How Do Institutions Address Collective-Action Problems? Bridging and Bonding in Institutional Design

Abstract

Collective-action problems affect the structure of stakeholder networks differently in policy settings (Berardo and Scholz 2010). However, interactions in policy settings do not usually occur in an institutional vacuum; instead, they are guided and constrained by agreed-on rules. Therefore, to better understand behavior in these settings, it is important to understand the parameters that guide and constrain it. Combining arguments from game theory and social network analysis, this paper focuses on how the nature of collective-action problems affect the design of formal institutional arrangements. The cases are two institutional arrangements for the provision of high-quality drinking water, in New York City and in Boston. The design of these arrangements is measured through Networks of Prescribed Interactions (NPIs), capturing patterns of interactions mandated by formal rules. NPI structures in each case are then compared analyzing their structural measures and applying exponential random graph models (ERGMs). By comparing these NPIs, the paper assesses the effects of collective-action problems on the design of formal institutional arrangements. Results show that cooperation problems are associated with designs prescribing redundant interactions that create a balanced distribution of responsibilities among the key actors to the agreement.

Keywords

social network analysis risk hypothesis institutional design institutional grammar tool

Introduction

Collective-action problems require different strategies to be addressed. When actors implement these strategies, it often results in specific and observable patterns of behavior. Researchers have used social network analysis tools to study how the type of collective-action problem affects the structure of policy and collaboration networks (Berardo 2014; Berardo and Scholz 2010; Feiock, Lee, and Park 2012; Lee 2011), or how the characteristics of different services (and the collective-action problems involved in their production or provision) influence the structure of networks of contracts for providing those services (Andrew 2010). In line with the “risk hypothesis” (Berardo and Scholz 2010), these studies have found that actors adopt specific network configurations depending on the nature of the perceived collective-action problem. When actors face “low risk” collaboration problems, their networks present bridging structures that facilitate coordination. On the other hand, when they face “high risk” cooperation problems, their networks adopt bonding structures that facilitate monitoring and enforcement.

However, the interactions studied by this literature do not always occur in an institutional vacuum, and are often guided and constrained by existing institutional arrangements. Formal or informal rules in place may encourage the creation of certain collaboration patterns. Therefore, it is important to disentangle the interactions that occur because a rule mandates them from those that are the product of other processes. Doing so would allow researchers to identify the extent to which the dynamics observed in policy or collaboration networks are a product of actor behavior, and which are a product of the institutions created to guide and constrain that behavior. Conversely, findings from the risk hypothesis literature can help define how institutions (rather than individual behavior) mandate interactions to address collective-action problems.

This paper focuses on how different types of collective-action problems affect the design of institutions created to address them. The research question in this paper is as follows:

Research Question 1: What patterns of interactions are prescribed by formal institutional arrangements when actors face different types of collective-action problems?

The collective-action problems studied take place in two cases referring to the access and provision of high-quality drinking water in New York and Boston. Settings where two or more governments must govern a shared natural resource are good examples where collective-action problems emerge (see, for instance, Baland and Platteau 1996; Ostrom 1990). In the two cases studied here, water providers developed institutional arrangements to access and provide high-quality drinking water sourced from other jurisdictions. The difference is that even though the goals of both arrangements are similar, actors in Boston faced different collective-action problems than actors in New York.

To answer the research question, the paper analyzes the context of collective-action problems faced in Boston and New York prior to the creation of their institutional arrangements, and then compares the patterns of interactions created by their arrangements. To do so, the paper introduces the notion of “Networks of Prescribed Interactions” (NPI) to capture the interactions mandated by a set of formal rules. NPIs are created by applying the institutional grammar tool (IGT; Crawford and Ostrom 1995, 2005) to a set of formal rules created to jointly govern a shared natural resource. By translating institutional design as a network, it is possible to assess whether institutions that were created to address cooperation problems mandate different patterns of interactions than institutions created to address coordination or division problems.

The main contributions of this paper are twofold: first, it provides evidence on the design of institutional arrangements in the face of different collective-action problems, showing that the risk hypothesis also applies to the design of formal rules. In particular, the paper shows that formal institutional arrangements mandate more redundant interactions among its members when they face cooperation problems. And second, the paper introduces a new tool for measuring and analyzing institutional design as networks. This is a step forward in the study of network dynamics in policy settings, highlighting the conceptual and methodological differences between observed behavior and actions mandated by formal institutions.

The paper is structured as follows: First, collective-action problems are defined, focusing on the differences between cooperation, coordination, and division problems. Second, the relationships between the parties involved in New York and in Boston are categorized using secondary sources and data from fourteen interviews with key actors. Third, a series of hypotheses are presented regarding how institutional arrangements can prescribe patterns of interactions to address different collective-action problems. Third, NPIs are created using the rules in the institutional arrangements in each case. Fourth, the New York and Boston NPIs are compared using structural network measures and exponential random graph models (ERGMs), showing that rules addressing cooperation problems foster more redundant and overlapping interactions than rules addressing coordination and division problems. Finally, a conclusion is presented discussing how institutions and behavior address cooperation problems in a similar fashion, and how this study opens the door to future research on the coevolution of behavior and institutional design networks.

Collective Action: What Is the Difference Between Coordination, Division, and Cooperation?

Collective-action problems involve two or more actors attempting to address a common concern, like the development and enforcement of rules for defining the use of a shared natural resource (Ostrom 2005, 219). These problems may assume different forms, for instance, when actors disagree on how to address such joint work and when following their self-interest may lead to socially suboptimal outcomes (Taylor 1987).

Game theory identifies three ideal types of collective-action problems by focusing on the sources of disagreement between actors and the consequences of their opportunistic behavior: coordination, cooperation, and division problems. Coordination problems occur whenever actors benefit from exchanging or producing goods or services because neither party could do so on their own. In these situations, transaction costs in the assignment of tasks and communication between the parties are the main cause of interpersonal risk, since a party acting in an uncoordinated fashion risks hampering the collective’s benefits. Division problems occur either when the achievement of specific gains will cause some parties to suffer more losses than the rest, or when parties have disagreements on how to distribute the costs or benefits of their joint work. The risk in division problems is in assigning retributions acceptable to all parties (Scharpf 1997). Finally, cooperation problems emerge when actors have different goals and face incentives to behave opportunistically. In this case, the risk is that behaving selfishly will yield fewer benefits than the costs paid by the group, leaving everyone worse off (Scharpf 1997).

