Looking for causes of effects in cases: Evaluating intermunicipal collaboration in The Netherlands applying QCA

Economic geography and regional planning suffer from a lack of clear answers with respect to the question what methods to use to systematically explain complex territorial phenomena such as regional development. Rather than the universal effect of causes, analysis should focus on patterns revealed through case-specific effects of enabling and disabling conditions. Using qualitative configurational analysis (QCA), this paper illustrates the relevance of such an approach examining the variable effectiveness of intermunicipal collaboration in the Netherlands. Drawing on a survey of a large sample of collaborations, complemented with case-specific inside knowledge, eight conditions are found supporting collaboration effectiveness: age, size, homogeneity, project/policy activity, mission and inclusiveness. The configurational analysis uncovers one dominant evolutionary pattern (evolving policy focus), one minor pattern (metropolitan collaboration) plus two somewhat unique cases. It also finds contributing roles for mission-orientation, inclusiveness, and, somewhat surprisingly, size. QCA proves a promising tool to study complex dynamics across a population of territorial cases.


Introduction
Rather than single linear explanations or pathways, regional development tends to follow multiple complex patterns. Moreover, complex spatial phenomena, such as territorial collaboration, depend primarily on indigenous dynamics and self-organisation in which an important part of causality is best captured in terms of enabling/ disabling conditions. Hence, instead of probing the effect of success factors, as universal causes, analysis should focus on the case-specific effect of conditions as causes, and examine which more general patterns transpire from examining multiple cases. A suitable method for this is Qualitative Comparative Analysis (QCA). QCA overcomes the limitations of traditional statistical techniques, such as poor qualification of effect sizes, applying unifinality and the danger of ecological fallacy (Rutten, 2019). QCA, in turn, embraces a contextual logic based on the principle of equifinality, both where it concerns the 'enablers' (there is more than one route to success) and outcomes (similar routes may lead to different destinations). Until now QCA has not been abundantly used in the field of economic geography. Some notable exceptions are provided by Rutten (2019Rutten ( , 2020, discussing the role of openness in regional innovation and Li & Bathelt (2019) on successful knowledge strategies across space. As we seek to illustrate here, QCA presents a fruitful approach to undertake regional socio-economic and policy research.
Establishing regional collaboration is seen as a strategy to achieve efficiency gains in the provision of public services, improving political clout and improving economic strength by aligning local investments and activities (Hospers & Beugelsdijk, 2000;Miller & Lee, 2009). The major drivers of collaboration are collective service provision, efficiency, and increasing importance of 'placebased (socio)economic development and branding' (Geertsema, 2017;Grindle, 2004;Hooghe & Marks, 2001;VROM et al., 2009). In their discussion of interjurisdictional collaboration across the US, Nunn and Rosentraub (1997) also present the motive of (re)distribution between more and less wealthy areas. The latter does not play a role in countries such as The Netherlands, where municipalities receive most income from national budgets. Nevertheless, while practical motives of mutual gain and efficiency prevail, and supporting conditions are present, establishing and maintaining effective collaboration is not easy. In practice, research has shown that only about half of all Dutch collaborations are successful (Duysters et al., 2012;Hospers, 2012). Besides illustrating the power of QCA, understanding this modest success presents the focus of this paper.
The paper is structured as follows. A review of theory on collaboration serves to sketch a broad spectrum of conditions. This is followed by an explanation of the configurational method (QCA) and data collection, and by a presentation of findings and conclusion.

