2010:001
M A S T E R ' S T H E S I S
The Effect of the Swedish Carbon Dioxide Tax
An econometric analysis
Jonas Grafström
Luleå University of Technology D Master thesis
Economics
Department of Business Administration and Social Sciences
2010:001 - ISSN: 1402-1552 - ISRN: LTU-DUPP--10/001--SE
D-UPPSATS NATIONALEKONOMI
The effect of the
Swedish carbon dioxide tax
An econometric analysis
Jonas Grafström 2009-05-13
Luleå tekniska universitet – Hanledare: Robert Lundmark
2 ABSTRACT
This thesis investigates the existing tax on Carbon Dioxide in Sweden and tries reveal if it has any effect on the emission levels of carbon dioxide. The thesis theoretical ideas rests on the thought that taxes is a way to internalize negative externalities. Global warming is the danger with carbon dioxide. To reach an answer on the effectiveness of the tax a regression analysis has been performed on the variables tax, emission levels and development of gross domestic product. In the household sector, which is the biggest source of pollution the regression result shows that taxes have effect. The attempt to investigate the aggregated level has however proven to be harder. The estimated model for the household sector showed on a statistically significant level that increases in the tax would lead to a reduction in emissions. A one percent increase in the tax would lead to a reduction of emissions with 0,128 percent.
3
SAMMANFATTNING
Denna uppsats undersöker den existerande skatten på koldioxid i Sverige och försöker visa på om de har någon effekt på utsläppsnivån av koldioxid. Uppsatsens teoretiska ansats bygger på tanken att skatter är ett sätt att internalisera negativa externaliteter.
Med detta menas den globala uppvärmningen som utsläpp av koldioxid tros leda till.
För att nå ett svar på funderingen om skatternas effektivitet har regressionsanalys utförts på dataset innehållande skattenivåer koldioxidutsläpp och förändringar av bruttonationalprodukten. I hushållsektorn som är den största källan till utsläpp bland de sektorer som har normal skattesats visar resultatet av undersökningen på att det fungerar. Försöket att undersöka den aggregerade nivån har däremot visat sig vara svårare. Den skattade modellen för hushållsektorn att på en statistiskt signifikant nivå går det att säga att en enprocentig ökning av skatten skulle leda till en reduktion av utsläppen på 0,128 procent.
4
TABLE OF CONTENTS
ABSTRACT ... 2
SAMMANFATTNING ... 3
TABLE OF CONTENTS ... 4
LIST OF FIGURES AND TABLES ... 6
Chapter 1 ... 7
INTRODUCTION ... 7
1.1 Background ... 7
1.2 Purpose ... 8
1.3 Methodology ... 9
1.4 Scope and Limitations ... 9
1.5 Outline ... 9
Chapter 2 ... 10
BACKGROUND ... 10
2.1The Global Warming ... 10
2.2 Intergovernmental Panel on Climate Change (IPCC) ... 11
2.2 Carbon Dioxide – a green house gas ... 12
2.3 The Swedish CO2 tax ... 12
2.4 Development of Swedish CO2 emissions ... 14
2.6 The trading and not trading sector ... 15
2.5 Kyoto protocol ... 15
Chapter 3 ... 17
THEORY ... 17
5
3.1 Supply and demand ... 17
3.1 Coase Theorem ... 18
3.2 Internalization of externalities ... 19
3.3 The Kuznets curve ... 21
3.4 Literature review ... 22
3.5 Problems with a CO2 tax ... 26
3.5.1 The Free rider problem ... 26
3.5.2Policy effect problems ... 28
3.6 Model of CO2 emissions ... 28
3.7 Reliability and Validity ... 28
Chapter 4 ... 31
RESULTS AND ANALYSIS ... 31
4.1 Does a Carbon dioxide tax affect the households? ... 31
4.2 Effects on an aggregated level ... 33
Chapter 5 ... 35
CONCLUSION ... 35
5.2 Suggestions for further research ... 36
REFERENCES ... 37
APPENDIX ... 41
6
LIST OF FIGURES AND TABLES
Figure 1- Energy intensity in the context of world history (Sun, 1999) ... 22
Figure 2 - Swedish CO2 emission development 1990-2005 ... 41
Figure 3- Predicted CO2 emissions, with and without tax ... 41
Table 1 Swedish CO2 tax 1991-2004 (Energimyndigheten/ Naturvårdsverket, 2004) .. 13
Table 2 - Summary of studies of the effect of CO2 taxes. (Pearce 1991) ... 23
Table 3 – Regression result on Household ... 31
Table 4 – Regression result on aggregated emissions ... 33
Table 5 - Tax table 1985-2008 ... 42
Table 6- Sectors in the regression, Number indicate the sectors SNI –number ... 43
Tabell 7 - Data for second regression ... 45
7 Chapter 1
INTRODUCTION
1.1 Background
The industrial revolution lead to the greatest productivity and growth increase that mankind had so far experienced. From barely having any population or production growth for two thousand years the population and production level exploded. In the time period between years 1000 to 1820 the average global GDP increase was around 0.05 percent per year with that development speed the GDP is expected to double in 1400 years. In the time period that sometimes is called the golden age, from 1950 to 1973 the GDP grow with 2.93 percent. With that growth rate the global GDP doubled in 24 years. (Fregert & Jonung, 2003) The growing population and production needed energy. The burning of wood and use of animal muscle power was not sufficient and other sources with higher energy content were explored. First came the burning of coal and later the use and development of petroleum based products.
The economic growth that lifted living standards for mankind, and still is, was of course positive but there are also downsides. When burning coal and oil a lot of Carbon dioxide, CO2, which earlier was in the ground is released. When freed into the air and not enclosed and concentrated underground the carbon dioxide that turn into gas form has to go somewhere. Carbon dioxide which is a gas goes up to the atmosphere where it stays for a very long time. Every action gives a reaction of some kind. After more than two hundred years of fossil fuel burning the effect is now starting to appear.
There have been concerns a long time that man made activities could affect our planet and its climate. In 1988 the Intergovernmental Panel on Climate Changes, IPCC, was created in an effort to give policy makers an objective view on climate change issues (Intergovernmental Panel on Climate Change). IPCC were awarded the Nobel Peace Prize in 2007 (Norwegian Nobel Institute, 2007) In the Fourth Assessment Report,
"Climate Change 2007", the IPCC stated that it is very likely that human activity affect the global climate. The report also stated that the global climate was likely to change
8
because of the human activity (Pachauri & Reisinger, 2007). Looking at historical data carbon dioxide is in a uniquely high level in the atmosphere compared to the resent 650.000 years (SMHI, 2007).
