Contracting Beyond the Market: Property Rights, Externalities, Historical Conflict, And

Contracting Beyond the Market: Property Rights, Externalities, Historical Conflict, and

Contractual Agreements between Firms and Nonmarket Actors

ABSTRACT: Despite firms’ growing engagement of nonmarket stakeholders, little attention has been devoted to understanding the emergence of contractual relationships between firms and nonmarket stakeholders. Considering that a very large number of contractual agreements with nonmarket actors are theoretically possible but only a small number of such contracts are observed, we seek to understand what factors explain the use of contracts to govern some firm-stakeholder relationships and not others. We ground our inquiry in transaction cost economics where governance is seen as a means to infuse order into a relation where potential conflict threatens value creation, and theorize three dimensions of a firm-stakeholder relationship that influence the potential for conflict: the property rights, externalities and history of conflict that characterize a firm-stakeholder relationship. We situate our inquiry in the Canadian mining sector and examine firm’s relationships to proximate indigenous communities to identify a plausible exhaustive set of stakeholders “at risk” of signing a contract with a firm, leverage the historical assignment of property rights, and to control for institutional-level factors that may drive stakeholder contracting. We find support for our propositions by examining which of the 4,414 dyads (involving 457 indigenous communities and 85 mining firms) have signed 190 contracts with between 1999 and 2013.

INTRODUCTION

Recent years have witnessed both increasing levels of tension between firms and some of their stakeholders and unprecedented levels of collaboration between market and nonmarket actors. Growing managerial attention to nonmarket issues and academic research on firms’ nonmarket strategies

(Dorobantu, Kaul, et al. 2017) provide additional recognition that firms’ ability to generate and appropriate value is, at least in part, conditioned by the level of political and social consent for their operations. To obtain such consent, firms have intensified their lobbying and campaign contributions, devoted more resources to corporate social responsibility programs, partnered with environmental and social organizations, developed public-private partnerships, and paid more attention to their relationships with surrounding communities. While many of these efforts are limited and informal interactions, more and more collaborations between market and nonmarket actors are being developed as long-term partnerships and formalized through contractual agreements.

A contract is “an agreement between two or more parties that is binding on those parties, to the degree that to renege on the agreement will be costly” (Argyres and Liebeskind 1999, p. 51). Contracts between firms and their stakeholders in the market space – employees, suppliers and consumers – have

1 been studied extensively by scholars working at the intersection of law and economics, including by those building on transaction cost economics (Coase 2008; Williamson 1985) and property rights theory (Hart

1995, Hart and Moore 1990). By contrast, little attention has been devoted towards understanding the emergence of contractual relationships between firms and nonmarket stakeholders (King 2007).

Considering that a very large number of contractual agreements with nonmarket actors are theoretically possible but only a small number of such contracts are observed, we seek to understand what factors explain the existence of some contracts and the absence of others. What economic, social and environmental conditions make a contract between a firm and one nonmarket entity more likely than a contractual agreement between the same firm and another nonmarket entity?

We build on prior research in transaction cost economics and property rights theory to examine when firms and nonmarket stakeholders use contractual agreements as mechanisms for governing their relationship. We start from the premise that “governance is a means by which to infuse order in a relation where potential conflict threatens to undo or upset opportunities to realize mutual gains,” (Williamson

1999). The firm aims to employ an interest-aligning governance mechanism that reduces opportunism and conflict within a stakeholder relationship to an “efficient” level for which the costs of further reduction outweigh the benefits (Jones 1995: 412). Formal contracts offer less flexibility and capacity for adaptation than relational governance; 1 and, however detailed, they are nonetheless incomplete and do not reduce the risk of future conflict entirely. But they reduce it considerably. The signing of a formal contractual agreement indicates that the two parties have converged on a mutually agreeable set of conditions that will define the processes of value creation and distribution and the resolution of future conflicts among the parties. Formal contracts are therefore more likely to be signed when the risk of conflict is high, even when relational governance might develop to complement the negotiation and implementation of a formal contract (Poppo & Zenger 2002).

1 We exclude hierarchy from consideration in the feasible choice set (Williamson 1999) because in the context of firm’s relationships with non-market stakeholders, hierarchy is typically not feasible.

In our theoretical framework, we first consider two sources of potential conflict that shape a firm- stakeholder relationship: property rights (ownership and use rights) and social costs (specifically, environmental externalities). Property rights define an owner’s rights over an asset, including their ability to restrict its use by others (Alchian 1965; Hart & Moore 1990). Externalities refer to the impact of an asset’s use on others and reflect the absence of property rights (Coase 1960; Demsetz 1967).

Disagreements over the use of an asset (and the distribution of the value its use generates) and over the externalities generated by its use are key sources of potential conflict between firms and their stakeholders. Because such conflict can interfere with the firm’s ability to create value, as in the case of employee strikes or disruptions resulting from community protests, firms have incentives to find ways to minimize the risk of conflict with stakeholders and its costs. Thus, in stakeholder relationships with heightened risk of conflict, firms are more likely to take on the additional costs of formal contracts in order minimize costs associated with such conflict.

We also consider in our theoretical development how the history of conflict between a firm and its stakeholders affects the probability of using a formal contract to govern a firm-stakeholder relationship. Past conflict is a strong indicator of perceived unfair distribution of value (i.e. disagreement over property rights or externalities) or perceived unfair treatment (i.e. a violation of norms) by the firm towards a stakeholder in past. A history of conflict between a firm and a stakeholder reduces trust, and may undermine the perceived credibility of commitments made through non-contractual (relational) governance. As such, where firm-stakeholder relations have a history of conflict, the probability of relying on a formal contract to govern future relations increases.

Part of the challenge in examining contractual agreements with nonmarket actors is the difficulty of defining nonmarket stakeholders and identifying those “at risk” of entering a contractual relationship with a firm. To overcome this challenge, we focus our study on a rather unlikely candidate as a contractual partner for a firm – local communities. In a number of industries (including agriculture and forestry; mining, oil and gas; and, more recently, real estate) such contracts, known colloquially as community benefit agreements (CBAs), have become a wide-spread industry practice. Nonetheless,

3 because CBAs are costly to the firm, it is unlikely that a firm will sign contracts with all local community groups in close proximity to a company’s operations. To focus our investigation on the economic, environmental and historical dimensions of the dyadic relationship between firms and the local communities in the proximity of their operations, we further limit our research design to one industry

(mining) and one group of stakeholders (indigenous communities) in one country (Canada). We investigate the relationships between 85 firms and 457 indigenous communities and seek to explain how the property rights, environmental externalities and social conflict characterizing these relationships influenced the signing of 190 CBAs between 1999 and 2013.

Our research builds upon and aims to contribute to different but related streams of strategy research. First, we aim to contribute to a growing stream of research on stakeholder value capture by studying when a community captures some value from the mining project through its contract with the firm. While attention to the relationship between value creation and value capture by different stakeholders has been growing in strategy research (Coff 1999; Gans & Ryall 2017), value capture by stakeholders outside the value chain – i.e. nonmarket stakeholders – has been discussed theoretically

(Freeman et al. 2007; Garcia-Castro & Aguilera 2015) but has not been examined empirically in prior research. We seek to advance this research by highlighting how the economic, environmental and social foundations of a firm-stakeholder relationship influence the probability of the stakeholder capturing some value from its relationship with the firm.

Second, our research also contributes to the extensive contracts literature by drawing attention to how variation in property rights, environmental externalities and social conflict influence the existence of contractual agreements. Prior research has theorized extensively about how property rights are defined by contracts (Bolton & Dewatripont 2005), but less about how property rights might affect the signing of new contracts. Argyres and Liebeskind (1999) argue that a firm’s previous contractual agreements condition the firm’s choice when signing new contracts. We build on this argument and related strategy research on contracts (Argyres & Zenger 2012; Poppo & Zenger 2002), but shift focus to how existing

4 property rights define the relationship between a firm and its stakeholders and, therefore, the probability of using a contract to govern their relationship.

