Power dynamics in transition : exploring the governance of complex socio-eco-technical systems : the case of biomass co-firing in coal plants in The Netherlands

(1)

Master Thesis Public Policy and Governance

Power Dynamics in Transition

Exploring the governance of complex socio-eco-technical systems The case of biomass co-firing in coal plants in the Netherlands

Author: Kristiaan Kok MSc Student Number: 10263519 Supervisor: dr. Anne Loeber Program: MSc Political Science Track:

Public Policy & Governance

Second examiner: prof. dr. John Grin

(2)

MSc Thesis Political Science

The research described in this thesis was carried out as part of the Master Research Project (30 EC) in order to obtain a Master of Science degree in Political Sci-ence at the University of Amsterdam.

(3)

Summary

This thesis aims to investigate the case of biomass co-firing in Dutch coal plants for under-standing the role of power in the dynamics of complex socio-eco-technical systems. That is an important research topic since the nature of the co-firing practice has become an important part of Dutch transition policy but at the same time is very much constested due to the uncertainty about the sustainable nature of the co-firing practice.

First an analytical framework for understanding the structural and dynamical proper-ties of complex systems is developed that aims to deal with some of the criticisms on existing frameworks such as the multi-level perspective (MLP). This new approach con-siders complex systems to have a societal function (e.g. energy) and to exist out of multiple constellations (e.g. coal, solar, biomass) with a distinct sub-functioning that in turn are constituted by human, physical, social and ecological components. Through testing hypotheses on the nature of the operationalized concept of functional dominance of constellations (indicating the relative share of systemic functioning of the constellation) it is argued that the approach developed in this thesis has empirical relevance for our case.

The concept of power is brought in to the complex systems approach and for that pur-pose this thesis reconceptualizes power as restructuration and by as an intrisicly rela-tional, emergent and symmetrical concept. Based on this proposition several dimensions of power-exertion are identified in the case that help to understand the complex relations between components-networks in the energy system. Finally and after systemic empirical analysis the concept of power is related explicitly to drivers and barriers for dynamics. Several mechanisms are identified in the case linking the role of power to those so called structural conditions. As a final proposition to be explored in further research, this thesis adopts the view that power exertion can be seen as a causal mechanism bringing about (transitional) systemic dynamics through constructing conditions for constellations within the functionalist system. Through this case study on biomass co-firing this thesis has aimed to contribute to a better understanding of both the complexity of politics as well as the politics of complexity.

(4)

Acknowledgements

Writing this thesis was a challenging and exciting endeavor and marks the completion of my masters degree in Political Science. I owe my great gratitude to several people who have helped me throughout this year and especially in completing this thesis.

First of all I owe great thanks to my supervisor Dr. Anne Loeber. Thank you for your contagious enthusiasm, the involved way in which you have been guiding and reas-suring me (and the others from our group) throughout the thesis process and our fruitful and insightful discussions. Secondly, I would like to thank Prof. Dr. John Grin. Thank you for being the second examiner of this thesis, but notably also for your guidance, enthousiasm and support throughout the year and for your tireless efforts in helping me finding a way to start my academic career.

I also thank all the interviewees who have shared their knowledge, insights and expe-rience with me. Obviously, without those insights the empirical work would not have been completed. Thanks to my fellow students from the thesis working group for inter-esting discussions and the mutual support in difficult times.

(5)

(6)

1 Introduction

This thesis aims to investigate the role of power in transitional dynamics in the case of biomass co-firing in Dutch coal plants. In this chapter, first the relevance of studies focusing on sustainability is highlighted (section 1.1) after which the role of biomass (co-firing) in the Dutch energy system is briefly described in section 1.2. Then a brief overview of transition studies as a guiding paradigm for Dutch transition policy as well as this thesis is presented, where some of the challenges for this field of research are indicated in section 1.3. Finally, the research objectives (section 1.4) as well as questions are presented together with an outline of the thesis (section 1.5).

1.1 Towards Sustainable Practices

The world faces a great challenge in transitioning towards sustainable practices in mul-tiple societal (often overlapping) functional domains including energy (EC, 2009; IEA, 2010; IEA Bioenergy, 2017; Panwar et al., 2011; Verbong and Loorbach, 2012), agri-culture (Tilman et al., 2002; Harvey and Pilgrim, 2011; Frison, 2016), fishery (Pauly et al., 2002), water management (Richter et al., 2003; Sendzimir et al., 2010), natural resource management (Rammel et al., 2007; Dold, 2008; Nabuurs et al., 2016), forest and ecosystem management (Searchinger et al., 2009; Nabuurs et al., 2016; Mai-Moulin et al., 2017; Fingerman et al., 2017), built environment (Sartori and Hestnes, 2007) and trans-portation (Kivits et al., 2010; Ragauskas et al., 2006; EC, 2010c). In the next decades a growing population including an (accompanied) expected increase in food, water and energy demand as well as the consequences of increased greenhouse gas emissions and the resulting climate change effects will increase the urgency of tackling these challenges even further (Hoffert et al., 1998; Goldemberg, 2000; Moss et al., 2010; IEA, 2010). Against this background, this thesis zooms in at the challenges involved with governing the tran-sition towards sustainable energy practices in the Netherlands. In particular, it focuses on the role of biomass co-firing in Dutch coal plants as part of that broader global context in which todays societies find themselves confronted with the need to transition towards a more sustainable energy system.

(9)

1.2 Biomass for Energy

In 2013 the Sociaal Economische Raad (SER) presented the Dutch Energieakkoord, a policy agreement between the Dutch government and over 40 organizations in the field of energy and sustainability. In the Energieakkoord, agreements where made to decrease carbon-dioxide emissions, reduce the effects of climate change and to foster new (techno-logical) pathways towards sustainable energy production in the Netherlands (SER, 2013). As agreed upon in the Energieakkoord, the Dutch government aims to produce 14% of its energy sustainably in 2020. To reach that target, one of the main pillars of Dutch energy policy is the stimulation of the use of biomass as a sustainable energy source as an alternative to fossil fuels. Biomass is a broad category, it for example encompasses the co-firing - the combustion of wood from forests or agricultural residues in former coal plants - but also the use of feedstock-based biofuels for transportation purposes. This focus on biomass is illustrated by the fact that several billions of euros are being given to big energy companies like RWE in the form of governmental SDE+ subsidies, in order for them to transition towards a practice of co-firing biomass in coal plants in the last five remaining coal plants in the Netherlands (Min EZ, 2017c). In 2016 alone, three ‘biomass co-firing projects’ received 3.6 billion euros (in two rounds), expected to contribute to the production of 14.4 PJs of energy, equal to 0.6% of energy production in 2020 (RVO, 2016, 2017; Min EZ, 2017b).

