Is the ball always round? A multi-site case study of local decision-making processes in Dutch football security

(1)

Is the ball always round?

A multi-site case study of local

decision-making processes in Dutch football security

Waardhuizen, J.M.D. van 9-6-2016

Master Thesis Crisis and Security Management Supervisor: M.L. Anderson, MSc Second reader: Drs. G.M. van Buuren

(2)

1

Is the ball always round?

A multi-site case study of local decision-making processes in Dutch football security

by

Jorg van Waardhuizen

Student Crisis and Security Management at Leiden University

June 2016

“Advocates fight for outcomes.” Allison & Zelikow, 1999, p. 303

(3)

2

Introduction

In August 2015, police in the Netherlands were on strike because of prolonged contract negotiations. One strike occurred strategically during the opening weekend of the Eredivisie, the highest football league in the Netherlands. As head of the security triangle, each mayor of a city has the power to allow or cancel a football match, depending on their assessment of the security situation. Because of the police strike and the resulting lack of football security, a lot of matches were cancelled in the interest of public safety. But the mayor of Alkmaar, Piet Bruinooge, chose to continue the match AZ-Ajax in Alkmaar, opting to deploy private security forces in place of absent public police forces. The mayor of Tilburg followed in the footsteps of Bruinooge allowing its football club, Willem II, to play. With the use of extra private security, the matches were allowed to continue and without real problems1_{. Ironically, in The Hague and}

Nijmegen, where matches were cancelled, there were problems with supporters protesting against the cancellation of their clubs’ matches, and the police had to step in anyway.

The majority of mayors did cancel the matches of their football clubs. Most football matches were of the same risk-level; there were no rival matches (derby’s) and they all faced the same problem: an insufficient number of public police able to be deployed. Then why did similar decision-making processes, with seemingly similar circumstances, lead to different outcomes? The main question in this research will be: how can the differences of decisions in August 2015 on cancelling football matches be explained? The options for the mayors were relatively straightforward: continue with the use of private security or cancel the match entirely. The differences in decisions across these Dutch cities illuminates the lack of an apparent uniform policy for decisions regarding the cancellation of football matches (Kok, 2015). To what extent would such a uniform policy be beneficial, and what would such a policy entail? The police was happy with the outcome in Alkmaar and Tilburg though; they feel that the 300,000 working hours spent per year on football security measures could better be used on other matters2. The situation reopened the discussion towards the societal costs of securing football; should society pay so much for such a repeating weekly private event as a football league? This research might contribute to that societal discussion. The Dutch national government (Rijksoverheid) also investigated the risk analysis of municipalities. They released a report in March 2016, with recommendations towards the current routines for risk analysis

1

www.ad.nl/ad/nl/1012/Nederland/article/detail/4117082/2015/08/10/Voetbal-kan-ook-zonder-standaard-politie-inzet.dhtml, last accessed 28-12-2015.

2_{www.volkskrant.nl/politiek/ook-los-van-cao-kwestie-is-politie-inzet-bij-voetbal-zat~a4115041/}_{, last accessed} 28-12-2015.

(5)

4

for professional football matches. Their conclusion is that the current ABC risk classification lacks professional means (Auditteam Voetbal en Veiligheid, 2016). Researching the decision-making processes of August 2015 may help decision-makers in government to address such questions and inform policy and risk analysis improvements. Further, this research also contributes to the gap in the literature on public versus private security provision in the context of football matches.

There has been a rise in the use of private security in The Netherlands more generally (De Waard, 1999). According to Krahmann (2008), security traditionally is seen as a pure public good, which is non-excludable and non-rival in nature, and will therefore not be provided by a free market. However, the rise of private security asks to look at security as a commodity. Police shifting public tasks towards the private sector is already happening, and is most likely here to stay (Loader, 2000, p. 324). What is the purpose of privatization? Shifting functions from the public sector to the private sector can either be seen from an administrative, an economic, or a political point of view (Feigenbaum & Henig, 1994). From an administrative perspective, privatization can be seen as a pragmatic way of solving societal problems. Privatization is then a ‘tool’ in the administrator’s toolbox for solving these problems. It is this tool of privatization of security for policy and decision-makers that is important in this research. The only option for the mayors was to either cancel the match, or to allow private measures to replace the absence of the public police.

One of the explanations for a rise in private security is the decline of social control by traditional institutions, like religion, schools, neighborhoods and families. Police forces need to take over the traditional social control mechanisms (De Waard, 1999, p. 146). One of the reasons football could equally be self-controlling their social environment, mixed with security provided by the clubs themselves, is because football is in a lot of ways comparable to a religious activity. The same needs are fulfilled as seen by church attendances or other religious gatherings (Taylor & Taylor, 1997). But there has not been done much research into sports leagues. Research in security for sports events often focuses on sports mass-events (SMEs), like the Olympic Games and the FIFA World Cup. However, there is a research gap in exploring diffuse SMEs, like large national football leagues. The difference between specific and diffuse SMEs is along a time and space dimension. The Olympic Games are set in a specific space and time, while the Dutch Eredivisie is a SME, but diffuse in time and space; it takes place most of the year on different dates and at different places (Giulianotti, 2011).

The events that took place in August 2015, and the resulting decisions by the mayors, provides a unique opportunity to study decision-making behavior in the context of private

(6)

5

security deployment during football matches. There has been much research into the decision-making behavior of decision-makers in public policy. One of the most read and cited works is that of Graham Allison (1999). In his work on the Cuban Missile crisis, he is able to develop three explanations for one string of single decisions: a rational model, an organizational behavior model and a governmental politics model. In this research his thee models will be used to find an explanation for different decision outcomes from the same processes. Allison’s models are based on foreign affairs, although he states that they can be applied in domestic matters as well (Allison & Zelikow, 1999, p. 7). Despite this remark, there haven’t been many scholars that used the models in this area, so this research yields more insight into the use of this framework in domestic matters. Also, the work of Allison is created to provide different “conceptual lenses” to look at a problem; it provides multiple explanations for a single decision-making process (Dobrow, Goel, & Upshur, 2004, p. 209). However, in this research we have one event: police strikes during football matches. But instead of studying one policy reaction or decision to this event, we study multiple decision-making processes who are by law the same, are all faced with the exact same problem, but yield different outcomes. Therefore, this research uses Allison’s models in a different way than has been done before, providing a unique opportunity to evaluate the robustness of his models.

This research will start with a review of relevant literature. First there will be a review of decision-making theory; how does decision-making fit in the world of policy making and what are the main bodies of work here? Then the framework of Allison will be introduced. The three models will be explored individually. This section will end with an overview of scholars who have previously applied or used Allison’s framework. Following the theoretical section, the research design will be introduced, including the empirical setting of the research, the data collection methods and the methodology. The following section focuses on a reconstruction of the decision-making processes and a summary of the report by the Rijksoverheid concerning the risk analysis for football matches, followed by analysis of the processes. This research will end with a discussion of the results, recommendations, and concluding remarks.

