2005:16
M A S T E R ' S T H E S I S
Capacity Changes in the Power Industry
Lessons from an Allowances Trading System
Jonas Tyrestad
Luleå University of Technology
Master's Thesis in Economics
Department of Business Administration and Social Sciences
ABSTRACT
The study focuses on how an allowances trading system affects the development in capacity in the power industry. This is done by analyzing the change in capacity in the US power industry during the Acid Rain Program, using an OLS regression on a model consisting of factors affecting investments in the power industry. The results obtained are thereafter used to discuss how the capacity in the European power indus- try will develop under the allowances trading scheme within the Kyoto Protocol. The results show that the change in capacity in the US power industry has been affected negatively during the Acid Rain Program, though on a small scale. How the European power industry will be affected is difficult to forecast, due to differences in abatement technologies between the different emissions targeted in the two schemes. Uncertainty about future policy and prices also add to the difficulty to make satisfying predictions.
SAMMANFATTNING
Uppsatsen behandlar hur ett system med utsläppsrättigheter påverkar utvecklingen av kapaciteten i el-industrin. Detta genomförs med att analysera förändringen i kapaciteten i den amerikanska el-industrin under det så kallade “Acid Rain Program”, med hjälp av en OLS-regression av en modell bestående av faktorer som påverkar investeringar inom el-industrin. Resultaten används sedan för att göra antaganden hur kapaciteten i den europeiska el-industrin kommer att utvecklas under systemet med utsläppsrättigheter inom Kyotoprotokollet. Resultaten visar att förändringen i kapaciteten i den amerikanska el-industrin har påverkats negativt under “Acid Rain Program”, dock i en liten skala. Hur den europeiska el-industrin kommer att påverkas är svårt att förutsäga, främst på grund av skillnader i reningsteknologi mellan de olika utsläpp som de två programmen fokuserar på. Osäkerhet om framtida policy och priser ökar också svårigheten att göra tillfredställande förutsägelser.
CONTENTS
ABSTRACT... I SAMMANFATTNING ...II CONTENTS... III TABLES AND FIGURES ... V
Chapter 1 INTRODUCTION ...1
1.1 Background ...1
1.2 Purpose...3
1.3 Methodological framework ...3
1.4 Scope...4
1.5 Outline of thesis...4
Chapter 2 BACKGROUND ...5
2.1 The Kyoto Protocol ...5
2.1.1 Commitments...6
2.1.2 Flexible Mechanisms ...7
Emission Trading ...8
Joint Implementation ...8
Clean Development Mechanism ...8
2.2 The Acid Rain Program ...9
2.2.1 The History behind the Program...9
2.2.1 Title IV in the 1990 CAAA, a market based solution...10
2.2.1 Results and experiences ...11
Chapter 3 THEORY ...13
3.1 Allowances trading...13
3.1.1 Graphical example of an allowances trading system...14
3.1.2 Cost efficiency ...15
3.1.3 Tradable allowances compared to charges ...15
3.1.4 Effects on investments ...16
3.2 Investment in capacity ...17
Chapter 4 MODEL SPECIFICATION ...21
4.1 Empirical specification ...21
4.2 Data ...22
Chapter 5 EMPIRICAL RESULTS AND DISCUSSION...24
5.1 Results ...24
5.2 Validity and reliability...28
5.3 Future capacity changes in the European power sector...29
5.3.1 Differences in technology ...29
5.3.2 Uncertainty...30
5.3.3 The development of the capacity in the European power sector ...30
5.3.4 Environmental effects ...31
Chapter 6 CONCLUSIONS ...33
REFERENCES...34
TABLES AND FIGURES
Figures
Figure 1.1 The development of the summer capacity in the US power industry, 1985 to 2003 ...2 Figure 3.1 Allowances trading system with two firms ...15 Figure 3.2 The Business Cycle ...19 Figure 5.1 Net installed total capacity in the European Union (15 countries), 1985 to 2002...31
Tables
Table 4.1 Definitions and Sources of the Variables ...22 Table 5.1 Results...24
Chapter 1 INTRODUCTION
1.1 Background
In 2005 a trial period with tradable allowances of carbon dioxide (CO2) will begin in the EU. This period of trial is based on obligations which follow the Kyoto Protocol, a pro- tocol that was adopted in December 1997 and which the European Union (EU) has rati- fied (Swedish Ministry of Industry, Employment and Communication, 2004). The pro- tocol is a legally binding commitment between concerned nations to limit greenhouse gas emissions, where CO2 constitutes the majority of such emissions. This controlling act is going to take place during a first period of five years, from 2008 to 2012. During these years the concerned nations in the Kyoto Protocol, those nations that have ratified the protocol, are going to reduce their yearly emissions of greenhouse gases to at least five percent below the levels of the base year of 1990. This reduction is partly going to be achieved what within the protocol is called “flexible mechanisms”, where one mechanism can be trade with allowances (Grubb et. al., 1999).1 The EU and its member states will therefore use the time from 2005 to 2007 to get valuable experience before the international trade with allowances, as a result of the Kyoto Protocol, starts in 2008.
As with every attempt to control a market, a system with tradable allowances will affect companies in all sectors where CO2 is emitted. With a permit the case is that the policy makers set a limit to the total amount that is allowed to be emitted, both for the industry as a whole and for the individual firm. In most situations this means that the firms in the industry have to lower their initial emission levels or to buy more allowances.
