(Mis-)matching framing foci: Understanding policy consensus among coastal governance frames

The push and pull between innovation and conservatism is a normal situation in Dutch coastal governance. In some cases this leads to conflicts over which policy solution to choose. This paper argues that coastal governance processes are shaped by policy frames. They influence what actors consider the policy problem, the appropriate solution, and the way to achieve this solutions. The paper features a Dutch coastal governance project called “Hondsbossche Duinen”, in which the mismatching of heterogeneous policy framing foci almost led to breakdown of the process. After a phase leading to controversy and shifts in project management, the project eventually became realised successfully. This near-breakdown and eventual success is studied from the perspective of framing foci: policy substance, policy process, and actors’ identities and relationships. In-depth interviews with policy actors allow for a nuanced analysis of those framing foci. We show that dynamics of emphasising and ignoring similarities and dissimilarities between actors’ framing foci explain the developments in the case. Emphasising dissimilarities and ignoring similarities produced controversy in the first phase of the case. In the end, emphasising similarities and the emergence of a consensus framing focus led to policy consensus.


Introduction
Implementing innovative coastal protection projects can be challenging. Coasts are often areas of bustling activity and especially in small countries such as the Netherlands, dense urbanised cores alternate with economic production space. Besides, beaches represent popular recreational destinations. Similarly there are ongoing administrative trends such as deliberative governance (Hajer, 2005), public participation (Rowe and Frewer, 2000), co-creation (Voorberg et al., 2015) and the rise of integrated management, e.g. integrated river basin management or integrated coastal zone management. As a result, decision making processes about the implementation of coastal protection projects involve an increasing number and diversity of stakeholders with heterogeneous interests, task assignments and approaches towards coastal governance. However, coming to terms with conflicting coastal governance approaches to achieve shared binding decisions is at the heart of the coastal governance process (cf. Kuhlmann, 2001). For example, in the past, coastal managers often relied on "hard" coastal defences to protect these coastal areas against storms andrecentlyclimate-change-induced sea-level rise. The Hondsbossche Duinen reflects such a coastal area where hard coastal defence, a 5.5 km seawall, was created in the 19th century already. Some coastal governance experts challenge this "hard" practice. They believe that more flexible, multifunctional, and ecologically integrated coastal protection solutions may be more appropriate (Van Slobbe et al., 2013). While innovation in coastal protection technology was crucial for the survival of the Netherlands as a country (cf. Disco, 2002;Gerritsen, 2005;Meijerink, 2005;Van Koningsveld et al., 2008), their implementation is not always welcomed by coastal managers. In principle, they prefer reliable, proven technologies. Bringing these conservative and innovative positions together is currently the bread and butter of coastal governance.
Understanding the dynamics of such a governance process is a complex undertaking. Although it increases predictability, the assumption of rational or even boundedly rational behaviour is an oversimplification (Schön, 1982). Furthermore, what comes out of a policy process has frequently been explained as a result of throwing everything in a garbage can in which solutions are achieved by muddling through (Cohen et al., 1972;Lindblom, 1979). Others have conceptualised different aspects of policymakingpolicy, politics and problemsto be independent of each other most of the time (Kingdon, 2014). The role of individual actors' perspectives in the process and outcome of policy processes does not play a role in these strands of literature. One exception is the concept of special policy entrepreneurs who can crucially steer policy processes (Aukes et al., 2017;Huitema and Meijerink, 2010;Mintrom, 1997;Mintrom and Norman, 2009;Winkel and Leipold, 2015). Rather, we argue in this paper that difficulties with the implementation of innovative coastal protection projects can be understood from the perspective of actors' policy frames and policy framing. Policy frames are not only crucial in understanding positions of actors. The development of frame alignment between actors also clarifies specifically why in some cases actors converge on some coastal protection technology and not on others (cf. Snow et al., 1986). This frame alignment becomes visible through the deconstruction of policy frames into framing foci and tracing their changes over times.
We subdivide actors' policy frames into three foci, i.e. policy substance, policy process, and identities and relationships (Van Hulst and Yanow, 2014). 2 The central research question we address in this paper is: why do some coastal protection technologies and the processes by which they are chosen lead to policy consensus while others do not? To illustrate this claim we apply this perspective to the case of the Hondsbossche Duinen in North-Holland. There, acts of matching and mismatching of policy frames led subsequently to near-breakdown and success of the process. We shall see how dynamics of emphasising and ignoring similarities and differences in framing foci between couples of actors ultimately elucidate some of the coalition-making and coalition-breaking in the case.
Hondsbossche Duinen is one of ten weak spots detected during inspections of the Dutch sea defences in the early 2000's. Within the socalled "Weak Links" framework programme, these weak spots were supposed not only to be reinforced with regard to coastal safety, but also to improve spatial qualitythe "double objective". In 2004, the Dutch Ministry of Infrastructure and Environment (MinI&E) mandated the Provincial Government of North-Holland to prepare a reinforcement proposal for this weak spot. The first proposal by the provincial government was turned down by the ministry in 2006. In the process of preparing the proposal, the provincial government had not taken into account detrimental effects to the local population, neglected additional expertise from the water board and Rijkswaterstaat, as well as exceeded the budget. After delegation of the project management to the water board in 2006, it took until 2014 for the first non-experimental meganourishment scheme to be realised in the Netherlands (Fig. 1). During these eight years, the water board succeeded in bringing together actors with different opinions regarding the solution to be chosen and the process by which this solution was to be chosen. While early on in the case, the project threatened to fail, it was finally realised with a broad epistemic community of actors supporting it. This is, in a nutshell, the case in which actors needed more than a decade to fix a weak spot in the coastal defence. We will analyse this case by emphasising the policy frames of actors and their interaction.
Framing has its root in Goffman's (1974Goffman's ( /1986) work on frame analysis. After having been picked up by psychologists Kahneman and Tversky (1984), the concept gradually diversified into different social-scientific fields, such as policy sciences (Schön and Rein, 1994), including water management (Dewulf et al., 2007;Isendahl et al., 2009;Vink et al., 2013). Framing is also common in the fields of social movements (Snow et al., 1986), communication, and media sciences (Entman, 1993;Scheufele, 1999). By now, two strands of framing literature have developed characterised by different conceptual outlooks . One focuses on the interactional component of the co-construction of meaning, studying the mechanisms by which the act of framing occurs. This line of thought prefers to study interactions of actors in the field. The other, "the cognitive-representational stance on framing focuses on the way that people experience, interpret, process or represent issues, relationships and interactions" .
This article continues with a discussion of framing foci followed by the methods. Then, we present the empirical results of the case study. We discuss the main actors' framing foci and how these changed over time. The development of framing foci throughout the case allows for an understanding not only of why the project almost broke down, but also of why success became more and more likely once the water authority became project manager.