(7)

6

Literature and Theoretical framework

Decision-making theory

There has been many research into decision-making in all kinds of fields: economics, psychology, business management. They all have developed different kind of models, but “we shall see that they all suggest that a variety of different decision-making process style exist, and that the likelihood of one being followed can be ascertained with some certainty by examining the nature of the actors making the decision and the constraints under which they operate” (Howlett, Ramesh, & Perl, 2009, pp. 142-143).

When looking at the policymaking process as a whole, scholars have named different stages in the process. The most common approach is to look at policymaking as a cycle, with five different stages and different key actors who are involved during that stage. The five stage are outlined below, with the key actors involved during that stage in brackets (Howlett, Ramesh, & Perl, 2009, p. 13):

1. Agenda-setting (policy universe) 2. Policy formulation (policy subsystems)

3. Decision-making (government decision-makers) 4. Policy implementation (policy subsystem) 5. Policy evaluation (policy universe)

The third stage, the decision-making, has the smallest amount of influencers. In all stages of the policy making process, actors can influence the process, like NGOs, lobbyists, and the media. However, when it comes to the actual decision, only the individual(s) remain with the authority to make public decisions. Of course, other organizations and individuals may still try to influence the decision-maker, but they have only a “voice” and not an actual “vote” (Howlett, Ramesh, & Perl, 2009, p. 140). Decision-making has been extensively studied in the field of psychology, economics and business management. During the 50s, which also marks the beginning of the development of policy studies, the first models of decision-making in public policy arose: the rational model and, in reaction to the former, the incremental model.

The rational model is derived from economic analysis, and consumer choice. In this model, decision-makers pursue maximization of the outcomes they have to make. According to this theory, they first set a goal for solving a particular problem. Then all possible alternatives are listed, together with a prediction of all consequences of each alternative. In the end, the alternative that solves the problem at the lowest cost, is the rational choice. Although the limits

(8)

7

to having all the knowledge to perfectly set out all alternatives and consequences were recognized early, the central idea remained that the approximations of the consequences and alternatives were close enough to perfect-rationality (Howlett, Ramesh, & Perl, 2009, pp. 144-145). Gulick (2007) was one of the first to apply this rational behavior into the workings of organizations. He developed the POSDCORB model, suggesting that organizations should maximize their output by careful planning, organizing, staffing, directing, coordinating, reporting and budgeting their activities.

One of the first real critics of this pure rational behavior model, was Herbert Simon (1946), who recognized there could never be something like an all knowing individual or organization. He introduced the term ‘bounded rationality’; decision-makers bounded rationale, in combination with their own preferences and cultural background, leads often to the choice of not a maximized alternative. Instead of maximizing an outcome, he would rather speak of a satisfycing outcome.

The incremental model developed by Lindblom (1959), rests on Simon’s explanation of bounded rationality. The results of decision-making are more based on political feasibility than on technical desirability. In this view decisions are not large, new ideas, but more adjusted existing ideas. This incremental view of the decision-making process rests on two principles. First, bargaining for new, radical ideas is too much of hassle and a waste of resources. The costs and benefits of existing arrangements are also easier identifiable and thus it requires less effort to continue the status quo, or only adjusting it a little. Secondly, organizations in which decision-makers operate tend to standardize practices, which leads to non-radical standard decisions.

One other author that has been deemed a classic in the field of policy decision-making, is Kingdon. First published in 1984, he proposes how an idea’s time has come. He identifies three streams: a political stream, a problems stream and an alternatives stream. Kingdon suggests that when these three streams come together, a window of opportunity arises for a policy adoption. These streams can be brought together at random, or can be influenced by policy entrepreneurs. These policy entrepreneurs can be anyone in the political process: a lobbyist, member of parliament etc. These policy entrepreneurs often look for the right problem that fits their alternative; in Kingdon’s model, the alternative or idea already exists and waits for the right problem to come along. One way a window of opportunity can open is by critical events: crises, elections or released reports are examples here (Kingdon, 2010).

Graham Allison borrowed from all previous authors (see table 1) in his work on the Cuban Missile Crisis in 1971, explaining the crisis from three different perspectives. The

(9)

8

revised work appeared in 1999. Allison and Zelikow (1999, p. viii) have a clear objective for their research: “How should citizens try to understand the actions of their government?” To do so, they provide three models for explaining decision-making: a rational actor model, an organizational behavior model, and a governmental politics model. In the next sections, the models are summarized.

Author Year Theory/ concepts

Luther Gulick 1937 Rational organizations; POSTCORB

Herbert Simon 1946 Bounded rationality

Charles Lindblom 1959 Incremental model

John Kingdon 1984 Three streams model; Critical events; policy entrepreneur

Graham Allison 1971, 1999 Rational Actor Model, Organizational Behavior Model, Governmental Politics Model

Table 1: Overview of decision-making theory classics

The Rational Actor Model

The rational actor model, or RAM or Model I, is in fact the rational model that is mentioned in the former section, and is based on rational choice by the decision-maker, which is supported by the assumption that “what humans do is at least intendedly rational” (Allison & Zelikow, 1999, p. 17). There are four core concepts within this model. The decision-maker has interests, which are mirrored in specific (1) goals and objectives. To achieve these, the decision-maker has a set of (2) alternatives out of which he may choose, and each alternative holds its own (3) consequences. Rational choice implies that the decision-maker will choose the alternative which leads to the highest ranking consequences: a (4) value-maximizing choice (Allison & Zelikow, 1999, p. 18). When applying this model, the basic unit analysis is “governmental action as a choice”. The rational decision-maker is a unified national actor: the government. One important proposition in this model is that when the estimated costs of an alternative rise, the likelihood of that alternative decreases, and vice versa. So, when government has made a decision, that decision must be seen as a value-maximizing choice out of different other options. That is the explanatory power of Model I and most analysts reason through this model (Allison & Zelikow, 1999, pp. 4, 24-25). There is, however, one limitation in this model, referred to as the “Rationality Theorem”:

(10)

9 …an imaginative analyst can construct an account of preference-maximizing choice for any action or set of actions performed by a government. (…) There exists no pattern of activity for which an imaginative analyst cannot write a large number of objective functions such that the pattern of activity maximizes each function. (…) he must insist on rules of evidence for making assertions about governmental objectives, options, and consequences that permit him to distinguish among the various accounts (Allison & Zelikow, 1999, pp. 25-26).

