The consumption impacts of imposing excise tax on sugar-sweetened beverages in Vietnam

In order to establish a demand system with eight F&B items taken into account, comprising sugar, candy, fresh milk, wine, beer, soft drinks, powered coffee and dried t[r]

VIETNAM NATIONAL UNIVERSITY, HANOI

VIETNAM JAPAN UNIVERSITY

LUONG NGOC MY LINH

THE CONSUMPTION IMPACTS OF

IMPOSING EXCISE TAX ON

SUGAR-SWEETENED BEVERAGES IN VIETNAM

MASTER'S THESIS

………

Master of Public Policy

Hanoi, 2019

VIETNAM NATIONAL UNIVERSITY, HANOI

VIETNAM JAPAN UNIVERSITY

LUONG NGOC MY LINH

THE CONSUMPTION IMPACTS OF IMPOSING EXCISE TAX ON SUGAR- SWEETENED BEVERAGES IN VIETNAM

MAJOR: MASTER OF PUBLIC POLICY

i

ACKNOWLEDGEMENT

Two years in Vietnam Japan University has come and gone It is my pleasure to become a part of second intake students of this university, where I had a chance to meet many interesting and unique friends as well as many enthusiastic and excellent professors Thanks to Vietnam Japan University, I was able to experience a totally different academic environment from other school, which is worth to spend my time and challenge myself

First, I would like to express my sincere gratitude to my supervisors, Assoc.Prof Kaida Naoko from University of Tsukuba and Dr Vu Hoang Linh from Vietnam Japan University for their enthusiasms, patience and support from the beginning to finishing of my master thesis With their numerous years of experiences and profound knowledge, both supervisors have always given me helpful advices whenever I crashed into a dead end

My honest appreciation goes to JICA, VNU, VJU, University of Tsukuba for giving

me great opportunity to study in VJU and experience three-month internship in Japan

I also would like to say thanks to all staff in VJU, especially in Program of Master of Public Policy and to the staff in Office of Global Initiatives in University of Tsukuba for their timely assistance and making my study time memorable

Besides this, I want to express my thankfulness to all second intake students, especially to become a part of my student life Hope that everyone will be successful and be happy in the future

Finally, it will be missing unless mentioning on the encouragement and support of

my parents, elder sister and the caring of my niece and nephew and lots of my friend despite of my ignorance when studying in VJU

ii

TABLE OF CONTENTS

ACKNOWLEDGEMENT i

LIST OF TABLES iv

LIST OF FIGURES v

LIST OF ABBREVIATIONS vi

CHAPTER 1: INTRODUCTION 1

1.1 Research background 1

1.2 Research Objective and Research Question 3

CHAPTER 2: LITERATURE REVIEW 4

2.1 An overview of sugar-sweetened beverage 4

2.2 Taxing sugar-sweetened beverages 5

2.3 Discussion over excise tax on SSBs 8

2.4 Consumption Impacts of Excise tax around the world 10

2.5 Previous Studies about Sugar Sweetened Beverages in Vietnam 15

2.6 Econometric model to estimate demand function 17

2.6.1 The Stone’s system 18

2.6.2 The Rotterdam models 19

2.6.3 Working-Lesser model 19

2.6.4 The Almost Ideal Demand System (AIDS) 20

CHAPTER 3: METHOD AND METHODOLOGY 23

3.1 Paradigm 23

3.2 Data Collection 25

3.2.1 An introduction of Vietnam Household Living Standard Survey 2016 25 3.2.2 Data Processing 27

3.3 Data Analysis 30

3.3.1 Overview of SSB consumption and SSB consumer 30

3.3.2 Expenditure, compensated and uncompensated elasticity 30

3.3.3 The differences in elasticity between different household patterns 31

iii

3.3.4 Welfare effect 32

CHAPTER 4: RESULTS 34

4.1 Sugar-sweetened beverages consumption in Vietnam 34

4.2 AIDS results 35

4.3 The impacts of household characteristics on SSB consumption 38

4.3.1 Urban and rural area 38

4.3.2 Majority and minority 39

4.3.3 The number of children 40

4.3.4 The expenditure quintile 42

4.3.5 Six regions in Vietnam 43

4.3.6 Regular and holiday consumption 44

4.4 Welfare effects 46

4.5 Discussion 48

CHAPTER 5: CONCLUSION 52

5.1 Policy implication 52

5.1.1 Imposing tax on SSB in combination with encouraging a healthy lifestyle 52 5.1.2 Learning from international experiences and pay attention to some specific groups 54

5.2 Further Study 54

5.3 Limitations 55

REFERENCES 56 ANNEX 1: LIST OF COUNTRIES THAT ARE APPLYING SPECIFIC TAX ON SUGAR-SWEETENED BEVERAGE

ANNEX 2: DATA PROCESSING SUMMARIZE

iv

LIST OF TABLES

Table 2.1 Controversy over SSB excise tax 8Table 3.1: Information about Vietnam Household Living Standard Survey 2016 25Table 3.2 The sample of VHLSS 2016 27Table 4.1 Elasticity result 36Table 4.2 Compensating Variation by categories if applying 10% excise tax on SSB 46

v

LIST OF FIGURES

Figure 1.1: Government budget and public debt in Vietnam 2011 – 2016 3

Figure 2.1 Health impacts of obesity 5

Figure 2.2 Pathway for the effect of fiscal policy intervention 6

Figure 2.3 The 5-tier health impact pyramid 8

Figure 2.4: Map of cities and countries that are implementing taxing sugary drinks 11

Figure 4.1 The price distribution of SSB consumed by households in VHLSS 2016 34

Figure 4.2 SSB consumption in Vietnam 2008 – 2016 through VHLSS 35

Figure 4.3 SSB’s expenditure, compensated and uncompensated elasticity be separated by urban and rural area 39

Figure 4.4 SSB’s expenditure, uncompensated and compensated elasticity separated by major or minor ethnicity 40