In the last two decades, different literatures have highlighted how collective-action dilemmas affect institutional design and which policy tools are available to address them. For instance, the literature on contracts for service delivery in metropolitan areas has used transaction-cost economics and game theory to study how governance mechanisms address collaboration risks (Rodriguez, Tavares, and Araújo 2012). Authors have also analyzed how the production or provision of certain goods affect governments’ contracting decisions and contract design (e.g., Brown and Potoski 2003; Brown, Potoski, and Van Slyke 2016), or how actors create mechanisms to address potential collaboration problems. For instance, Maser’s (1998) study of 145 U.S. municipal charters analyzed how these documents included formal safeguards to address coordination, division, and defection problems.

Another approach, the Institutional Collective Action (ICA) (Feiock 2013; Feiock and Scholz 2010), incorporates ideas from transaction cost economics to understand how local governments define the production and provision of goods and services in metropolitan areas (Andrew and Hawkins 2012; Carr and Hawkins 2013). For the ICA, coordination, cooperation, and division are common problems arising whenever governments create joint arrangements (Carr and Hawkins 2013) and entail different types of risk between the actors involved, such as defection (when actors do not comply with the rules), division (when actors disagree in assigning costs or distributing benefits of their joint work), or coordination (regarding the aggregation of responsibilities for achieving complex tasks). Each problem requires specific responses: cooperative problems demand credible commitments that eliminate or reduce the possibilities of mutual defection (Scharpf 1997). Coordination problems require mechanisms that facilitate communication (Feiock 2013) so that interjurisdictional goods or services are provided efficiently. Division problems require mechanisms that favor decisiveness over the distribution of resources, such as hierarchical structures that can unilaterally allocate costs and benefits (Feiock 2013; Miller 1992).

The Cases: Collective-Action Dilemmas in the Production of Unfiltered Drinking Water

Two institutional arrangements for the provision of unfiltered water in the United States are analyzed to identify how they address different collective-action problems. U.S. water providers must filter their water prior to delivery to comply with federal regulations. However, since filtration is an expensive treatment, water providers can receive a waiver if their water is of the highest quality at the source, and they develop programs for maintaining quality and establishing ownership responsibilities (Surface Water Treatment Rule 1989). New York and Boston are two of the largest cities in the country operating under such a provision (Alcott, Ashton, and Gentry 2013; U.S. Environmental Protection Agency [EPA], 1999). In both cities, water is sourced from other jurisdictions, and the agency in charge of providing or managing drinking water (in New York largely through the New York City Department of Environmental Protection; and the Massachusetts Water Resources Authority in Boston) has developed formal arrangements with landowners in the watersheds to secure access and acquire land to implement water protection programs.

To identify the collective-action problems faced by actors in Boston and New York prior to developing their institutional arrangement, fourteen semistructured interviews were conducted by the author between June and August 2016 with organizations that are parties to the agreements in Boston and New York.¹ The interviews were complemented with secondary data. Interviewees were asked to describe the relationship between the parties prior to defining their current institutional arrangements for obtaining the filtration waiver. Also, they were presented with three hypothetical scenarios and asked which of the three better adjusted to their case. The scenarios were as follows: (A) We get along well, and we share the same goals, but we disagree on how to do and implement activities; (B) We generally agree on how to do things, but we tend to disagree on how to distribute the benefits of our joint work; or we disagree because achieving the goals of the agreement often entails that one of the parties will be more negatively affected than the other; (C) We tend to disagree on our perspectives regarding the use of the resource. In addition, if either party makes unilateral decisions, it will have consequences for the other party. This forces us to spend time making sure that the other party is doing what they are supposed to do. Each scenario represents the core dilemma occurring in coordination (A), division (B), and cooperation (C) problems. This helped contextualize respondents’ answers given the three categories of collective-action problems.

New York

New York City obtains its drinking water from the Catskill, Croton, and Delaware watersheds, 125 miles north of the city. Of the two cases studied here, New York has the longest history of conflict (see, for instance, Galusha 2016; Soll 2013). In the late nineteenth century, the State granted New York City the power of eminent domain to claim rights to watershed lands, which allowed for the construction of reservoirs and flooding of existing towns, forcing residents in the watersheds to relocate.

Throughout the twentieth century, the city and watershed communities were in constant conflict over tax payments for land owned by the city and over regulations curtailing local communities’ ability to use the watersheds (Galusha 2016). During this time, the city’s “refusal to accept the fundamental premise of local assessors—that its waterworks properties were substantially more valuable than the homes and businesses that dotted the Catskills—unnerved Catskill officials responsible for providing services to watershed residents” (Soll 2013, 134). As one interviewee mentioned, “There was a lot of animosity between the locals. They hated the city. I grew up hating the city” (Catskill Watershed Corporation representative, personal communication, June 17, 2016).

When asked about the relationships between the city and watershed communities during this time, interviewees highlighted the deep disagreement over how to govern the resource. Using the provided ideal types of problems, one respondent mentioned,

It’s been pretty well documented that folks did not agree, the upstate watershed communities and the city of New York. There was pretty much complete disagreement on what needed to be done and how they needed to be implemented in order to sort of protect the city’s drinking water supply, because obviously [the city] was looking at it strictly or unilaterally from a water quality perspective. And it didn’t really look at what the potential implications of what they were proposing would have on the communities and what would that mean for the future of these communities going forward. (New York State Department of Environmental Conservation representative, personal communication, July 12, 2016)

Even city representatives categorized it as a cooperation problem: “Prior to 1993 the third statement [option C] was probably the closest. I certainly think that the communities would say that the city acted unilaterally without any consideration for potential community impacts and without any real opportunity for the communities to have input” (New York City Bureau of Water Supply representative, personal communication, July 7, 2016). A representative from the watershed communities was even more explicit, stating that “it’s probably [option] C” (Coalition of Watershed Towns’ representative, personal communication, June 16, 2016).

In the early 1990s, to comply with the Surface Water Treatment Rule, New York city distributed a draft of their new watershed Rules and Regulations. These regulations were detailed and stringent, eliciting a negative response from watershed communities:

Suddenly, because of the [federal filtration mandate], the city drops new regulations out of the blue on the towns. No groundwork, no discussion. Suddenly “this is what we’re planning to do.” And they were incredibly draconian. They would have essentially forced out all agricultural uses, they had ridiculous restrictions on septic systems far greater than anything else on the state. Far higher requirements for treatment levels on wastewater treatment plants and anywhere else in the State, and no means to pay for anything. (Coalition of Watershed Towns’ representative, personal communication, June 16, 2016)

Local anger toward the city, paired with New York’s negligible interest in listening, fueled antagonistic interests. The city prioritized maintaining its unfiltered status at any cost, but the communities had no intention of collaborating under the city’s imposed terms. The city’s preference for individual rationality over group rationality joined with the antagonistic goals of the watershed communities resulted in a conflictive cooperation problem.²

