How rule directions influence actors to achieve collective action: an analysis of Dutch collective infrastructure decision-making

ABSTRACT While institutional analyses often assess whether collective action occurs, scant literature exists on key characteristics of institutional rules and how they affect collective action. Building on the Institutional Analysis and Development framework, this paper aims to identify rule directions and demonstrate how rule directions influence collective action. A rule direction is the particular direction that is provided to the decision-making process by the aggregated rules-in-use of a rule type. We compare two Dutch infrastructure projects, where collective decision-making either was or was not achieved. Our study demonstrates that rule directions provide a systematic and context-sensitive explanation for how institutions influence collective action. Collective action requires active agency regarding rule directions – i.e. (re)directing the diversity of positions, soft-closing the exit of boundaries, sharing and assimilating information, establishing process symmetry in aggregation, and managing reciprocity regarding scope – which can transform the institutional predisposition of sectoral action towards collective action.


Introduction
The organization of the spatial domain is characterized by coordination, cooperation and collaboration challenges: different societal needs are fulfilled by various organizations, such as governments and public-service agencies, with competing interests and choices (Feiock 2013;Hooghe and Marks 2003;Keast and Mandell 2013). Often, these organizations make independent choices in interdependent situations. Consequently, overseeing the planning domain and using common-pool resources effectively and efficiently becomes challenging; and as a result, social dilemmas and institutional fragmentation emerge (De Bruijne and Van Eeten 2007;Hardin 1968;Klijn and Koppenjan 2016;Reilly, Samuel, and Guikema 2015).
It is commonly argued that the solution to social dilemmaswhich are characterized by decisions that maximize individual short-term self-interests but yield a collectively inefficient outcomeis collective action (Feiock 2013). Collective action can generally be understood as the level, extent, or degree of collective activity to which 'the individuals involved intend to create a joint product or to achieve a common purpose' (Scharpf 1977, 54; see also Ostrom 2005;Poteete and Ostrom 2004;Theesfeld 2004;Varughese and Ostrom 2001). However, collective action may be hampered by institutional barriers created by institutional fragmentation (Olsen 2009;Shrestha, Berardo, and Feiock 2014). Hence, to study collective action in spatial domain implies by its very nature the study of institutions, which fits with the institutional turn in planning (Tennekes, Harbers, and Buitelaar 2013;Alexander 2005;Hall and Taylor 1996;Salet 2018;van Karnenbeek and Janssen-Jansen 2018).
The Institutional Analysis and Development (IAD) framework (Ostrom 2011) is a leading approach that aids a broad analysis for the systematical study, identification and categorization of institutions in relation to collective action in contexts where multiple actors affect commons. Institutions can be defined as durable systems of established, embedded and codifiable humanly devised rules and norms that inform actors to do X in circumstance Y, hence structuring social interactions (Crawford and Ostrom 1995;Hodgson 2006;North 1991). Institutions cover 'the entire continuum from legally enforced (regulative) institutions to informal but taken-for-granted (cognitive and normative) institutions' (Raven et al. 2019, 260). The IAD allows for many institutional rules-in-use to be identified per rule type, with qualitatively rich perspectives that defy (straightforward) quantification and a plurality of mechanisms that explain the impact of institutions on collective action. This means that finding a general set of conditions under which collective action may occur is challenging (Minkoff 2013;Provan and Kenis 2008), while simultaneously, in Ostrom's own words: 'Simply saying that "context matters" is not a satisfactory theoretical approach ' (2005, 287). While there is ample literature on the effects of publicness of goods, repeated interactions and reciprocity, group size and composition (Olson 1965;Hardin 1968;Axelrod 1984;Sander 2004), less is known about how key characteristics of each rule type of the IAD provide a particular direction to the decision-making process and may affect collective action.
Thus, a knowledge gap exists regarding the directions of rule types. A more intricate understanding and classification of rule types in rule directions, as well as how these directions may influence collective action, has the potential to enhance the systematic analysis of how institutions affect collective action, and to deliver more concrete and focused recommendations for public-service delivery. We focus on individual rule directions; whilst empirically interdependencies between directions exist, a prerequisite for understanding interactions is that individual directions are analytically understood. The aim of this study is, therefore, to identify what and explore how rule directions influence actors to achieve collective action. The empirical context of our study is Dutch infrastructure networks. They provide numerous potential situations for establishing collective renewal decision-making, which promises functional-spatial and financial advantageous alternatives to sectoral action of single infrastructure administrationsdue to the networks' high spatial density, functional interconnectedness and renovation need (Buitelaar and Bregman 2016;Neef et al. 2020). However, such collective action is also hampered by the sectoral orientation of the infrastructure administrations that manage different networks, and the institutional fragmentation of Dutch infrastructure planning (Heeres, Tillema, and Arts 2016;Hiteva and Watson 2019;Shrestha, Berardo, and Feiock 2014;WRR 2008). This challenge of major renewal in infrastructure networks in institutionally challenging settings is present and urgent in many countriesespecially those with scarcity of space, such as The Netherlands. The selection of the Netherlands as a case aids improving best practice because 'the Dutch system of landuse and transport planning is often highly regarded among academic and practitioners' (Pojabi and Stead 2014, 2;WEF, 2022). Studying collective infrastructure decisionmaking in this context holds the promise of enhancing collective action. Our research question is: What institutional rule directions can be identified and how do they influence collective action in the case of collective renewal in Dutch infrastructure decision-making?