Finally, our study also highlights how differences within stakeholder groups affect a firm’s relations with its stakeholders. Stakeholder theory has long argued that the cooperation of different stakeholders is a necessary condition for firm survival (Clarkson 1995; Donaldson & Preston 1995;

Freeman 1984) and has explored the theoretical differences between different stakeholder groups

(Frooman 1999, Mitchell et al. 1997, Rowley 2016). Our study contributes to this line of research on stakeholder relations by highlighting that even when firms interact with seemingly similar stakeholders

(e.g., indigenous communities) important differences exist in the relationship between the firm and these stakeholders. These differences are partly attributable to characteristics of the stakeholder itself

(Dorobantu and Odziemkowska 2017, Kassinis and Vafeas 2006) and partly due to differences in the foundations of the dyadic firm-community relationship. As a result, the firm’s relationship with one stakeholder (e.g., community A) may look very different than the firm’s relationship with a seemingly similar stakeholder (e.g., community B). We emphasize that firm-stakeholder relationships are multidimensional and theorize about property rights, externalities and historical social conflict as foundations of firm-stakeholder relationships, and we examine empirically how they affect the probability of using contracts to govern these relationships.

CONTRACTUAL AGREEMENTS AND VALUE CAPTURE BY NONMARKET STAKEHOLDERS

Property rights theory, stakeholder theory, and the more narrowly-focused agency theory, all view the firm as a nexus of contracts intended to reconcile divergent interests (Hill and Jones 1992).

Whereas agency theory focuses on the reconciliation of interests between managers and shareholders

(Jensen and Meckling 1979) and property rights theory views the firm as a collection of contracts that define rights to specific assets (Hart 1995, Hart and Moore 1990), stakeholder theory proposes that stakeholders (including suppliers, employees, local communities and the broader public) that are affected

5 by, or can affect, value creation have implicit or explicit contracts with the firm (Freeman & Evan 1990;

Freeman 1984; Hill & Jones 1992).

A formal (explicit) contract seeks to align the expectations of the parties involved and reduce their incentives to take advantage of the exchange partner. Contracts are written to clarify the obligations of each side under a wide (but nonetheless incomplete) range of future conditions and to define means of enforcement and mechanisms for resolving potential disagreements. As such, contracts allow exchange partners to jointly define a mutually agreeable framework for their interactions and thus to better govern their relationship (Schepker et al. 2014). As Mayer and Xing (2017, p. 101) emphasize, “contracts need to provide incentives for both parties to fulfill their promises, seek to prevent costly disputes from arising, and provide a mechanism to resolve disputes which arise from opportunism or from honest misunderstanding in spite of best efforts (Macaulay 1963, MacNeil 1973).”

Contracts with nonmarket stakeholders are no different. They provide the firm and the stakeholder an opportunity to reach a mutual understanding about the terms of their exchange or collaboration, to reduce incentives for opportunistic behavior on either side, and to agree on mechanisms to resolve disagreements should such disagreements arise. But while stakeholders within the firm (employees, shareholders, managers) and within the value chain (suppliers and customers) typically have formal contracts with the firm, stakeholders outside the value chain – nonmarket stakeholders – most often have only implicit contracts with the firm. “Stakeholders supply the firm with resources on the implicit (tacit) understanding that their claims on the organization will be recognized” (Hill and Jones 1992, emphasis added). When interests are largely aligned, the risks of opportunism or conflict are low, and implicit contracts suffice to govern the relationship between the firm and its nonmarket stakeholders. Local communities, for instance, offer their implicit consent to most economic developments and no formal contracts exists to govern these relationships. But when interests diverge and the risk of conflict is high, implicit contracts may not suffice to prevent conflict escalation with severe implications for the firm (Dorobantu, Henisz, et al. 2017).

The signing of a formal contract attests that the stakeholder and the firm have converged on a mutually agreeable set of conditions for the development of a project and the distribution of the value it

6 generates. Not only have the two parties developed a mechanism to govern their relationship, but they have also agreed on mutually agreeable distribution of value. Recent academic research has devoted increasing attention to the dynamics of value creation and appropriation (Brandenburger & Stuart 1996; Coff 1999;

Gans & Ryall 2017; Garcia-Castro & Aguilera 2015). In addition to considering the resources and capabilities that enable a firm to create value (Amit and Schoemaker 1993, Barney 1991, Castanias and

Helfat 1991), this body of scholarship seeks to understand both how value is created through the combination of resources and capabilities contributed by different stakeholders and how these stakeholders appropriate some of the value created through their exchange relationship with the firm (Chatain & Zemsky

2011; Lieberman et al. 2016; Coff 1999). But while new research has started to build on stakeholder theory to emphasize the role of various stakeholders in the process of value creation and appropriation (Asher et al.

2005; Tantalo & Priem 2016; Garcia-Castro & Aguilera 2015), attention to value appropriation by nonmarket stakeholders has received relatively little attention in both theoretical and empirical research.

We focus our theoretical attention and empirical analysis on contractual relationships between firms and one type of nonmarket stakeholder: local communities. By definition, local communities are critical stakeholders. All firms are affected by and affect the welfare of the communities in close proximity to their operations. In certain industries, especially those involving intensive use of natural resources (e.g., agriculture, forestry, fishing) or their extraction (mining, oil and gas), and those where large physical developments disturb the local community (e.g., infrastructure, real estate), the interests between private firms and local communities can easily diverge. Social opposition from local community groups can lead to significant disruptions and delays for the firm. For instance, protests by the Standing

Rock Sioux indigenous community in North Dakota and other groups supporting them led to significant delays in the construction of the 1,200-mile Dakota Access Pipeline (Dorobantu and Walter 2018).

To limit the risk of disruptions and delays resulting from potential conflict with local communities, many companies in these industries have used contractual agreements to govern their relationships with local communities (O’Faircheallaigh 2013). Such contracts, known as community

7 benefit agreements, or CBAs,2 have been negotiated in mining and extractive industries worldwide, have accompanied the development of onshore windfarms in the United Kingdom, and have also been used in the development of large-scale real estate projects in the United States. In the Canadian mining sector – the empirical context of our research – over 300 CBAs have been signed between mining firms and indigenous communities. In 1995, Falconbridge, the company operating the Raglan nickel mine, signed a

CBA with the local Inuit community in the Nunavik region of northern Quebec; in 1996, BHP signed four

CBAs with aboriginal communities in the vicinity of its Ekati diamond mine3; and, in 2002, Inco signed two CBAs with the Innu and the Labrador Inuit for the Voisey Bay nickel mine in Newfoundland.

The negotiation of these contracts offer firms and local communities an opportunity to reach a shared understanding of the terms that will govern the development of a project, the responsibilities of the two parties, and the best ways to resolve disagreements should they arise along the way. A CBA therefore represents a mutually agreed-upon formula for the project development and the distribution of value generated by it. By signing a CBA, the local community consents to the development of the project under the agreed conditions. In exchange, the CBA enables local communities to appropriate value in several ways.

First, most CBAs include provisions for direct financial disbursement in the form of fixed-amount payments, revenue sharing, or profit sharing. Second, CBAs typically provide for local employment and training opportunities, ensuring that members of the local community capture value through jobs. Third, CBAs can also include provisions for local procurement that give local suppliers preferential treatment and therefore enable additional channels for the local community to appropriate value from the development of the project.

Thus, the signing of a formal contract between a firm and a local community is an unambiguous signal that the local community captures some value from its relationship with the firm and that the firm

2 We use the term community benefits agreements (Cain 2014; Parks & Warren 2009; Salkin & Lavine 2008) to also include agreements referred to as community development agreements (O’Faircheallaigh 2015) and impact and benefit agreements (Sosa and Keenan 2001). 3 The four agreements were signed with the Dogrib Treaty 11 Council, the Akaitcho Treaty 8, the North Slave Metis Alliance, and the Inuit of Kugluktuk and the Kitikmeot Inuit Association (Natural Resources Canada, “Ekati Diamond Mine, Northwest Territories: Partnership Agreements,” Available: http://www.nrcan.gc.ca/mining- materials/publications/aboriginal/bulletin/8822).