The Global Warming constitutes a great problem because we cannot reverse it, with our current technology, just slow it down or reduce the effects. The cost of the damage must be measured over long time and there are worries that the warming might self escalate when the temperature rises. (Pearce, 1991) It was mankind who created the problem and is the cause behind it and therefore there also might be a possibility to change. The CO2 emissions are a major part of the problem and a reduction of the emissions would be a way to reduce or even stop the global warming. Finding methods to reduce emissions is urgent since the energy use is predicted to rise with around two percent on a global scale. (Chapman, 2000)
One classical way to reduce the demand for products that leads to emissions of CO2 for example gasoline for cars or coal fire heating for houses is to raise prices. Governments can archive this with several different policies. One such policy is taxes on fuels that lead to emissions. When using this method government usually creates a tax on fossil fuels that is set after the proportions of carbon dioxide that is produced when the fuel is combusted. This action is supposed to stimulate households and firms to shift their fuel mix toward less carbon intensive once. The emissions create negative externalities on a global scale and the tax is there to internalize it. (Jorgenson, Slesnick, & Wilcoxen, 1992)
This thesis will examine the Swedish CO2 tax in certain sectors and try to look at how it affects CO2 emissions. How much would the emissions drop if the tax was raised with one percent? Policy makers hope to effect emissions but taxes could also lead to undesired effects as reduced economic growth and capital investment. Which tax amount is the right one is a very important issue, does the tax give the desired effect or is it to low?
1.2 Purpose
The purpose of this thesis is to examine if there is any pattern between CO2 emission levels and taxes on the same. The thesis will make an ex-post analysis of the effect of CO2 tax on the households and the economy as a whole. The empirical investigation is based on data from emission levels, real GDP development and taxation levels. This
9
thesis will also go through literature on the subject and aims at giving a review on earlier research in an effort to collect findings and present them for others to learn from.
1.3 Methodology
This thesis will take an econometric investigation approach on one CO2 emission model. Ordinary Lest Squares regressions will be carried out on two data sets with slightly different components and variables. The theoretical framework concerns taxes and how they by supply and demand is intended to internalize the damage caused by pollutions in an effort to create abatement and shifts to reduction of CO2emissions.
1.4 Scope and Limitations
This thesis is limited to Sweden and the effects of tax in the country. Major parts of the earlier research on CO2 taxes have been made abroad. Therefore foreign research will play an important role in the review of earlier research. When speaking about climate changes it is defined as only man created not the part that is naturally occurring. There are many gases that contribute to global warming, sulfur dioxide, SO2 and Nitrogen oxide, NOX for instance however CO2 is the most common and this thesis will be limited to it.
1.5 Outline
This thesis is divided into five chapters. Chapter 1 consists of a general description of the problem the thesis is meant to investigate and why it is relevant and interesting.
Chapter 2 describes a relevant background of issues that is important regarding why CO2 emissions are a problem and information about related subjects. Chapter 3 is intended to give a broad understanding of the issue. It contains a review of earlier related research and focus on some specifically important thoughts in the environmental economics that is important for the thesis. The empirical model that is investigated is also presented. In chapter 4 the empirical result from the regression on the dataset is presented and analyzed. Chapter 5 contains conclusion and suggestions for further research.
10 Chapter 2
BACKGROUND
2.1The Global Warming
CO2 emissions are a major problem since it is one driving force behind Global warming which leads to climate change. There are different definitions on climate change the IPCC defines it as:
“A change in the state of the climate that can be identified (e.g. using statistical tests) by changes in the mean and/or the variability of its properties, and that persists for an extended period, typically decades or longer. It refers to any change in climate over time, whether due to natural variability or as a result of human activity.” (Pachauri &
Reisinger, 2007)
There is also another definition which is used by the United Nations Framework Convention on Climate Change (UNFCCC) they refer climate changes to:
“A change of climate that is attributed directly or indirectly to human activity that alters the composition of the global atmosphere and that is in addition to natural climate variability observed over comparable time periods.” (Pachauri & Reisinger, 2007)
The climate on the planet is of course not static even without human activities. There have been ice ages and warm periods. This paper only addresses human created climate changes and more specifically the CO2 emissions that partly create the problem. The scientific opinion about if the global warming is real also continues. Although according to Pearce (1991) the IPCC report give anyone who choosing not to believe it a formidable intellectual challenge. (Pearce, 1991)
According to Chapman (2000) there was no clear growth in greenhouse gases in the atmosphere until the beginning of the 20th century when the use of coal and petroleum products started to accelerate. Since then the average temperature on the earth´s surface
11
have risen with 0.6 degrees Celsius. This means that the average global temperature has now gone up to 15 degrees Celsius and is projected to raise even further (Chapman, 2000). Chapman use IPCC ´s findings which show a trend where global climate is rising. The 100- year linear trend shows that in the period between1901-2000 the trend showed a rise in temperature with 0.6 degrees. Further findings were that the climate change seems to be accelerating. If instead the years 1906-2005 were used the climate trend showed that the global average temperature had risen with 0.74 degrees. Out of the twelve years between 1995-2006 eleven were ranked as the warmest years since the temperature measurement data started in 1850 (Pachauri & Reisinger, 2007)
The climate changes naturally but it becomes a problem if it happened to rapidly for people to adapt. Global warming could make the ices at the north and the south pool melt. If that happened the sea levels would be affected. Since most cities are build close to water they would be threatened by rising sea levels. That cost would not outwait the benefit of for example new shipping rout in the Arctic. (Schinder, 2007)
The Global Warming is quite obvious as the name say global. There are not only negative effects of this change, in certain areas positive changes can be seen. In a cost benefit analysis of the presumed Global warming the conclusion by IPCC and several others is that the cost will outwait the benefits. (Pachauri & Reisinger, 2007) In Kyoto, Japan most of the worlds countries agreed on goals to follow in an effort to reduce their greenhouse gases which contributes to the global warming. The agreement is known as the Kyoto Climate Treaty or the Kyoto Protocol. (Chapman, 2000)
2.2 Intergovernmental Panel on Climate Change (IPCC)
The Intergovernmental Panel on Climate Change (IPCC) was established in 1988. Its mission is to give policy maker objective sources of information for decisions regarding the possible climate changes. The founders was World Meteorological Organization (WMO) and by the United Nations Environment Program (UNEP). IPCC´s reports is as they put it based on scientific evidence and should be reflecting the existing viewpoints of the scientific community, the material come from all regions in the world and from different research fields. The panel do not do any own research, they gather and quality check research. The first Assessment report was published in 1990 and more have followed. The reports are supposed to be neutral about policy and only give scientific material. (IPCC, 2009)
12 2.2 Carbon Dioxide – a green house gas
Carbon dioxide has the chemical formula CO2. It consists of two oxygen atoms bonded to one carbon atom. It is a gas in standard pressure and temperature and it exist in the earth´s atmosphere in this state. (Chang, 2006) Carbon dioxide is a natural thing that exists in the nature. Allot of it is stored under ground and when we burn the natural process of CO2 fixation is broken and the concentration increases in the atmosphere.
