The Emissions Trading Scheme and Renewable Energy Source Support in the EU: Two minds, one body?

(1)

Leiden University – Faculty Governance and Global Affairs

MSc Thesis

The Emissions Trading Scheme and Renewable Energy

Source support in the EU: Two minds, one body?

An empirical analysis of the spatial effect of the interaction of EU

ETS and RES-support in the EU

Student: Frederik

Student number: s<=>=?@>

Date: BB-=>-<=B?

Word count: B<,>FB words (ex. references and footnotes) Thesis supervisor: Dr. P.W. van Wijck

(2)

(3)

DECLARATION BY CANDIDATE

I hereby declare that this master thesis, “The Emissions Trading Scheme and Renewable Energy Source support: Two minds, one body?” is my own work and my own effort and that it has not been accepted anywhere else for the award of any other degree or diploma. Where sources of information have been used, they have been acknowledged.

Name: Frederik Lubbersen

BB-=>-<=B?

(4)

(5)

Abstract

To lower greenhouse gas emissions and encounter global warming and the consequences thereof, the European Union has implemented the European Union Emission Trading Scheme (EU ETS) as well as renewable energy source (RES) support instruments. Since both instruments co-exist from 2005 onwards and their goals largely overlap, the question arises what the interaction effect is of these two policy instruments is. This paper focusses on the possible spatial interaction effect, which entails a transfer of production and GHG emissions as a result of differences in carbon emission costs between member states as a result of differences in levels of support on renewable energy sources and the presence of the EU ETS and its binding cap. The interaction is tested empirically by analysing the effect of differences in levels of RES-support amongst the <? EU member states on the GHG emissions. As the effect is expected within the EU ETS, the verified emissions within the EU ETS are used as primary dependent variable. Panel estimation results on the period <==?-<=B< show that the RES-support in other EU member states than a particular one do not have an effect on verified emissions in that particular member state. The spatial interaction effect thus does not exist in the observed sample and the possible interaction effects of the combination of RES-support and the EU ETS need to be examined further.

(6)

1. Introduction

One of the greatest challenges facing mankind today is undoubtedly climate change and the implications and consequences thereof. Temperature changes, climate migrants, food and water security and sea level rise are just some of the negative effects (EEA, <=B@). This creates an ever louder call for efficient, targeted and comprehensive policies to encounter these changes and their causes. The urge to combat these negative effects to humans and the planet becomes more imperative as they materialize more extremely and suddenly.

International acknowledgement of global warming and its perils has already started to form decades ago, indicated by the founding of the Intergovernmental Panel on Climate Change (IPCC) and the adoption of the Kyoto Protocol, f= and <= years ago, respectively. The IPCC was set up in BF?? to provide policymakers with regular assessments of the scientific basis of climate change, its impacts and future risks, and options for adaptation and mitigation (IPCC, <=Bf). The Kyoto Protocol is aimed more specifically at the reduction of GHG emissions and commits industrialized countries to reduction goals in order to achieve sustainable development (UNFCCC, <=BB). It requires that all affected plants in the B@= countries that signed the Protocol limit their GHG emissions according to specified limits. More recently, in December <=Bg, during the <=Bg United Nations Climate Change Conference, the global community signed the so called ‘Paris Agreement’ on climate change, thereby committing to new climate change policies and reduction goals. This agreement of <== nations is partially binding, as submitting an emissions reduction target and the regular review of that goal is required. This agreement aims to achieve that the global average temperature stays well below a <°C, and preferably B.g°C, increase compared to pre-industrial levels. Additionally, it is expected that the ability to adapt to adverse impacts of climate change, to

(9)

fostering climate resilience and to low GHG emissions development is achieved without threatening food production. In addition, finance flows should stimulate low GHG emissions and climate-resilient development (UNFCCC, <=Bg).

Following these international agreements, many countries and international organizations have made the reduction of GHG emissions, and in particular carbon dioxide (CO#), an important goal in order to encounter global warming. Formulating policies to achieve this goal has been a significant challenge, but most view the issue as an economical problem and thus address this issue accordingly. Pollution is commonly viewed as an externality leading to market failure, as the costs of polluting is not passed on through price incentives to final energy users. Therefore, these users are unable to respond adequately and suboptimal amounts are consumed. To address this issue and reduce GHG emissions, different levels of government implement various policies. In the European Union (EU), two highly relevant policies - also the policies of interest for this research - are the national support schemes for renewable energy sources (RES) and the increasingly important and international emissions trading scheme (ETS) (EP & EC, <==f) (EP & EC, <==F).

In fact, the Kyoto protocol proposes that companies reduce their emissions in the least costly and most efficient way by using trading and market mechanisms (United Nations, BFF?). Today, inter alia New Zealand, Japan, multiple states in the USA and Canadian provinces have implemented such a scheme. The EU has the largest and best known ETS, the European Union Emission Trading Scheme (EU ETS), in order to comply with the Kyoto Protocol. Parallel to this, government support for RES is commonly applied, encouraging RES use amongst the <? European member states (EU<?)1. In the pursuit of a substantial share of RES in energy production it is used as a significant complementary measure in the effort to reduce GHG emissions. In <==?, the EU adopted its

1

In this thesis, as the RES-support considers public interventions in energy markets in all 28 EU Member States, the EU28 is the area of interest.

(10)

ambitious “<=-<=-<=” plan intending to increase the share of renewables in overall EU primary energy consumption to <= % by <=<= and a <=% cut in greenhouse gas emissions2. This EU wide plan specifies ambitious RES-targets per member state and commits them to the development and use of RES.

Since the main objective of both policy instruments is effectively the same, namely reducing emission of GHGs, the issue of counterproductive overlapping regulation arises (Tinbergen, BFg<). The interaction between RES-support and the emissions trading scheme may cause conflict or synergy, as one policy can affect the other. If so, policy coordination may be necessary to be able to control the interaction effects and ensure the effective functioning of both policy instruments in order for them to reach the goals at a minimum social cost. A large stream of theoretical and a smaller stream of empirical literature has been dedicated to the interaction of these environmental policy instruments.

However, where most empirical literature focusses on effects through the permit prices and not on spatial interactions, this research will focus on the spatial interaction effect. As the spatial interaction effect is not found in previous literature, it is defined as follows. The spatial interaction effect is the transfer of production and emissions as a result of differences in carbon emission costs between member states as a result of differences in levels of support on renewable energy sources and the presence of the EU ETS and its binding cap. Nota bene, the binding cap is the predetermined and limited number of total allowances circulating in the EU ETS that participating companies in the EU can use for emitting GHG (European Union, <=Bg). More precisely, this paper will answer the following research question:

2

Initiated in 2007, the 2020 package is meant to ensure the EU meets its climate and energy targets for the year 2020 by pursuing a 20% cut in greenhouse gas emissions (from 1990 levels), sourcing 20% of EU energy from renewables and ensure a 20% improvement in energy efficiency.