Theory: rule directions
In this section, we use the seven rule types as distinguished in Ostrom's (2011) IAD framework as our basis to identify what rule directions may exist per rule. 1 We discuss how each of the particular rule directions may influence collective action. A rule direction may be understood as the direction provided to a decision-making process and collective action due to characteristics inherent to a rule type. Table 1 summarizes the section.

Position rules
Position rules establish the specific positions that actors can fulfil (Ostrom 2005). Each position constitutes 'a unique combination of resources, opportunities, preferences, and responsibilities' (McGinnis 2011, 174). Position rules may either converge or diverge the diversity of unique combinations in positions. Both convergent and divergent diversity may be important for collective action (Varughese and Ostrom 2001). For divergence to incentivize collective action, for example imagine a game of baseball, where position rules dictate at least two positions: the positions of the pitcher and of the batter. In this metaphor, collective action is unlikely to emerge when only the position of 'batter' can be fulfilled. Collective action also requires alternative positions to be fulfilled, i.e. the position of 'pitcher'. Contrastingly, for convergence to incentivize collective action, imagine a public participation process for area development where many mutually exclusive spatial claims are intertwined. Here, the complexity of the position diversity can inhibit benefitting from possible synergies of that diversity (De Roo 2003). Thus, position rules should allow for a mix of positions in order to be able to match the requirements of the specific situation at hand (Klijn and Koppenjan 2016).

Boundary rules
Boundary rules outline the conditions under which actors can access or leave a position. When conditions are assigned, boundary rules are closed; when no conditions are assigned, actors can enter and exit freely, and boundary rules are open. Open entry rules welcome position diversity, and vice versa, closed entry rules prevent further divergence and may encourage position convergence (depending on the specific exit rules) (cf. Cole and McGinnis 2018). Regarding exit, closed exit rules are seen to contribute to building commitment for collective action by constraining opportunistic behaviour (Klijn and Koppenjan 2016). Contrasting, closed exit rules are also seen to constrain participating in or achieving collective action, because an excessive number of participants can cause increased complexity (De Roo 2003) and 'exponentially rising and eventually prohibitive transaction costs' (Scharpf 1977, 70;Feiock 2013). Therefore, literature also argues that exit rules should be open so as to allow actors to leave the process if it no longer serves their interests (Cole and McGinnis 2018).

Choice rules
Choice rules specify what a participant occupying a position must, must not, or may do (Ostrom 2005). Actors commonly conform to their own institutionalized templates in order to be legitimate in their actions (Hodgson 2006;Olsen 2009). In a collective action situation, 2 actors may have to conform to divergent institutionalized templates. Consequently, actors may have to address conflicting goals or have mutually exclusive preferences and solutions (Feiock, Krause, and Hawkins 2017;Hooghe and Marks 2003;Shrestha, Berardo, and Feiock 2014). Choice rules therefore need to provide a flexible direction for addressing the specific task-at-hand (Feiock 2013;Hawkins, Hu, and Feiock 2016;Mullin and Daley 2010). As explained by Feiock (2009) (Shrestha, Berardo, and Feiock 2014, 51). The dilemma inherent to choice-rule directions therefore is to be flexible and adapt to the task-at-hand but be perceived as illegitimate, versus to be rigid and legitimate but not finding the conditions required for a specific ICA situation.

Information rules
Information rules regulate what information is sent and received (Ostrom 2005). As such, information rules can provide a restrictive or facilitative direction to information sharing. Shared information is possessed by multiple actors; unshared information is possessed by one actor (Laughlin and Ellis 1986;Schittekatte and Van Hiel 1996). Examples of information regulation are systematic regulatory oversight and knowledge management techniques (Reilly, Samuel, and Guikema 2015;Schittekatte and Van Hiel 1996). Facilitating shared information is important for collective action because it helps to overcome hidden profiles (Stasser, Titus, and College 1985, 426): 'In a hidden profile, a superior decision alternative exists but its superiority is hidden from individual group members because they each have only a portion of information that supports this superior alternative' (Brodbeck et al. 2007;Bushouse 2011;Chan et al. 1999;Fraidin 2004;Gigone and Hastie 1993). However, information sharing can be excessively facilitated, leading to an increase in either white noise or substantive complexity through 'report wars' where research provokes more and contradicting research (Veenma 2021;Klijn and Koppenjan 2016), which can disrupt and inhibit achieving collective action.

Aggregation rules
Aggregation rules influence how actors jointly affect collective decision-making by clarifying the influence of each actor when multiple positions have partial control over the same decision. In a collective action situation, actors have to give up some of their control, i.e. autonomy to achieve collective benefits that might not benefit each actor proportionally (Herzberg and Ostrom 1991). Aggregation rules can regulate this control either in a symmetric or non-symmetric direction. Symmetric rules value each participant's input equally (e.g. simple majority, or supra majority); non-symmetric rules allow some participants to exert more influence than others (e.g. weighted voting, or subgroups with chairmen). Symmetric aggregation rules can conduce collective action, because actors want to exert sufficient influence on the decision-making process (Ostrom 2005;Straffin 1977;Cox 2012;Levin and Nalebuff 1995). In contrast, a nonsymmetric direction may contribute to collective action when the participant that enjoys the relatively higher influence is especially interested in collective action. A special aggregation rule is the no-agreement rule that determines the outcome when no agreement is reached. No-agreement rules may either change or maintain a current situation (Ostrom 2005).