Policy framing foci
In our analysis, we focus on policy frames. We define policy frames as actors' "implicit theories of a [policy-making] situation" (cf. Nie, 2003;Van Hulst and Yanow, 2014). Thus, policy frames are representations of actors' understanding of the policy situation at hand, including a definition of the problem, possible solutions, and an idea of how to reach those solutions. In the form of individual stories (Stone, 2002), "frames (…) guide the ways [actors] perceive their social reality and (re)present it to themselves and to others; (…) they structure the ways in which segments of social reality are attended to" (Van Hulst and Yanow, 2014). Which segments of social reality stand out to actors depends on characteristics, such as education, upbringing, organisational membership, or experience. The selected segments of social reality form the basis for a definition of the policy problem at hand. In turn, the scope of acceptable policy solutions is encapsulated in this problem definition (Kurt, 2004).
A policy frame and the problem definition it contains deals with three policy "framing foci", as we call them Van Hulst and Yanow, 2014). First, the meaning actors give to the policy content is 'policy substance' (Van Hulst and Yanow, 2014). Second, actors' sense-making sense of other involved actors and how these are networked represents the framing focus of 'actors' identities and relationships' (Van Hulst and Yanow, 2014). The third policy framing focus is an actor's perception of the 'policy process', e.g. how and which actors should be involved in the process (Van Hulst and Yanow, 2014). These framing foci serve as conceptual tools to subdivide actors' frames.
Using the language of 'reframing', we can study the changes in policy framing foci. Reframing, i.e. a change in the implicit theories of the policy-making situation, "resolves the controversies that arise in policy practice" (Schön and Rein, 1994). Policy controversies occur over differences in policy framing foci, e.g. a different framing of the policy substance. Actors' policy frames converge, once tensions between policy framing foci become resolved. A mutually accepted solution, then, becomes more likely. By disassembling actors' frames into afore-mentioned framing foci over time during a policy process, we can follow the development of policy consensus or controversy in a case (cf. Schön and Rein, 1994).
Alternative concepts to account for the presence or absence of policy consensus are the concepts of "master frames" or "consensus frames". These connect otherwise divided policy frames on a more abstract level by coupling them with "widely shared terms and concepts" (Candel et al., 2014) or a "broad scope" (Benford and Snow, 2000). Frames that are brought together by master frames or consensus frames share "equifinal meaning" (Donnellon et al., 1986), i.e. they are "interpretations that are dissimilar but that have similar behavioural implications". Linking framing to action, consensus frames or master frames can lead to the same results. For example, in situations where coastal actors' problem definitions differ, there may be coastal protection technologies solving all defined problems without necessarily converging on the problem definition itself. 3 We argue that these alternative concepts are not exclusively useful. We contend that an approach centring on the share of the overlap between framesas consensus frames and master frames dofails to increase our understanding of why projects threaten to break down. On the contrary, we will show that the different framing foci have to be taken into account. We shall see that even a large overlap in framing foci may still put actors in opposing camps and that minimal frame overlaps may also lead to policy consensus.

