Reforming emission trading systems : interest group influence on the reform of the EU ETS for 2021-2030

(1)

Reforming emission trading systems:

Interest group influence on the reform

of the EU ETS for 2021-2030

Student name: Mia Bivered Student number: 11937580 Supervisor: Dr. Robin Pistorius Second reader: Dr. Luc Fransen

Master thesis Political Science specialisation Political Economy University of Amsterdam

(2)

2

Abstract

The aim of this thesis is twofold: to study how the EU ETS has evolved since its implementation in 2005 and to analyse the role of interest group influence on the policy reform of the EU ETS between phase three and four. The findings show that the EU ETS has been environmentally efficient since its first phase, in terms of corresponding to the relevant emission reduction targets over time. The impact of the EU ETS on abatement investment was found to be small, indicating that higher permit prices will be necessary to incentivise abatement investment. The effect of the EU ETS on the competitiveness of regulated firms was found to have little to no carbon leakage effect. Carbon leakage has thus not been an issue, but industry and policy makers agree that the risk of carbon leakage motivates protection through continued free allocation.

Interest group influence on the policy reform of the EU ETS between phase three and four was analysed using the preference attainment method. The findings did not show any empirical evidence of differences in influence between interest groups representing business interests and other interest groups, nor that business associations had more influence than individual firms. However the analysis showed a structural difference in influence between the two main regulated sectors, the energy intensive industry sector and the energy sector, where interest groups from the energy intensive industry were found to have a statistically significantly higher probability of preference attainment. This finding can be interpreted as this type of interest group having had more influence in the latest revision of the EU ETS than interest groups from the energy sector. This is of policy relevance as stakeholder consultations should provide an equal opportunity for interest groups to have their preferences heard. A policy recommendation for the EU ETS and other ETS is thus to strive for more levelled influence between different sectors. Otherwise there is a risk that policy reforms lead to preferential policy designs for certain sectors, which could undermine the political acceptance of the policy as well as the legitimacy of the regulator.

(3)

3

Acknowledgements

I would like to thank my supervisor Robin Pistorius for your valuable support and guidance throughout the thesis writing process. Your mentoring helped me narrow down my (too) broad interests, to overcome the challenging parts of the thesis process, and to produce a piece of research that I would not have been able to write before this thesis project. I would also like to thank Dr. Luc Fransen for being the second reader of this thesis.

I also want to thank Matthew for taking the time to help me by continuously discussing the progress and the difficulties, as well as looking after me in busy writing times. Last but not least, I want to thank my amazing family and friends for your support.

List of abbreviations

CCS Carbon Capture and Storage CCU Carbon Capture and Utilisation CO2e Equivalent carbon dioxide

EEA European Environmental Agency EEX European Energy Exchange EIB European Investment Bank EC European Commission EU European Union

EU ETS European Union Emissions Trading Scheme GHG Greenhouse gas

MSR Market Stability Reserve NAP National Allocation Plan R&D Research and development

(4)

4

1. Introduction

Since its implementation, the European Union emission trading scheme (EU ETS) has been the largest greenhouse gas (GHG) cap-and-trade system, covering more than 11000 high-energy consuming power stations and industrial plants in 31 countries, accounting for around 45% of total GHG emissions in the EU (EC, 2015a). The policy was introduced in 2005, and has since been reformed several times. In the three phases of the EU ETS, the policy has changed in terms of emission cap, what countries, sectors and GHGs are covered, the method for allocation of permits and emission permit prices, and these will change again in the fourth phase starting in 2021. These policy changes affect the environmental effectiveness, abatement investment, competitiveness and political acceptance of the policy.

The aim of this thesis is to study how the EU ETS has evolved since its implementation in 2005 and analyse the role of interest group influence on the policy reform of the EU ETS between phase three and four. This will be studied in three steps: First, by exploring how the policy design has been reformed and how this has affected policy outcomes, in terms of environmental effectiveness, abatement investment, competitiveness and political acceptance (chapter 4). Second, by analysing the political space of the policy reform between phase three and four in terms of what factors were driving the reform, what the topics of the stakeholder consultation were and how these policy issues were put forward in the European Commission’s proposal following the consultation (section 5.1). Third, by comparing interest groups’ influence in the policy reform using the preference attainment method (section 5.2). The results and limitations of this analysis will be discussed in chapter 6, before concluding this thesis with recommendations for future research and policy recommendations in chapter 7.

The theoretical framework of this essay combines economic theory on cap-and-trade systems and political economy theory on interest groups influence on regulation. Economic cap-and-trade theory outlines what design options there are in creating and reforming such a system (see section 2.1), while the interest group influence theory offers theoretical explanation as to why certain designs are chosen over other ones (see section 2.2). Interest group influence is analysed using the preference attainment method and survey data from the European Commission’s stakeholder consultation on the revision of the EU ETS for the fourth phase of the policy (see chapter 3).

(7)

7

1.1 Relevance

Market-based policy instruments, as well as cap-and-trade systems more specifically, are an increasingly important type of environmental governance as they are increasingly implemented at the national and subnational level around the world (World Bank, 2018). Previous studies have analysed the performance of various ETS of different designs (e.g. Ellerman et al., 2016; Schmalensee & Stavins, 2015; Newell et al., 2013; Aldy & Stavins, 2012; Fromm & Hansjürgens, 1996). This research area offers some explanation of the politics of choosing ETS design, but does not systematically study the policy reform process. Another smaller strand of literature instead focuses on different aspects of the EU ETS reform process, including the effect of interest group influence (e.g. Jevnaker & Wettestad, 2017; Drew, 2010). As economic policy instruments are becoming more prominent in climate change governance, the importance of understanding the ETS reform processes and the role of interest groups increases. Efficient climate policies are critical in order to avoid harmful levels of climate change, and it is therefore of societal relevance to study the conflict between environmental and economic interests and whether economic interests are more influential in the policy reform process.

Dür has argued that “an analysis of interest group influence remains of crucial importance to the understanding of political processes.” (2008 p.560), while the importance of measuring interest group influence was described by March as “Influence is to the study of decision-making what force is to the study of motion – a generic explanation for the basic observable phenomena.” (1955 p.432). However, methodological difficulties of measuring influence have led to few attempts to measure interest group influence on policy making in the EU, and the lobbying success and policy influence of interest groups in the EU is considered an under-researched topic (Bunea & Baumgartner, 2014; Dür & De Bièvre 2007; Dür, 2008). Even less attention has been given to the more specific topic of interest group influence on the reform of the EU ETS, with the main exceptions being the studies by Drew (2010) and Jevnaker and Wettestad (2017), which both used the process-tracing method.