With regard to the reduction in green house gases and more specifically the reduction in CO2, which is stated in the Kyoto Protocol, this reduction is going to affect the power
1 The other mechanisms are Joint Implementation and the Clean Development Mechanism. These three mechanisms are further discussed in chapter 2 and 3.
industry to a great extent. The reason behind this is that the power industry is a major contributor of the total emission levels of CO2 and also one of the sectors that is specifi- cally targeted in the Kyoto Protocol (UNFCCC, 1997).
Given the introduction of a tradable allowances scheme it becomes interesting to ana- lyze how the power industry will react, especially regarding investments in new capac- ity. This can be translated as the change in capacity during a certain period of time.
Since the system with allowances has not yet started, it is difficult to forecast how the future change in capacity is going to be affected. Therefore, it is of great value to study past experiences with allowances trading. One of the few past experiences is the United States (US) Sulfur (SO2) Emission Trading Program, also known as the Acid Rain Pro- gram, which began in 1995 and is going to be fully implemented in 2010. The main ob- jective of the program is to reduce the emission levels of SO2 in electric generating fa- cilities, the power industry, in the US (Benkovic & Kruger, 2001). The consequences of the Acid Rain Program will therefore give help to shed light on what possibly will hap- pen with the power industry if a similar program is being implemented.
The change in capacity in the US electric power sector is therefore the key focus in this study. This change is shown in figure 1.1 during the years included in this study, 1985 to 2003. In 1985 the summer capacity was 655 Gigawatts. 18 years later in 2003 the ca- pacity had reached 923 Gigawatts, an increase in capacity with 45 percent. The change can be divided into three trends. Between the years 1995 to approximately 1999 there seems to have been a lower growth in capacity compared to prior years. This is followed by a more rapid growth in capacity compared to both of the earlier parts of the major trend. According to these observations the change in capacity has behaved differently during the Acid Rain Program, compared to before the program was initiated. If the program to some extent is an explanation to this change in capacity, and if the change has been a positive or a negative one overall, is a question for a further study, including important factors that play a part in the change in capacity in the power sector. This is done in the thesis’s following chapters.
Figure 1.1 The development of summer capacity in the US power industry, 1985 to 2003
0 100 200 300 400 500 600 700 800 900 1000
1985 1987 1989 1991 1993 1995 1997 1999 2001 2003
Year Net summer capacity (Gigawatt)
Source: EIA, 2004
1.2 Purpose
The main purpose of this study is to analyze if the Acid Rain Program has had any im- pact to the capacity changes in the US power industry. The study will also analyze pos- sible implications of the future allowances trading scheme within the Kyoto Protocol for the European power sector, based on the development of capacity changes during the Acid Rain Program.
1.3 Methodological framework
The theoretical framework for this study is environmental economic theory regarding pollution control policy instruments. The theoretical discussion focuses on allowances trading and a comparison with other forms of instruments is being done. Investment theory is also briefly discussed in order to explain investment behavior in the power sec- tor.
Given the theoretical discussion and data regarding the electricity sector in the US, an ordinary least square (OLS) regression is used to explain changes in the investment pat- tern during the period with a trade in allowances. The purpose of the model is to explain if the selected variables can explain changes in the investment pattern during the period with tradable permits. The model contains data of the change in yearly capacity and dif- ferent input costs as the price of electricity, coal, natural gas and oil. The demand for electricity, level of interest rate and policy factors are also included in the model.
1.4 Scope
The data used range from 1985 to 2003 and covers the US power sector. Earlier data are not being used in the study since it is likely that such data may be influenced by the Oil crisis in the seventies, and therefore not showing normal investment behavior.
1.5 Outline of thesis
Chapter 2 gives a background to the Kyoto Protocol and the Acid Rain Program. In the third chapter the theory behind the pollution control mechanisms and theory behind in- vestments are studied. In chapter 4 the model that is being described. The data that are included in the model are also introduced and discussed. Chapter 5 presents the results and chapter 6 sums up the main findings.
Chapter 2 BACKGROUND
In this chapter the study is placed in a context, which explains the current status of the chosen field of study. Firstly, the Kyoto Protocol is presented and secondly the US ex- perience of a trade in allowances, the Acid Rain Program, is discussed.
2.1 The Kyoto Protocol
Before the end of the 1950’s greenhouse gases2 and their influence on the earth’s at- mosphere attracted little interest by both scientists and the public. In 1957 though, sci- entists began to monitor the increasing concentration of CO2 in the atmosphere. During the 1960s and 1970s concern was growing about the warming effects that the green house gases would have on the climate and the humanity. This concern gave birth to the First World Climate Conference in 1979, which aimed to stimulate research in the area of global warming. Environmental issues became more and more popular at this time and as a consequence of this concern, the Intergovernmental Panel on Climate Change (IPCC) was established in 1988. The purpose of the IPCC was to provide assessments for governments based on the latest knowledge in the field (Grubb et. al., 1999). In 1990 the IPCC presented its first report. The report provided scientific knowledge about cli- mate changes and the possibility of following environmental and economic conse- quences with such changes. It also provided possible measures to prevent the situation.
This report was important both for the policy development and the public awareness regarding the negative effects following a change in the climate (SOU 2003:60). The report also made the conclusion that the rising concentration of greenhouse gases in the atmosphere was caused by human activity, or so called anthropogenic emissions (Grubb et. al., 1999).