Figure 4.5 The SSB’s expenditure, uncompensated and compensated elasticity be separated by the number of children in the household 41

Figure 4.6 The SSB’s expenditure, uncompensated and compensated elasticity be separated by the quintile of household 42

Figure 4.7 The SSB’s expenditure, uncompensated and compensated elasticity be separated by six regions in Vietnam 43

Figure 4.8 The SSB’s expenditure, uncompensated and compensated elasticity be separated by time of consuming 45

Figure 5.1 The number of hours that Vietnamese spend on doing exercise 53

IH Income household MOF Ministry of Finance NCD Non-communication disease OLS Ordinary least squares

O&O Overweight and obesity PHPT Public health product tax SSB Sugar-sweetened beverage AIDS The Almost Ideal Demand System VHLSS Vietnam Household Living Standard Survey

1990 (World Bank, 2002) Parallel with successful economic achievements, Vietnam

is also facing to some problems such as the increasing of the prevalence of O&O, particularly among children and adolescent in urban areas According to a survey conducted by Ministry of Health, the ratio of Vietnamese adults (18 to 69 years old) who are suffering from O&O (Body Mass Index – BMI ≥ 25) is 15.6%, in which14.9% of men and 16.9% of women) The situation seems worse than 5 year before when these ratios in 2010 were 12.5% of men and 11.4% of women In

2

addition to that, Vietnam Health Scheme has set the target for the percentage of adult who suffers from O&O under 12% until 2025 and under 10% until 2030 (Vietnam Prime Minister, 2019), which means that Vietnamese authority has a lot of things to

do to deal with this situation The rising trend was also observed among children and adolescents, with the prevalence of O&O among both genders increased from 2.6%

in 2002 to 9.7% in 2016 (The Department of Preventive Medicine, 2017) In big cities such as Hanoi and Ho Chi Minh city, the ratio of children and adolescents who are suffering from O&O was 43% in 2009, increased from 12% in 1996 In 2014 – 2015, this ratio in the inner area of Ho Chi Minh city was 50% and a recent survey in the inner area of Hanoi showed that it was up to 40.7% (Hanoi Health Department, 2017) Although the average BMI of Vietnamese was still far less than O&O threshold, there are existing a concern that Vietnam may reach this ceiling soon

As an effort in improving Vietnamese health condition, Ministry of Finance (MOF) had introduced a new excise tax applied on SSBs with the tax rate at 10% or 20% in

2016 According to Lax on Excise tax of Vietnam, excise tax is a tax that is applied

on specific goods and services that (i) are not encouraged due to their bad effects on health; (ii) are normally consumed by high-income consumers in the society; (iii) are unnecessary services (Vietnamese Assembly, 2008) The rationale for levying excise tax was its high-containing sugar amount in each item, which is one of the reasons for O&O in Vietnam and around the world Added to that, imposing excise tax on SSBs is a new trend in the world and being strongly encouraged by international health organizations, such as WHO This was a part of Six-Tax-Law Amendment in order to reform tax system in Vietnam The tax law was introduced in the context of increasing government budget deficit and high public debt ratio in Vietnam, which was reflected by figure 1.1 Therefore, this reforming scheme was expected to increase government revenue and solve these problems

3

Figure 1.1: Government budget and public debt in Vietnam 2011 – 2016 Data for

the Ministry of Finance (2012, 2013a, 2013b, 2013c, 2014a, 2014b, 2015, 2016a, 2016b, 2017a, 2017b)

1.2 Research Objective and Research Question

The objective of this paper is to estimate the consumption impacts of imposing SSBs’ excise tax in Vietnam It will focus on the main research question “How does excise tax on SSB impact on Vietnamese consumers?” and below sub-questions:

▪ To what extent excise tax on SSBs influence on consumers of this kind of product?

▪ If then this tax policy is implemented, what should Vietnamese government pay attention to, from perspective of consumption?

4

CHAPTER 2: LITERATURE REVIEW

2.1 An overview of sugar-sweetened beverage

According to World Health Organization (WHO), free sugar, which is usually added

to foods and beverages or naturally contained in honey, syrups and fruit juice and its relative products, includes monosaccharides and (C6H12O6) and disaccharides (C12H22O11) It has been proved that higher intake of free sugar provides significant energy to diets without specific nutrients, which may harm to the balance of their intake energy That is one of the reasons for O&O in recent years In addition, free sugar in foods and beverages is also related to dental caries Therefore, the intake of free sugar should be limited to 10% of total energy intake in near future and to less than 5% in further (World Health Organization, 2015)

SSBs include, but are not limited to sport drinks, energy drinks, lemonade or fruit drinks, sweetened iced tea, vitamin water drinks and carbonates drinks, which contain added caloric sweeteners, for example, sucrose, high-fructose corn syrup (HFCS), which is produced from corn syrup (glucose) or fruit-juice concentrates It is said that SSB is one of the highest free sugar containing products In case of Vietnam, the governmental authority has not given any official definition of SSB However, according to WHO, SSB referred to any beverage item containing added sugars, for example sucrose of high fructose corn syrup (Lobstein, 2014) This ingredient will create a sweet flavor to beverages and satisfy consumers’ taste

In combination with insufficient physical activities, SSBs could causes to obesity and diabetes because it leads to the excess of intake calorie, particularly to children, adolescents and adults (World Health Organization, 2016) Furthermore, obesity will affect to human lives and health on different dimensions, physical health, mental health as well as social and spiritual aspects as figure 2.1 (Djalalinia, Qorbani, Peykari,

& Kelishadi, 2015) That is the rationale for governmental intervention into food and beverage industry

5

Figure 2.1 Health impacts of obesity Reprinted from Health impacts of obesity by

Djalalinia, Qorbani, Peykari, & Kelishadi (2015)