The conflict would be addressed in 1997 with the signing of the New York City Watersheds Memorandum of Agreement (MOA), between the city, watershed communities, the State of New York, and the U.S. EPA. The agreement created a complex set of mechanisms for collective decision making and payment for watershed services, where the city provided the funding for economic and infrastructure development in the watersheds. One interviewee defined how the MOA addressed these collective-action problems by defining it as a “belt and suspenders” system, “to show that there’s an ample regulatory jurisdiction that the city has to comport with in order to make sure that it’s doing what it’s supposed to be doing” (New York State Department of Conservation representative, personal communication, July 12, 2016). Another interviewee added,

if something happened and the city was no longer able to acquire land, that could jeopardize the [filtration waiver], which would also then jeopardize the funding for the water quality programs. On the other hand, if the city didn’t pay for the water quality programs, that would jeopardize their [waiver] and also they would lose their Land Acquisition Program. (Catskill Watershed Corporation representative, personal communication, June 17, 2016)

Boston

Water in Boston is managed by a state and a regional agency: the Department of Conservation and Recreation (DCR) and the Massachusetts Water Resource Authority (MWRA). Unlike New York, the City of Boston is not in charge of demonstrating compliance with filtration regulations; rather, it is MWRA’s responsibility. In 2004, MWRA and DCR signed a memorandum of understanding (MOU) defining responsibilities for the management of the water sources.

Boston’s water is obtained from three sources: the Quabbin reservoir, the Wachusett reservoir, and the Ware River. These sources are on lands owned mostly by local governments, with a small percentage owned by MWRA and DCR. In this scenario, the city of Boston’s and watershed towns’ involvement is indirect at best, with DCR acting as the owner of protected lands and MWRA representing ratepayer communities in the metropolitan area and central Massachusetts.

Even though city interests (indirectly through a state agency) settled on eastern Massachusetts as their source of drinking water, differences between the water agencies and the watershed communities never reached the levels of antagonism seen in New York (Nesson 1983). A study conducted in 1995 surveyed 765 residents of the Wachusett reservoir, the most urbanized of the water sources, looking at attitudes toward watershed managers. Although they exhibited some distrust toward urban interests, a majority of residents expressed interest in maintaining a partnership with watershed managers to increase water protection (Steinberg and Clark 1999). When comparing the relationships between towns and watershed managers with the New York experience, an MWRA representative said, “We never had that meltdown. I wouldn’t say that we had a perfect relationship with the towns in the watershed, [but] we never got to that level of political disaster . . . It was a much more friendly [sic], a much more responsive relationship” (MWRA representative, personal communication, June 3, 2016). This does not mean the total absence of conflict between local communities and watershed agencies, but it is a good indicator of the lack of an “overt antagonism” (Steinberg and Clark 1999).

The current bipartite arrangement for managing Boston’s water was not always this way. For almost sixty-five years, water sources were single-handedly managed by the Metropolitan District Commission (MDC), a state agency. In 1984, MWRA assumed responsibility for all water and sewer, while MDC managed the watersheds. Finally, in 2003, DCR replaced MDC. The personal relationships among employees of these agencies, built on a shared administrative past, facilitated collaboration between the two organizations (Water Supply Citizens Advisory Committee representative, personal communication, August 15, 2016). One interviewee defined the history of relationships between DCR (previously MDC) and MWRA by saying: “I would say that [option] A is the closest. In that we get along well, we share the same goals, and we occasionally disagree on how to do and implement things. It’s much more narrow [sic] than that” (MWRA representative, personal communication, June 3, 2016). Other respondents found it harder to place the history in one of the given categories; a DCR representative mentioned, “boy, I’m not sure any of those [options] fit . . . . I’d say the first one, but only we sometimes disagree” (DCR representative, personal communication, June 11, 2016).

In the early 1990s, MDC and MWRA faced conflicts arising from differences in their organizational structure and the operationalization of their joint mandate (Roberts 1990). According to one interviewee, in the 1990s “they [MDC] were making all of the decisions, and conversely weren’t able to accomplish what they needed to accomplish because they didn’t have the resources to do it” (MWRA Advisory Board representative, personal communication, July 14, 2016). On one hand, conflict came from the side of funding responsibilities: state legislation (Chapter 92A ½ of Massachusetts General Laws) mandated MWRA to fund MDC/DCR, but curtailed MWRA’s ability to determine how those funds would be used. This caused frustration within MWRA, as one interviewee attests:

When [Bill Parcells] was coach of the Patriots, there was this whole thing, and when he left, he said: “you know, if you’re not going to allow me to pick the groceries, how can I make the dinner?” So, similar, the frustration arose dramatically that if we [MWRA] are not allowed, even though we’re spending the money, we’re not allowed to participate in the decisions, how are we going to progress? (MWRA Advisory Board representative, personal communication, July 14, 2016)

These division problems arose because even though costs were assigned, the benefits of participating in decision-making were not equally distributed among the parties. In response, the state legislature created a Water Supply Protection Trust in 2004, placing DCR under the scope of the newly created trust. This allowed MWRA more control over DCR’s budget requests and insulated DCR from state-wide managerial decisions.

The other source of conflict was the operationalization of their shared goals. During this time “there were places where we [MWRA] had a crew and a land mower, but didn’t have any land, and other places where they [MDC] had a crew and a land mower and we had land” (MWRA representative, personal communication, June 3, 2016). Roberts (1990, 100) defined this coordination problem, writing, “although jurisdictional boundaries . . . are specified in their enabling legislation, the resulting operational interpretation became a valid controversy because the two organizations share similar goals.” This also came up when asking interviewees about the main purpose of the institutional arrangements. New York respondents mentioned that their agreement allowed them mostly to control for noncompliant behavior, whereas Boston respondents emphasized the need for organizational clarity:

what we were trying to do with the first couple of versions of the MOU, was make sure that everything that needed to be done was being done by someone and that to the extent that we could make that efficient . . . we try to limit the number of places where we both deal with our job. The DCR is responsible for the land around the water . . . once the water leaves the reservoirs, it’s clearly our [MWRA] job. (MWRA representative, personal communication, June 3, 2016).

Even though the two agencies managing Boston’s water had collaborated extensively, conflict arose over the funding and operationalization of shared goals, in the form of coordination and division problems. Unlike in New York, however, these problems were not due to conflicting goals, but differences on how to implement shared goals. Table 1 summarizes the main features of both cases.

Table 1.

Features of Each Case.

City	Actors involved	Dominant dilemma
Boston	MWRA and DCR	Coordination with elements of division
New York	New York City, New York State, U.S. EPA, Watershed communities, Environmental NGOs	Cooperation

MWRA = Massachusetts Water Resource Authority; DCR = Department of Conservation and Recreation; U.S. EPA = U.S. Environmental Protection Agency.