However the use of biomass for energy production especially in co-firing projects is con-tested (Stupak et al., 2007; Woutersen et al., 2017; Straver, 2018). Environmentalists and scientists have argued that the use of biomass for energy has negative effects on biodiversity, destroys forests and associated ecosystems and interferes with other land-use practices such as agriculture (Harvey and Pilgrim, 2011; Schulze et al., 2012). Some say that there is not enough residual biomass in forestry and agriculture to make a sig-nificant impact on a sustainable energy sector (Ter-Mikaelian et al., 2015), while other research argues there is (Ecofys, 2008, 2017). Furthermore, scientists are currently debat-ing whether in fact the co-firdebat-ing of biomass is CO2-neutral. Many scientists argue that the use of residual forest wood in fact contributes to a large CO2 debt, extending up to several decades, because removing the residual wood causes disturbances in the carbon-metabolism of forests (Searchinger et al., 2009; Cherubini et al., 2011; Repo et al., 2015; Chatham House, 2017). Others argue that those effects are relatively weak or simply not present (Nabuurs et al., 2017; IEA Bioenergy, 2017). Some scientists in favor of using biomass for energy acknowledge barriers to implementation, by for example focusing on solving technical uncertainties (Bentsen and Felby, 2012) or by focusing on political and

(10)

legal barriers (Sikkema et al., 2014). This brief introduction illustrates the complexity of the practice of biomass co-firing, which ncompasses social, ecological and technical components.

1.3 Transition Studies as a Guiding Paradigm

Interestingly though despite its very much contested nature the practice of biomass-to-energy has rapidly become an highly institutionalized practice in the biomass-to-energy system in recent years. In fact, the practice of biomass-for-energy is considered a vital part of the Dutch energy transition (SER, 2013; Ecofys, 2015). Dutch transition policy is guided by the paradigm of transition studies (Kemp et al., 2007; Kern and Howlett, 2009; Verbong and Loorbach, 2012). Transitions here are considered to be non-linear processes of struc-tural change in socio-technical systems (Geels and Schot, 2007; Rotmans and Loorbach, 2010). The energy sector with its complex socio-technical nature is often the empirical object of transition studies literature: Markard et al. (2012) in their review article on this emerging field indicate that 36% of transition studies literature involves research in to the energy sector.

Transition studies by its very nature is an interdisciplinary field. Insights from soci-ology, political science, economics, science and technology studies, as well as complexity science are combined in aiming to understand the behavior of complex socio-technical sys-tems. A central concept in transition studies is the multi-level perspective (MLP) (Geels and Schot, 2010; Grin, 2010; Geels, 2011). In this perspective, used as an analytical framework, transitional dynamics in systems is driven by the emergence technological innovations (niches, the microlevel), that compete with each other in challenging domi-nant material, discursive and social structures and practices in the system (regimes, the mesolevel). These levels themselves are embedded and under continuous influence of ex-ternal social, physical or cultural processes beyond the direct scope of agency (landscape trends, the macrolevel).

However, in recent years two important and related branches of criticism have emerged that require theoretical and empirical attention involving the capacity of the MLP to understand both systemic structure as well as systemic dynamics. The first criticism involves the MLP as an analytical framework for understanding systemic configuration and structural properties. Scholars have argued that the rigidly discretized nature of the multi-level framework is questionable or even untenable (Smith et al., 2005; Genus and Coles, 2008; Schatzki, 2011; Markard et al., 2012; Hoffman and Loeber, 2016; Patterson

(11)

et al., 2017) since for example it is difficult to operationalize levels in the system (Genus and Coles, 2008; Ten Pierick and van Mil, 2009; Schatzki, 2011) or because regime ac-tors play an important role in bringing about transitional systemic dynamics too (Raven, 2007; Hoffman, 2013; Hoffman and Loeber, 2016). The second criticism has an origin in the field political science and involves the until recently often neglected role that power plays in the system dynamics (Avelino and Rotmans, 2009; Grin, 2010; Markard et al., 2012; Hoffman and Loeber, 2016). For example, Avelino and Rotmans (2009) developed a power-in-transition framework, drawing upon the work of Arts and Van Tatenhove (2004) by describing the power-interactions between niche and regime actors. Grin (2010) then also takes into account structural power embedded in landscape trends. A power-oriented focus on these politics of transitions has increased the understanding of dynamics in com-plex systems, but has also reinforced the first criticism on the rigid level-demarcation of the MLP (Smith et al., 2005; Markard et al., 2012; Patterson et al., 2017).

When discarding the a-priori assumption that niche innovations are solely responsible for purposefully bringing about transitional change and thus moving beyond the strict dialectic demarcation of niche and regimes levels while taking explicitly in to account the role of power a different picture of dynamics in complex systems emerges. That in turn might lead to a better understanding of the politics and governance of transitions in com-plex systems. How to do that exactly in a coherent way still remains a challenge (Geels, 2011; De Haan and Rotmans, 2018). Connecting power to the systemic dynamics within a complex systems approach beyond the MLP is exactly the main puzzle this thesis aims to address.

1.4 Research Objectives

The theoretical objective of this thesis is to contribute to the understanding of both structure and transitional dynamics of complex systems simultaneously and in coherent fashion by constructing a complex systems framework and explicitly linking power to the systemic dynamics within that framework. To do so this thesis conducts single case study research using the case of biomass co-firing as an empirical referent for iteratively developing theory by taking on an abductive approach, as is further specified in Chapter 2.

The main objective of the thesis can be divided in to three specific goals. First, it aims to contribute to the understanding of systemic conceptualizations and interactions as well as the mechanisms driving complex socio-eco-technical systems into transitional dynamics, in order to contribute to the further development of a complex systems

(12)

ap-proach in transition studies. Second, it aims to elaborate on the conceptualization of power as a factor contributing to the dynamics of these complex systems, in order to contribute to political science. A deepening of the understanding of power dynamics in complex socio-eco-technical systems is crucial to better understand both the politics of complexity and the complexity of politics. Thirdly, this thesis aims to explore the case of biomass co-firing in coal plants in order to provide policy pathways to enable the consti-tution of sustainable practices in the energy sector in the Netherlands. To contribute to that practical goal is of utter importance, given the environmental, social and economic challenges imposed by both the fossil fuel regime and the biomass co-firing practice.

It should also be mentioned that the goal of this thesis is thus not to either directly solve the sustainability controversy of the biomass co-firing practice, to normatively ar-gue against or in favor of particular policies or systemic dynamics or to construct a full model of systemic dynamics that predicts future developments of the case. It merely aims to understand how power relates to the systemic dynamics by taking a complex systems perspective on the case of biomass co-firing in the Netherlands.

1.5 Research Questions and Thesis Outline

Because the extent to which the biomass co-firing practice is ‘sustainable’ is strongly con-tested (possibly suggesting a sub-optimal configuration for the energy system) the rapid emergence of biomass-to-energy as one of the main pillars of the Dutch energy transition policy is surprising. It makes one wonder how this practice came to surface, as well as what mechanisms contribute to enforcing and maintaining this new configuration.

Taking on a power-in-transition approach, this thesis then aims to answer the following central research question: in which ways does power play a role in enabling the practice of biomass co-firing in coal plants to become a dominant tenet of Dutch transition policy? This case, in which the sustainable nature of the practice is contested and where niche-regime dynamics seems rather complex, might therefore help to shed light on the role of power in driving transitional dynamics in socio-eco-technical systems more general. To answer that central puzzle, several sub-questions are postulated.