Methods
A case study approach is suitable for studying framing for three reasons (Yin, 2018). First, studying the dynamics of policy consensus or breakdown from a framing perspective requires nuanced knowledge about the context of the policy situation in question (Flyvbjerg, 2006). Second, the argumentative dynamics occurring between actors in a policy situation are suitably understood by having those actors describe and explain the policy situation to the researcher. Third, as (coastal) governance does not follow the predictability of rational behaviour, the diversity of actors, interests and policy networks involved turns it into a complex social phenomenon. Furthermore, the Hondsbossche Duinen case was nearly unique at the time of realisation with only one other (even slightly different) instance in the form of the Sand Engine near The Hague (Aukes et al., 2017;Bontje and Slinger, 2017). Taking the Zeitgeist into account, it can be characterised as an extreme case (Flyvbjerg, 2006).

Access and data generation
We gained access to the case through contract research for the Dutch public works agency Rijkswaterstaat. From there, snowball sampling led us to interview eleven policy-relevant respondents (Noy, 2008). All respondents' organisations are to a certain extent involved in infrastructural decision-making in the Netherlands (see Table 1). 4 We 3 Cf. Kingdon's multiple streams approach posing the relative disconnectedness of policy solutions from problems and the need to bring these together at the right moment (Kingdon, 2014). Kingdon's approach differs from the policy frames approach taken here in that it disavows the teleology of policy problems that the close connection of policy problem and solution scope in a policy frame represents. 4 We aggregated frames to the organizational level.
attempted to gain an all-around view of the case to cover not only a large variety of organisations, but also the wide variety of meanings they attributed to the case. In a metaphorical sense, this leads to an exposure map (Schwartz-Shea and Yanow, 2012). Fig. 2 shows such an abstract exposure map, in which the central triangle denotes the case and the arrows pointing at it from different angles are the wide variety of meanings elucidated by interviews or from other sources. Each interview exhibits a unique view of the case depending on respondents location on the map. Once additional interviews did not yield substantially new perspectives anymore, the point of saturation was reached and we ceased interviewing. Studying the case retrospectively, we had to rely on semi-structured interviews and policy documents from the case (Schwartz-Shea and Yanow, 2012). Policy framing foci can be derived from conversational, semi-structured interviews, because these reveal the policy process from the point of view of the actors. The interview guideline included questions about the respondent's experience and daily work, expectations about the project, important controversies and collaboration among actors. For the contract research, the guideline also touched upon the construction task, design and organizational tasks and responsibilities. 5 However, we also included these latter topics in the analysis, as they constitute relevant contextual information. We recorded and transcribed the interviews. Transcripts were returned to respondents for fact-checking, if requested. 6 The interviews took place between January and September 2014.
Besides semi-structured interviews we further detailed our exposure by reviewing relevant policy documents (cf. Böhm, 2005). We triangulated within the interviews and with consulted additional project documentation, background literature and parliamentary reports (Lincoln and Guba, 1985).

Data analysis
For the analysis of the interviews, we used the policy framing foci as an initial code list (Miles and Huberman, 1994). The interview transcripts were systematically coded using ATLAS.ti. 7 We looked for extracts dealing with the policy substance, policy process, and actors' identities and relationships. Besides plain descriptions, stylistic devices such as metaphors served as signposts for a framing focus. We distinguished between extracts explaining actors' own framing focus and extracts explaining other actors' framing focus. Reported framing foci were used as triangulation. Table 2 shows coding examples, explaining how we linked parts of the transcripts to policy framing foci. Often, extracts referred to more than one framing focus at the same time. Such extracts were double-coded and included for both framing foci. This elicited actors' policy frames, argumentative dynamics and socio-political networks in the case (Noy, 2008).
After coding, all codes were pooled per respondent and framing focus to develop a label for the framing focus. We discerned variation in framing focus among respondents in the case. Nonetheless, respondents from the same organisations had roughly the same framing foci.
We assumed that organisations in the role of project manager have strong influence on which framing foci dominate. Hence, changes in which organization leads the project also mean shifts in the dominance of framing foci. Thus, we divided the case into three phases marked by changes in the organisation in the lead of the project. In the first phase, the Provincial Government of North-Holland led the project, followed by the water board in phase two, and the water board together with Rijkswaterstaat in phase three. Furthermore, distinguishing project phases as such led to a more fine-grained overview of framing focus changes.