The Organizational Behavior Model

The main difference of Model I with the Organizational Behavior Model (OBM or Model II) is that government is not seen as an individual or an unitary actor. Government action is performed by a variety of organizations, all created for specific purposes. These organizations create programs and routines to deal with their daily work. Analysis through this model has organizations as basis unit of analysis, like a ministry, agencies, or a subunit of these. Situations are not treated individually by organizations, but are classified to fit a certain routine in which they can be treated further, the so-called “logic of appropriateness”. This differs from the “logic of action/consequence” by which decisions are made in Model I (Allison & Zelikow, 1999, p. 146). In this sense, the available programs and routines form the options to choose from for government leaders, and this makes their decisions seldom unique. The existing routines and programs actually influence the choices of government leaders; standard procedures are low in cost and often proven effective. The influence of government leaders is lower than one might expect; organizations “are not turned on or off just by pulling a switch” (Allison & Zelikow, 1999, p. 173). Although these organizations all specialize in some problem they want to solve, important problems often hit multiple organizations, both vertically and horizontally (Allison & Zelikow, 1999, pp. 143-148).

Model II sees governmental action as organizational output. The routines created by organizations are called “Standard Operating Procedures” or SOPs. Because of the SOPs, organizational behavior often seems slow and inefficient in the eyes of an outsider. A collection of SOPs designed to effectively coordinate the behavior of numerous individuals is called a program. Change in routines and programs often come from budgetary drops or rises, and of course performance failures. The reason that organizations create SOPs and programs, is because they want to avoid uncertainty: “by arranging a negotiated environment, organizations try to maximize autonomy and regularize the reactions of others with whom they must deal” (Allison & Zelikow, 1999, p. 170). When faced with a problem they cannot solve within existing routines or programs, organizations tend to search for other solutions by a problem-oriented approach: the problem they cannot solve becomes an uncanny problem for them. Often

(11)

10

the then resulting solutions are biased, because of the specialized behavior, training and capabilities of the organization. The explanatory power of Model II lies in the uncovering of the standard routines and capabilities of organizations. Then action can be predicted, because organizations behave in a standardized way: “the best explanation of an organization’s behavior at t is t-1; the best prediction of what will happen at t+1 is t” (Allison & Zelikow, 1999, p. 175). A flaw in this model is that when decisions were made through existing routines, it can tell the analyst nothing about the intentions of the leadership (Allison & Zelikow, 1999, pp. 168-175). Also important is the position of the organization in its environment and its (power) relations with other organizations, and their mission or the goals of the organization and its capabilities to achieve those goals. The responsibility for specific problems and the required capabilities to solve them create strong beliefs among the members of the organization in the mission of the organization and lead to organizational culture. This culture is further strengthened by the organization’s definition of “acceptable performance”, also defined as compliance with existing rules, the available information, special systems and specific recruitment (Allison & Zelikow, 1999, pp. 167-168).

The Governmental Politics Model

The governmental politics model focuses again on the individual, but not as an unitary actor. Instead of an individual rationally seeking to fulfil the interests of its organization, they have their own personal interests, and are players in a game: politics. Governmental behavior in this model can best be described as resultants from a bargaining game of “competing preferences”. To fully understand complicated decisions, there is often a combination of resultants of smaller and larger bargaining games needed to fully explain the decision (Allison & Zelikow, 1999, pp. 255-257). The most important factor in this model is to know who plays the game. Who are the influencers, what organization do they belong to, what are their preferences, what are their roles? It is not only dependent on the position of the player, but also on its personality; how does he or she cope with problems, what is his/her style? First, main preferences and priorities can often be predicted by the organizational goals the player is part of: “where you stand depends on where you sit” (Allison & Zelikow, 1999, p. 307). Second, the player has its own ideas and opinions on matters, also outside their expertise. Third, during the game, decisions and actions influence interests, objectives and personal concerns. Finally, problems must first become problems: “Deadlines and events raise issues and force busy players to take stands”. These events and deadlines can be raised by immediate crises, political speeches or changes in budget .

(12)

11

The impact of players on the game can be stablished by skill, responsibilities and access to information (Allison & Zelikow, 1999, pp. 294-299).

Then what is this game? Decisions are often made through action-channels; regularized means of governmental action. These action-channels can structure the game by selecting players, and their proposed course of action. The game also has a set of rules. Decisions can be made formally and informally. Within the formal setting there can be a majority vote system, an unanimous vote system, or the decision may be made by an single individual (Allison & Zelikow, 1999, pp. 300-303). There are factors that influence the results of group decision-making. Decisions are often better when made in a group, but also lead to the principal-agent problem. The decision-maker, the principal, relies on expertise and often information by his advisors, the agents. So agents exerts their information and expertise as power over the principal, while they need the principal to make a formal decision. Another risk of group decision-making is the concept of groupthink. Homogenous groups with shared expertise often have a high degree of cohesion which make their analysis on problems often biased. They often bolster the positive consequences of preferred alternatives. Important to acknowledge here is that the game is never over, even after an important decision is made; policy decisions are an ongoing, never-ending process (Allison & Zelikow, 1999, pp. 265-283).

Since power relations are often confusing and take a long time to write down on paper, some scholars have attempted to develop a more systematic way to describe power relations. Michaud (2002) has actually developed a graphic way to project power relations within a governmental system, combining Allison and Lemieux (1989). However, this model would only be helpful if power relations within a decision-making process would be numerous and non-hierarchal: when the decision-making process is complicated.

Review of work using Allison’s models

Allison’s work has been cited a lot within the policy research field. Often, it is used to acknowledge the fact that government action isn’t as rational as it is often perceived to be (Forsynth, 2002, p. 2527) or to point out the fact that government officials look at problems through a governmental lens, often directly cited as “where you stand depends on where you sit”. One of the researches of the latter kind also shows an overlap in field of study. Decision-making happens of course in all kind of organizations. In the field of Operations Management, Dilts and Pence (2006) have tried to explore the role of decision-makers in failed projects. They researched whether the role of (1) project manager or (2) executive sponsor, leads to different decisions when terminating an overdue or over costly project in the public sector. Or as they

(13)

12

say, “when terminating a project, does where you stand depend upon where you sit” (Dilts & Pence, 2006, p. 378). They incorporate Allison and Zelikow’s multitheory perspective on decision-making. Allison’s account of the Cuban Missile Crisis is also often used as comparison to other crises or foreign affairs problems (Jacobides, 2007).

There are also examples of direct use of Allison’s models. Axelrod and Borzutzky (2006) analyzed NATO’s approach to the war on terror. They think that Model II, the Organizational Behavior model, applies to NATO’s approach to this war. NATO has implemented a lot of new policies and techniques in order to counter terrorist organization Al Qaeda. However, Axelrod and Borzutzky see these policies and techniques as new routines as in Model II. The slow, incremental change of routines by organizations make these policies and techniques obsolete before they are used. Al Qaeda has already transformed into “Al Qaeda 2.0”, while the new routines and techniques are developed with the old Al Qaeda 1.0 in mind. This proves Model II predictability, that organizations behavior at t can best be predicted by its behavior at t-1.