2.2 Taxing sugar-sweetened beverages

There are two main kinds of policy tools that government can utilize to improve diet, they are taxation and subsidies By taxing on unhealthy food and subsiding for healthy food, consumer will be encouraged to gradually change their consumed baskets to healthier products Then, it is hoped that this transformation can make significant effects on food industry and society The mechanism of how tax and subsidy on food and beverage effect on consumption was expressed by figure 2.2

•Lower spiritual well-being

•Impaired eating behaviors

6

Figure 2.2 Pathway for the effect of fiscal policy intervention Reprinted from Fiscal Policies for Diet and Prevention of Noncommunicable Diseases: Technical Meeting Report by World Health Organization (2016)

In relation to SSBs, many countries are applying tax on this kind of products There are two types of tax: ad-valorem and specific tax Many countries are imposing an ad-valorem tax on SSB, which means that the tax rate is calculated as a percentage of product value As a result, the price of SSB will be relatively higher than other products Whereas, a specific tax is fixed at a constant by unit For example, in Mexico, tax on SSB is 10 Mexico dollars by unit In general, both kind of taxes aim

to increase relative price of SSBs in expectation to decrease their consumed volume (World Health Organization, 2018)

Increased purchase and consumption od

healthy foods

Decreased purchase and consumption

of less healthy food

Financial implications for manufactures and producers – incentive for decreased production and reformulation

Decreased related risk factors for

diet-NCDs

Financial implications

for manufactures and

producers – incentive for

7

Rationale for taxing SSBs is related to their negative impacts on health as mention above To be more detailed, Lustig and his colleagues have pointed out four characteristics of SSBs that are similar to tobacco and alcohol that make them to be taxed and regulated Firstly, it is said that sugar is unavoidable, but rather than natural-made, man-made sugar is becoming more popular than ever before and is added to nearly all processed food Secondly, the toxicity of sugar has been mentioned by giving that excessive sugar consuming induce disease related to metabolic syndrome such as hypertension, high triglycerides and insulin resistance, diabetes and aging process Added to that, it is doubted that it also links to cancer and cognitive decline Thirdly, sugar is potential for abuse because it signals brain to consume more and make impacts on the feeling of satiety Finally, as consequence of healthcare and human costs of metabolic syndrome, consuming excessive amount of sugar possibly cause bad effects on society (Lustig, Schmidt, & Brindis, 2012) From perspective of public health, there are several ways to intervene into choices of consumers The figure 2.3 shows a framework to conduct a public health campaign The 5-tier health impact pyramid describes how different in the impacts of public health intervention methods and can be a framework for public action that related to improve public health The first tier of this pyramid is socio-economic factors, which refers to address socio-economic status and educational status that are important determinants of health The second tier of pyramid is to change the context to encourage healthy decision, which suggests the interventions to make the healthy choice become the default choices of citizens, without concerning on societal factors like education or income The third one mention on long-lasting protective interventions, which no longer reaches people collectively but is an individual choice The next level is clinical interventions and the highest is counselling and educational interventions The interventions belong to lower level of pyramid are said to be more effective in changing citizens’ habits due to wide range of impacts, whereas the higher levels are depended on individual decisions and require more personal autonomy, of which people are normally lack (Frieden, 2010) The SSB excise tax belongs to

by high-income consumers; and (c) are unnecessary services (Vietnamese Assembly, 2008) According to proposal by Ministry of Finance (MOF), an ad-valorem tax at

10 or 20% should be applied on SSBs because of their bad effects on health On the view of MOF, they prefer the first plan (The Ministry of Finance, 2017c)

2.3 Discussion over excise tax on SSBs

Despite of positive policy objectives, this tax policy has raised fierce controversy and dialogue not only among researchers, policy makers but also community around the world Both sides, approval and opposition, have their own opinions to protect their arguments against the another and Barry, Niederdeppe, & Gollust (2013) In the research in 2017, the author based on the opposing arguments to SSB excise tax given

by Autralian Beverages Council, which is a lobby group for sugary beverage

Counseling and educationClinical interventionLong-lasting perspective InterventionChanging the context to encourage

healthy decisionSocio-economic factors

9

enterprises, to counter This organization tried to keep the tax away from policy table,

so the author of the research had to establish the counterpoints for the opposing arguments, depended on the actual SSB consumption situation in Australia Another public opinion survey conducted by Barry, Niederdeppe, & Gollust showed that American citizens tend to agree with pro-tax than anti-tax arguments because they believed that the tax would not affect to the consumption of unhealthy products (60%), the tax was just a way to fill in the government budget (58%) The majority of them also thought that it was unacceptable way for government to intervene into their lives (53.8%), would be opposed by most of American (53%) and had bad effects on the poor (51%) However, some of them said that the tax contributed largest to O&O (49%) and helped to raise budget for prevention this NCD (41%) due to the severence

in the suffering of O&O in this country

Table 2.1 summarizes the key arguments regarding to SSB excise tax around the

world, which were summarized by Backholder and Martin (2017) and Barry, Niederdeppe, & Gollust (2013) In the research in 2017, the author based on the opposing arguments to SSB excise tax given by Autralian Beverages Council, which

is a lobby group for sugary beverage enterprises, to counter This organization tried

to keep the tax away from policy table, so the author of the research had to establish the counterpoints for the opposing arguments, depended on the actual SSB consumption situation in Australia (Backholer & Martin, 2017) Another public opinion survey conducted by Barry, Niederdeppe, & Gollust showed that American citizens tend to agree with pro-tax than anti-tax arguments because they believed that the tax would not affect to the consumption of unhealthy products (60%), the tax was just a way to fill in the government budget (58%) The majority of them also thought that it was unacceptable way for government to intervene into their lives (53.8%), would be opposed by most of American (53%) and had bad effects on the poor (51%) However, some of them said that the tax contributed largest to O&O (49%) and helped to raise budget for prevention this NCD (41%) due to the severence in the suffering of O&O in this country (Barry, Niederdeppe, & Gollust, 2013)

