Stakeholder Involvement in Risk Governance: Studying the effect of Risk Governance on Stakeholder Involvement in the European Commission

(1)

Stakeholder Involvement in Risk Governance:

Studying the effect of Risk Governance on Stakeholder Involvement in the

European Commission.

Master Thesis

Name: Hannah Kroes Student Number: s2092611 Supervisor: Dr. Caelesta Braun Word Count: 15910 Date: 11-01-2019

(2)

2

Introduction

What to do when scientific information about a risk does not match the public perception about that risk? The Fipronil crisis showed that perceptions matter: while there was no scientific indication for panic about the presence of Fipronil in foodstuffs, public perception indicated otherwise. How do you ensure that public and stakeholder perception matches the policies you make to mitigate certain risks? The field of risk governance theory, developed by Klinke and Renn in 2012, attempts to provide a framework to deal with risks by balancing science and perception. In the following

paragraphs I will discuss how I came to my specific research question.

Because of the commotion around the Fipronil cases, I decided to explore the field of risk and crises for my thesis. During the literature review on the risk governance I realized there were some major gaps in the literature with regards to its empirical testing. First, the framework has not often been tested empirically overall. Second, the role of stakeholders, while emphasized greatly in the literature has not been researched empirically at all. Even though this research has not been done, the International Risk Governance Council, the leading advising authority on risk governance within public agencies bases its advice on this framework (IRGC, n.d.). Therefore, both practical and academic relevance is achieved by testing the theory with a focus on stakeholders since it addresses the empirical gap in the literature and provides justification or evidence against the use of the risk governance framework as a basis for public agencies in dealing with risk.

As a result, the research question I chose to devote this masters’ thesis to is as follows:

To what extent does stakeholder involvement constitute part of risk governance?

The methods with which I will attempt to address this research question are as follows. The European Commission is my unit of analysis. In dealing with transboundary risks, the Commission often deals with systemic risks: those risks that cannot be solved by science alone. From the Commission, public consultations are used as the data from which the involvement of stakeholders is analysed. This data are then analysed using the Cramer’s V statistical test and descriptive

methods. This will be elaborated more on in the Research Design section of this thesis.

This thesis consists of five chapters: the literature review, theory, the research design, results and analysis and the conclusion. In the literature review I will take the reader on a journey through the evolution of the field of risk: from objective risk to subjective risk; from simple risks to systemic risks; and from the management of risk to risk governance. Throughout the chapter I discuss limitations to the research that I hope to address using this thesis. In the theory I discuss the theoretical

expectations arising from the literature on risk governance, supplemented by literature on stakeholder engagement. Furthermore, I construct hypotheses. In my research design section, I justify the type of research I choose to analyse my research question, discuss my case selection and data collection methods and operationalize my variables and justify this operationalization and finally, discuss my coding methods. In the results and analysis section I discuss the data gathered, the method of analysis and the results in relation to the hypotheses developed in the theory chapter. Lastly, I elaborate on the implications of these results and give some recommendations for future research. Finally, in my conclusion I briefly summarize the thesis, answer the research question, and discuss the limitations of my research.

(5)

5

Literature Review

Introduction

This literature review elaborates on the different research fields that are implied in my research question: to what extent does stakeholder constitute part of risk governance. The first major field that I discuss is the field of risk itself. The second is the evolution of the management of risk into risk governance. Throughout, I will elaborate on the relevance of the perspectives for my research. Furthermore, I will discuss the research methods used in the different articles. Finally, I will discuss the gap in the literature that I address with my research: the distinct lack of empirical research in the field of risk governance, with the explicit focus on stakeholder involvement.

Risk and Uncertainty

The Realism versus Constructivism debate

One of the first scientific works on risk was done by Economist Frank H. Knight in 1921. In his work, he distinguishes between risk and uncertainty. Risk is defined as an event where the consequences are possible to know (Knight, 1921, p.233). This is impossible in the case of uncertainty (Knight, 1921, p. 233). Thus, risk is measurable, either through calculation of probability or through

experiences from past events (Knight, 1921, p.233). This notion of risk as measurable was also taken on by the British Royal Society, which focused on the use of experts when assessing risk (Adams, 1995, p.8).1_{This aided a field of risk research in which authors argued it unnecessary to include} different perspectives into account (Adams, 1995, p.16).

This notion of measurable, objective risk was countered by multiple authors. One of the first of these was Mary Douglas (Thomson and Rayner, 1997, p.139). She argued that it is impossible to make a distinction between objective and subjective risk as risk is inherently socially constructed (Thomson and Rayner, 1997, p.139).2_{On the other side of the pond, in the United States, the National Research} Council (NRC) (1982) also argued that it is impossible to deal with (social) risks in a scientific manner, because there are always different interests at play and that even risk analysis by scientists is subject to subjectivity (p.27-29). Furthermore, relying on scientific analysis of risk may be dangerous

because risk calculations might seem more precise that they actually are (NRC, 1982, p.42).

However, if they are assessed qualitatively, the results might be perceived as less precise than they actually are (NRC, 1982, p.42). Both are unwanted, and therefore require a balancing act (NRC, 1982, 42-43).3_{In 1997, the British Royal Society finally conceded its previous position that risk should only} be looked at using objective assessment by arguing in favour of stakeholder and public involvement in risk assessment (Thomson and Rayner, 1997, p.140).

The balancing act and possible conflict between quantitative and qualitative risk assessment is addressed by many authors (Stirling, Renn, Rip and Salo, 1999; Beck, 2002). In practice the inclusion of perceptions means the inclusion of stakeholders in the risk assessment process (Canadian Council of Science and Technology Advisors (CCSTA), 1999; National Research Council, 1996; OECD, 2003).

1_{The Society distinguishes objective and subjective risk and argued that policymaking should only be done based on}

objective risk, meaning expert assessment, and alternative (stakeholder or public) perspectives were not to be included (Adams, 1995, p.8)

2_{Douglas and Wildavsky (1983) give examples for this position on risk as subjective: How are risks selected for assessment}

and why? And how come that even scientists often disagree on the nature and intensity of risk and their solution? (p.6; p.63)?

3_{That there is a subjective part in risk assessment is echoed by different authors as well (Amendola, Contini, Ziomas, 1992;}

(6)

6 An important contributor to this position that stakeholders should be included in the risk assessment and management process is the TRUSTNET project from the European Commission which was the start of a new perspective on the management of risk (European Commission (EC), 2000a).4_{In this} project, a new paradigm called the Mutual Trust paradigm was tested, as opposed to the traditional top-down perspective. In this new paradigm, stakeholders gain a bigger role in the risk assessment and management process. Science and expertise no longer played the determining factor in decision-making, making room for different perspectives as well (EC, 2000a, p.15).