2Carbon dioxide (CO2), Methane (CH4), Nitrous oxide (N2O), Hydro fluorocarbons (HFCs), Per fluoro- carbons (PFCs) Sulfur hexafluoride (SF6) (UNFCCC, 1997)
At the Second World Climate Conference in 1990 the assessments and recommenda- tions from IPCC were accepted and representatives from the participating countries of the conference, requested that the United Nations (UN) would begin negotiations on an international agreement. The requested agreement would later become the UN Frame- work Convention on Climate Change (FCCC) and was signed at the Rio Earth Summit in 1992 (ibid). According to the UN FCCC the industrialized countries should conduct a national climate policy that aims to prevent climate influence (SOU 2003:60).
In Berlin 1995 the first annual Conference of Parties was held with the parties in the UN FCCC. At the conference it was agreed to establish a pilot phase of legal binding com- mitments among the industrialized countries. It was also agreed that the commitment under the UN FCCC was insufficient and there needed to be discussion how to take ap- propriate action for the period after 2000, such as setting quantified emission limitations and reduction objectives within a specified timeframe. These new negotiations should be completed as early as possible in 1997. This resulted in the Kyoto Protocol that was adopted in December 1997, at the third annual Conference of Parties in Kyoto (Grubb et. al., 1999).
2.1.1 Commitments
The Kyoto Protocol, as earlier mentioned, aims to control the green house gas emis- sions. The main commitment, which is legally binding, limits emission levels during the period of 2008 to 2012. The commitment applies only to Annex I3 countries and those countries’ individual commitments, or level of reduction, are listed in Annex B4 in the protocol. The countries in Annex I are according to the protocol obligated to reduce their overall emissions of green house gases with at least 5 percent below 1990 levels, as an average during the commitment period. However, many countries especially USA, Japan and the EU countries decided during the negotiations to make a lager reduction.
3 These countries are (in 1997): Australia, Austria, Belarus, Bulgaria, Canada, Croatia, Czech Republic, Estonia, EU, Hungary, Iceland, Japan, Latvia, Liechtenstein, Lithuania, Monaco, New Zealand, Norway, Poland, Romania, Russian Federation, Slovakia, Slovenia, Switzerland, Turkey, United States of America (SOU 2003:60)
4 All countries in Annex I except for Belarus and Turkey (ibid.)
Other countries, especially Australia and Iceland, were on the other hand allowed emis- sion levels above that of the base year 1990. For the Russian federation, among other countries, no change would be made compared to the 1990 emission levels (Grubb et.
al., 1999).
According to the protocol a group of countries can form a “bubble” where the total emission out from the “bubble” is at least five percent below 1990 levels. However, the distribution of the reduction in emission can freely be decided between these countries.
An example of this is the EU Bubble where the eight percent reduction for each EU country works as a level for the EU as a whole. This implies that the distribution of the emission reductions varies from country to country in the EU. Denmark and Germany among others have to make a larger percentage reduction, whereas countries like Swe- den, Greece and Spain are allowed to emit more compared to 1990 levels (ibid).
The commitment is put into action only if it is ratified by at least 55 countries that con- stitute 55 percent of the total level of emissions in Annex B (Ellerman, 2000). In 2003 the protocol had been ratified by 108 parties, which were meeting the first of the above stated requirements. However, these parties only constituted 42.9 percent of the total emissions that is stated in Annex B and since the USA in 2001 announced that it was not interested in ratifying the protocol, the only chance for the Kyoto Protocol to be- come operational was that the Russian Federation, which contributes with 17 percent of the emissions in Annex B, would ratify the protocol (SOU 2003:60). This happened in October 2004 when the Russian Duma ratified the protocol and thereby making way for the protocol to go into effect (UN, 2004).
The protocol also states that the parties should make progress in actually achieving its commitments no later than 2005 (Grubb et. al., 1999). This is now partly done in the EU, as has been discussed in chapter one, during a trial period with tradable allowances.
2.1.2 Flexible Mechanisms
To achieve the reduction in the emission of green house gases, the protocol states three basic mechanisms; Emission Trading, Joint Implementation and the Clean Development Mechanism. These mechanisms are also mechanisms for international transfer (ibid.).
According to Article 2 in the protocol, the mechanisms are means for energy efficiency and to stimulate research on new technologies and renewable energy.
Emission Trading
This is a trade in, or an exchange of, emission allowances between parties in the proto- col. Parties in Annex B may participate in emission trading to fulfill their commitments.
Emission trading implies a redistribution of the emission allowances (ibid.). This mechanism is being discussed in detail in chapter 3.
Joint Implementation
This is a system of investments in projects between Annex I countries that reduces emissions. The country that makes the investment can claim the reduction to meet its own commitment. In other words, allowed emissions are transferred from the host coun- try to the investing country, which is similar to emission trading. The receiver of the investment has an incentive to participate because of the positive economic and envi- ronmental effects following the investment. All investments have to be sanctioned by both the concerned parties’ governments to become legal in the context of the Kyoto Protocol (ibid.).
Clean Development Mechanism
This is a flexible mechanism similar to Joint Implementation. The difference is that the Clean Development Mechanism allows Annex I countries to invest in non-Annex I countries, i.e. the worlds’ developing countries. The investments are aimed to achieve sustainable development in these countries and also to help the Annex I countries fulfill their commitments. Thus, the developing countries shall benefit from these investments and the reduction in emission is added as allowed emission for the investing country.
This requires a long-term control that the reduction results in a real climate effect and also that the reduction in emission is a reduction that otherwise would not have occurred (ibid.).