10

Table 2.1 Controversy over SSB excise tax

Note: Reprinted from Commentary: Sugar-sweetened beverage tax: The inconvenient truth by Backholer & Martin (2017) and Taxes on Sugar-Sweetened Beverages:

• Reason of O&O is not only SSB

• Free sugar consumption is declining • SSB is the largest contributor to O&O • SSB encourages people to consume

more

• Quick way to fill in budget holes • Raising government revenue

• Unacceptable government intrusion

into personal lives and individual

• The tax policy is failed in some

countries

• Successful in some countries

• Take time to measure its long-term effects

• Regressive tax and hurt the poor

• Hurt the economy • Equalize the cost between unhealthy

and healthy drinks

• SSB industry has supported to

reduce O&O by calorie information

on packages, removing SSB from

school in some countries

• Tax on SSB will result more

• Job lost • Increase the production in non-sugar

drink industry

• The reinvestment of revenue will create job in other sectors

• Long-term effects

11

Result from a 2011 National Public Opinion Survey by Barry, Niederdeppe, &

Gollust (2013)

Theoritically, the tax policy had both positive and negative impacts on the

socio-economy, which had been foreseen in many researches However, these impacts were different depending on the characteristics of each region or country

2.4 Consumption Impacts of Excise tax around the world

Officially, it was been recorded that tax on soft drinks like SSB had been placed since 1930s in Denmark (The Department of Health, Social Services and Public Safety of Northern Ireland, 2015) Taxing SSB is now becoming popular around the world due

to its negative effects on health, especially among countries which are having severe O&O According to a Research Program on global food, there are 41 countries and territories are now issued this regulation, which will be showed by figure 2.4 To be

more detailed, the effective date, tax rate and taxed items in each country was showed

in Annex 1

Figure 2.4 Map of cities and countries that are implementing taxing sugary drinks

Reprinted from Sugary drink taxes around the world by Global Food Research

Firstly, it has been observed and demonstrated that the tax policy makes the price increase which lead to a decrease in the consumption

The case of Mexico should be a typical example for this outcome, which was proved out by few researches For example, a research by Grogger in 2015, which was focused on the price changes in post-tax period, had found out that after implementing, the tax policy had affected on the price of regular sodas and other sugary sodas by different levels To regular sodas, the tax took effect let the prices rise more than the amount of the tax whereas the prices or others sugary beverages rise less than that amount [Grogger, 2015] The increasing trend in price were agreed by a research team of World Bank However, this group simultaneously claimed that the prices were not high enough to move up prices of the whole market [Bonilla-Chacin, Iglesias, Suaya, Trezza, Macias, 2016] By comparing the observed consumed volume in post-tax period with the estimated volume in case of not issuing the tax policy, the observational study by Colchero et al after one-year implementation has shown that purchases of taxed beverages in 2014 had decreased by 6% in average at an increasing rate up to 12% by December 2014 The decreasing trend had been detected among three socio-economic groups, but the lower status group was affected more than the others, with an average at 9% decline in 2014 and up to 17% by December 2014

13

(Colchero, Popkin, Rivera, & Ng, 2016) Also focusing on price-change effects of the tax in Mexico, aimed to two different group of items, high caloric foods (HCF) and/or sugary drinks, Aguilar et al have found that the price of sugary drinks had increased but more than 100 percent of tax had passed through on consumer prices After 2 years of implementation, they found that the purchased liters of sugary drinks decreased by 6% (Aguilar, Emilio, & Seira, 2016)

In France, in a study of Nutrinet for the General Directorate of Health in 2013, soda consumption has declined (World Health Organization, 2016) In Hungary, the country has already implemented a public health product tax (PHPT) since 2011, prices had increased and partly led to 26-32% of consumers who decrease the intake

of subjected items (Martos, et al., 2015) In Berkeley (California), one of the first cities in America issued SSB tax, after four-month of implementation, the consumption of SSB had decreased 21% (Falbe, et al., 2016)

Secondly, there is a shift from SSB to healthier items A study by Martos et al in

2015 has shown that due to public health in Hungary, a lot of consumer choose to substitute their choices by healthier products, for example, mineral water accounted for 63%, fruity and vegetable drinks accounted for 82 – 86%, do-it-yourself sweet drinks accounted for 95%, and green herbs or spices accounted for 84% (Martos, et al., 2015) In Mexico, per capital consumption of bottled water increased (World Health Organization, 2016) In America, after four-month of implementation, the amount of water consumption in Berkeley raised more than compared cities, +63%

in comparison with +19% (Falbe, et al., 2016)

Thirdly, the tax income contributes to government budget For example, in 2014, the tax help French raise €300 million (World Health Organization, 2016) In case of Hungary, SSBs contributed the highest proportion (65%) to the reported tax HUF 61.3 billion (approximately €200 million) of PHPT The second and the third is pre-packaged sweets and energy drinks with the proportion of 19% and 8%, respectively (Martos, et al., 2015)

14

Fourthly, the tax policy force food producers to reformulate their products For instance, the response of consumers to government’ public health scheme had made 40% of food producers reformulate their products, 30% of them remove and 70% of them decrease unflattering ingredients (Martos, et al., 2015)

Fifthly, the tax affected on targeted groups For example, in France, the change in behaviors regarding to the tax had been shown clearly among young people, households that has adolescents ang lower income group (World Health Organization, 2016) In Hungary, overweight and adolescent adult were more likely to change their behavior regarding to targeted products (1.8 – 2.7 times) in a positive way than the others (Martos, et al., 2015) In case of Mexico, the effects on different social status households were varied Aguilar and his colleagues proved that households with higher income were affected most (Aguilar, Emilio, & Seira, 2016) while Colchero

et al said that household in low level of socio-economy have been affected largest (Colchero, Popkin, Rivera, & Ng, 2016)