Another example of direct use of Allison’s models is that of the acquisition of new aerial tankers for the US Air Force, researched by Franck and Udis (2015). This is one of the few researches that applies Allison’s framework in a national decision-making process. Defense acquisitions are often highly technical and difficult politicized acquisitions, but aerial tankers are deemed a rather simple acquisition within this category. But the attempt to select a new tanker by the US Air force started in 2001 and the decision was made in 2011, with the tankers only be airborne in 2017. Why did the decision-making process of a simple acquisition took so long? Franck and Udis use Allison’s models as explanations to why the selection criteria of the aerial tanker were changed two times. Once in 2003, when a former government employee who got a job at Boeing and favored them above Airbus when she was working in government. The other time was in 2008, when Boeing successfully asked for a revaluation of the tanker criteria, since their tanker was more suited to the original criteria than EADS, their competitor who got the contract in 2007. The contract was revoked after an investigation, and again, a new bidding attempt was made by the government in 2010 with new selection criteria, when Boeing finally got the contract. The changing of the criteria have been explained by all three models. First, the changing of the selection criteria was based on rational behavior; changing circumstances needed new criteria according to rational reasoning, a Model I explanation. Second, the change of the selection criteria is an incremental change of the routine of acquisition within the US Air Force. Although the organization could better take a look at itself, it made small changes to a routine, a model II explanation. Finally, congressional powers made the shift possible; Boeing’s presence in Congress was larger than its competitor, EADS, and they made the shift in selection

(14)

13

criteria possible by publicly criticizing the US Air Force on this matter. This is an result of political maneuvering, making it a Model III explanation (Franck & Udis, 2015, pp. 15-18).

Allison’s work has not been without critique. The main critique on the 1971 edition has been the line “where you stand depends on where you sit”. Critics have, and rightfully so, pointed out that this might be true in some cases, but remains incorrect in most instances. But this critique is countered in the 1999 edition, where Allison and Zelikow remind their readers that “depends” does not mean “is always determined by”, but rather “is substantially affected by” (Allison & Zelikow, 1999, p. 307). A critique on his 1999 work is the acknowledgement of factual errors about the Missile Crisis made in his first publication of the theories in 1971, but the failure to name these errors in detail. For example, Allison and Zelikow have excised most errors, but often don’t mention them. This is important because the removal of these errors removes important evidence for Model II (Houghton, 2002, p. 162). Also the adding of the Groupthink framework is a point of critique. Although the concept of groupthink fits well into Model III, the author of the groupthink approach sees no groupthink during the Cuban Missile Crisis, but rather “vigilant appraisal” which fits more into Model I (Houghton, 2002, p. 166). Although these critiques all focus on the Cuban Missile Crisis’ facts, one must not forget that the theories of Allison and Zelikow rest heavily on the empirical evidence found in this case. A last critique is towards Model II and III. After the first publication in 1971, Allison later merged Model II and III into a single model: the Bureaucratic Politics model. However, in the 1999 revision they are separated again, leaving questions to the purpose of combining them in the first place (Houghton, 2002, p. 178; Michaud, 2002, p. 275).

After studying various works that have used Allison and Zelikow’s framework, three points can be made. First, although Allison and Zelikow’s work has been cited extensively, there are not many studies that directly apply the models. One explanation for this could be that the models are more practical in nature than theoretical, an issue the authors also address themselves (Allison & Zelikow, 1999, p. viii). Second, most studies that have applied the models directly stem from the area of foreign policy, and very few from the domestic policy area. Also a remark made by the authors themselves in the book, although they do add that their models can be used “across the wider array of governmental actions” (Allison & Zelikow, 1999, p. x). Applying the models in domestic decision-making processes might yield more insight in the usefulness of the models in this area. Third, all models and examples mentioned above look at a single decision outcome, which seems logical since Allison and Zelikow have used their models in a single case as well, the Cuban Missile Crisis. The strong side of Allison’s model is that it can fully explain a single decision outcome through different lenses, resulting

(15)

14

in different explanations (Dobrow, Goel, & Upshur, 2004, p. 209). However, in this research we have similar decision-making processes, with similar circumstances, happening at the same time, faced with the same critical event. Yet some outcomes are different from the others. This setting allows for an unique opportunity to evaluate Allison’s framework and contribute to the wider discussion on policy decision-making processes. So the scientific contribution of this research is threefold. It contributes to the body of knowledge that has applied Allison’s framework, specifically in the domestic area, and in an unique empirical setting. In the next section, the design of this research is elaborated on, starting with a description of the empirical context, followed by the methodology and data collection methods.

(16)

15

Research Design

Empirical context

Football security has always been an issue in The Netherlands, as a traditional football nation with notorious clubs like Ajax and Feyenoord. The most notable result of ongoing attention by Dutch government for this issue, is the introduction of the Voetbalwet (Football-bill) in 2007. This bill makes it possible for BVOs (abbreviation for professional football clubs) to punish hooligans more severely, like lifelong stadium- and area bans3_{. The bill was even further}

amended in 2015, giving the mayors opportunity to forbid fans from their home club to travel to away games4. Dutch government has been trying to bring down the police deployment with these bills, according to answers of questions by MPs5. The Dutch National Football Association (KNVB) states on its website that both the police deployment and the number of incidents regarding football vandalism have declined with 10 percent in the 2012-2013 season6. However, the police union NPB said in August 2015 that football security still costs over 300,000 working hours a year7. Most of the questions asked in parliament and news articles focus on hooliganism, but this is outside the scope of this research. Although the measures to reduce police deployment during football matches essentially replace public police with private security companies and stewards, it is never framed in a way of privatization.

Security during football matches is separated: the BVO is responsible for security inside the stadium, government outside the stadium8. The current system for assessing risks of football matches is done by the municipality of the home-playing BVO. In municipalities, the mayor is responsible for security, and has special powers in this area. The same counts for football security. Normally, security issues in any field are assessed during meetings of the Security Triangle, which includes the mayor, the police and the Public Prosecutor. The police, Public Prosecutor and the mayor’s staff advise the mayor on the issue discussed (see figure 1).

3_{www.nu.nl/sport/1130588/nieuwe-voetbalwet-maakt-stadionverbod-mogelijk.html}_{, last accessed 19-5-2015} 4_{www.vi.nl/nieuws/eerste-kamer-akkoord-met-aangescherpte-voetbalwet.htm}_{, last accessed 19-5-2015} 5_{Aanhangsel van de Handelingen II, 2013-2014, 786}

6_{www.knvb.nl/themas/veiligheid/cijfers-en-onderzoeken}_{, last accessed 31-12-2015}

7

www.ad.nl/ad/nl/1012/Nederland/article/detail/4117082/2015/08/10/Voetbal-kan-ook-zonder-standaard-politie-inzet.dhtml, last accessed 28-12-15.