2.2 The Acid Rain Program
The term acid rain refers to when emissions react with water, oxygen, and other chemi- cals in the atmosphere. This process results in various acidic compositions that fall out of the atmosphere, sometimes several hundreds of miles from the emission source.
These compositions can take either a wet or a dry form and affect humans, wildlife, plants, lakes, streams and even buildings in a harmful way5. The main cause of this en- vironmental problem is anthropogenic emissions of SO2 and nitrogen oxides (NOx) (EPA (A), 2004). Thereby, the emissions can be a subject for regulations.
2.2.1 The History behind the Program
The Acid Rain Program has its roots from the Clean Air Act Amendments (CAAA) program, dated 1970. The program was the first US federal air pollution legislation of importance and was implemented to set national maximum standards of SO2 and other emissions6 in the USA, which was related with the generation of electricity. At this point of time the concern was about human health and welfare and not about the envi- ronmental effects. However, many states could not comply with their obligations, mainly because of the extended use of older coal plants. This gave as a result that Con- gress in the 1977 CAAA made adjustments answering to this issue and also to respond to the new concern arising from environmental groups. After this period of time acid rain grew as both an environmental and a political issue, resulting in several financed research projects conducted by the National Acid Precipitation Assessment Program (NAPAP), established in 1980. Acid rain was now not only associated with negative human effects, but also negative environmental effects. Despite the new scientific knowledge gained from the research projects during the 1980’s any further proposal to control the emission of SO2 was a subject of great conflict and dispute. This situation seemed to change when George Bush became president, in 1988. It is argued that the new president and his administration were more environmentally conscious, compared to his predecessor Ronald Reagan. The Bush Administration came up with a clean air proposal that included a trade with emission allowances of SO2 in 1989. After negotia- tions between the Senate, the President’s Administration, the House of Representatives
5 For example; respiratory diseases, dissolving nutrients from trees, acidic levels in lakes and streams that can be deadly for aquatic life (EPA (A), 2004).
6 Carbon monoxide, nitrogen, dioxide, particulates, ozone and lead (ibid.)
and different interest groups, a modified version of the clean air proposal earlier written by the Administration, was passed both by the Senate and the House of Representatives and finally signed by the President in late 1990. Included in this legislation was the title IV in the 1990 CAAA, the Acid Rain Program, which included the system with emis- sion allowances concerning the emission of SO2 (Ellerman et. al., 2000).
2.2.1 Title IV in the 1990 CAAA, a market based solution
To reduce the negative effects caused by acid rain the Acid Rain Program was aimed to lower the total emission level of SO2 by half the amount recorded in 1980, by the year 20107 (EPA (B), 2004). This was going to be achieved through a system with a market based cap and trade. This implies that the total emission from electricity generating units is subject to a cap, a maximum allowed level of emission, and that the emission allowances can be traded freely amongst the parts in the program8. This is going to be implemented in two phases (Schmalensee et. al., 1998). Phase I was running from the years 1995 to 1999 and the second phase is running between the years 2000 to 2010 (EPA (B), 2004). The two phases differ in the numbers of plants included. During Phase I only the 263 biggest and filthiest SO2 emitting generating units, located in 110 gener- ating plants, were affected by the program. The plants were required to reduce their yearly emission by approximately 3.5 million tons. Phase II includes almost all fossil- fueled electric generating plants in continental USA. This phase also implies a larger yearly reduction of almost 9 million tons yearly. It is not until the end of this phase in 2010 that a permanent emission cap regarding SO2 takes effect (Ellerman et. al., 2000).
The initial distribution of allowances is based on historical data and is distributed yearly and is conducted by US Environmental Protection Agency (EPA). One allowance per- mits the holder to emit one ton of SO2 in any given year. Thereby, the allowances can be banked for later use but cannot be borrowed from the future. At the end of January the data of the last year’s emission is handed in to EPA. If the required target of the level of emission is not met, there exists a penalty (ibid.).
7 Title IV also included a reduction of NOx by two million tons below 1980 levels (Ellerman et. al., 2000).
8 See section 3.1 for a further discussion
The affected units have a number of choices in order to achieve the emission reduction.
A unit can install flue gas desulfurization systems, also known as “scrubbers”. There are several different technologies9 that rely on either a dry or a wet system. Besides being an expansive alternative, the system with scrubbers has its advantages. One advantage is that it is relativity easy to shift production between units within a plant, from a unit with scrubbers to a unit without such facilities. Another way of complying is changing fuel to a low-sulfur option. This strategy is much cheaper than the alternative with scrubbers. It also gives an option to invest in scrubber systems at a later date. However, changing fuel can require investments in equipment modifications so that the plant’s different parts can handle the change in fuel. There is of course also another way of meeting up with the reduction and that is that a plant can buy allowances from another plant.
Thereby, a plant that buys allowances can emit more than its original assigned level of emission cap (Diltz, 2002). Moreover, the plant owners are free to choose any amount of allowances transactions or reduction technologies to meet each year’s requirements (Ellerman et. al., 2000).
There also exists an auction of allowances, conducted annually by the EPA. The amount of allowances is about three percent of the total amount of the assigned yearly quantity.
The auction is a mean to assure that new units have a public source of allowances other then the initially assigned amount. It also helps the EPA to set the price of allowances that is used when trade is initiated between generating units (EPA (A), 2004). The auc- tion of allowances is also not only a matter for the affected utilities, anyone can buy or sell allowances. This has resulted in brokers buying allowances in hope of a future price rise and environmentalists buying to further reduce the level of emission (Ellerman et.
al., 2000).