Besides above positive results, there were several unexpected impacts of the tax had been recorded Despite of aiming to reduce the prevalence of O&O, there is nothing change in BMI and even higher after implementing the tax policy This fact has been claimed in Mexico (Aguilar, Emilio, & Seira, 2016) or Finland (Northern Ireland Assembly, 2015)

In Denmark, the authorities even had to abandon their “fat tax policy” due to several disadvantaged consequences The tax policy was said to be ineffective because it was not enough to make consumers change their consumed habits They could easily switch to cheaper products or do cross-shopping Added to that, the administrative cost was too high (10% of the tax revenue) and even led to lost 1,300 jobs (Snowdon, 2013)

Several studies have tried to estimate the own price elasticities of SSB in order to measure the impacts of SSB excise tax on consumption According to a systematic review on SSB excise tax in middle income countries like Vietnam in the period of

15

1990 - 2016, including Brazil, Ecuador, India, Mexico, Peru and South Africa, the author group had found out that the own elasticities of SSB in these nations varied from -0.6 to -1.2, which meant that when the price of SSB increased by 1%, the consumed amount would decline by 0.6 to 1.2% Moreover, the research also pointed out that whereas milk is a substitute beverage of SSB, food away from home (FAFH), candy and snack were complementary to SSB (Nakhimovsky, et al., 2016) Another systematic review by Andreyeva, Long and Brownell (2010) collected data from 160 studies related to price elasticities of foods and beverages in America from 1938 to

2010 showed that soft drinks have the elasticity from -0.8 to -1.0 (Andreyeva, Long,

& Brownell, 2010) The, several studies had also estimated the own price elasticity

in American in different household group Lin et al (2011) estimated that own-price elasticity of low-income group in USA (-0.95) were lower than high-income group (-1.29) (Lin, Smith, Lee, & Hall, 2011) In UK, Brigg et al (2013) measured elasticity for low, middle and high-income group by -0.79, -0.80 and -0.85, respectively (Briggs, et al., 2013)

2.5 Previous Studies about Sugar Sweetened Beverages in Vietnam

Due to bad effects of sugary drinks on health, MOF had introduced excise tax on this kind of products since 2014, but this proposal only covered non-alcoholic carbonated sugary drinks Finally, because of strong oppositions of researchers, enterprises and citizens, the implementation of this tax policy had been postponed One of the most important bases for those protest was the research conducted by Nguyen Dinh Chuc (2014) named “Soft drink industry and the economic impact of excise tax on non-alcoholic carbonated soft drink in Vietnam” In this research, Chuc had approached the issue from two directions – elasticity approach and general equilibrium approach From elasticity approach, he tried to estimate price elasticity of SSB through the equation 1:

ln 𝑌 = 𝛽1+ 𝛽2ln 𝑃 + 𝛽3ln 𝑋1+ ⋯ + 𝛽𝑚𝑙𝑛 𝑋𝑛+ 𝛼𝑘𝑙𝑛(𝑋𝑝𝑋𝑞) + 𝑒 (1)

𝑌𝑆𝑆𝐷 = −2.81𝑃𝑆𝑆𝐷 + 0.0009𝑑𝐺𝐷𝑃 + 3.98𝑃𝐸𝐷 − 0.96𝑃𝐹𝐽 − 1.98𝑃𝑃𝑊

In which, 𝑑𝐺𝐷𝑃, 𝑃𝐸𝐷, 𝑃𝐹𝐽, 𝑃𝑃𝑊, 𝑃𝑆𝑀, 𝑃𝑃𝑆𝑆𝐷𝐸𝐷 represents the growth of GDP, price

of energy drink, fruit juice, packaged water, soya milk and the interaction between its own price and price of energy drink, respectively The result showed that carbonated soft drinks were sensitive to price with its price elasticity at −2.8, which means that increasing 1% in price of this drink would lead to decrease 2.8% in demanded volume From general equilibrium approach, he estimated the impact of the tax policy on the whole economy According to the results, taxing 10% on SSBs would results in 12.1 million dollars decrease in GDP, 8.46 million dollars increasing in government revenue, whereas the soft industry revenue might loss 40.5 million dollars On economical aspect, it showed that this tax policy had both negative and positive impacts on Vietnamese economy Besides measurable impacts, there might have some other negative consequences, employment effect, income effect, influences on other drinks and unknown administrative cost, for instance (Chuc, 2014)

In 2016, excise tax on unhealthy drinks was re-proposed but wider coverage to all sugary drinks MOF has proposed two alternatives, imposing excise tax on sugary drink at 10% or 20% tax rate However, it was more partial to the first option For the purpose of “Economic assessment of the possibility to apply the excise tax on non-alcoholic beverage products in Vietnam”, Phong, T N and Thai, P T has conducted

a research in 2017, in which they had used The Almost Ideal Demand System (AIDS)

to estimate owned-price elasticity of SSB in Vietnamese market, which had the formation as equation 3:

17

𝑤𝑖 = 𝛼𝑖+ ∑ 𝛾𝑖𝑗ln(𝑝𝑗)

𝑛 𝑗=1

2.6 Econometric model to estimate demand function

Estimating demand function had always been an attractive topic among economists because demand analysis was a comprehensive way to approach to consumer choice The main objective of this approach was fundamentally to measuring pure numbers so-called elasticities, which were used to evaluate the reaction of consumers to a change in affecting factors like income or prices However, it had never been an easy

18

work due to complicated technique and strict restrictions, including adding-up, restriction, homogeneity restriction and symmetry restriction Basically, there were two forms of a demand function, Marshallian demand function and Hicksian demand function In Marshallian demand function, a giving quantity (𝑞𝑖) was a function of prices (𝑝) and total expenditure (𝑥) as equation 4, which meant that it reflected the relationship between consumed quantity and its price (Deaton & Muellbauer, 1980b):

