using AIDS due to various ways of applying the AIDS to estimate elasticities. Comparing the results of Colchero et al. (2015) with Guerrero-López et. al (2017), a different application of AIDS could be an explanation for the substantial difference in estimated elasticities.
tax. This can be done by educational programs to reduce the asymmetric distribution of information and by making drinking water publicly available.
Andreyeva, Chaloupka & Brownell (2011) present a method to estimate tax revenues from an excise tax on SSBs. They used data on regional beverage consumption in the US, historic trends and recent price elasticity estimates of SSB to calculate the total tax revenue. They conclude that the public health impact of taxes on SSBs could be substantial. A 20% price increase would result in a reduction of SSB intake by 24%. Consumption among youth and lower-income groups in particular could be influenced by fiscal policy. This conclusion is supported by Powell and Chaloupka (2009) who argue that especially children, lower-income groups and those most at risk for obesity respond to price changes. They add that investing the tax revenue in obesity prevention programs could lead to more pronounced benefits.
Ireland is one of the countries where a tax on SSBs is part of the political debate. Several propositions in favor of an enforcement of a sugar tax have been discussed. Briggs, Mytton, Madden, O’Shea, Rayner & Scarborough (2013) have examined the potential impact on obesity of a 10% tax on SSBs. They used price elasticity estimates to determine the effect of a 10% tax considering a pass-on-rate to consumers of 90%. The -0.9 price elasticity of SSBs is predicted to reduce obesity by 1.3% and overweight by 0.7%. An interesting finding is that there are no significant differences related to gender and income disparities. On the other hand, they found differences regarding age. Young adults react more strongly to the tax, and thus obesity reductions are greater for them.
Similar results are presented by Briggs, Mytton, Kehlbacher, Tiffin, Rayner, & Scarborough (2013) for the UK. A tax on SSBs of 20% leads to a reduction of obesity that amounts to 1.3% and 0.9% for overweight people. A tax of 10% will result in a reduction of obese people of 0.6%. Ireland and the UK seem to be comparable countries. A price increase is expected to have similar effects on SSB consumption. However, the effects of a SSB tax to reduce the obese population are estimated to be different. A plausible explanation is that the relative contribution of sugary drink consumption that causes obesity can be different. Also, the estimated reduction in energy intake from a similar SSB tax is different. Further research is required to understand why a 20% tax in the UK has the same result as a 10% tax in Ireland regarding the reduction of the obese population.
Comparing the estimated price elasticities in the US to Ireland and the UK, a
considerable disparity is noted. Andreyeve et al. (2011) estimate that a 20% price increase of SSBs results in a 24% consumptive reduction. A similar price increase lessens SSB intake by
about 15% in Ireland and the UK (Briggs et al., 2013a; Briggs et al., 2013b). Explanations for this dissimilarity include price differences for substitutes7 and distinct preferences. Also, Andreyeve et al. (2011) focus on the effect of an excise tax. Briggs et al. (2013a) and Briggs et al. (2013b) do not specifically focus on an excise tax.
These findings show that the possible effectiveness of a soft drink tax seem to vary from country to country. Applied to the Netherlands, Waterlander, Steenhuis, de Boer, Schuit, & Seidell (2012) and Waterlander, Elzeline, Mhurchu, & Steenhuis (2014) have examined the effect of a price increase on the demand for unhealthy food and beverages, including soft drinks. Rather than estimating price elasticities, they showed in a virtual
supermarket experiment that price changes influence consumer behavior. There is widespread academic consensus on the correlation between price and consumption of SSBs. However, to make recommendations for policy makers, more detailed knowledge of price elasticity estimates in the Netherlands are required. Therefore, my research focuses on the relation between price and demand for soft drinks. The estimation of price elasticities adds useful empirical evidence to the available literature. To account for variations between countries, public policy suggestions require country-specific data. My analysis uses data from the RIVM8 household survey. This allows me to focus on the effect of a sugar tax in the Netherlands. In addition, the data facilitates the necessary information – regarding gender, age, and BMI – to calculate differential effects.
In short, the objective of my paper is to estimate the price elasticity of demand for soft drinks in the Netherlands, and thus examine the effectiveness of a soft drink tax. Results from this paper could be used in the governments’ considerations to introduce additional fiscal policies to reduce consumption of soft drinks. It could also add valuable information to the evaluation of current public policy.