2.2.1 Results and experiences
The program has worked well so far with almost 100 percent compliance. Several posi- tive environmental effects have also been observed (EPA (B), 2004). SO2 emissions had been reduced far more than required during the first phase and this can be seen as a di- rect effect of the program. This situation has created banked allowances that can be used
9 Dry-injection, spray-dryer, wet-limestone and wet-lime systems (Diltz, 2002).
to smooth excess emissions in Phase II and if a larger early reduction compared to a later one is preferred this can be seen as an additional positive environmental effect (Ellerman, 2000). The cost of compliance has been assumed to be 25 to 34 percent lower compared with a scenario without trade (Schmalensee et. al., 1998).
Furthermore, concerns over the development of a market have proven to be unneces- sary. The market for allowances has progressed rapidly and has encouraged the belief that a market will appear when there is a need. This is argued to have much to do with the fact that the allowances have been viewed as property rights. The reason behind this is the long scope of time, up to 30 years, an allowance is valid for (Ellerman, 2000).
The success of the Acid Rain Program has as a result made the use of a trade with al- lowances a serious and popular alternative to achieve environmental goals, not only by economists but also amongst environmentalists. This has for instance resulted in the use of a trade with allowances as the most important ingredient of the Kyoto Protocol (ibid.).
Chapter 3 THEORY
In this chapter the theoretical explanation of a trade with allowances is presented. The chapter also includes a discussion regarding important parts that have to be reconsidered when making investment decisions.
3.1 Allowances trading
A trade with allowances is one of the economical pollution controlling instruments available to achieve an emission reduction target. One allowance gives a polluter the right to emit a certain quantity. Other instruments of importance are taxes and pollution abatement subsidies. The major difference between tradable allowances and the other two is that there exists a cap within a system with tradable allowances, where the cap is equal to the amount of issued allowances. Taxes and pollution abatement subsidies, also called charges, are more focused on price differences between controlled and uncon- trolled emission and the effect these differences have on the emitted quantity (Perman et. al., 1999).
Within a system with tradable allowances no unit is allowed to emit more than its issued allowances, possibly followed by a penalty if the rule is broken. The size of the penalty plays an important role in achieving the cap and the environmental goal. If the size of the penalty is small a firm may emit more then its initiated allowances. Thus, paying the penalty instead of buying further allowances or finding technologies to reduce the emis- sion can become more profitable for the company (Ellerman, 2000). The controlling au- thorities decide how the allowances are first distributed, for example the EPA in the USA. The allowances can be handed out with or without the historical levels taken into consideration. Taking historical emission levels into account is also referred to as grand- fathering and has been the common case in past initiated allowances trading schemes (Baumol & Oates, 1988). Moreover, the allowances are required to be freely transfer- able so that a market for buying and selling allowances appears (Tietenberg, 1992).
3.1.1 Graphical example of an allowances trading system
In figure 3.1 a hypothetical system with tradable allowances is demonstrated graphi- cally. It shows two firms where the first firm has the marginal costs of controlling abatement MC1 and the second firm has the marginal cost MC2. If there is no reduction in emission the two firms emit fifteen units each. The authority then sets a cap on the total level of emission and distributes the allowances according to e.g. the grandfather- ing principle. The first firm is thereby in this example allowed to emit seven units, the firm gets seven allowances, and the second firm is allowed to emit eight units, equal to eight allowances. Thus, the first firm has to control eight units and the second firm has to control seven units. The marginal cost of the first firm is given by point A and the cost of the second firm is given by point B. Since the different marginal costs between the two firms, point C is higher than point A, there exists an incentive for trading allow- ances. In other words, it is cheaper for the first firm, according to its relatively flat mar- ginal cost curve, to reduce emission compared to the second firm, which has a steeper marginal cost curve. The first firm can profit by selling its distributed allowances for a price higher than A and the second firm could lower its costs by buying allowances from the first firm, at a lower cost than C. Because of these circumstances a market of allowances will appear between the two firms. The trade with allowances will go on un- til point B is reached, where the two firms cannot be better off trading allowances with each other. At this point the first firm will have reduced its emission with ten units and the second firm with five units. Thus, the final distribution of the allowances is five al- lowances to the first firm and ten allowances to the second firm (Tietenberg, 1992.).
Figure 3.1 Allowances trading system with two firms
0 15
1 14
2 13
3 12
4 11
5 10
6 9
7 8
8 7
9 6
10 5
11 4
12 3
13 2
14 1
15 0 MC1
MC2
Marginal Cost (dollars per unit)
Source 1
Souce 2 Quantity of
Emissions Reduced B
A C
Source: Tietenberg, 1992
3.1.2 Cost efficiency
The final allocation of the allowances between the two firms has been reached at a minimum cost and is thereby a cost-efficient allocation, since both of the firms’ mar- ginal costs have been equalized in point B. Any other mix in emission reduction will in this case lead to a higher combined cost for the firms. With a charge this is also the case but with an important exception, uncertainty. To reach a cost-efficient allocation with a pollution control instrument that relies on charges, each firm’s marginal cost curves have to be identified by the authorities. If the firm’s marginal cost curves are not identi- fied it is hard for the authorities to set the right amount of charge to reach the desired environmental goal (Baumol & Oates, 1988). This results in a trial-and-error process over time to find the proper charge, whereas in a system with allowance trading the market interaction automatically results in the right price of an allowance (Tietenberg, 1992).