Conditions of intermunicipal collaboration
In the Netherlands, intermunicipal collaboration has a long history. Since 1851, when the first law on municipal collaboration was established, municipalities have joined forces and resources in specific areas such as mobility, spatial planning (housing, business estates) and environment (waste, water). In recent decades, devolution in welfare and care provision from national and provincial to municipal level has induced further collaboration. Since 1950, collaboration is formally arranged through the Law on Common Provisions (Wet Gemeenschappelijke Regelingen, WGR), revised many times since. Currently, three developments stand out. First, due to increased administrative and competitive pressures, municipalities either choose to merge or to enhance collaboration notably in the field of common service provision and/or economic profiling and strategy-making (e. g. through establishing Economic Boards). It should be said that merger has always been a strong alternative to collaboration, resulting in a decline of the number of municipalities from almost 1200 in the early 1900 s to 355 in 2019. Second, also in light of the broadening spectrum of collaboration (from services to economic strategies), a broader variation of institutional arrangements is used for intermunicipal collaboration besides the WGR. This is part of a constant search for new variations in the nature, scale, establishment, organisation and performances of inter-municipal collaborations (Hes & Van Sabee, 2015;Johnston & Acharya, 2007). Third, there is increasing concern about the effectiveness of collaboration, also in light of mounting (cost) pressures on municipalities. Despite the long-standing institutionalisation and experience with collaborations, municipalities and other stakeholders find it hard to work together. This raises the question of how effective collaboration can be established.
Which conditions play a role in turning intermunicipal collaboration more or less successful? What commonalities and variations can we see in the achievement of successful collaboration versus failures? This section will revisit the literature on collaboration, starting from the broad view that collaboration embraces three broad dimensions, namely political, institutional and social-cultural. In turn, these dimensions are each synthetically covered by three perspectives, namely institutional collective action theory, collaborative governance and regional regimes (Kuijpers, 2017). Our review, accordingly, leads to a conceptual matrix of three perspectives shedding light on three aspects of collaboration (Table 1).
The three perspective are as follows. First, collective action theory adopts a primarily instrumentalist approach, with a focus on the distribution of costs and benefits between collaboration partners, focusing on the long-term (Boogers, 2013;Feiock, 2009Feiock, , 2013Nicholls, 2005;Olson, 1971;Van den Berg et al., 2002). It sees collaboration in terms of a deal in which costs are shared to create and distribute benefits among the partners. Such a deal is well anchored in institutional-legal frameworks, based on a certain homogeneity in the scope, power and resources and a fitting level of mutual trust. Second, collaborative governance theory features the role of organisational-cultural conditions of 'stakeholders' in the way a collaborative arrangement functions as a social arrangement (Emerson et al., 2015). To do so, members should manifest a certain equality in their stakes and resources for collaborating, and the network should possess adequate levels of binding leadership, ways of trust-creation and conflict regulation. Finally, drawing on urban regime theories, regional regime theory addresses the way collaboration yields, and is sustained by, a political coalition of interests, resources and a shared spatial-economic agenda (Boogers, 2013;Hamilton, 2004;Stone, 1989). The impetus for collaboration stems from commitment going beyond formal administrative motives, and captures the transition from 'managerial' to more 'entrepreneurial' forms of territorial governance (Harvey, 1989). Regimes derive their strength, in particular, from the input and quality of stakeholders acting as civic 'leaders' or 'policy entrepreneurs', and the umbrella organisations channelling their views and interests (Keating, 1998).
Turning to the dimensions, political conditions encompass the grounds and logics for choosing a particular form and scope for collaboration, the creation of political alliances and leadership, and the advocacy for intermunicipal collaboration from the wider political economy of the region. Three core factors stand out, namely 'value propositions', political motivation and the sense of urgency. First, for all municipal partners, collaboration, as a form of collective action, must present explicit and convincing 'value propositions', based on realistic and achievable aims (Von Malmborg, 2003). Collaborations manifest, in general, a strong instrumentality, rooting collaboration in calculative logics of costs and benefits at partner level. Just solidarity based on benefits accruing to the region as a whole presents insufficient motive for partners to join or stay on. Such instrumentality, however, exceeds the economic-financial domain. Value refers here to the broad range of societal-spatial issues for which municipalities are responsible. Second, how these values are pursued depends on the role of motivational factors, power and leadership in forging processes of communication, decision-making and implementation, topics generally discussed under the header of collaborative governance theory (Ansell & Gash, 2012;Boogers, 2013;Boogers et al., 2016;Van den Berg et al., 2002). Third, a key motivational trigger is the way prominent leaders and organisations shape a certain sense of urgency prompting collaboration. The sense of urgency fuels the more formal steps towards collaboration, as well as the more concrete collaborative policy processes and measures (Hamilton, 2004). An important aspect of the sense of urgency is a broad awareness of individual actors that they do not have the ability to solve core regional problems on their own, so that collaboration is imperative (Von Malmborg, 2003). An overarching dimension, finally, is the political-economic setting of the intermunicipal arena, in which political motive and drive are rooted (as discussed in Regime Theory). This setting includes the views and interest of leaders and representatives of business, non-state organisations (science, education, utilities, etc.) and civil society, from the 'captains of industry' to the transition advocates of NGOs (Van de Laar, 2010).
Institutional conditions refer to the way interaction is structured and performing. For collective action, this warrants a well-structured governance framework, with clear (limited) membership and a fitting legal framework (Olson, 1971). Such a framework does not only entail decision-making on policy focus and design, but also the way implementation and activities are coordinated. Collaborative governance adds to this framework the grounds and skills for negotiation, including the competencies for organising the decision-making and communication processes and the mechanisms of conflict resolution. Moreover, at a more informal level and intersecting with social-cultural conditions, the evolution of mutual conventions ('accepted ways of working together') and mutual trust is important. Trust generally manifests a high level of recursivity. That is, mediated by the interdependency between stakeholders and the voluntariness of a collaboration, trust is fuelled by tangible success of the collaboration (Ansell & Gash, 2012). Extending the institutional roots beyond the formal boundaries of the municipal state, finally, three conditions come into play: -the strengths of regional networks, notably their institutional capacity to channel issues and fuel senses of need and urgency; strong regional networks heavily depend on the willingness to act together, flexibility and trust (Van den Berg et al., 2002) and can be characterised by organisational features as homogeneity and scope (Boogers, 2013); -possibilities for negotiation outside the formal meetings of the collaboration, involving other leading figures and representatives from the region; -a liaison or even fusion of the state-only intermunicipal collaboration with a wider regional umbrella organisation, such as a Triple Helix configuration (for instance in the form of an 'Economic Board'). -Strong (socio-economic) regional networks; regional umbrella organisation 5 -Room for informal meetings * Social-cultural -Homogenous population structure 3 -Regional social unity * -Bonding between partners 1 -Regional solidarity * -Wide social commitment 7 -Quality of stakeholders' contacts and networks *