(17)

16

Figure 1: Local security provision process: the security triangle

When assessing football security, the BVO is added to that triangle and is therefore called the Security Quadrilateral (see figure 2).

Figure 2: Local security provision process during football matches: the security Quadrilateral

Security

Triangle

Public

Prosecutor

Mayor

& Staff

Police

Security

Quadrilateral

Public

Prosecutor

Mayor

& Staff

Police

BVO

(football club)

(18)

17

When analyzing the risk of a certain match, municipalities use the ABC classification risk profile since the introduction in the 2003-2004 season. An A-match has low risk, a B-match some risk and a C-match a high risk. This classification is based on earlier matches and experiences, often of the former season’s matches. Depending on the risk assessment, the mayor can ask for more or less measures for security, for example extra police deployment and more security measures in stadiums by the BVOs, like increasing the number of stewards. This ABC classification has been in use since 2003. This process of risk analysis was created to find a balance between the following goals: prevention of incidents by applying appropriate measures, preventing unnecessary restrictions for fans to visit a match, and preventing unnecessary high costs for matches, like police deployment. The Auditteam Football and Security has released a report in March 2016, on the use and effectiveness of the ABC risk profile, because they saw a lack of flexibility and precision in the use of the profile in several places (Auditteam Voetbal en Veiligheid, 2016).

The report has identified the following bottlenecks of the ABC risk profiling. First, the risk analysis is performed in a very routinized way; there is no specific analysis of each separate match, and often a standard package of measures is taken. Analysis is often based on matches against the same opponent last season, and the reputation of the club and its fans, and not on the actual facts and current information. Second, there is a lack of precise measures. The ABC profile contributes to this, because each risk level has its own standardized package of measures. An A-profile often means no measures and C a lot of measures. Most matches are consequently deemed a B-match, because only this classification allows for a combination of measures (Auditteam Voetbal en Veiligheid, 2016, pp. 4-5).

The risk assessment of the matches in the Eredivisie weekend in August 2015 are also briefly discussed in the report:

…the preparations of the matches, including the decision of continuation, have been different from the usual process. De circumstances (no police deployment, preparations done by others due to vacations) allowed to look at the process and the accompanying risks from a new, fresh viewpoint. The decision-making process was not based on routine, but on the basis of a solid risk analysis (Auditteam Voetbal en Veiligheid, 2016, p. 6).

The suggestion of the Auditteam, is to create a system where each preparation of a match starts with a clean sheet. They opt for a ‘button’ model to replace the ABC risk model. In the button model, each of the measures the actors can take are identified in different classes. Each class has its own buttons that can be turned, meaning each button is a different measure that can be

(19)

18

applied to some certain degree, like police deployment in the Oversight & Control class. See figure 3 for the complete model.

Figure 3: the proposed Button model. Source: Auditteam Voetbal & Veiligheid (2016)

Why did the police strikes in August 2015 take place? New contract negotiations have been going rather slow, and in winter 2015, the police unions rejected the final offer by the ministry of Security and Justice, and went on strike. At first, the strikes were small, focused for example on less speed checks and overall less fining, but soon they turned to harder measures like strikes during events and protest marches. The police unions first targeted professional football in May, where a strike was held during the Eredivsie matches Feyenoord-Vitesse and FC Groningen-PEC Zwolle. The mayor of Rotterdam, Aboutaleb, thought the risks involved were too high and cancelled the match, but the match in Groningen was allowed to continue. In absence of a new offer, finally the unions decided to do a total strike during the opening weekend of the new 2015-2016 season of the professional football leagues, the Eredivisie and the Jupiler League, in August (Schildkamp & Winterman, 'Chaos in voetbal niet onze schuld', 2015).

(20)

19

A last important remark here is the clear division of the police leadership and the police unions. The police unions feel that the leadership, especially national police chief Gerard Bouman, is often on the side of the ministry in the matter of the contract negotiations. This remark is important, because it is the police leadership that advises the mayors in the security triangles and quadrilaterals (Ruzie bij politie loopt hoog op door cao-acties, 2015).

Research objective

The events in August 2015 represent an interesting research puzzle. There was one critical event -the police strike - which affected many cities in the same way. The lack of available public police forced multiple mayors to decide to cancel their local football match, or continue with the use of private security. While some mayors chose the former, there were those who chose the latter. Further, despite the existence of a uniform risk analysis tool, decision outcomes differed. How can the difference of decisions in August 2015 on cancelling football matches be explained? This research will try to find an explanation for these decisions, and should provide feedback for the assessment of risk for security quadrilaterals in the Netherlands, and provide more insight in the use of Allison’s framework in domestic affairs, and its use in multiple case-studies.

Methodology and data collection

This research will be conducted using a qualitative multiple case-study design. The first and foremost reason for this approach, is that Allison’s framework only lends itself for such a design, and is therefore the only way to make this research feasible; we cannot simulate the events of August 2015 (Swanborn, 2010, p. 34).

The strike in August focused on all matches in professional football in the weekend of August 7 to 9, 2015. However, due to time and distance limitations, a selection of cases has to be made. Selection criteria will be on pragmatic grounds and on selection of critical cases, which is useful when testing a theory (Swanborn, 2010, p. 71). First of all, the matches in the Eredivise, the highest professional football league, were prominently given attention in the news, and clubs here have a higher need of police deployment than in the lower league9. So the first selection criterion would be an Eredivisie match. Second, the decision of continuing matches were the exception; three out of nine matches in the Eredivisie were allowed to continue, but one of them, Feyenoord-FC Utrecht, was allowed to continue because there were

(21)

20

no police strikes in Rotterdam, because of earlier strikes there10_{. So both matches that were}

allowed to continue are a critical case; they are the main reason for this research. These matches, AZ-Ajax and Willem II-Vitesse were both on Sunday. There were four matches in total on Sunday August 9. One of these matches that did not continue, NEC-Excelsior, is included in this selection as well, to compare the critical cases with a case that has the most equal circumstances . The fourth and last addition of cases is the match ADO Den Haag-PSV that should have been played on Saturday August 8. The reason for this is that ADO is situated in The Hague, the largest relevant city of all matches that were played that weekend and were affected by the strike. Large cities have often more experience in security issues and are therefore relevant to this research. In table 1, all the selected matches and their characteristics can be found.

Match Date City (residents)11 _Continued?