What becomes more and more apparent, is that the traditional definition of risk by Knight (1921) is not necessarily wrong, but that an evolution has taken place. Currently, when talking about risks, uncertainty is almost always included, instead of there being a distinction between the two. When discussing contemporary risk, one almost talks about uncertainty and the inability of science to address this uncertainty. This necessitates the inclusion of different perspectives in risk assessment and management (EC, 2000a, p.20). Out of this principle, the precautionary principle arose. This is the principle that the decision to reduce risk should not have to await scientific certainty, because this certainty might be impossible to reach (CCSTA, 1999, p.7).5_{This precautionary principle has also} been taken on by the European Commission (2000b). Thus, the European Commission argues that science and expertise are no longer the determining factor in decision-making.

Over all, the idea that risk and its assessment are completely objective has been countered and is no longer found often in articles. The importance of including stakeholders and the public is also consistently affirmed. When looking at the research in the field of risk, specifically in this debate a few criticisms come to mind. First of all, the lack of empirical research is quite apparent. Of the articles discussed above, very few authors conducted empirical research, except for the articles about European research projects (Amendola, Contini, Ziomas, 1992; Commission, 2000a) and the OECD (2003). Furthermore, the subject of most of these articles is chemical risk assessment, thus excluding many fields from research. Finally, despite the debate, it is still difficult to find an exact definition of risk. This is because there is no such thing as a singular, agreed upon definition of risk, and there are many different classifications of risk.

Possible definitions of risk

A first example of a broader definition of risk is as follows: the probability or expectation of future damage or loss (van Asselt, 2007, p.10; Adams, 1995, p.8; van Asselt and Renn, 2011, p.437). However, in most definitions uncertainty plays a main role, contrasting this first definition.

This uncertainty comes back in the risk classification scheme of Douglas and Wildavsy in 1983, which classifies risk based on knowledge and consent. Knowledge here can be equated to certainty. Certain risks, upon which everyone agrees, are technical risks and only require calculation to solve them.

4_{This new perspective steps away from the traditional top-down perspective that the public authorities are the main}

governing actor in risk management, where stakeholders are only allowed to represent their specific interest and there is a reliance on science and expert justification (EC, 2000a, p.11-12). However, due to an increase in uncertain risks, (p. 20) this is no longer sufficient, and because of this, issues arise such as the possibility of reliance on incomplete science and stakeholder conflicts (EC, 2000a, p.12-13).

5_{One criticism to this precautionary principle in its legal formulation is that there is often a ‘knowledge condition’ included,}

which implies the necessity to appeal to scientists and experts to reduce uncertainty (van Asselt and Vos, 2006, p.317). However, implicit to the context of the precautionary principle is that it is impossible to reduce uncertainty, which therefore warrants the precautionary reduction of risk without scientific certainty (van Asselt and Vos, 2006, p.316). Thus, there is somewhat of a contradiction in precautionary principle and the way it is legally formulated, which shows some of the struggle between quantitative and qualitative risk assessment that is still present in decision-making about risk.

(7)

7 (Douglas and Wildavsky, 1983, p.5). This is the closest to the definition by Knight in 1921. Uncertain risks, upon which everyone agrees, can be solved through more research (Douglas and Wildavsky, 1983, p.5). However, uncertain risks upon which there is disagreement, a solution is difficult to find. (Douglas and Wildavsky, 1983, p.5). Thus, here we see a mix of uncertainty and consent, or

ambiguity when it comes to risk classification.

There is also often a distinction made between systemic risks and simple risks. Simple risks are similar to the original definition of Knight (1921). Regarding systemic risks, there are two relevant perspectives. On the one hand, Beck in his Risk Society argues that risks are consequences of modernization. Dealing with risks means to deal systematically with dangers arising from

modernization itself (Beck, 1992, p.21). These dangers are often unknown and unintended, and their solution requires an interdisciplinary approach (Beck, 1992, p.29). This lays the groundwork for the development of systemic risks. Due to globalization and societal changes, many risks are no longer simple and measurable (OECD, 2003, p.8). Van Asselt and Renn in 2011 describe systemic risks as “embedded in larger contexts of societal processes” (p.436). Therefore it is necessary to deal with interdependencies and spillover effects (van Asselt and Renn, 2011, p.436; OECD, 2003, p.8). They combine different, often mentioned aspects of risk into one conceptualization of systemic risks. These are uncertainty, ambiguity and complexity (van Asselt and Renn, 2011, p.432).

Uncertainty means risks for which science cannot provide an easy solution, and which have not been previously encountered by society (van Asselt, 2007, p.10; van Asselt and Vos, 2008, p.281).6

Uncertainty increases with long-term risks (Underdal, 2010, p.387).

Ambiguity implies that there are many different perspectives to a risk, a plurality of viewpoints (van Asselt and Renn, 2011, p.444). Risks that include economic and ethical issues are often subject to high levels of ambiguity, with the explicit mention of food safety (Mahmoudi et al., 2015, 747-749; 753).

Finally, complexity means that events result from multiple causes, making it impossible to establish causal relationships (van Asselt and Renn, 2011, p.437; Klinke and Renn, 2012, p.275-276).7

Outcomes cannot simply be only reported in quantitative results (Connolly, Reid and Mooney, 2015, p.75).

There is another work that brings these same aspects into one conceptualization of risk, but adds seriousness to the mix (Klinke, Dreyer, Renn, Stirling, 2006, p.378-380).8_{The potential of harm is} already implied by the definition of van Asselt and Renn, and therefore does not have to be repeated in the conceptualization (van Asselt and Renn, 2011, p.437). This conceptualization of systemic risk through complexity, ambiguity and uncertainty will be used throughout my thesis and as a basis for my research.

6_{Uncertainty can result from variability, through natural and societal randomness, value diversity, behavioural} variability and technological surprises; from a lack of knowledge, through inexactness, limited observations, practical immeasurability, reducible and irreducible ignorance and indeterminacy (Van Asselt, 2000, p.236). 7_{This can be due to feedback loops, nonlinear relationships, and for example multiple variables affecting a} relationship (Klinke and Renn, 2012, p.275-276).

8_{With seriousness they mean the potential of harm, however in their model for risk analysis they see} seriousness more as a condition for the classification of something as a risk (Klinke et al., 2006, p.380).

(8)

8

Risk Management versus Risk Governance

In this section, I will firstly elaborate on the traditional risk management cycle. Second, I will discuss the risk governance framework in detail.

The traditional Risk Management framework

The traditional risk management framework always consists of three phases. However, the content of these phases is disputed. Two varieties of the framework are discussed here. The first consists of risk research, risk assessment and risk management, which was developed by the NRC and focuses on chemical exposure risks.9_{This approach heavily relies on scientific research to assess and manage} the risk (NRC, 1983, p.21). The second consists of risk assessment, risk management and risk

communication.10_{Here, the approach is top-down, which means that assessment and management} are performed by the public agency itself, and the results are communicated vertically to all actors affected by the risk (Haimes, 2012, p.1457).