First, not as a subquestion but as a basis for further developing the central ideas in this thesis an methodological framework as well as an analytical framework will be devel-oped in Chapter 2 and Chapter 3 respectively. Chapter 2 will elaborate on the design of the thesis, the epistemological approach and the process of data gathering and analysis.

(13)

Chapter 3 then will construct the foundation of a complex systems approach in the politi-cal sciences. It will elaborate on functional properties, systemic anatomy, transitions and other features (adaptation, self-organization, non-linearity) commonly associated with complex systems (Holling, 2001; Byrne, 2002; De Haan, 2010; Coenen et al., 2012; Raven et al., 2012; Cairney, 2012). These systemic properties will be evaluated against the background of interpretations of the MLP in transition studies research (e.g. De Haan (2010); Geels and Schot (2010); Grin (2010); De Haan and Rotmans (2011); Schatzki (2011); Markard et al. (2012); Hoffman and Loeber (2016)) and I will draw on litera-ture (De Haan, 2010; De Haan and Rotmans, 2011) in order to further develop a complex systems framework beyond the MLP. I will also formulate hypotheses for the purpose of evaluating the added value of this new framework in understanding empirical reality.

Chapters 4, 5 and 6 this thesis form the bulk of this thesis. They aim to understand the structure of complex systems (Chapter 4), power in complex systems (Chapter 5) and the link between power and dynamics in complex systems (Chapter 6) through the following three sub-questions:

1 | How can we understand the case of biomass co-firing as a complex socio-eco-technical system? Through this question this thesis aims to understand the structure of complex systems. In Chapter 4 we zoom in at both the complexity of the case as well as the case as a complex system. First the extent to which the biomass co-firing practice can be considered a complex system must be investigated. First the case will be projected on to the framework as developed in Chapter 3 to understand the case itself as a complex system. Then to answer this sub-question in more detail we will explore the multi-actor, multi-scalar, and multi-process nature of the case as well as the knowledge uncertainty associated with the practice in order to understand the complexity of the case. Finally, based on this case study we evaluate the added value of the newly developed approach to complex systems based on the formulated hypotheses.

2 | How can we can conceptualize power dynamics in this complex socio-eco-technical system? In Chapter 5 we will draw upon the conceptualization of our case as a complex system for developing a conceptualization of power dynamics within complex socio-eco-technical systems. Drawing upon a power in transition framework (Avelino and Rotmans, 2009; Grin, 2010), broader conceptualizations of power (Lukes, 2005; Giddens, 1986; Arts and Van Tatenhove, 2004; Clegg and Haugaard, 2009) and the role of politics in science and technology studies (Latour, 2004; Brown, 2009) we aim to identify (manifestations) of power in the case. That might also help in contributing to new conceptualizations of

(14)

power in socio-eco-technical systems based on the observable empirical reality of power exertion in the case.

3 | How can we understand the role of power in transitional dynamics in the biomass co-firing practice? In Chapter 6 this thesis aims to understand the mechanisms by which and the extent to which power drives transitional dynamics. The point of departure for answering this question is the conceptualization of power and the identified power dy-namics in the case (empirics) as well as the developed complex systems approach. Based on those conceptualizations, this thesis explicitly aims to connect the role of power in the system to the systemic dynamics through a through empirical analysis of the case. That in turn might also enable us to construct a dynamical narrative the case of biomass co-firing. The resulting conceptualization of the relation between power and systemic dy-namics will be formulated as a proposition, thereby explicitly inviting further evaluation and development the power-in-transitions approach presented in this thesis.

Then in Chapter 7 I will present a brief conclusion of the work in this thesis. For the purpose of reflection and outlook Chapter 8 will then first consider some of the policy recommendations for governing this specific case as well as draw on this case to make some more general observations on the governance of complex socio-technical systems. Finally, after that prescriptive contribution the discussion aims to reflect on the research conducted in this thesis through formulating an outlook on directions for future research.

(15)

2 Methodological Approach

This chapter elaborates on the methodological framework upon which this thesis draws. First, the research objectives and the methodological considerations regarding the nature of the case study research are set forth in section 2.1. Secondly, the methodological choices regarding abduction and process tracing are explained in sections 2.2 and 2.3. Then in section 2.4, methods for data gathering and data analysis are presented. The chapter will be concluded by paragraphs on the ethical considerations (section 2.5) and challenges regarding the research methodology presented in this thesis (section 2.6).

2.1 Case Study Research

The proposed research design is qualitative in nature. This thesis aims to answer a research question that is essentially descriptive in nature. It aims to shed light on how power plays a role in transitional dynamics in complex systems. However, this thesis also serves a practical goal: to contribute to the construction of future policy pathways by which these dynamics can be governed. This practical nature where research is done on what ‘is’ and what ‘ought’, is a common feature of transition studies (Markard et al., 2012; Avelino and Grin, 2017). To do so, a single case study will be used as an empirical referent to further develop theoretical understanding of complex systems. That case study still will allow for the observation of causal mechanisms and deepen the understanding of the dynamics in the case, as well as provide opportunities to contribute to theory (Flyvbjerg, 2011), in our case in the field of transition studies. Case studies are useful for analyzing causal mechanisms in phenomena that are entangled with their context (Baxter and Jack, 2008). They furthermore allow for the deepening of understanding of system and the linking of causes and consequences (Flyvbjerg, 2011). This research aims to conduct a single-case study, whereby the entire practice of biomass co-firing in Dutch coal plants is considered to be the case. As from 2017, there are only 5 coal plants left in the Netherlands (Min EZ, 2017c) and the co-firing subsidies are given to the three energy companies RWE, Uniper and Engie (Woutersen et al., 2017). That means that even though the case is important and these coal plants are responsible for a significant amount of the Dutch energy production, the case in terms of actors and scope is comprehensible

(16)

enough for the purpose of this thesis.

2.2 Abductive Approach

This thesis takes on a critical realist perspective. Instead of the more common deductive or inductive approaches, this thesis aims to use an abductive approach. In a deductive approach one starts with a rule or thesis, after which one tries to test that theory on a case (leGall and Landley, 2015), whilst by taking an inductive approach the starting point is a case, after which one aims to construct theory based on that case. In the abductive approach however, as elaborately described by for example leGall and Landley (2015), researchers are not so much testing or constructing theory, but rather they aim to further elaborate on pre-existing theoretical notions, therefore contributing to development of theory in an iterative exercise, which is a “continuous process based on the interplay between theories and data.” (leGall and Landley, 2015, 38). Dubois and Gadde (2002, 559) argue that an abductive approach should not be seen as a simple mix between induction and deduction. It serves a distinctly different purpose: to discover new variables and relationships. Abduction is an appropriate approach for the case of biomass co-firing in coal plants since there is a rather developed theoretical framework on transition studies and political science, but in this case we are confronted with a surprising empirical finding (the rapid institutionalization of a controversial practice, which at first sight also does not seem to be an technical innovative niche). The expectation thus is that using this case as an empirical referent will deepen the understanding of the case, as well as contribute to theory development. In that case an abductive approach is appropriate (leGall and Landley, 2015). This case study then is what Venesson (2008) would describe as a disciplined configurative case study. That case study type “uses theoretical frameworks to provide an explanation of particular cases, which can lead as well to an evaluation and refinement of theories” (Venesson, 2008, 227). The use of the iterative abductive approach can thus be justified in the case of a single case study.