8 views this distribution as preferable to assuming the risks associated with conflict with the local community. Since such contractual arrangements are not costless to the firm, they are not likely to govern the relationship between a firm and all the communities in close proximity to its operations. Therefore, a key question concerns the choice of formal contract over implicit contracting to govern some community relations, and not others. In answering this question, we assume that the cost associated with negotiating and enforcing a contract are similar across different communities – that is, a firm would incur transaction costs in the signing of a contract with any community. We therefore focus our theoretical inquiry on the firm-stakeholder dyadic relationship, acknowledging that actor-specific characteristics and the institutional context also play a role in the choice of governance mechanism. Specifically, we theorize about the choice of formal contract to govern a relationship by considering three dimensions of a firm’s relationship with a local community: (1) the property rights affecting the firm-community relationship;

(2) the level of environmental externalities involved; and (3) the degree of social conflict describing the historical relationship between the firm and the community.

We further emphasize that a similar mechanism connects each of the three dimensions to the probability of signing a contractual agreement: the presence (or absence) of property rights, externalities, and historical conflict in a firm-stakeholder relationship are indicators of the extent of divergent

(conflicting) interests that must be resolved through governance arrangements. Viewed from this perspective, the property rights, the externalities and the history that characterize a firm-stakeholder relationship are used as evaluative tools in the firm’s assessment of the probability and costs of conflict against the costs and benefits of formal contracts.

Property rights and contractual agreements with nonmarket stakeholders

Stakeholders own many of the resources that a firm needs to access and use in order to generate value

(Pfeffer and Salancik 1978). Different types of stakeholders contribute different types of assets: employees provide time, effort and specialized knowledge; suppliers provide intermediary inputs; financiers provide capital; governments provide regulatory approval and oversight; local communities provide local infrastructure and access to locations valuable to the firm. Value creation is the process of

9 combining existing assets in ways that lead to the total being more than the sum of the different parts

(Amit and Schoemaker 1993, Penrose 1959). A firm emerges as the nexus of contracts with different stakeholders that define or transfer ownership rights over existing and new (created) assets. By contributing to the activities of the firm, a stakeholder places some asset “at risk” in the firm, expecting to be compensated for the risk it assumes (Orts & Strudler 2002).

Contracts are negotiated and signed to define property rights over new assets (e.g., new technology) or transfer property rights over existing ones (e.g., land, equipment, patents). With the refinement of property rights theory through closer attention to the implications of incomplete contracts

(Hart 1995, Hart and Moore 1990), increasing emphasis has been placed on the idea that ownership of an asset confers the owner residual decision rights over it. Because all contracts are incomplete, they cannot explicitly specify what the parties should do in all future contingencies. Instead, contracts specify which party has decision rights in situations when the two parties cannot agree. As Hart (1989) explains,

“ownership of an asset goes together with the possession of residual rights of control over the asset; the owner has the right to use the asset in any way not inconsistent with a prior contract, custom, or any law.”

As a result, the party in possession of residual control rights over an asset has bargaining power over its exchange partner. Imagine, for instance, a situation in which a firm wants to build a mine that requires the use of private land. The land owner has, by law, property rights over the land. She can use it

“in any way not inconsistent with a prior contract, custom, or any law,” and she has the right to restrict the access and use of the land by others. She can also choose to transfer her property rights to another party either in full (through a sale) or in part (through a leasing agreement). A firm interested in developing a mine on that land would have to negotiate terms for accessing and using the land with the land owner, or assume the risk of conflict with this stakeholder.

While the preceding may suggest that the use of formal contracts is the most efficient governance mechanism between firms and stakeholders with property rights, the relationship becomes more complicated when one considers property rights as bundles of rights (Asher et al., 2005; Coase 1960).

Specifically, property rights typically include three privileges: (a) the right to use the asset (usus); (b) the

10 right to appropriate returns from the asset (usus fructus); and, (c) the right to change its form, substance and location (abusus) (Libecap 1999). As such, there are important differences between stakeholders not only in whether they have (or not) property rights over assets that the firm wants to access, but also in the range of privileges afforded by the property rights they possess. In our empirical context – indigenous communities in Canada – the degree to which each element is represented in these communities’ property rights over land varies considerably for historical reasons. While some communities possess property rights that include all three elements, others have use rights only, and others no property rights at all. A similar ordering of rights emerges within non-indigenous communities where, for instance, members may poses full property rights to their private dwelling, but only use rights to a public park. We expect differences in the privileges contained in stakeholders’ property rights to an asset sought for use by a firm to influence the governance mechanism the firm employs in its relationship with that stakeholder.

First, where a stakeholder possesses full property rights to an asset, a firm is likely to sign a contract with that stakeholder because the stakeholder can restrict the access and use of the asset by others. As such, the contract is a mechanism to transfer use rights on a valuable asset to the firm.

Conversely, a stakeholder that possesses only use rights to an asset, cannot in the same way restrict access to others’ use of the asset. However, use rights still provide some bargaining power if the stakeholder’s own use rights are impinged by the use of the asset by others. As such, a firm may still sign a contract with a stakeholder that has only use rights over an asset. While the probability of contracting is lower than in the case of full property rights, we argue that it is higher than in the case of no property rights. The preceding suggests an ordering of probabilities of formal contracting that mirrors the ordering of the degree of privileges contained in a stakeholder’s property rights. Specifically, we expect to observe CBAs when the mining firm’s operations are located on land over which the local community has full property rights, and, to a lesser extent, when the community has use rights to the land, while CBAs are unlikely to be signed when the local community has no property rights over the land. We hypothesize that:

Hypothesis 1: A firm is more likely to sign a contract with a community that has full property

rights, and, to a lesser extent with a community that has use rights, over assets that are valuable to

the firm, than with a community that has no property rights over these assets.

Contractual agreements as resolutions for environmental externalities

In a highly influential 1960 article entitled “The Problem of Social Cost,” Ronald Coase argued that firms and stakeholders can organize mutually beneficial exchanges that minimized the social costs of economic activities and ultimately ensured the preservation of the natural environment and social cohesion (Coase

1960, 1992). Coase defined social costs as “harmful effects on others” and considered the standard example “that of a factory the smoke from which has harmful effects on those occupying neighboring properties” (Coase 1960, p. 1). Coase’s argument was aimed to counter that of scholars who claimed, following Pigou (Pigou 1932), that the only solution for dealing with externalities was intervention from a central authority. Government regulations, their argument went, would restrict economic activities that imposed high social costs on others. Coase believed that even in the absence of government regulation or legislation assigning liabilities for social costs, affected parties would work out a mutually beneficial agreement if transaction costs associated with negotiating and enforcing the agreement were not too high.

More recently, Ostrom (1990) made a similar argument about the management of common-pool resources

(e.g., common pastures, irrigation systems, fisheries), which, given their shared or common property nature, are constantly at risk of depletion because of externalities. Ostrom showed throughout her career that institutions for the governance of these resources can be established by and enforced through the participation of those most involved in their use, without intervention from a central government authority

(Ostrom 1990, 2010, Ostrom et al. 1994).

Coase’s argument applies directly to interactions between firms and their stakeholders. A firm’s interest to pursue its activities even in the presence of social costs is clearly at odds with the interest of stakeholders who prefer that the firm cease activities that result in social costs. Coase would expect the firm and these stakeholders to negotiate a mutually beneficial agreement – i.e. a contract – in which either the firm would compensate stakeholders for the harm inflicted or stakeholders would incentivize the firm

12 to discontinue its economic activity in order to avoid the harm. For Coase, the agreement would maximize social welfare (Stigler 1989).

In our empirical context, we expect mining firms to sign contractual agreements with the indigenous communities directly affected by the mining firms’ activities. As noted before, we assume that transaction costs are similar across different communities – that is, a firm and any community would incur costs associated with negotiating and enforcing an agreement. Thus, relevant variation across communities is associated with the extent to which the firm’s mining activities impacts their welfare.

Mining activities can be highly polluting and affect the nearby environmental ecosystem, especially through their effects on water quality and availability. As a result, negative environmental externalities may be considerable not only for proximate communities, but also for more distant communities located in the same watershed as the mining development. For these communities, a chemical spill at the mine site or a fracture in the tailings dam can have devastating effects. In the absence of an agreement between a firm and a community subject to such negative externalities, the likelihood of community consent for the operation is reduced and the likelihood of conflict heightened. Therefore, we expect that firms are more likely to view formal contracts as a preferred governance mechanism for relationships with communities subject to negative externalities because they attenuate the probability of conflict.