7 Price differences for substitutes include non-sweetened dairy products, 100% fruit juices, mineral waters,
coffee and tea.
8 Rijksinstituut voor Volksgezondheid and Milieu (The National Institute for Public Health and the
Methodology
In the following section, I will describe the dataset I have used and explain all variables that are included in my empirical model. Also the analytical method applied to estimate
elasticities will be presented.
Consumption and population
Consumption and population data (pooled data) I have collected from the Dutch National Food Consumption Survey conducted by the RIVM. The survey was held over five time periods9 and includes information on the consumption of soft drinks for different sex, age and Body Mass Index (BMI). The RIVM collected food and beverage consumption of all
participants for two days. The analytical sample includes 2161 individuals.
Price data
To collect data on the prices of soft drinks I had to convert different indexes. First I have used the CBS’s soft drink consumer price index (CPI) with base year 2015 to calculate the relative changes in price for Dutch consumers during the period of my examination. The average price of soft drinks in the Netherlands is not available. Therefore, I have used the average price of soft drinks in the UK (Statista, 2015) to calculate the price in the Netherlands. One reason why I think that UK price level can be used as a proxy for the Netherlands is due to the European internal market. Free trade allows retailers from both countries to buy products overseas without being charged an import duty. Therefore, great price dispersions give retailers an incentive to buy the product abroad which results in a domestic price decrease. This way, prices in the UK and the Netherlands will converge. Another reason why I believe that UK prices can be used as a proxy is the comparable inflation rate in both countries. Also the degree of competition influences soft drink prices. Herfindahl-Hirschman indices (HHIs) are calculated to compare market competitiveness among the two countries. Data is retrieved from Euromonitor. The HHI for the UK soft drink market amounts to ± 0.06310 versus ± 0.05411 in the Netherlands, and thus market competitiveness among the UK and the Netherlands is expected to be comparable. On the other hand, the assumption could be
9 Time periods are 2007, 2008, 2009, 2013 and 2014.
10 HHIs are calculated for 2007, 2008, 2009, 2013 and 2014. The HHIs vary from 0.058 to 0.069. A HHI < 0.15
indicates an unconcentrated industry.
problematic due to a tax scheme change in the Netherlands in 2012. However, statistics from the UK’s Office for National Statistics show a similar change in tax regime.
To convert the UK price to an appropriate Dutch one, I have assumed that the Eurostat report12 on “comparative price levels for food, beverages and tobacco” is correct. The report states that UK price level for non-alcoholic beverages is 126 versus 98 in the Netherlands. Note that I assume the comparative price level for soft drinks to be similar to the one of non-alcoholic beverages. Available data to test this assumption was limited. I have used consumer price data from the CBS to compare soft drinks to non-alcoholic beverages. This data showed a marginal difference in price level.
The relative price level (NL/UK) multiplied by the average soft drink price in the UK for 2015 resulted in a converted price of 0.824 British pound per litre. Finally, I have taken the average exchange rate (£/€) of 2015, reported by the European Central Bank (ECB), which amounted to 1.3785. Multiplying the converted price (in £/litre) by the average exchange rate (£/€) resulted in the average soft drink price in the Netherlands of €1.136 in 2015. I have used this price adjusted by the change in CPI for soft drinks to determine the price level for each year.
Empirical model: effect of a tax rise on soft drink purchases
To determine whether a tax on soft drinks will be effective, price elasticities of demand are estimated. I am particularly interested in the effect that such a tax will have on different types of consumers. To find out whether gender, age and BMI change the effectiveness of a soft drink tax, separate analyses are performed. I formed groups of age and BMI to make interpretations clearer13 and to create sufficiently large sample sizes. Hereafter, I computed the consumed quantities for each time period and group through which I can apply the empirical model that I present in the next paragraph.
The log-log model is a logarithmic transformation of a regression model that fixes the requirement of linearity in parameters. In other words, the model allows me to set up a regression function with a slope that is not constant. An increase in the independent variable (price) does not necessarily result in an increase in the dependent variable (quantity
demanded). This model is appropriate for my data as they do not show a constant relation
12 See appendix 5 13 See appendix 1.
between price and quantity demanded14. Additionally, the interpretation of the model suits the goal of my research. Due to the connection between logarithms and percentages, 𝛽" can be interpreted as the elasticity of the independent variable (price) with respect to the
dependent variable (quantity demanded) (Stock & Watson, 2015)15.