2.3 Systems, Processes and Causality

For explaining history-dependent social phenomena scholars argue one can take on roughly three kinds of theoretical approaches: variance theories, process theories and systems the-ories (Morris, 2005; Burton-Jones et al., 2011). Variance theory focuses on stochastical causality and explains outcomes (dependent variable) as a direct result of the influence of causal factors (independent variables) (Abbott, 2001). Process theory "focuses on narratives and aims to find typical patterns" (Abbott, 2001, 164) through analyzing se-quences of events (Burton-Jones et al., 2011, 6). Finally systems theory focuses on

(17)

interactions and (reciprocal) relationships between parts of the system and their emer-gent properties (Burton-Jones et al., 2011, 6). These rather idealtypical approaches are often combined in scholarly reality to enhance the completeness and ecological validity of research (Burton-Jones et al., 2011, 28). Scholars of transition studies in their work gen-erally combine elements of process theories and system theories (Geels and Schot, 2010; Rotmans and Loorbach, 2010), some focusing more on processes and chains of events while others focus more on systemic interactions. Following that work, this thesis also explicitly combines systemic and process oriented approaches. It focuses on processes and narratives in order to understand complex interactions in a changing contexts (Geels and Schot, 2010, 99) but also takes on a systemic approach in aiming to understand causal mechanisms through which systemic phenomena occur (Burton-Jones et al., 2011, 15-17).

For understanding the narratives of the case while allowing for the existence of sys-temic relations and complexity, this thesis follows (Geels and Schot, 2010, 99) in doing process tracing as a method. Process tracing is considered a robust method for under-standing causal mechanisms in within-case studies, as well as for small-N comparative case-studies (Kay and Baker, 2015, 2). It is a method suitable for this case, since process tracing allows for analysis of multi-stakeholder and multi-level systems in a complex pub-lic popub-licy context while it at the same time allows for the consideration of the temporal dimension (Cairney, 2013; Kay and Baker, 2015). It is thus often used to identify and understand the emergence of particular (single or multiple) policy pathways, through understanding of the causal mechanisms driving that path dependency (Hall, 2006; Col-lier, 2011). It does so by focusing subsequently on (1) detailed narratives (2) hypotheses on theoretical mechanisms to explain that narrative (3) analytical explanations linking patterns of events and finally (4) the formulation of generalizable explanations about the phenomena (George and Bennett, 2005, 210-212). As Venesson (2008, 235) puts it: “the goal of process tracing is ultimately to provide a narrative explanation of a causal path that leads to a specific outcome”. In this thesis taking on a process tracing approach does then not mean that we aim for the observation of immediate variable-based causation, tunable independent variables or deterministic pathways (Morris, 2005; Geels and Schot, 2010; Rotmans and Loorbach, 2010; Burton-Jones et al., 2011).

Literature generally identifies three different variants of process tracing (George and Bennett, 2005; Hall, 2006; Beach and Pedersen, 2013): one case-centered variant and two theory-centered variants. The case-centered process tracing variant aims to con-struct a detailed historical narrative to understand how a particular systemic configura-tion emerged. It aims to offer an explanaconfigura-tion on “what mechanistic explanaconfigura-tion accounts

(18)

for the outcome” (Kay and Baker, 2015, 7). It also aims to provide the minimally suf-ficient explanation for the outcome in the single case, but has no solid ambitions for generalizability of the findings to other cases (Beach and Pedersen, 2013, 15-20). In this thesis, the aim is to contribute to development of transition studies theory. Some degree of generalizability of the findings in our case, to more general causal mechanisms in complex socio-technical systemic transitions is therefore preferable. We therefore dis-card this case-centric analytical approach, and take on a theory-centric one. The first theory-centered process tracing variant, theory-testing, is rather deductive of nature and aims to confirm or falsify hypotheses that are formulated based on theory. The second one, theory-building, has an inductive nature and aims to identify causal mechanisms and construct theory by observing a particular case. As Lin (1998, 166-169) argues, process tracing also allows for an unique opportunity to bridge the gap between positivist and interpretivist approaches. According to Venesson (2008, 232), process tracing in that way is able to contribute to understanding about “both the causal ‘what’ and the causal ‘how’ ”. In this light, an abductive variant of process tracing, iterating between theory building and theory testing whilst aiming to contribute to theory development, seems ap-propriate. These theory-based process tracing variants thus generally involve either the formulation of testable hypotheses beforehand or hypotheses that are formulated based on the causal mechanisms that are being identified in the case (Kay and Baker, 2015) as a result of the research. In this thesis, hypotheses will be formulated based on the analytical framework (Chapter 3). They will be tested on the case (Chapter 4) and will allow for sharpening of the analytical perspective of the framework for understanding the causal ‘what’. That sharpened analytical framework (of which Venesson (2008, 237) might argue it is set of "broad guidelines rather than clearly formulated hypotheses") then will be deployed troughout the rest of the thesis to identify and understand the dynamics of our system. Subsequently, based on the findings of Chapters 5 and 6 a conceptualiza-tion of the relaconceptualiza-tion between the variables ‘power’ and ‘system dynamics’ will be presented that in future research can be operationalized as a hypothesis which can be tested. That will then account as an explanation for understanding the causal ‘how’ of the case of biomass co-firing in the Netherlands. Through formulation of hypotheses both for testing the framework and as a resulting proposition of empirical work the abductive nature of the research is illustrated.

2.4 Data Gathering and Analysis

For the purpose of this thesis, two kinds of data sources will be used. First of all, documental analysis will be performed. Secondly, semi-structured interviews will be

(19)

con-ducted. Both documental analysis and the use of open-ended semi-structured interviews are generally considered reliable and effective methods within the analytical framework of process tracing research (Venesson, 2008; Kay and Baker, 2015). Plenty of documen-tal information is available in terms of scientific research (Ringel, 2006; Stupak et al., 2007; Searchinger et al., 2009; Henry, 2010; Cherubini et al., 2011; Verkerk et al., 2011; Bentsen and Felby, 2012; Schulze et al., 2012; Jonker et al., 2014; Repo et al., 2015; Ter-Mikaelian et al., 2015; Nabuurs et al., 2017), policy research (Ecofys, 2008, 2015, 2017; PBL, 2013; CE Delft, 2016; Chatham House, 2017; IEA Bioenergy, 2017), policy reports (SER, 2013; CBS, 2015, 2017a; NEV, 2017; Min EZ, 2008, 2010, 2017c,b, 2018b), transcripts of meetings of the Parlimentary Committee on Economic Affairs (Tweede Kamer, 2017), and investigative journalism (Woutersen et al., 2017; Straver, 2018). For the purpose of gaining inside information into the decision making processes involving biomass co-firing, several expert-interviews are be conducted. In order to allow for tri-angulation of viewpoints this group of experts make up a diverse stakeholder field, with actors from industry, government, politics and environmental organizations. A list of interviewees is presented in Appendix I.