CO2 is mainly created when fossil fuel is burned but also by agricultural activities and the decay of any organism (Cato, 2009). CO2 is called a Greenhouse Gas since when it go up to the atmosphere since it contribute to heat retaining. Because of this an analogue to greenhouses is made. It should be noted that without this effect no life would be able to be on earth. There have always been some amounts of CO2 in the atmosphere, the problem arises when it becomes too much (Chapman, 2000).
2.3 The Swedish CO2 tax
The Swedish first CO2 tax was introduced in 1991. There were other energy taxes before but the policy makers now decided to consider the amount of CO2 emissions that the fuel produced. (Bohlin, 1998) The tax is designed to differentiate between the energy sources and the tax level depends on the carbon load of the fuel. The tax is levied on most products that produce CO2 when combusted. Such products are gasoline, oil, natural-gas, coal and burning of household garbage. (Skatteverket, 2009)
The energy intense industries have exceptions from certain taxation in the trading sector. Harrison and Kriström (1997) find in their research that the exceptions are not that important for the total aggregated carbon CO2 emissions. When they considered the effects on Swedish import and export they found that the Sweden would lose welfare but not making any improvement on the global emission level if the industries were heavily taxed (Harrison & Kriström, 1997)
Since the tax level is set after the average carbon content of the fuel some fuels receive a heavier tax load. This is because the policy makers wanted to make it reflect the damage cost of global warming. The real effect when the taxes were implemented was that the tax cost for natural gas doubled, the taxes on coal increased with 80 % and the tax on oil with less than 20 percent (Bohlin, 1998). The CO2 tax which was implemented in January 1991 was set to 0.25 SEK per kilogram of emitted CO2. (Harrison & Kriström, 1997) To make the Swedish industry competitive the CO2 taxes
13
was reduced and the general energy tax was removed in 1993. This put the tax on the households which had to pay 4.4 cent per kilo of CO2 compared to the industry who paid 1.1 cent per kilo (Bohlin, 1998). In Swedish currency, 0.34 SEK per kilogram for the sectors that was not exempted and 0.083 SEK per kilogram for the manufacturing sector (Kriström & Lundgren, 2005). The tax level has change upward over the years. In 2007 the tax for the industry was 0.19 SEK and the other sectors 91 SEK. (Byman, Grunfeldt, Grönquist, & Stenquist, 2007) In the budget of 2008 the government raised the tax with 0.06 SEK per kilogram to 1.01 SEK per kilogram of emissions.
(Miljödepartementet, 2007) Virtually anything that we consume contributes to CO2 emissions one Swedish hamburger restaurant for example claims that one of their standard hamburgers contributes with 1.8 kg of CO2.counting on the tax level of 2004 then the product would be taxed by 1.9 SEK per burger (MAX, 2009).
Table 1 Swedish CO2 tax 1991-2004 (Energimyndigheten/ Naturvårdsverket, 2004)
In the table above the development of the Swedish CO2 tax is showed. The numbers is in SEK per ton so to read it in cost per kilo the numbers have to be divided by one thousand. The first row show the tax levied on sectors that is not excluded for any reason. The second row show how many percent the excluded industry pay. The last row reveal how many SEK the industry have to pay per ton (Energimyndigheten/
Naturvårdsverket, 2004).
In an article Folke Bohlin (1998) discusses the effects on the Swedish carbon dioxide taxes and some problems in the correct evaluation of the effectiveness (Bohlin, 1998).
Bohlin finds that there is a problem in evaluation since there is nothing to compare with.
Since either the tax or no tax alternative is picked in a country you cannot really compare with the other. In sum Bohlin claims that the CO2 taxes lead to abatement, according to him it was between 0.5 and 1.5 million tons per year. He found that the CO2 emissions would have been 2-3 percent higher without the government policy interventions in the year 1994. Bohlin found that even though the taxes were
14
implemented in 1991 the aggregated CO2 emission continued to grow from 55.4 million tons in 1990 to 58.1 million tons in 1995. (Bohlin, 1998).
In a model by the swedish department of envrionmental care and the department of Energy the conclusion is that the tax, if held at 2004 years level would reduce emissions each year and that the tax will lower emissions by 5 tons in the year of 2023 if held at the same rate. At that level the emssion level would however rise a little, see apendix:
Figure 3- Predicted CO2 emissions, with and without tax. (Energimyndigheten/
Naturvårdsverket, 2004)Figure 3
Different fuel types have different emission rate when producing the same amount of energy. Coal and coke is the heaviest polluters with 92 and 103 grams of CO2 emissions per MJ of energy. Raw oil has 45 grams and natural gas have 56 grams. Sweden has shifted away from the high-carbon low-effective fuels to cleaner alternatives. Emissions from firewood are generally not included in the statistics since it is considered to not contribute to a net increase of CO2. As long as new trees grow new CO2 will be bound.
(Kander, 2002)
2.4 Development of Swedish CO2 emissions
There is no certain numbers of how big the Swedish CO2 emissions were during the beginning of the 19th century, guesses could be made by import and production statistics. Firewood and animal muscle energy was substituted by modern energy carriers such as coal and oil. The coal imports started in the 1820s and accelerated in the 1850s at the same time as oil extraction started in the United States. The Swedish CO2
emissions have increased almost the whole 20th century. (Kander, 2002).
The First and Second World War had negative impact on the amount of emissions.
During the war years the emission level was at century low. This can be explained by that Sweden was isolated and could not import petroleum products at any large amount.