Hypothesis 2: A firm is more likely to sign a contract with a community affected by environmental

externalities from its operations than with a community that is not similarly affected.

Contractual agreements as a response to social conflict

Finally, our theoretical framework also considers the historical relationship between the firm and its stakeholders. Our previous two hypotheses build on property rights and externalities as foundations underlying the relationship between a firm and a stakeholder, and consider both the use of assets and the externalities generated by this use as potential sources of conflict in firm-stakeholder relationships. Both arguments, however, assume a contemporaneous perspective on conflict: current disagreement over the use of assets and the externalities generated by these uses. In the theoretical perspectives we build upon, conflict disappears, for the most part, once a well-written contract has been signed to align the incentives

13 of the two parties and define their property rights (i.e. decision rights) over the assets exchanged and/or the externalities generated. At that point, there is no need for any additional – i.e. future – contract.

In reality, however, conflict between two parties develops over time, either because small disagreements over property rights and externalities are not immediately addressed, or because the interactions between the two parties violate one or both parties’ expectations about the other party’s behavior. A history of conflict indicates that trust among the two parties is low and suggests that at least one of the parties views the other’s actions as inequitable or not constituting fair dealing (Ring & Van de

Ven, 1994). If the agglomeration of mutual trust, respect and understanding that limit opportunistic behavior in interorganizational relationships (Kale et al. 2000) is an evolutionary process of repeated interactions (Ring and Van De Ven 1994), then prior conflict may suggest that relational governance is an insufficient or ineffective mechanism for governing the relationship. The damages that the two parties inflict on each other by continuing in a conflictual relationship can escalate, and the necessity of finding ways to resolve the disagreements increases. A formal contract is likely to ensue.

A common instance of firm-stakeholder conflict involves firms and their employees, represented by union organizations. The conflict concerns the distribution of value between them – specifically, how much value would be appropriated by the employees in exchange for their labor and skills (Coff 1999).

Agreements on this issue are not always easy to reach, and failure to do so can easily result in union- organized protests, strikes and walkouts. Such social conflict results in disruptions and delays that can be very expensive for both the firm (forgone value) and for the employees (forgone wages), so both parties have strong incentives to reach an agreement to resolve the conflict. Moreover, conflict with one set of stakeholders (e.g., employees) can easily escalate into conflict with other stakeholders as well, magnifying the financial implications of the original conflict situation (Dorobantu, Henisz, et al. 2017).

In our context, the relationship between a mining firm and an indigenous community is shaped by its history. In parts of the world where mining activities have been carried out for decades or longer, local communities and mining firms share long histories of acrimonious interactions and social conflict. The lives of local communities are often dramatically transformed by large-scale mining operations, with

14 positive impacts (e.g., jobs, local economic development) not always compensating for negative externalities (e.g., social disruption, pollution, water shortages). In addition, many mining firms have tarnished reputations in the eyes of their local communities because of the inconsiderate ways of their managers in past interactions. The combination of perceived unfair past distribution of value (or the lack of distributional justice) and unfair treatment (or the lack of procedural justice) in the past might underlie the conflictual relationship between many local communities and firms in this industry.

We propose that a contractual agreement between a firm and a community is more likely when the two parties share a history of social conflict. Efforts by firm managers to reach agreements with such communities are largely driven by incentives to minimize the negative effects of renewed social conflict: disruptions and delays of the mine operations or its proposed development, and the financial implications of such delays. A community that has mobilized in the past through social protests or blockades at the mine site is likely to mobilize again. To reduce such risk, managers may be willing to secure the consent of the local community in exchange for sharing some of the value generated by the project with the community itself (Amengual 2018). This effort can also be construed as an effort by the firm to co-opt sources of social opposition to large-scale projects (Selznick 1949). In addition, past conflict suggests that the relationship between the firm and the community is characterized by low levels of mutual trust. As a result, contractual governance becomes the preferred option as it is considerably more cost effective than efforts to rebuild the trust necessary for relational governance to work. We hypothesize that:

Hypothesis 3: A firm is more likely to sign a contract with a community with which it shares a

history of conflict than with a community with which it does not.

METHODS AND ANALYSIS Empirical Context and Sample Description

Although CBAs are becoming prevalent in a number of industries (renewable energy, oil and gas, residential and transportation infrastructure) and institutional settings (Australia, Canada, and the United

States), we focus our empirical inquiry in one country (Canada) and one industry (mining). Our single

15 country setting allows us to minimize differences in the enforcement of property rights and regulatory or legal underpinnings of CBAs across countries. Focusing on one industry also ensures our results are not driven by differences in time-varying unobservable norms around stakeholder contracting across industries.

We chose to situate our study in Canada and one particular type of stakeholder, indigenous communities, to overcome a critical source of endogeneity associated with property rights. The value that a stakeholder can appropriate via property rights to a particular asset (e.g., land, human capital) fluctuate with the value of the underlying asset. As such stakeholders with good foresight or information are more likely to hold property rights to valuable assets (e.g., employee invests in valuable human capital acquisition through training). Since stakeholder foresight and information is not likely to be observable in most contexts, the researcher cannot disentangle the effects of the ‘savviness’ of stakeholders from their possession of property rights over the asset, inherently biasing any results. By situating our empirical inquiry in the context of Canadian indigenous communities, we are able to exploit historical assignment of property rights to land to minimize such bias.

Specifically, between 1780 and 1921 the British government, and subsequently the Canadian government, signed treaties with indigenous communities to acquire indigenous lands (Alcantara 2003,

2008).4 Commonly referred to as ‘historic treaties’, they included the ceding of title to large tracts of land by indigenous communities in exchange for monetary payments, while often maintaining hunting and fishing rights (i.e., use rights) over the land (Sosa and Keenan 2001). As the British government or its representatives acquired land from indigenous Canadians via historic treaties, indigenous populations were moved onto reserves where they received rights of occupation to the land (Alcantara 2003). Initially reserves were located in isolated areas far away from colonial settlements, and beginning in 1850 the British government decided reserves should be closer to colonial settlements (ibid. 399). Although the allocation of reserve lands varied across provinces and time, and no comprehensive description of the strategies exists

4 The treaties signed prior to the 1780s focused on military alliances (e.g. “Peace and Friendship” treaties with the Mi'kmaq and Maliseet tribes between 1725 and 1779), but nevertheless, eventually those indigenous communities were moved to reserves in the mid-1800s.

(Wilson and Peters 2005), an overarching ideology of separation of indigenous and non-indigenous populations resulted in only 5.1% of indigenous peoples residing in urban spaces in 1901 (ibid. 399). These historically imposed land boundaries over which indigenous communities possess different classes of rights

(full property rights versus use rights) maintain today, and were determined in a manner plausibly exogenous to present day characteristics (e.g. savviness) of the communities that hold those rights, because they were agreed to by the distant ancestors of current community members and the allocation of reserve lands was determined by the patterns of colonial settlement and government officials. While more recently some indigenous communities have sought greater rights to their ancestral lands by negotiating ‘modern treaties’ with the Canadian government, our results are robust to their exclusion, giving us confidence that our results are not biased due to unobservable community characteristics that may correlate with the possession of property rights.

Within this context of historically assigned property rights, several characteristics of mining recommend it as a useful industry in which to test our hypotheses. First, the well documented negative externalities that mining creates on water enable us to exploit variation in the co-location of mines and communities’ on the same watershed basins to more precisely estimate the role of externalities in stakeholder contracting. Secondly, location choice in extractive industries is constrained to geological anomalies that result in mineral formations. Unlike the siting of developments such as roads, oil and gas pipelines, or retail outlets, where the firm can avoid certain areas (i.e. subject to property rights, or to minimize conflict or externalities) by changing the path of the linear development or location of a store

(Yue et al. 2013), mineral developers are constrained to building a mine on a reserve of minerals in which they have expertise. We do not suggest that firms in extractives are not sensitive to other institution-level factors such as rule of law or political risk in choosing where to locate, but within a single country context with stable developed institutions, location choice is typically constrained to a few mineral rich areas and more likely to be driven by the potential value of a mineral deposit. We do not assume the as-if random assignment of firms to mineral deposit locations, but deal with these choices by exploiting the ownership

17 of multiple mineral properties by the same firm to include firm fixed effects that deal with time-invariant unobservable characteristics that may drive a firm’s choice of one mineral reserve over another.