For reference numbers see conditions in table 4 * -not directly covered by conditions used in the analysis
Source: own elaboration Social-cultural conditions primarily play an enabling role. Issues like shared regional identities, solidarity and mutual understanding can certainly lubricate the formation of a collaboration and its actual performance, but do not commonly play a decisive role. That is because intermunicipal collaboration is primarily about creating shared benefits, not about the establishment of political-cultural regions. Nevertheless, within the environment of the collaboration, what may help is a certain homogeneity in population structures and the perception of social unity. Also helpful are a good fabric of interpersonal contacts and networking amongst stakeholders (Hes & Van Sabee, 2015), and a wide social commitment to doing things together, as part of deeper regional anchoring. Within the setting of the collaboration, social bonding and solidarity play important enabling roles (Paasi, 1986). The elements listed in Table 1 should not be seen as exerting determinate effects, but as conditions that in different combinations may (dis)serve to the way stakeholders collaborate. Because of its intricate and complex nature, intermunicipal collaboration is a phenomenon in which, to quote Nunn and Rosentraub (1997, p. 217), "different tactics and approaches may achieve similar outcomes". Hence, collaboration is rooted in different combinations of conditions, enabling agency and resources contributing to more or less effectiveness. They do so in complex ways. The same condition may contribute to opposite outcomes (effective vs less-effective) in conjunctions with other conditions, while the same outcome may be supported by different configurations (equifinality).
To deal with such causal complexity and equifinality, this paper draws on Qualitative Comparative Analysis (Marx et al., 2014;Ragin et al., 2003). QCA's key aim is to reveal necessary and sufficient conditions for an outcome, bundling cases in configurational sets underpinning certain 'recipes' or 'paths' or, as called here, 'types of collaboration'. More technically, configurations present sets of cases marked by the presence and/or absence of certain conditions yielding one outcome. For these configurations, the conditions are called 'INUS-conditions', single conditions being insufficient but necessary parts of a compound configuration which is itself unnecessary but sufficient for the outcome. The application of INUS-conditions enables systematic and robust interpretations based on equifinality (multiple paths towards one outcome) and in which conditions explain in contingent and combinatory manners. Importantly, and unlike regression analysis, QCA's path (or pattern) recognition occurs bottom-up, through reducing the number of configurations needed to explain the outcomes for all cases. In doing so, QCA helps to "gather in-depth insight into the different cases and capture their complexity, while still attempting to produce some level of generalization" (Marx et al., 2014, p. 129).
The bottom-up processes adopted here consists of a two-step empirical procedure. The first step entails a primary pre-screening of the total corpus of cases with the help of a broader set of conditions. On the basis of the pre-screening results, the second step narrows down both the corpus and list of conditions, thus constituting the final corpus and set of conditions for the configurational analysis.

Measuring effectiveness and its conditions
How do we define effectiveness? As Nunn and Rosentraub (1997, p. 208) argue, the success of collaboration strongly depends on local contexts and ambitions: "measuring the outcomes of interjurisdictional cooperation must be based on knowledge of the original objectives of each region's attempt at cooperation, and not be driven (…) by some set of preconceived notions about what is and is not good cooperation". The analysis here thus adopts an instrumental approach to the characterisation of effectiveness and its conditions. Effectiveness is defined as reaching, on balance, a positive benefit for the partners involved. Theoretically, effectiveness entails the extent to which a region succeeds in translating self-determined collective interests into concrete initiatives by using a regional cooperation structure (Levelt et al., 2012). In more practice terms, a collaboration is experienced as effective and well-functioning when a situation is created in which the benefits for a region do outweigh the perceived costs of a cooperation (Feiock, 2013;Lowery, 2000). Costs and benefits go beyond financial and economic surpluses. Financial costs concern the fees of municipalities and inhabitants of a region to achieve objectives; yet these weigh less (in terms of amounts) than non-financial or collaboration costs. Boogers et al. (2016) distinguish five types of collaboration costs, namely information, coordination, negotiations, compliance and representation costs and four types of collaboration benefits. First, information costs are incurred by the need to obtain insights into the resources and objectives of all stakeholders. Second, coordination costs stem from the adaption of various policies of municipalities and other organisations to the specific nature of a region. Third, negotiation costs are needed to achieve consensus regarding the content and form and distribu-tion of costs and benefits. Fourth, compliance costs originate from efforts to control stakeholders concerning the appointments made and their contribution to intermunicipal collaboration. Fifth, representation costs cover the meetings and discussions with boards of municipalities and other stakeholders. Opposite these costs, Boogers distinguishes between the following types of benefits: -Strategic benefits: the tackling of regional objectives and social issues; -Tactical benefits: the development of regional effective policies; -Operational benefits: efficiency and mass in involving; -Economic benefits: impact on gross regional product.
In our configurational analysis, outcome is based on the overall balance of costs and benefits. A positive surplus is associated with effectiveness; a negative surplus indicates less-effectiveness.
How do we select conditions? To limit the number of possible configurations, QCA is commonly executed with 5-8 conditions (Berg-Schlosser et al., 2009). In view of the encompassing exploration of possible conditions (Table  1), our pre-screening started with a broader inventory. Subsequently, the relevance, calculability and availability of optional conditions have been discussed in detail with experts in the field (as part of an internship by one of the authors). This has resulted in a selection of 10 conditions, covering a large part of the spectrum covered by Table 1. We recognise that finding good and doable conditions is a difficult task which always leaves certain aspects wanting. In the final selection, every cell is covered by at least one condition, with some more emphasis on institutional and political conditions than on socio-cultural conditions. The conditions are:

Age [YOF]
Many aspects of collaboration evolve through time (Boogers, 2013). This includes the creation of mutual understanding, solidarity, trust and commitment, the creation of effective means of informal and formal communication, and the development of shared governance and decision-making structures. A basic expectation thus is that intermunicipal collaborations learn how to be more effective over time by gaining maturity (Boogers, 2013;Bremekamp, 2014;Hes & Van Sabee, 2015). However, age may also erode effectiveness. The sense of urgency prompting the initial steps towards collaboration may evaporate. Feelings of solidarity and commitment may be tempered by competing interests or mounting irritation. Trust may be broken due to negative experiences. Collaborations may also become subject to institutional sclerosis. Accordingly, how effectiveness changes through time, if at all, may vary.

Size: number of partners [NOP]
A second condition of effective collaboration is the number of participating municipalities. A general expectation is that homogeneity in scope, power and resources, the building of a consistent network, informal communication and solidarity all stand to benefit from a smaller group. So a first expectation is that it is easier for smaller collaborations to be more effective than larger ones (Boogers, 2013;Boogers et al., 2016). There are also counterarguments, however. A larger network may benefit from scale economies, have more political clout, and hence muster more resources from local, provincial and national levels. There are more possibilities for distributing costs and benefits amongst a large group of partners. Larger groups may also be better able to regulate and stem conflicts, because it is more difficult for one or two partners to overpower the network. So, like time, size may work both ways, or not work at all, and depend strongly on specific issues and challenges.

Group homogeneity [HOM]
A closely related condition is the homogeneity amongst the partners in population size (the extent to which partners have comparable numbers of inhabitants. The basic expectation here is that collaborative structures manifesting more homogeneity are more effective (Boogers, 2013). The main reasons for this are threefold: -equal power positions, -the operation of governance structures (including conflict regulation) and leadership at a distance of the individual partners, and -the creation of balanced mutual trust.
Opposite mechanisms could also be at work, however. A large core municipality surrounded by several smaller (suburban) ones may benefit from the sense of urgency stemming from typical metropolitan issues such as congestion, land shortage for housing and business estates, environmental problems, etc. Such an 'unequal' collaboration may also benefit from the resources and political clout of the dominant municipality.

WGR formalisation [WGR]
Three conditions deal with coordination aspects, covering formalisation, involvement of regional stakeholders and coordination. On the first aspect, intermunicipal collaboration warrants an agreement stipulating the identity and role of the parties, its overall scope and specific domains, and its procedures of communication and decision-mak-ing. This is provided by the legal framework on intermunicipal collaboration, the 'Wet Gemeenschappelijke Regelingen' (WGR). The expectation here is that the effectiveness of intermunicipal collaborations benefits from the formal structure of a regional cooperation (Hamilton, 2004;Hes & Van Sabee, 2015). Formal anchoring is a signal of commitment and helps to fix the basic terms and expectations of the collaboration. Theoretically, it could limit flexibility, although municipalities always have the possibility to extend or change the scope of their joint activities in an informal way, followed by adaptation of the formal contract. A negative impact therefore seems remote. What is less remote is that WGR has no impact. Although the 'WGR' legal framework prescribes strong collaboration agreements with many obligations, this can also be met by other legal frameworks (creating a foundation, using a variety of management agreements, etc). Also given the widespread use of other frameworks, it is an open question whether WGR makes a noticeable difference.