Sweden was also cut off from most of its exports and imports. The emission levels increases rapidly after the end of World War 2 until the late 1960s, it peaked at 1969.
After the peak the emission levels started to decline fast for several years. Reasons for that was the new nuclear power which produced large amounts of energy without any CO2 emissions. The price shock on oil after OPEC´s decision to increase oil prices further increased demand for alternatives. The production processes in the industry changed because of technological progress and the amount of electricity declined there
15
to. From the 1980s the emission level has stabilized and is just growing slowly.
(Kriström & Lundgren, 2005) The Swedish CO2 emissions are above the world average but half of the OECD average (SCB, 2009).
The emission level has been below the 1990s level so far during the first years of the 21st century. The economic recession which is believed to be long and hard is most likely to lower the emission level even further. The industry production is falling and unemployment is spreading across the world. See appendix for curve over CO2 emission development 1990-2005. (European-Comission, 2007)
2.6 The trading and not trading sector
All sectors of the Swedish economy are not affected equally by the CO2 tax. The economy is divided into the trading and no-trading sector. This since some industries is part of the European emission permit system. Therefore they only pay a small part or is excused from the Swedish tax. In 2005 the European Union implemented a system where certain industry sectors and producers of heat and electric energy have to trade with allowances to be permitted to have emissions. If the companies want to emit CO2 they have to buy a permit to do so. The “not trading sector” contains transports, agriculture, housing, waste management and certain energy businesses. Since 2008 Sweden I obligated to work for a reduction of 17 percent in the non-trading sector. The number of renewable energy should increase to 49 percent from 40 percent. (Lagerstedt, 2008)
2.5 Kyoto protocol
The Kyoto Protocol was established after a UN convention on climate change where 160 countries and over 10.000 people participated. It created and defined the general goals for the developed countries with some exceptions, some countries refused to sign it. The participating countries are supposed to do efforts to reduce CO2 emission and try to stabilize their emissions at the 1990 level to the target year 2010. (Chapman, 2000) Sweden decided to aim for even higher goals (Kriström & Lundgren, 2005) . One note is that Western Europe and The United States have had a stable not changing emission level since 1990. Russia and Ukraine have since the downfall of communism and the economic breakdown after the fall of the Soviet Union actually had decrees in emission. In 1996 their emission levels were 30 percent lower than in 1990. The developing countries showed a major growth in their CO2 emissions but did not have to
16
commit themselves to any reduction (Chapman, 2000). One major setback for the Kyoto protocol is that the United States withdrew from the agreement in 2001. The protocol covers only approximately 30 percent of global emissions. Those percents are located in countries where emissions are not really increasing at all even thought the economies is growing (Nordhause, 2007).
There are arguments against the effectiveness and use of the Kyoto protocol. Since the climate is affected by total emissions it does not matter if some countries reduce their emissions if other countries take the opportunity to raise theirs. If taxes are levied on some countries it simply gives the effect that production that creates emissions is relocated to other countries. Therefore there are arguments that tradable resources should face a lower tax. (Hoel, 1996)
17 Chapter 3
THEORY
This chapter begins with discussing the general supply and demand theory and will be followed by a discussion of taxes and how they should be formed. The Coase theorem about externalities will be presented, this since CO2 is an externality and taxes are a way to solve it. It will be followed by a literature review which presents other research on the area. Since more than taxes can affect CO2 emissions the Kuznets curve will be presented, it discuss if economic development leads to reduced energy intensity. This chapter is intended to give a broad understanding of issues related to CO2taxes and viewpoints on this kind of environmental issue.
3.1 Supply and demand
There are many factors that affect the price of a good. Generally the demand is derived first of all from the price of the good, the price of substitutes is also important. If the price for a good rise the quantity demanded will fall, this is known as the law of demand. There are two main reasons for this, the income effect and the substitution effect. When the price of a good go up peoples relative wage go down, they will feel poorer and will not be able to consume as much. When price for one good go up the relative price for another good goes down, this is called the substitution effect, people will change their consumption mix. (Sloman, 2003)
The factors that affect demand are mathematically specified in the demand function.
The demand function describes the relative factor of each of the factors. The demand function can be described graphically. This becomes the demand curve. The supply curve can be derived in a similar way. Those curves show at any given price what would be demanded and supplied ceteris paribus. Where those two curves intersect the real quantity demanded is reveled. (Sloman, 2003) It should be noted that there is no direct demand for CO2. CO2 is an externality from the use of and production of other desired goods. Taxes are a way to put a price on the externality and by that affect the demand for it.
18 3.1 Coase Theorem
In 1960 Ronald Coase wrote the article “The Problem of Social Cost”. It has become a milestone article in environmental economics. Coase´s basic idea is that if rights are well defined the market will take care of the resources use in a proper way. The main problem with CO2 emissions is that everybody is able to combust products that release CO2 to the atmosphere without paying anything for the damage they cause others. Even thought the direct effect of one individual’s actions is hard to see the aggregated emissions give an effect that can be seen and measured. The air is free so to say and since nobody can claim an ownership anyone can use it and dump their cleaning cost, a negative externality is created.
Externalities are a problem that the Coase theorem seeks to solve. The Coase theorem states that if property rights are established the different actors in the economy will try to bargain for the right to use it. This to get a solution where both benefit and the use of resources is optimized. According to Coase it does not matter who own the property right, economic efficiency will be archived anyway. For the Coase theorem to work two properties must be fulfilled. First the marginal benefit and cost for participants must be known and second: there should be no transaction cost in the bargaining. (Coase, 1960) Nicholson (2004) illustrates this by an example with two firms. Firm X and Y. in Nicholson example firm X gets 𝑋𝑇 rights to produce (and pollute). The firm can decide to use some of those rights for its own production (𝑋0) or sell to firm Y, that amount is given by 𝑋𝑇− 𝑋0. The gross profit for firm X is given by this profit equation:
𝜋𝑥 = 𝑝𝑥𝑥0+ 𝑟 𝑥𝑇− 𝑋0 = 𝑝𝑥− 𝑟 𝑥0+ 𝑟𝑥𝑇 = 𝑝𝑥− 𝑟 𝑓 𝑦𝑖 + 𝑟𝑥𝑇 (1)
For firm Y the function is given by:
𝜋𝑦 = 𝑝𝑦𝑔 𝑥𝑖𝑥0 − 𝑟 𝑥𝑇− 𝑥0 (2)
Profit maximization will lead to the same solution for both firms. It doesn’t matter which firm who get the right. (Nicholson, 2004)
Coase points out that the usual economics approach with the assumption that there is no transaction cost is not very realistic. Coase means that if anyone that want to carry out
19
market transactions must identify information with the ones they want to deal with and what to deal about. Contracts must be written. In the example with CO2 emissions which is a global problem it is hard to say who is damaged by someone’s emissions and how should they come to an agreement that benefits both?