This thesis can thus fill two gaps in the literature. It will fill a gap in the literature by studying the design and functioning of the EU ETS over a longer time period than previous studies. A further gap in the literature can be filled by researching interest group influence in the latest reform of the policy using the preference attainment method for analysing interest group influence on the reform of the EU ETS, rather than process-tracing as in previous literature. This will allow for comparison of findings across methods.

(8)

8

1.2 Research questions

This thesis aims at answering two research questions, which combined offer some insight into how the policy design has evolved, what effect the policy changes have had on policy outcomes, and what role interest groups played in the latest reform of the EU ETS. The first research question will be answered in chapter 4. This research question is:

How has thedesign and policy outcomes of the EU ETS changed since its implementation in 2005?

This research question will be answered in two steps. First, by analysing how the policy design has changed between the four phases of the policy (analysed in section 4.1), followed by an analysis of what effect the policy design changes have had on the studied policy outcomes, namely environmental effectiveness, abatement investment, competitiveness and political acceptance (analysed in section 4.2). The second research question is:

What interest groups were more influential in the policy reform between the third and the fourth phase of the EU ETS?

In order to answer this research question, the focus will be on comparing the influence of different types of interest groups. This will be done by testing the following three hypotheses (derived from theory and previous literature in section 2.2.2 and tested in section 5.2):

H1: Business interests had more influence on the policy reform outcome than other interest groups.

H2: Business representative associations had more influence on the policy reform outcome than individual firms.

H3: The energy intensive industry sector had more influence on the policy reform outcome than the energy sector.

An analysis of the policy space is important in order to distinguish the counterfactual, i.e. what would have happened without interest group influence, and section 5.1 will therefore analyse what the reasons for reforming the policy between the third and the fourth phase were, what policy changes were decided before the consultation and what the policy outcomes were, in terms of what policy changes were suggested in the European Commission’s proposal following the consultation.

(9)

9

1.3 Summary of main findings

The findings relating to the first research question, about the policy evolution of the EU ETS and its effect on the studied policy outcomes, show that the EU ETS has been environmentally efficient since its first phase, in terms of corresponding to the relevant emission reduction targets over time. The impact of the EU ETS on abatement investment was found to be small, indicating that higher permit prices will be necessary to incentivise abatement investment. The effect of the EU ETS on the competitiveness of regulated firms was found to have little to no carbon leakage effect. Carbon leakage has thus not been an issue, but industry and policy makers agree that the risk of carbon leakage motivates protection through continued free allocation to sectors exposed to the risk of carbon leakage.

The findings relating to the second research question, which focuses on comparing levels of influence between different interest groups on the policy reform of the EU ETS between phase three and four, did not show any empirical evidence of differences in influence between interest groups representing business interests and other interest groups, nor that business associations had more influence than individual firms. However the analysis showed a structural difference in influence between the two main regulated sectors, the energy intensive industry sector and the energy sector, where interest groups from the energy intensive industry were found to have a statistically significantly higher probability of preference attainment. This finding can be interpreted as this type of interest group having had more influence in the latest revision of the EU ETS than interest groups representing the energy sector.

1.4 Thesis outline

This first chapter has so far introduced the research topic, the relevance of this research, the research questions and their focus as well as a summary of the main findings. The final section of this chapter provides an extensive literature overview of the previous literature relating to interest group influence on the reform of the EU ETS. Chapter 2 presents the theoretical framework which combines economic cap-and-trade theory and theory on interest group influence in EU policy making. This framework has been used for identifying a relevant research topic and operationalising the research questions. This chapter also presents hypotheses derived from theory and previous literature. Chapter 3 contains the research design, where the chosen research method is presented and justified, and the research questions are

(10)

10

operationalised. Chapter 4 aims at answering the first research question, by first describing how the EU ETS design has changed between the different phases of the policy and then analysing how these changes has affected the environmental effectiveness, abatement investment and competitiveness of regulated firms. In chapter 5, interest group influence on the latest reform of the policy is analysed using the preference attainment method, with the aim of answering the second research question. The results from chapter 5 is discussed in chapter 6, which furthermore discusses the limitations of the chosen theoretical and methodological approach. Chapter 7 concludes by summarising the findings and answering the research questions, and provides recommendations for future research and policy recommendations.

1.5 Literature overview of interest group influence on the reform of the EU ETS This section will explore previous literature on EU ETS reforms and the role of interest group influence, which has been used for finding a relevant research topic (section 1.1) and will be used for deriving hypotheses (section 2.2.2) and operationalising the research questions (section 3.2 and 3.3).

Relatively few attempts have been made by scholars to study interest group influence on the policy formation and reform of the EU ETS, and these previous studies all use the process-tracing method. Interest group influence on the initial implementation of the EU ETS has been studied by Christiansen and Wettestad (2003) and Markussen and Svendsen (2005). Christiansen and Wettestad (2003) focused on the driving factors leading up to the EU ETS proposal and discussed its design options (regarding cap, coverage, allocation method, linkages to other countries and project-based credits) and their impact on political acceptability, environmental effectiveness, cost-effectiveness and market efficiency. They argue that the proposal for an EU level ETS was affected by the EU trying to take a greater leadership role at the international level, especially due to the uncertainty of US participation in the Kyoto agreement. Furthermore, some member states had already been pursuing an EU level carbon tax, which was prevented by the unanimous voting rule for fiscal policies at the EU level. A non-fiscal ETS on the other hand only requires a majority vote. They also argue that interest groups influenced the ETS design in the EC proposal, particularly the choice of allocation method. Markussen and Svendsen (2005) instead compared the green paper and the final EU ETS directive to analyse the effect of lobbying. They argue that dominant interest groups

(11)

11

influenced the design of the ETS, and that industrial rent-seeking led to the design choice to use a grandfathering permit allocation method over auctioning.