These interviews are semi-structured interviews. That means that the the questions are prepared in advance, but that the order of the questions and the nature of the formu-lation, as well as follow-up questions, is determined based on whether the researcher con-siders it appropriate in both context and timing (Van Teijlingen, 2014). Semi-structured interviews are often used in policy research, especially when the goal is to deepen the understanding of a particular case or phenomenon (Harrell and Bradley, 2009), as is the case in this thesis. They are also considered appropriate for the exploration of for example beliefs, values and motives (Van Teijlingen, 2014). That could be beneficial when aiming for the uncovering of sensitive topics related to power dynamics in the practice. When it comes to data analysis of the interview transcripts, a multi-method combination of con-tent analysis and discourse analysis is used. This combination, as illustrated by Bennett (2015), aims to bridge the gap between interpretative and positivist approaches of data analysis in the social sciences. It might therefore suit the abductive approach of this thesis. For the purpose of the interview transcription, software-assisted coding (through the program Atlas.ti 7) of text might help to identify particular relevant words or text fragments, but also to order the claims of interviewees in themes. Content analysis, as a rather formal and rule-governed procedure might thus help to advance the inter-coding reliability and the systematic nature of the analysis (Mayring, 2004), which might for example help to avoid inaccuracies by the researcher or even confirmation bias when an-alyzing semi-structured interviews (Schmidt, 2004). Discourse analysis then might add

(20)

value beyond the analysis of socio-linguistic and psycho-linguistic processes and is able to unravel power relations that are constituted in linguistic practices as well as the pat-terns that reproduce those relations (Parker, 2004). Discourse analysis thus allows for the study the involved narratives, storylines, meaning, epistemic figures and policy vo-cabularies (Hajer and Wagenaar, 2003). According to Bennett (2015) a combination of software-assisted content analysis and discourse analysis might be a powerful tool for an-alyzing complex social phenomena. An overview of the coding-labels used for the analysis of the interviews is presented in Appendix II.

2.5 Ethical Considerations

When dealing with research on power relations in complex systems and when interacting with people in order to gain access to data, ethical issues are an important issue to address properly. First of all, the author declares no competing (financial) interests in writing this thesis. For the interviews, only relevant actors with whom the author has no friendship or familial relation with are selected in order to avoid conflict of interest. With regard to the collection of interview data, the author has always asked for permission before any recording took place. Interviewees are given full openness on the procedure for data analysis, data collection and the process of writing the thesis. All interviewees are also granted anonimity. In that way, interviewees hopefully feel free to speak as they like, not limited by complex power-relations in their scientific field or organization. That in turn might help the researcher to gain more understanding of the case, benefitting not only the ethical but also the scientific quality of the thesis.

2.6 Identifying Methodological Challenges

There are several methodological challenges that need to be addressed. The first chal-lenge has to do with choosing the appropriate approach. This thesis adopts an abductive approach for reasons explained in section 2.2, but at the same time I am aware that the abductive approach is one that is less developed in methodological literature when compared to for example the widely used deductive or inductive approaches in case study research. When considering single case studies, one can choose a case study design based on the case typology. Flyvbjerg (2011) for example considers several different types of case typologies, such as deviant cases, critical cases or paradigmatic cases. However, the tricky challenge here is to choose the ‘right’ methodology for the ‘preferred typology’, when one can hardly fully know how to characterize the case in advance when the re-search has not yet been completed. Thirdly, several challenges are related to the process tracing analytical approach. The challenges here is to formulate hypotheses based on

(21)

ei-ther existing theory, or expectations based on case-related information on possible causal inference. Again, the abductive approach here is justified, but requires the researcher to skillfully ‘hop’ between theory and case. When considering data gathering within process tracing, triangulation is an important aspect in order to more fully grasp the narrative and the historical path-dependency of the systemic development. However, selecting documents and interviewees remains an analytical choice that is severely constrained by practical matters (time, access, availability). A final challenge then will be the integra-tion of this abductive approach as well as the findings and outcomes of the thesis into a coherent power-in-transition narrative and framework. The key challenge there will be to integrate methods, theories and data-analytical tools from different disciplinary backgrounds, while keeping the focus on power in a public policy context.

(22)

3 Analytical Approach to Complex Systems

This thesis aims to develop new insights on the role of power in transitions in complex systems. That requires not only understanding of power and its role in transitional dynamics, but also a coherent analytical framework that serves as the foundation for grasping complexity and systems thinking. It therefore first focuses on constructing a solid complex system framework in section 3.1. Then I will debate the multi-level perspective (MLP) and argue how to construct a framework that takes in to account the criticisms on the MLPs capacity to understand systemic structure while retaining some of the key insights from that approach in section 3.2. In sections 3.3 and 3.4 I will argue how a complex systems framework might help to understand the dynamics of systemic transitions. Finally in section 3.5 I will formulate hypotheses to empirically test the added value of the developed complex systems approach.

3.1 Foundations: Building Blocks of Complex Systems

The complexity view is a relatively new scientific paradigm that has entered the realm of social sciences in the last few decades (Byrne, 2002; Mitchell, 2009). Complex systems are considered a special class of social systems (Rotmans and Loorbach, 2010). There are many different ways to characterize and classify complex societal systems. De Haan (2010) distinguishes systems for example based on function (healthcare, education), ge-ography (regions, river-delta) or regulatory level (spatial planning, law). These different classifications of course overlap in the empirical world, just as that one system can be a subset of another social system. The key argument here is that demarcating what consti-tutes the ‘system’ is a matter of choice and depending on that demarcation of the system, the systemic analysis takes a particular perspective. And for exactly that systemic anal-ysis, the complexity paradigm offers a useful framework. As Cairney (2012, p. 1) puts it: “complexity theory suggests that we shift our analysis from individual parts of a system to the system as a whole”. That is necessary since a key feature of complex systems is that the systemic behavior can not simply be understood as a linear superposition of the behavior of the parts that constitute the system (Cairney, 2012). This non-linearity also means that small changes in one component do not linearly translate to the same

(23)

small changes in the systemic behavior (Byrne, 2002). This is related to the long-range interaction-networks that can exist between components in the systems. Due to those interactions systemic feedback-loops can emerge in the system, allowing small changes to either accelerate systemic change (positive feedback-loop) or dampen systemic change (negative feedback-loop). In other words: small changes can have large effects, large changes can have small effects and it is very difficult to predict such dynamics (Cairney, 2012).