(11)

Do differences in levels of Renewable Energy Source support amongst the ;< EU member states in combination with the EU ETS lead to displacement of GHG emissions?

Using regression analysis on a dataset covering the period from <==?-<=B<, it is determined whether the level of RES-support in a certain EU member state influences the emission levels of the other member states. It will establish an empirical understanding of a spatial interaction of the two policies, by using a general panel model with the emission levels within the EU ETS as main independent variable.

Accordingly, this paper is structured as follows. The considered RES-support and EU ETS are described in part <. Next, the theoretical framework is presented in part f, elaborating on the technical aspects of EU ETS and the RES-support policies and covering the interaction literature. Further, part s explains the causal mechanism of the hypothesized adverse spatial interaction effect. Data are described in part g and part > discusses the method to test the interaction empirically. Part @ will give the results, followed by the concluding remarks in part ? and F.

(12)

2. Measures taken: RES-support and EU ETS

This section presents the unfolding and characteristics of the considered EU ETS and RES-support policies currently used in the EU.

2.1 Kyoto Protocol

The Kyoto Protocol treaty became legally binding in February <==g and commits each participating member state to the GHG emission reduction targets. Given these targets, it provides spatial flexibility through the so-called Kyoto Mechanisms: countries may buy and sell their assigned amounts of emissions, which is called international emissions trading (IET). The clean development mechanism (CDM) allows for the selling of GHG emission reductions achieved in projects in developing countries to an industrialized country. These project-based trades between industrial countries are named joint implementation (JI). The quantified reduction commitments combined with the JI creates an international system of tradable GHG permits (Springer, ;DDE).

In order to comply with the commitments made by the European Union to the protocol, the European Commission (EC) poses a suggestion in the form of the document “Greenhouse gas emissions trading within the European Union” in <===. This entails the first EU ETS scheme that started to develop (UNFCCC, ;DJJ). In <==f, the first EU ETS Directive was adopted and in the EU ETS was introduced in <==g (EP & EC, ;DDE).

2.2 The EU ETS policy

(13)

change. It was set up to be able to comply with the Kyoto Protocol in a cost-efficient and effective way and to achieve its goal of reducing emissions of GHGs by ? percent over the period <==?-<=B< (NERA, <==g). The EU ETS is a carbon market, which Mol (<=B<, p. Bf) defines as a “system where permits or allowances to emit greenhouse gasses or credits earned by not emitting greenhouse gasses (offsets) are traded”. The EC describes the policy instrument as “a cornerstone of the European Union's policy to combat climate change and its main tool for reducing industrial greenhouse gas emissions cost-effectively” (EC, <=B>, p. B). Its key function is reducing the emission of CO# (EC, <=B>, p. s). The EU ETS includes more than BB,=== power stations and industrial plants and it includes aviation between participating countries as well. It is the first and largest emissions trading system for reducing GHG emissions (European Union, <=Bg). Yet, the EU still has a different ETS for aviation, and it still excludes shipping and states other exclusion conditions (EP & EC, <==f). Ultimately, the goal is to make the ambitious “<=-<=-<=” targets of increasing the share of renewables in overall EU primary energy consumption to <= % by <=<= and make a <=% cut in GHG emissions. After <=<=, the reduction goals of GHG emissions are at least s= percent by <=f= (European Union, <=Bg).

The EU ETS is divided up into four phases, each phase introducing revisions adding and changing specific policy regulations, ultimately increasing the effectiveness of the EU ETS. The first phase ran from <==g to <==?, the second from <==? to <=B<, the third trading phase has started in <=Bf and will end in <=<= and the fourth phase will start in <=<B and has no end date yet.

2.3 Considered phases

The period of interest, <==?-<=B<, coincides with the second phase of implementation of the ETS system. As mentioned before, this has implications, since the EU ETS is adjusted every phase.

(14)

The second phase was the first commitment period in Kyoto Protocol after the pilot period from <==g-<==@. In this period, Iceland, Liechtenstein and Norway from EEA-EFTA, and therefore non-members of the EU<?, were introduced to the scheme.

In addition, the cap was reduced by >.g% for the period compared to <==g. In essence, this increases scarcity, but due to the economic downturn, allowance surpluses emerged (European Union, <=Bg).

Also, nitrous oxide was added to widen the scope of the system and transition started from free allocation to auctioning. However, most allowances were still given out to participants for free. In addition, the Linking Directive3 allowed businesses to use offset credits from the joint implementation (JI) mechanism to meet their EU ETS obligations. Since the credits used in the clean development mechanism (CDM) in the Kyoto Protocol were already allowed to use for EU ETS compliance, cost-effectiveness was further improved by allowing compliance through these international crediting mechanism (European Union, <=Bg). Also, the aviation industry is covered from <=B< onwards, widening the sector scope of the system.

2.4 Renewable energy source support policies

Government support for RES can regard the production and consumption side of renewable energy, as well as the entire energy system. This considers any public intervention in the energy market by public actors such as national and regional governments that influences the market price (Ecofys, <=Bs, p. xiii). These interventions, aiming to develop RES-technology and reduce GHG emissions, are widespread and increasing in the EU. In fact, interventions to support

RES-

3

Directive 2004/101/EC of the European Parliament and of the Council amending Directive 2003/87/EC establishing a scheme for greenhouse gas emission allowance trading within the Community, in respect of the Kyoto Protocol's project mechanisms.

(15)

production and consumption had a value of €s= billion in <=B< compared to €B> billion to support coal, natural gas, oil products and other fossil fuels in the EU (Ecofys, <=Bs).

A wide array of different RES-support schemes can be observed, and often these interventions are applied simultaneously. The types of support considered and calculated in the Ecofys report are direct transfer of funds, government tax and other government revenue foregone, transfer of risk to government, installation income or price support and non-financial measures (Ecofys, <=Bs).

The main instrument of support in the EU historically and over the considered period are feed-in tariffs, which are direct transfers of funds. In a feed-in tariff (FIT) scheme, a fixed payment is made for each unit of energy generated, power plant operators receive a fixed payment for each unit of electricity, heat and/or biogas generated, in spite of the market price of the energy products (Ecofys, <=Bs).

Other important instruments of support are investment grants, which are awarded to installations reaching a specific goal like installed capacity and entail a transfer of risk to government. And exemptions from energy taxes, a form of government revenue foregone, can be given on RES, making the production of RES-energy cheaper (Ecofys, <=Bs).