Analysis: Project managers and their framing foci
Different project managers have their own approaches to managing  projects. This is especially the case if it concerns the various organisations responsible for coastal governance in the Netherlands. Consequentially, changes in who manages a project also mean shifts in the framing constellation of the policy situation. For example, certain policy substances may then be preferred over others that were dominant under previous project managers. This may mean shifts in coalitions of organisations working towards one coastal protection solution or another. Hence, to trace the evolution of the framing foci of involved organisations, we divided the case into three phases of the project marked by transitions of the formal project manager role from one organisation to another. In the first phase, the Provincial Government of North-Holland led the project, followed by the water board in phase two, and the water board together with Rijkswaterstaat in phase three. In addition, distinguishing these different phases enabled a fine-grained overview of framing focus changes.

Project phase 1: Provincial government in charge
As afore-mentioned, the seawall called the Hondsbossche en Pettemer zeewering had to be repaired. It had to be prepared for strong storms while contributing to spatial quality, which was termed a "double objective". This double objective was formalised in an agreement of all involved actors in 2004. The first proposal by the provincial government in 2006 meant the partial destruction of the near-by village. This proposal not only led to unrest among the local population, but also to its rejection by the responsible minister of Infrastructure and the Environment. Additionally, it was judged too expensive and designed with too little public participation.
In the same year, Rijkswaterstaat introduced a new national flood protection programme. In its role as project evaluator, this programme saw heightening of the seawall as the only acceptable solution: "Primarily, at the flood protection programme we said: 'sand is (…) efficient, but not cost-effective, functional, and robust'. Because that is really only the seawall" (Project manager regional Rijkswaterstaat). "Functional and robust" is somewhat of a motto at Rijkswaterstaat. Hence, any solution that also included other objectives did not align with their view. In addition, Rijkswaterstaat continuously had to justify its passive evaluative role vis-á-vis municipalities who kept appealing for help against the provincial government's rigorous plan (project manager regional Rijkswaterstaat). At the same time, Rijkswaterstaat disapproved of the provincial government's top-down approach, which diverged from the participatory national policy-making tradition (project manager regional Rijkswaterstaat).
Flood protection is not a municipal task in the Dutch coastal protection system. Nevertheless, flood safety is of course a basic condition for living at the sea and hence an important topic for municipalities. Additionally, the concerning region in North-Holland is perceived as economically weak. This led the municipality to lobby for an economically beneficial solution and to point out the potentially disastrous consequences of following the provincial proposal. To achieve this, it used its formal political means as well as informal ways of lobbying. It advocated what was in its eyes the best solution for the region: sandy coastal protection with economic benefits. Despite clearly taking a position with regard to the policy substance, the municipality was also aware of its limited influence in the policy process.

Project phase 2: Water board opens up
In early 2007, the provincial government delegated the project to the water board. The official reason for this was the reformulation of the hydraulic conditions that the seawall needed to meet. Some interpreted this as shirking responsibility to cover up the emerging technical infeasibility of the provincial proposal (policy advisor MinI&E). In this way, the provincial government could evade public backlash by "escap [ing] with the newest hydraulic conditions" (policy advisor MinI&E).
The water board began its project management task with a pragmatic perspective: "If there is a seawall somewhere, (…) and it has to be reinforced, you think: 'I am going to reinforce that seawall'" (water board executive). Including hard seawall reinforcement in its initial considerations "did not make [the water board] popular in the region" (water board executive). In the end, the water board came to the same conclusions as the previous provincial proposal: the additional height necessary to make the seawall safe would destroy part of the village. The water board's resulting conclusion was different, though: "If you talk about doubling the size of a seawall, then that is a new seawall from my point of view. So, if you talk about a new seawall, you may also think of new solutions" (Water board executive). Thus, sandy and hybrid solutions suddenly became a viable alternative. The water board involved civil society and Rijkswaterstaat in the process. Future orientation, flexibility and landscape fit were important arguments for the water board to open up towards sandy solutions. Benefits for the local population were not mentioned as such. Nevertheless, the fact that destruction of a part of the village was off the table carried away appreciation with the local population. This change of perspective was referred to as a "lifesaver" for the project (Policy advisor MinI&E).
In the meantime, the provincial government began lobbying for sandy solutions, too. The provincial official claims the provincial government advocated a sandy solution all along. Success of the sandy solution would even be due to the provincial government's commitment. At the beginning, the water board was the opponent, because it "was still rather conservative at that moment" (provincial official).
The afore-mentioned double objective sparked a lasting conflict between the provincial government and the ministry over the allocation of costs which erupted in this phase. The provincial government insisted that the ministry should pay for the complete project based on the double-objective agreement. Conversely, the ministry stressed the clearcut task division between ministry and provincial governmentflood safety and spatial quality, respectively. This task division would imply the distribution of costs to both organisations. The provincial government still hoped that Rijkswaterstaat agreed with its own interpretation of the double-objective consensus. It hoped for the use of available national funds for both safety and spatial quality. This hope turned out to be in vain when Rijkswaterstaat proclaimed to invest in safety only. It expected the provincial government and water board to pay for nonsafety-related elements. In the end, the provincial government conformed and focused on spatial quality measures, which were also expected to be cheaper. By 2008, the provincial government gravitated heavily towards sandy solutions.