The extent to which systemic structure and dynamics is affected by internal or exter-nal pressures on the system as a whole or its constitutive components is closely related to the complexity-related concepts of resilience and adaptation (Holling, 2001; Foxon et al., 2009; De Haan, 2010; Cairney, 2012). Resilience is a concept with an origin in ecology, that refers to the capacity of a system to retain its structural and dynamical configuration under the influence of perturbations (shocks, external and internal pres-sures). For an ecological system, this can for example be the capacity of a coral reef to retain its characteristics under the influence of increasing carbon-dioxide concentra-tions in the oceanic waters or the capacity of a forest to restore itself after a wildfire. When considering socio-technical systems, resilience can be understood for example as the capacity of an energy system to retain its reliance on oil, when confronted with a sudden depletion of resources. Systems exhibit a degree of resilience because the local interactions between the systemic components have the ability to adapt to perturbations by self-organizing their configuration in a new way that is compatible with the perturba-tion or are able to resist the perturbaperturba-tion, allowing the system to retain its funcperturba-tional or structural configuration (Holling, 2001; Foxon et al., 2009; De Haan, 2010). Since societal and socio-technical systems are key examples of systems where a multitude of actors and structures (material, discursive, social) is involved in self-organizing processes in com-plex interaction-networks, the comcom-plexity paradigm has recently gained ample attention in the field of policy studies (Cairney, 2012). Not only because policy-making systems themselves are complex systems, but also because they increasing deal with systems and problems that are fundamentally complex - or ’wicked’ - of nature (Rittel and Webber, 1973).

When internal or external perturbations manage to disturb the stable configuration that a complex system finds itself in, the system can undergo a transition. In physical systems, phase transitions are processes of non-linear dynamics whereby the system fundamentally changes it structural configuration from one (meta-)stable state to another (meta-)stable state. Following that line of reasoning, scholars of transitions in societal systems have

(24)

thus defined socio-technical transitions as non-linear processes of structural change (Geels and Schot, 2007; Rotmans and Loorbach, 2010), as a “profound change in various or all aspects of a systems functioning” (De Haan, 2010, p. 59) and processes of change that lead to “far-reaching changes in the system along different dimensions: technological, mate-rial, organizational, institutional, political, economic, and socio-cultural” (Markard et al., 2012, p. 956).

Before elaborating further on the definition and nature of transitional dynamics in sec-tion 3.3, I first proceed to define a coherent language for discussing systems, subsystems, systemic components etcetera. For the purpose of that common language, I follow the etymology of De Haan (2010) in his work on developing an integrated formal framework of transition theory. De Haan and Rotmans (2011, p. 92) define a societal system as “a part of society that can be attributed a functioning and functioning is the way a societal system meets a societal need”. In this definition a system is thus inherently functionalist in nature (De Haan, 2010). The functions of complex adaptive systems such as societal systems can then best be seen as the product of the emergent self-organizing processes evolving from the interaction-networks of the components in the system. This function-alist systemic approach, although widely embraced by scholars in the studies of complex adaptive systems (Foxon et al., 2009), has also drawn criticism for overemphasizing struc-tural and generalized functional system properties above contextual factors (Smith et al., 2005). Perspectives on functionalism of systems itself are worthy of scholarly debate, but for the purpose of this thesis I adopt the functionalist approach. In the systemic frame-work of De Haan (2010) the complex system is made up of a collection of sub-systems, each contributing to the overall functioning of the system in their own way. These sub-systems are defined to be constellations. They each have their own functioning, interact with other constellations and might find themselves to be constituted of several distinct constellations as well. I explicitly point out that these constellations “can be small . . . they can [be] comparable in scale to the entire system. Or anything in between, continuously scalable” (De Haan, 2010, p. 62). As I will extensively argue in section 3.2 and chapter 4 this is an important characteristic, since constellations thus do not require specific char-acteristic traits in terms of spatial size, geographical distribution, temporal dimension, institutionalization or degree of structuration in order for them to be considered a rele-vant constellation. In fact, what exactly demarcates constellations (and those constituting constellations) from systems of course depends on the focal point and the level of analysis.

Then what substance constitutes these constellations? According to Rotmans and Loor-bach (2009) all sub-systems contain attributes of structures (institutional and

(25)

phys-ical), cultures (prevailing perspectives and discourses) and practices (rules, routines, habits). De Haan (2010) considers a dualism containing structures (how it works) and cultures (how it is perceived ). He considers practices to be an excess category since prac-tices refer to the individual level (as a microstate unit of analysis) and the functioning of complex systems due to their very nature emerges from a configuration of micro-states: local interactions of individual practices. Regardless of the precize categorization, I ar-gue that the substance of which these constellations are constructed are microstates, their interacting networks and functional superpositions. Those microstates I define as compo-nents, that can be classified as (1) physical or non-physical (2) human or non-human and (3) structural, discursive, psychological, social, ecological or material of nature. This neu-tral definition, not favoring human actors above other actants in the system, I explicitly draw from the work in Actor Network Theory (ANT) as developed by for example Latour (1987, 2004); Murdoch (1997). As an example of how these components, constellations and systems relate to one another, Table 3.1 elaborates on their nature by virtue of the example of the mobility sector as a system.

Typology Definition Examples System A part of society that can be

attributed a functioning and functioning is the way a so-cietal system meets a soso-cietal need.

Mobility sector

Constellation A societal sub-system, con-tributing to the overall func-tioning of the system in one or multiple ways.

Public transport | Electrical car industry | Aviation sector

Component Societal microstates, their interacting networks and functional superpositions that constitute constellations.

Table 3.1: Example of systemic anatomy for the mobility sector as a system

3.2 Beyond a Multi-Level Perspective on Complex Systems

Since this thesis aims to understand transitional dynamics in socio-technical systems, there is no possibility of evading a discussion of the MLP for three reasons. First of

(26)

all, the MLP as first developed most notably by Rip and Kemp (1998) as well as Geels (2002) has by far been the most influential analytical tool for understanding transitional dynamics in complex systems (Grin et al., 2010; Markard et al., 2012). Second, in recent years many criticisms have emerged that have pointed out that the original MLP needs some refining inter alia in terms of conceptualizations of the levels and the interactions between them (Smith et al., 2005; Genus and Coles, 2008; Shove and Walker, 2010; Geels, 2011; Schatzki, 2011; Raven et al., 2012; Hoffman and Loeber, 2016). Thirdly because as I will argue in subsection 3.2.1, the systemic complexity framework of section 3.1 allows for acceding some criticisms of the MLP while at the same time retaining the valuable analytical contributions of multi-level conceptualizations in complex systems. This refining and reconceptualization might help to further understand our case of the biomass co-firing practice in coal plants.

3.2.1 Multi-Level Perspective

The MLP was first developed by Rip and Kemp (1998) in an attempt to combine el-ements from Science & Technology Studies (STS) and evolutionary economics in order to provide an analytical tool for understanding transitions in socio-technical systems. The perspective has since then been developed extensively by the work of Geels (2002); Rotmans and Loorbach (2009); Geels and Schot (2010); Grin (2010) and others. The perspective assumes sociotechnical systems to have three different levels: (1) micro -technological niches (2) meso - socio-technical regimes and (3) macro - socio-technical landscape. These levels find themselves in a nested hierarchy (Geels and Schot, 2010, 18), in which niches are embedded in regimes and regimes in their turn are embedded in the landscape. One should consider these levels to span functional or structuration space, rather than levels extending across spatial, geographical or governmental levels. The three levels are thus seen to represent functional scales (Rotmans and Loorbach, 2010), or degrees in stability (Geels and Schot, 2010; Geels, 2011). A schematic overview of the nested hierarchy representation in the MLP is presented in Figure 3.1.