3.1.3 Tradable allowances compared to charges
Tradable allowances also have characteristics, other than cost-efficiency, making it a more preferable alternative in most cases compared to charges. Both Baumol & Oates (1988) and Tietenberg (1992) describe such characteristics, including the earlier discus- sion about differences in reaching cost-effectiveness. One of these characteristics con-
cerns inflation and growth in production. Inflation lowers the real value of charges and an increased production increases the demand of emission, thus forcing the authorities to make adjustments of the charges, which can be costly. With allowances, inflation and increased production results in a higher allowance price, assuring that the environmental goal is reached.
Another characteristic is uncertainty of the damage caused by the pollution aimed to control. If the negative consequences of a deviation from the environmental target are large, the most important thing is that the emission level does not exceed this target.
This is done more securely with tradable allowances than with charges, because of the necessity of knowing the marginal cost curves to reach an environmental goal when a charge is used. However, charges have some desired effects since it is a source of in- come for the public. If allowances are initiated according to the grandfathering princi- ple, as has been the common case, it involves no source of income to the public, it is a free initial distribution. Despite the problems reaching a cost-efficient allocation with charges compared to tradable allowances, charges can thereby be used to reach an envi- ronmental goal both in a cost-efficient manner and to serve as public revenues. How- ever, auctions can be combined with the grandfathering principle, which has been the case within the Acid Rain Program, and thus bring in an income to the public.
One area where no distinction can be made between the two instruments is the existence of incentives for technical progress. New technologies that lower the cost of emission reduction are desired both in the case with tradable allowances and charges. With al- lowances new emission reduction technologies can make the firm’s initial distributed allowances superfluous and can thereby be sold to other firms not possessing these tech- nologies. With charges the lower emission caused by the new technologies results in fewer charged emitted units and as a result lowering the firm’s costs.
3.1.4 Effects on investments
All pollution controlling systems hinder, in most cases, the optimal level of emissions for firms. In figure 3.1 the optimal level of emissions for the firms are fifteen units each.
However, this level of emission is not, in this example, considered optimal from soci- ety’s point of view. Thus, an environmental goal is set up, using a pollution controlling
system to reach the desired goal. How this controlling act affects investments has to do with different factors. In the section below some of these factors are tried to be identi- fied.
One factor is the availability of emission reduction technologies and the cost of using these. If the reduction target is set relatively strict and no proper technologies at a proper price are available, the possibility of a lower production or a slower expansion in pro- duction can be the case for the firms targeted. On the other hand, as has been discussed in the section above, both an allowances trading system and a system relying on charges gives strong incentives for technical progress, so this discussion is ambiguous.
The preceding aspect is related to the actual cost of emission abatement, the size of the charges and the price of allowances. If the size of the charges is high, production, which results in emitted units, can be costly. The situation within an allowances trading system is somehow different. The initial distribution of allowances can be without costs for the firms. However, if an effective market does not appear this can result in a high prize of an allowance and thereby be costly for the targeted firms. These things can also be con- sidered to affect production and thereby to some extent investments. For this aspect a similar discussion concerning technological progress can be considered.
The conclusion of this brief discussion is that pollution controlling instruments hinder the normal production and in most cases introduces costs for the firms that otherwise would not occur. This is analyzed in later chapters taking the Acid Rain Program as an example if differences in investment behavior in the US power industry can be noticed during this program.
3.2 Investment in capacity
In order to build a model that tries to explain the investment behavior in the US power industry, the factors that can be considered to play a role in investment decisions have to be studied.
An investment means that a power company changes its long-run capacity, though the problem is to explain this theoretically. According to the report Nordleden (2002:4), in-
vestment decisions in the power industry are a complex issue. One approach states that investment decisions in the power industry can be made by examining marginal costs for respective technology, on the other hand a more modern approach assumes that the power industry has the same decision making problems as every other commercial company. However, according to de Vries and Hakvoort (2002), electricity has charac- teristics that differ from other products. Firstly, electricity cannot be stored successfully in a scale that is profitable. Secondly, the short-term price elasticity of demand is low, thus there is a limited possibility to substitute electricity with other products. This makes the investment behavior for the power industry different from other sectors. Nev- ertheless, in this study it is assumed that there are some basic factors that are important for the companies’ investment decisions and these are briefly discussed below.
Firstly, some microeconomic considerations are being done. An essential part in the firms’ investment decisions is the demand for the product. In the case with the power industry this is represented by the demand for electricity. Another part is the prices of inputs and outputs, in other words the cost of the fuel used when producing electricity and the charged price of the produced electricity (Schwartz & Trigeorgis, 2001). With an increase of demand output will increase that can lead to investments in for example new capacity. If input prices become more expensive the production costs increases, which can lead to a reduction in investments and thereby a smaller change in capacity.
If the price of the produced output increase it becomes more profitable for the company to produce the product. This naturally increases the will to invest in new capacity.
Secondly, not only factors associated directly to the companies’ microenvironment af- fect investment decisions, the surrounding macroeconomic environment also has to be taken into account. One part of this environment is the different stages in the economy, also known as the business cycle (Mayo, 2003). The business cycle can be defined as
“…repeated fluctuations in employment, output and the composition of output, associ- ated with a certain typical pattern of co-movements in prices and other variables.” (Lu- cas, 1981 p.232). The business cycle is graphically shown in Figure 3.3.
Figure 3.2 The Business Cycle
Source: Mayo, 2003
The economy expands to a certain level, the peak. After this point the economy de- clines. This stage in the economy is the recession, where the level of employment and output declines. After this phase, when the recession has hit the bottom, there is a new expansion in the economy (Mayo, 2003). Thus, in a period of recession the investments decline because of the drop in output and employment, therefore there are no incentives to invest in new capacity. The opposite can be said when the economy is in a stage of expansion.