𝐿𝑛(𝑄𝑢𝑎𝑛𝑡𝑖𝑡𝑦-,/)) = 𝛽2+ 𝛽" 𝐿𝑛(𝑃𝑟𝑖𝑐𝑒-) + 𝜀-16
Results
The elasticity estimates will be demonstrated to analyze the consequences of a sugar tax on soft drinks. Also, the expected change in soft drink consumption will be displayed assuming a 100% pass-on-rate to consumers.
Table 1: Descriptive statistics 2007
n = 485 2008 n = 483 2009 n = 439 2013 n = 375 2014 n = 379 All years n = 2161 % of individuals with soft
drink consumption > 0
64.5 66.9 63.3 62.1 58.6 63.4
Average price (€/litre) 0.946 0.982 0.981 1.117 1.136 1.024 Mean consumption (litre) 103.3
(5.86) 111.83 (6.27) 109.89 (6.85) 86.33 (6.02) 85.26 (6.06) 100.44 (2.81)
Table 1 shows a downward trend in the percentage of people that consume soft drinks starting from 2009. The consumer tax on soft drinks, enforced since 2012, is a plausible explanation for the observed increase in average price per litre for soft drinks. Comparing 2009 with 2014, a significant difference in mean consumption is noted. To get a better understanding of the decrease in mean consumption, I have estimated price elasticities. Also the effects of future price rises can be evaluated due to the elasticity estimates.
Table 2 presents estimates of price elasticities of demand for different groups of individuals (see appendix 2 for regression outputs). I have found an overall elasticity of -1.43 for the population. It appears that men will react stronger to price changes than women. I also found evidence hinting that young people (8-36 years) are likely to respond more heavily to
14 Comparing 2008 to 2007, an increase in consumption and price for soft drinks is observed. Comparing 2009
to 2008, a decrease in consumption and price is observed. Comparing 2013 to 2009, a decrease in consumption and an increase in price is observed. In short, there is no constant relation between price and quantity demanded in my dataset.
15Ln(Y + DY) – ln(Y) = [β
2+ β" Ln(X +DX)]− [β2+ β" Ln(X)] = 𝛽" [ln(X +DX) – ln(X).
price changes. In particular, men aged between 8-18 and women aged between 19-36. Noteworthy is that I have not found a relation between soft drink consumption and BMI.
Table 2: Price elasticities of demand
Price elasticity
Overall -1.43* (0.36)
Male -1.67* (0.46)
Female -1.09** (0.43)
Age Group 1 (8-18 year) -1.84** (0.68) Age Group 2 (19-36 year) -1.62* (0.39) Age Group 3 (37-59 year) 0.36 (0.90) Age Group 4 (60+ year) - 1.16 (1.76) Male, Age Group 1 -2.87** (1.08) Male, Age Group 2 -1.33 (1.11) Male, Age Group 3 1.17 (1.08) Male, Age Group 4 -1.04 (2.40) Female, Age Group 1 -0.41 (0.42) Female, Age Group 2 -2.21** (0.72) Female, Age Group 3 -1.11 (1.52) Female, Age Group 4 -1.11 (2.39) BMI 1 (< 18,5) -6.01 (2.72) BMI 2 (18,5 £ BMI < 25) -1.84 (0.36) BMI 3 (25 £ BMI < 30) -0.76 (0.58) BMI 4 (³ 30) -0.02 (0.87) *significant at a = 0.05 **significant at a = 0.1
The potential after-tax percentage reduction in soft drink consumption depends on the share of the tax that is passed on to consumers. I assumed a 100% pass-on-rate to consumers because previous research has shown that the pass-on-rate to consumers in the UK, US and France equals 100% (Besley & Rosen, 1998; Crawford, Keen & Smith, 2010; Berardi, Sevestre, Tepaut, & Vigneron, 2016). A sugar tax that amounts to 10% would result in a 14.3% reduction in soft drink consumption. Table 3 shows the estimated percentage reduction in demand for soft drinks after a 10%. I have calculated17 the tax effect for all statistically significant results (presented in table 2).