On the micro-level, so called niches are considered to be the drivers of radical systemic change (Geels, 2002). Niches are novel configurations with a low degree of structuration, often representing new (technological) innovations. The term niche finds its origin in evolutionary theories (via ecology and evolutionairy economics) where it is considered a rather safe space in which a new configuration has the oppurtunity to further develop without being subject to the same selection mechanisms (e.g. market prices, competition) as for example regime-constellations. These protective spaces where the novelty resides niches actively constructed (Geels and Schot, 2010, 22) and require forms of (public)

(27)

Figure 3.1: Schematic overview the nested hierarchy in the MLP. Taken from (Geels, 2002, 1261).

management. Transition Management (TM) and Strategic Niche Management (SNM) are such steering philosophies that are developed by transition studies scholars in order to help create protected niches allowing novel desired socio-technical innovations to first gain strength and finally to radically transform the system through a transition (Geels and Schot, 2007; Raven, 2012). That strengthening of the niche happens through at least three different internal niche-processes: (1) construction of social networks that carry and support the innovation (2) learning processes to improve the functioning of the socio-technical niche-constellation and (3) articulating visions and expectations about the processes of learning and future transitional pathways (Raven, 2012, 135-144).

Socio-technical regimes play a key role in the (un)stability of a socio-technical system and their tendency to be sensitive to niche developments and transitional dynamics. The term regime refers to the dominant material, discursive and social structures and practices in the system. It builds on the work of Rip and Kemp (1998, 340) who pro-posed the following popular definition for a technological regime:"A technological regime is the rule-set or grammar embedded in a complex of engineering practices, production process technologies, product characteristics, skills and precudures, ways of handling rele-vant artefacts and persons, ways of defining problems; all of thm embedded in institutions and infrastructures". Socio-technical regimes are rather similar, but generally involve a greater multitude of actors, belief systems and social structures. Following the work

(28)

of Geels (2002); Geels and Schot (2007); Rotmans and Loorbach (2010) I take on a rather open definition where I consider a socio-technical regime to be "systems of domi-nant structures, culture and practices that are shared by groups of actors" (Rotmans and Loorbach, 2010, 132). Regime-networks consisting of actors as well as of social, discursive and material structures, cultures and practices are generally considered to enhance the stability of systems. This systemic lock-in emerges from the interactions and alignment of regime-components and enhances the systemic resistance to transitional change (Rip and Kemp, 1998; Geels, 2002; Van der Vleuten and Högselius, 2012). However, recent empirical and theoretical work has shown that the degree of lock-in is heterogeneously distributed within and between (components of) regimes (Van der Vleuten and Högselius, 2012). Furthermore it has been pointed out that regime-networks can play crucial roles in accelerating transitional change (Smith et al., 2005; Geels and Schot, 2007; Avelino and Rotmans, 2009; Geels and Schot, 2010; Verbong and Loorbach, 2012). Thus while regimes reflect a rather high degree of functional structuration, it is a matter of empir-ical investigation whether (parts of) regime-constellations contribute to accelerating or inhibiting transitional systemic dynamics.

In the nested hierarchy paradigm, both niches and regime-constellations are embedded in an exogenous macro-structure: the landscape. The landscape accounts for macrodimen-sional physical, societal and cultural processes that have a high degree of structuration and that are beyond the direct influence of actors in the niche- and regime-levels. Land-scape trends however can have crucial impact upon the dynamics of regime-constellations and niche-developments (Grin, 2010; Geels and Schot, 2010; Verbong and Loorbach, 2012). In their conceptualization of landscape factors, Van Driel and Schot (2005) dis-tinguish between three different landscape manifestations: (1) rapid external shocks such as an earthquake or a financial crisis (2) constant (or slowly changing) structural trends, such as the climate or a geographical landscape and (3) long-term changes such as glob-alization. These landscape trends interact with dynamics on the other levels in the MLP, thus creating conditions that either enforce or decrease the structural stability of regime-constellations as well as conditions that either enhance or decrease the developmental processes of niches (Geels, 2002; Geels and Schot, 2010; Grin, 2010; De Haan, 2010; Verbong and Loorbach, 2012).

3.2.2 Debating the Multi-Level Perspective

Several criticisms, refinements and reconceptualizations have been suggested for the ‘origi-nal’ multi-level perspective as developed by among others Rip and Kemp (1998) and Geels and Schot (2010). The scholarly debate and the resulting incremental refinement of the

(29)

multi-level perspective are worthy of a thesis themselves. For the purpose of this thesis however I will first proceed by following Ten Pierick and van Mil (2009) by broadly defin-ing three categories of MLP-sub-debates and subsequently discuss some of the critical issues raised on the (1) conceptualization (2) operationalization and (3) utilization of the MLP. I will conclude this section by adding some critical remarks on what I call the (4) expansion of the MLP. Then in subsection 3.2.3, I will address how I will take these criticisms into account in finalizing the construction of the analytical framework that will allow for the analysis of power and transitional dynamics in complex systems.

i) Conceptualization | When it comes to conceptualization of the levels in the MLP, there are at least two different criticisms to take into account. The first one relates to the nested hierarchy implied by the introducting discretized levels (Shove and Walker, 2010). Advocates of practice-oriented approaches to complex systems argue that an a-priori level discretization in a nested hierarchy is troublesome (Røpke, 2009; Shove and Walker, 2010; Hoffman and Loeber, 2016). From an ontological point of view, it would therefore be more consistent and logically sensible to consider not discretized levels but a conti-nous scale of structuration (Schatzki, 2011): a flatter ontology (Shove and Walker, 2010). According to Schatzki (2011) social macro-structures consist of the exact same substance as meso- and micro-structures: bundles of practices and material arrangements. That notion in fact still allows for large social phenomena (e.g. socio-technical regimes) to ex-hibit fundamentally different behavior than small ones (e.g. niches). The emergence of so called structural length scales, where macro-scale behavior is fundamentally different from micro-scale behavior even though they consist of constellations of the exact same sub-stance are found to be a characteristic feature of complex physical systems too (Berthier et al., 2005). However, holding on to strict level-demarcation and thereby especially link-ing meso-phenomena to stability and micro-phenomena to dynamics such as in the bulk of MLP-work, is ontologically suspect (Schatzki, 2011, 15). Geels (2011, 38) acknowledges the value of the practice-approach as an alternative to the MLP, but notes that "because the ontology is (even more) complex than the MLP, there may be trade-offs with regard to accuracy, generalization, and empirical operationalization." In addition to the ontological objections to the nested hierarchy as dominant paradigm, empirical work has shown that holding on to the trialectic poses some explanatory problems. Scholars have argued that it leads to oversimplication of actor-roles as either niche- or regime-actors (De Haan and Rotmans, 2018), that several sub-systems are not accurately depicted as either niche or regime (leading to some scholars adopting the concept of niche-regime (Haxeltine et al., 2008; Avelino and Rotmans, 2009; Rotmans and Loorbach, 2010) and that there is hetero-geneity of e.g. actors, structuration, stability within constellations or in fact a multitude

(30)

of relevant regimes (Raven, 2007; Verbong and Loorbach, 2012; Jørgensen, 2012).