Criticism and a new framework

Both of these approaches were criticized. The first approach of risk research, assessment and management received criticism that it fundamentally distinguishes between values and facts, while this approach ignores the inherent subjectivity of selection of risks to worry about and the selection of the methods to assess them (Amendola, 2002, p.21). Furthermore, decisions of risk reductions based on probability and fact ignores concerns of trust and liability (Amendola, 2002, p.21). Thus, in response to this criticism, the NRC developed a new framework, based on deliberative-analytic processes (NRC, 1996). This model focuses on the notion that judgements are inherent to risk analysis, and different perspectives should be welcomed instead of shunned (Amendola, 2002, p.22). Thus, early in the process different stakeholders are involved in the characterization of risk, while there is also still place for quantification (Amendola, 2002, p.23). This participatory approach considers the subjectivity of experts, as well as reducing the possibility of dissatisfaction by the public and stakeholders with the technical risk assessment process (Amendola, 2002, p.22). The second risk management cycle, consisting of assessment, management and communication is similarly criticized. Again, the role of stakeholders and societal values is argued to be underexposed in this framework (Klinke and Renn, 2014, p.443). Furthermore, this framework lacks the

instruments to deal with issues of complexity and ambiguity (Klinke and Renn, 2014, p.443). Thus, when it comes to the criticisms to both frameworks the realism versus constructivism debate outlined in the previous section on risk is reflected.

What happened to make these traditional approaches no longer sufficient? First of all, there are more possibilities to express risk perceptions for stakeholders and the public (Clarke and Short, 1993). This has increased the importance of taking into account stakeholder and public perceptions in policymaking. This is also seen as the democratization of policymaking. These possibilities consist of for example, a more open public debate and more public access to governmental decision-making (Hutter, 2006, 212).

This shift from decision-making solely by the public agency to a more participatory form of

governance is sometimes also called the governance turn of policymaking. This entails the idea that

9_{Used by the following authors: Douglas and Wildavsky, 1983 and NRC, 1983.}

(9)

9 governance should be done using networks and horizontal interactions between public and private actors (van Asselt, 2007, p.7-8; Boholm, Corvellec, Karlsson, 2012, p.1; Lyall and Tait, 2004, p.3). The government could no longer deal with risks by itself because science no longer provide solutions to risks because of the increase of uncertainty, complexity and ambiguity (van Asselt and Renn, 2011, p.433). The increase of these risks is in turn caused by globalization and the resulting transboundary issues (Walls, O’Riordan, Horlick-Jones and Niewöhner, 2005, p.636). While this governance turn seems to have gained a place in the literature, in practice this turn does not seem to be quite as clear (Stirling, 1998, p.100; Walls et al., 2005, p.656).11

In the literature, this shift to governance instead of government has led to the creation of the risk governance framework by Klinke and Renn (2012). This is now one of the leading frameworks on risk in the field. Looking back at the literature on the risk management cycle, it again can be said that there is a lack of empirical research to be found. Here, only the NRC (1996) and Walls et al. (2005) conducted empirical research.

The Risk Governance framework

The Risk Governance framework was first developed by van Asselt and Renn in 2011 and is focused on systemic risks: risks that are complex, uncertain and/or ambiguous. It is based on the

constructivist notion that risks should be managed by accepting scientific controversy and

researching alternatives solutions and policies (Van Asselt and Renn, 2011, p.438). Furthermore, it considers the institutional structures and policy processes that direct collective activities of public agencies and other actors, such as the public and stakeholders (Klinke and Renn, 2012, p.439). Van Asselt and Renn (2011) mention three principles guiding the risk governance framework: the principle of communication and inclusion, the principle of integration and the principle of reflection (van Asselt and Renn, 2011, p .439).

The principle of communication and inclusion entails that there should be a flow of communication between stakeholder and agency in a social learning process (van Asselt and Renn, 2011, p.439). Inclusion entails the inclusion of stakeholders that are affected by risk in the risk governance process (van Asselt and Renn, 2011, p.440).

The principle of integration entails the integration of all relevant knowledge and experience in the risk governance process (van Asselt and Renn, 2011, p.442). It points to the idea that the different stages of the risk governance process (assessment, management and communication) are all interlinked (van Asselt and Renn, 2011, p.442).

Finally, the principle of reflection establishes the need for constant reassessment and consideration (van Asselt and Renn, 2011, pp.442-443). These principles thus emphasize the need to involve stakeholders in the risk governance framework.

This first risk governance framework still employs the three phases of assessment, management and communication (Van Asselt and Renn, 2011, p.442). However, these phases did evolve overtime. In 2012, Klinke and Renn redeveloped the risk governance phases. From then on, the five risk

governance phases are pre-estimation, interdisciplinary risk estimation, risk characterization, risk

11_{In the early days of this turn, Stirling (1998) argued that there was more focus on communication to the} public and classifying perceptions of the public instead of actual public participation (p.100). Second, Walls et al. (2005) argue that there is no simplistic move towards governance, but the extent to which ‘governance’ is used differs among cases where different variables are at play (p.656).

(10)

10 evaluation and risk management and monitoring and control (Klinke and Renn, 2012, p.279).

Through all these phases communication and inclusion, integration and reflection is important (Klinke and Renn, 2012, p.279). Their framework is now also in use by the International Risk Governance Council (International Risk Governance Council, n.d.).

The five phases of Risk Governance explained

Pre-estimation phase:

As mentioned before, the risk governance framework was founded on the notion that risks are socially constructed (van Asselt and Renn, 2011, p.438). This makes it necessary to establish institutions that are sensitive to cultural early warnings (Klinke and Renn, 2012, p.279). This detection of early warnings is done in the pre-estimation phase. To do this sufficiently, first a large amount of actions and problems must be screened to look for those that could be perceived as risky (Klinke and Renn, 2012, p.280). Then, an exploration must be done of what public agencies,

stakeholders, the public and other actors perceive as risks and what types of problems they associate with this risk and related uncertainties (Klinke and Renn, 2012, p.280).

Interdisciplinary risk estimation phase

This phase consists of a scientific risk assessment based on the natural sciences, but also of a concern assessment based on the social and economic sciences (Klinke and Renn, 2012, p.281). The scientific risk assessment should focus on the aspect of physical harm that could result from the consequences of a risk (Ibid.). The concern assessment is about the issues that the society could link to a certain risk (Ibid.). While this is mainly a scientific assessment, this could include for example public surveys or stakeholder hearings (Ibid.). After this phase, there is a process of risk

characterization which consists of the evidence-based component of making the judgement on the tolerability of risk (Klinke and Renn, 2012, p.284).