Project phase 3: Cooperation saves the day
This phase begins in 2009 with the water board presenting its preferred solutiona sandy coastal protectionas a result of the environmental impact assessment. A new water board director, who previously worked at Rijkswaterstaat, stepped up the same year. His avid endorsement of sandy solutions, only reinforced the water board's preference.
Rijkswaterstaat was by then the only remaining actor with substantive objections to the preferred solution. At the national flood protection programme there was disbelief about the water board's additional claim that the sandy solution would be less expensive than the hard reinforcement. This position triggered staff of the programme to rerun the cost calculations. Although Rijkswaterstaat insisted on reconsidering the preferred solution, the water board carried on. In this situation, the water board director's experience with Rijkswaterstaat's organisational culture and language was of no help. The difference of opinion about the preferred solution meant that cooperation was still difficult.
Regardless of Rijkswaterstaat's ongoing opposition, the water board's new solution satisfied the ministry's demands. In 2010, the ministry consented to the sandy preferred solution, but added two financial conditions. It set a budget cap of €250 000 000 and demanded the inclusion of 20 years of maintenance in the contract. Especially the latter led to renewed conflict between the water board and Rijkswaterstaat. Tasking the water board with organising the maintenance of the project challenged Rijkswaterstaat's legally enshrined responsibilities in that field. Additionally, Rijkswaterstaat questioned the water board's experience and capability to manage a project of this size and complexity on its own. Consequentially, Rijkswaterstaat demanded more influence in the project. This positioning led to resistance bordering on annoyance on the part of the water board (water board executive; project manager regional Rijkswaterstaat). The turning point in this situation was a high-level meeting at the end of 2011. During this meeting Rijkswaterstaat and the water board agreed to finalise the project together. Both sides termed this "big change" the "best of both worlds" to "carry the risk together" (stakeholder manager water board; project manager regional Rijkswaterstaat).
The tendering procedure concludes this project phase. For this tender, the project management team only set framework conditions such as a maximum budget of €170 000 000 and "economically and societally sound investments". Contractors could use creativity, expertise, and their own ideas for their plans. 8 This process incentivised contractors to include as many ideas as possible at as low a cost as possible. With €140 000 000, this tendering approach resulted in project costs of merely 56% of the larger ministerial budget cap. The ministry very much appreciated this achievement. The final plan features a sandy mega-nourishment of about 30 000 000 m 3 (Fig. 1). The existing seawall loses its function, but stays intact as a cultural-historic landmark.

Types of framing foci
The case of the Hondsbossche Duinen shows different positions for each of the framing foci. Policy substance emerged in three positions: safety, regional development, and multifunctionality. We categorised policy substance as "safety", if it revolved around cost-effective coastal safety solutions. "Regional development" related to livelihoods of the local population and economic development. Policy solutions that linked both coastal safety, other functions, and intended to maximise their synergy were termed "multifunctionality".
For the other two types of framing foci we found two positions. The policy process was either envisioned in an inclusive, participatory way, which we termed "process-oriented". In others, it entailed an exclusive perspective termed "task-oriented", where involving other actors was seen as detrimental to quick realisation. Furthermore, there were two positions for actors' identities and relationships. These positions included "empathy" and "indifference" concerning other actors. Frames labelled with "Empathy" were aware of and interested in other actors. "Indifference" was used, if no signs of interest in other actors or relationships were observable. It turns out that framing foci of policy process and actors' identities and relationships are related. For example, the analysis reveals that a task-oriented framing of the policy process often relates to an indifference with regard to other actors' identities and relationships. On the other hand, there is an interest in other actors' identities and relationships among actors with a more process-oriented framing of the policy process. The results are summarised in Table 3 and described in detail below.