Whereas, the Hisksian demand function was seemed quantity (𝑞𝑖) as a function of utility (𝑢𝑖) and price (𝑝) In which, utility was a function of prices (𝑝) and total expenditure (𝑥), so Hicksian demand function could be written as equation 5 (Deaton

& Muellbauer, Economics and consumer behavior, 1980b):

𝑞𝑖 = ℎ𝑖(𝑢, 𝑝) = ℎ𝑖(𝑢(𝑥, 𝑝), 𝑝) (5) Despite of simple definition, to establish a demand function was a challenge to any researcher due to its restriction as named above Generally, the econometric studies

of demand might be a single equation or a system of demand equations The most basic demand function was linear demand function which simply estimated the relationship between quantity and price and other independent variables However, it was obviously that this function was hard to satisfy the adding-up restriction

2.6.1 The Stone’s system

In 1954 Richard Stone firstly introduced his achievement in estimating demand function The function had the form as equation 6:

log 𝑞𝑖 = 𝛼𝑖+ 𝑒𝑖log𝑥

𝑃+ ∑ 𝑒𝑖𝑘

∗ 𝑛 𝑘=1

log𝑝𝑘

In which P is the price index, calculated by equation 7

19

log 𝑃 = ∑ 𝑤𝑘

𝑘

It was easily seen that 𝑒𝑖 was the total expenditure elasticity The compensated

elasticity of item i in respect to item k was calculated as equation 8 (Stone, 1954):

Despite of satisfying some restriction, actual estimation results showed that there existed conflict in the relationship between different food items For example, food item A might be substitution of B, but B was complementary product of A (Deaton

& Muellbauer, Economics and consumer behavior, 1980b) Therefore, the most popular application of this system was so-called the Stone index, which was utilized

to solve the problem of nonlinearity of inherited systems

2.6.2 The Rotterdam models

The Rotterdam model was developed by Theil (1965) and Barten (1969), which used the similar approach as Stone but rather than using the function of logarithms, this model used differentials as equation 9 (Theil, 1965):

𝑑(log 𝑞𝑖) = 𝑒𝑖(𝑑(log 𝑥) − ∑ 𝑤𝑘

𝑘

𝑑(log 𝑝𝑘)) + ∑ 𝑒𝑖𝑗∗

𝑗𝑑(log 𝑝𝑗) (9)

In which, 𝑤𝑘 was the budget share of good k If we multiplied this function to 𝑤𝑖, it would become equation 10:

2.6.3 Working-Lesser model

The original form of Working-Lesser model expressed budget share of a specific item

by a linear function of the logarithm of total expenditure and price index of all products that were under consideration (Working, 1943; Leser, 1963) Despite of many different versions, they were all based on and developed from the most basic model, which was written as equation 14

Whereas, 𝐻𝑘 referred to household characteristics or demographic factors, which were used to evaluate the effects of these factors to real expenditure Using this model,

it could be easy to measure cross-price elasticity and expenditure elasticity of each i

item To estimate this model, ordinary least squares (OLS) was normally utilized This model had become one of the basic foundations for other models and demand systems later However, because Working-Lesser model was a single equation model,

it might not satisfy the restrictions of demand function if each item is estimated individually

21

2.6.4 The Almost Ideal Demand System (AIDS)

It could be said that AIDS was inherited from its predecessors as mentioned above The system was firstly introduced by Deaton and Muellbauer in 1980 (Deaton & Muellbauer, An Almost Ideal Demand System, 1980a) The cost function of AIDS 𝑐(𝑢, 𝑝), which defined the minimum expenditure level to get a fixed utility level at a given price, was as function 16

(18)

The most important advantage of this system was its “flexibility” and its satisfaction

of aggregation across consumers The AIDS was said to automatically meet with three restrictions of a demand function (Moschini, 1995; Vu, 2008):

• Adding-up restriction which meant that the total value of items should equal

to the total expenditure In AIDS, the dependent variables were the budget share of each item, so the adding-up restriction should be as equation 19a, 19b, 19c

∑ 𝛽𝑖 = 0𝑛

𝑖=1

(19a, b, c)

• Homogeneous restriction (equation 19) – the Hicksian demands are

homogeneous to degree zero

∑ 𝛾𝑖𝑗 = 0𝑗

(20)

• Slutsky symmetry restriction (equation 20), which refers to the symmetry in

the cross-price derivatives of the demands

𝛾𝑖𝑗 = 𝛾𝑗𝑖 =1

Due to this advantage, many researchers were still employing this demand system into their studies after decades However, there were existing concerns referring to the price index 𝑃 because of possibility of the non-linearity once using equation 18 Therefore, Deaton and Muellbauer suggest to us use the Stone index (Deaton & Muellbauer, 1980b), which so-called LA/AIDS In reality, this system were more popular than its original one (Vu, 2008) Nevertheless, the Stone index was said to be invariant to the unit of measurement if it was changed, so it could not be called as

“index” Meanwhile, the change in unit of measurement only affected to the intercept

of system in case of the nonlinearity one (equation 7), which was not significant to the demand properties (Moschini, 1995) Therefore, this research would run the original AIDS model with the function of price index as equation 18

23

CHAPTER 3: METHOD AND METHODOLOGY

This research aimed to estimate the consumption impacts of imposing excise tax on sugar sweetened beverages – which was proposed by Vietnamese MOF but has not been enforced yet due to severe controversy among government authority, scholars

as well as citizens The research focuses on answering the questions: How does excise tax imposed on SSBs impact on Vietnamese consumers?