If the business cycle has this effect on investments, it is of interest to be able to measure the economic activity. This is done with aggregated indicators such as the level of pro- duction and national output, as can be seen in figure 3.2. Alternative measures are fo- cused on prices and unemployment, where unemployment is considered the most impor- tant factor. The unemployment is often measured in the percentage rate of unemploy- ment and measures output lost due to the decline in economic activity (ibid.).
The macro-economic environment also includes other factors such as inflation, mone- tary policy and fiscal policy. Inflation is a measure of the price development in the economy. The inflation can have different effects on investments. One effect is that an increase in the inflation rate increases the interest rate. This increase makes investment less attractive. An increase in the inflation rate also makes the price of inputs go up, which makes production more expensive. Monetary policy is used to alter the economic activity and changing the interest rate is the method. A higher interest rate makes it
more expensive for companies to finance investments. A higher interest rate also in- creases consumers’ incentives to save instead of consume. These effects lower the ag- gregated demand. The opposite discussion is valid for a lower interest rate. Fiscal policy is the use of taxes, regulations and government expenditures to try to change the eco- nomic activity.
It is not only the above stated factors that influence investment in capacity in the power industry, it is also the uncertainty associated with these factors. This is uncertainty about which policy – such as future regulations, future environmental rules and maximum price – that is going to be implemented by the policy makers. A higher degree of uncer- tainty is linked with a higher risk, thereby probably generating lower investments (de Vries and Hakvoort, 2002). Radetzki (2004) writes about this issue in the context of the conducted Swedish energy policy during the last 30 years. In the study it is concluded that, because of the irrational behaviour of the policy makers, the energy market in Sweden has taken damage. The policy makers have used the energy policy as a tool to achieve goals different from a properly working energy market and they have also in- tervened to a greater extent than is valid for. This has caused an instability that has re- duced the efficiency in the Swedish electricity market. Such efficiency is especially damaging for the energy sector where investments take a long time to implement and also considering the relatively long operational time of the invested capacity. Such in- stability results in lower investments and Radetzki argues that the former conducted en- ergy policy in Sweden is going to lead to problems for the electricity supply in Sweden the coming years.
The real effect uncertainty has regarding investments is hard to measure, though it has to be taken into consideration when analysing and explaining investment behaviour.
Chapter 4
MODEL SPECIFICATION
This chapter states the model that is being used to examine if a trade in allowances has had an effect of the changes in capacity in the US power industry. An explanation of the different variables included in the model and their sources is also being done.
4.1 Empirical specification
Taking section 3.2 into consideration, a model attempting to explain the changing ca- pacity in the electric power industry can be built up. There are no theoretical reasons to suspect some specific functional form of the regression model, thus this study assumes a nonlinear relationship. The explaining variables are lagged one period. The rational be- hind this assumption is that investment decisions, to some extant, are based on historical circumstances. The model can be written as shown below:
υ α
β β
β β
β β
β β
8 7
6 5
3 4 2
1
) (
) (
) (
) (
) (
) (
) (
) (
1 1
1
1 1
1 1
ACIDRAIN UNEMPL
FED DEMAND
ELECTRP OILP
COALP NAGASP
CAP
−
−
−
−
−
−
= −
(1)
CAP is the dependent variable in the model and α is a constant. NAGASP, COALP, OILP, ELECTRP, DEMAND, FED , and UNEMPL are the explanatory variables and ACIDRAIN is a dummy variable. The parameter υ is a disturbance term and β1 to β8 are coefficients. An explanation of the different variables is being done in table 4.1.The model can be transformed into a form that is linear in parameters:
υ β
β
β β
β
β β
β α
log ) log(
) log(
) log(
) log(
) log(
) log(
) log(
) log(
log log
8 1 7
1 6
1 5
1 4
1 3
1 2
1 1
+ +
+ +
+ +
+ +
+
=
−
−
−
−
−
−
−
ACIDRAIN UNEMPL
FED DEMAND
ELECTRP
OILP COALP
NAGASP CAP
(2)
4.2 Data
The data used in the model are data for the electricity sector in USA during the period 1985 to 2003. In table 4.1, descriptions and the sources of the variables are explained.
The data have been collected from the Energy Information Agency (EIA), the Bureau of Labor Statistics (BLS) and the Federal Reserves. These are all US government authori- ties, which means that the reliability of the collected data can be considered as relativity high.
Table 4.1 Definitions and sources of the variables
Variable Definitions Sources
CAP The maximum capacity during the summer peak demand in the US power industry, measured in Gigawatts
EIA (2004)
NAGASP The real natural gas price, in chained (2000) US dollars, in the electric power sector, measured in dollars per thou- sand cubic feet
EIA (2004)
COALP The average real coal price, in chained (2000) US dollars, measured in dollars per short ton
EIA (2004)
OILP Crude oil refiner acquisition costs, in chained (2000) US dollars, measured in dollars per barrel
EIA (2004)
ELECTRP The average real retail price, in chained (2000) US dol- lars, of electricity measured in cents per Kilowatt hours
EIA (2004)
DEMAND End use of electricity, measured in thousands Kilowatt hours
EIA (2004)
FED Federal funds rate Federal
Reserv (2004) UNEMPL The annual unemployment rate of people over 16 years. BLS (2004)
ACIDRAIN Dummy variable for the Acid Rain Program, with the value 0 for the period before and the value 1 for the pe- riod during the program
All the variables included in this model have been selected based on the discussion in section 3.2. This means that the model includes both micro and macro economic vari- ables. The micro variables are first being explained in the following part.