How the EU ETS managed to improve through policy reform has been studied by Drew (2010) and Jevnaker and Wettestad (2017). Drew (2010) applied the Wilson’s typology theory on the case of achieving more stringent allocation method rules in the EU ETS, and argued that this change can best be explained by a shift from a situation of clientelist politics in phase one to interest group politics in phase three. Through document analysis and expert interviews, he found that the energy sector was the lobbying winner in the initial policy formation as they managed to avoid differential allocation treatment in the first and second phase of the policy, leading to significant windfall profits. In the revision for the third phase, the energy intensive industries instead had more influence on the policy reform, as they succeeded in establishing differential rules with the energy sector subject to buying their permits through auctioning, while free allocation continued for the energy intensive industries. He argues that interest group politics did not emerge in phase one because the windfall profit mechanism was unexpected for all actors apart from the energy sector and the energy intensive industry’s lobbying efforts were less coordinated than the energy sector’s lobbying. Informational and transaction costs of collective action in this case thus led to one interest group type having lobbying advantages, which they used to economically benefit from the policy at the cost of diffuse interests (energy consumers) while not being exposed to a carbon price. The policy reform process for phase three was instead characterised by more even collective action and information costs, and competing interest group lobbying resulted in a more balanced reform in terms of rent-seeking. Drew (2010) argues that this case thus showcases the self-destruction process of client politics and the potential to improve an ETS through reform, suggesting that implementing a ‘weak’ ETS might be a better approach than trying to overcome the political opposition to implement a stringent ETS.

Jevnaker and Wettestad (2017) studied the politics of reforming the EU ETS in terms of introducing the market stability reserve, focusing on the roles and interests of the main actors in the policy reform. They used the process-tracing method to analyse what political mechanisms caused the conditions for the adoption of the market stability reserve. Business interests were found to be split between energy sector and energy intensive industries, with the energy sector pro-reform and the latter against. They argue that the energy intensive industries knew they could not stop the MSR, so instead strategically lobbied for an innovation fund and continued negotiations on the carbon leakage rules. They conclude that “The feasibility of

(12)

12

carbon market reform is a function of how well it matches the views of actors within the relevant political system.” (Jevnaker and Wettestad, 2017 p.119) and argue that the tightening of an ETS can be done in multiple reform rounds in order to improve the chances of a supporting majority. More literature is available about the determinants of interest group influence on EU policy-making in general (such as Klüver, 2013; Bunea, 2013; Mahoney, 2008). Mahoney (2008) compared lobbying in the EU and the US. She used the preference attainment method to analyse interest group influence on policy making, and based this analysis on interviews with involved interest groups. She analysed the effects of institutional, issue-specific and interest group specific factors. The findings suggest that interest groups in the EU are likely to achieve 60% of their preferences, which indicates that interest group influence is higher in the EU than in the US. She found support for the hypothesis that institutional factors such as democratic accountability and legislative process affect interest group influence. In terms of issue characteristics, such as the scope of an issue, polarisation of preferences on an issue and the salience of the issue, her findings show that increases in these factors decreases the level of lobbying success. While these effects are similar in the US, they are stronger in the EU. Results for interest group specific factors were more varied. No empirical evidence was found for the effect of interest group resources, such as financial resources and membership. However, the advocate type was found to affect policy influence; interest groups representing citizens’ interests and business associations were found to be more influential in the EU than in the US. Advocates lobbying for the status quo were found to be more likely to succeed, and business interests were more likely to lobby for the status quo while citizen interest groups were more likely to lobby for policy change.

Bunea (2013) researched the determinants of interest groups preference attainment in environmental policy formation in the EU, using stakeholder consultation data to identify preferences and the policy proposal following the consultation to identify policy outcomes. She focused on interest group characteristics and issue-related factors, and found empirical evidence that preferences are more likely to be attained when taking a median preference or a preference to maintain the status quo. She argues that this has implications for the influence of business vs. citizen interests, as business interests thus can influence the implementation of new regulation that would lead to concentrated costs for certain business sectors. Findings also indicate that advocacy type affects preference attainment, as actors representing concentrated interests, such as business interests, were more successful in attaining their preferences than interest groups representing diffuse interests, such as NGOs. However, no empirical evidence

(13)

13

was found for issue-related factors, such as issue salience, polarisation of preferences and inter-organisational ties.

Klüver (2013) also analysed interest group influence using the preference attainment method, focusing on what determines the level of preference attainment. She used stakeholder consultation data to conduct a large N study using a new computerised text analysis method with the aim of testing the validity of this new measurement approach. The findings show empirical evidence that information supply, citizen support and economic power has a positive effect on preference attainment. The effect of the latter two was of a similar magnitude, while the effect of information supply was smaller. This indicates that influence can be gained through information supply to the EC. Contrary to the expectation that businesses have more power, these findings also suggest that interest group influence on policy formulation in the EU is balanced between interest groups with economic power and those with citizen support.

To conclude, this section has explored previous literature relevant for this thesis. Previous literature that focused on the reform of the EU ETS using the process-tracing method (Drew, 2010; Jevnaker & Wettestad, 2017) found that the policy has evolved to become more stringent and that interest groups have influenced these reforms. Both these studies found that the energy intensive industry was more successfully in influencing reforms of the EU ETS than the energy sector. This is an important finding, and will be used for the derivation of hypotheses (in section 2.2.2). Previous studies focusing on interest group influence on EU policy making in general (Klüver, 2013; Bunea, 2013; Mahoney, 2008), further offers insights on what factors are expected to affect interest group influence relevant for this thesis. Some of the determinants of interest group influence used by these studies will be used for deriving hypotheses (in section 2.2.2) and used as control variables (in section 5.2.2). This opens up the possibility of comparing findings between studies, which will be done in section 6.1.

(14)

14

2. Theoretical framework: cap-and-trade systems under interest

group influence

This chapter will present the theoretical framework of this thesis, which combines economic theory on the design and functioning of cap-and-trade systems (section 2.1) and theory on interest group politics and interest groups influence on policy making in the EU (section 2.2). This theoretical framework has been used to identify a relevant research topic and will be used for operationalising the research questions (in chapter 3). Economic cap-and-trade theory focuses on the expected effects of economic incentives, and assumes that the process of forming and implementing an ETS is neutral, allowing the most efficient design to be chosen. In practise, this is not the case as the policy-making process can be influenced by interest groups lobbying their preferred policy design. Combining cap-and-trade theory with theory on interest group influence thus opens up the possibility to systematically analyse the reform process of a cap-and-trade policy in practise.