ADO - PSV August 8 The Hague (520,704) No

AZ - Ajax August 9 Alkmaar (107,822) Yes

NEC - Excelsior August 9 Nijmegen (172,322) No

Willem II - Vitesse August 9 Tilburg (212,943) Yes

Table 2: selection of cases

The cases are reconstructed, and then analyzed and compared through one of the models of Allison’s framework, through a best-fit, or a combination of elements of the models. The data for this research is derived from interviews held with stakeholders, policy documents, and media articles. The interviews are qualitative and in-depth. This provides information that would otherwise stay hidden. For each case an interview was held with an advisor of the mayor, and in some cases also with the home-playing clubs. The interviews can be found in the Appendix. The interviews are anonymous on request by the respondents, but the function of each respondent is included. The relevant policy documents are answers to questions asked in municipality councils regarding the decisions for August 8 and 9, 2016. The media articles are all from the period just before and just after the weekend. There are a lot of articles available,

10_{www.vi.nl/nieuws/overzicht-gevolgen-politiestaking-voor-start-eredivisie.htm}_{, last accessed 20-5-2016} 11_{CBS (2013),}_{statline.cbs.nl/statweb/publication/?dm=slnl&pa=82245ned}_{, last accessed 20-5-2016}

(22)

21

because it was a holiday period and this was a hot case to report about in a period with traditionally less news available.

(23)

22

The Cases of Football Security in August 2015

In this section all decision-making processes of the four cities are reconstructed, using interviews, policy documents, and newspaper articles.

ADO Den Haag-PSV

In The Hague, ADO Den Haag is the professional football club (BVO). If ADO has a match, they have to send a notification to the municipality. This notification is then reviewed by and discussed in the security quadrilateral, whether extra measures are necessary in regard to the risk profile of the match. For this risk profile, the ABC classification is used. The classification of the different opponents can change during the season. After the consultation in the security quadrilateral, a decision is made by the mayor, Jozias van Aartsen, to approve the match and the measures that need to be taken. Mayor van Aartsen is of the right-wing VVD party. He gives ADO much space, until incidents happen. Police should be there, but invisible. He thinks the bad guys need to be punished in such a way that the good fans do not have to suffer by unpopular measures of the mayor. The ADO management is very happy with this approach. Afterwards, matches are evaluated by the executive level and incidents are discussed in the security triangle. Two times a year there is a strategic meeting on the board level. Also, ADO has good relationship with the ADO fan association, with whom they regularly talk about measures during matches. The reputation of ADO fans was not good, but has been improved since they moved to their current stadium some years ago. Some struggles with extreme fans remain, and therefore both ADO and the government are keen on personal punishments like stadium bans (Appendix A1 & B).

The notification of the KNVB to inform the municipality on police strikes during the ADO-PSV game, was received relatively late, on Thursday August 6, and the final decision to cancel the match early on Friday August 7. The match ADO-PSV, to be played Saturday August 8, was deemed an A-B risk game, which means there is no to low risk. The mayor made a decision before the situation was discussed in the security triangle, later that Friday. Apparently, the concerned policy advisor experienced pressure by the KNVB to make a quick decision. There was also another factor that played a role in the early decision: “… mayor van Aartsen would leave the country for a week that weekend and did not want to bother his deputy mayor, Ingrid van Engelshoven, with this decision” (Appendix A1). The mayor also did not let his decision get influenced by decisions made in other cities. There has been contact with ADO as well, but neither party tried to influence each other into a certain direction of decision. ADO

(24)

23

wanted to continue the match, but this was more based on commercial concerns; moving a match to another date would be very expensive for ADO (Appendix A1 & B).

In the end, the main reasons for cancelling the match was the advice by the police leadership to do so:

It was still the opening match of the season against last season’s champion. Also, the mayor would not be there that weekend; 90% chance that everything would turn out okay, but the decision also has to be prepared for that last 10%, and that is the responsibility of the mayor (Appendix A1)

The direct reasons for cancelling the match are also discussed in answers to questions asked by the municipality council. Here it is said that the police strike has been “the only reason” to cancel the match, and there is also the remark that not all police tasks can be handled by private security and stewards (Gemeente Den Haag, 2015). The policy advisor also said that the atmosphere during matches is very dependent on the way the match unfolds (Appendix A1).

The most eye-catching response of this decision came from the fans. Because their match was cancelled, the ADO association organized a protest in the center of The Hague during the time the original match was planned. Their motto was “No match for us, no strike for you”. It worked out the way they planned. Because of the protests, police had to deploy towards the center to secure the premises. The police unions would only ignore deployment needed for the football match, so in this case they had to move out towards the center to secure order. Although the fan association made a deal with the municipality to stay on Square, a main street in The Hague, the fans went on marching in the city anyway, also threatening journalists of the local news outlet Omroep West. The national broadcast station NOS also hired private security, just in case. Also, the bars and restaurants at Square had to prepare in case the protest turned grim (Broos & Roos, 2015). There was a fear amongst police and municipality that the fans would march towards a police office, but in the end they returned to Square before escalation (Modderkolk, 2015).

AZ-Ajax

The municipality of Alkmaar also works with the ABC risk profile for assessing risk for matches of their professional football club AZ. C-matches are Ajax and Feyenoord. B-matches are PSV, FC Twente, ADO Den Haag and FC Utrecht. The 11 other clubs are A rated matches. The opinion of mayor Piet Bruinooge is:

(25)

24 …“football is a party”. Security and law and order is the responsibility of the mayor, but the deployment of security for a BVO is of such a structural level, that a very good steward organization has been developed to control and secure the AFAS stadium (stadium of AZ, red.) and its surroundings. Therefore the mayor leaves a lot of control and security to the BVO itself. It is a goal of the mayor to reduce the police deployment during professional football matches to a minimum (Appendix A2).

The notification of the upcoming strikes came in about two weeks before the match. The mayor saw the police strike as an ideal scenario to test whether a football match could safely continue without the use of police. He has asked the football clubs, AZ and Ajax, about their opinions on the matter. Both clubs wanted to continue the match as well. The mayor then asked for an action-plan before Thursday August 6, 17.00, wherein both clubs had to address their extra measures for securing the match without police. The mayor asked maximal deployment by both clubs: extra stewards and deals with the fan associations. To the fans the mayor said: “You want me to allow alcohol in stadiums, so now you can show you handle responsibility”. Although AZ did stated they wanted to continue the match, they possibly hoped for a cancellation; they had been playing midweek in the Europa League. The police leadership had no involvement. They felt the strike was of their unions, and they had no intention to join the debate on contract negotiations. They let the mayor make his own judgement. There was no consultation within the security triangle.