The way the government interventions are financed can also differ per member state. Some systems use tax revenue from non-renewable energy generators and others make the end consumer pay a mark-up on their energy bills to gather funds (Butler & Neuhoff, <==s) (Fischer & Preonas, <=B=).

Generally, energy support instruments can be applied to energy production or energy consumption side. However, government support on renewables are often directed at the production side. Therefore, like the EU ETS, it affects not directly the emissions of consumers, but that of the industry (Ecofys, <=Bs).

(16)

Importantly, RES-support schemes are aimed to only affect the <? European member states’ national RES-industries (Ecofys, <=Bs). Whereas the EU ETS affects all participating countries at the same time, including the non-member countries Norway, Iceland and Liechtenstein. As these are non-members of the EU<?, this means the area considered for the EU ETS data slightly deviates from the area considered for the RES-support data.

Again, as for the EU ETS, reaching the ambitious “<=-<=-<=” targets of increasing the share of renewables in overall EU primary energy consumption to <= % by <=<= and make a <=% cut in GHG emissions is the main goal of the RES-support policies (Ecofys, <=Bs).

3. Theoretical Framework

This section describes the theoretical framework of both policies and the potential interactions between them.

3.1 Economics of the EU ETS

Participating companies subject to the ‘cap-and trade’ scheme that the EU ETS is, receive allowances for emitting that are tradable, while the cap is limited to ensure that there is permit scarcity (European Union, <=Bg, p. B>). The member states draw up their own national allocation plan (NAP), in which they specify the allocation to installations and set a cap. In brief, governments in the member states will first determine the cap, and then allocate the allowances to installations in energy-intensive industrie and the power and heat generation sectors (Christiansen, 2005). These NAPs are subject to review by the EC (Ellerman & Buchner, <==@).

(17)

by doing this, the European Union Allowances (EUAs) become valuable and participants are willing to pay a price that permits them to emit CO#. Crucially, in order to create the necessary financial incentives and therefore pressure for emission reductions and trade, the cap needs to be tight enough to increase carbon emission costs and push participants and member states to comply to the cap and reduce carbon emissions or trade accordingly (Grubb & Neuhoff, <==>). The initial goal is to cut CO# emissions by <B% by <=<= by reducing the amount of allowances over time, so the EU ETS attempts to create the necessary incentives.

In addition, the participating companies are allowed to trade these EUAs. When participating companies emit less GHGs than they were granted allowances for, they generate a surplus of allowances that can be sold on the allowance market, and vice versa if companies overproduce. This flexibility ensures that emission reductions are made by the installations that are most efficient and effective at doing so. This decreases overall costs of GHG reduction in the system.

An additional crucial design feature of the policy is the allocation of the allowances. The two main types of allocation are auctioning or free allocation. During phases B and <, the majority of allowances were handed out for free based on the historical levels of emissions4. This so-called ‘grandfathering’ is suboptimal, as it allocates allowances based on historical emissions, thereby rewarding the larger polluters and reducing allowance scarcity.

In phase f (<=Bf-<=<=), a new base was adopted where the free allocation was based on benchmarking; GHG intensive participants receive less free allowances relative to their production in order to stimulate them to reduce emissions (European Union, <=Bg, p. s=).

Auctioning is the second allocation strategy, where participants have to acquire the allowances at the relevant exchange markets. Only a limited part of allowances allocated was

4

(18)

auctioned in the first two phases. However, in the third phase, the Auctioning Regulation was set up to govern the increased auctioning by specifying the timing, administration and other aspects of open, transparent, harmonise and non-discriminatory auctioning (European Union, <=Bg, p. <?). The actual allocation strategy is the responsibility of each individual member state through the NAP for each phase of the scheme.

The mode of allocation of the allowances is deemed very important in the economic design of the EU ETS. In an early analysis of the system, Grubb classifies the allocation strategy as the “Achilles’ heel” of the EU ETS as the amount of permits available in the market and thus permit scarcity depends directly on government decisions about allocations (Grubb & Neuhoff, <==>). In addition, in an empirical analysis of the allocation system specifically, Ellerman and Buchner (<==?) raise concern about over-allocation in most of the NAPs. This lack of scarcity might cause economic dysfunction, since the cap would be nonbinding and there is no reduction in emissions. Differences in allocation strategies are therefore crucial to keep in mind, since they can have important implications for the distributional effects and the effectiveness and efficiency of the system as a whole.

3.2 Economics of RES-support

RES-development is needed to reduce GHG emissions and to provide a clean environment. From an economic point of view, clean air is a so-called public good, meaning that people can use the good without reducing the availability for others. This means that individuals cannot be effectively excluded from using clean air and where use by one individual does not reduce availability to others. This causes insufficient incentives to produce or sustain clean air voluntarily. Therefore, from an economic point of view, the goal of government RES-support is to create sufficient incentives let society benefit from cleaner air through the implementation of renewable

(19)

energy and to increase the diffusion of these technologies (Menanteau, Finon, & Lamy, <==f). More specifically, while private generation costs are higher for renewable energy than for conventional techniques, renewable energy provides benefits that are not valued by the market, leading to a lower actual social cost for renewable energy (Del Río González, <==@)5. Furthermore, market entities (investors, generators, suppliers and consumers) are guided by the incentives provided by the market, where decisions are taken on the basis of the abovementioned private and not social costs. In order to level the playing field with respect to conventional technologies, public support to renewable energy is used to internalise the positive externalities of renewable energy in the decisions taken by economic actors and allows renewable energy to penetrate the energy market (Del Río González, <==@).

It is intuitive and logical that the RES-support instruments are aimed at increasing the share of RES in energy production. Additionally, as the share of energy consumption derived from fossil fuels becomes smaller and the share of RES in energy production becomes bigger, the decrease in GHG emissions get larger and the adverse environmental impact gets smaller. Effectiveness of the RES-support is therefore defined as the ability of the RES-support to reach the GHG emissions reduction targets by developing RES-production capacity.

3.3 Interaction between policies

Normally, as explained above, by introducing the EU ETS, externalities of the production and consumption of energy, in the form of the emissions, are internalized and through the trading mechanism, the technology that can lower emissions at the lowest cost, will do so the first (Abrell,

5

Additional obstacles to the deployment of RES-production are the availability of the resource, the appropriateness of the technology, lead times and delays caused by institutional and regulatory barriers, difficulties in accessing the grid and the socio-cultural acceptance of renewable energy technologies (Del Río González, 2007).

(20)

<==?). Also, emission intensive activities get relocated to expensive technologies. On the other hand, the RES-support closes the gap between the social and private costs of renewable energy. However, when both are combined, complex interactions occur.