The cap-and-trade theory section (2.1) will outline what design options there are in creating and reforming an ETS and what effects these design variables are expected to have on policy outcomes, which will be used to answer the first research question about the policy design changes for the EU ETS and their effect on policy outcomes. This section also presents explanations of technical aspects of cap-and-trade systems, which will be referred to throughout following chapters of this thesis. Theory about interest groups influence on policy making in the EU (section 2.2) explains how interest groups influence policy-making in the European context and outlines the factors that are expected to affect the level of influence of interest groups. Throughout section 2.2, hypotheses about the influence of different interest groups will be derived from this theoretical framework and the literature overview. These hypotheses will be used for answering the second research question, about what interest groups had more influence in the policy reform between phase three and four of the EU ETS.

2.1 Cap-and-trade theory

Cap-and-trade systems are a type of market based policy instrument that in principle facilitates reaching emission reduction targets at the lowest possible cost. Cap-and-trade theory builds on Coase’s (1960) argument that pollution externalities can be resolved by defining clear property rights and that the outcome would be the same whether the right to pollute is allocated to the

(15)

15

polluter or the victim. Coase argued that the method of initial allocation of permits does not matter as people and firms will trade the right to pollute until it equals the marginal abatement cost, which would mean that the maximum welfare has been achieved. Compared to a tax on emissions, cap-and-trade systems allow for more flexibility for regulated actors. In principle, a cap-and-trade system creates an incentive to invest in emission abatement technology through the pricing of emissions. Regulated actors face different investment costs, and those who can reduce emissions at the lowest cost will therefore invest in abatement and sell their emission permits while those who face high investment costs will prefer to buy emission permits as long as this is the least costly alternative. Thus the permit price level is very important for the functioning of an emission trading system. Cap-and-trade systems do not set a price on permits, instead the permit prices depend on the availability of permits. An important assumption for achieving emission abatement through an ETS is thus that there is no oversupply of permits, as this leads to too low permit prices to induce abatement investments, or undersupply, causing very high permit prices that might lead to carbon leakage. According to economic theory, the larger the coverage of an ETS the more efficient the policy will be and if an ETS does not cover all sectors or geographical entities, competition distortions arise. The concept of carbon leakage can be defined as “a shift of CO2 emissions from a region with emission constraints to an unregulated area, via a change in relative competitiveness in an open global economy” (Naegele & Zaklan, 2019 p.125). Carbon leakage would thus in the case of the EU ETS be a shift of emissions to industry outside Europe, through the loss of net exports. This section has introduced the main principles of cap-and-trade systems and the next section will describe the policy design options in more detail.

2.1.1 Cap-and-trade policy design variables

Designing and implementing a cap-and-trade system involves four main steps (Fromm & Hansjürgens, 1996). First the total emissions must be decided to create the cap. Second, emission allowances need to be transferred to the covered actors. Third, it has to be made sure that market transactions can take place. Finally, there has to be compliance regulation and monitoring to ensure the trust in the system.

The emission cap is importantly the main design feature that affects the environmental outcome, as this will be the maximum level of pollution assuming compliance. The cap should therefore be a direct translation of emission reduction targets. Cap-and-trade systems are thus a quantity

(16)

16

based type of market policy instrument. This means that prices will be determined through the demand for a fixed supply. If demand is high in relation to available permits, prices will be high and vice versa. These permits are then allocated to the actors that are covered by the system. Coverage plays an important role for market efficiency, as market liquidity increases when more actors and greenhouse gases are included.

There are different methods for allocating emission permits: auctioning, free allocation through grandfathering, and benchmarking. Auctioning has several advantages, it creates incentives for abatement already at the initial allocation, it treats incumbents and new entrants the same, the initial allocation auctioning prices can help planning for the second market (where permits are traded between actors) and auctioning revenues can be reinvested to e.g. subsidise abatement technology to further strengthen the effectiveness of the ETS (Fromm & Hansjürgens, 1996). The pitfall of the auctioning method is that it is difficult to implement politically in practise, as regulated industries tend to object to the induced costs of auctioning (ibid.). Free allocation based on historic emissions, also called grandfathering, is a more politically feasible design choice. However, free allocation can inhibit both incentive effects and distributive targets (ibid.). Free allocation benefits high emitters more relative to actors that have already invested in abatement, and can therefore be seen as unfair treatment of the best performing actors. Using historical emissions is problematic as demand for permits depend on many factors, some fluctuate with the business cycle and can cause over or undersupply of permits. Grandfathering can also be discriminating against new entrants, if they have to buy their permits while incumbents get theirs for free. Benchmarking is an alternative method for determining free allocation. It is a reference value for emissions related to a specific production activity. Benchmarks are set at a level so that the best performing actors are given free allocation, while actors that emit more have to buy permits for the exceeding emissions. Benchmarking can thus be a more efficient way of providing free allocation of permits, however it demands access to detailed emission data from industry which can be difficult to access since this is sensitive information.

To conclude, this section has introduced cap-and-trade theory that will be used to answer the first research question about the policy design changes for the EU ETS and their effect on policy outcomes (in chapter 4). This section has furthermore provided some explanation of technical aspects of cap-and-trade systems, which will be referred to throughout the following chapters of this thesis. This section has also demonstrated that economic cap-and-trade theory focuses on the expected effects of economic incentives, with little attention given to the political process

(17)

17

of designing and implementing an ETS. Instead it is assumed that the most efficient design will be chosen. In practise, this is not the case as the policy-making process can be influenced by interest groups lobbying their preferred policy design. To be able to systematically analyse the reform process of the EU ETS it is thus necessary to combine cap-and-trade theory with theory about interest group influence, which is introduced in the next section.

2.2 Interest group politics and the influence on policy-making

The concept of interest groups go under many names in the literature and a definition is therefore useful to ensure clarity. Beyers et al. (2009) argue that three factors determine whether an actor is an interest group: organisation, political interest and informality. Organisation refers to that the actor has to represent aggregated individual preferences, political interest refers to that the actor is trying to influence policy outcomes and informality refers to how this influence is being pursued without taking public office or being subject to the state (ibid.). This definition covers actors such as companies, business associations, trade unions, NGOs and research institutes.