Sample

We test our hypotheses on a sample of 101 mines and all indigenous communities within a 500km radius of those mines. We began with the complete list of 187 mines that signed CBAs with Canadian indigenous communities between 1999 and 2013 as reported by Natural Resources Canada (NRCAN), the government agency responsible for resource development in Canada. We chose the period 1999 to 2013 because it corresponds to a stable period in the legal interpretation of indigenous and resource industry property rights in Canada. Following the recognition of indigenous rights in the 1982 Constitution Act, the interpretation of indigenous property rights and the security of property rights of resource industries in Canadian law have evolved through several landmark Supreme Court cases up to 1997 (Keay and Metcalf 2011), and then again in the Canadian Supreme Court’s 2014 Tsilhqot’in v. British Columbia decision. Focusing our inquiry from 1999 to 2013 ensures that legal interpretations of both indigenous and resource industries property rights were consistent during our study period. After excluding mines for which no public announcement of the CBA existed, and those owned by private firms (i.e. no financial data available), 101 mines remained.

To investigate which communities firms chose to sign CBAs with, we identify communities located within 500km of the focal mine. We employ a 500km radius because, while almost 90 percent of CBAs are with communities within 300km of the mine, slightly over 10 percent of the CBAs are signed with communities located between 300 and 500 km of a mine. In supplemental analysis, we restrict the choice set to those communities within 300km of the mine. We use ArcGIS software to determine the geodesic distance between each mine and indigenous community by mapping the mine location coordinates onto the

Canadian Aboriginal Lands map from NRCAN’s Geogratis database. The Aboriginal Lands map consists of polygons that depict the administrative boundaries of lands, both reserves and lands covered by modern treaties, where the title has been vested in specific First Nations or lands set aside for their exclusive benefit

(e.g. reserves).5 The matching of communities to the 101 mines in our sample resulted in 4,414 mine– community dyads, or an average of 44 indigenous communities within 500 km of each mine.

Dependent Variable

Our outcome of interest is a CBA between a firm operating a mine and a proximate community. Therefore, our dependent variable is a dyad-level dummy coded 1 if a mine-community dyad has a CBA signed by the focal firm in a given year, and 0 otherwise (CBA within firm-community dyad). We relied on the list of all

CBAs between indigenous communities and mine developers maintained and updated annually by NRCAN using public information and input from analysts intimately familiar with each mineral development. We confirmed the signing year and signatories to the CBA contained in the NRCAN list, using announcements associated with each CBA which we obtained by searching the Dow Jones Factiva database, and firms’ financial disclosures and reports.

Independent Variables

Property rights. To test our hypothesis regarding property rights driving CBA formation, we distinguish between two classes of property rights an indigenous community can have to the land where a mine is located. Property rights include “any social institutions that define or delimit the range of privileges regarding specific resources granted to individuals.” (Asher et al. 2005: 8). 6 In Canada, indigenous communities possess full property rights to land located on modern treaty and reserve lands. Modern treaties, the first of which was signed in 1975, are negotiated agreements between indigenous communities

(often multiple communities) and the Canadian government that delineate a community’s rights with respect to specific lands, including surface and subsurface (i.e. mineral) rights (Sosa & Keenan 2001).

Property rights to lands demarcated in modern treaties represent use rights, rights to appropriate returns and the right to change their form (i.e. full property rights). On reserves, indigenous communities possess what

5 For a small number of communities that have neither reserve lands nor modern treaties, we supplemented the lands map with latitude and longitude coordinates for the geographic center of the community to ensure we include them in the sample. 6 Typically, those privileges include three elements: (a) the right to use the asset (usus); (b) the right to appropriate returns from the asset (usus fructus); and, (c) the right to change its form, substance and location (abusus) (Libecap 1999).

19 may be considered de facto full property rights. In essence, the Canadian government retains title of reserves for the use and benefit of indigenous communities, however, in practice, communities through their elected band councils exercise usage and occupancy rights to reserve lands. Band councils, which are similar to municipal governments, control “economic development, zoning, housing, and most importantly, the administration of reserve lands” (Alcantara 2007).

In addition to the full property rights category, there is another category of rights relevant in our context: ‘use’ rights given to signatory communities of some historic treaties. Of the 70 historic treaties, more than half established continuing rights to hunt and fish on lands defined in the historic treaty. While we expect these to have a lower impact on the probability of a CBA being signed because the attenuation of property rights in an asset affect the terms of trade (Libecap 1999: 6), they nevertheless represent use rights over land.

In order to determine the property rights held by each community to the land where a mine was located, we supplemented the Canadian Aboriginal Lands map that includes reserves and modern treaty boundaries (described above) with the Historic Treaties map containing polygons of the geographic boundaries of historic treaties. Using the coordinates for each mine in our sample, and drawing a 1 km radius around each coordinate to better represent a typical footprint of a mine, we identified if the land on which the mine was located was on reserve or modern treaty lands (full property rights), on historic treaty lands (use rights) or neither (no property rights). Matching each reserve, modern treaty and historic treaty on which a mine was located to each of 637 indigenous communities, we created an ordinal variable of property rights, representing communities with full property rights over the land where the mine is located

(2), communities with use rights over the land where the mine is located (1), and communities with no property rights over that land (0).

Externalities. We identify communities more likely to be subject to negative externalities from the mining project as those located in the same watershed basin as the mine. As our interest is in contracts typically signed before the start of operations at a mine, we cannot rely on measures of externalities produced after the start of the mine (e.g., the Toxics Release Inventory is one popular measure used by

20 others), but instead those indicating higher probabilities of externalities ex ante. Mining’s greatest environmental impact is on water quality and quantity (Franks et al. 2014). Pollution originating at the mine is carried by drainage within a watershed basin resulting in negative externalities well beyond the originating source of the pollution (i.e. the mine).

To identify whether the mine and the community are co-located in the same watershed basin, we map each community’s land boundaries and each mine’s geographic coordinates on the Watersheds in

Canada map available from ArcGIS. The map is based on the Water Survey of Canada data, and includes main drainage and sub-drainage areas. The 637 indigenous communities in Canada are located on 135 distinct sub-drainage areas, while our 101 mines are located on 44 sub-drainage areas. Approximately 13 percent of our mine-community dyads are co-located in the same sub-drainage area. Of those dyads, 19 percent have CBAs, compared to only 1 percent in the entire set of dyads. Externalities is a dummy coded

1 when the mine and community are located on the same watershed sub-drainage area, and 0 otherwise.

Social conflict. Following common practice in social movements research (Earl et al. 2004), we rely on media reports of social conflict events (e.g. protest, blockade) by a community against the firm and/or mine to test hypothesis 3. Since the early 1980s, indigenous peoples have engaged in widespread social mobilization in Canada, much of which has been reported in the media (Wilkes et al. 2010). To overcome the selection bias and description bias associated with media reports (Earl et al. 2004), we include in our source list the entire Factiva database, which covers over 25,000 media outlets and press wires, and rely only on the “hard facts” of the event (e.g. who, what, when), which is relatively accurate in media reports (Earl et al. 2004: 65). We searched Factiva for reports where the community name appears within

10 words of terms typically associated with social conflict (e.g., strike, rally, demonstration, protest, blockade) (Wilkes et al., 2010) using multiple spellings of the same community name (e.g., “Na-cho Nyak

Dun” or “Nacho Nyak Dun”), historical names (e.g., Snowdrift was renamed to Łutselk'e, also spelled

Łutsel K’e), and names of tribal councils to which the community belongs to ensure no events were missed.

Each article was read to ensure it referred to social conflict, and coded for the date it took place, and the target of the conflict (e.g. firm, mine, government). Social conflict is an indicator variable coded 1 if the

21 focal community engaged in social conflict targeting the firm or mine, as mining projects can change owners frequently, as of the year preceding the CBA, and 0 otherwise.

Control Variables

We control for a number of additional factors likely to impact contract formation between a firm and a community, pertaining to the firm-community dyad, community, mine and firm characteristics and the broader institutional setting and changes therein over time.