Risk Evaluation

This is the value-based component of risk governance. Here, a public agency tries to assess the risk and provide an evaluation about its acceptability or tolerability in society (Klinke and Renn, 2012, p.282). For the risk evaluation, Renn, (2008, p.149) developed the traffic light model (see fig. 1), where risks are situated in a diagram as a function of probability and impact. Risks can be accepted, tolerable but in need of risk reduction and intolerable (Klinke and Renn, 2012, p.282). The risk evaluation phase is the most contested phase, and most susceptible to ambiguity and mandates extensive stakeholder and public involvement (Klinke and Renn, 2012, p.283-284).

(11)

11

Figure 1: Klinke and Renn, 2012, p.149

Risk Management and Monitoring and Control

During risk management all the information of the previous stages is reviewed. Based on this, decisions are made for the further dealing with the risk. The main goal is here is to ensure a

consensus among stakeholders and the public and to ensure a legitimate process to justify decisions on acceptability and tolerability that affect stakeholders and society (Klinke and Renn, 2012, p.286). However, even though this framework is in use by the International Risk Governance Council, it has been susceptible to criticism. It has been criticized for being both too complex and too simple, inconsistent, and lacking a coherent definition of risk (Boholm, Corvellec and Karllson, 2012, p.2). Furthermore, the participatory mode of decision-making could lead to a higher amount of lobbying and actually reduce a balance in interests (Boholm, Corvellec and Karllson, 2012, p.2-3). Also, there is the criticism that the risk governance framework is not grounded in reality and is unable to adapt to different contexts (Boholm, Corvellec and Karllson, 2012, p.3). Finally, there is the observation that there has been a very limited amount of empirical research done using the risk governance framework (van der Vegt, 2017, p.6). After doing this literature review, in my opinion, this criticism is valid.

Analysis

Clearly, there is a gap in the literature because of a lack of empirical research.

Throughout all the different sections it is noticeable that not many authors have conducted

empirical research to support their claims. This is the case for the field of risk itself, risk management and the risk governance framework. Therefore, the empirical foundations for the evolution towards the risk governance framework can be considered shaky. Out of all (approximately 35) the articles discussed, only nine used their own empirical research to provide evidence for their argument (Amendola, Contini and Ziomas, 1992; OECD, 2003; European Commission, 2000a; Beck, Asinova and Dickson, 2005; Klinke et al., 2006; Boholm, Corvellec and Karlsson, 2012; Walls et al, 2005; and

(12)

12 Tavares and Santos, 2014; NRC, 1996). Of these, only two specifically consider the risk governance framework as discussed previously in this chapter, consisting of pre-estimation, interdisciplinary risk estimation, risk evaluation and risk management (Tavares and Santos, 2014; Boholm, Corvellec and Karlsson, 2012). It is interesting that the authors of the works on this framework, nor the

International Risk Governance Council have conducted empirical research towards their risk governance framework. When it comes to the empirical research done regarding the risk

governance framework, there is a focus on the application of the framework on policymaking itself (Tavares and Santos, 2014; Boholm, Corvellec and Karlsson, 2012). Thus, no empirical research has been done when it comes to stakeholder involvement specifically in the risk governance framework, even though it plays a significant role in the framework.

The empirical articles that don’t discuss the risk governance framework are about risk assessment (Beck, Asinova and Dickson, 2005; Amendola, Contini and Ziomas, 1992), other risk analysis frameworks (NRC, 1996; European Commission, 2000a; Walls et al., 2005), the precautionary principle (Klinke et al., 2006) and systemic risks (OECD, 2003). Finally, the non-empirical articles are either literature reviews (e.g. Hutter, 2006; Stirling, 1998; Van der Vegt, 2017), theory building (e.g. Klinke and Renn, 2012; Renn and Schweizer, 2009; van Asselt and Renn, 2011) and other theoretical endeavors.

Therefore, it is necessary that research is done to test the risk governance framework empirically. After all, the framework is already internationally used and advocated for by the International Risk Governance Council. Of utmost importance as well is the inclusion of stakeholder involvement in the empirical literature on risk governance as stakeholders play a major role in the framework. Thus, their role in empirical applications of the framework is should be addressed.

The research that has been done regarding the risk governance framework has been primarily qualitative (e.g. Tavares and Santos, 2014). To develop more general findings on the framework more quantitative research should be done. When looking at the criticisms of other authors on the risk governance framework, for example that the concepts are not grounded in reality or adaptable across different contexts (Boholm, Corvellec, and Karllson, 2012), most of these criticisms can be addressed by empirical research.

Conclusion

In this chapter I have looked at the evolution in risk research: from an objectivist perspective on risk towards a constructivist approach to risk and from risk management towards risk governance. Overall, I concluded that there is a distinct lack of empirical research throughout these fields. Since the field of risk and dealing with risk has now evolved into the risk governance framework, it is important to at least test this framework and apply this framework in empirical research.

Furthermore, while the engagement of stakeholders is of main importance in the risk governance framework there has not only been a lack of empirical research, there has been none at all. The role of stakeholders in the framework should be researched. Finally, most research that has been done, has been qualitative. In order to develop a more general empirical validation of the framework, quantitative research can be of use. In my research I will attempt to fill these gaps. First, I will test the risk governance theory empirically especially focusing on the role of stakeholders within the framework. Second, I will do this using quantitative methods.

(13)

13

Theory

Introduction

In this section I discuss the theoretical foundations for my thesis and the hypotheses that follow from these. This will help answer my research question: To what extent does stakeholder

involvement constitute part of EU risk governance? In my literature review I concluded that risk

governance on its own has not been empirically researched sufficiently. The situation for

stakeholder involvement within risk governance is especially dire. Stakeholder involvement in risk governance is emphasized greatly in risk governance theory. Thus, this is my research focus. For my dependent variable of stakeholder involvement, I use two different operationalizations: consultation instruments and types of stakeholders. In doing so I differentiate within my dependent variable. First, I take an agency perspective in defining stakeholder involvement as different types of consultation. Second, I take a stakeholder perspective, by operationalizing stakeholder involvement as stakeholder types.12

I formulate two sets of hypotheses. The first set considers the relationship between consultation types and risk governance phases. The second set considers the relationship between stakeholder types and risk governance phases. Finally, I finish this chapter with a conclusion.

Consultation Instruments and Risk Governance

My first set of hypotheses considers the relationship between consultation types and risk governance stages. First, I discuss the different types of consultations, their audience and their objectives. Second, I will discuss stakeholder literature to explain what we can expect from these types of consultations and link this to the different risk governance stages. Third, I will provide my hypotheses.

Consultation Instruments

The four different types of consultations available for my research from the European Commission, are open public consultations, targeted stakeholder consultations, calls for experts and calls for information.13_{Open public consultations are open for everyone, citizens and stakeholders alike.} They can have any objective. This is different for calls for information and calls for experts. In those types there is a pre-implied goal: a goal of gathering information or expertise. Targeted stakeholder consultations don’t differ from open public consultations in their objective, they don’t have a pre-implied goal, but they are targeted towards specific types of stakeholders. The different

consultations will form the dependent variables in my research. The different phases of risk governance form my independent variables.