The second criticism relates to this level-debate and involves the somewhat residual nature of the landscape-level and the epistemological and ontological complications of that conceptualization (Røpke, 2009; Shove and Walker, 2010; Jørgensen, 2012). In influential MLP conceptualizations by e.g. Geels (2002); Geels and Schot (2007); Rot-mans and Loorbach (2010) the landscape is consired exogenous and its developments "cannot be influenced by niche and regime actors in the particular domain that is the object of study" (Geels and Schot, 2010, 24). This entails a rather asymmetrical re-lation between the landscape and the other levels and while that might perfectly well be the result of an empirical analysis, some scholars argue it is unappropriate as an a-priori assumption exactly because macro-phenomena consist of the same structure as micro-phenomena (Schatzki, 2011). Some scholars such as Shove and Walker (2010) therefore argue that an exogenous landscape environment does not exist because those macro-structures are merely constituted of constellations of micro-structures. Others, like Berkhout et al. (2009); Jørgensen (2012) argue that we can still consider landscapes as fundamentally different from regimes but state that a more symmetrical approach is appropriate and that such an approach can be developed by studying how regimes and niches in their turn influence and shape landscape trends.

ii) Operationalization | If one manages to conceptualize the systemic levels of analy-sis in a satisfactory fashion, there is still a lot of effort needed to operationalize the units of analysis properly. According to several critics (Smith et al., 2005; Genus and Coles, 2008; Shove and Walker, 2010) this has been a critical issue in studies that have adopted the MLP as an analytical tool. Genus and Coles (2008) argue that most (case study) literature does not explicitly and systematically apply the MLP to the case because for example levels are not clearly demarcated, described or accounted for, partly because it is difficult to identify a group of similar ‘niches’ and a stable ‘regime’ in a diverse and com-plex case (De Haan and Rotmans, 2018). Smith et al. (2005, 1493) note that the notions ‘niche’ and ‘regime’ also depend on the level of analysis. What in a particular study can be considered a regime, can be considered a niche in a different one. That stresses the need for careful elaboration on the operationalization. What is furthermore difficult to operationalize, is the exact trajectory of a transition including its start point, end point and relevant systemic components. This transition-demarcating issue gets complicated further because in empirical systems where timescales can stretch up to several decades it is often difficult to disentangle radical change, incremental change, long-term change etcetera (Genus and Coles, 2008).

(31)

iii) Utilization | Finally, when it comes to the topic of utilization, there are roughly two major issues: (1) how to do systemic analyses of transitions? and (2) how to influ-ence systemic transitions? Often coupled in the field of transition studies, I will treat these descriptive and prescriptive dimensions of transitions studies seperately as far as possible. As a matter of fact however, that is rather difficult. Genus and Coles (2008) have critized previous work on the MLP for adopting a too global model whereby there is no place for the uncertainty and complexity of local practices. At the same time when researchers are distancing themselves from the systemic dynamics by adopting such a global model, they should also be aware of the multitude of rather subjective choices on how to do the systemic analysis that lie with the researchers (Genus and Coles, 2008; Ten Pierick and van Mil, 2009). Researchers choose the operationalization of analyti-cal levels, systems, cases, components, relevant actors, transition pathways, transition timeframe etcetera. Especially in cases where researchers aim to address the prescriptive question as well, complex power dynamics involving researcher and the object of study might occur Shdaimah and Stahl (2012).

iv) Expansion of the MLP | Many scholars have attempted to unify the MLP with their fields of study, or have aimed to expand - or broaden the scope of the MLP (Geels, 2017). Political scientists have argued that the MLP should be able to account for pol-itics and power dynamics (Smith et al., 2005; Avelino and Rotmans, 2009; Grin, 2010; Hoffman and Loeber, 2016; Avelino and Grin, 2017), while those with an interest in (so-cial) geography have argued that the MLP should be expanded to take into account the spatial and geographical dimensions (Coenen et al., 2012; Raven et al., 2012; de Boer and Zuidema, 2015; Chandrashekeran, 2016). Scholars of international relations have opted for a transnational view on the multi-level perspective that incorporates insights from state-of-the-art transnational governance studies (Van der Vleuten and Högselius, 2012), along with scholars of political economy who aim for the incorporation of the role of elites and power-coalitions (Osunmuyiwa et al., 2018). In a review article Geels (2017, 227) mentions dozens of other studies that sought to expand the MLP by introducing views on for example discourses, narratives, firms, organizations, institutions, consumers and households.

All these insights have of course been very valuable in increasing the understanding of complex systems and their transitional dynamics. However, when reviewing some of these contributions I can not help to notice that scholars (though mostly rightfully so) attribute particular properties of interest (e.g. a geographical scale or a particular

(32)

manifestation of power) to a level in the MLP, since they explicitly or intrinsicly aim to broaden the scope of the MLP. The risk at hand here is that when taking the sum of all the contributions of the authors, oversimplified correlations emerge. For example as a result of overlapping Grin (2010) and Raven et al. (2012), one could start to think that the typical regime timescale is roughly a decade, regimes manifests themselves on a national scale with networks that exhibit high degrees of proximity and regimes generally contain relational power dynamics. The obvious risk I am pointing to here is that the tendency of generalizing specific findings (which can perfectly well be the justified result of a thorough case study) when always starting with the MLP as a point of departure has a risk of evolving into an unjustified projection of all kinds of properties on these levels in the MLP. In that way these properties end up being - maybe unintentionally or unjustifi-ably - correlated in idealtypical MLP-levels. This is problematic because (1) idealtypical levels only exist in theory (2) the criticisms listed in i)-iii) have shown that subtle in-teractions, relations and case-specific complexities are very important for understanding systemic dynamics and (3) as I pointed out using these strictly demarcated levels is not overwhelmingly justified based on both theoretical and empirical considerations. In the next section I therefore aim to find a way to decouple (decorrelate if one whishes) these scales of complexity (actors, geographical scale, spatial scale, power manifestations, struc-turation - or MLP - scale, functional scales) by taking on the complex systems approach as postulated by De Haan (2010) as point of departure for analyzing systemic dynamics instead of departing from the MLP. This decoupling might be able to free us from the risk of "correlation due to projection" as previously described, while at the same time retaining the very valuable contributions from previous work on structuration, stability, dynamics, power, scale, geography, governance etcetera in the field of transition studies.

3.2.3 Towards a New Framework

Instead of taking the MLP as the paradigmatic point of departure, I argue that it is more fruitful to start with the foundations of complex systems as postulated most notably by (De Haan, 2010) and further developed in section 3.1. In taking on this complex-ity approach, one would identify all relevant constellations as such. There is no need to a-priori distinguish between single or multiple niches, niche-regimes and regimes in the system. According to De Haan (2010, 62) this approach "eliminates the need for a rigid level division such as employed in various emanations of the multi-level perspective". There is therefore no need to a-priori project all kinds of systemic properties that gen-erally belong to particular levels in the MLP on to the identified constellations. That of course does not mean that all constellations should be treated as similarly contributing to systemic dynamics or systemic functioning. Of course, the degree of structuration

Power dynamics in transition : exploring the governance of complex socio-eco-technical systems : the case of biomass co-firing in coal plants in The Netherlands

Master Thesis Public Policy and Governance