At the dyad-level, we control for the geodesic distance between the nearest land boundary of the community and the mine, which not only is expected to drive contract formation but likely correlates with externalities and possibly mobilization (i.e. communities closer to the mine may find it less costly to mobilize at the mine site). Mine-community geodesic distance is scaled by 100km (i.e. distance in km divided by 100) to ease the presentation of results. Secondly, regulatory or legal institutions can be used to gain leverage over firms, and may correlate with social conflict. Therefore, we control for whether a community took any actions against the firm or mine via legal or regulatory channels. We take an identical approach to that for social conflict, reading each article retrieved from a search of Factiva where the name of the community appears within 10 words of terms typically associated with regulatory or legal actions

(e.g., petition, grievance, investigation, injunction, lawsuit, legal action, court). Legal conflict is an indicator variable coded 1 if the focal community took action against the firm or mine via regulatory or legal channels as of the previous year, and 0 if it took no such actions, as reported in the media.

We control for several community characteristics we expect to influence the probability of a contract with that community. We control for the community’s history of social and legal conflict, in general (i.e. not against the focal firm), because firms may be motivated to sign CBAs where the risk of future action by the community is elevated. Repeated episodes of social mobilization build capacity for future collective action

(McAdam 1982, Zald and McCarthy 1987) and lower its costs (Rowley and Moldoveanu 2003). Therefore, even in instances where the firm or mine has not been the target of social or legal conflict, managers may attribute higher risks of future conflict to communities with a history of engaging in such conflict with other parties. We constructed this variable from the data we obtained from our search and reading of media reports

22 of community social conflict (e.g. strike, rally, demonstration, protest, blockade) and legal conflict (e.g., petition, grievance, investigation, injunction, lawsuit, legal action). The community’s social and legal conflict history is the sum of all unique media-reported social or legal conflict events the focal community participated in as of the previous year, not including those targeting the focal firm or mine.

We also expect that communities learn through their past experiences with CBAs and through the experience of proximate communities. Learning through past experience can facilitate future CBAs by increasing the efficiency of the negotiating process, or increase community resources for future negotiations, while the observance of proximate communities’ CBAs increases information diffusion about the negotiation process and rent-capture opportunities. Therefore, we control for the focal community’s experience by summing the number of CBAs it has signed (community’s past CBAs), as well as the number of CBAs communities within 300km of the focal community have signed (proximate community CBAs).

Finally, we control for several socio-demographic characteristics of the community that may be predictive of a CBA. We control for the community’s population, which is an annual measure logged due to the skewness of the data. We obtain population data from Aboriginal Affairs and Northern Development

Canada (AANDC), the government agency responsible for indigenous issues in Canada, which has a statutory duty to maintain a record of all registered indigenous persons by community under the Indian Act.

We control for the employment rate of indigenous peoples in the community, which we obtain from the

2006 Census, because this might be related to the degree to which the community is poised to capture positive externalities from the mine in the form of local employment. Finally, the degree to which externalities matter may vary by the extent to which residents practice traditional lifestyles. We proxy this using the percent of indigenous residents that can speak an indigenous language (indigenous language speakers), which we obtain from the 2006 Census.

We control for the value of the asset under contract by estimating the portion of the value of the firm that is represented in the mining project. Mine value to the firm is the value of mineral resources in the mine divided by the firm’s market capitalization in a given year. Each mine’s estimated volume of mineral resources is obtained from the Raw Materials Group database, and supplemented by resource estimates

23 released by firms compliant with Canada’s Standards of Disclosure for Mineral Projects. We multiplied the volume of mineral resources (measured in billions of tonnes) by the prevailing spot price for the commodity in the previous year, using commodity prices from the London Metal Exchange, and for those commodities not traded on an exchange (e.g., diamonds), estimated prices based on U.S. Geological Survey statistics.

We control for the phase of the mine because it determines the certainty of cash flows associated with the mine, the impact of conflict with the local community (Franks et al. 2014), and the regulatory levers that a community can employ. Mining projects move through several stages, from early exploration to pre- feasibility and feasibility studies, to development (which includes permitting and planning), to construction and operations. We obtain data on the phase of each mine in a given year by reading the mine’s feasibility, pre-feasibility, and resource estimate reports obtained from firms’ websites and SEDAR, as well as news reports and press releases where resource estimate reports were not available. Mine phase is a continuous variable reflecting whether the mine is awaiting exploration or suspended; in exploration; pre-feasibility or feasibility; in construction; or in operation. We also control for firm size because large firms are more likely targets of collective mobilization (Bartley and Child 2014). Firm size is the firm’s market capitalization, log-transformed to address skewness, and obtained from the Thomson Reuters Worldscope Database.

We include in our estimation the number of social or legal conflicts the firm has experienced from indigenous communities other than the focal one, as well as the number of CBAs the firm has already signed, as of the preceding year. Given limits on firm’s resources, both those devoted to negotiations as well as those disbursed via CBAs, we expect that there is a theoretical limit on the number of CBAs a firm will negotiate and sign. As such, in the face of social conflict with other communities it may be more likely to prioritize those communities claims over the focal community, and where it has signed CBAs previously it may be less motivated to sign subsequent CBAs that disburse more of the mine’s value. Other conflict against firm is the sum of all unique media-reported social or legal conflict against the firm by other indigenous communities (i.e. not focal community) as of the previous year. We constructed this variable from the data we obtained from our search of Factiva for social and legal conflict described above, using

24 the ‘target’ field to filter for actions against the focal firm or mine. Firm’s past CBAs is the number of CBAs the firm has already signed for the focal mine.

In line with past findings that changes in the institutional environment moderate demands for, and returns to, corporate social responsibility (Flammer 2013), we control for institutional-level factors that may drive CBAs. First, we control for media attention given to indigenous peoples and issues in Canada which we obtain from the Global Database on Events, Language, and Tone. Media attention to indigenous issues is the number of media mentions of events involving North American indigenous peoples in Canada in the year preceding the CBA, normalized by dividing by all media-reported events, and multiplied by

1,000 to ensure readability of coefficients. Similarly, we expect that the general diffusion of CBAs as a governance instrument for managing stakeholder relations to impact the propensity of its use. As such we control for the number of CBAs in the province in which the mine is located (CBAs in province), as the effects of diffusion are stronger when past adoption is proximal.

RESULTS

Table 1 shows descriptive statistics, and Table 2 presents correlations for the variables in our sample. The dependent variable, CBA within firm-community dyad, is a binary variable that takes on the value 1 if a mine-community dyad has a CBA in a given year, and 0 otherwise. As such, we use logistic regression to test our hypotheses regarding the probability that a firm signs a CBA with a community.

Specifically, we use the conditional maximum likelihood estimator for fixed effects logit, which avoids the incidental parameters problem of logit estimators with fixed effects (i.e. dummies) (Greene 2014) and allows us to control for firm time-invariant unobservable characteristics.

------Insert Tables 1 & 2 ------Table 3 presents the results from the regressions. Model 1 includes the control variables, of which the distance between the mine and community is negatively associated with the probability of a firm signing a CBA with that community (p=0.000), as expected. Conversely, the community’s experience with CBAs and media attention to indigenous issues, are associated with an increase in the probability of a CBA. We

25 test our first hypothesis in Model 2 by including the continuous measure of property rights in the regression.

We find that the greater the property rights of the community to the mine, the greater the probability of a firm signing a CBA with that community (p=0.001). In Model 3 we disaggregate property rights into use rights and full property rights, with the comparison being communities with no property rights. Again we observe the positive effect of both use rights (p=0.010) and full property rights (p=0.016) on the probability of a CBA with a community. Having use rights over the land on which the mine is located compared with no property rights, is associated with an increase of 4.98 in the odds of a CBA, whereas, communities with full property rights have 6.34 greater odds of getting a CBA, compared to those that only have use rights, holding all other variables constant.

We next turn to our second hypothesis, where we posited that firms are more likely to sign CBAs with communities affected by environmental externalities. In Model 4, we observe that co-location on the same watershed basin is indeed associated with a higher probability of a firm-community dyad signing a

CBA (p=0.001). The estimate corresponds to an odds-ratio of 3.15, or stated differently, being subject to environmental externalities is associated with a 3.15 increase in the odds that a firm signs a CBA with a community. Finally, in Model 5 we find support for hypothesis 3 that past social conflict with the community increases the probability of a CBA (p=0.000). Specifically, the presence of social conflict in a firm-community dyad, is associated with an increase of 0.98 in the probability of the community having a

CBA with the firm (other variables held at sample means).