Collective-Action Problems and Their Effects on Institutional Design

Myriad studies in the past two decades have analyzed how collective-action problems affect behavior. In the policy networks literature, the “risk hypothesis” (Berardo and Scholz 2010) posits that actors select ties to mitigate interpersonal risk. “Low risk” coordination dilemmas occur whenever actors share the same goals, have little incentive to defect, but have trouble agreeing to a shared course of action (Berardo 2014). Networks in these settings tend to adopt bridging structures that allow accessing and distributing unique information. In high-risk cooperation situations where actors do not share the same goals and have little incentive to collaborate, however, networks adopt bonding structures with more redundant and overlapping structures that allow controlling the veracity of information.

Since cooperation problems require reducing defection, institutions designed to address them should favor the generation and distribution of redundant information to monitor and sanction opportunistic behavior. In these circumstances, the literature has found that actors opt for developing bonding relationships (Berardo 2014; Berardo and Scholz 2010) that increase an actor’s ability to influence or control its neighbors (Berardo et al. 2016). In addition, bonding relationships have been found to help actors share risk and adapt to environmental hazards (Henry and Vollan 2014). A common way of increasing monitoring and influence is by ensuring that an actor can have direct and indirect access to information about the behavior of its peers (Berardo 2014). At the most basic structural level, actors may seek to reciprocate linkages to obtain information about their peers directly from them. This results in networks with a high level of redundancy. However, if there are reasons not to trust the peer, an actor may seek to contact the peer’s contacts to add more information about the peer’s behavior. In consequence, a network in this scenario should present a high number of closed triads, where actors seek indirect information about their peer’s behavior by contacting the peer directly and also by contacting the peer’s contacts.

In coordination problems, the source of the dilemma falls not on the possibility of defection but on the risk that uncoordinated actions may have for the collective. In these situations, actors must be aware of the group’s actions, to align their actions to them and effectively coordinate with each other. Doing so requires that information reaches everyone in the network quickly and efficiently, which can be achieved through structures that facilitate bridging over bonding (Berardo and Scholz 2010).

Division problems are similar to coordination in that the dominant strategy is to collaborate (as opposed to self-interested behavior). The difference is that, in division problems, efficient communication is not enough to address the problem. Instead, institutional arrangements addressing division problems should rely on mechanisms that facilitate quick decision-making. In network terms, however, behavioral responses to coordination and division problems should yield similar network structures. For different reasons, both types of problems are addressed through high levels of centralization where a small group of actors occupy central positions. Centralization can facilitate the spread of information by having a core of actors centralizing and distributing it (Berardo and Scholz 2010), and the transaction costs of decision-making can be reduced by having a centralized authority (Feiock 2013) deciding and communicating their decisions to the rest of the network (Scholz, Berardo, and Kyle 2008).

By mandating how actors must or may interact, rules create patterns of expected interactions, reflecting how the institutional arrangement defines the solution to the underlying problem. The logic of the risk hypothesis can, thus, be used to understand institutional design and how institutions address specific collective-action problems. As stated above, the research question in this paper is what patterns of interactions are prescribed by formal institutional arrangements when actors face different types of collective-action problems? This paper introduces the notion of NPIs to capture how formal rules mandate specific relations between actors. NPIs capture how rules create patterns of information flow, resource exchange, and relationships of authority between actors, by looking at formal rules.

Bonding allows access to multiple sources of information, increasing the overall redundancy of linkages in the network (Andrew and Carr 2013). This results in dense and overlapping networks. Bridging, on the other side, facilitates information sharing and reduces the transaction costs of decision-making by creating efficient channels for the distribution of information and responsibilities. As a result, a few actors will possess linkages to several alters in the network, connecting otherwise disconnected actors and facilitating efficient access to distant parts of the network (Berardo et al. 2016). These concepts can be understood in institutional design terms: bonding will result in designs where most parties to the agreement will be involved in most aspects of the institution, resulting in redundancy and overlap in the activities to be performed. On the other hand, more bridging would result in institutions that favor the distribution of responsibilities and information among the parties, reducing overlap and establishing clear lines of authority.

Hypothesis 1A: NPIs created to address cooperation problems will characterize by bonding network structures, with redundant and overlapping ties between actors.

Hypothesis 1B: NPIs created to address coordination or division problems will characterize by bridging structures, with a small group of actors occupying central positions and connecting otherwise disconnected actors.

Institutional design can also make some actors more popular and active than others. Popular actors are those who receive several ties from all actors in a network, whereas active actors are those who send many ties to everyone else in a network. The risk hypothesis and the general literature on behavioral policy networks highlight the importance of resource differentials in making some actors more active or popular than others (for instance, Henry 2011 or Leifeld and Schneider 2012). This effect is often observed for governmental actors in policy networks because of their access to larger organizational resources. Governmental actors often have the authority to make and impose decisions over others in a network, thus making them the target of other actors in the network who want to influence those decisions. Those effects, however, have been observed in behavioral networks where actors have flexibility in deciding who to interact with. In the case of NPIs, relationships are mandated by formal rules, so actors cannot change them in the short run. Moreover, the cases studied here focus on intergovernmental agreements, and, thus, all (or most) actors in these NPIs represent a level of government. Notwithstanding, NPIs are a product of their social and political contexts and of the power imbalances between actors with access to different resources, and, thus, they should reproduce some of these effects.

The cases studied here have one common denominator: a metropolitan water provider interacts with federal, state, or local actors to protect water quality sourced from other jurisdictions. Depending on the nature of the dilemma, institutions should divide responsibilities differently. Since New York city and the MWRA have a special interest in making the agreements work, the expectation is that actors related to New York city and the MWRA will be more active and popular than the average actor in the NPI. However, in the face of cooperation problems, actors want to create mechanisms to prevent opportunistic behavior. Therefore, their institutions should evenly assign responsibilities to prevent a group from imposing its decisions on the other. According to this argument, since actors respond to cooperation by creating redundant linkages, the level of popularity and activity of city actors should be indistinguishable from others in the NPI. For coordination and division problems, since actors share similar goals, their concerns are less about one group imposing its decisions on the other and more about facilitating information-sharing and decision-making. In consequence, institutions should assign specialized roles for each actor, with some actors presenting higher levels of centralization than others.

Hypothesis 2A: Because they are facing mostly cooperation problems, New York city actors will be equally as active and popular than noncity actors in the NPI.

Hypothesis 2B: Because they are facing mostly coordination and division problems, either MWRA or non-MWRA actors will be more active and popular than their counterparts in the NPI.