17 Assuming ceteris paribus, I made use of this formula to calculate the effect of a 10% tax. 𝛽
" refers to the price
elasticity estimate. 100 ∙ DEFGHI-IJK,L
EFGHI-IJK,L = 100 ∙ ( DMN-OPL
Table 3: Estimated % reduction in soft drink consumption after tax t = 10% Overall 14.3 Male 16.7 Female 10.9 Age Group 1 18.4 Age Group 2 16.2
Male, Age Group 1 28.7 Female, Age Group 2 22.1
Conclusion
My price elasticity estimates infer that the demand for soft drinks is elastic. A tax of 10%, assuming a pass-on-rate to consumers of 100%, results in a demand reduction of 14.3%. People between 8-36 will be effected the most. More men than women are expected to respond to price changes. No evidence for a correlation between BMI and the price
sensitivity for soft drinks has been found. All in all, a tax on soft drinks could be an effective policy design if the goal is to reduce consumption among young people. Presuming that obesity and heart diseases occur at an older age, a soft drink tax could lower future healthcare costs and cause net benefits in the long run.
Discussion
I have estimated an overall price elasticity of -1.43 in the Netherlands. The biggest impact of a sugar tax would be observed for teenage men (8-18) and young women (19-36). There is no evidence to suggest that the effectiveness of a sugar tax differs across various BMI groups. Based on my findings, middle-age and older people are not expected to react strongly to a sugar tax. Consequently, my conclusions provide useful input into the current policy debate on the merits of a tax on soft drinks. However, additional research on the effectiveness of a sugar tax is required to formulate policy implications.
Strengths and limitations
The strengths of my paper include the following. Foremost, it is the first study using the National Dutch Food Consumption Survey of 2007-2010 and 2013-2014 to examine the effect of a sugar tax on soft drinks. Previous research has been done for the Netherlands by means of an experiment in a virtual supermarket (Waterlander et al., 2014), not by estimating price elasticities. Secondly, it considers the effects of a tax for disparate groups, sorted by gender, age and BMI.
My paper has several limitations. Due to limited availability of useful data, my elasticity estimates are merely based on 5 inconsecutive time periods. Possibly, this results in biased price elasticity estimates18. Ideally the sample size would be larger as well.
The empirical model that I have used to estimate price elasticities for demand is rather simple. A more sophisticated model is likely to result in more reliable estimates, and thus more useful results for policy makers.
Next, the way I had to estimate average soft drink price is quite cumbersome. As I had to convert UK prices and currency, the precision of my average price estimate could be weakened. I had to assume that the UK market is similar to the Dutch market in order to use UK prices as a proxy for the Netherlands. In addition, I assumed the price level for soft drinks to be similar to the one of non-alcoholic beverages.
I have only focused on own price elasticity. Cross-price elasticities should be
estimated for all substitute goods in order to oversee all consequences of a tax. If soft drinks are replaced by other SSBs, taxation should be levied over these unhealthy substitutes as well. Although my estimates add value to the existing knowledge regarding the effectiveness of a sugar tax, it is crucial to know what individuals consume instead of soft drinks.
Policy implications
My research is not sufficiently strong to result in policy implications. However, I have found some useful results. My research shows strong indications regarding the value of a sugar tax. I found out that in the Netherlands, there is no evidence to support the claim that a tax would reduce consumption by obese people more than non-obese. On the other hand, it is expected to affect young people who are the main consumers. This way it could be seen as an
instrument to limit the number of obese people in the future, and thus result in long-run health benefits and healthcare cost savings.
To present a concrete tax proposal, further research is required to determine the type of tax and to calculate the required level of taxation.
18 Due to unavailability of data, I could not include time periods 2010, 2011 and 2012 in my analysis. Therefore,
Further research
Many questions stay unanswered. Price elasticity estimates to influence public policy design should be based on more time periods. On top of that, cross-price elasticity estimates are required to understand the substitute effect.
I have calculated differential effects for gender, age and BMI groups. More factors should be included in the estimation of elasticities. Counting for income level and urban versus rural living conditions would add up to the applicability of elasticity estimates.
To test the effectiveness of a tax on soft drinks, multiple tax types should be tested. Research in the US by Sharma, Hauck, Hollingsworth, & Siciliani (2014) has shown that a flat tax on sales has distinct effects as compared to a volumetric tax. Moreover, the response of soft drink manufacturers to a sugar tax is ambiguous. Taxing producers rather than consumers is expected to have different results. Also, the tax pass-on-rate to consumers should be computed to determine the real effects of a sugar tax.
Acknowledgement
I sincerely thank Andro Rilović for the supervision of my thesis. His comments have significantly contributed to the quality of this research.
Bibliography
Alston, J. M., Foster, K. A., & Green, R. D. (1994). Estimating elasticities with the linear approximate almost ideal demand system: some Monte Carlo results. The review of
Economics and Statistics, 351-356.