One logical solution to avoid interaction complications, is to only implement one policy to achieve the objectives (Jensen & Skytte, <==f). However, it is important to note that while both policies’ objectives overlap by the goal of reducing GHG emissions, they are not identical. RES-promotion has the additional goal of developing RES-production capacity. Therefore, if the EU ETS were the only policy instrument, while it would reduce the difference in cost between conventional and RES-energy, it would not promote RES-deployment enough (Linares & Santos, <==?). Conversely, merely using the RES-promotion policy would mean that incentives are lost to strive for all other potential emission reduction opportunities than just by increasing production from renewable energy sources (Linares & Santos, <==?). Therefore, with the EU ETS ensuring short term, cost-efficient solutions towards emission reductions and the RES-support providing incentives to develop long term reductions through RES-deployment, the policies are commonly viewed as complementary.

However, there are multiple studies that raise doubt whether they truly are an effective combination of policies considering their overlapping goals. For example, it is argued that there would be no incentive to further reduce emissions from additional RES-policies when there is a binding cap (Morris, <==F) (Pethig & Wittlich, <=B=) (Sijm, <==g). Considering that the participants of the EU ETS will already use the maximum amount of allowances, despite the RES-policy incentives (Fischer & Preonas, <=B=). This would only increase the compliance costs, and ultimately social costs, for the participating firms, as participants lose flexibility to choose the lowest cost methods to reduce their emissions (Fischer & Preonas, <=B=).

(21)

implemented, managed and have effect on different levels, namely national and European. This would not be a problem, as long as each policy focusses at the geographical level appropriate for its objectives (Linares & Santos, <==?). However, as pointed out before, the national RES-support policies, besides having a national objective of stimulating RES-deployment and supply, try to reduce GHG emissions and mitigate climate change. The latter goals clearly surpass the national level and influence the European level as well. This shows that, because their core objectives overlap, when both the RES-support and the EU ETS are implemented, interaction effects need to be taken into account. While the EU ETS and the national RES-support try to reach the same goal, namely reducing GHG emissions and ultimately reduce global warming, the interaction effect of these European and national policies might be undesirable (IEA, <=BB) (Del Río González, <==@) (Sijm, <==g).

The theoretical stream of research on the interaction effects of the ETS and RES-support is plentiful and include (Skytte, <==>), (Boots, <==f), (Morthorst, <===a), (Morthorst, <===b), (Morthorst, <==B), (Morthorst, <==fa), (Morthorst, <==fb) and (Del Río, F, & Gual, <==g). While there are different conclusions and outcomes, a general important finding is that the interactions are very complex since they depend on the particular country-specific instruments and the ever evolving EU ETS (Del Río González, <==@). In addition, some argue the possible ineffectiveness of any other policies on reducing CO#emissions of the participating sectors, like the considered national RES-support, once the EU ETS becomes operational (Sijm, <==g).

Altogether, there is consensus that there are advantages to use the EU ETS in combination with RES-promotion, but that the involved policies need be precisely coordinated to avoid adverse interaction effects (Linares & Santos, <==?) (Böhringer, <=B=) (IEA, <=BB) (Fischer & Preonas, <=B=).

(22)

Whilst the abovementioned discussion explains important interaction effects, hardly any literature is available on possible spatial effects. However, the well-known free-riding problem often found in climate change policies and their coordination does present rationale to investigate this. The theory stipulates that since the earth’s climate system is an almost absolute public good, countries are tempted to pollute more than what is socially optimal on this planet (Vanderheiden, <=B>). Therefore, countries that do choose to reduce emissions bear the full costs of this carbon emission reduction and receive only a small part of the benefits, since polluting activities simply shift to cheap-carbon-emitting countries and the overall emission reduction is only a fraction of what the emission reducing country is paying for.

A more specific form of free-riding in climate change policies is formulated in the carbon leakage theory, which states that in absence of coordinated climate change policy, before reducing carbon emissions in countries where climate change policy is in more stringent, production of goods and services based on fossil fuels will transfer to countries with less stringent environmental regulation and therefore less carbon emission costs (Tirole, <=B<) (Foster, et al., <=B@).

Applying this rationale to the analysed combination of policies implies that in absence of coordinated RES-support policy, before reducing carbon emissions in countries where RES-support is higher, production of goods and services based on fossil fuels will transfer to countries with lower RES-support.

Crucially, whereas without the EU ETS the RES-support level differences only create very small relative price differences of RES-energy in terms of non-RES-energy (and thus carbon emission costs) between countries, the incentive to move activities between countries is therefore assumed negligible. However, the existence of the EU ETS strengthens the incentive to move activities as its cap-and-trade mechanism generates scarcity of legal rights to emit, and thus raises the carbon emission costs. This creates higher relative price differences of RES-energy in terms of

(23)

non-RES-energy between member states. Therefore, industries in member states where RES-support is higher will more hastily move their activities to member states where RES-support, and thus carbon emission costs, are lower.

3.5 Empirical Literature

In the growing amount of empirical literature, it is regularly found that the RES-support increases demand for RES-energy, thereby decreasing demand for non-RES-energy, which in turn has a downward effect on the demand and prices of permits. This reduces the strength of the EU ETS as it makes polluting activities cheaper (NERA, <==g) (Rathmann, <==>) (Kim & Koo, <=B=). However, there are a few limitations on the empirical literature so far, which are outlined below. Also, it is pointed out how they will be addressed by this research.

Firstly, the literature mainly focuses on the interaction effect of renewable electricity (RES-E) promotion as opposed to RES-support for all energy types. This is because the electricity market is a significant part of the energy system in most countries, which makes the potential impact of effectively encouraging RES significant. Another reason is that the RES-E promotion policies and EU ETS interact here through their respective effect on the transparent and regulated electricity market variables (i.e. prices and volume), for which the data are accessible and abundant (Del Río González, <==@). However, in <=Bs, in only four countries in the EU did the RES-E represent over half of all RES consumption (EEA, <=B@). Therefore, it is evident this disregards a significant amount of markets for non-RES. This analysis addresses this issue by considering RES-support for all energy markets.

Secondly, inherent to investigating RES-E, only a specific selection of RES-technologies (and support policies) are observed. For instance, in RES-E in <=Bs, hydropower is prevalent, whereas in renewable heating and cooling solid biomass is dominant and in European transportation biodiesel

(24)

is the most used renewable alternative fuel (EEA, <=B@). Other than most empirical literature investigating the interaction between both policies, this research will focus on the effect of the member states’ support for the complete array of RES-technologies.

Thirdly, as mentioned above, most research analyses the downward effect of the RES-support schemes on the allowance prices. However, as will be shown in this research, there is an adverse spatial effect amongst member states that needs to be considered as well.