There are several ways to evaluate the impacts of a tax policy, including both qualitative and quantitative methods This research is trying to apply qualitative method in order to measure clearly these impacts and establish accurate policy implication

3.1 Paradigm

Because of the advantages over other models, the AIDS were still applied popularly

in many researches and this research was also not a bystander However, it is undeniable that the household characteristic will significantly effects on consumption behaviors and choices Therefore, this research would employ Ray’s method (1983)

to give demographics into the system The fundamental concept of Ray was to convert household with different contexts into equivalent units, in term of some reference household, which was called equivalent scale This scale had an important meaning

in welfare economics as well as public policy in terms of equity, justice and resource relocation (Ray, 1983) According to Ray’s method, the cost function of a household

h at a given price level p and utility level u was proportional to the cost function of

representative household 𝑐𝑅 by a general equivalence scale 𝑚̅̅̅̅ as function …, in 0which 𝑐𝑅 shared the same function as function 16

𝑐ℎ(𝑢, 𝑝, 𝑧) = 𝑚̅̅̅̅(𝑧)𝑐0 𝑅(𝑢, 𝑝) (22) This research used maximum-likelihood estimator with the equation 21 of the AIDS, which was incorporated with Ray’s method:

24

𝑤𝑖 = 𝛼𝑖+ ∑ 𝛾𝑖𝑗ln 𝑝𝑗

8 𝑗=1

+ (𝛽𝑖 + 𝜂′𝑧) ln [ 𝑚

𝑚̅0(𝑧)𝑎(𝑝)] + 𝜀𝑖 (23)

In which 𝑖 and 𝑗 run from 1 to k corresponding to k items that were taken into account

In this research, there were 8 items, including sugar (𝑖 = 𝑗 = 1), candy (𝑖 = 𝑗 = 2), fresh milk (𝑖 = 𝑗 = 3) , wine (𝑖 = 𝑗 = 4) , beer (𝑖 = 𝑗 = 5) , bottle beverages (𝑖 = 𝑗 = 6), powered coffee (𝑖 = 𝑗 = 7) and tea (𝑖 = 𝑗 = 2)

The dependent variable (𝑤𝑖) denoted the proportion of consumed value of item i th

over all covered items The independent variable (𝑝𝑖) denoted the average price of

item i th that was consumed by the household And, the equivalent scale 𝑚̅0(𝑧) was calculated as a function of household characteristics 𝑧𝑥 (function 24) In this research, there were 10 household characteristics that would be mentioned in the next part

𝑚̅0(𝑧) = 1 + ∑ 𝜌𝑥𝑧𝑥

10

(24) And ln[𝑎(𝑝𝑖)] was a trans-log function as mentioned in equation 23:

ln[𝑎(𝑝)] = 𝛼0+ ∑ 𝛼𝑖ln 𝑝𝑖

8 𝑖=1

+1

2∑ ∑ 𝛾𝑖𝑗ln 𝑝𝑖ln 𝑝𝑗

8 𝑗=1

8 𝑖=1

(25)

By estimating above system, this research would be able to measure approximately the compensated price elasticity, uncompensated price elasticity and expenditure

elasticity for item i th

In order to estimate the AIDS, this research used the QUAIDS package, which was developed by Poi (2012) This package allowed user to estimate the basic AIDS developed by Deaton and Muellbauer (1980) and the quadratic AIDS model of Banks, Blundell and Lewbell (1997) In addition, user can add to the model several demographic variables to estimate In the post estimation stage, this package can establish a matrix of expenditure elasticity, uncompensated elasticity as well as compensated elasticity (Poi, 2012)

25

3.2 Data Collection

This research used the data collected by GSO every two-year, which is called Vietnam Household Living Standard Survey (VHLSS) The main utilized version is the one conducted in 2016 but other years’ surveys were also processed in order to show the trend of SSB consumption in recent years As the method of collection and procession was the same among years, this part only mentioned about VHLSS 2016

3.2.1 An introduction of Vietnam Household Living Standard Survey 2016

VHLSS was a nation-wide survey in Vietnam that was collected every two years by GSO, in order to inquire information about income, expenditure as well as other living conditions of Vietnamese household The most recent survey with available data was conducted in 2016, following Decision 1095/QD-TCTK on Conducting Household Living Standard Survey in 2016 The detail of VHLSS 2016 was summarized in table 3.1

Table 3.1 Information about Vietnam Household Living Standard Survey 2016

Survey object - Households and members of households

- Communes that had survey households Survey scope Across 63 provinces and cities

Time of survey The VHLSS 2016 was conducted within one month for an area

There were four periods of time: March, June, September and December

Survey period Thirty days or twelve months before survey

Survey method Interview

Survey content To households: Collected information related to their living

standard:

26

- Household income: income and income classified by source (from salary; agriculture, forestry and fishery industry; other production and service industry; other)

- Household expenditure: expenditure and expenditure classified by purpose and by type of expenditure (for meal, clothes, accommodation, commute, education, medicine, culture and other)

- Other characteristics of households and members of households: main demographic characteristics (age, gender, ethnic, marital status); educational level; disease status and medical service using; employment; electric, water and sanitary condition; voluntary and aid scheme participation

To commune: Collect information that affected to household living standard

- Overall demographic and ethnic condition

- Basic socio-economic infrastructure: system of electric, road, school, medical office, market, post office and water source

- Economic situation: agricultural production (land, main crops, production supporting conditions like irrigation and agricultural extension), non-agriculture jobs

- Other information related to social safety, environment, credit and savings

Note: Reprinted from Decision 1095/QD_TCTK on Conducting Household Living Standard 2016 by General Statistics Office (2015)

The VHLSS 2016 had been conducted widely to help authority understand deeply about living condition of their citizens There were 46.995 households being interviewed in 2016, in which 37.596 households being asked about income solely (income household - IH) and other 9.399 households being inquired on income and

27

expenditure information (income and expenditure household – IEH) [General Statistics Office, 2018b] As a requirement of necessary variables, this research would utilize data of 9.399 IEHs The table 3.2 summarized information about the sample of VHLSS 2016

Table 3.2 The sample of VHLSS 2016

Note: Reprinted from Result of the Vietnam Household Living Standards Survey 2016

by General Statistics Office (2018b)