Scope rules
Scope rules define the solution space, the outcomes which may, must, or must not occur. This includes the timeframe in which decisions regarding these outcomes need to be made. Scope rules can have a narrow or a broad direction. For both directions, there must be some degree of homophily. Homophily means outcome preferences can match, interests of agencies and decision-makers can align, and goal consensus can occur (Gerber, Henry, and Lubell 2013;Gerber and Gibson 2009;Mullin and Daley 2010). The entire point of independent actors jointly producing policies is to achieve outcomes that they could not achieve alone (Hawkins 2010), allowing actors to capture positive externalities, economies of scale and to coordinate service provision (Minkoff 2013). In a narrow direction, outcomes are delimitated and the number of possible outcomes is limited. This direction clarifies whether and in what shape(s) homophily can occur, yet also excludes certain collective action possibilities. Contrastingly, in a broad direction, the number of collective action possibilities is expansive, yet what the actual collective action outcome may be is less clear and precise. Both directions may benefit from repeated interactions, i.e. when actors meet each other outside of the collective action arena. The reason is that rewards of defection decrease over time (Axelrod 1984;Hawkins 2010) due to, amongst others, reciprocity. In repeated interactions, social capital can grow and trust can rise allowing for greater assurance, reliability, commitment and loyalty (Graddy and Chen 2006;Hawkins 2010;Milward and Provan 2000;North 1991).

Payoff rules
Payoff rules establish the costs and benefits actors incur and yield. Payoff rules can provide a proximate or distal direction, both regulating the availability of resources. A proximate direction means that actors are directed towards resources that are readily available: actors have access to earmarked resources where these earmarks match the intended spending of the corresponding fund. Earmarks are a means to efficacious and appropriate spending of available budgetse.g. using a general rail infrastructure fund for renewal of rail infrastructure. In contrast, a distal direction means that the access to resources is strenuously available. In complex situations where ownership is not evident, earmarked funds may not be readily available. In interdisciplinary settings and settings where resources are scarce, distal payoff rules may invite actors to make efficient use of dispersed resources. Moreover, the willingness of an actor to try to access strenuously available resources is affected by the degree of interest actors have in a particular outcome: 'high-interest persons are motivated to do almost everything they can to ensure provision of the good' (Marwell andAmes 1979, 1356;Graddy and Chen 2006;Oliver 1980;Shadmehr and Bernhardt 2011). The ability to incur costs is further enhanced by a proper distribution of investment risk, which helps actors to share responsibility by allocating risks to those actors who are best able to manage specific risks (Cruz and Marques 2013;Shen, Platten, and Deng 2006). Finally, the ability to incur costs is strengthened by rewards. These 'positive incentives are especially efficient for motivating cooperation by a relatively small proportion of a group and, in many instances, generate pressures toward collective action by a small group of large contribution' (Oliver 1980(Oliver , 1373.

Methodological approach and case selection
We employed a comparative case study approach to identify rules-in-use and rule directions, and to analyze their effect on collective action in practice. Comparative case studies are especially useful for institutional analysis because they allow studying context-sensitive, complex action situations that span a longer period of time (Ostrom 2005;Theesfeld 2004). We differentiated the case outcome: in one case collective action was achieved, whereas in the other it was not. Moreover, we sought out cases with high contextual similarity to enhance comparability and learning.
To identify cases where multiple infrastructure administrations were involved in a single renewal opportunity, we employed research stays from November 2019 to March 2020. These also allowed us to familiarize ourselves with the organizational structures of the infrastructure administrations. Research stays are particularly useful for institutional analysis as 'obtaining information about rules-in-use requires spending time at a site and learning how to ask non-threatening, context-specific questions about rule configurations' (Ostrom 1999, 53), and because participants often understand rules-in-use tacitly and implicitly and may even refer to in order to explain and justify their actions (Ostrom 2005, 19). The first author was present at two infrastructure administrations -ProRail and the Port of Rotterdam Authorityeach for approximately two days per week for six months. ProRail is the government task organization that, as single concessionaire, is responsible for the management, maintenance, development and extension of the national railway infrastructure. The Port of Rotterdam Authority (PoR) is an unlisted public limited company that manages, operates and develops the Rotterdam port and its industrial complex for safe and smooth handling of all shipping (de Gooyert 2020).
The following case selection criteria were used: cases concern renewal in co-located networks of ProRail and PoR, cases concern infrastructures that were near end-of-life, and in one case the renewal had to be conducted collectively and in the other case not. The cases of the bridges 'Calandbrug' and 'Suurhoffbrug' were found to meet all criteria. Both bridges are crucial connectors in the railway and shipping networks of both administrators to their respective hinterlands, and impact their respective network capacities due to their aged status. The Suurhoffbrug carries a stretch of the highway A15, which also involved Rijkswaterstaat, the executive agency of the Dutch Ministry of Infrastructure and Water Management, responsible for the design, construction, management and maintenance of the Dutch main road network. We consider the Calandbrug to have achieved collective action because actors joined goals and resources, and the Suurhoffbrug to lack collective action because administrations only permitted that their infrastructure is affected by another administration, which is a precondition to any intervention in infrastructure networks in the Netherlands, but not collective action. Table 2 summarizes the cases.