Interest group politics is characterised by concentrated costs and benefits of regulation, which incentivises groups to organise and lobby their interests, as the distribution of costs and benefits of regulation will be uneven and the group with the most influence on policy therefore will benefit from policy-making outcomes at the expense of other groups (Wilson, 1980). This situation has less favourable conditions for rent-seeking than in situations with client politics, defined as situations when costs of regulation are spread out and benefits are concentrated (ibid.).

Interest group influence can be defined as “a causal relation between the preferences of an actor regarding an outcome and the outcome itself” (Nagel, 1975 p.29). However, interest groups’ preference attainment does not necessarily equal influence, it can also be a case of luck or alignment with other interest groups’ lobbying efforts (Klüver, 2013). To be able to distinguish whether the alignment of an actor’s preference and policy outcome is due to influence or luck, it must therefore be possible to empirically identify a statistically significant systematic effect of actor properties and the policy outcome. The following sections in this chapter go deeper into the theory on interest group influence and hypotheses are derived (in section 2.2.2) which will be tested in chapter 5.

(18)

18

2.2.1 The challenge of measuring interest group influence

There are three main challenges to measuring interest group influence: “the existence of different channels of influence, the occurrence of counteractive lobbying and the fact that influence can be wielded at different stages of the policy process.” (Dür, 2008 p.561). Interest groups can lobby using channels such as direct lobbying, outside lobbying, engage in the selection of decision-makers or use their structural power (ibid.). Counteractive lobbying makes measuring influence difficult because it means that the influence of one interest group might be counteracted by the influence of another, causing an underestimation or even non-identification of the first group’s influence. Lobbying at different stages of a policy process, i.e. in the policy formation or decision-making stage, also makes it difficult to estimate total influence. If interest groups are successful in the policy formation stage, they may not need to lobby further in the policy making process (ibid.). To limit the impact of these problems on the results of this thesis, the measurement of interest group influence will be limited to a measurement of interest group influence in the policy formation stage of the latest reform of the EU ETS. This will be further discussed in chapter 3.

2.2.2 Interest group influence determinants and hypotheses

Dür and De Bièvre (2007) present three categories of theoretical factors that are expected to explain interest group influence: the institutional context, interest group properties, and issue-specific factors. The institutional context affects the influence of interest groups as different institutional contexts may favour different balances of interests and thus empower or disenfranchise certain interest groups (ibid.). Schneider and Baltz (2004) have for example found that concentrated interests are advantaged by the EU institutional structure, at the cost of more diffuse interests. Furthermore the institutional context affects interest groups’ access to policymakers and policymakers need for resources from interest groups, such as information (Dür & De Bièvre, 2007). For example, the stakeholder consultations held by the European Commission is a way of giving interest groups access to the policy making process, in exchange for information about their views on policy proposals and technical information that can help making policies more efficient. Including interest groups in policy making can also be viewed as an attempt by the European Commission to increase its democratic legitimacy (Dür & De Bièvre, 2007). By involving stakeholders in the policy making process and allowing some influence, the political acceptance of a policy can be increased.

(19)

19

Interest group properties is the second category of factors affecting interest group influence, and these revolve around access to useful resources and advocate type, referring to what type of interest an actor represents. Interest groups with access to more resources, such as funding, expertise and knowledge about the policy area and other actors’ positions, should have more influence than groups with less resources (Dür & De Bièvre, 2007). Whether an interest group represents concentrated or diffuse interests also affects influence, as those representing diffuse interests face higher bargaining costs and lower marginal benefits and therefore are less likely to organise and lobby their preferences (Olson, 1965). Diffuse interest representatives may also have less access to funding and expertise (Dür & De Bièvre, 2007). On this basis it is expected that business interests, which are concentrated and generally have access to more resources, have more influence than interest groups representing diffuse interests, such as NGOs (Bunea, 2013; Mahoney, 2008). This will be tested by the following hypothesis:

H1: Business interests had more influence on the policy reform outcome than other interest groups.

Similarly certain organisational structures theoretically have more lobbying success because they have more resources and higher levels of legitimacy at their disposal, e.g. business associations versus individual firms (Coen, 2009). Klüver (2013) also argues for an analytical distinction between associations and companies, as these differ in terms of membership, which might affect their lobbying success. Business associations are therefore expected to have more influence than individual firms in the policy reform. This will be tested by the following hypothesis:

H2: Business representative associations had more influence on the policy reform outcome than individual firms.

It is also possible that there are divisions of interests and influence between different sectors. Put concisely by Bunea: “The concentration of lobbying efforts on the emerging lobbying sides translates into different levels of preference attainment depending on the strength of the advocacy side” (2013, p.555). In the case of the EU ETS, as suggested by the literature review (section 1.2), there is a division of interests between the regulated sectors, i.e. between the energy sector and the energy intensive industries (Jevnaker and Wettestad, 2017; Drew, 2010). Because the energy intensive industries had more at stake in the revision for the fourth phase (Jevnaker and Wettestad, 2017), they are expected to have put more effort into lobbying the

(20)

20

reform and therefore had more influence on the policy reform. This will be tested by the following hypothesis:

H3: The energy intensive industry sector had more influence on the policy reform outcome than the energy sector.

Issue-specific factors is the third category of factors that affect interest group influence. It is anticipated that influence varies depending on whether a policy issue is distributive, regulatory or redistributive, since these shapes the lobbying landscape. Dür and De Bièvre (2007) argue that regulatory policies are more likely to involve opposing interest groups, while interest group coalitions are more likely to be formed on distributive policy issues and little collective action at all is likely for redistributive policy issues. Interest group influence is also expected to depend on how salient a policy issue is, referring to the attention given to a policy issue by the public (Dür & De Bièvre, 2007). This concept is difficult to measure. Therefore, it has for example been operationalised by measuring the salience of each policy issue as the share of interest groups that expressed their preference on an issue (Bunea, 2013). Furthermore the technicality of a policy issue also affects interest group influence, as more technical issues demand more input from actors in terms of information which should lead to more influence if interest groups supply this information (Dür & De Bièvre, 2007). Issue-specific factors will be controlled for in the statistical analysis (in section 5.2.2).