Andreyeva, T., Chaloupka, F. J., & Brownell, K. D. (2011). Estimating the potential of taxes on sugar-sweetened beverages to reduce consumption and generate revenue. Preventive
medicine, 52(6), 413-416.
Berardi, N., Sevestre, P., Tepaut, M., & Vigneron, A. (2016). The impact of a ‘soda tax’ on prices: evidence from French micro data. Applied Economics, 48(41), 3976-3994.
Besley, T. J., & Rosen, H. S. (1998). Sales taxes and prices: an empirical analysis (No. w6667). National Bureau of Economic Research.
Briggs, A. D., Mytton, O. T., Madden, D., O’Shea, D., Rayner, M., & Scarborough, P. (2013a). The potential impact on obesity of a 10% tax on sugar-sweetened beverages in Ireland, an effect assessment modelling study. BMC Public Health, 13(1), 860.
Briggs, A. D., Mytton, O. T., Kehlbacher, A., Tiffin, R., Rayner, M., & Scarborough, P. (2013b). Overall and income specific effect on prevalence of overweight and obesity of 20% sugar sweetened drink tax in UK: econometric and comparative risk assessment modelling study. Bmj, 347, f6189.
Briggs, A. (2016). Sugar tax could sweeten a market failure. Nature, 531(7596), 551-551. Centers for disease control and prevention. (April 7, 2017). Get the Facts: Sugar-Sweetened Beverages and Consumption. Retrieved from
Colchero, M. A., Salgado, J. C., Unar-Munguia, M., Hernandez-Avila, M., & Rivera-Dommarco, J. A. (2015). Price elasticity of the demand for sugar sweetened beverages and soft drinks in Mexico. Economics & Human Biology, 19, 129-137.
Crawford, I., Keen, M., & Smith, S. (2010). Value added tax and excises. Dimensions of tax
design: The Mirrlees review, 275-362.
Deaton, A., & Muellbauer, J. (1980). An almost ideal demand system. The American
economic review, 70(3), 312-326.
FWS. (2015). Kerngegevens 2015. Publication of the Dutch association for soft drinks, waters and juices.
Green, R., & Alston, J. M. (1990). Elasticities in AIDS models. American Journal of
Agricultural Economics, 72(2), 442-445.
Guerrero-López, C. M., Unar-Munguía, M., & Colchero, M. A. (2017). Price elasticity of the demand for soft drinks, other sugar-sweetened beverages and energy dense food in
Chile. BMC public health, 17(1), 180.
Hirschler, B. (2015, November 11). Diabetes experts tell G20 to tax sugar to save lives and money. Reuters. Retrieved from http://www.reuters.com/article/us-health-diabetes-sugar-idUSKCN0T100K20151112
International Diabetes Federation Europe. (2016). IDF Europe position on added sugar.
Malik, V. S., Schulze, M. B., & Hu, F. B. (2006). Intake of sugar-sweetened beverages and weight gain: a systematic review. The American journal of clinical nutrition, 84(2), 274-288. (n.d.). Sugar-sweetened beverages. Department of Health, State of Rhode Island. Retrieved from http://www.health.ri.gov/healthrisks/sugarsweetenedbeverages/
Powell, L. M., & Chaloupka, F. J. (2009). Food prices and obesity: evidence and policy implications for taxes and subsidies. Milbank Quarterly, 87(1), 229-257.
Sharma, A., Hauck, K., Hollingsworth, B., & Siciliani, L. (2014). The effects of taxing sugar-sweetened beverages across different income groups. Health economics, 23(9), 1159-1184.
Stock, J.H. and Watson, M.W. (2015). Introduction to Econometrics (updated third edition (global edition)). Essex: Pearson Education.
Statista. (2017). Average price of soft drinks in the United Kingdom (UK) from 2008 to 2015 (in GBP per litre). Retrieved from https://www.statista.com/statistics/308926/average-soft-drink-price-in-the-united-kingdom-uk/
Vartanian, L. R., Schwartz, M. B., & Brownell, K. D. (2007). Effects of soft drink consumption on nutrition and health: a systematic review and meta-analysis. American
journal of public health, 97(4), 667-675.
Waterlander, W. E., Steenhuis, I. H., de Boer, M. R., Schuit, A. J., & Seidell, J. C. (2012). Introducing taxes, subsidies or both: the effects of various food pricing strategies in a web-based supermarket randomized trial. Preventive Medicine, 54(5), 323-330.