Phase 1
We find a scattered framing foci situation in the first project phase (Table 3). In this phase, most framing interactions occurred between the provincial government and the ministry. Other actors were either not included at all (water board) or did not have a formal say in the process (Rijkswaterstaat and municipalities). Although the actors were off to a flying start with the collective agreement on the double objective, this solidarity is not reflected in the following interactions between actors. The provincial government's policy substance in the first phase was based on safety considerations. Unfortunately, other actors perceived the proposed solution as detrimental for the already alarming economic situation in the region. It was unclear, how the proposed solution would contribute to the double objective and the provincial government's general interests (provincial official). In addition, the provincial government followed a top-down and exclusive policy process. According to the ministry, it neglected other actors, such as municipalities, Rijkswaterstaat, or the water board (policy advisor MinI&E). This task-oriented policy process focus also limited the scope of the policy substance. Excluding other actors from the process prevented the provincial government from confronting other policy substance foci. According to Rijkswaterstaat, the provincial government overrated its own capacity to realise a project in a policy domain which it is usually not responsible for. Additionally, it seems to have misinterpreted the framing identities of other actors. Relationships with other actors worsened due to this purposeful self-isolation by the provincial government.
The policy substance focus of the ministry is vague throughout the first phase. It disapproved of the solution advocated by the provincial government, due to its intrusiveness. The interviews did not elucidate whether this framing focus came from a disagreement with the exclusion of other actors (policy process) or a generic support of multifunctional coastal governance solutions (policy substance). However, the latter is probable due to enthusiastic ministerial support of multifunctional solutions in other projects (Aukes et al., 2017). The ministry also advocated an open and inclusive policy process. It was dissatisfied with the way in which the provincial government dealt with the project (policy advisor MinI&E). The interest in including other actors and hearing their frame renders the ministry an empathic actor. Nonetheless, the ministry's role remained in the background.
Rijkswaterstaat thought an innovative, sandy solution would be efficient. However, it did not match the other standard criterion of robustness (see 4.1). Thus, Rijkswaterstaat advocated a traditional safety solution as policy substance focus. It did not matter how safety This approach is uncommon insofar as usually the specific parameters of the project are laid out by the project management. In those cases the lowest price is often the decisive argument for selecting the contractor.
was achieved, as long as the chosen solution would fit its financial and safety performance criteria. Rijkswaterstaat found the provincial government's top-down policy process approach inappropriate. Furthermore, it had a clear idea of the distribution of tasks and responsibilities in the policy process. Towards the municipalities, Rijkswaterstaat felt it had to justify its passive role in the process continually. The water board was quite clear about its position in the project. It did not identify as an actor in the project in this phase. As a result, it did not yet have relationships with other actors, nor did it take position concerning policy substance or the way in which to approach the process.
The municipality also opposed the provincial proposal due to its sole focus on safety and the partial destruction of the village. It advocated sandy solutions in its policy substance framing focus. These were said to be most advantageous for economic development. While aware of its limited influence in the project, the municipality reached out to its political relationships as the spokesperson for the local population. For this, it used the formal and informal tools at its disposition. The municipality realised it had a small role in coastal governance, but tried to make the most of the options it had.

Phase 2
In phase 2, with the water board as project manager, the framing foci experienced considerable reframing (Table 3). It argued that inclusion of all possible types of coastal governance solutionstraditional, sandy, and hybrid solutionswas a reflection of thorough project management. This made its original policy substance a combination of the safety and multifunctionality framing focus. The water board also appreciated the participation of other actors. Later on, concluding that "seawall reinforcement" would equal "new seawall construction" mirrors the reframing of the policy substance on part of the water board. Additionally, looking for new solutions warranted a search for a coastal governance solution which would also respect other actors' needs and interests. A multifunctional solution was more in line with the prevailing double objective. It also led to improved relationships with the municipality and local population. Overall, the openness with which the water board approached policy substance and policy process considerably reduced the tensions among the involved actors.
The provincial government reframed to a multifunctional policy substance framing focus in this phase. It began criticising actors who did not explicitly advocate sandy solutions. This included the water board with its broad view on possible coastal governance solutions. The provincial government intensified its focus on the spatial quality aspect of the project. Thereby, it joined the municipalities in the framing focus on maximising the economic and spatial benefits of the projectregional development (Table 3). This improved the previously severed relationship between provincial government and local population.
Rijkswaterstaat did not incur substantial reframing in its framing foci. It remained sceptical of sandy solutions. When the water board's outlook on policy substance moved towards sandy solutions, Rijkswaterstaat urged to reconsider: it would be too uncertain that sandy solutions could be cost-effective investments. Rijkswaterstaat risked further impairment of its relationship with the local population, but policy substance seemed to have precedence over actors' identities and relationships in this case.
Regional development remained the municipality's policy substance focus. It regarded the water board's policy substance with ambivalence. On the one hand, it welcomed the water board's inclusion of sandy solutions. This increased the chances of selecting a minimally destructive solution. On the other hand, there was still a possibility that a non-sandy solution would be chosen. The municipality stayed alert and kept making use of its participatory opportunities to try and influence the policy process. In this phase, its relationships with both the water board and the provincial government improved.
Regardless of its reduced role, the ministry thought the openness with which the water board approached the project essential to its success. It was glad to see a new project manager with policy substance and policy process framing foci similar to its own.