Involvement of business and education: Triple Helix [TH]
Besides providing for horizontal ties between municipalities, collaboration may also entail lateral ties with organisations from business and knowledge domains. This generally occurs through the establishment of a 'Triple Helix' umbrella organisation, in which decision-making on regional development is taken by a collective of municipalities and stakeholders from business, science and education. Sometimes also civil society (NGOs, CSOs) is involved as a fourth domain, giving rise to a so-called 'Quadruple Helix'. Such lateral engagements can be expected to bolster collaboration, making it better supported and hence effective (Boogers, 2013;Hamilton, 2004). Market, education and civil society parties may yield insights, support and resources for a wide variety of strategy developments and service provisions. Through their broader basis, Triple Helix structures also make it easier for leading figure from business, education and society to express their views on current states and future direction of a region. This facilitates a role for so-called 'policy entrepreneurs' striving for transitions (mobility, energy, housing) and 'civic entrepreneurs' engaging in the development of economic clusters (Ebbekink & Lagendijk, 2013). A major advantage of a Triple Helix nexus is that such critical stakeholders will not use other, rival platforms to advocate their ideas (Janssen-Jansen, 2010). There is also a major caveat, however. Inclusion of non-state partners raises major issues on scope, legitimacy and responsibility. Where state-only collaborations may work from a clear joint idea about overall ambit and purpose, Triple Helix structures often face the need to reconcile rather different views on both purpose and means of collaboration (Forsberg & Lindgren, 2015). As spelled out by urban regime theory, there is a danger to fall prey to partisan interests and narrow political aspirations (Stone, 1989).

Operational coordination [CON]
The final coordination issue is rather specific. It concerns the use of contractual means of policy/project implementation complementing the joint decision-making at strategic level. Such contractual means can entail the use of an executive body, the reliance on tenders and competitions, or other forms of market coordination (Buitelaar, 2003;Exworthy & Halford, 1999;Feiock, 2009;Janssen-Jansen, 2010). While much literature has been devoted to network coordination and means of joint-decision-making, literature on intermunicipal collaboration has argued that, notably for the actual policy implementation and service provision, the use of contractual modes of coordination may be beneficial (Buitelaar, 2003;Hamilton, 2004). Such other modes may help to improve transparency, delivers conflict mechanisms, and enhance efficiency through a good separation of strategic and operational processes.

Policy orientation (instead of project orientation) [KOA]
Intermunicipal collaborations face a basic choice in what kind of collaborative ventures are pursued. They can share resources and partake in common projects, or they can step up their ambitions and develop and implement common policies. The expectations are that the latter yields a higher level of effectiveness (Boogers, 2013;Boogers et al., 2016;Miller & Lee, 2009). Resource pooling and joint projects also bring benefits, but these accrue more to the level of the individual municipalities than the group as a whole. Joint policymaking, on the other hand, makes the partners focus on joint benefits for the territory as a whole, calling on regional solidarity and a wider, more intense commitment. The latter is also prompted because, while resource pooling and joint projects primarily require coordination of operational activities, policymaking covers the full span from policy-design to implementation. Yet, embarking on the full span also comes with a downside. Joint policymaking considerably raises the efforts and costs of policy coordination within the respective municipal administrations. From a pure utilitarian perspective, the need for collaboration may be felt more directly in the case of resource pooling and joint projects than for joint policymaking. Starting with less ambition is thus often presented as a way to achieve results more easily and quickly. A problem, however, is that the results, because of the limited collaboration, may not meet the problems at hand. Especially when there is a sense of urgency about certain spatial-envi-ronment issues (congestion, housing shortage, detrimental competition between business locations, pollution, etc), a more ambitious approach is likely to be more effective.

Issue focus [FOC]
Another choice collaborating municipalities have to make entails the scope and breadth of policy fields addressed. Do collaborations focus on selected policy fields or do they collaborate across the board? Both options show pros and cons. A prevailing idea is that issue focus may raise the chance that collaborations will work effectively (Ebbekink & Lagendijk, 2013;Hes & Van Sabee, 2015). In general, issue focus can anchor itself more easily in a perception of need and sense of urgency. It also makes it easier to set up coordination and legal arrangements and bring relevant actors together, sometimes even in different spatial configurations (Keating, 1998). A disadvantage of (too much) focus, however, is that it inhibits coordination and alignment between sectoral policies. Spatial-environmental challenges faced by municipalities often warrant cross-sectoral, even integral approaches. Pursuing collective goals may thus benefit from collaborating in multiple fields. More generally, integral approaches are often based on synthetic visions of major regional challenges and directions. If such a vision is well-crafted, and well-supported and conveyed by all stakeholders, it can make a broader approach highly effective (Van den Berg et al., 2002). If a vision merely presents a paper tiger, the effect may be the reverse. The role of the issue spectrum, accordingly, is strongly dependent on contextual factors and the kind of collaboration pursued.

Mission orientation [MIS]
A related issue is the kind of objectives pursued by collaboration: is the collaboration explicitly aiming at collective strategic aims and missions, generally driven by a sense of urgency, or does it pursue more tactical-operational objectives to create mutual benefits? This condition can work two-ways. On the hand one, strategic orientation generally requires more in terms of commitment, policy design, support, and implementation, and are thus more prone to resistance and failure. According to Boogers (2016), intermunicipal collaborations that mainly aim at strategic aims run the risk of building insufficient capacities for effective collaboration, in comparison with collaborations formulating primarily tactical or operational objectives. Strategic forms of collaboration usually are more entrusted with political choices and policies resulting in a more difficult and often ineffective decision-making process. On the other hand, strategic goals may call upon shared drives, a commons sense of urgency and direction, and visionary leadership. This may set a collaboration on a roller-coaster of appealing and motivating visioning and strategy-making, joint commitment and resourcing, strong networking and positive results which can be attributed to joint strategic orientation. While strategy-making is more demanding, it may also be more rewarding.
10 Inclusive decision-making process [DMP] Balance of power plays an important role in the shaping of collaboration and its effectiveness. So, the question is whether intermunicipal collaborations become more effective when all partners participate in the decision-making process, something which is expected in the literature (Boogers, 2013). The basic idea is that inclusive decision making underpins a more equal power position of partners, better possibilities to distribute costs and benefits, more mutual trust and stronger overall commitment. A problem may be, however, that inclusion of more parties reduces overall governance and strategic capacities. In certain context, it can be more effective to confine decision-making to a small number of core players, like the larger, most competent and resourced municipalities. Where fitting, the latter can be part of a Triple Helix umbrella structure. This condition, indeed, can be expected to mirror, to some extent, the scores on Triple Helix.