Waterlander, Wilma Elzeline, Cliona Ni Mhurchu, and Ingrid HM Steenhuis. "Effects of a price increase on purchases of sugar sweetened beverages. Results from a randomized controlled trial." Appetite 78 (2014): 32-39.
World Health Organization. (2015). Fiscal policies for diet and prevention of noncommunicable diseases. Geneva, Switzerland.
Zizzo, D. J., Parravano, M., Nakamura, R., Forwood, S., & Suhrcke, M. (2016). The impact
of taxation and signposting on diet: an online field study with breakfast cereals and soft drinks (No. 131cherp).
Appendix 1: Age groups & BMI groups
Age groups:
Age Group 1 8-18 year Age Group 2 19-36 year Age Group 3 37-59 year Age Group 4 60+ year BMI groups:
BMI Group 1 < 18,5 Underweight
BMI Group 2 18,5 £ BMI < 25 Normal weight BMI Group 3 25 £ BMI < 30 Overweight
Appendix 2: regression outputs a) Overall b) Male Regression Statistics Multiple R 0,902208 R Square 0,813979 Adjusted R Square 0,751972 Standard Error 0,077288 Observations 5 ANOVA df SS MS F Significance F Regression 1 0,078414 0,078414 13,127194 0,036167 Residual 3 0,017920 0,005973 Total 4 0,096334 Coefficients Standard
Error t Stat P-value Lower 95%
Upper 95% Intercept 4,773321 0,037074 128,752309 0,000001 4,655336 4,891306 Price(ln) -1,671696 0,461393 -3,623147 0,036167 -3,140055 -0,203336 Regression Statistics Multiple R 0,918081 R Square 0,842872 Adjusted R Square 0,790496 Standard Error 0,059886 Observations 5 ANOVA df SS MS F Significance F Regression 1 0,057714 0,057714 16,092699 0,027797 Residual 3 0,010759 0,003586 Total 4 0,068473
Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Intercept 4,633262 0,028726 161,289253 0,000001 4,541842 4,724682 Price (ln) -1,434173 0,357509 -4,011571 0,027797 -2,571926 -0,296419
c) Female Regression Statistics Multiple R 0,822761 R Square 0,676935 Adjusted R Square 0,569247 Standard Error 0,072864 Observations 5 ANOVA df SS MS F Significance F Regression 1 0,033374 0,033374 6,286071 0,087152 Residual 3 0,015928 0,005309 Total 4 0,049302
Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Intercept 4,457526 0,034952 127,533358 0,000001 4,346293 4,568758 Price (ln) -1,090600 0,434986 -2,507204 0,087152 -2,474921 0,293721 d) Age Group 1 Regression Statistics Multiple R 0,842222 R Square 0,709338 Adjusted R Square 0,612450 Standard Error 0,113953 Observations 5 ANOVA df SS MS F Significance F Regression 1 0,095068 0,095068 7,321252 0,073426 Residual 3 0,038956 0,012985 Total 4 0,134024 Coefficients Standard
Error t Stat P-value Lower 95%
Upper 95% Intercept 4,897040 0,054661 89,588742 0,000003 4,723083 5,070996 Price (ln) -1,840681 0,680277 -2,705781 0,073426 -4,005626 0,324264
e) Age Group 2 f) Age Group 3 Regression Statistics Multiple R 0,224444 R Square 0,050375 Adjusted R Square -0,266166 Standard Error 0,151384 Observations 5,000000 ANOVA df SS MS F Significance F Regression 1,000000 0,003647 0,003647 0,159143 0,716646 Residual 3,000000 0,068751 0,022917 Total 4,000000 0,072398 Coefficients Standard
Error t Stat P-value Lower 95%
Upper 95% Intercept 4,206150 0,072616 57,922923 0,000011 3,975052 4,437248 Price (ln) 0,360523 0,903733 0,398927 0,716646 -2,515557 3,236604 Regression Statistics Multiple R 0,921585 R Square 0,849319 Adjusted R Square 0,799092 Standard Error 0,065920 Observations 5 ANOVA df SS MS F Significance F Regression 1 0,073480 0,073480 16,909617 0,026047 Residual 3 0,013036 0,004345 Total 4 0,086516 Coefficients Standard
Error t Stat P-value Lower 95%