The plan was brought in on time and the mayor was satisfied enough, and decided to continue the match. In the end, Ajax came up with 25 stewards instead of ten and AZ doubled their normal amount of stewards. There were also extra traffic controllers during arrival and departure of the fans. Also, voluntarily and undercover police officers were in the stadium as well (Modderkolk, 2015). There have been no incidents (Appendix A2). The mayor, however, does realize he is not the mayor of Rotterdam or Utrecht, where fans have a bad reputation. He implies here that his football club has a less extreme fan base, and also a smaller city:

So if I would’ve let ADO-PSV continue if I were the mayor of The Hague I can’t really tell. I’m not in that position. But I would’ve let AZ here play against any club, including ADO. I would only have doubts if Feyenoord was our opponent. Feyenoord in Noord-Holland (the province, red.) that is always a difficult situation (Schildkamp, Voetbal zonder politie: het kan toch, 2015).

Also, the match was won by Ajax with 0-3, so the Ajax fans had nothing to be angry about (Schildkamp, Voetbal zonder politie: het kan toch, 2015). AZ fans have a good reputation: “there is always a good atmosphere in the stadium” (Appendix A2).

(26)

25

The reactions after the match were quite positive. A lot of newspapers published articles with interviews with the mayor after the successfully played match. The only stakeholders involved who were not happy were the police unions. They felt the decision was irresponsible, pointing out that it would have been the responsibility of the mayor if things took a wrong turn (Appendix A2). The KNVB was very enthusiastic. In a letter to the mayor they applaud his decision, and also state they want to move even further in bringing down police deployment during professional football matches:

We want to use the matches that continued despite the police strike, to investigate together with the ministry of security and justice and police to what extent it is possible to structurally have less police deployment in the future. (…) There was a need of trust in each other, collaboration with all involved parties, including fans, but above all courage. You (the mayor, red.) have given a nice example (Hoogendoorn, 2015).

However, AZ director Robert Eenhoorn has stated that the next match will be played with police deployment again: “Security in the stadium is mainly done by us with stewards. It is not our decision to evaluate whether police is necessary or not” (Fontein & Van Gruijthuijsen, 2015).

Mayor Piet Bruinooge is of the Christian-democratic party CDA, slightly to the right of the political spectrum. He is a great proponent of giving more responsibility to citizens and organizations. He feels that they are much more resilient and self-sufficient than the government is often inclined to think. Another example of action in this line of thought was with a discovery of asbestos in the city center. Normal routines require to evacuate a certain area surrounding the find. However, the mayor decided to only inform the residents about the situation, and the accompanying risks; he let them decide for themselves whether to evacuate or not (Appendix A2).

NEC-Excelsior

The normal procedure in assessing risk of a football match in Nijmegen is as follows. Usually, information about certain threats is given by the police. That information then needs to be processed. The information is then discussed within the security triangle; the risk level is assessed using the ABC classification, and then an advice for the mayor is formed. There has not been a case in Nijmegen in which the participants of the triangle, police, public prosecutor, municipality, did not agree: “I cannot remember we were ever in a disagreement during the security triangle meetings” (Appendix A3).

(27)

26

The notification about the police strike was received Wednesday August 5. Both the mayor, Hubert Bruls, and his first security advisor were on leave for summer holidays. The decision had to made by deputy-mayor Ben van Hees of the local party, and he was advised by a replacement security advisor, who’s main activities normally were social security, and not law and order. The police leadership has advised against continuation of the match, not only because of the strikes, but also overall low availability of police forces due to the summer holiday period. So although the match had an low-risk A profile, the replacements decided to cancel the match. Another matter that played a role was the reputation of the NEC fans, which is not good. At the time, the fans were also in conflict with the NEC board, so tensions were there (Appendix A3).

NEC wanted to continue the match, but has made no offer on extra measures. Also, during the security quadrilateral meeting on Thursday August 6, the NEC representative made no efforts to use the meeting to plead for continuation. The decision to cancel the match was made at the end of this meeting, even though they knew that fans had announced protests if the match would be cancelled. However, the police strike focused only on the football match, so police would be available if that protest would escalate (Gemeente Nijmegen, 2015). Deputy-mayor van Hees said he feels that football without standard police deployment is still impossible due to the incalculable nature of the NEC fans (Graat, 2015).The fact that both the mayor and the first security advisor were on leave made the option for cancelling the match even more compelling: “You really have to see the decision in context: everybody was on vacation and we then have a deputy-mayor that does not want to take this risk” (Appendix A3).

The municipality experienced a lot of reactions to the decision, by local media and by opposition parties in the municipal council. The protest by the NEC fans led to some incidents in the city center. The protests were not organized by the fan associations; they were spontaneous (Van Ginneken, 2015). Police deployment was necessary to disband the protesting fans. Some newspaper articles were critical. To cite one: “Did Hubert Bruls really think a van with Excelsior fans would stir up the city of Nijmegen” (Driessen, 2015).

Willem II-Vitesse

Willem II is the football club of Tilburg. The deputy-mayor of Tilburg initially cancelled the match Willem II-Vitesse, because of the security concerns by the absence of police (Gemeente Tilburg, 2015). The mayor, Peter Noordanus of the Labour party (PvdA), and the deputy-mayor, Berend de Vries of the Democrats (D66), both played a role here. The mayor was on leave for holiday during the week of the match and the ultimate decision, but played a vital part in

(28)

27

meetings with Willem II and its supporters in the week before. Although the deputy-mayor made the final decision, the groundwork for that decision was laid by the mayor (Appendix A4).

The match had an B-risk profile. The security triangle was of the opinion that organizing a football match is a delicate interplay of different actors, and with the absence of one important actor, the police, the football match could not continue. Also, there were concerns of stepping on the toes of the police. After all, it was their strike and the mayor was afraid that the police would be agitated if he ignored their strike. There is extra sensitivity in this issue, because the mayor of Tilburg is also the head of region that the municipality of Tilburg is part of. So the mayor does not only represent the city but also the region as a whole. So, during that meeting of the triangle in the last week of July, the mayor decided to cancel the match (Appendix A4).

Although it was decided that the match would not continue, behind the scenes, the mayor, together with the football fans and Willem II board, worked on a solution. When early August the board of Willem II made promises for extra private security and measures like extra traffic controllers, and the football fans of Willem II promised to stay quiet, the deputy-mayor, together with the supporting administrative staff, decided to continue the match. Although Willem II had to make a lot of extra costs in providing extra security, they were great advocates of continuing the match. The costs of moving a match to another date would likely to be even higher (Appendix A4). Also, in the week before the match, the Willem II fan association had been threatening with organized protests in the city center during match time, should the match have been cancelled, using the same motto as in The Hague: no match for us, no strike for you (Willem II-fans lokken politie naar centrum, 2015).