Literature on Stakeholder Involvement

The literature on stakeholder involvement discusses various types of information public agencies need to gather from stakeholders. Bouwen (2002) defines different types of resources that EU agencies need to be able to function. This is based on resource-dependence theory. This theory emphasizes the exchange of resources between public and private actors and the interdependence that results from this. The different resources are expert knowledge, information about the

European encompassing interest and information about the domestic encompassing interest (Bouwen, 2002, p.369). According to Bouwen, the European Commission mainly requires expert knowledge and the European Encompassing interest (Bouwen, 2002, p.3). Expert Knowledge here

12_{As operationalized in the Research Design}

(14)

14 concerns the provision of technical expertise and thus relates most to calls for expertise and calls for information. Since open public consultations and targeted stakeholder consultations do not have a pre-implied goal, we cannot say directly that one type of information is required more than others in the different types of consultations. Similarly, de Bruycker (2016) makes the distinction between expertise-based information exchanges which consist of technical, legal and economic information and political information exchanges.

When we link these stakeholder theories to the risk governance stages and their goals, technical expertise mainly relates to the scientific risk assessment stage. As mentioned in the previous paragraph, technical information mainly relates to calls for information and calls for expertise. Therefore, it is to be expected that, within the calls for information and calls for expertise the main observations will be of the scientific risk assessment stage.

Since open public consultations and targeted stakeholder consultations do not have pre-implied goals to draw from theoretically, we have to look more in depth into the risk governance stages to see which stages we are likely to see in these consultations.

The goal of the first phase of risk governance, pre-estimation, is to detect risks as early as possible and to determine the possible stakeholder and public perception of this risk (Klinke and Renn, 2012, p.280). The use of public consultations here would seem likely, as it allows a broad selection of stakeholders to answer the call, thus giving the Commission a broad view of the different

perspectives. However, the Commission guidelines on consultations state that public stakeholder consultations are only mandatory in specific cases. This is only mandatory in case of initiatives with impact assessments, evaluations, fitness checks, commission communications with the explicit goal of launching a consultation process and green papers (European Commission (EC), 2017, p.70-71). This is about legislation/initiatives that have already been proposed. Thus, this is not applicable to the risk governance phase of pre-estimation.

The interdisciplinary risk estimation phase firstly aims to assess risks using a scientific risk assessment based on the natural sciences (Klinke and Renn, 2012, p.281). General stakeholder involvement (such as targeted stakeholder consultations and open public consultations) is therefore not expected in this phase.

Secondly, the interdisciplinary risk estimation aims to assess the economic and social aspect of established risks (Klinke and Renn, 2012, p.281). This is the concern assessment. Here some stakeholder involvement can be expected. Experts of economic and social sciences can use non-public forms of stakeholder involvement such as focus groups to gather perspectives (Klinke and Renn, p.281). This aligns with targeted stakeholder consultations. These forms of stakeholder involvement also include surveys, which can be seen as a form of public consultations. Therefore, there is a chance that the Commission uses public consultations as a form of surveys for the concern assessment stage of risk governance. Furthermore, Commission impact assessments seem to fit in with the concern assessment stage and impact assessments are often the subject of public

consultations. This also increases the likelihood that there will be a significant amount of public consultations in the concern assessment phase (EC, n.d.).

Thirdly, the risk evaluation phase. Here, the possible impact and its social acceptability of a risk is judged (Klinke and Renn, 2012, p.285). This phase of risk governance is highly disputed, because of the difficulty to make value-based judgements (Klinke and Renn, 2012, p.282). In their view this is a stage that is suitable to include stakeholders in a broader way. This in contrast with the more selective stakeholder involvement of the interdisciplinary risk estimation stage (Klinke and Renn,

(15)

15 2012, p.287). Both public consultations and targeted stakeholder consultations would be suitable for this stage. For the Commission’s public consultations specifically, it also seems possible that risk evaluation is present in these consultations. Especially in consultations on green papers, because there is no specific proposal set in this case and is therefore not yet part of the management phase (EUR-lex, n.d.; Klinke and Renn, 2012, p.28). Regarding calls for information and calls for experts, their goals do not match the goal of the risk evaluation phase and an association is therefore not expected.

The fourth phase, risk management is composed of the review of the results of phase 2 (the

interdisciplinary risk estimation phase) and phase 3 (the risk evaluation phase). Also, the design and implementation of actions to manage the risk are involved in this stage (Klinke and Renn, 2012, p.284-285). Its specific stages are the identification and generation of different management options, the assessment and evaluation of those options, the selection of the management option, the implementation of that option and monitoring and control (Klinke and Renn, 2012, p.285). In theory monitoring and control is part of the risk management phase. However, in practice it comes after this phase. In the monitoring and control phase the implementation of the chosen management option is monitored and evaluated. Thus, in the hypotheses, I separate the risk management phase and the monitoring and control phase, as I did with the interdisciplinary risk estimation phase previously.

When it comes to the risk management phase where management options are identified, evaluated and implemented, Klinke and Renn argue that there is room for stakeholder involvement. They especially mention broader consultations when a risk is characterized by ambiguity (2012, p.285). Thus, it is likely that there is a significant amount of public consultations and targeted stakeholder consultations in the risk management phase of risk governance in the European Commission. Two of the mandatory subjects of public stakeholder consultations are evaluation and fitness checks This happens after an intervention is implemented. Thus, one would expect that monitoring and control play a significant role in the risk governance of the EU during public consultations and targeted stakeholder consultations.

Hypotheses

First, we can expect that most calls for information and calls for experts will have scientific risk assessment as their goal. Furthermore, we expect that pre-estimation is not present in any of the consultations. This results in the following hypotheses:

Hypothesis 1a: Scientific risk assessment will play the major role in calls for information and calls for experts.

Hypothesis 1b: There will be observations of all risk governance stages except for pre-estimation and

scientific risk assessment in open public consultations and targeted stakeholder consultations.

Stakeholder Types and Risk Governance

For my second dependent variable, I chose to look into the different types of stakeholders.

The types of stakeholders that I will look at in my research are more deeply discussed in the research design section. In the following paragraphs I use stakeholder literature and link this to risk

governance theory to develop the second set of expectations for this thesis. Since I do not expect pre-estimation to play a major role in consultations, which is my data source, I cannot test hypotheses regarding pre-estimation.

(16)

16 Bouwen (2002) argues there are three types of information necessary for EU public agencies of which there two applicable to the European Commission. Expert Knowledge, which is about technical expertise and information about the European Encompassing interest. According to Bouwen (2002), firms are the most likely provider of expert knowledge, while associations are best able to provide information about the encompassing interests. In the operationalisation of the types of stakeholders this would apply to citizen groups, professional associations and business

associations. Adding to this, if additional scientific knowledge is necessary, it is important to include natural scientists, thereby including research institutes (Klinke and Renn, 2012, p.281).