When producing electricity from fossil fuels the inputs are coal, natural gas and petro- leum. The prices of these inputs are likely to in some extent influences the investment decisions in the electric power industry, and are therefore included in the model. Prices are shown in chained (2000) dollars which is a measure used to express real prices (EIA, 2004). The price of electricity is also in chained (2000) dollars and shows how the output produced by the power industry has changed in value during the period. The prices also reflect the inflation rate in the economy.
In the model it is assumed that the electricity demand is equal to the total consumed electricity in the USA. This is called Electricity Overview in the EIA statistics and in- cludes domestic produced electricity as well as imported electricity.
As being discussed in section 3.2, the interest rate is a key player concerning investment decision. The interest rate is represented by the federal funds rate in the model. This is the rate the Federal Reserve uses to alter the supply of money and credit. The Federal Reserve also uses the discount rate for this task. However, a change in the discount rate is seldom made and such a change is argued to have a more symbolic than a real effect on the economy. Therefore, the federal funds rate rather than the discount rate is in- cluded in the model. The federal funds rate is the rate of interest a bank charges another for borrowing reserves, while the discount rate is the rate of interest the Federal Reserve charges banks for borrowing reserves. Another macro economic factor that can explain the investment pattern, and thereby the change in capacity in the power industry, is un- employment. The unemployment rate can also be seen as an indicator of the state of the economy, if there is a recession or an expansion in the economy (Mayo, 2003). The data of the unemployment rate is for the whole population in the USA.
The effect of the Acid Rain Program is modeled as a dummy variable. The variable is in other words used to differ the two periods, before the Acid Rain Program and the period from 1994 to 2003. Officially the Acid Rain Program started in 1995 but EPA has con- ducted allowances auctions since 1993 (Ellerman et al., 2000). Therefore, to measure the change in capacity due to a trade in emission allowances, the dummy variable has the value 1 from 1994 to 2003. The reason for lagging the variable follows the same idea as have been discussed in 4.1. The idea is that investment seems to respond one year after changes of some kind in the economic environment.
Chapter 5
EMPIRICAL RESULTS AND DISCUSSION
5.1 Results
In this chapter the results from the regression analysis of the model in chapter 4 are pre- sented. The results from each variable are shown in table 5.1.
Table 5.1 Results
Variable Coefficient Expected Sign t-statistics Significance Constant -7.344
Natural Gas Price -0.0372 - -0.8424 0.4214
Coal Price -0.1893 - -0.858 0.4091
Oil Price 0.0689 - 1.8194 0.1022
Electricity Price 2.0097** + 4.0740 0.0028
Electricity Demand 1.1439** + 3.7529 0.0045
Federal Funds Rate -0.1502** - -5.6561 0.0003
Unemployment -0.3443** - -4.1323 0.0026
Acid Rain Program -0.0518* - -2.5524 0.0311
R2-Adjusted 0.9667
F-statistic 62.6301
Durbin-Watson 2.2024
* Statistically significant at a 5% level
** Statistically significant at a 1% level
The regression analysis consists of 18 observations, which equals the number of years from 1985 to 2003. The value of the R2-Adjusted measures to what degree the explana- tory variables can explain the variance in the dependent variable, from 0 to 1. A large value explains to a greater extend than a low value. The adjusted R2 value takes into consideration the number of parameters and automatically lowers the R2 value with re- gard to the number of parameters included in the model (Dougherty, 1992). In table 5.1 the value of the R2-Adjusted shows to be 0.9667. This means that 96.7 percent of the variance in the change in capacity can be explained by the explanatory variables in the model. The rest of the variance, 3.3 percent, cannot be explained by the explanatory variables. Hence, the model has a good fit. However, even though the model has a god fit it is a simplification of the real relationship, like all other models are. This, for in- stance, implies that there are omitted variables that ought to be included in the model to state the real relationship.
A problem when using time series in a regression analysis, such as the stated model, is autocorrelation. This means that the disturbance term is influenced by other variables omitted in the model. As a result the regression coefficients become inefficient and their standard errors are estimated wrongly. There are two types of autocorrelation, positive and negative. Positive auto correlation is the most common type of auto correlation in economic analysis, and is a result of observations that seem to follow a cycle. Negative auto correlation is when one positive value is followed by one negative and so on. This type of autocorrelation is uncommon in economic situations and the reason for negative autocorrelation is mainly when the model have been transformed to fit in the regression analysis in a better way (ibid.). A method to measure autocorrelation is the Durbin- Watson stat. If the value is close to 2 it can be assumed that there exists no autocorrela- tion. Values under and over 2 are translated into a negative respectively positive auto- correlation. In table 5.1 the value of the Durbin-Watson stat is 2.2024 and this indicates a negative autocorrelation. This value slightly lies between the upper and lower critical values of the Durbin-Watson Statistic, 3.48 and 2.20. Hence, the null hypothesis cannot be rejected. In other words, there may or may not exist autocorrelation. There are sev- eral methods to lower the autocorrelation. One is to reconsider the model specifications and change the variables used in the model. After revising the variables statistical tech- niques dealing with autocorrelation can be used. One of these techniques is the AR(1) technique (Eviews, 1996). When using this correction technique on the model, no sig-