Data collection
Data collection comprised document analysis and interviews. The former comprised searching news articles in general, and internal databases of ProRail and PoR, specifically for the cases, and selecting those documents that contributed to reconstructing a timeline towards the final collective action decision. The latter comprised semi-structured interviews: the sequence of questions on specific institutional rule types was adapted depending on cues by interviewees. The respondents were in part identified through the research stays 3 assisted by infrastructure administrators who helped the researchers to get acquainted with the organizationsand in part through snowball sampling. The interviews were conducted in July and August 2020. The interviews focused on the rules and rule directions that were present in the cases. The interviews were recorded and transcribed.

Data analysis
The interview transcripts and documents were coded using ATLAS.ti software. We used the rule types and rule directions as described in Table 1 as codes. We assigned codes when either the explicit or latent content of the data reflected the meaning of the  (Babbie 2013;cf. Spijkerboer et al. 2019). Next, we used the assigned codes to formulate the rules-in-use and classify the rules directions. 4 Thereafter, we assessed per case how each separate rule direction affected collective action (see Results), by assessing what rule direction was encountered in each case, and how that individual rule direction affected collective action. While an interdependency could exist when two codes applied to one textual element, we assessed separate effects. Finally, we compared that direction between the cases (Discussion), to conclude how a rule direction influences collective action (Conclusions). Table 3 shows the main results: the rule directions that were identified in both cases. Below, we describe why single rule types were classified as providing a particular direction, and how that rule direction affected to collective action.

Position rules
In both cases, the position rules stimulated diversity. The crucial difference, however, is whether and how the mix of positions affected collective action. In the Calandbrug-case, Facilitative information rules allowed actors to actively and mutually and collectively receive and send information, whereas restrictive information rules inhibited consistency of passing on information when alteration occurred of specific persons occupying a position, across disparate arenas and assimilation of that information Aggregation Nonsymmetric

Nonsymmetric
Infrastructure administrations commonly apply non-symmetric aggregation rules for outcome decisions. However, symmetry in process continuation conduced collective action, whereas nonsymmetric process aggregation rules did not Scope Broad Narrow Broad scope rules conduced collective action because they enabled problem and solution spaces to conjoin to homophily, whereas the narrow scope rules were unable to conjoin problem and solution spaces due to heterophily. Repeated interactions affected collective action differently as they established a path-dependent interaction rather than either positively or negatively incentivizing collective action Payoff Proximate Distal Proximate conduced collective action by allowing for both the willingness and ability to collectively incur costs and yield benefits, whereas distal payoff rules did not establish the willingness and ability to collectively incur costs and benefits the diversity of (civil-)technical and managerial positions at various organizations (Cp2; Cp3;Cp4;Cp5;Cp6;Cp7;Cp8) was joined to establish both technically sound and managerially feasible solutions (Ca1;Ca3). The Port of Rotterdam Authority (PoR) had a lead role (Cp10) in making a mix and match of the diversity towards collective action as they 'created a unity amongst the involved actors' (Ci#2), and subsequently this 'direction role got actors in motion' (Ci#4). In the Suurhoffbrug-case, a rich diversity of positions existed because many different thematic issues were considered throughout the Suurhoffbrug's long decision-making process (Sp2;Sp3;Sp4;Sp5;Sp7;Sp8). However, here the diversity was not effectively mixed and matched to collective action. The actor occupying the lead role (Sp10) was maneuvered into that position in the later stages of the process because 'that is the party who really must act' (Si#3). Subsequently, the lead actor was unsuccessful to create unity amongst the involved actors, and was preoccupied with its own interests rather than incorporating others' interests and getting actors in motion. Concluding, the mix and match of diverse positions conduced collective action in the Calandbrug-case partly by PoR's lead role, and the lacking mix and match of the many diverse positions did not conduce collective action in the Suurhoffbrug-case partly due to a lacking unifying lead position.

Boundary rules
In the Calandbrug-case, the infrastructure administrations experienced unconditioned access to the arena and all positions (Cb2;Cb3;Cb5;Cb6;Cb7)including the lead position (Cb9). In the Suurhoffbrug-case, however, not all infrastructure administrations were able to enter the arena (Sb5; Sb7). In fact, it was difficult to find a clearly identifiable arena regarding the decision-making about the Suurhoffbrug. One respondent stated that 'the administrations don't have any role at all' (Ci#6), i.e. no arena should exist with access for administrations before any official ministerial decision. Instead, decision-making was discussed in multiple, disparate, other arenas, to which not all actors were invited. Especially PoR felt excluded. For example, while all infrastructure administrators are 'involved in a National Meeting (in Dutch: Rijksoverleg) that is meant to prevent dissonances to occur between national administrations as early as possible (…), PoR is not involved' (Si#4). In conclusion, we can state that the dispersion of decision-making arenas created closed entry rules that were less welcoming to infrastructure administration (diversity) than the open entry rules. Next, in the Calandbrug-case, the infrastructure administrations experienced unconditioned possibility to exit, but the build-up commitment prevented actors from doing so. In the Suurhoffbrug-case, positions came and went and limited commitment was created as a certain liberty to exit existed (Sb4;Sb1). One respondent clearly describes this exit possibility: 'in the beginning, we wanted to collaborate with ProRail, but they quickly mentioned that renovation is time-wise too far away for us, so that is reason enough for us to not participate' (Si#7). Open exit even applied to the primary problem owner, Rijkswaterstaat, as the process of the Suurhoffbrug had been at a standstill from 2011 to 2014 (Si#1). Concluding, a closed exit emerged for the Calandbrug-case reflecting build-up commitment, which did lead to collective action. In the Suurhoffbrugcase open exit allowed actors to leave if their interests are not served by collective action. Therefore, as a separate rule direction, closed exit reflected progress towards collective action and open exit did not.