Linking this information to the different risk governance stages, the result is the following. Expert knowledge is mainly present in scientific risk assessment. Thus, an expectation arises that mainly firms and research institutes are part of scientific risk assessment. The encompassing interest resonates with the objectives of most of the other risk governance stages. This is because concern assessment, risk evaluation, risk management and monitoring and control all take into account the needs and interests of the relevant stakeholders (Bouwen, 2002; Klinke and Renn, 2012). Thus, it is expected that in these stages, citizen groups, professional associations and business associations are most often found to be part of the involved stakeholders. This results in the following hypotheses:

Hypotheses

Hypothesis 2a: In the scientific risk assessment phase, mostly firms and research institutes are involved.

Hypothesis 2b: In the other phases, citizen groups, professional associations and business associations and are most involved.

Conclusion

In this theory chapter I have discussed the expectations that I have derived from risk governance theory and stakeholder literature. I have supplemented this with some documents from the Commission itself to ensure that the hypotheses consider the Commission context. In the

hypotheses, risk governance is my independent variable. The variations on stakeholder involvement are my dependent variables: the different types of consultations and the different types of

stakeholders. Regarding the relationship between the different types of consultation and risk governance I first expect that scientific risk assessment is found most in calls for information and calls for expertise. Furthermore, except for pre-estimation, all phases will be present in public consultations and targeted stakeholder consultations. With regards to the relationship between the types of stakeholders and risk governance, I expect firms and research institutes in the scientific risk assessment. For the other phases it is expected that citizen groups, professional associations, business associations and trade unions are most involved. For the concern assessment phase, research institutes are also added to that list.

(17)

17

Research Design and Data Collection

In this section I will discuss the research design and data collection methods used for this thesis. Furthermore, I operationalize the theoretical concepts mentioned in the previous chapter.

Justification of Research Type

First of all, I need to discuss some general principles upon which my thesis is based. My research design is a positivist one, which means that I analyse reality instead of taking a normative approach where an argument would be made for how reality should be. This positivist approach becomes apparent in my research question: To what extent does stakeholder involvement form part of risk

governance? The main aim is to discover the association between different risk governance stages

and stakeholder involvement, using theory to provide a possible explanation for the association, making it an explorative research design. Since I attempt to test the risk governance theory against an empirical case, this is a theory-testing and deductive thesis. I chose this design because risk governance theory has not been empirically tested with regards to its arguments about stakeholder and public involvement. I mainly take an agency perspective, meaning that I look at stakeholder involvement and risk governance with the perspective of the European Commission. Thus, I mainly look at what the Commission would expect from different risk governance stages and how that relates to stakeholder involvement: what types of consultations would be used by the Commission and what types of stakeholders would be targeted by the Commission. However, for this last part I also look at what types of stakeholders respond to consultations, to make a comparison.

In the context of my research question, a quantitative research design is more useful than a qualitative design, as this allows for the analysis of a large amount of cases to provide a general overview of risk governance. A qualitative research design would be more useful for an in-depth view of a few cases. Since my focus is not on the outliers, a large-N design rather than a small-N design is suitable.

Case Selection and Data Collection

In this section, the case selection will be discussed, where I elaborate on my choice for the focus on the DG SANTE as the source of my observations. Then, I will discuss data collection.

I decided to focus on a few closely related policy fields within the European Commission. These policy fields are Public Health and Food Safety, which are grouped in the DG SANTE. This decision was made for various reasons. These policy fields are often exposed to risk because of their direct effect on human health (Martinez, Verbruggen, Fearne, 2013; de Sa, Mounier-Jack, Coker, 2009). This also makes them more susceptible to public perception and discussion because these risks can affect the public. Since my aim is to research risk governance, looking into a field in which risk plays a role makes sense and increases external validity.

In order to construct my dataset, I needed to collect data on Commission stakeholder involvement. Fortunately, the Commission publishes most of their consultations open for a broader audience online. This often includes consultation documents, the type of consultation, consultation objectives, targeted stakeholders and responding stakeholders. When finished, the responses to the

consultation are published on the official Commission consultation page

(https://ec.europa.eu/info/consultations). Furthermore, I looked at the specific consultation page for the DG SANTE for additional resources (https://ec.europa.eu/health/consultations_en). Overall, I looked at 150 consultations, ranging from 2010 to 2019. This includes some consultations that were still open at the time of coding. Thus, for those consultations I could not include information on the stakeholders who responded to the consultation, but I was able to provide the rest.

(18)

18

Operationalization and Coding

To transform the different risk governance stages into variables that we can research, I must operationalize them. However, since the field of risk governance knows little empirical research, especially on the use of stakeholders in risk governance it is difficult to draw from theory for my own research. Thus, I searched for an operationalization framework myself and came across the

International Risk Governance Council (IRGC). This council bases their work on the framework of Klinke and Renn (2012). The IRGC (2017) identifies several questions that a public agency should ask for each risk governance stage. These questions will be used to operationalize the stages in my thesis: if answering these questions is one of the objectives the Commission mentions in the consultation texts and documents, the consultation belongs to one of the risk governance stages. These questions do not have to be reiterated word for word, neither need all questions belonging to one risk governance stage be present in the consultation. Regarding the wording of the questions themselves, it is most important that the essence of the questions comes back in the objectives of the consultations. Furthermore, not all of the questions for a single stage have to be present in the consultations because it is unlikely that an agency touches on all questions in a single consultation. Also, even if just one, or a few questions of a single stage are discussed, this can be taken as sufficient evidence that the corresponding stage is involved in the consultation. This is because the essence of the stages, or their main goal is present in the separate questions as well. The variables of stakeholder involvement were easier to operationalize. For the consultation types I can draw from the Commission typology itself, and for the stakeholder types I draw from the guidelines of dr. Braun.

Independent variable: Risk Governance Stages

Pre-estimation

As previously mentioned in the literature review and theory chapter, the pre-estimation stage is the first stage of the risk governance framework and is about exploring a large amount of actions, policies or problems to look for those that include risks and to see what stakeholders and the public identify as problems associated with this risk (Klinke and Renn, 2012, p.279-280). The exploration stage is called the screening stage, and the identification of the public and stakeholder problems with risk is called framing (Klinke and Renn, 2012, p.279-280). Similarly, the IRGC identifies the pre-estimation stage as the issues that stakeholders and the public associate with a risk and the exploration of existing indicators, routines and conventions that could help to identify the risk and the ways the risk could be addressed (IRGC, 2017, p.11).