To conclude, this section has focused on interest group influence theory and how this can be applied to the research topic of this thesis. Three hypotheses have been derived from this theory in combination with previous literature, which will be tested in the preference attainment analysis in chapter 5. These relate to whether the interest group represents business interests or other interests, whether the interest group organisation type is a business association or an individual firm and whether the interest group represents the energy intensive industry or the energy sector. Furthermore, this section has presented issue-specific factors that are expected to affect interest group influence, which will be included in the preference attainment analysis in section 5.2.2 as control variables. This chapter has thus provided a theoretical framework that can be used to systematically study the impact of interest group influence on the reform of the EU ETS. The next chapter outlines how this analysis will be carried out.

(21)

21

3. Research methods

This chapter will first describe and justify the chosen methods, then summarise the conceptual framework (section 3.1), operationalise the research questions and introduce what sources will be used (section 3.2 and 3.3). The limitations of the chosen approach will be discussed in section 6.2. This thesis will be based on a single case study and the preference attainment method, for the first and second research question respectively. The single case study method can be defined as "an in-depth study of a single unit (a relatively bounded phenomenon) where the scholar’s aim is to elucidate features of a larger class of similar phenomena." (Gerring, 2004 p.341). The case study method is appropriate for this thesis in its aim to explore how a cap-and-trade system can be reformed by studying how the EU ETS has evolved. While the single case study method has limited generalisability, the EU ETS can be seen as an exceptional case and findings regarding the potential and limitations for policy reform of such a large and well established ETS system can still provide lessons for other, current and future, ETS systems.

There are three main methodological approaches to study the influence of interest groups on policy-making in the EU: process-tracing, attributed influence and the preference attainment method (Dür, 2008). As demonstrated in the literature overview, process-tracing is the most common method of researching interest group influence on the reform of the EU ETS. This method has advantages such as the ability to take rival explanations into consideration to argue for causality and the possibility to use interviews for information gathering that can provide insights that are unavailable from other sources (Dür, 2008). There are however also methodological problems, such as “collecting empirical evidence that is precise enough to cover all steps of a causal process; cross-checking evidence gained from interviews against other sources; identifying a yardstick to assess what ‘influential’ means; avoiding inferences about influence from the level of interest group activity; and generalizing from small-N studies.” (Dür, 2008 p.563).

The attributed influence method usually relies on surveys where interest groups are asked to self-assess their influence or the influence of other groups (ibid.). It importantly produces a measurement of perceived influence rather than actual influence, and therefore has low validity if the aim is to study influence rather than perceived influence. This method is associated with problems of self-estimation bias, in terms of both over and underestimation of influence, for strategical reasons or due to a lack of information (ibid.).

(22)

22

This thesis will instead use the preference attainment method to analyse the influence of different types of interest groups in the latest reform of the policy. Put simply, “the idea is that the distance between an outcome and the ideal point of an actor reflects the influence of this actor.” (Dür, 2008 p.566). Other factors that shift outcomes can also be controlled for. This method has several advantages compared to the methods described above. It allows for detecting ‘invisible’ influence, such as secret lobbying and structural power, as well as influence achieved through different channels. Large-N studies are possible with this method because the preference attainment of many interest groups can be estimated. Using many cases decreases the risk of assessment errors affecting the estimate, and makes it possible to study the level of influence (ibid.). Process-tracing in comparison is often limited to whether an interest group had influence or not.

The main problems associated with the preference attainment method are how to determine actors’ preferences, control for other factors affecting the policy outcome and not being able to identify the channels of influence (Dür, 2008). For this thesis, the determination of preferences problem is minimised by using data from a stakeholder consultation survey (EC, 2015b), where interest groups are asked for their preferences on specific policy issues which facilitates comparable preference information from many interest groups on the same policy issues. This data furthermore provides actual policy preferences of interest groups, presented to policy-makers in the hope of influencing policy, and is thus a good data source for identifying interest group policy preferences. Which channel influence is exerted through will not be identified, since this is not within the scope of the thesis. The problem of alternative explanations is difficult to overcome and will be discussed in section 6.2, where the limitations of the research will be discussed.

3.1 Conceptual framework and operationalisation of research questions

The conceptualisation of the theoretical framework applied to the reform of the EU ETS can be summarised by figure 1, which describes the effect the studied policy design variables have on certain policy outcomes, as well as where in the policy reform process interest group influence can be exerted. Blue arrows represent a direct effect on a policy outcome, green arrows represent a direct effect of policy variables that will be introduced in phase four on a policy outcome and red arrows represent the indirect effects of overallocation on policy outcomes. The figure highlights that policy reform is a circular process of deciding on a certain design that

(23)

23

lead to certain outcomes, which have implications for the political acceptance of the policy. If the acceptance of the policy design and outcomes is too low, reform will take place again to decide on rules that in turn lead to a new balance of interests. Interest groups participate in this process to try to influence the policy reform so that it benefits their interests.

This figure is further explained through the operationalisation of the research questions in the following sections of this chapter. The first research question (operationalised in section 3.2 and analysed in chapter 4), about the evolution of the EU ETS design and its effect on policy outcomes, fits within the top half of this scheme, as the focus is on how these policy design changes has affected the policy outcomes on the right side. The second research question (operationalised in section 3.3 and analysed in chapter 5) about interest group influence on the latest reform of the EU ETS can be placed in the bottom right corner of this scheme, as the focus is on interest group influence in the policy formation stage of the policy reform process.

Figure 1: Conceptualisation of the link between ETS design variables and policy outcomes,

and interests group influence on the policy reform.

(24)

24

3.2 Operationalisation of the first research question

How has the design and policy outcomes of the EU ETS changed since its implementation in 2005?

To answer the first research question, the policy design in the four phases of the EU ETS will be compared in terms of emission cap, coverage (sectors, countries, GHGs), allocation of permits (method and quantities), permit prices and other significant policy changes. These findings will be summarised and compared to see how the policy has evolved through the four policy phases since its implementation in 2005. The focus on these policy design variables was chosen because of their prominence in previous literature and centrality in economic cap-and-trade theory.

To analyse how the policy changes have affected the environmental effectiveness, abatement stimulation, competitiveness, market efficiency and political acceptance of the policy, theory introduced in chapter 2 will be linked to the findings about how the policy design has changed. Literature research will be employed to answer this research question, using a mix of primary and secondary sources and this research question will be explored in chapter 4. Which policy design variables affect what policy outcomes are summarised in figure 1 and these expected effects are described in the following paragraphs.