There are eight food items that were under consideration, including sugar, candy, fresh milk, wine, beer, soft drinks, powered coffee and dry tea This research focused

on SSBs, so its main concern was soft drinks, which was defined as all bottled beverages in the market such as carbonated drinks, fruit juice, energy drinks and fresh water, excluding instant tea and coffee Unfortunately, this research could not extract fresh water from other sugary drinks

3.2.2 Data Processing

This research followed three-step procedure in processing data to estimate the consumption impacts of excise tax on SSBs Firstly, the necessary data would be extracted from VHLSS 2016 dataset There were two main groups of figures, the number and the amount of money that households spent on SSB and other household characteristics that might affect to their consumption Secondly, these variables

28

would be given into AIDS in order to estimate the income, compensated and uncompensated price elasticity of SSBs Finally, the research would analyze the dataset and result of AIDS to give some policy implications

In order to estimate AIDS by using VHLSS 2016, eight food items were added into the model, including sugar, candy, fresh milk, wine, beer, bottled beverages, instant coffee and dried tea Regarding to these products, VHLSS 2016 only investigated the consumed amount and the consumed value of each household Except fresh milk, VHLSS asked household representative about how much they consumed in holiday, which was assumed to last for 15 days per year, and how much they consumed regularly in the last 30 days Therefore, for these items, the estimated consumed

quantity of item i (𝑞𝑖) and the estimated amount of money spending on item i (𝑣𝑖) for one year were calculated as equation 26 and 27:

𝑞𝑖 = 𝑞30𝑑𝑎𝑦𝑠×365

𝑣𝑖 = 𝑣30𝑑𝑎𝑦𝑠×365

For price of each item, VHLSS did not survey the actual price of each food item so

in this research, the price of item i th (𝑝𝑖) will be seemed as the average price (Equation 30) of that item in case of positive amount of consuming

𝑝𝑖 =𝑣𝑖

𝑞𝑖 (∀𝑞𝑖 > 0, 𝑣𝑖 > 0) (30)

29

However, because not all household consumed all item in selected basket, so in case

of zero consumption, the price of item i th was the median of 𝑝𝑖 by district If the median value of 𝑝𝑖 by district were unable to be calculated, it was replaced by the median value of 𝑝𝑖 by province The reason for using median value for 𝑝𝑖 was because of exceptional high and low average value in the price range, which could be show later (Figure 4.1)

For the household characteristics, there were six household characteristics that were taken into account, including:

- The variable of “urban” was used to separate the living area of household, whether

they are living in urban or rural area In case of urban, “urban” equaled to 1 and

in the opposite case, “urban” equaled to 0

- The variable of “majority” was used to distinguish the ethnicity of household If they are Kinh, they will be majority people in the society and these households

will have “majority” equal to 1 In the other cases, “majority” equals to 0

- The variable of “hhsize” measures the number of people in each household

- The variable of “exp_quan” were utilized to rank the expenditure quintile of households from 1 to 5, corresponding to the highest one to lowest one

- The variable of “nchild” indicated the number of children in each household

- Regarding to six geographical area of Vietnam, there were 5 variables, “reg6_1”

to “reg6_5” The variable of “reg6_1” equaled to 1 if the household was living in Red River Delta and equaled to 0 in other cases The variable of “reg6_2” equaled

to 1 if the household was living in Midlands and Northern Mountains area and equaled to 0 in other cases The variable of “reg6_3” equaled to 1 if household was living in North Central and Central Coast and equaled to 0 in other cases The variable of “reg6_4” equaled to 1 if the household was living in Central Highlands and equaled to 0 in other cases The variable of “reg6_5” equaled to 1 if the household was living in South East and equaled to 0 in other cases

The data processing method was summarized by Annex 2

30

3.3 Data Analysis

3.3.1 Overview of SSB consumption and SSB consumer

Because VHLSS collected data from selected citizens in nation-wide, utilizing the data in VHLSS 2010 – 2016 could help us to see the trend in consuming a specific product and also SSB in particular

Here was some information that could be extracted from the database:

- Price of SSB: The average price of one liter of SSB, the price range of SSB that consumers spent on, the price level at which majority of consumer consumed

- The trend in consuming SSB from 2010 – 2016

- The characteristics of SSB consuming households: household structure, urban or rural area and so on

By analyzing these features, this research aimed to provide an overview over SSB consumption in Vietnam Through this step, we could also predict which factors were affecting to this market and who were affected most by the increasing price created

by new tax policy

3.3.2 Expenditure, compensated and uncompensated elasticity

After estimating the AIDS, the elasticities of each item would be calculated by using equation 29, 30 and 31 (Poi, 2012) The first elasticity showed that if the total

expenditure of a household increased by 1%, price of good i would increase by 𝜇𝑖%

Therefore, the expenditure elasticity for good i was:

means that the expenditure for good i will increase when its price increases Whereas,

inferior goods refer to the case that if the expenditure elasticity is negative, the

expenditure for good i will decrease when its price increases

]

(32)

Where 𝛿𝑖𝑗 is the Kronecker delta (𝛿𝑖𝑗 = 1 for 𝑖 = 𝑗 and 𝛿𝑖𝑗 = 0 for 𝑖 ≠ 𝑗) (Green & Alston, 1990) The uncompensated elasticity showed how demand would change once the price changes by 1% with the assumption of unchanged income It solves the problem of maximize utility of consumer, with subjecting to a specific amount of expenditure level

Approaching from another aspect, the Hicksian demand solved the problem of cost minimization, with subjecting to a specific level of utility The Hicksian (compensated) elasticity was calculated as function … It said that if the price increased by 1%, the demand on that item would change at 𝜖𝑖𝑗𝐶%

3.3.3 The differences in elasticity between different household patterns

In order to compare the impacts of new tax policy on different household patterns, this research would employ AIDS model to each group of households to estimate elasticity for each group There were five criteria that were taken into consideration, including urban or rural area, majority or minority ethnicity, the number of children