In summary, whilst previous empirical research does find interaction effects between support policies and the EU ETS, the scope of considered energy types, the scope of RES-technologies and the type of impact considered on the EU ETS system is incomplete. The next section will describe the spatial interaction theory, which will describe a spatial interaction effect supplementary to the previous demonstrated and described effects.

4. The spatial interaction theory

This section describes how the spatial interaction theory is derived using the GHG determinants model and the theory presented in the previous section.

4.1 Theory in short

The functioning of both policies explained above give rise to the development of the spatial interaction effect. As mentioned above, the supply of allowances is fixed by governments through the EU ETS and the subsequent NAPs. Therefore, firstly, there is a predetermined amount of GHG emissions per period, known by all participants and decreasing over time. This creates the scarcity of allowances, crucial for the spatial interaction effect. Within these restrictions, an allowance to

(25)

legally emit a small part of the total is traded among and within member states (European Union, 2015).

Secondly, as the considered RES-support is aimed at the RES-industry within the member state’s borders, the relative prices of RES-energy in terms of non-RES-energy in that particular member state decline. With an active EU ETS creating scarcity in allowances, increased RES-support in a certain member state will reduce the emissions within the EU ETS in that member state. Therefore, since the RES-support also alters emission patterns, the possibility arises for both policies to interact.

Thirdly, the single market in the EU provides relatively easy circumstances to migrate activities within the EU.

In summary, when RES-support and therefore relative prices differ amongst member states, the EU ETS cap and trading possibility are creating scarcity, and the characteristics of the single market in the EU are considered, polluting industries are moved across the EU. Namely, due to the differing support levels and therefore relative prices of RES, member states will trade allowances and move the polluting industries, thereby affecting GHG emissions in other member states within the EU ETS which establishes the spatial interaction effect. More specifically, this gives rise to the following hypothesis: With an active EU ETS creating scarcity in allowances, increased RES-support in a certain member state will increase the emissions in other member states within the EU ETS. This will be illustrated and further explained in the model presented in the next sections.

4.2 The GHG emission determinants

In order to estimate the effects of the RES-support levels on the GHG emission levels across the member states, established GHG determinants models are used as vehicle. The models are created to investigate the factors contributing to the GHG emission levels. It is found that

(26)

population, energy intensity, fuel mix and emission intensity of energy generation and economic size are determinants of GHG emissions and are therefore used in this research (Rafaj, Amann, & Siri, <=Bs) (Zakarya, Mostefa, & Seghir, <=Bg).

There is a long academic debate on the characteristics of the effect of economic size on pollution levels, an important determinant of GHG emission levels. A prominent theory on this matter is the Environmental Kuznets curve (EKC) hypothesis, suggesting that there is an inverse U-shaped relationship between emissions and gross domestic product (GDP)6 (Grossman & Krueger, BFFB). Therefore, when the EKC theory is followed, depending on the phase of the economic development a member state is in, this relationship can either be positive or negative. However, in most research it is found that the inverted U-shaped relationship holds when a per capita income is considered (Kaufmann, Davidsdottir, Garnham, & Pauly, BFF?) (Viguier, BFFF); (Bruvoll & Medin, <==f) (Markandya, Golub, & Pedroso-Galinato, <==>). Also, similar research investigating SO#, NO_', and CO# emissions suggest that emissions tend to increase with economic growth (De Bruyn,

Van Den Bergh, & Opschoor, 1998) (Stern, 2006). Therefore, since this research considers GDP not

per capita and aims to explain the abovementioned greenhouse gases, the economic size is expected to increase emissions in this research. In addition, it not the focus of this research to investigate the existence of the EKC.

4.3 The model used to establish the spatial interaction effect

The conditions for the occurrence of the spatial interaction effect are described in the discussion above. It is further explained how the GHG determinants model is used as a vehicle. In

6

This is due to autonomous technological progress, structural changes in national economies, behavioural changes and dedicated environmental policies.

(27)

this section, the derived model and its reasoning are explained.

Firstly, it is important to understand that effective RES-support is assumed, meaning they will help reach the goal of reducing GHG emissions. Secondly, the cap of the EU ETS and the possibility to trade are introduced. The amount of verified emissions within the EU ETS is used as a measure of GHG emissions within the system. In contrast to the total GHG emissions, which include emissions from entities outside of and unaffected by the EU ETS as well.

To be able to test these dynamics, the amount of RES-support of member state i and that of the rest of the EU (EU-i) are considered. Also, when it comes to the abovementioned GHG determinants, as economic size is one of the most important and the other controls increase the risk of endogeneity and overestimation, GDP is used as a control variable in this research. So, member state i impacted by RES-support and by the EU ETS, the model looks as follows:

()*+,+)- )/+00+120₃₄ = 6 + 8₉:;<0₃₄ + 8_#:;<0_{=>?3 4}+ 8_@ABC₃₄ + D₃₄ (B) Where ()*+,+)- )/+00+120₃₄ is the amount of verified emissions in member state i in period t, α is the intercept, :;<0₃₄ is the amount of support on RES in member state i period t, ABC34 is the gross domestic product in million euros in member state i in period t, and D34 is the standard error in member state i in period t.

In this model, assuming effective RES-support, 8₉ is expected to be negative, reducing the GHG emissions levels as it increases. 8# is expected to be positive as the RES-support levels in the rest of the EU affect the GHG emissions member state i through the spatial interaction effect. Furthermore, 8@ is assumed to be positive, as GDP is expected to increase emissions.

To illustrate the RES-support effects, consider a national RES-support in member state i in year t of S and a higher RES-support in the rest of the EU of 2S. Using (B), assuming the RES-support only affects the respective national industries and assuming a present and functioning EU ETS, the

(28)

following will happen.

The RES-support in the rest of the EU is higher and therefore will have relatively cheaper RES-energy and more expensive non-RES-energy. Vice versa, member state i has lower RES-support and therefore will have relatively more expensive RES-energy and relatively cheaper non-energy. So, having cheaper energy in the rest of the EU, participants there will use more RES-energy and less non-RES-RES-energy in their system and therefore demand less allowances and emit less GHG in production. Vice versa, the more expensive RES-energy in member state i will mean that they emit more GHG in production.

Therefore, because of the capped emissions in the EU and the generated scarcity of permits to emit, both countries will use the allowance market to trade and the rest of the EU will sell allowances to member state i, allowing it to emit more GHG. This is the hypothesised spatial interaction effect the RES-support level in the rest of the EU has on the GHG emissions within member state i. Therefore, in addition to 81, 82 plays a role and is expected to be positive. In terms of the model the change of verified emissions within the EU ETS in member state i after the illustrative year is:

8₉∗ < + 8_#∗ 2< ^

_ 0 (<)

So, member state i will use more allowances and emit more GHG because of the difference in support levels with EU-i (8#), meaning that there is an adverse spatial interaction effect of having both policies in place at the same time. Nota bene, a member state will have higher GHG emissions in its country because of a higher level of RES-support in the other EU member states. As the magnitudes of 8# and 89 are ambiguous, the total change in verified emissions is ambiguous as well.