Phase 3
In the third phase, few framing foci changes happened. However, an important organisational change occurred: Rijkswaterstaat joined the project management. At that point, it remained the only actor without sandy solutions in its policy substance framing focus. Nevertheless, the ministry's conditional acceptance of the project proposal influenced this. It triggered reframing of Rijkswaterstaat's policy substance: sandy coastal governance solutions were accepted as a valid safety alternative. Now it saw its identity threatened by the water board. Rijkswaterstaat feared the proposed takeover of its coastal maintenance responsibilities. Joining the project management reduced this threat. The choice of tender reflects Rijkswaterstaat's and water board's process focus with regard to the policy process.
Meanwhile, the sandy solution become undisputed at the water board. Before, the water board's framing focus was policy process based ("for good project management, you need to include all possible options"). With the new director, the framing focus was grounded more strongly in the policy substance ("sandy solutions are the preferred option"). Moreover, the water board's framing of the policy substance reframed the water board's identity as seen by provincial government and municipalities to "ally". Although Rijkswaterstaat challenged the water board's leading role in the third phase, it was now easier for the water board to understand Rijkswaterstaat's identity due to the new director. This improved their relationship.
The provincial government now focused completely on the orphaned spatial quality part of the double objective. Because of reframing the policy substance, the provincial government retreated from the safety project. By now, the provincial government saw itself more as a spokesperson for the interests of the local population. In this phase of the project, there were officially two sub-projects dealing with coastal safety and spatial quality separately. The latter was in the hands of the provincial government, municipalities, and nature organisations.

(Mis-)matching framing foci
Throughout the case actors reframed their framing foci. Failure or success in the project phases depended on mismatching or matching framing foci, as a comparison of framing foci constellations reveals (Table 3; cf. Entman, 1993).
The provincial government as the project manager looked for a coastal governance solution promoting safety in the first project phase. Its search can be characterised by a task focus and indifference for other actors' positions. Unfortunately, this was an isolated stance. Rijkswaterstaat was the actor with the most similar framing foci, converging on the policy substance and actors' identities and relationships. However, its process orientation was decisive for Rijkswaterstaat to oppose the provincial government's proposal in 2006. Rijkswaterstaat prioritised one framing focus over others. Other actors were either not involved, such as the water board, or differed from the provincial government on all framing foci. The municipality's opposing stance on policy substance and the policy process may have triggered the rejection of the provincial government's proposal by the ministry. As the second-most important actor for the course of the project in the first phase, the ministry disapproved of the task-orientation of the provincial government and the resulting exclusion of other ideas about policy substance in the case. This difference in framing foci explains the negative reception of the provincial proposal marking the end of the first phase.
Actors converged towards a state of consensus in the second phase. A main factor is the entrance of the water board as project manager, which had a much more open-minded framing with regards to policy substance and policy process. This increased other actors' willingness to cooperate.
Especially the initial undecidedness of the water board regarding the policy substance kept the other actors on board. On the one hand, Rijkswaterstaat expected that safety would be safeguarded by the water board. On the other hand, other actors advocating multifunctionality and regional development could still hope for a project solution in their favour. The water board also welcomed participation much more and was concerned to find a solution that would benefit all.
The final framing setting, after which the project was realised, shows a number of partial overlaps among framing foci. The water board's multifunctional solution was acceptable for the provincial government and municipalities, because it promised new space for recreation and potential for the tourist industry. Meanwhile, Rijkswaterstaat also reframed the policy substance to the multifunctional solution, owing to its membership in project management, the strong preference of the ministry for multifunctional solutions and the increasing certainty that the safety performance was indeed warranted. The framing differences in the other two domains weighed less strongly for Rijkswaterstaat, because its main aim was to guarantee coastal safety, regardless how this state was achieved. This is different from the first phase, where the provincial government's task orientation was a barrier for Rijkswaterstaat to support the proposal. The water board's approach to the policy process as well as its awareness of the interests and needs of other actors, made for a smoother decision-making process towards realisation of the project.