Method

The goal of this paper is to assess how collective-action problems affect the design of institutions. One way to measure institutional design is by observing how rules mandate relationships between groups of actors, which allows for identifying how rules provide solutions to underlying collective-action problems. The formal rules in each case were distinguished by asking interviewees to indicate the main documents guiding and constraining their behavior regarding watershed protection. Table 2 lists these documents and their roles within each institutional arrangement.

Table 2.

Rules Used for the Creation of NPIs.

Case	Document (year)	Purpose
Boston	Watershed Protection Act (1992)^a	Defines responsibilities between MDC (now DCR) and MWRA
	DCR Watershed Protection Regulations (1994)	Operational rules for protecting water quality on watershed lands
	Memorandum of Agreement between DCR and MWRA (2004)	Clarifies jurisdictional and operational responsibilities
	Water Protection Supply Act (2004)	Creates Water Supply Protection Trust
New York	New York City Watersheds Memorandum of Agreement (1997)	Defines the relationship between city, watershed communities, regulators, and environmental NGOs regarding watershed management
	New York City watersheds rules and regulations (2010)	Operational rules for protecting water quality on watershed lands
	New York State Department of Environmental Conservation Water Supply Permit (2014)	Authorizes the city to acquire land from willing sellers in the watersheds

NPIs = Networks of Prescribed Interactions; MDC = Metropolitan District Commission; DCR = Department of Conservation and Recreation; MWRA = Massachusetts Water Resource Authority; NGO = nongovernmental organization.

Amended in 2003 after the creation of the Department of Conservation and Recreation (DCR).

Each document was coded applying the IGT (Basurto et al. 2010; Crawford and Ostrom 1995, 2005; Siddiki et al. 2011), which provides a systematic and theoretically driven approach for the analysis of formal rules. The IGT was chosen because it provides a level of granularity in the analysis that for each rule allows identifying the actors involved, the nature of the relationship defined by the rule, and the contexts in which that relationship should, may, or should not occur. This level of detail helps capture the relationships that according to the institutional arrangement should address the underlying collective-action problems.

The unit of analysis in the IGT is an institutional statement, which often adopts the form of a sentence in a formal document. Within each institutional statement, the IGT identifies the actor who must (or not) do a certain action and who (or what) is the recipient of that action. The “doer” of the action is termed Attribute by the IGT, and the recipient (usually another actor or inanimate element) is defined as Object.”³ For example, an institutional statement in the 2004 Boston MOU states that “On a quarterly basis, the WATERSHED AGENCY shall provide MWRA with a progress report identifying the activities undertaken in the preceding three months toward implementing the Annual Work Plan” (2004 MOU). In this example, the watershed agency is the Attribute in charge of providing MWRA (the Object) with a progress report. By defining who must do an action and the recipients of those actions, rules carry an intrinsic relational purpose, making it possible to create dyads between the Attribute of an institutional statement and its Object. These dyads form the fundamental relational structure for the creation of NPIs.

However, not all institutional statements create linkages between animate actors. For instance, a statement might prescribe that someone must build a bridge, so the bridge is the Object in the statement. In addition, rules often forbid certain relations. To create NPIs, only statements mandating positive relations (i.e., those that may or must occur) and linking animate actors were considered. Table 3 lists the percentage of total statements used in the creation of both NPIs.⁴

Table 3.

Institutional Statements Used for the Creation of NPIs.

	Institutional statements	Statements used for NPI (% of total)
Boston	570	129 (22.6)
New York	2,083	675 (32.4)

NPI = Networks of Prescribed Interaction.

After identifying the dyad in each institutional statement, they were aggregated into a Network of Prescribed Interactions (NPI) capturing relationships mandated by a set of formal rules. Figure 1 provides a graphical representation of the Boston and New York NPIs. Both networks are directed and have been dichotomized for analytical purposes.

Figure 1.

Networks of Prescribed Interactions in Boston and New York.

Results

Both hypotheses were addressed by analyzing the structural coefficients of each NPI, and by fitting an ERGM on each. The structural coefficients, listed in Table 4, provide a straightforward approach for analyzing the topography of each network. However, structural coefficients can be influenced by factors such as size (number of actors) and network density. As a result, it may be difficult to distinguish between effects caused by network dynamics from effects that are a byproduct of size or density (Anderson, Butts, and Carley 1999). To control for this, ERGMs (Lusher, Koskinen, and Robins 2013; Robins et al. 2007) were fitted to each NPI. Rather than comparing structural coefficients between networks of different size, ERGMs offer a way of comparing a network’s features with those of multiple randomly generated networks of similar characteristics, providing a more nuanced approach to assess their structures.

Table 4.

Measures Used to Assess Concepts of Interest.

Network concept	Measures used
Bonding	Transitivity, reciprocity
Bridging	Degree centralization, betweenness centralization
Actor activity	Out-degree
Actor popularity	In-degree

Structural coefficients (Table 5) show higher bonding scores in New York than in Boston. Reciprocity captures the proportion of dyads with reciprocal ties (a tie is reciprocated whenever a tie i → j is joined by a tie i ← j). Rules in New York mandate that ties should be reciprocated in 48.9 percent of the cases. This coefficient presents a stark difference with Boston, where only 18.5 percent of all mandated interactions are reciprocal. Such difference is also present for transitivity, which captures the percentage of paths of length two (ties connecting actors i ↔ j ↔ k) for which a tie between i and k is present, “closing” the triangle. In New York, rules prescribe three times more the transitive interactions (28.9%) than in Boston (9.3%). The reciprocity and transitivity values are in line with theoretical expectations about bonding.

Table 5.

Structural Measures.

Measure	Boston	New York
Number of nodes	82	139
Density	0.024	0.027
Reciprocity	0.1852	0.4893
Transitivity	0.0931	0.2821
Degree centralization	0.2976	0.3036
Betweenness centralization	0.2576	0.2752

To assess bridging, two centralization measures were used that capture how much a network depends on a small number of actors. Degree centralization focuses on how evenly distributed degree levels (number of ties) are among the nodes in a network. An NPI with high centralization will concentrate its institutional statements on a small number of actors, granting them a role in most of the activities mandated by the agreement. This is expected to occur in coordination and division NPIs, where the main goal is not controlling defection but to facilitate information flow and decision-making.

Since these are institutional design networks, an actor’s centrality in the arrangement may also depend on something more than the direct linkages it sends or receives. An alternative measure focuses on an actor’s ability to connect otherwise disconnected nodes through the shortest path. This measure, termed betweenness centrality, is another common indicator of bridging in behavioral networks (Scholz, Berardo, and Kyle 2008). An NPI with high betweenness centralization would indicate that the arrangement fosters specialization of responsibilities by placing actors in positions that keep the agreement together, especially in the absence of actors with high levels of degree centrality. For instance, an agreement may not concentrate responsibilities on an actor through centralizing activities but rather by granting that actor the ability to control information flows. Both scores were obtained using the centralization routine in the igraph package for R (version 1.1.2) (Csardi 2015; Csardi and Nepusz 2006) and normalized by dividing the observed values by the maximum theoretical score for a network with the same number of nodes. New York presents higher centralization than Boston in both measurements. These differences, albeit small, do not follow the theoretical expectations.