Upper 95% Intercept 4,955731 0,031621 156,723745 0,000001 4,855099 5,056362 Price (ln) -1,618249 0,393531 -4,112131 0,026047 -2,870640 -0,365859
g) Age Group 4
h) Male, Age Group 1 Regression Statistics Multiple R 0,839196 R Square 0,704250 Adjusted R Square 0,605667 Standard Error 0,180077 Observations 5 ANOVA df SS MS F Significance F Regression 1 0,231653 0,231653 7,143714 0,075512 Residual 3 0,097283 0,032428 Total 4 0,328936 Coefficients Standard
Error t Stat P-value Lower 95%
Upper 95% Intercept 5,036942 0,086380 58,311573 0,000011 4,762043 5,311841 Price (ln) -2,873293 1,075023 -2,672773 0,075512 -6,294496 0,547911 Regression Statistics Multiple R 0,355859 R Square 0,126636 Adjusted R Square -0,164486 Standard Error 0,293987 Observations 5,000000 ANOVA df SS MS F Significance F Regression 1,000000 0,037596 0,037596 0,434993 0,556659 Residual 3,000000 0,259286 0,086429 Total 4,000000 0,296882 Coefficients Standard
Error t Stat P-value Lower 95%
Upper 95% Intercept 3,652945 0,141021 25,903557 0,000126 3,204153 4,101737 Price (ln) -1,157525 1,755050 -0,659540 0,556659 -6,742876 4,427825
i) Male, Age Group 2 Regression Statistics Multiple R 0,567463 R Square 0,322014 Adjusted R Square 0,096019 Standard Error 0,186348 Observations 5,000000 ANOVA df SS MS F Significance F Regression 1,000000 0,049479 0,049479 1,424869 0,318391 Residual 3,000000 0,104177 0,034726 Total 4,000000 0,153656
Coefficients Standard Error t Stat P-value Lower 95% Upper 95%
Intercept 5,068995 0,089388 56,707708 0,000012 4,784522 5,353467
Price (ln) -1,327923 1,112463 -1,193679 0,318391 -4,868275 2,212429
j) Male, Age Group 3 Regression Statistics Multiple R 0,529962 R Square 0,280860 Adjusted R Square 0,041146 Standard Error 0,180672 Observations 5,000000 ANOVA df SS MS F Significance F Regression 1,000000 0,038245 0,038245 1,171647 0,358303 Residual 3,000000 0,097927 0,032642 Total 4,000000 0,136173
Coefficients Standard Error t Stat P-value Lower 95%
Upper 95%
Intercept 4,328803 0,086665 49,948443 0,000018 4,052995 4,604611
k) Male, Age Group 4 Regression Statistics Multiple R 0,242227 R Square 0,058674 Adjusted R Square -0,255101 Standard Error 0,401737 Observations 5,000000 ANOVA df SS MS F Significance F Regression 1,000000 0,030179 0,030179 0,186994 0,694630 Residual 3,000000 0,484177 0,161392 Total 4,000000 0,514357 Coefficients Standard
Error t Stat P-value Lower 95%
Upper 95% Intercept 3,916760 0,192707 20,324993 0,000260 3,303481 4,530038 Price (ln) -1,037089 2,398293 -0,432428 0,694630 -8,669526 6,595349
l) Female, Age Group 1 Regression Statistics Multiple R 0,487823 R Square 0,237971 Adjusted R Square -0,016038 Standard Error 0,070335 Observations 5,000000 ANOVA df SS MS F Significance F Regression 1,000000 0,004635 0,004635 0,936860 0,404483 Residual 3,000000 0,014841 0,004947 Total 4,000000 0,019476
Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Intercept 4,693769 0,033739 139,121380 0,000001 4,586398 4,801141 Price (ln) -0,406416 0,419888 -0,967916 0,404483 -1,742687 0,929855
m) Female, Age Group 2
n) Female, Age Group 3 Regression Statistics Multiple R 0,387433 R Square 0,150104 Adjusted R Square -0,133194 Standard Error 0,255355 Observations 5,000000 ANOVA df SS MS F Significance F Regression 1,000000 0,034549 0,034549 0,529846 0,519340 Residual 3,000000 0,195618 0,065206 Total 4,000000 0,230167 Coefficients Standard
Error t Stat P-value Lower 95%