The IRGC identifies several questions a public agency should ask in this stage. The presence of these questions in public consultation texts would therefore point to the public consultation being in this phase. For the pre-estimation these are as follows:

1. What are the risks to be addressed?

2. Who are the stakeholders, what are the power relationships between them and how do they view the risks?

3. Which stakeholders are mobilized by the risk?

4. What are the boundaries of the evaluation in terms of scope, scale and time? 5. Is there already a problem arising from the risk that needs urgent action?

6. Are there scientific and analytic methods available to assess the risk or do we need new ones?

7. What is the current legal or regulatory context and how does this affect the risk? 8. Does the organisation try to identify emerging risks?

(19)

19 9. What is the organizational capacity of the public agencies and stakeholders involved? (IRGC,

2017, p.11-12).

Interdisciplinary risk estimation

The interdisciplinary risk estimation phase consists of two stages: the risk assessment and the concern assessment (Klinke and Renn, 2012, p.281). The risk assessment phase is the scientific part, where the physical harm of a risk is calculated by scientists of the natural sciences (Klinke and Renn, 2012, p.281). The concern assessment is done by scientists of the social and economic sciences and considers the broader social impact of a risk (Klinke and Renn, 2012, p.281). For the scientific risk assessment the IRGC questions are:

1. What are the potential damages of the risk? 2. What processes could create and control the risk? 3. How vulnerable is the system that should absorb the risk? 4. How severe are the accident scenarios?

5. Can the risk be quantified?

6. How confident can we be in the risk assessment process?

7. How much uncertainty is there regarding the probability of the risk?

8. Is scenario development used to assess the prospects of the risk? (IRGC, 2017, p.14). For the concern assessment these are:

1. What are the perspectives of stakeholders and how involved are they in the risk? 2. Are there biases that may affect this perspective?

3. How does the public react to the risk?

4. What role do current governmental structures and the media play in the perception of the risk?

5. Are there controversies regarding the risk that risk managers might have to deal with? (IRGC, 2017, p.15).

Risk Evaluation

The risk evaluation phase consists of the risk characterization and the evaluation of a risk. The risk characterization is the phase where the risk is concluded to be either simple or systemic (IRGC, 2017, p.17). The risk evaluation itself is the procedure where a risk is judged based on its tolerability. The overall risk evaluation is a function of tolerability and probability (Klinke and Renn, 2012, p.282-284). The IRGC questions for this stage are:

1. Are there other ethical issues to consider, beyond those considered in the concern assessment?

2. What are the societal norms and values for judgements about tolerability and acceptability and are these changing?

3. Do stakeholders have reasons to want a particular outcome of the risk governance process? 4. What are the time, budget and contextual constraints?

5. What are the political and strategic ideas about possible benefits and risks of certain actions?

6. Is it possible to change an action in such a way that you can substitute the risk for another,

(20)

20

Risk Management

In the risk management phase different options for addressing the risk are analysed and selected and the chosen option is implemented (Klinke and Renn, 2012, p.284-286). The specific stages of this phase as defined by Klinke and Renn (2012) are: identifying risk management options, assessing these options, evaluating these options, selecting the appropriate management options,

implementing the risk management options and finally monitoring and controlling the performance of this option (p.285).

The IRGC questions for this final stage are:

1. Which actors and stakeholders should be involved in the risk management process? 2. What are the management options and based on what criteria are these evaluated? 3. What is the impact of risk-reduction options?

4. What are the trade-offs between the risks, benefits and risk-reduction options? 5. Is there an appropriate level of international cooperation for systemic risks?

6. What measures are needed to ensure effectiveness in the long run? (IRGC, 2017, p.23) While for the previous phases, the questions of the IRGC have been sufficient and address all parts of the phases, in my opinion the questions for this phase are somewhat lacking. The questions here are focused on the selection and evaluation of management options, while this stage also comprises the actual implementation and monitoring of the chosen option. When discussing the monitoring and control of the implemented policy, the IRGC argues that decisions should be monitored for their effectiveness, reviewed, and feedback should be integrated in this process (IRGC, 2017, p.25). Thus, for this research I will include consultations which aim to gather perspectives on the evaluation of policies related to risk in this stage.

Dependent variables: Consultation Instruments and Stakeholder Types

Finally, stakeholder involvement has to be defined within the context of this research. I use two measures to measure stakeholder involvement. First, I use the types of Commission consultations as an operationalization for stakeholder involvement; the Commission makes a distinction between open public consultations, targeted stakeholder consultations, calls for information and calls for expertise. Second, I use a measure which looks at the type of stakeholder to operationalize stakeholder involvement instead of the type of consultation. For this, I use the coding scheme established by dr. Braun in the Coding Instructions Media Analysis 2-capture. In this coding scheme, a distinction is made between citizen groups, trade unions, professional associations, business associations, firms, research institutes, institutions with a public function, national institutions of EU countries, European institutions, courts, foreign public authorities, intergovernmental organizations and other stakeholders. Citizen groups include NGO’s and for my research individual citizens as well. A professional association organizes members of a profession, while a business association organizes businesses in an industry. When the Commission did not specify the stakeholders requested or responding, I coded the data as missing.

Coding

The coding sheet I used can be found in Appendix 1. First of all, the coding sheet includes descriptive data of the consultations in order to ensure that they can be retrieved again online. This increases replicability and therefore the reliability of my research. This descriptive information includes the name of the consultation as mentioned on the Commission website, the period of consultation, the policy field(s) of the consultation and the URL of the consultation page.

(21)

21 Regarding the Risk Governance stages, I first coded whether the consultation was applicable to this thesis, by looking at whether any of the risk governance stages was present in the consultation objective at all. This was coded in binary form. If this resulted in a 0, the risk governance stages were not looked at and coded as missing. If one of the risk governance stages was involved, I used the operationalization provided in the previous part of this chapter to see which risk governance stage the consultation belonged to. This was coded in binary. Thus, all the risk governance stages the consultation did not belong to were coded with a 0, and the risk governance stages it did belong to were coded with a 1. Finally, I included a column with evidence taken from the text of the

consultation provided on the consultation page, or from the consultation documents provided on the consultation page. If the evidence came from the consultation document, I included a URL to that document. If the evidence came from the consultation text on the consultation page, I did not include an URL, since the URL to the consultation text was already provided under the URL column in my coding sheet. When it comes to the different consultation types, I coded them by name in the first column, and the following columns were titled with the different types of consultations and coded them in binary form (0 if not present, 1 if present). I did the same with the type of stakeholders, which I coded by name in the first column about stakeholder types and in the following columns I titled them with the stakeholder names and coded them in binary.