Choice rules
Infrastructure administrations commonly adhere to choice rules that provide a rigid direction (Cc2). For example, ProRail 'usually does everything regarding its own infrastructure, so nothing has to be decided except the minister who goes to ProRail and says here you go friends here is money now go build' (Ci#5). Nonetheless, in the Calandbrug-case, PoR was able to stimulate ProRail to allow for discussion on how their design norms were used and what problems were to be involved (Cc1;Cc3;Cc4), which created enough room for both infrastructure administrations to solve their issues. In other words, the flexible direction provided by the choice rules enabled collective action. Contrasting, throughout the Suurhoffbrug decision-making process, all actors strictly focused on their own obligations (Sp1;Sc1;Sc7) and consequently choice rules provided a rigid direction throughout the entire process. For example, ProRail adhered to their technical norms, their reserved train paths and utilization of the remaining lifecycle of their rail bridge, (Sc3;Sc4) because this sufficed to achieve their individual goals. Similarly, other respondents indicated rigidity by, on the one hand, explaining that lengthy planning procedures were not adapted: 'the administrations were, as figure of speech, taken hostage by the procedure' (Si#2). On the other hand, rigidity ultimately resulted in postponing action, a 'sticking-plaster approach' (Si#5;Sc2). Actions delayed are actions denied: sectoral action was the only possible option due to the end-of-life of the road bridge: 'At a certain moment you are tired of all those procedures (…) and if you are the last one to have the problem, then you take action' (Si#7). In conclusion, choice rules that provided a flexible direction incentivized collective action as joint renewal decisionsbecause they reflected a search for mutual problems, action and solutionswhereas a rigid choice rule direction had a postponing characteristic that first delayed and then denied collective action, and reflected a sectoral orientation.

Information rules
In the Calandbrug-case, information rules facilitated information sharing, which in turn helped to overcome hidden profiles (Ci3;Ci4;Ci6). PoR actively ensured that information rules were of facilitative nature. A case in point is that, initially, no information was shared that indicated the need for a renovation decision either within ProRail, within the Ministry or between these two actors (Ci1;Ci2). PoR first arranged meetings with ProRail to collect information themselves: 'we first made up our own mind' (Ci#3; Ci#1). PoR identified the hidden profile, being the collective action solution, through uniting all information: 'Finding an actor who can unite all perspectives in one solution, we coincidently happened to be that actor (…) I just happened to be in the position to identify the problem' (Ci#1). The active and facilitative information rule manipulation of PoR continued as a collective exploration of alternatives for solutions with ProRail and the Ministry and thereafter shared sending of information (Ci6;Ci7).
By contrast, the information rules for the Suurhoffbrug restricted information sharing. No active organization existed to facilitate information sharing. Rather, the information available for sharing declined over time due to the frequent change of specific persons who occupied specific positions (Si2). The result was that 'the Ministry did not even know the Suurhoffbrug was still on the agenda. (…) recently we had to point out on the map where that thing was for God's sake!' (Si#1). Moreover, when information was shared, this occurred within disparate arenas where no actor felt the responsibility to actively share information across arenas (Si1;Si3;Si4;Si5). Finally, the results indicate that just sharing information does not suffice: actors need to assimilate information as well. For example, PoR states that the Ministry, Rijkswaterstaat and ProRail do not understand the business climate consequences of one-on-one replacement: 'they do not know this case, hence they do not act accordingly' (Si#4). In conclusion, information rules that provided a facilitative direction permitted an actor to actively extract information from one organization to identify a hidden profile and then collectively send information, whereas information rules that provided a restrictive direction inhibited active information sharing and assimilation.