The final decision was made Thursday August 6. It was not formally decided within the security triangle, but was discussed and approved informally with the same actors, including the police leadership. When asked what the turning point of deciding to continue the match was, it became clear that it was not only the promises made by the supporters and Willem II. The mayor and his staff had been extensively following the public opinion. At first, they were afraid that continuing the match would lead to angry reactions by the police unions. But when the public opinion turned, and more media and commentators were pointing to the fact that football should be able to be played without police, the mayor and its staff dared to continue the match. Also because of the decision in Alkmaar:

We did look at the rest of the country; if we were the only ones to continue the match, then the critique from the police unions would have been greater and only focused on us. But then a sentiment developed

(29)

28 nationwide that football should be played also without police and the police unions did not seem to be bothered that their strike would be undermined, so it turned out not be an issue (Appendix A4).

The match was played without any problems. There were 80 stewards deployed and the normal amount of twenty security guards was raised to 40. Pim Dirckx of the Willem II fan association: “We didn’t miss the police. I think we showed it has been a good decision to continue the match” (Van der Lee, 2015).

(30)

29

Analysis of the Decision-making processes

In this section, each match is fitted into one or more of Allison’s models, the rational actor model, the organizational behavior model or the governmental politics model through a best-fit approach. Although it is imaginable to apply model I to every situation, because of the rationality theorem (Allison & Zelikow, 1999, p. 26), this has been hard to do so in the AZ-Ajax case. The main findings of each case are summarized in table 3.

Match Decision-maker (party) Risk profile Reputation home-fans Continue Consequence

ADO – PSV Mayor Van Aartsen (VVD)

A-B Average No

Protest by angry supporters, police deployment needed

AZ – Ajax Mayor Bruinooge (CDA)

C Good Yes None

NEC – Excelsior Deputy-mayor van Hees (local party)

A Bad No Protest by angry supporters, police deployment needed Willem II – Vitesse Mayor Noordanus (PvdA) & deputy-mayor De Vries (D66)

B Good Yes None

Table 3: Summary of findings

ADO Den Haag-PSV: A Model I and II explanation

When looking through a rational actor perspective, the city of The Hague made a rational decision when faced with the absence of police during the football match of ADO Den Haag. The unified governmental actor here is the municipality of The Hague, not only the mayor, but his whole bureaucratic apparatus and advisors combined. First, what were the goals of the city?. In this case, the wider goal is to keep law and order. The absence of police during a large event influences this goal, so governmental action is required. Second, what were the alternatives to choose from? There were two real alternatives: continue the match with extra security measures or cancel the match and find a new date for the match. Third, what were the consequences of these two alternatives? To decide the risk-level of a match, municipalities use the ABC risk profile. The match ADO-PSV had an A-B risk level profile, meaning low to some risk. The ADO fans used to have a bad reputation, but that reputation has become better in the last years. There are still some high-risk groups of fans though. Also, next to the ordinary risk assessment and fan reputation, the security advisor said that the risk might be higher because it was the

(31)

30

opening match of the season, against last season’s champions, and the mayor would be on leave that weekend. So, the consequence for the alternative of continuing the match, would be letting the match play with about 10% chance that it would escalate, according to the advisor. The consequence of cancelling the match would be finding another date for the match. Only ADO itself would undergo negative consequences of that decision, from a commercial perspective; moving a match would cost the club a lot of money due to refund of tickets and other event costs. However, there would be no costs for the municipality of The Hague. A negative consequence for the municipality would be the protests that were announced by fan associations. They threatened to occupy the Square in The Hague during the time of the match, thereby drawing police to the city center to make sure their strike would be in vain. So, the main consequence of cancelling the match would be protests by fans in the city center with possibility of escalation. So, what was the value-maximizing choice? The municipality decided to cancel the match. Even though they knew the fans would protest if they did, they had the opportunity to deploy the police in the city center if needed. That option was not available for the stadium, should the match continue and escalate. The match had an A-B profile, so some risk was expected. Taking into account that the mayor, the leader in crisis and security events, would be on leave during the match and its aftermath, the most value-maximizing choice was to cancel the match.

When looking through a model II perspective, the organizations that were involved have to be defined. First, there are the governmental organizations: the public prosecutor, the police, and the municipality, who is also the main decision-maker. Private actors involved are the BVOs: ADO and PSV. Let’s start with the main actor, the municipality. The first characteristic is the routinized way in which ADO has to apply for playing a match at the municipality. Even though every match is being discussed in the security triangle in every city, The Hague is the only city where its BVO has to make an official notification for every match they play. This can be described as an SOP (Allison & Zelikow, 1999, p. 169). This routinized behavior might also explain why ADO did not put pressure on the municipality to continue the match, even though this would cost them a lot of money. Also, the choice of cancelling the match was a risk avoiding choice:

It was still the opening match of the season against last season’s champion. Also, the mayor would not be there that weekend; 90% chance that everything would turn out okay, but the decision also has to be prepared for that last 10%, and that is the responsibility of the mayor (Appendix A1)

(32)

31

This fits into the notion that organizations are risk avoiding machines, trying to routinize practices to avoid uncertainty (Allison & Zelikow, 1999, p. 170). The problem that the municipality faced did not fit into any of the routines available, thereby becoming an uncanny problem. Problem are fitted into ordinary routines. Take for example the risk profiling system that is created with the idea of at least some availability of police. Letting a match play without police was not a routine available. Therefore, the municipality’s behavior is indeed best described at t by t-1; it would always cancel the match, because a match without police fits no existing routine (Allison & Zelikow, 1999, p. 175). Dealing with protests of fans as a consequence of cancelling the match, however, did fit in existing routines. The municipality often talks with ADO and its fan associations, so agreements were made with the fan associations regarding the protests, like they would not leave the Square, and should that agreements not be held on to, the police was available to deploy in the city center, because their strike was restricted to the stadium.

AZ-Ajax: A Model III explanation

To explain the decision-making process in Alkmaar through a governmental politics model, first the players in the game have to be identified, together with the organization they belong to, and their interests. The main player in the game of law and order in a municipality is the mayor. The mayor can decide on its own whether the match should continue or not. Also, in the decision-making stage of the policy cycle, we have acknowledged that there are not many influencers left (Howlett, Ramesh, & Perl, 2009). Mayor Piet Bruinooge of Alkmaar is an advocate the least amount of police deployment as possible during matches. Instead, he rather points out the individual responsibility of citizens and organizations. This idea is something he has done before during a case of asbestos in the city center. Other players were the representatives of the football clubs AZ and Ajax. It was in both their interest to continue the match, even though there was a suspicion that AZ felt they would be better off playing the match on a later date due to their participation in the Europa League earlier that week. But that feeling was never officially expressed. The last important player here is the police, which can be separated into the police leadership and the police unions. The leadership let the mayor make his own decision; they did not want to get involved in the political debate of the contract negotiations, or step on the toes of the police unions. The police unions were against continuing the match, but had no official stance as they are not part of any of the action-channels, which brings us to the next element.

Is the ball always round? A multi-site case study of local decision-making processes in Dutch football security