Conclusion

In this chapter I have discussed the research methods, data collection and variable

operationalization of my thesis. With regards to the research methods, I discussed that I take a theory-testing, inductive approach. Furthermore, I take an agency perspective. The data collection is done using the online Commission consultation database. These consultations can be divided into open public consultations, targeted stakeholder consultations, calls for information and calls for experts. The risk governance stages, consultation types and stakeholder consultations were coded in nominally and in binary and multiple descriptive variables were included in the dataset to ensure replicability of my research. Finally, the risk governance stages were operationalized using the questions of the International Risk Governance Council established in 2017, which are questions that agencies are advised to ask in the different risk governance stages. There was no need to

operationalize the consultation instruments, as I could use the typology that the Commission provides itself. For the types of stakeholders, I used the typology of stakeholders as written in the Coding Instructions by dr. Braun.

(22)

22

Results and Analysis

Introduction

In this chapter I discuss the results from my research in relation the hypotheses constructed in the theoretical chapter. First, the data are discussed. Then, the statistical test is described, and the results of these tests and the methodology are given. Finally, the results are discussed in relation to the theory. Furthermore, some implications and recommendations for further research are given.

The Data

To test the hypotheses laid out in the Theory chapter I used the consultations from the European Commission for data. These consultations took place between 2012 and 2018. As explained in the Research Design chapter, these consultations were taken from the policy fields of public health and food safety. This is the domain of DG SANTE. In total, 146 consultations were looked at. Within those consultations, 162 observations were made regarding the types of consultations. 374 observations were made for the requested stakeholders and 500 were made for the responding stakeholders. The high number of observations for the requested and responding stakeholders resulted from the various stakeholder types looked at. Each consultation could have multiple types of stakeholders involved, and each type of stakeholder connected to a consultation had to be coded separately. Thus, for each consultation multiple observations could be made.

Unfortunately, I frequently found missing data in the datasets as well. For the types of consultations this remained quite limited, to 19 missing datapoints. These resulted from consultations where no risk governance stage was applicable. However, for the types of stakeholders missing data was found more frequently. Of the types of stakeholder requested, 141 observations were missing, 37.7% of the data. Likewise, the data for the responding stakeholders had 151 missing cases, which was 31% of the data. This was mainly a result from the Commission not specifying which stakeholders

responded to the calls for experts, targeted stakeholder consultations and calls for information. This is not an issue for this research. The hypotheses regarding stakeholder types do not discuss the relationship between stakeholder types and consultations types, but instead focus on the association between stakeholder types and risk governance stages.

Finally, the subjects of the consultations almost always had some type of risk explicitly involved. This contributes to the idea that both food safety and public health as policy fields are subject to risk as described in the literature review and research design (Mahmoudi et al., 2015, 747-749; 753; Martinez, Verbruggen, Fearne, 2013; de Sa, Mounier-Jack, Coker, 2009).

Risk Governance Stages

As is shown in figure 2, the distribution of the risk governance stages is heavily skewed. Scientific risk assessment is by far most often present in the consultations, with 67 appearances. In contrast, pre-estimation and risk evaluation were far less often visible, respectively appearing three and six times in the consultations overall. Concern assessment was seen somewhat more often, with 11

appearances. Finally, risk management and monitoring and control respectively appeared 37 and 24 times.

Thus, from this data we can see that scientific risk assessment, monitoring and control and risk management play the main role in the consultations. When we link this back to the literature on risk governance, this is unexpected. First, there is only limited room for stakeholder involvement in scientific risk assessment (Klinke and Renn, 2012, 281). Furthermore, while Klinke and Renn (2012) argue that pre-estimation and risk evaluation are both suited for extensive stakeholder involvement, this is not visible in the data. Even more, the data actually shows a relative lack of pre-estimation

(23)

23 and risk evaluation. For pre-estimation this was expected due to the nature of Commission

consultations. This was deliberated on in the Theory chapter.

Figure 2

Consultation Types

Figure 3 shows the distribution of the consultation instruments, my first dependent variable. Again, we see that the distribution is heavily skewed. By far, most consultations are open public

consultations. This type of consultation is seen 104 times. In contrast, calls for information, calls for experts and targeted stakeholder consultations were only seen respectively 15, 15 and 12 times. The lack of calls for information and calls for experts can be explained by the fact that the scientific committees of DG SANTE were the only ones to publish these types of consultations. The Commission itself did not. However, both the open public consultations and the targeted

stakeholder consultations can be published by the Commission itself and the scientific committees. Thus, it is interesting that there were so little targeted stakeholder consultations. Unfortunately, this skewed distribution was impossible to avoid by sampling, since there were very little consultations of these types in the whole population of consultations as well.

Types of Stakeholders

My second dependent variable, stakeholder types, is distributed less skewed. For the stakeholders requested (figure 4) only foreign public authorities, European institutions and trade unions have limited presence in the dataset. Theoretically, this is not surprising. Bouwen (2002) argues that the Commission will primarily seek out associations and firms for the provision of expert knowledge and the provision of information about the encompassing interest. However, the role of EU national public authorities in the requested stakeholders is surprising, since these represent Domestic interest. This result contrasts with the argument of Bouwen (2002) that the Commission primarily focuses on gathering information about the European Encompassing interests.

The responding stakeholders (figure 5) are similarly distributed: less skewed. Again, foreign public authorities, European institutions and trade unions have limited presence. Furthermore, the EU national authorities have a larger than expected role. Thus, the same theoretical considerations apply. Since the data on requested stakeholders and responding stakeholders are relatively similar,

(24)

24 the Commission receives responses from the stakeholders it targets. This is an interesting finding and might be of interest for further research.

Figure 3

(25)

25

Figure 5

The Test

The data collected using the data collection methods that I will discuss in the next part of this chapter will be analysed using a statistical test. Since I am calculating the association between nominal variables, the Pearson’s R test that is usually used for correlation analyses is not applicable. This only works when data is numerical. However. the Cramérs V test can be used to test the association between nominal data. The missing data in my dataset is not an issue, as these data were manually dropped from the analysis in SPSS. My alpha-value is set to 0.05. This means that the p-value must be below 0.05 to accept the hypotheses. Furthermore, with regards to the association: an association of 0.1 is considered weak, an association of 0.3 is considered a medium association and an association of 0.5 is considered a strong association. Regarding the hypotheses and the expected higher or lower presence of different consultation and stakeholder types, those will be commented on descriptively using the crosstabulations generated by SPSS. While the hypotheses could be addressed using only the crosstabulation, I chose to use the association test as well. This is necessary because an association indicates a relationship between two variables: one variable affects the other. In order to make judgements about whether the hypotheses are valid, I must make sure that risk governance stages and stakeholder involvement are related. Otherwise, the results could be completely random: even though the hypotheses might seem valid, the results could be caused by something other than the risk governance stages entirely. Taking this approach increases internal validity.

The Results

In the previous chapters the hypotheses, data collection, the data and the test have been discussed. Now, the results will be discussed. First, I will look at the association between the consultation types and the risk governance stages. Second, I will look at the association between the stakeholder types and risk governance.

Stakeholder Involvement in Risk Governance: Studying the effect of Risk Governance on Stakeholder Involvement in the European Commission