Aggregation rules
The aggregation rules provided a non-symmetric direction in both cases. Either infrastructure administrations or the Ministry had a veto for renewal decisions by respectively retreating their own assets and resources from the decision or by deciding to not assign a budget (Ca1;Ca2;Ca4;Ca5;Sa1;Sa2;Sa4). Effectively, therefore, collective action was only achieved upon unanimous agreement. A difference in aggregation rules between the cases concerns actors' influence on the continuation of the process. In the Calandbrug-case, support of either ProRail or PoR sufficed to further the process (Ca3). Hence, aggregation symmetry by simple majority existed to continue the process. By contrast, these process aggregation rules provided a non-symmetric direction in the Suurhoffbrug-case (Sa5). Illustrating, PoR indicated that 'the only thing we can do as stakeholder is to shout loudly and move people in The Hague at the Ministry to lobby for our interests' (Si#2). ProRail indicates that all influence is allocated to the Ministry: 'when push comes to shove, we have no say in it' (Si#4). Interestingly, Rijkswaterstaat experienced symmetry: 'we only make such a decision with support from other parties, (…) You must act together' (Si#5). However, as the road bridge's remaining time to its technical end-of-life declined, Rijkswaterstaat provided a non-symmetric direction to ensure safety of the bridge. Therefore, a symmetric direction to continue the process conduced collective action, whereas a non-symmetric direction of process aggregation rules did not.
Additionally, no-agreement rules vitally affected decision-making. Near the Suurhoffbrug's end-of-life, PoR experienced that the no-agreement rule had shifted (Sa6): 'We have always acted against a temporary bridge, but what if no bridge is realized and that boomerangs to us (…) Then we (…) really have something to explain' (Si#1). Subsequently, PoR reluctantly changed their effect on the collective decision to agreeing to a non-elevated temporary road bridge. In the Calandbrug-case, the no-agreement rule was that ProRail will conduct a like-for-like replacement if collective action is not achieved (Ca7). PoR indicated that they absolutely wanted to deviate from this outcome (Cs3), hence striving for collective action. Therefore, no-agreement rules steered the actors towards a solution that did not negatively affect their accountability, hence incentivizing collective action when they render sectoral solutions undesirable.

Scope rules
In the Calandbrug-case, the scope rules provided a broad direction. PoR and ProRail acknowledged the legitimacy of each other's problem space and constraints (Cs1;Cs2; Cs9;Cs10), and they allowed for the construction of a homophilic solution space that represents the desires (Cs3;Cs4;Cs5;Cs6) of both organizations. This broadness was further enhanced by repeated interactions between PoR and ProRail: their previous involvement in the Tweede Maasvlakte project initially identified the Calandbrug as key connection hence reducing their required effort for discovering joint activities. Another important repeated interaction, which reduced bargaining for joint activities, was the collaboration agreement between ProRail and PoR that 'makes it easier to make up your mind first and in the end produce an outcome in a suitable structure for the project' (Ci#2).
Contrasting, for the Suurhoffbrug, the scope rules provided a narrow direction. Heterophily persisted as actors did not simultaneously manage to join their distinct, sectoral problem spaces (Ss3;Ss4;Ss6;Ss7) in a collective solution space. Ultimately, the non-collective decision for a temporary road bridge was taken when there was not enough time left to complete complex collective decision-making procedures as compared to the remaining lifetime of the bridge (Ss8). This narrowness was further strengthened by repeated interactions that actually adversely affected collective action. PoR was frustrated because they expected reciprocity of a previous investment rather than a precedency: 'We invested a hundred something million euros, that is a boomerang we now get back (…) as others say 'you have invested before, so jump in this time as well' (Si#1). In conclusion, scope rules provided a broad direction supporting collective action as they enabled problem and solution spaces to conjoin and become homophilic, whereas a narrow direction of scope rules inhibited collective action through a heterophily that separated problem and solution spaces. Additionally, repeated interactions are a double-edged sword that established a path-dependent interaction pattern between actors that may either conduce or impede collective action.

Payoff rules
For the Calandbrug, the payoff rules provided a proximate direction. ProRail was able to request resources through earmarked MIRT financing (Cy2;Cy3). PoR forged the willingness of ProRail to do so, as their perspective changed from 'we as ProRail don't want a bypass (the chosen collective action solution, red.), so PoR has to arrange that with the minister' to 'Just renovation would lead to quite some disorder' (Ci#4)'. Hence, ProRail changed to wanting to make a substantial investment. PoR made a substantial investment themselves to indicate their major interest in the collective action solution (Cy4). Contrasting, the Suurhoffbrug payoff rules provided a distal direction (Sy1; Sy2;Sy3;Sy4). The willingness to collectively invest was insufficient, because the infrastructure administrations considered the potential sectoral benefits of collective action to be too disproportionate to their sectoral costs at any specific moment in time during the process. Ultimately, PoR considered the chosen solution of a temporary road bridge as a disinvestment: 'initially, €40 million was reserved, which is now almost €80 million. The situation for 2030 is thus already -€40 million. (…) And then in 2042 its ProRail's turn, and thereafter ours. It simply doesn't work like this, it is very unhealthy' (Si#2). Importantly, the ability to invest was constrained by a payoff rule that was paramount in both cases: 'Who pays, determines' (Cy5;Sy5). Collective action only occurred when all involved infrastructure administrations financially contributed. This rule dominantly influenced actors' ability to invest, because the specific earmarks of infrastructure funds determined who could access particular funds, to what end. Sectoral earmarks inhibited accessing budgets for collective ends: 'for the Suurhoffbrug, one part of the budget comes from the ministerial department "roads", and another from the ministerial department "public transport and rail", and when the scope changes, then the discussion is about whose budget is used' (Si#2). Finally, in the Calandbrug-case, a proximate direction of payoff rules conduced collective action by allowing for both the willingness and ability to collectively make costs and yield benefits, whereas in the Suurhoffbrug-case a distal direction of payoff rules did not conduce collective action as willingness and ability to collectively make costs and benefits lacked.