Methods and data
Data was collected through surveys sent to municipal staff and regional stakeholders with key positions for all collaborations recorded by the Netherlands Association of Municipalities, VNG. Of 530 surveys sent, 143 were fully completed. This covered 58 of 72 known collaborations. To enable triangulation, only those collaborations with multiple responses have been included in the analysis, resulting in 37 cases. Data has also been checked through available documents and websites, and follow-up inquiries with local experts accessed through the internship organisation (referred to as 'field knowledge').
Following a 'mutual gain' (payoff) logic, effectiveness was measured in terms of perceived costs and benefits. As explained above, costs were divided into information, coordination, negotiation, compliance and representation costs and benefits into strategic, tactical, operational and economic benefits.
The municipal staff and regional stakeholders contacted were asked to assess effectiveness in three ways (all on a 1-10 scale), namely through  1. scores per cost category (information, coordination, negotiation, compliance and representation) and benefits (strategic, tactical, operational and economic); a balance score CAT is calculated by subtracting the average of all costs from the average of all benefits. 2. scores of overall total costs and total benefits, yielding the balance score 'TOT' and 3. grading of overall effectiveness (MARK), The scores per region have been calculated by averaging the scores for CAT, TOT and MARK over all respondents, yielding aCAT, aTOT and aMARK. Then, through various rounds of expert consultation, positive effectiveness has been defined as the cases with -aCAT or aTOT showing one positive and one at least neutral score, or -with one positive and one negative and aMARK scoring 6 or higher, -or aCAT and aTOT scoring both slightly negative (between 0 and -1), but aMARK showing 8 or higher. Cases with lower scores are considered to be non-effective.
Absolute grading scores turned out as follows. On a scale from 1 to 10, cost items scored between 5,3 and 6,2, with an average of 6; total costs score 5,8. Benefit items scored between 5,8 and 6,2, while total benefits scored somewhat higher at 6,5. Overall effectiveness was valued at 6,1. Using the effectiveness formula, 22 collaborations appeared effective, and 15 appear less effective at pre-screening. Turning to the conditions, seven out of ten were measured as binary variables to create a crisp set, in the sense that the attribute either exists or not: 4, 5, 6, 7, 8, 9, and 10 (see above). For the other three conditions, cut-off values have been determined, on the basis of expert consultation and further calibration. The latter means that cut-off points, with a range deemed reasonable by the experts, have been set at such a value that similar configurations yield equal outcomes. For the time factor (1), collaborations with more than six years (before 2011) are considered to have a 'long history'; a larger collaboration (2) is considered one with 15 or more partners. Homogeneity (3) is based on a scoring scale running from 0 to 1, with a cut-off point at 0.55.
Subsequently, all scores have been combined in a truth table. A truth table contains all combinations of conditions present in the dataset plus the outcome (1, 0 or contradictory). Applying QCA rules , this table has been modified in two ways, to address the presence of some configurations with conflicting outcomes (all equal condition values with contrasting outcomes). Two conflicts were solved by changing the cut-off points for time and homogeneity. A third conflict was solved by removing what was considered an 'odd' result, namely the positive score of 8KTD (a contracting of the names of the two constituting municipalities of Achtkarspelen and Tytsjerksteradiel).
The final set, accordingly, measured 11 columns (10 conditions plus one outcome) by 36 cases (See appendix). These cases are combined in the truth table (Table 2).
The analysis was carried out following crisp set QCA (csQCA), with binary scores for conditions and outcome (present/absent). The use of the crisp set approach, rather than the more complicated fuzzy set, has been deliberate. A crisp set provides a straightforward and parsimonious analysis of which configurations contribute to the outcome or its absence, given the collaborations' aim to yield an overall benefit for the partners involved. Moreover, as nearly all the original data used in this study were binary in nature, and those which were not appeared to perform well with single cut-off points, a crisp analysis permits a straightforward presentations of the findings through an easily interpretable truth table (Kraus et al., 2018;Marx et al., 2013). Separate analyses took place for effective and less-effective outcomes, using the intermediate solutions (including logical remainders excluding contradictory cases from the logical minimization process).
As QCA is best applied to sets containing not too many conditions, the original dataset was scrutinized if the number of conditions could be reduced. A first 'visual' and somewhat crude inspection learned that the condition Operational Coordination (CON) only applied to two cases where the intermunicipal collaboration is organised in the form of Economic Boards. Strategic decision-making takes place here through Triple Helix umbrella organisations, and policy implementation, with a focus on economic branding, business support and networking, takes places via a separate agency (the executive of the EBs) and market contracts. Decision-making, moreover, is in the hands of group of core partners. This in line with theoretical insights that effective collaboration is supported by autonomous institutions and the use of private agencies and strong leaders. From a QCA perspective, this configuration proved relatively marginal since only two cases are part of this configuration; to test the strength of the condition 'non-network coordination' would warrant the exploration of more intermunicipal 'Economic Board' cases. Field knowledge, as well as a quick web scan, revealed that cooperation-based Economic Boards were mushrooming across the country, presenting a newly emerging, post-WGR collaboration type. It was decided to leave out this type from the QCA analysis, effectively also removing the CON-condition. A first inventory further revealed that the Triple Helix and WGR-formalisation conditions also were not really prominent and discriminatory, resulting in the removal of two further conditions. This removal meets theoretical expectations. As explained above, WGR has lost its historically prominent role, due to the rise of alternatives. In the data (Appendix), Triple Helix indeed mirrors inclusiveness (DMP). This all resulted in a dataset of 34 cases and 7 conditions and a truth-table with 22 configurations (Table 2), meeting the criteria of cases-condition relation as set by Marx et. al. (2013).