Hypothesis 2 focuses on the role of city and MWRA actors in each NPI because they are the most interested in securing access to and managing the resources. Unfiltered water providers must prove that they own or manage their watershed lands; otherwise, they risk losing the waiver.

Table 6 presents a comparison of average in-degrees (incoming ties) and out-degrees (outgoing ties) for city or MWRA actors versus noncity or non-MWRA actors. Out-degree and in-degree in an NPI are associated with whether an actor occupies the role of the attribute or object in an institutional statement. Depending on the context of each institutional statement, an argument can be made about whether the statement is granting formal power to the attribute or the object. Because this paper focuses on the overall structure of mandated interactions, the focus is on whether actors occupy central positions both as attributes and objects, which would indicate they are granted a predominant role in the overall design of the arrangement.

Table 6.

Degree Coefficients.

	Boston		New York
	MWRA	Non-MWRA	City	Noncity
Number of actors	5	77	13	126
Average in-degree	2.4	1.948	4.385	3.635
Average out-degree	4	1.844	4.692	3.603

MWRA = Massachusetts Water Resource Authority.

The noncity actors’ category groups actors not directly associated with New York city or with MWRA, which includes other municipalities; federal, state, and local agencies; and nongovernmental organizations. In both cases, city and MWRA actors have higher average out-degrees than their counterparts. Moreover, the difference between city and noncity actors’ in-degree in New York is larger than in Boston, at odds with the theoretical expectation. Furthermore, t-tests for the difference in means for in-degree and out-degree showed no statistically significant differences in either case, not supporting Hypothesis 2.

ERGMs, although usually applied to identifying the underlying dynamics of tie formation in a network, can be used to assess a network’s structural features. In this paper, the application is conceptually different in that the goal is to assess whether specific structural features are present, rather than explaining partner selection dynamics in a network.⁵ Moreover, the nature of NPIs requires mentioning two caveats in terms of ERG modelling: first, coefficients should not be interpreted as the probability of an actor creating a tie with another but as the presence or absence of ties mandated by formal rules. This conceptual difference is because NPIs do not capture patterns of behavior; instead, they capture mandated interactions. Second, model parameters are chosen based on theoretical expectations and tangentially on which parameters allow a model to converge (reproduce characteristics of the network of choice in the randomly generated networks that will work as the baseline for comparison).⁶ The NPIs studied here differ in size and density (see Table 5), and therefore, finding parameters that consistently converge across models is a challenge. Models in Table 7 directly test Hypothesis 1a and Hypothesis 2 while providing the best goodness-of-fit among other similar model configurations.

Table 7.

ERGM Coefficients.

	Boston	New York
Edges	−4.4252***	−5.308***
Mutual	1.1431**	2.704***
Transitive ties	1.5541***	1.899***
Out-degree city/MWRA	0.4882*	0.1318
In-degree city/MWRA	−0.2017	7.825e-05

ERGM = exponential random graph models; MWRA = Massachusetts Water Resource Authority.

p < .05. **p < .01. ***p < .001.

ERGMs were calculated with five parameters using the ergm package for R (version 3.8.0) (Handcock et al. 2017; Hunter et al. 2008). These parameters can be interpreted as the coefficients of a linear logistic regression. The edges parameter is a baseline coefficient capturing the observed number of ties in the network and works as an intercept in the model. Reciprocity and transitivity capture two basic bonding structures: reciprocity focuses on the number of reciprocal ties, with a positive and statistically significant effect on both NPIs. The New York NPI has the highest relative coefficient, supporting Hypothesis 1a. Transitivity focuses on the number of dyads i → j also connected by a two-path i → k → j, identifying the presence of overlapping ties. The parameter behaves according to Hypothesis 1a; the New York NPI has the highest propensity to form transitive ties.

Finally, the model includes two centrality parameters for city and MWRA actors in each NPI. The out-degree city/MWRA parameter indicates the propensity of city or MWRA actors to send ties, whereas in-degree city/MWRA focuses on the propensity of city and MWRA actors to receive ties. These effects are relative to noncity actors. The ERGMs show statistically significant effects only when MWRA is prescribed to send ties, and no significant effects for New York, supporting Hypothesis 2.

Discussion

To date, no study has tested whether the nature of the social dilemma affects how rules mandate how actors should interact. This paper shows that just like in behavioral networks, institutional design is similarly affected by cooperation dilemmas by favoring redundancy in the assignment of responsibilities.

When actors face cooperation problems, their formal institutional arrangements tend to create more redundant and reciprocal interactions. On the contrary, when actors face a mix of coordination and division problems, arrangements prescribe fewer redundant interactions and assign responsibilities unevenly. This finding is in line with the risk hypothesis (Berardo and Scholz 2010).

Interestingly, NPIs showed positive and statistically significant coefficients for reciprocity and transitivity even in the absence of cooperation problems. This finding may indicate the presence of a common baseline of redundancy in institutional arrangements between governments. Recent studies have argued that the difference between bonding and bridging may not be antagonistic; networks may simultaneously address both cooperation and coordination problems (Angst and Hirschi 2017; Berardo et al. 2016).

In Hypothesis 2, rules in the cooperation case assigned indistinguishable amounts of responsibilities for city actors versus noncity actors. In Boston, the institutional arrangement did assign MWRA actors a significantly higher level of out-degree. This result should be further investigated for two reasons. First, it should be assessed whether the finding is a byproduct of having only two main actors involved, since the New York agreement includes multiple actors. Second, research should address whether this finding is present only in cases where actors experienced coordination and division problems, or if it is also present in cases of coordination-only or division-only problems. This invites more theoretical work to understand how we can identify coordination and division problems and how actors address them in practice. Although the game theory literature clearly distinguishes between the two (for instance, Scharpf 1997), other authors define bargaining and division problems as a subtype of coordination dilemmas (see Miller 1992), thus highlighting that more precise definitions are required to distinguish the two in real-life scenarios.

Conclusion

The literature studying how governments respond to different collective-action problems (i.e., Feiock 2013) has linked different institutional designs to specific collective-action problems. However, no study thus far has analyzed institutional design in terms of the relations prescribed by formal rules, and how these relations are structured in the presence of different collective-action problems. This paper analyzed whether different collective-action problems result in variations in the design of formal institutional arrangements. Following arguments from the policy networks literature, two sets of hypotheses were developed focusing on whether cooperation problems require different institutional designs than coordination and division problems. A novel approach was developed to address these hypotheses, capturing how formal rules prescribe interactions among actors participating in an institutional arrangement for governing the sources of high-quality drinking water in New York and Boston. Findings show that cooperation problems result in more redundant and overlapping networks than coordination and division problems. In addition, since coordination and division problems do not require a high level of redundancy, rules addressing these problems grant certain actors more responsibilities than others.