(29)

5. Data

This section describes the input data used for the model and their properties.

5.1 Data used

A panel data set with data for the <? EU member states over time is used for the analysis. Data is collected for RES-support levels, GHG levels, verified emissions and GDP. The period observed is from <==?-<=B<. This time period is bound by the availability of data on government interventions on RES.

RES-support

In order to capture the full effect of RES-support in the EU, all the different sources of renewable energy used in the EU are considered in this research. For this, the complete dataset containing the quantified extent of public interventions in energy markets in all <? member states for all energy use excluding transport is used (Ecofys, <=Bs). This is the first and only complete dataset for the EU<? detailing and quantifying government interventions in the energy market and therefore the only data source on government intervention in this empirical analysis. The data is extracted from the table “Total support per Member State <==?-<=B<” found in the annex of the Ecofys report (Ecofys, <=Bs). It is yearly data and the unit of measure is EUR million <=B<. The data is shown in figure B.

(30)

Figure 1. Total RES-support per Member State <==?-<=B<.

Verified emissions

As a measure of emissions within the EU ETS, verified CO_# emissions from installations and aircraft operators from European Union Transaction Log (EUTL). Extracted from the “EU Emissions Trading System (ETS) data viewer” provided by the European Environment Agency (EEA, <=B?). Data is in tonnes of CO#equivalent and covers the period <==?-<=Bf. The data is shown in figure <.

(31)

Figure 2. Verified emissions from all stationary installations in tonnes CO_#equivalent for <==?-<=B<.

Total GHG emissions

As a measure of GHG emissions in total produced by the member states, greenhouse gasses (CO#, N#O in CO#equivalent, CHs in CO#equivalent, HFC in CO#equivalent, PFC in CO#equivalent, SF> in CO#equivalent, NFf in CO#equivalent) emitted per member state for all sectors and indirect CO_# (excluding LULUCF and memo items, including international aviation) are used. Data is in tonnes of CO#equivalent and covers the period <==?-<=Bf. Source of data is the European environment agency (<=B@). The data is shown in figure f.

(32)

Figure 3. GHG emissions in tonnes of CO_#equivalent for <==?-<=B<.

GDP

As a measure of economic size, gross domestic product at current market prices in million euros is used for the period <==?-<=Bf (Eurostat, 2018).

5.2 Data properties

(33)

Table D. Descriptive statistics.

Variable Obs. Mean Std. Dev. Minimum Maximum

Verified emissions Bs= >.?ge+=@ F.@Fe+=@ = s.@fe+=?

Total GHG emissions Bs= B.@se+=? <.f=e+=? f<g@gg= B.==e+=F

RES-support Bs= B=f>.=@B <<<s.@F< = Bf,g==

GDP Bs= s>f,s<?.s >Fg,s?F.@ >B<?.@ <,@g?,<>=

As can be seen, there are Bs= observations for every variable, stemming from the g observed years for all <? member states.

The mean of verified emissions of >.?ge+=@ tonnes of CO#equivalent is roughly half of the mean of the total GHG emissions, which is B.@se+=? tonnes of CO#equivalent. The standard deviations for verified emissions and total GHG emissions are F.@Fe+=@ and <.f=e+=? tonnes of CO_#equivalent, respectively.

The mean of RES-support is B=f>.=@B million euros with quite some variation, as the standard deviation is <<<s.@F< million euros.

The mean GDP is s>f,s<?.s= million euros with a standard deviation of >Fg,s?F.@= million euros.

(34)

6. Method

This section describes the method used to test the hypotheses. Firstly, the analysis will be explained. Secondly, the specific models forming the empirical method will be explained. Lastly, the limitations of the research design will be explained.

6.1 Analysis reasoning

Using panel estimation, this research will test the abovementioned hypotheses for the effect of national RES-support on the EU ETS participants’ emission levels in member state i in year t and about RES-support in the rest of Europe having an effect on the EU ETS participants’ emission levels in member state i in year t.

Firstly, the effectiveness of RES-support within the EU ETS is analysed by testing whether it reduces the verified GHG emissions within the system. More specifically, it is tested whether 89is negative in (B). Next, the hypothesized spatial interaction effect of the RES-support and the EU ETS on the verified GHG emission levels within the EU ETS cap, a positive 8# in (B), is tested.

Furthermore, derived from the GHG-determinants model discussed above, GDP is used as control variable. The variables of interest, capturing the effect of member state’s RES-support on member state’s GHG emissions within the binding EU ETS (indicated by the verified emissions) are added.

Additionally, it is important to note that the spatial interaction effect is expected conditioned on both the national RES-support and the EU ETS being present. So, to more compellingly establish whether the spatial interaction effect is invoked by this specific combination, a comparison of two regressions is made. First, the verified emissions per member state as

(35)

dependent variable is used, capturing the GHG emissions within the EU ETS. Then, the total GHG emissions are used as dependent variables. Logically, the hypothesized effect is only expected when using verified emissions per member state as dependent variable, which is produced by participants of the EU ETS.

6.2 Empirical Method

In general, a panel estimation can be performed using a random or a fixed effects model. Since most likely, the independent variables in these models are not accounting for all factors influencing the GHG emission levels or verified emission levels, the model could be biased by omitted variables. This would mean that the error term and independent variables would correlate. Additionally, the Hausman test, which compares the fixed and random effect estimates and tests whether the errors are correlated with the independent variables, pointed out that the fixed effects model is the appropriate model to use (Wooldridge, <==@). For this reason, the fixed effects model is used for all regression analyses that follow.

In the following sections, the empirical test and used models are explained. Firstly, it is tested whether the RES-support is effective within the EU ETS system using verified emissions within the EU ETS as dependent variable in model B. In addition, the spatial interaction effect and the time trend robustness of model B is tested. After that, the same RES-support effectiveness, spatial interaction effect and time trend robustness are tested using the total GHG emissions of member state i as dependent variable.

6.3 Verified emissions EU ETS in member state i as dependent variable

(36)

variable. The model looks as follows:

()*+,+)- )/+00+120₃₄ = 6 + 8₉:;<0₃₄ + 8_#:;<0_{=>?3 4}+ 8_@ABC₃₄ + D₃₄ (f) Where ()*+,+)- )/+00+12034 is the amount of verified emissions in tonnes of CO#equivalent in member state i in period t, α is the intercept, RESs₃₄ is the amount of support on RES in million euros in member state i period t, , :;<0_{=>?3 4?9} is the amount of support on RES in million euros in all other member states (EU-i) in period t and ABC ₃₄is the gross domestic product in million euros in member state i in period t and D₃₄ is the standard error in member state i in period t.