Findings
The analysis has been carried out using Tosmana, a QCA software, followed by an extensive exploration and discussion of possible configurations. Based on empirical knowledge, the third outcome of the analysis for the effective and the first for the ineffective outcomes proved the most sensible, in which the configurations give rise to eight types of collaboration, four explaining effective outcomes and four explaining less-effective outcomes, as shown in Table 3. The analysis makes a distinction between core 'conditions', defining the configuration, and 'peripheral' conditions, which additionally characterise the configuration. Table 3 also lists raw coverage (RC, share in the outcome of all cases covered by the type), unique coverage (same share restricted to cases only appearing in one type) and consistency (share of cases in the type-set with the corresponding outcome)

Types of effective collaboration
Focus and maturity (Type A: YOF[1]+FOC[1]) A prevalent configuration for effective intermunicipal collaboration is the co-occurrence of a long history and issue focus. This sizeable type of eight collaborations (RC=0.42) manifest a policy-and mission-orientation, with only one exception, DOWR, which is more service oriented. Overall, this strong type confirms initial theoretical expectations regarding effective collaborations: maturity, issue focus and (for most) mission.

Metropolitan (Type B: YOF[1]+NOP[1])
The second configuration explains effectiveness through the compound condition of a large number of partners and absence of issue focus. The set comprises the two unique cases of the Metropolitan Areas of Amsterdam and Eindhoven, which differ from Type A in adopting broad rather than focused issue focus. These two major urban areas, with their heterogenous composition and massive resource base, manage to effectively collaborate in a broad range of policy domains and issues. A general finding to report here is that of all the larger collaborations only one appears less-effective, namely AgriFood Capital, which presents a special case further explained below (H). Size thus appears an almost sufficient condition for gaining effectiveness, contracting the notion that it is easier for smaller collaborations to gain effectiveness. Field knowledge indicates that size supports effective decision-making because there is less risk that collaborations are hijacked by minority positions.
Policy coherence (Type C: YOF[0]+HOM[1]+KOA[1]) Effective collaboration is also the hallmark of two groups of younger partners, manifesting policy orientation and inclusive decision-making. Type C includes four collaborations, Keyport 2020, BUCH, Werkorganisatie CGM and Noaberkracht, which all have established effective structures to work together in delineated policy fields. Field knowledge reveals that for the last three cases, this is based on a dedication to joint service provision enabled by what is described as an 'administrative fusion' between municipal processes. Keyport 2020 has a stronger focus on economic strategy-making. The four cases manifest a small number of partners (between 2 and 6), which, sustained by group homogeneity, facilitate prompt results. Moreover, they are located in non-core areas of the Netherlands, all suffering some kind of demographic problem (out-migration of young, highly skilled population or even shrinkage), spurring them, according to field knowledge, to develop collective strength. More generally, policy orientation manifests itself as a nearly necessary conditions for effectiveness (A-D, with one deviant case in A).