This paper provides both methodological and theoretical contributions. Methodologically, it applies a novel approach for analyzing institutional design, by understanding the interactions prescribed by formal rules as NPIs. Theoretically, this paper discussed the effects of types of collective-action problems on the design of institutions. Results have shown that institutional design responds to collective-action problems in a way similar to how individual behavior does it. This sheds light onto how patterns of interactions observed in the behavioral world might affect and be influenced by patterns of interactions prescribed by formal institutional arrangements. In the end, the bonding and bridging effects observed by the policy networks do not occur in an institutional vacuum and are rather influenced by formal rules in place. The NPI approach and the findings in this paper are the first step in disentangling that interdependence. By “translating” rules into networks, NPIs have the potential to help analyze and explain dynamic processes of institutional change and evolution. For instance, an analysis over time of NPIs and behavioral networks could explain the coevolution of behavior and institutional change.

Finally, results also highlight the need for more empirical and theoretical work on the effects of division and coordination dilemmas on institutional design. Specifically, more conceptual work is needed to refine the expected effects of coordination and division problems in cases of pure coordination and pure division.

Supplemental Material

Olivier_Replication_Data – Supplemental material for How Do Institutions Address Collective-Action Problems? Bridging and Bonding in Institutional Design

Supplemental material, Olivier_Replication_Data for How Do Institutions Address Collective-Action Problems? Bridging and Bonding in Institutional Design by Tomás Olivier in Political Research Quarterly

Footnotes

Appendix

Acknowledgements

I would like to thank Edella Schlager, Ramiro Berardo, Craig Smith, Brint Milward, and Tyler Scott for their comments on previous versions of this article. I also want to thank Edella Schlager and Jeff Hanlon for allowing me to use the coded data from New York, and Ute Brady for her help as second coder. Finally, thanks to the editors of Political Research Quarterly (PRQ) and the three anonymous reviewers who helped enormously to strengthen this paper. All errors remain my own.

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

Notes

Supplemental Material

Replication data for this article are available with the manuscript on the Political Research Quarterly (PRQ) website.

ORCID iD

Tomás Olivier

References

Alcott

Emily

Ashton

Mark S.

Gentry

Bradford S.

, eds. 2013. Natural and Engineered Solutions for Drinking Water Supplies: Lessons from the Northwestern United States and Directions for Global Watershed Management. Boca Raton: CRC Press.

Anderson

Brigham S.

Butts

Carter

Carley

Kathleen

. 1999. “The Interaction of Size and Density with Graph-Level Indices.” Social Networks 21 (3): 239–67.

Andrew

Simon

. 2010. “Adaptive versus Restrictive Contracts: Can They Resolve Different Risk Problems? In Self-Organizing Federalism: Collaborative Mechanisms to Mitigate Institutional Collective Action, edited by Feiock

Richard

Scholz

John T.

, 91–113. New York: Cambridge University Press.

Andrew

Simon

Carr

Jered B.

2013. “Mitigating Uncertainty and Risk in Planning for Regional Preparedness: The Role of Bonding and Bridging Relationships.” Urban Studies 50 (4): 709–24.

Andrew

Simon

Hawkins

Christopher V.

2012. “Regional Cooperation and Multilateral Agreements in the Provision of Public Safety.” The American Review of Public Administration 43 (4): 460–75.

Angst

Mario

Hirschi

Christian

. 2017. “Network Dynamics in Natural Resource Governance: A Case Study of Swiss Landscape Management.” Policy Studies Journal 45 (2): 315–36.

Baland

Jean-Marie

Platteau

Jean-Philippe

. 1996. Halting Degradation of Natural Resources: Is There a role for Rural Communities? New York: Food and Agriculture Organization of the United Nations.

Basurto

Xavier

Kingsley

Gordon

McQueen

Kelly

Smith

Mshadoni

Weible

Christopher M.

2010. “A Systematic Approach to Institutional Analysis: Applying Crawford and Ostrom’s Grammatical Syntax.” Political Research Quarterly 63:523–37.

Berardo

Ramiro

. 2014. “The Evolution of Self-Organizing Communication Networks in High-Risk Social-Ecological Systems.” International Journal of the Commons 8 (1): 236–58.

10.

Berardo

Ramiro

Alcañiz

Isabella

Hadden

Jennifer

Jasny

Lorien

. 2016. “Networks and the Politics of the Environment.” In The Oxford Handbook of Political Networks, edited by Victor

Jennifer Nicoll

Montgomery

Alexander H.

Lubell

Mark

, 611–28. Oxford: Oxford University Press.

11.

Berardo

Ramiro

Scholz

John T.

2010. “Self-Organizing Policy Networks: Risk, Partner Selection, and Cooperation in Estuaries.” American Journal of Political Science 54 (3): 632–49.

12.

Brown

Trevor L.

Potoski

Matthew

. 2003. “Contract-Management Capacity in Municipal and County Governments.” Public Administration Review 63 (2): 156–64.

13.

Brown

Trevor L.

Potoski

Matthew

Van Slyke

David

. 2016. “Managing Complex Contracts: A Theoretical Approach.” Journal of Public Administration Research and Theory 26 (2): 294–308.

14.

Burnett

Nancy

. 2005. Behind the Scenes: The Inside Story of the Watershed Agreement. Margaretville, NY: Catskill Watershed Corporation.

15.

Carr

Jered B.

Hawkins

Christopher V.

2013. “The Costs of Cooperation: What the Research Tells Us about Managing the Risks of Service Collaborations in the U.S.” State and Local Government Review 45 (4): 224–39.

16.

Carter

David P.

Weible

Christopher M.

Siddiki

Saba N.

Basurto

Xavier

. 2016. “Integrating Core Concepts from the Institutional Analysis and Development Framework for the Systematic Analysis of Policy Designs: An Illustration from the US National Organic Program Regulation.” Journal of Theoretical Politics 28 (1): 159–185.

17.

Crawford

Sue

Ostrom

Elinor

. 1995. “A Grammar of Institutions.” American Political Science Review 89 (3): 582–600.

18.

Crawford

Sue

Ostrom

Elinor

. 2005. “A Grammar of Institutions.” In Understanding Institutional Diversity, edited by Ostrom

Elinor

, 137–174. Princeton: Princeton University Press.

19.