Emphasising (dis-)similarity
Throughout the case we see actors agreeing and disagreeing with each other over one framing focus or another. However, the project phases show different patterns with regard to the emphasis on similar or dissimilar framing foci (Table 4). We understand the mechanisms of process breakdown or success as a result of these emphasising motions.
In phase 1, there were largely two coalitions. First, there was the provincial government. Other actors who were against the provincial government's policy substance and policy process formed the second coalition. This latter, larger coalition was not a unified whole, though, with differences in the policy substance and the interest in other actors and relationships with them. In this phase, the provincial government was the only actor focused on mono-functional safety solutions, without including other actors in the decision-making process or interest in their positions. The other actors all framed a broad, inclusive policy process necessary. Most actors also disagreed on the policy substance and were empathic of other actors. However, Rijkswaterstaat overlapped with the provincial government in two of the three framing foci. Both wanted a coastal governance solution dealing with the safety problem and were not very interested in other actors. But the difference in policy process framing focus between the two turned out to be crucial for Rijkswaterstaat's opposition against the provincial government. Agreeing on the policy substance was not enough for Rijkswaterstaat to condone the policy process framing focus of the provincial government. Rijkswaterstaatand other actorsemphasised the dissimilarity between the provincial government's and their own framing foci (Table 4). For the ministry, with framing foci centring on innovative, multifunctional solutions, and an inclusive process, the differences with the provincial government were too large. The framing foci constellations of the ministry and Rijkswaterstaat also reveal another mechanism. In their case, it was enough to agree on the policy process framing focus to take the same position. In this phase, two emphasising mechanisms occurred. First, emphasising the dissimilarity of two actors in their policy process, but ignoring the similarity in the other two framing foci, defined an opposing relationship. Second, emphasising the similarity of two actors in the policy process, but ignoring the dissimilarity in the other framing foci, defined a coalescing relationship.
In the following two phases, actors began to emphasise similarities more than dissimilarities. Most importantly, the new project manager, the water board, had a more connective policy substance focusmultifunctionality. There was no necessity for other actors to reframe their policy substance framing focus, because it was broad enough to accommodate other policy substances. The multifunctionality policy substance acted as a consensus frame on the framing foci level, by emphasising the similarities in the policy substance (Table 4). Another factor was that other dissimilarities between actors in policy process and actors' identities and relationships decreased. It was less necessary for actors to emphasise or ignore agreements or disagreements, because they were heading towards a total consensus. The only disagreement in policy substance was reconciled with a consensus framing focus.

Conclusion
In this paper, we set out to show what policy frames and policy framing can contribute to our understanding of the complex dynamics of coastal governance leading to policy consensus or policy controversy. We argued that it is appropriate to break down policy frames into their framing foci. The analysis has shown that a focus on either of the three framing focipolicy substance, policy process and actors' identities and relationshipsalone would not have increased our understanding of why the sandy solution was chosen in the Hondsbossche Duinen case. Instead, looking at all three framing foci and their interplay is more fruitful. Understanding how actors in a coastal governance case actively and subconsciously match or mismatch their framing foci explains certain frictions in the process. This had not been possible without looking at the way framing foci change over time. Breaking up the case into three phases enabled us to see this. Framing foci are thus useful to understand actors' policy frames at a certain point in time. To understand how policy frames contribute to policy controversies and their resolution the dynamic perspective is necessary.
In the coastal governance case of the Hondsbossche Duinen, we found the policy substance framing focus of multifunctionality to Table 4 Dynamics of emphasising and ignoring framing foci.
become dominant, because it accommodates other policy substance positions. While this case illustrates that complete agreement in all framing foci is not necessary for actors to reach consensus, it also becomes clear that consensus frames or master frames do not tell the whole story either. It may be enough for actors to differ in one of the framing foci to increase the potential for conflict in a process. The further development of a decision-making processtowards consensus or towards controversythen depends on which framing foci are emphasised and which are ignored. Multifunctionality has proven a versatile framing focus in the case presented here. It is also an example of the interconnectedness of the framing foci. The broad view on possible functions connected with multifunctionality also increases the probability of perceiving the stakes of other actors and the willingness to include them in the process.
While the Hondsbossche Duinen case is atypical in the set of other Weak Links projects, which have been realised in less time and with less conflict, it still shows the signs of a typical case for the Dutch policymaking arena, in which participation of affected and concerned actors is important. Provincial governments in the Netherlands usually take into account the interest of their population and are responsible for regional development as well. The fact that the first phase ended in controversy with a policy solution found with limited participation is rather surprising from this perspective. Digging their heels in, provincial project managers avoided opinions and advice from outside their own organisation. The water board, on the other hand, is an institution with a bottom-up history, originally established by farmers to protect their life and work surroundings. In the water board's case, organisational history was part of the reason for the choice of approach.
The final aspect at play that this analysis revealed, is the interplay between framing foci and actors' institutional power. In the first project phase, the ministry's authoritative veto power prevented the realisation of a project that was not in the national government's interest. Other actors also had no formal power in the process. They could only have participated with the goodwill of the provincial government, which did happen in the following phases. The framing foci perspective clarified the reasons for the use of formal power in the decision-making process.
Understanding actors' ability to reach policy consensus in the Hondsbossche Duinen project despite their differences in framing foci requires the exploration of the assemblage of frames. Our framing foci perspective enabled us to specify which constellations of framing foci between actors led to conflict and why the project finally succeeded. It produced an in-depth analysis of the elements of policy framing which may otherwise be convoluted. We observed different mechanisms with regard to emphasising or ignoring similarities and dissimilarities between actors' framing foci. These mechanisms deliver an understanding of why the Hondsbossche Duinen project almost broke down in the first project phase, while it thrived in the later phases. By extension, the approach we present underlines the value of a policy framing foci perspective in understanding how coastal governance processes play out.

Declaration of competing interest
The authors declare no conflict of interest.