In this model, the hypotheses, derived in the previous chapters, are as follows:

§ 8₉: Negative. Effective RES-support is expected to reduce the amount of verified emissions. § 8_#: Positive. Adverse spatial interaction effect where support levels in the rest of the EU increases

the amount of verified emissions in member state i.

§ 8_@: Positive. GHG emissions are expected to increase with income.

In order to test whether model B is robust for a time trend, the model is extended by s year dummies, so that the model looks as follows:

()*+,+)- )/+00+120₃₄ = 6 + 8₉:;<0₃₄ + 8_#:;<0_{=>?3 4}+ 8_@ABC₃₄ + 8_bc)d*2009 + 8_fc)d*2010 + 8_gc)d*2011 + 8_hc)d*2012 + D₃₄ (s) It is expected that the model is robust, so the year dummies are of no significant influence and therefore the hypotheses stay the same:

§ 8₉: Negative. Effective RES-support are expected to reduce the amount of verified emissions. § 8_#: Positive. Adverse spatial interaction effect where support levels in the rest of the EU increases

the amount of verified emissions in member state i.

(37)

6.4 Total GHG emissions of member state i as dependent variable

In model B the amount of verified emissions per member state is used as dependent variable, capturing the GHG emissions within the EU ETS. In model <, however, the total GHG emissions are used as dependent variable. The model looks as follows:

AiA ₃₄ = 6 + 8₉:;<0₃₄ + 8_#:;<0_{=>?3 4}+ 8_@ABC₃₄ + D₃₄ (g) Where AiA 34 is the amount of total GHG emissions of member state i in period t, and the remaining variables are the same as in model B.

In this model, the hypotheses, derived in the previous chapters, are as follows: § 8₉: Negative. Effective RES-support is expected to reduce the total amount GHG emissions. § 8_#: Ambiguous. As the EU ETS cap is absent, the effect of RES-support levels in the rest of the

EU on total GHG is uncertain.

§ 8_@: Positive. GHG emissions are expected to increase with income.

In order to test whether model < is robust for a time trend, again, the model is extended by s year dummies, so that the model looks as follows:

AiA ₃₄ = 6 + 8₉:;<0₃₄ + 8_#:;<0_{=>?3 4}+ 8_@ABC₃₄ + 8_bc)d*2009 + 8_fc)d*2010 +

8_gc)d*2011 + 8_hc)d*2012 + D₃₄ (>)

It is expected that the model is robust, the year dummies are of no significant influence and therefore the first three hypotheses stay the same:

§ 8₉: Negative. Effective RES-support are expected to reduce the amount of verified emissions. § 8_#: Ambiguous. As the EU ETS cap is absent, the effect of RES-support levels in the rest of the

EU on total GHG is uncertain.

(38)

6.5 Study limitations

The complex and multifaceted nature of the interaction of the EU ETS and RES-support in the EU poses several threats to the validity of this research that need to be addressed. Therefore, it should be noted that there were some limitations in the design of the study.

First of all, as it was the first and only source of comparable quantified data on RES-support in the EU, meaning there was no alternative way to operationalise the amount of RES-support. Therefore, the selections made by this report automatically defined the selections considering RES-technologies, RES-support instruments and industries in this research. As there are multiple definitions of RES and RES-support, this selection could be (politically) biased, altering the results. Secondly, while the EU ETS covers the EU <?, Norway, Liechtenstein and Iceland, the RES-support data is merely about the <? EU member states. Depending on the level of RES-RES-support in those three countries, this could influence the results.

Fourthly, the EU ETS allowances permit a participant to emit CO# only. In the total GHG emissions data all GHG are incorporated in CO_# equivalent amounts. As different types of GHG develop differently over time and are affected differently by RES-support policies, this could alter the results.

Having acknowledged these limitations, it is important to note that this research still contributes to the existing literature as discussed in previous sections.

7. Results

In this section, the results of the panel regressions of the abovementioned models are presented. First, the results of model B are shown and explained in table <. Secondly, the results of

(39)

model < are shown and explained in table f. Finally, the results are clarified using illustrative calculations.

7.1 Empirical results

Table E. Estimated coefficients, t-values and r-squared of model J. Verified emissions as dependent

variable. Second column includes year effects.

Note. An (†) indicates significance at the B=% level, (††) indicates significance at the g% level and

(†††) indicates significance at the B% level. MS is defined as member state. *Year effects to control for temporal variation in dependent variable that are not adequately captured by the explanatory variables in the model. ** Year<=B< omitted because of collinearity.

Model B Model B* Coefficient (t-value) Coefficient (t-value) Constant s.F>e+=@††† (f.>@) g.@fe+=@††† (f.s@) RES-support per MS -s,g<f.FF?††† (-s.sg) -s,<fg.B<?††† (s.=B) RES-support rest EU _-B>g.<f?F†† (-<.>?) -B?=.fsB (-B.ff) GDP ?=.Bs??g††† (<.F@) >>.>s?gg†† (<.g?) Year <==F - -fg@f>ss†† (-<.@<) Year <=B= - -B<>gg>.s (-=.BB) Year <=BB - -B>sf?<.@ (-=.B?) Year <=B< - Omitted** R-squared =.@< =.@B Observations (N) Bs= Bs=

(40)

When examining the results of model B table <, firstly, it can be seen that the coefficient estimate for the variable for RES-support per member state has the expected negative sign and is significant on the B% level, which indicates that the verified emissions under the EU ETS decrease when RES-support per member state increases.

Secondly, the coefficient estimate for the RES-support for the rest of Europe is also negative and is significant on the g% level. This indicates a that the verified GHG emissions under the EU ETS decrease when RES-support for the rest of Europe increases, not showing the hypothesized positive influence and spatial interaction effect.

Thirdly, the coefficient estimate for the GDP variable is positive and significant on the B% level, which indicates that an increased economic size increases the verified emissions under the EU ETS, which is intuitive and as expected.

As a robustness test of model B, the regression results of model B* show the estimates for model B when year effects are included to control for the effects of temporal variation on the dependent variable, verified emissions, that are not adequately captured by the explanatory variables in the model. This is specified in equation (s) in the previous chapter.

It can be seen that the significance for RES-support in the rest of the EU decreases. This indicates that the explanatory variable for RES-support in the rest of the EU accounted for some temporal variation.