The environmental effectiveness of the EU ETS is mainly affected by the setting of the cap, as this will be the maximum emissions from the sources covered by the system. Therefore it is important that the cap is coherent with the emission reduction targets of the EU. The environmental effectiveness will thus be analysed by comparing the cap with the relevant emission reduction targets over time. Compliance also affects the environmental effectiveness, as the cap would otherwise be exceeded. However due to the scope of this thesis, compliance will be assumed.

It is further relevant to compare the yearly cap with the yearly verified emissions to identify overallocation which in theory causes low prices, and which in turn causes lower incentives to invest in abatement technology. This is because cost-minimising firms will choose the cheapest option between buying permits or invest in reducing their emissions to avoid having to buy permits. Descriptive statistics will be used to explore overallocation and permit prices between 2005 and 2018 and the effect of low prices on abatement investments will be researched through a review of empirical literature on the topic, where methods, data and results will be compared

(25)

25

to see if the findings are consistent. Furthermore, the potential effects of the phase four market stability reserve, investment fund and modernisation fund on abatement investment will be discussed.

The competitiveness of regulated firms is in theory affected by the carbon price, as competitors not facing the same carbon costs have a cost advantage that make them more competitive. In practice, this carbon leakage issue has been dealt with by giving free allocation to exposed industries. The effect of changes in allocation method on competitiveness will be researched through a review of the empirical literature on carbon leakage, where methods, data and results will be compared to see if the findings are consistent.

Market efficiency is affected by coverage, as market liquidity increases when more actors and greenhouse gases are included. However, because the EU ETS has covered many installations since its implementation and has not changed remarkably in this regard, this link between coverage and market efficiency will not be analysed further.

As indicated in figure 1 above, most of the cap-and-trade design variables have an effect on the political acceptance of the policy. Essentially, political acceptance of the policy depends on how economic and environmental interests are balanced, and to what extent these interests overlap with the interests of the other stakeholders. The political acceptance of the policy will be studied by analysing the effect of the policy changes in the four phases. Support for a stringent cap depends on the level of climate mitigation ambition, which varies between stakeholders. The coverage affects the support from industry, as there might be sectorial preferences of being excluded, but when that is not possible firms may argue for broader coverage. Initial allocation method is very important for acceptance from businesses, which prefer free allocation over auctioning. While permit prices are not set by the policy, it still affects the political acceptance as lower permit prices means lower induced costs from the policy. The market stability reserve will deal with the permit surplus, which would increase permit prices and thus affect political acceptance from industry as described above. Two new subsidies will be introduced in phase four which are likely to increase political acceptance.

(26)

26

3.3 Operationalisation of the second research question

What interest groups were more influential in the policy reform between the third and the fourth phase of the EU ETS?

This research question is limited to interest group influence in the latest policy reform. While studying interest group influence in multiple EU ETS reforms could have provided some explanation as to why the policy has changed the way it has, this was not possible within the scope of this thesis and this is therefore left to future research. Furthermore, interest groups can try to influence the policy reform process both in the policy formation stage and the decision-making stage. These cannot be studied simultaneously, so this thesis will focus on influence exercised in the policy formation stage. Previous research has justified focusing on the policy formation stage as it is considered to be the stage where interest groups have the best chances of influencing policy outcomes (Bunea, 2013; Klüver, 2013).

In order to answer this research question about the influence of different interest groups in the latest reform of the EU ETS, an original dataset will be constructed from data on actors’ preferences on specific policy issues in the latest policy reform from the stakeholder consultation about the policy reform (EC, 2015b). This consultation consisted of an online survey with open-ended questions. It took place between the 19 December 2014 and 16 March 2015 and citizens, industry stakeholders, trade unions, NGOs and public authorities were invited to contribute. Nine policy issues were identified from the survey questions (presented in section 5.1), using the following definition of a policy issue: “An issue represents a potential controversy centred around a particular aspect of a legislative proposal and for which stakeholders and policy-makers may disagree about a desirable political outcome” (Klüver et al., 2015 p.451). Appendix 1 contains the stakeholder consultation survey questions and shows how policy issue identification was done.

Content analysis is very time consuming and due to time constraints, a smaller sample will therefore be analysed. The sample will be cleared of survey entries that have not answered most survey questions, as it would not be possible to assess their preferences. Furthermore the analysis will focus on actors that fit the definition of interest group used in this thesis (see section 2.2), leading to the exclusion of citizen and government institution entries. Random sampling will be used to create a smaller sample. However since the main focus of the research question is to compare levels of influence between different types of interest groups, all entries from NGOs and academic/research institutions will be included. Random sampling on the

(27)

27

whole dataset would have led to a loss of an already small numbers of participants of these interest group types.

Table 1 describes the survey data and the sample used for the analysis. The number of entries from each interest group is a valuable finding in itself, as it shows what interests succeeded with mobilising collective action to try to influence the policy reform and might tell something about who has most at stake in this reform. Table 1 shows that business interests were far more represented than diffuse interests and that the energy intensive industry representation was several times higher than representation from the energy sector in the stakeholder consultation regarding the latest reform of the EU ETS.

Table 1: Description of full survey dataset and sample,

divided by interest group type and sector.

The identification of interest groups’ preferences from EU stakeholder consultations can be justified by the high degree of institutionalisation in the dialogue between stakeholders and policy-makers within the EU (Bunea, 2013). Furthermore, while lobbying in many cases is not documented and therefore hard to measure, data from stakeholder consultations provides actual policy preferences of interest groups presented to policy-makers in the hope of influencing policy and is thus a good data source for identifying interest group policy preferences. Policy outcomes on the identified policy issues were identified from the European Commission’s policy reform proposal (EC, 2015c) released following the stakeholder consultation. To estimate interest groups’ preference attainment, a binary variable describing whether or not the

Interest group divisions Survey data Sample

Total N 440 163

Interest group

type

Business 154 55

Trade association representing businesses 162 55

Small and medium enterprise 43 20

SME business organisation 5 0

Government institution/regulatory authority 15 0

Academic/research institution 7 5

NGO 33 28

Citizen 4 0

Sector

Energy intensive industry 235 81

Energy sector 60 55

(28)

28

policy preference of the actor was achieved was created for each policy issue, where 1=preference attained and 0=preference not attained.