------Insert Table 3 ------Model 6 contains all our hypothesized and control variables. Therein, a community’s property rights

(p=0.001), environmental externalities (p=0.004) and social conflict with the firm or mine (p=0.000), are positively associated with the probability of the firm signing a CBA with that community. To aid interpretation,

Figure 1 contains plots of the predicted probabilities for a CBA being signed within a firm-community dyad for different levels of our hypothesized variables, holding all other covariates at their sample means. In Panel

1, we observe that while the probability of a CBA is low (5.6%) for communities with no property rights, it

26 rises to 21.2% if the community has use rights, and 58.9% if the community has full property rights (H1). The probability of a firm signing a CBA is 19.7% if the community is subject to externalities (H2), in comparison to 8.1% if it is not (Panel 2). Finally, in Panel 3, we see that in the presence of historical social conflict between the firm and community, the predicted probability of the firm signing a CBA with that community is 99.8%.

This high probability reflects the fact that all firm-community dyads with social conflict sign a CBA.

------Insert Figure 1 ------Robustness to Alternative Samples & Estimation

We perform additional analyses to ensure the robustness of our results to alternatively constructed samples, as well as alternative estimation techniques. We first consider the possibility that there may be something unique about firms that chose to employ formal contracts (i.e. CBAs) to govern their relations with communities. As such, our sample, which is built from a database of all mining projects with at least one

CBA, conditions on firms that have some propensity to use these governance mechanisms. In order to assess whether our results are biased, or not generalizable to the broader population of mining firms, we constructed a quasi-control group of matched mining projects in Canada that had no CBAs. Beginning with the 101 mines in our original sample, we used coarsened exact matching (Blackwell et al. 2010) to match each mine to another mine in the same province, in the same phase of the mining life cycle, and with the same mineral type. We matched to mining projects listed in the Raw Materials Group database, which contains over 1,000 mining projects at various stages of development in Canada. We matched on province because provincial regulatory or other institutions may drive incentives for CBAs. We match on mining phase because the value of CBAs to firms varies by the phase of the mining lifecycle (Dorobantu and

Odziemkowska 2017). Finally we matched on mineral type because the salience of externalities may vary by mineral type (e.g., uranium and gold are highly polluting). Of the 101 mining projects in our original sample, 87 were matched, resulting in a total of 174 mines in our matched sample.

In Model 7, we replicate our main logistic regression model with year and firm fixed effects on the matched sample. Our results remain substantively unchanged. Because observations for which there is no

27 variance in the outcome of interest within a given firm (i.e. the firm has never had a CBA with any indigenous community) do not contribute to the fixed effects logit estimation, the results are based on a subset of our matched sample (i.e. those mines that were owned by a firm that signed a CBA during our sample period). Mines that were never owned by firms that had signed CBAs, therefore are automatically dropped from the effective sample of estimation. To leverage all of the data in our matched sample, we estimate our specifications as linear probability models (LPM) using ordinary least squares with firm fixed effects, which has the advantage of retaining all perfectly predicted groups (Model 8). In Model 9, we simultaneously cluster our standard errors on both the firm and community using two-way clustering

(Kleinbaum, Stuart, and Tushman 2013) to account for the non-independence of firms and communities appearing in multiple dyads.

We also investigated the robustness of our results to the inclusion of only those communities located within 300km of the focal mine. While over 10% of the CBAs in our sample are with communities located more than 300km away, it may be the drivers of such CBAs are materially different from the majority of the sample and may bias our findings. In Model 10, we show the results of a linear probability model on the reduced 300km community sample – the fixed effects logit estimation did not converge. All of the variables of interest remained significant and in the same direction.

In the remaining models, we return to our original sample, although all results are robust to the matched and 300km samples (results available from authors). In Model 11, we include province fixed effects to account for the fact that the historical assignment of property rights in some instances clusters around provincial boundaries. Our results remain substantively unchanged. Finally, to address the fact that our data include few actual events (i.e. CBAs) and a large number of nonevents, we check the robustness of our results to a rare events logistic regression (King and Zeng 2001). Rare events bias can produce inflated standard errors for the coefficients responsible for the infrequently occurring outcome of a CBA between a firm and community. Our results remain essentially unchanged (Model 12) using the rate events logistic regression.

------Insert Table 4 ------

DISCUSSION AND CONCLUSION

Our research investigates how different foundations underlying a firm-community relationship – and, more broadly, a firm-stakeholder relationship – influence the probability that they use a contractual agreement to govern their relationship. A formal contract allows the parties to define the conditions under which value is created and how it is distributed among them. Our findings suggest, first, that firms and local communities are likely to sign a contractual agreement when the community has full property rights over the asset (land) that the firm wants to use. A contract is also likely to be signed, but to a lesser extent, when the community only has use rights over the asset. Second, firms and local communities are likely to sign a contractual agreement when the relationship between them is defined by important externalities, or social costs. Finally, we find that firms and local communities are also likely to use a contract to govern their relationship when they share a history of conflict.

Our research builds and extends prior work on contracting, stakeholder relations, and the more recent research on the relationship between value creation and value capture. Scholars in economics and strategy have studied contracts as a mechanism of governance for a long time. Building on property rights theory (Alchian 1965, Demsetz 1967, Hart and Moore 1990) and transaction costs economics

(Williamson 1985), this research focused on understanding the choice of different contract designs and their implications (Bolton and Dewatripont 2005, Macher and Richman 2008). The question of whether a formal contract is needed or will be signed in the first place has largely been framed through a debate at the intersection of research on interorganizational collaborations (Gulati 1995), trust and social capital

(Adler 2001, Adler and Kwon 2002), and relational governance (Baker et al. 2002, Gibbons and

Henderson 2012). While prior research rooted in sociology has emphasized that the trust developed through repeated interactions may render the signing of contracts irrelevant (Adler 2001, Gulati 1995), economists and strategy scholars have argued that contractual agreements and relational governance complement each other (Baker et al. 2002, Poppo and Zenger 2002, Ryall and Sampson 2009).

In our empirical setting – as in many relationships between firms and nonmarket stakeholders – the process of negotiating and signing a contract takes place either in the absence of other mechanisms of

29 governance (i.e., no prior relationship exists) or against the background of a tense or conflictual historical relationship between the firm and the local community. Our empirical observation leads us to conclude that contractual agreements between firms and indigenous communities are a defining element of the relationship between the two parties. The negotiation, signing and enforcement of these contracts provide a much needed forum for the two parties to come together, adjust their expectations of each other, and reach a shared understanding of the conditions under which they can work together. Trust may or may not develop in the process; to the extent that it does, it is the result of observing the behavior of the other party in the process of negotiating and enforcing the contract itself.

In addition, while our findings that property rights and historical conflict drive the use of formal governance complement and extend existing theories of contracting to nonmarket stakeholders, we also highlight another antecedent to contracting: externalities. While externalities find theoretical grounding in arguments made by both Coase (1960) and Ostrom (1990), they have not garnered much attention in empirical research on firm-stakeholder relationships. This is likely to be an important theoretical oversight, given stakeholders are “any group or individual, that can influence, or is influenced by, the operations of the firm” (Freeman 1984, p.25; emphasis added).

Our research also contributes to a more recent but growing literature on the relationship between value creation and appropriation (see Gans and Ryall 2017 for a recent review). As nicely emphasized by

Coff (1999, p. 120), “[it] is not enough to predict when rent will be generated. In order to predict performance differentials among firms, it is just as important to understand who will appropriate the rent.”