Inclusive-strategy (Type D: YOF[0] +MIS[1]+DMP[1])
The final effective group of young collaborations stands out by mission-orientation and inclusive decision making, supported by a policy orientation. This type involves four cases, all manifesting a certain specialisation in manufacturing in locations outside of the Randstad, which informs and motivates joint agenda-setting and policymaking. Inclusive decision-making in this group is based on engagement with wider societal and business circles, which will contribute to support and effectiveness. A special case here is Regio Zwolle. While younger and with issue focus, Zwolle performs similar to Eindhoven and thus approaches Type B. Our field knowledge indicates that Zwolle's inclusive decision-making roots the collaborations strongly and effectively in the wider regional community, including networks of entrepreneurs.  5, 7, 8, 9, 10, 11, 12 3, 6 13, 14, 15, 16 13, 17, 18, 19 Overall solution consistency 1.00 Overall Cases (unique in italic) 20, 21, 22, 31, 32, 33, 34, 35, 36 24, 25, 26, 27, 28, 29 30, 31, 32 22, 23 Overall solution consistency 0.88 Overall solution coverage 0.93 (14/15) Note: Black circles ("•") indicate the presence of a condition, and open circles ("⊗") indicate its absence. A hyphen ("-") a condition that is not relevant to that particular configuration in regard to the outcome (it is both present and absent in the set). Large circles suggest (set-defining) 'core' conditions, while small circles indicate (emerging) 'peripheral' conditions.

Conclusion
Regional or intermunicipal collaboration plays an important role in local government and policy-implementation in the Netherlands, but also shows quite mixed results. What explains this variability? Using a large sample of registered initiatives in the Netherlands, this study has explored conditions for more or less effective collabora- So, the question is not what defines success or failure, but whether, amongst all the unique cases explored, one can detect certain commonalities and patterns. This then amounts to a genuine geographical understanding of the phenomenon of collaboration, which sees causality in substantive terms -probing the causes of effects in cases -rather-than in probabilistic terms -predicting the effects of causes (Goertz & Mahoney, 2012; see also see also Rutten, 2019). Such an approach, we would argue, is suitable for all geographical phenomena marked by high levels of internal agency and self-organisation (urban/ regional development, innovation, policy), for which too often inappropriate probabilistic-universal methods are applied through regression and econometrical analysis. Following such a pattern-seeking, qualitative approach, this study has employed a crisp-set configurational analysis (QCA). Results come in the form of collaboration types, that is, configurations of condition explaining effectiveness and less-effectiveness. Not unexpectedly, quite a number of types emerged, four matching effectiveness and four less effectiveness, manifesting a number of patterns. The most prominent pattern here is that effectiveness takes time, provided with time comes focus (type A versus F). Another prominent pattern is how effectiveness requires policy orientation (as a near-necessary condition for A-D), while project orientation appears clearly detrimental for young collaborations (E), although these findings do not come with full consistency. So, the core pattern emerging is that policy and mission orientation are beneficial for young collaborations (C, D, versus E), which, when collaborations mature, warrant further issue focus (A vs F). This pattern covers 30 of our 34 cases. Corroborated by their low unique coverages, the other types manifest some additional patterns. Type B's unique cases, the Metropolitan areas of Amsterdam and Eindhoven, reflect the capacity of these two major urban areas to be effective without focus. Types G and H each cover the one unique less-effective case, namely ABG-municipalities (Gilze-Rijen) (G) and Agrifood Capital (H). This lesser performance can be attributed to mission-orientation and inclusiveness respectively. In sum, we find one dominant evolutionary pattern (evolving policy focus), one minor pattern (metropolitan) and the specific characteristics of two unique cases.
What do these patterns mean in the context of the conditions listed in Table 1? First it should be said that such an assessment does not entail a theoretical (dis) confirmation or qualification. The only thing we can do is reflecting upon the role of conditions as transpiring from our cases. As summarized above, our synthesis is relational-configurational (see also Table 4). While some conditions manifest more prominence, such as policy orientation for all, and issue focus for older effective collaborations, all these results remain contextual. Nevertheless, a clear match can be found regarding the emphasis of collective action theory on political and institutional aspects, notably focus, mass for cost-benefit distribution, and formalisation (albeit not through the WGR). Perhaps somewhat surprisingly, size proves to play a rather positive, supporting role of capacity-building. The regime perspective sheds light, in particular, on the mushrooming of economically oriented cooperations (partly set apart in our analysis). Both along more formal and informal lines (inclusive decision-making yet focused and mission-driven), these collaborations show how business interests and economic positioning infuse cooperation. After our data collection, we have witnessed how many more regions established Economic Boards. While these may quickly gain effectiveness, this development also raises concern. In times when inclusiveness and comprehensive spatial approaches are so important, one can question the desirability and even legitimacy of such 'neoliberal' ventures (Harvey, 1989). What seems to weigh less, as far as accounted for (Table 1), are 'softer' socio-cultural and collaborative aspects. The overall impression is that bonding, trust and solidarity are complementary aspects, which assist in sustaining more virtuous or vicious cycles. Our study thus clearly bears out the instrumental nature of intermunicipal collaborations.
In conclusion, we are convinced that the approach applied here presents a promising way to study these kinds of contextual and relational spatial phenomena. By focusing on combinations of conditions, as 'causes of effects in cases', QCA has allowed to discover a set of patterns (major and minor) and unique case positions fitting the complexity and variability of the phenomenon of intermunicipal collaboration. This also brings home the message that success and failure do not come from one-size-fits-all solutions or general best-practices. Policy insights stem from contextual knowledge of the combined roles and paths of enabling conditions in view of the particular trajectory of collaboration.