For the replicability and transparency of this study, the coding of interest groups’ preferences attainment is described extensively and examples of interpretations are given in the code book (attached in appendix 2). To ensure that the reliability of the coding scheme is satisfactory an inter-coder reliability test was conducted, where another person was asked to code the preference attainment for 100 interest group-issue dyads. The purpose of this test was to verify whether the coding scheme is satisfactory. A good coding scheme should guide coders towards the same interpretation of the text analysed. Perfect agreement between coders is uncommon, and in the case of this thesis the main concern was the technical nature of the content to be analysed. Considering the difficulties of achieving perfect agreement, the question of what a suitable level of reliability is arises. One commonly used method to test reliability is Krippendorff’s alpha. This method measures agreement between ratings whilst also accounting for the probability of giving the same rating. Krippendorff (2013) argues that intercoder reliability tests that score above α=0.8 can be interpreted as reliable, whereas scores between α=0.667 and α=0.8 should be interpreted carefully. The inter-coder reliability test results showed that ratings converged on 85% of the cases, and these results give a Krippendorff’s alpha of 0.696. This is on the lower end of the acceptable reliability score and the results of the analysis should therefore be interpreted with certain caution. As expected, the second coder reported struggling with the technical nature of the content, which would partly explain this relatively low score and justifies moving forward with the present code book.

3.3.1 Variables and model specifications

First, the level of preference attainment of different stakeholder groups will be explored descriptively. In order to do this, the preference attainment of each interest group in the sample was translated into a share of total preferences expressed. Using total preferences expressed rather than total possible preferences attained is motivated on the basis that preferences often were not expressed about all policy issues. This is likely because actors do not necessarily have preferences about policy issues that does not affect them, and using total possible preferences attained would therefore likely have created an underestimation bias. This will be further discussed in section 6.2. Actors that expressed preferences on less than half of the policy issues

(29)

29

were excluded from the descriptive analysis as their preference attainment levels were more extreme and therefore excluding them increases the reliability of the findings.

This will be followed by a statistical analysis testing the hypotheses formulated in section 2.2.2. Three model specifications will be used:

Model 1: Preference attainment = β1*Business interest + e

Model 2: Preference attainment = β1*E-I Industry + β2*Association + e

Model 3: Preference attainment = β1*E-I Industry + β2*Association + β3*Salience +

β4*Polarisation + e

The dependent variable, preference attainment, is binary and reflects whether an actor’s preference was attained for a specific policy issue. The independent variables for whether the actor represent business interests, whether the actor represent energy intensive industry and whether the actor is a business association are binary variables. Model 1 tests whether business interests had more influence on the policy reform outcome than other interest groups. Model 2 tests whether the energy intensive industry sector had more influence on the policy reform outcome than the energy sector and whether business representative associations had more influence on the policy reform outcome than individual firms. Klüver et al. (2015) argue for the necessity of understanding how contextual factors affects lobbying, including the effect of policy-related factors such as the salience and degree of conflict on an issue as well as institutional features. Model 3 therefore also includes policy issue related control variables for the salience of an issue, operationalised as the share of interest groups that expressed an opinion on a specific issue, and the degree of polarisation on an issue, operationalised as the deviation of the mean from 0.5 (maximum polarisation of preferences occurs when the mean preference attainment for a policy issue equals 0.5, the further away from 0.5 the mean is the more agreement there is on the issue). Institutional features cannot be controlled for as these remain constant. Table 2 describes the variables used in the statistical analysis.

(30)

30

Table 2: Description of variables

Variable Type of variable Mean SD Min Max

Dependent variable:

Preference attainment Binary 0.543 0.498 0 1

Independent variables:

Business interest Binary 0.795 0.402 0 1 Energy intensive industry Binary 0.608 0.488 0 1

Association Binary 0.423 0.494 0 1

Salience Continuous 59.1 17.8 27.0 83.4

Polarisation Continuous 0.188 0.133 0.004 0.396

The mean of the preference attainment variable in table 2 can be interpreted as the mean preference attainment across the sample, which expressed in percentage is 54%. Similarly, 79.5% of the expressed preferences were expressed by interest groups representing business interests. From the business interest representative interest groups, 60.8% of preferences expressed came from the energy intensive industry and 42.3% from associations. The salience of the policy issues in the analysis, expressed as the share of interest groups that expressed a preference on each policy issue, range from 27-83.4% with an average of 59.1%. The interpretation of the polarisation variable is less direct. Expressed on a scale of 0-0.5, where 0 represents the maximum polarisation on a policy issue, this variable ranged from 0.004-0.396 with an average of 0.188.

(31)

31

4. The policy evolution of the EU ETS

This chapter aims at answering the first research question, about how the design and policy outcomes of the EU ETS have changed over time and what policy issues remain. This will be done in two steps; section 4.1 outlines the policy design in the four phases of the policy and section 4.2 analyses how these design changes have affected the studied policy outcomes. The chapter is summarised and concluded in section 4.3.

4.1 Policy design in the four phases of the EU ETS 4.1.1 Phase one

The first phase of the EU ETS (2005-2007) was intended to be a trial period, to gather data, experience and build the infrastructure for monitoring and verification. The policy design was decided when the EU ETS directive was adopted by the European Council in 2003 (Directive 2003/87/EC). Permit allocation was managed at the national level, as each member state decided on a National Allocation Plan (NAP) which had to be approved by the EU. The EU wide emission cap was thus the sum of all the NAPs, rather than a predetermined cap as suggested by economic theory. In this initial phase, the EU ETS covered the EU-25 countries. The regulated emitters were electricity generators and energy intensive industrial installations (more specifically production and processing of ferrous metals, cement, lime, glass, bricks, ceramics, pulp, paper and board), and the policy only covered carbon dioxide emissions (Ellerman et al., 2010). Member states were allowed to exclude small emitters since these face higher regulation costs in relative terms. The allocation method in this phase was free allocation based on grandfathering, but because of the lack of data on historical emissions businesses were in practise able to report higher emissions than their actual emissions, which led to a permit surplus (EC, n.d.). Permit prices averaged between €10-15 and varied between €0-30 as prices crashed at the end of the period (see figure 1 in appendix 3). This was because banking of permits between phase one and two was not allowed and all actors therefore tried to sell any excess permits in the final year.

Reforming emission trading systems : interest group influence on the reform of the EU ETS for 2021-2030