Discussion
In this paper, we set out to identify rule directions to explore how a multitude of rules-inuse affect collective action. Table 4 shows the empirically identified incentivizing directions and their descriptions. Below, we elaborate the ways in which separate rule directions affected collective action, additional to the ways identified in the theoretical section.

Rule directions and collective action: theorized mechanics
The rule directions for boundary entry rules, choice rules and payoff rules function in line with theory in both cases. Open entry rules welcomed diversity, whereas closed entry rules did so insufficiently. Flexible choice rules allowed actors to adapt to the specific task-at-hand, whereas rigid choice rules excluded sufficient action customization. Proximate payoff rules granted actors sufficient access to resources, whereas distal payoff rules did not. Moreover, for the collective action case, no additional mechanics were found for information and scope rules. This means that facilitative information rules conduced collective action because they allowed for sharing information, which in turn identified a superior decision alternative; broad scope rules conduced collective action because Proximity enhanced willingness and ability to access resources they allowed for homophily and because the repeated interactions reduced bargaining efforts for joint activities.

Rule directions and collective action: different mechanics
Position rule directions: active agents and diversity Our analysis demonstrates that a divergent rule direction is insufficient to conduce collective action. An active organizationi.e. a lead positionwas required to achieve collective action. This position actively redirected existing diversity towards unifying the potential of the resources, opportunities, preferences and responsibilities that actors bring to the decision-making process. Importantly, simply assigning an actor to be in the lead is insufficient: being maneuvered into a lead position only assigned responsibility for any action, not for unifying diversity towards collective action. Therefore, our data suggest that a divergent position rule direction can incentivize collective action when active agents reflect and act on how diversity relates to the institutional status quo and are tenacious in (re)directing diversity to collective action.

Boundary and information rule directions: actively organize commitment and information sharing
The results suggest that, similar to position rules, neither closed boundary exit nor facilitative information rule directions suffice to conduce collective action. An active agent who redirects those rules is required. Closed exit conduced collective action, not because closed exit built commitment, but because closed exit reflects built-up commitment: the active organization (of the lead position) built trust among actors that their interests would be served, hence preventing actors from exiting the decision-making process. This organization of commitment that closes the arena may be called a softlyorganized closed exit rule direction, which contrasts a hard-closed-exit rule direction of a formally closed arena. Our comparative case study also revealed that without an active organization of ensuring that interests will be served by collective action, open exit does allow actors to adapt to a changing context and may reduce complexity, but open exit does not build commitment and subsequently incentivize collective action. Therefore, our data suggest that the boundary exit rule direction that incentivizes collective action is soft-closed exit. Similarly, an active organization of information sharing promoted identification of a collective option that was superior to a sectoral optioni.e. a hidden profile. Importantly, a passive organization of information sharingi.e. making information accessible but not actively sharingreduced assimilation of information. Consequently, nonassimilation inhibited actors to mutually understand their problems. Moreover, without an active organization, disparate arenas inhibited information rules to take a facilitative direction. Therefore, our data suggest that the information rule direction that incentivized collective action is an actively organized facilitative direction. More generally, an active approach thus refers to the role of the active agent in institutional change where actors purposively reflect on whether the institutional configuration favours or disfavours their goals, and subsequently (re)direct their agency to attempt to break free from automatic rule-reproduction. In contrast, a passive approach refers to institutions as enduring, inert and sometimes even self-reproducing.
Aggregation rule directions: apply process symmetry We identified a non-symmetric aggregation rule direction in both cases. A final renewal decision itself can be vetoed by any actor. However, for the decision to be vetoed, the collective action process had to be continued to a stage when there actually was something to decide on. These aggregation rules for continuing the process were symmetric when collective action occurred, and non-symmetric when it did not. Because of process symmetry, the process reached a collective action decision moment in one case, whereas in the other case process non-symmetry hindered reaching a decision moment. Possibly, as with soft-closed exit rules, process symmetry reflects commitment and actors' trust that their interests may be served by collective action. Therefore, our data suggest that the aggregation rule direction that incentivizes collective action is a symmetric process direction.
No-agreement rule directions: discourage solitary alternatives Our analysis reveals significant influence of no-agreement rules. In both cases, no-agreement rules disincentivized collective action if the no-agreement solution negatively affected actors' their individual objectives. Hence, a characteristic of no-agreement rules is their discouraging effect. Thus, for no-agreement rules to make collective action possible, they should constitute discouragement of a sectoral alternative to decision-outcomes. Therefore, rule directions that may be identified for no-agreement rules are 'discourage' or 'encourage' solitary alternatives. Our study suggests that the former conduces collective action. Moreover, this understanding emphasizes the outcome; i.e. scope rule character of no-agreement rules. However, no-agreement rules are positioned within aggregation rules in the IAD (Ostrom 2005), because a change in no-agreement rules can change actors' vote on a decision. This study suggests that no-agreement rules may be more logically positioned under scope rules because noagreement rules concern the outcomes themselves.
Scope rule directions: actively organize tit-for-tat, keep time tight While our literature study indicated incentivizing effects of repeated interactions on collective action, our empirical study also demonstrated disincentivizing effects. Whereas one actor expected to be rewarded on the basis of previous investments, the other actor instead expected previous investments to be repeated. Hence, precedency prevailed over reciprocity and the 'tit-for-tat' element of repeated interactions is not self-evident. An interesting avenue for future research is to investigate the conditions under which repeated interactions incentivize or disincentive collective action.
Another finding in the scope rules concerns the timeframe. The conducive case achieved collective action in four years, whereas the non-conducive case achieved non-collective action after seventeen years. At first sight, it might be expected that more time is beneficial for achieving collective action due to the complexity of the cases. However, the reverse seems true in our study. The little time available for collective action actually incentivized actors to be more active in the conducive casea 'pressure cooker'whereas the plethora of time caused actors to first delay collective action, and later deny collective action through a temporary bridge. In fact, the now chosen temporary, sectoral solution may again be conceived as further delay of collective action. Therefore, this study suggests that a tight time frame may incentivize collective action, where a broad time frame does not.