Coase means that an alternative solution to individual bargaining is to use the government which Coase think in a sense is a super firm. He thinks so because it can influence the factors of production through decisions. The government can regulate the market with different policies. It can be regulations about how firms and households must act or taxes if they do a certain activity. The government can make decisions that apply to everyone and in that way reducing the free rider problem. Coase give one fictional example with a big factory and a thousand property owners nearby. The factory is given the right to pollute and the property owners can pay them to reduce their pollution. Some households might refuse to pay or try to downplay the effect of the pollution to be charged less. (Coase, 1960)
3.2 Internalization of externalities
Taxes on CO2 are a mean for the government to try to internalize the cost associated with CO2 emissions to those that pollute. The argument is that pollution causes damage and that if a cost is introduced the consumption will hopefully shift to a substitute that cause less damage.
Hoel (1996) studies how the optimal consumption and production patterns should be formed. Hoel means that utility is gained from consumption but it also creates negative environmental externalities that effect utility negatively. Welfare (W) is derived from consumption of different kind, tradable and non-tradable. Consumption on an aggregated level is 𝐶 = (𝑐0, 𝑐1, … , 𝑐𝑛) where 𝑐0consumption of fossil fuel products is.
From the Utility, which makes up the welfare, the cost of external effects should be subtracted. The environmental cost of total emission is denoted z.
𝑊 = 𝑈 𝐶 − 𝐸(𝑧) (3)
In the model constructed by Hoel the total emissions consists of domestic consumption of fossil fuel (𝑐0), domestic use of fossil fuels as input into production which Hoel denotes as v, and foreign emissions, denoted e. that give the formula:
20
𝑧 = 𝑐0+ 𝑣 + 𝑒 (4)
In the case of Global warming the negative externalities could give large Utility losses.
Environmental impacts often occur on a very big scale and therefore could create a very big cost. The climate change that might come from global warming is predicted to influence the occurrence of hurricanes (Cato, 2009). One example a costly weather incident, even if it not for certain an effect of global warming, is the hurricane Katrina.
Katrina hit the United States and caused massive damage and severe loss of life. The monetary damage was calculated to almost 156 billion US dollar (Burton & Hicks, 2005).
Taxes are one way to internalize the externalities. This is illustrated by Brännlund and Kriström (1998),they show a firms simple profit function. The firm produces the good Q. Then the company’s profit could be written as:
𝜋𝑓 = 𝑃 × 𝑄 − 𝐶 𝑄 (5)
Where P is the price of the product and C(Q) is the companies cost for producing the product. The firm will produce until the marginal cost for producing one more unit is equal to P. the production will cause emission which is denoted as z = k x Q, those emissions are harmful to society. The society also profit from production, the profit function is written as:
𝜋𝑠 = 𝑃 × 𝑄 − 𝐶 𝑄 − 𝐷 𝑧 (6)
Assuming that:
𝐷 𝑧 = 𝑎 × 𝑧 = 𝑎 × 𝑘 × 𝑄 (7)
One more unit of production increases the damage to society with a. assume now that a tax is introduced, t. the tax is set to a, which is the societies cost. The profit function becomes:
𝜋𝑓 = 𝑃 × 𝑄 − 𝐶 𝑄 − 𝑡 × 𝑘 × 𝑄 (8)
In the company’s perspective their cost now increases with t. this reduce their profit and production. The society will get compensated for the negative effect of the firm’s
21
production. The tax becomes in this way an option for internalization of the negative externality. (Brännlund & Kriström, 1998)
3.3 The Kuznets curve
There is a hypothesis is called the environmental Kuznets curve, it proposes that there is an inverted U-shaped relationship between pollution intensity and per capita GDP. In this hypothesis intensity is defined as pollutant per GDP. The hypothesis states that if GDP grow to a certain level then the emission levels will start to decline per capita. Not necessarily in absolute terms however since there is economic development and population growth. (Sun, 1999). In a study made by Holtz-Eakin and Selden (1995) the relationship between economic growth and CO2 were studied. They and found that the ratio between income and CO2 never had turned downwards. However there was a slower growth of CO2 emissions at higher income level in their sample (Holzt-Eakin &
Selden, 1995).
Sun (1999) observe that when a country is in a period of industrialization the energy intensity per GDP tends to go up, more consumed energy give the result that more emissions is created, at least in a pre-industrial state. After a while the industry structure in the country tends to shift away from high energy intensive industries to low intensive industries. The production mixed shifted from general value-add to a more knowledge intensive production. The energy peak occurred at different times for different countries. The United Kingdom which was the first country to start its industrialization reached its peak at 1880 while the United States reached it at 1920. Countries like Japan which was heavily damaged during the Second World War reached it at 1972. (Sun, 1999)
22
Figure 1- Energy intensity in the context of world history (Sun, 1999)
From this Sun demonstrate that the CO2 environmental Kuznets curve can be derived from the peak energy theory of energy intensity. Sun shows that the energy intensity has gone down on average in both OECD countries and non-OECD countries. Sun makes the assumption that W represent aggregated energy use, E is aggregates CO2 emissions and C is the CO2 emission coefficient, he define the relationship as𝐶 = 𝑊𝐸. This gives CO2 emission intensity (𝐺𝐷𝑃𝐸 ) = 𝐶 ×𝐺𝐷𝑃𝑊 = C x Energy intensity. He found that worldwide the CO2 emission coefficient was reduced by 0.41 percent and in the OECD countries the reduction number was even higher, 0.57 percent. (Sun, 1999)
Energy intensity seems to be correlated with economic development. Energy use leads to emissions but the energy intensity seems to go down according to Sun after the economy in a country has reached a certain level. Sun finish with pointing out the fact that this pattern is only a historic development and do not for certain tell when a country will start to improve its environmental conditions. (Sun, 1999)
3.4 Literature review
Even thought research on CO2 taxes is a rather new field, since the problem with global warming has not until recently been a major concern, there is several interesting articles with different angles. Pearce (1991) have presented a summary of research of how different tax levels would affect the emission rates. It is presented in the table below.