Also, as the Year <==F dummy is negative and significant on the g% level, this indicates that the verified emissions are significantly lower in that particular year. However, as the other year dummies are not significant, there is no time trend in the verified emissions that is not accounted for. Additionally, Year <=B< is collinear and omitted, since most likely, one of the other variables is following a temporal trend.

(41)

equal. This means that, besides the change in RES-support in the rest of the EU, model B is robust for a time trend. Therefore, it can be concluded that the regression results indicate that both RES-support variables are negative, meaning 89 and 8# are negative and 8@ is positive in equation (s) and all but 8_# are significant.

Table F. Estimated coefficients, t-values and r-squared of model ;. Total GHG emissions as dependent

variable. Second column includes year effects.

Note. An (†) indicates significance at the B=% level, (††) indicates significance at the g% level and

(†††) indicates significance at the B% level. MS is defined as member state. *Year effects to control for temporal variation in dependent variable that are not adequately captured by the explanatory variables in the model. ** Year<=B< omitted because of collinearity.

Next, the results of model < are shown in table f and explained. In model <, the dependent

Model < Model <* Coefficient (t-value) Coefficient (t-value) Constant B.>se+=?††† (@.@<) B.?=e+=?††† (?.=?) RES-support per MS -?g??.Bsf††† (-s.?s) -?Bs?.=?<††† (-s,>f) RES-support rest EU _-fg>.>f?>††† (-f.f<) -sBB.?@@g††† (-f.s<) GDP B=f.B<??†† (<.><) ?B.>=?=>†† (<.=?) Year <==F - -><s?sf?††† (-f.s>) Year <=B= - BB@g@Bg (B.=g) Year <=BB - -BBsB<>@ (-=.>?) Year <=B< - Omitted** R-squared =.?@ =.?? Observations (N) Bs= Bs=

(42)

variable is replaced by the total amount of GHG emissions.

When examining the results of model < table f, first of all, the results show that the variable RES-support per member state has a negative sign and is significant on a B% level, indicating that the total GHG emissions decrease when RES-support per member state increases.

Secondly, the estimate for the RES-support in the rest of Europe is negative and significant on the B% level. This indicates that the total GHG emissions decrease when RES-support for the rest of Europe increase.

Thirdly, the estimate for the GDP variable is positive and significant on a g% level, which indicates that an increased economic size increases the total GHG emissions.

Again, the robustness test of model < shows the estimates for model < when year effects are included to control for the effects of temporal variation on the dependent variable, total GHG emissions, that are not adequately captured by the explanatory variables in the model. This is specified in equation (>) in the previous chapter.

Like in model B, the Year <==F dummy is negative and significant, this time on the B=% level, indicating that the total GHG emissions are significantly lower in that particular year. However, as the other year dummies are not significant, there is no time trend in the verified emissions that is not accounted for. Additionally, Year <=B< is again collinear and omitted, probably indicating again that one of the other variables is following a temporal trend.

Moreover, the signs of the other coefficients and their significance levels stay the similar or equal. This time including the explanatory variable for RES-support in the rest of the EU. This means that model < is robust for a time trend. Therefore, it can be concluded that the regression results indicate that both RES-support variables are negative and significant, meaning 89 and 8# are negative and 8@ is positive in equation (>).

(43)

7.2 Comparing verified emissions within EU ETS and total GHG emissions

In order to establish the effects on emissions within the EU ETS and the effects of total GHG emissions, model B using the amount of verified emissions within the EU ETS as dependent variable and model < using the amount of total GHG emissions as dependent variable, are compared.

It can be seen that results are similar. RES-support per member state is negative and significant in both models, indicating that they decrease emissions both within the EU ETS and total GHG emissions. The RES-support in the rest of Europe is negative in both models, and significant in model <. Meaning that in the spatial interaction effect hypothesis is rejected in both models.

Additionally, the GDP variable is positive and significant in both models, showing similar results for both models.

7.3 Interpreting the results

The results show that the RES-support levels significantly affect the amount of GHG emitted both within the EU ETS and outside. In order to establish whether the magnitudes of effects of RES-support within and outside the EU ETS differ, consider the following calculation.

Regarding the relative magnitude, the estimated coefficient for the effect of RES-support per member state within the EU ETS (89) in equation (f) is divided by the mean of verified emissions. The coefficient shows the amount of reduced verified emissions of an increase of B million in RES-support in an average member state in an average year. Therefore, by dividing the coefficient by the mean of verified emissions emitted in an average member state in an average year, an indication of the relative impact of RES-support is established.

lm

nopq (sot3u3ov ow3xx3yqx) =

?b,#@f.9#z

(44)

The result shows that an increase of B million in RES-support in an average member state in an average year reduces verified emissions by =.==><% of the mean of verified emissions emitted in an average member state in an average year.

When applying the same calculation to the estimated coefficient for the effect of RES-support per member state on the total GHG emissions, 89 in equation (s), the following result is obtained:

lm

nopq (4y4pÑ ÖÜÖ ow3xx3yqx) =

?z9bz.}z#

9.hb{|}z ∗ 100% =-=.==s@%

The result shows that an increase of B million in RES-support in an average member state in an average year reduces total GHG emissions by =.==s@% of the mean of total GHG emissions emitted in an average member state in an average year.

This means that the effect of an increase in the RES-support per member state on verified emissions is B.f< times as big as the effect of an increase in the RES-support per member state on total GHG emissions.

(45)

8. Concluding remarks

8.1 Conclusion

This thesis has been written with the aim to investigate the possible spatial interaction between RES-support policies and the EU ETS in the EU. Related to that aim, the research question is ‘Do differences in levels of Renewable Energy Source support amongst the ;< EU member states lead

to displacement of GHG emissions?’

Firstly, the most important conclusion is that the spatial interaction effect of applying both the RES-support and EU ETS policies is not found. Therefore, the rationale of the carbon leakage theory is not found in this research on this policy combination, as the production of goods and services based on fossil fuels is not shown to transfer to member states with less stringent environmental regulation. This could well be caused by the absence of well-functioning EU ETS market mechanisms creating sufficient allowance scarcity, as many results of other papers have raised doubt on the effectiveness of the market mechanisms of the EU ETS as well. A small result which does point in the direction of a functioning EU ETS, market mechanisms and the right incentives is the fact that the relative effect of an increase in the RES-support per member state on verified emissions is B.f< times as big as the relative effect of an increase in the RES-support per member state on total GHG emissions.

Secondly, this thesis shows that RES-support decreases verfied emissions within the EU ETS and total GHG. This is caused by the efficient closing of the gap between the private and social costs of renewable energy which creates growth in RES-energy production, reducing GHG emissions. This is in line with the theory and findings of other literature concerning this issue.

The Emissions Trading Scheme and Renewable Energy Source Support in the EU: Two minds, one body?

MSc Thesis