Rule directions and interrelatedness
In addition to our study's focus on individual rule directions, our data point to interrelations between rule directions. A key example is the interrelation between a rigid choice direction and a narrow scope direction: a rigid choice means that actors adhere to their own legitimized template, and a narrow scope implies that actors adhere to their individual preferred outcomes. Rigid choice and narrow scope directionsand, conversely, flexible choice and broad scope directionsco-occur and influence each other influencing on collective action. A more systematic analysis based on a theoretical framework on rule (direction) interrelations promises to improve understanding how rule constellations influence collective action.

Conclusion
Social dilemmas are abundant in the organization of the spatial domaincertainly in infrastructure planning. Collective action is commonly considered to be the way forward. Institutions significantly affect collective action, yet research is often conducted on whether collective action occurs, while the set of characteristics of institutional rule types under which collective action can occur is less well understood. To increase the understanding of how institutional rule types provide direction to a decision-making process and influence collective action, we set out to identify rule directionsi.e. the direction provided to a decision-making process and collective action due to characteristics inherent to a rule type. These rule directions further operationalize the IAD. Our inquiry into collective action as infrastructure renewal decisions by multiple infrastructure administrations produced three main conclusions.
First, our study adds to the literature on the IAD by demonstrating the power of rule directions as an operational construct for analyzing how institutions affect collective action. We identified eight rule directions: converge-diverge (position rules), closeopen (boundary rules), rigid-flexible (choice rules), restrictive-facilitative (information rules), non-symmetric-symmetric (aggregation rules), narrow-broad (scope rules), distal-proximate (payoff rules) and discourage-encourage (no-agreement rules). The classification of institutions in their directions helps to move beyond the obvious statement that 'context matters'; rule directions both confine the institutional analysis and remain context-sensitive because the direction that may be expected to positively affect collective action can be studied per specific collective action situation at hand. For example, in one situation divergent positions may be expected to contribute to collective action, whereas in others convergent positions may be expected to do so.
Secondly, our study confirms and adds to contemporary literature on the characteristics of institutional rules and their impact on collective action in the particular case of joint infrastructure decision-making. Our analysis showed that rule directions of open entry, flexible choice and proximate payoff incentivize collective action, whereas rule directions of closed entry, rigid choice and distal payoff do not. Our research adds that in addition to a particular direction of position, boundary, information, aggregation and scope rules, these rule types' directions require active agents to incentivize collective action. Position rules require active unification of diversity; simply having the lead does not suffice for collective action. Boundary exit rules require a soft-closed exit to build up commitment; a hard-closed exit does not suffice. Information rules require an actively organized facilitative direction to share and assimilate information across arenas; just putting information 'out there' does not suffice. Aggregation rules require a symmetric process direction; symmetric aggregation itself does not suffice to progress the decision-making process to a collective action decision moment. Scope rules require repeated interactions that formulate clear expectations; otherwise, a precedent can be established instead of reciprocity. Additionally, our research suggests that no-agreement rules are better understood as scope than aggregation rule characteristics, and that their directions can be identified as 'discourage' or 'encourage' solitary alternatives, of which the former incentivizes collective action.
Third, for collective infrastructure renewal specifically, our findings highlight that preexisting institutional rules incentivize sectoral rather than collective action. Exemplary are boundary rules that establish disparate arenas, choice rules that first delay and then deny action, aggregation rules that require the demanding precondition of unanimity, and payoff rules that inhibit any actor who doesn't pay to co-determine outcomes. However, infrastructure administrations can turn this institutional predisposition around by actively reflecting on the institutional status quo and directing their agency towards incentivizing rule directions and towards joint decision-making. Additionally, the analysis yields that, particularly for infrastructure planning, a scope rule that attributes a little instead of (too) much time may act as a pressure cooker to incentivize collective action.
This study established firm ground for rule directions to analyze how institutions affect collective action. Still, 'a single optimal design of institutional rules does not exist' (Poteete and Ostrom 2004, 454). To enhance the robustness and understanding of rule directions, further research can test more precise conditions of when rule directions incentivize collective action. They should be further studied in different sectors of the spatial domain and towards different forms of collective action. Also, a temporaldynamic perspective may deepen our understanding of whether different rule directions are required throughout the process. Additionally, such a perspective may provide insights into how rules evolve over time, and may provide tools to better attune an institutional design throughout a decision-making process. Finally, the interaction between rule directionsi.e. the rule constellationmay be further studied to assess whether and how configurations of rule directions affect collective action.