23
Table 2 - Summary of studies of the effect of CO2 taxes. (Pearce 1991)
The finding in the different research seems to show that taxes do have effect. But considering the spread of the result it is hard to draw any conclusion about which tax level is the right one. Comparing Cline (1989) and Howarth (1989) both focus on the year 2050 and reach very different results. According to the result presented by Pearce (1991), Cline´s result indicates that a tax set to 158 dollars per ton of carbon would reduce the CO2 emission levels with 57 percent in the world. Howarth on the other hand reach a much lower reduction to a much higher price. Howarth find that a tax set to 623 dollars would result in a reduction of 26 percent to the year of 2050. (Pearce, 1991)
24
One effect that could be considered as undesirable is that general tax on carbon dioxide consumption might be regressive, this happened in the Danish case. Regressive taxes make low income people pay a bigger portion of their salary than high income people.
Low income families were expected to pay approximately 0.8 % of their disposable income in CO2 taxes while the high income families paid only 0.3 % (Wier, Birr- Pedersen, Jacobsen, & Klok, 2005). According to Harrison and Kriström (1997) most household expenditure surreys over developed countries including Sweden shows that direct energy consumption is a larger share of total household budget in low income families (Harrison & Kriström, 1997).
Similar conclusions are drawn by Jorgenson, Slesnick and Wilcoxen (1992) who also find CO2 taxes to be regressive. They investigate the demand effect on 35 different goods and found that the demand for them went down for all in different magnitude except for imported goods. As for example a household with only one member and a low personal wealth would lose 1396 dollars if a tax was levied in the year of 2010 with the objective to push down the emission levels to the level of 1990. One of their conclusions is that the CO2 tax in the US, which is the investigated country, must go up every year if the policy makers want to stay at the 1990 level. They predicted the economy and the population to grew and with that the CO2 emission. (Jorgenson, Slesnick, & Wilcoxen, 1992).
According to Pearce (1991) there are several positive things about environmental taxes like the carbon dioxide tax. First of all according to him, unlike regular tax on goods like enterprise and labor it does not distort incentives. A tax on labor reduces the willingness to work which is consider as a negative by most policy makers. The environmental tax is there to solve the distortion that arise since the common good, the environment, is overused. There are also arguments that the environmental tax can be made more acceptable by the public if the revenue is used to reduce taxes which have a negative effect on incentives, such as the corporate tax. Regular taxes create dead weight losses on more than twenty percent. The CO2 tax itself also can create a dead weight lose, it have to be measured against the gains from reducing the negative externalities. (Pearce, 1991)
Pearce (1991) used a general equilibrium approach to investigate the effect of different tax levels at the United States economy. He found that a small, 5$ tax per ton of CO2
25
would create a welfare lose of 0.01 percent of the US gross national product. That would be $280 million dollars. A tax on $ 450 dollars per ton would lead to a welfare lose of 1.2 percent of the US GNP. (Pearce, 1991)
Technological development is a contributor to the reduction in CO2 emissions. In the US the energy use by the households has declined with 16 percent since the 1980s, this despite a growing economy. The decline was primarily related to reduction in energy usage for home heating, it fell by 33 percent. (Chapman, 2000). Taxes instead of regulations create incentives for technological advances. When a regulation level is introduced there is no incentives to go beyond that level, taxes give continuous incentives to abate. The total amount that has to be paid in taxes decline when the emissions decline (Pearce, 1991). Evidence of technological development is for example the “Snövit” project in Norway. A part of the gas field’s onshore production facilities has the mission to separate the carbon dioxide from the natural gas and pumped back underground. Another case is the development of more effective power plants. In 1890 3.5kilograms of coal was needed to produce one KWh of electricity in the most advanced countries. In 1980 only 0.343 kilos was needed in the. The production has become more than ten times more effective. (Kander, 2002)
Bruvoll and Larsen (2004) found in a study of the Norwegian CO2 emissions that the Norwegian emission level per capita gross domestic product had declined with twenty percent the last nine years. Their analysis was conducted in a three step and the combined two different methods. In their first step they first subdivided the observed changes in emissions in their target years which were 1990 to 1999 to eight different driving forces. Their goal was to reveal the main mechanisms that had changed the emissions in that decade. Such mechanism could be taxes CO2 and energy taxes. To be able to quantify the isolated effects they applied a disaggregated general equilibrium model.
On average the energy intensity was reduced with 7.2 percent after the tax was introduced during the years from 1990 to 1999. The sectors that were not included in the tax regime experienced an increase in energy intensity. Like in Sweden the energy intense industries is not taxed. When the tax was introduced 25-30 percent of Norway’s CO2 emission came from such sources. They found that even thought that gains were made in areas such as technological development and energy efficiency this was not
26
sufficient to match other contributing factors. The population growth which Brunvoll and Larsen thought was responsible for a five percent emission growth and the economic development in the country which gave another 30 percent growth. (Bruvoll
& Larsen, 2004)
The households could not avert the taxes in the same way as the industry. The households’ main expenditure back then was related to heating and transportation.
(Bohlin, 1998). The environmental taxes are effective since it have a potential to be an opportunity for industries and households to comply with the tax in many ways. Anyone using the good that is taxes will be faced with the decision to abate or to continue to pollute but. As long as the marginal cost to abate is lower than not to the consumer will abate. The high marginal cost polluters will pay the tax and low cost polluters will start to abate. (Pearce, 1991).
3.5 Problems with a CO2 tax
There are a number of factors associated to policies involving taxes on CO2 might render the taxes ineffective. Taxes should be formed so they cannot be avoided by just buying from another market or moving production. In the case of CO2 emissions a tax in one country might lead to a reallocation of production to another country where the taxes is lower. A number of problems that has been presented in the literature will be discussed below.
3.5.1 The Free rider problem
Countries have an incentive to free ride on other countries policies. If a couple of countries agree on inserting taxes there will be a positive effect for all countries. If a single country stays out of the agreement it can enjoy the same benefits as the others without having to give up any production pay for any actions taken. The country without the emission standard may gain investments since it is more profitable for a firm to produce where there are no taxes. This is considered a major problem when creating environmental policies. Countries in the developing world might start to race to the bottom to get investments. If nothing is done about the free rider problem the agreement might collapse. There is no international law that can force countries to do anything it does not want to. (Hoel, 1996)
27
It would in fact be rational for most developing countries to not participate in an agreement. In their developing process extensive use of energy which causes CO2 emission occurs. Their economy usually relies on production of goods not as much services as in the developed world. That kind of production is more energy intense.