Csardi

Gabor

. 2015. Package ‘igraph’. http://igraph.org/r/doc/igraph.pdf.

20.

Csardi

Gabor

Nepusz

Tamas

. 2006. “The igraph Software Package for Complex Network Research.” InterJournal, Complex Systems 1695 (5): 1–9.

21.

Feiock

Richard C.

2013. “The Institutional Collective Action Framework.” Policy Studies Journal 41 (3): 397–425.

22.

Feiock

Richard C.

Lee

In Won

Park

Hyung Jun

. 2012. “Administrators’ and Elected Officials’ Collaboration Networks: Selecting Partners to Reduce Risk in Economic Development.” Special Issue, Public Administration Review 72:S58–68.

23.

Feiock

Richard C.

Scholz

John

, eds. 2010. Self-Organizing Federalism. Washington, DC: Georgetown University Press.

24.

Galusha

Diane

. 2016. Liquid Assets: A History of New York City’s Water System. Expanded ed. Fleischmanns, NY: Purple Mountain Press.

25.

Handcock

Mark S.

Hunter

David R.

Butts

Carter T.

Goodreau

Steven M.

Krivitsky

Pavel N.

Morris

Martina

. 2017. “ergm: Fit, Simulate and Diagnose Exponential-Family Models for Networks.” The Statnet Project. http://www.statnet.org. R package version 3.8.0, https://CRAN.R-project.org/package=ergm.

26.

Hayes

Adam L.

Scott

Tyler A.

2018. “Multiplex Network Analysis for Complex Governance Systems Using Surveys and Online Behavior.” Policy Studies Journal 46 (2): 327–53.

27.

Henry

Adam D.

2011. “Ideology, Power, and the Structure of Policy Networks.” Policy Studies Journal 39 (3): 361–83.

28.

Henry

Adam D.

Vollan

Björn

. 2014. “Networks and the Challenge of Sustainable Development.” Annual Review of Environment and Resources 39:583–610.

29.

Hunter

David R.

Handcock

Mark S.

Butts

Carter T.

Goodreau

Steven M.

Morris

Martina

. 2008. “ergm: A Package to Fit, Simulate and Diagnose Exponential-Family Models for Networks.” Journal of Statistical Software 24 (3): 1–29.

30.

Lee

Youngmi

. 2011. “Economic Development Networks among Local Governments: The Structure of Collaboration Networks in the Tampa Bay Metropolitan Area.” International Review of Public Administration 16 (1): 113–34.

31.

Leifeld

Philip

Schneider

Volker

. 2012. “Information Exchange in Policy Networks.” American Journal of Political Science 56 (3): 731–44.

32.

Lusher

Dean

Koskinen

Johan

Robins

Garry

, eds. 2013. Exponential Random Graph Models for Social Networks: Theory, Methods, and Applications. New York: Cambridge University Press.

33.

Maser

Steven M.

1998. “Constitutions as Relational Contracts: Explaining Procedural Safeguards in Municipal Charters.” Journal of Public Administration Research and Theory 8 (4): 527–64.

34.

Massachusetts Water Resources Authority representative. Personal communication. June 3, 2016.

35.

Massachusetts Water Resources Authority Advisory Board representative. Personal communication. July 14, 2016.

36.

Memorandum of Understanding between the Commonwealth of Massachusetts Department of Conservation and Recreation and the Massachusetts Water Resources Authority. 2004. https://www.mass.gov/files/documents/2017/10/11/memorandum.pdf.

37.

Miller

Gary

. 1992. Managerial Dilemmas: The Political Economy of Hierarchy. New York: Cambridge University Press.

38.

Nesson

Fern L.

1983. Great Waters: A History of Boston’s Water Supply. Hanover, New Hampshire: University Press of New England.

39.

New York State Department of Environmental Conservation representative. Personal communication. July 12, 2016.

40.

Ostrom

Elinor

. 1990. Governing the Commons: The Evolution of Institutions for Collective Action. New York: Cambridge University Press.

41.

Ostrom

Elinor

. 2005. Understanding Institutional Diversity. Princeton: Princeton University Press.

42.

Roberts

Christine L.

1990. Conflict and Cooperation in Watershed Management: Case Study of Metropolitan Boston’s Water Supplies. Unpublished Masters in City Planning thesis, Department of Urban Studies and Planning, Massachusetts Institute of Technology.

43.

Robins

Garry

Pattison

Philippa

Kalish

Yuval

Lusher

Dean

. 2007. “An Introduction to Exponential Random Graph (p*) Models for Social Networks.” Social Networks 29 (2): 173–91.

44.

Rodriguez

Miguel

Tavares

Antonio F.

Araújo

J. Filipe

. 2012. “Municipal Service Delivery: The Role of Transaction Costs in the Choice between Alternative Governance Mechanisms.” Local Government Studies 38 (5): 615–38.

45.

Scharpf

Fritz W.

1997. Games Real Actors Play: Actor-Centered Institutionalism in Policy Research. Boulder, CO: Westview Press.

46.

Scholz

John T.

Berardo

Ramiro

Kyle

Brad

. 2008. “Do Networks Solve Collective Action Problems? Credibility, Search, and Collaboration.” The Journal of Politics 70 (2): 393–406.

47.

Siddiki

Saba

Weible

Christopher M.

Basurto

Xavier

Calanni

John

. 2011. “Dissecting Policy Designs: An Application of the Institutional Grammar Tool.” Policy Studies Journal 39 (1): 79–103.

48.

Siddiki

Saba N.

Koski

Chris

Carboni

Julia L.

Sadiq

Abdul-Akeem

. 2015. “How Policy Rules Shape the Structure and Performance of Collaborative Governance Arrangements.” Public Administration Review 75 (4): 536–547.

49.

Soll

David

. 2013. Empire of Water: An Environmental and Political History of the New York City Water Supply. Ithaca, NY: Cornell University Press.

50.

Steinberg

Phillip E.

Clark

George E.

1999. “Troubled Water? Acquiescence, Conflict, and the Politics of Place in Watershed Management” Political Geography 18 (4): 477–508.

51.

Surface Water Treatment Rule, 40 C.F.R §141.71. 1989.

52.

Taylor

Michael

. 1987. The Possibility of Cooperation. New York: Cambridge University Press.

53.

U.S. Environmental Protection Agency. 1999. “Protecting Sources of Drinking Water: Selected Case Studies in Watershed Management.” http://nepis.epa.gov/Exe/ZyPURL.cgi?Dockey=200027IX.txt.

54.

Hongtao

Scholz

John T.

2016. “Policy Networks in Complex Governance Subsystems: Observing and Comparing Hyperlink, Media, and Partnership Networks.” Policy Studies Journal 44 (3): 248–79.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.01 MB