Upper 95% Intercept 4,056777 0,122489 33,119417 0,000061 3,666961 4,446593 Price (ln) -1,109632 1,524418 -0,727905 0,519340 -5,961011 3,741748 Regression Statistics Multiple R 0,870797 R Square 0,758288 Adjusted R Square 0,677717 Standard Error 0,120522 Observations 5 ANOVA df SS MS F Significance F Regression 1 0,136706 0,136706 9,411452 0,054656 Residual 3 0,043576 0,014525 Total 4 0,180282
Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Intercept 4,803487 0,057812 83,087680 0,000004 4,619503 4,987471931 Price (ln) -2,207263 0,719492 -3,067809 0,054656 -4,497006 0,082480634
o) Female, Age Group 4 Regression Statistics Multiple R 0,258989 R Square 0,067075 Adjusted R Square -0,243900 Standard Error 0,400644 Observations 5,000000 ANOVA df SS MS F Significance F Regression 1,000000 0,034622 0,034622 0,215693 0,673970 Residual 3,000000 0,481548 0,160516 Total 4,000000 0,516170
Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Intercept 3,152809 0,192183 16,405284 0,000493 2,541198 3,764420 Price (ln) -1,110804 2,391771 -0,464428 0,673970 -8,722486 6,500878 p) BMI 1 Regression Statistics Multiple R 0,787390 R Square 0,619983 Adjusted R Square 0,493310 Standard Error 0,455327 Observations 5,000000 ANOVA df SS MS F Significance F Regression 1,000000 1,014716 1,014716 4,894379 0,113855 Residual 3,000000 0,621968 0,207323 Total 4,000000 1,636683
Coefficients Standard Error t Stat P-value Lower 95%
Upper 95% Intercept 4,932047 0,218413 22,581302 0,000190 4,236960 5,627134 Price (ln) -6,013575 2,718216 -2,212324 0,113855 -14,664150 2,637000
q) BMI 2 r) BMI 3 Regression Statistics Multiple R 0,600458 R Square 0,360550 Adjusted R Square 0,147400 Standard Error 0,097967 Observations 5,000000 ANOVA df SS MS F Significance F Regression 1,000000 0,016235 0,016235 1,691532 0,284290 Residual 3,000000 0,028793 0,009598 Total 4,000000 0,045027 Coefficients Standard
Error t Stat P-value Lower 95%
Upper 95% Intercept 4,500963 0,046993 95,778849 0,000003 4,351410 4,650517 Price (ln) -0,760644 0,584846 -1,300589 0,284290 -2,621885 1,100597 Regression Statistics Multiple R 0,947482 R Square 0,897723 Adjusted R Square 0,863630 Standard Error 0,060136 Observations 5,000000 ANOVA df SS MS F Significance F Regression 1,000000 0,095227 0,095227 26,332027 0,014333 Residual 3,000000 0,010849 0,003616 Total 4,000000 0,106076 Coefficients Standard
Error t Stat P-value Lower 95%
Upper 95% Intercept 4,668122 0,028846 161,826710 0,000001 4,576320 4,759924 Price (ln) -1,842213 0,359003 -5,131474 0,014333 -2,984720 -0,699706
s) BMI 4 Regression Statistics Multiple R 0,011513 R Square 0,000133 Adjusted R Square -0,333157 Standard Error 0,144931 Observations 5,000000 ANOVA df SS MS F Significance F Regression 1,000000 0,000008 0,000008 0,000398 0,985342 Residual 3,000000 0,063015 0,021005 Total 4,000000 0,063023 Coefficients Standard
Error t Stat P-value Lower 95%
Upper 95% Intercept 4,640710 0,069521 66,752671 0,000007 4,419463 4,861956 Price (ln) -0,017254 0,865210 -0,019942 0,985342 -2,770737 2,736230
Appendix 3: Euromonitor statistics
Retrieved from http://www.portal.euromonitor.com.rps.hva.nl:2048/portal/statistics/tab
Appendix 4: Differences among SSBs
Types of sugar-sweetened beverage (SSB) Description
Soft drink Non-alcoholic, flavored, carbonated or
non-carbonated beverage
Fruit drinks, excluding 100% fruit juice Sweetened beverage of dilute fruit juice Ready to drink (RTD) coffee and tea Coffee and tea that include caloric
sweeteners
Sport drink Beverage that includes sugar and
electrolytes
Energy drink (carbonated) drink that contains high
amount of sugar and caffeine
Sweetened milks Milk that contains sweetened power or syrup
Appendix 5: Eurostat report
Retrieved from