Countries have incentives to participate when their cost to do so is smaller than their benefit. Island counties like the Maldives for example have a big incentive since they could be severely affected if the water rise which is one predicted thing to happen if the temperature goes up and the ice starts to melt. The Maldives is an island nation with a low shore line (Pearce, 1991).
To solve this problem the taxes has usually been differentiated between sectors. Hoel (1996) argues that that is not necessary if the countries solve it with tariffs instead. Then there can be unified taxes on all sectors. The tariffs could work since the non- cooperating countries will lose some of the benefits involved in free riding and make it an economic choice to actually agree on cooperating. The cheating countries do not have a strong position to argue either. Since the tariffs and taxes are intended to save the environment it is easy to argue for the cooperating countries.
Sweden is a small and open economy. The country is affected by the outside world.
This is something that has to be considered when creating a tax. All electric power that is consumed is not produced in the country. Sweden buys and sells electricity on a European market. If a large tax is put on energy production in Sweden consumers can substitute to foreign producers which is not affected by the tax in their home country (Harrison & Kriström, 1997). This effect have also been shown in Denmark were taxes on petroleum were not used since the geographical closeness to the German market.
Higher prices on Danish petroleum were expected to lead to cross border shopping.
That would render the tax inefficient and divert the consumption to another market where the taxes were lower. (Wier, Birr-Pedersen, Jacobsen, & Klok, 2005)
This problem can be seen in the context of the prisoner’s dilemma. Individual countries can gain advantages if they don’t care about the global warming and let other do the reductions and pay the price for a better climate for both of them. If both cheat both will have to take the consequence when the global climate changes. The best decision for al is to cooperate but there are incentives to cheat.
28 3.5.2Policy effect problems
If a target for the desired reduction percent is set it is hard to know which level the tax should be in. The elasticity must be known at least to a reasonable certainty for the policy to reach the desired objective. Energy demand varies much and since the carbon content of the fuel is the thing that is usually taxed the policymaker must have an idea about the inter-fuel substitution elasticity. The prices of the different alternatives plays a major role as long as the cost is lower per energy unit the producer will stick to that source.
3.6 Model of CO2 emissions
The estimated model will be based on the Cobb-Douglas function. The amount of CO2
emissions is estimated to depend on taxes and the industrial output plus a dummy variable.
𝐸 =∝ 𝑇𝛽1𝑄𝛽2𝑢 (9)
E = CO2 emissions T = Tax
Q = Real GD
U = dummy variable
∝ = Intercept
To be able to see the changes in percent the regression is made in logarithmic for. In the ordinary Cobb-Douglas shape the model shows changes in real numbers. For example if the tax is raised with one SEK per kilo of emissions how many kilos less would be emitted. It is hard to get a picture of the effect in this way. Therefore the logarithm form is useful, it shows the result in percent. The Cobb-Douglas function is made linier and gets the form:
𝐿𝑛 𝐸 = 𝐿𝑛 ∝ +𝛽1𝐿𝑛 𝑇 + 𝛽2 𝐿𝑛 𝑄 + 𝐿𝑛 𝑢 (10) 3.7 Reliability and Validity
The data that is used for the first regression and in the first model is collected from several different sources mostly from Sweden’s Statistical Central Bureau (SCB). The data should most likely be quite reliable. The major part of the data of the taxes is taken from the table presented by the Swedish department of Environmental protection and the Energy department in a rapport the two governmental departments cooperated with.
It covers the tax level in the years 1993-2004. The year 1991 and 1992 is from the Tax statistical year book from 1994. The tax level from 2004 and forth is gathered from the
29
Swedish government’s budget propositions. The tax level has been described in both kilos and tons. To calculate in tons the kilo price has been multiplied with one thousand.
Since no monthly or quarter-year data over emissions have been found data over the aggregated year emissions have been made. Since the tax has not existed that many years and have been levied on different sectors in different ways the number of observations makes the regression result somewhat unreliable as an effect of the low number of observations. On the other hand if the data would have been presented in months it would not compensate for the changes during the year in the same way. There is a huge difference of the use for heating in the winter and the summer even thought that the tax level is the same. The reader should observe that in the regression only 14 years is included and tested. To make the regression more reliable more observations would have been required.
The data over emissions could also be suspected of consisting of some errors. First of all it is hard to measure the correct amount of emissions. The numbers must be based on estimates to some extent since it would be very hard and costly to monitor all possible emission sources. However there should be possible to calculate on the imports of different energy sources, the main problem is to know when they are used. A dataset from Sweden’s Statistical bureau is used for the emission levels. (SCB, 2008) It originally contains several sectors, both from the trading and non trading sector. The ones from the trading sector will be removed. Sectors with the SNI-code 10-37 and some agricultural- and hydro-sectors have a reduced tax rate and some parts of them are included in the trading sector. (Byman, Grunfeldt, Grönquist, & Stenquist, 2007) To be able to give an accurate measurement those sectors are therefore excluded. The result of a test for how the tax effect emission would turn out biased if some sectors that only pay approximately 25 percent of the tax is included. From the data some sectors have also been taken out. Those are SNI 61 – shipping companies, SNI 62 Airlines and SNI 95-99 work in the home, foreign embassy. See included sectors in Table 6- Sectors in the regression, Number indicate the sectors SNI –number. Those have recorded emissions but no tax revenue. Airlines are for example excluded from the CO2 tax (Svedlund, 2007).
One problem with the data is that in the years of 2007 the SNI codes where changed.
This was made in an effort to standardize the Swedish codes and statistics to the rest of
30
the world (SCB, 2007). Those changes would affect the regression since there have been hard to see the new statistic and know where emissions have been counted. To avoid the measurement error the year of 2007-2008 has been excluded.
Real GDP numbers come from Sweden’s Statistical Central Bureau, this should make it reliable since the source should be considered as especially reliable. A major part of SCB´s job is to keep track on the economic development in the country.
One major problem when analyzing the effectiveness of the Swedish CO2 taxes is that there are many different policies and political issues that make the full effect of the tax scheme hard to see. Some sectors is not affected by the taxes other pay a reduced sum.
The model can be criticized for emitting factors like technological development and energy efficiency which would be way to reduce emissions.
In the second regression the model is partly the same but the model contains more years some other data and the extra dummy variable ETS which stands for the years that the European Emission trading system has been in effect for Sweden. It stretches from1960 to 2006. The tax is set to one in the years that it was not implemented thus when taken into logarithmic form transformed to zero. The data for the emission levels is collected from OECD´s database over aggregated emissions in the different countries of the world (OECD, 2008).
31 Chapter 4
RESULTS AND ANALYSIS
4.1 Does a Carbon dioxide tax affect the households?
The households are the main source of emissions in the no-trading sector. It is also the final line of consumption where taxes have big effect. A firm could pass on a tax to the consumer. When the firm is faced with a tax it can raise the price on their product and charge more so it does not have to face the whole cost themselves. For most households electricity, gasoline and heating are big parts of the budget. Therefore the tax is suspected to affect their behavior. The household sector is also paying almost half of the total CO2 taxes collected by the government (SCB, 2008).
The regression result below comes from a regression on the logarithmic values of the carbon dioxide emissions the households have had the years 1993-2006. Equation (X) is used to estimate the effect of the tax during the time period. Table X shows the result of the regression on households.
Variable Coefficient t-ratio Stastisticly Significant
LNTAX -0,12864 -2,533 ´´
LNGDP -0,60907 -3,05 ´´´
Constant 26,1596 9,921 ´´´
Adjusted R-squared 0,916
´ Statisticly significant at a 10 percent level
´´ Statisticly significant at a 5 percent level
´´´ Statisticly significant at a 1 percent level
Table 3 – Regression result on Household
The regression shows that if the coefficient for both and GDP is negative. This means that if any of those coefficients would increase it would create a reduction in CO2 emissions. The tax coefficient shows that if the tax on Carbon dioxide is raised by one percent the households CO2 emissions is expected to decrease with 0.128 percent. This goes in line with the earlier studies that were presented in the literature review and also
32
with the expectations according to the law of supply and demand. The result on taxes is statistically significant at a 95 percent level. The result is not however certain at a 99 percent level. the result suggest that taxes do effect the emissions that comes from the household but that it does not seems to be that effective. A doubling of the CO2 tax if demand elasticity is assumed to be fixed would reduce CO2 emissions with 12.8 percent.
From the ministry of treasure a tax on Carbon dioxide for the households could be considered as a good thing to use if collection of tax money is just the goal. The households need energy and transportation so they cannot avoid the taxation. The tax base is stable since the consumption is not that sensitive to the tax. The downside is how well the tax could be said to achieve its goal.
In the case of the GDP variable the result might seem a bit odd at first sight. Economic growth is increased action in the economy, logically if all factors would be fixed an economic growth would lead to more CO2 emissions and a shrinking economy would lead to less emissions. This does not however seem to be the case. In all the years in the regression Sweden’s economy have grown but the emission trend has go down. The emissions have gone down on an aggregated level and on the household level. The result shows similarities with the ones the Kuznets curve points at. This implies that after a country has reached a certain economic level their energy intensity will start to fall. This in combination with technological advances and the reduction goals that Sweden have and the measures that is taken to reach them it is perfectly possible that emissions could go down when GDP does it. The GDP growth is an indicator of reactions in the economy and a way to but a price on the value of it. To argue that the GDP growth itself is the driving factor behind emission reductions is most likely wrong.
It is most likely the development of the economy that is driving the process. Economies tend to shift from low-tech energy consuming production to production mixes that is more knowledge based when the country get richer.
According the adjusted R-squared value which is 0.916 the model provides a very good fit when estimating the variables as explanatory. However since there few observations available the result could be affected by this. More observations would give a better estimate. In this regression only 14 observations is measured. This since the tax has only been in effect for a few years. A data set with the emissions for every month during that time period would most likely have provided a more secure result. However no such data were to be found, there were only aggregated emissions per year.
33 4.2 Effects on an aggregated level
To come around the problem with the low number of observations the data set is modified and widened. In an effort to solve the problem with the few observations
Variable Coefficient t-ratio Stastisticly Significant
LNTAX 0,13561 2 ´
LNGDP 0,01552 0,148 Not statisticaly significant
Constant 4,053 5,367 ´´´
ETS -0,26625 -1,868 ´
Adjusted R-squared 0,1109
´ Statisticly significant at a 10 percent level
´´ Statisticly significant at a 5 percent level
´´´ Statisticly significant at a 1 percent level
Table 4 – Regression result on aggregated emissions
In this regression some very odd results comes out of the regression. The values are not very logical. In the tax case the result actually is pointing on that an increase in the tax level would give an increased emission level. This is however only significant at a 90 percent level. in the GDP case the result show that if GDP increase with one percent then the emissions would increase with 0.01 percent this is first of all incredibly little and secondly the result is not even close to be statistically significant. ETS is a dummy variable that is intended to sort out results when the European Emission trading system was implemented. The R-squared value is very low 0.1109. This indicate that this model does not even is close to have explanatory power. The model fails to indicate any relation between taxes and reduction in CO2 emissions.
One explanation for this could be that since time series is analyzed there could be structural breaks that that come from outside the model that affects the results. One such structural break in CO2 emission data could be the oil shocks in the 1970s. Dougherty (2007) points out five different reasons why the model might be wrong. The second explanation why the model might not work is that the variables are aggregated. This could explain the weakness o this attempted model with a longer time series data. The emission levels are aggregated and represent total emissions in the country. Therefore sectors that are not exposed to the tax and sectors that are will be mixed together.
Different sector can also have different reactions to emission taxes since some is less price insensitive. (Dougherty, 2007)
34
This is something to consider for anyone that is trying to solve the problem with the few observations. To increase the number of years beyond the once that was effected by the taxes do not seem to solve the problem. Winter temperature which could be a driving force behind energy use could not be used. Since the regression model could not handle the negative numbers when they were transformed to a logarithmic form. Cold winters lead to a larger energy use compared to milder ones.
Looking at the real numbers of emissions on an aggregated level evidence of a decrease in emission is easy to spot (Tabell 7 - Data for second regression). Many active policies is at work that effect the emission levels and it would not be unlikely that the tax is one of them even thought indicating the opposite in the regression result.