Research that developed this insight, leveraging the theoretical power of cooperative game theory

(Brandenburger and Stuart 1996), has sought to illuminate some of the conditions under which different stakeholders appropriate value created by the firm (Chatain and Zemsky 2011, Garcia-Castro and Aguilera

2015, Lieberman et al. 2016). The focus in this research, however, has been almost entirely on stakeholders within the value chain – customers, capital providers, employees, suppliers, management – with an occasional mention of the government as a stakeholder that captures some of the value through taxation

(Garcia-Castro and Aguilera 2015). Our research highlights that other nonmarket stakeholders also partake

30 in the value creation and appropriation process. We focus our theorizing and empirical analysis on local communities, but we argue that the consideration of all nonmarket stakeholders (local communities, nongovernmental groups, and activists) is an important extension of value creation-appropriation models.

Our examination of contractual agreements between firms and local communities provides insights into the conditions under which local communities are more likely to appropriate some of the value created by a firm not only indirectly through jobs and increased local economic activity, but also directly, through financial transfers and contractually-defined terms for employment, training and procurement. Because contractual agreements between mining firms and indigenous communities in

Canada are confidential, we can only establish that the local community captures some value without being able to identify how much value it captures. Measuring value capture in transactions that involve nonmarket stakeholders will be a challenge for all research in this area. Nonetheless, research that will successfully overcome this challenge will provide important contributions not only to our understanding of value distribution among stakeholders, but also of the conditions that make “shared value” (Porter and

Kramer 2011) more likely.

Finally, our research contributes to the growing literature on stakeholder relations. Stakeholder theory (Freeman 1984, Freeman et al. 2010) has long emphasized that the cooperation of various stakeholders is critical for the survival of a firm, and empirical research has provided strong support for this claim through the examination of the financial value associated with stakeholder cooperation (Dorobantu,

Henisz, et al. 2017, Henisz et al. 2014, Hillman and Keim 2001). But, while recent research has examined various firm responses to stakeholder pressures (Delmas and Toffel 2008, Eesley and Lenox 2006, Hiatt et al. 2015), the use of contractual agreements by firms in managing nonmarket stakeholders has received considerably less attention. Our focus on contractual agreements brings us back to the conceptualization of the firm as a “nexus of contracts” and to stakeholder theory as a paradigm that considers both “the implicit and explicit contractual relationships between all stakeholders” (Hill and Jones 1992: 132).

REFERENCES

Adler PS (2001) Market, Hierarchy, and Trust: The Knowledge Economy and the Future of Capitalism. Organ. Sci. 12(2):215–234. Adler PS, Kwon SW (2002) Social Capital: Prospects for a New Concept. Acad. Manage. Rev. 27(1):17–40. Alcantara C (2003) Individual Property Rights on Canadian Indian Reserves : The Historical Emergence and Jurisprudence of Certificates of Possession. Can. J. Native Stud. 23(2):391–424. Alcantara C (2007) Reduce Transaction Costs? Yes. Strengthen Property Rights? Maybe: The First Nations Land Management Act and Economic Development on Canadian Indian Reserves. Public Choice 132(3):421– 432. Alcantara C (2008) Deal? Or No Deal? Explaining Comprehensive Land Claims Negotiation Outcomes in Canada. (University of Toronto). Alchian AA (1965) Some economics of property rights. Il Polit. 30(4):816–829. Amengual M (2018) Buying stability: The distributive outcomes of private politics in the Bolivian mining industry. World Dev. 104:31–45. Amit R, Schoemaker PJH (1993) Strategic Assets and Organizational Rent. Strateg. Manag. J. 14(1):33–46. Argyres NS, Liebeskind JP (1999) Contractual Commitments, Bargaining Power, and Governance Inseparability: Incorporating History Into Transaction Cost Theory. Acad. Manage. Rev. 24(1):49–63. Argyres NS, Zenger TR (2012) Capabilities, transaction costs, and firm boundaries. Organ. Sci. 23(6):1643–1657. Asher CC, Mahoney JM, Mahoney JT (2005) Towards a property rights foundation for a stakeholder theory of the firm. J. Manag. Gov. 9(1):5–32. Baker G, Gibbons R, Murphy KJ (2002) Relational Contracts and the Theory of the Firm. Q. J. Econ. 117(1):39–84. Barney J (1991) Firm Resources and Sustained Competitive Advantage. J. Manag. 17(1):99–120. Bartley T, Child C (2014) Shaming the Corporation The Social Production of Targets and the Anti-Sweatshop Movement. Am. Sociol. Rev. 79(4):653–679. Blackwell M, Iacus S, King G, Porro G (2010) Cem: Coarsened exact matching in Stata. Stata J. 9(4):524–546. Bolton P, Dewatripont M (2005) Contract Theory. MIT Press. Brandenburger AM, Stuart HW (1996) Value-Based Business Strategy. J. Econ. Manag. Strategy 5(1):5–24. Cain C (2014) Negotiating With the Growth Machine: Community Benefits Agreements and Value-Conscious Growth. Sociol. Forum 29(4):937–958. Castanias RP, Helfat CE (1991) Managerial Resources and Rents. J. Manag. 17(1):155–171. Chatain O, Zemsky P (2011) Value creation and value capture with frictions. Strateg. Manag. J. 32(11):1206–1231. Clarkson ME (1995) A stakeholder framework for analyzing and evaluating corporate social performance. Acad. Manage. Rev. 20(1):92–117. Coase RH (1960) The Problem of Social Cost. J. Law Econ. 3:1–44. Coase RH (1992) The Institutional Structure of Production. Am. Econ. Rev. 82(4):713–719. Coase RH (2008) The Nature of the Firm. Economica 4(16):386–405. Coff RW (1999) When competitive advantage doesn’t lead to performance: The resource-based view and stakeholder bargaining power. Organ. Sci. 10(2):119–133. Delmas MA, Toffel MW (2008) Organizational responses to environmental demands: opening the black box. Strateg. Manag. J. 29(10):1027–1055. Demsetz H (1967) Toward a Theory of Property Rights. Am. Econ. Rev. 57(2):347–359. Donaldson T, Preston LE (1995) The stakeholder theory of the corporation: concepts, evidence, and implications. Acad. Manage. Rev. 20(1):65–91. Dorobantu S, Henisz WJ, Nartey LJ (2017) Not All Sparks Light a Fire: Stakeholder and Shareholder Reactions to Critical Events in Contested Markets. Adm. Sci. Q. 62(3):561–597. Dorobantu S, Kaul A, Zelner BA (2017) Nonmarket strategy through the lens of new institutional economics: A review and future directions. Strateg. Manag. J. 38(1):114–140. Dorobantu S, Odziemkowska K (2017) Valuing Stakeholder Governance: Property Rights, Community Mobilization, and Firm Value. Strateg. Manag. J. 38(13):2682–2703. Dorobantu S, Walter I (2018) The Dakota Access Pipeline. Earl J, Martin A, McCarthy JD, Soule SA (2004) The Use of Newspaper Data in the Study of Collective Action. Annu. Rev. Sociol. 30(1):65–80. Eesley C, Lenox MJ (2006) Firm responses to secondary stakeholder action. Strateg. Manag. J. 27(8):765–781. Flammer C (2013) Corporate Social Responsibility and Shareholder Reaction: The Environmental Awareness of Investors. Acad. Manage. J. 56(3):758–781.

Franks DM, Davis R, Bebbington AJ, Ali SH, Kemp D, Scurrah M (2014) Conflict translates environmental and social risk into business costs. Proc. Natl. Acad. Sci. 111(21):7576–7581. Freeman RE (1984) Strategic management: a stakeholder approach (Pitman, Boston). Freeman RE, Evan WM (1990) Corporate governance: A stakeholder interpretation. J. Behav. Econ. 19(4):337–359. Freeman RE, Harrison JS, Wicks AC (2007) Managing for Stakeholders: Survival, Reputation, and Success (Yale University Press, New Haven). Freeman RE, Harrison JS, Wicks AC, Parmar BL, Colle S de (2010) Stakeholder Theory: The State of the Art (Cambridge University Press, Cambridge ; New York). Frooman J (1999) Stakeholder Influence Strategies. Acad. Manage. Rev. 24(2):1991–205. Gans J, Ryall MD (2017) Value capture theory: A strategic management review. Strateg. Manag. J. 38(1):17–41. Garcia-Castro R, Aguilera R V. (2015) Incremental value creation and appropriation in a world with multiple stakeholders. Strateg. Manag. J. 36(1):137–147. Gibbons R, Henderson R (2012) Relational Contracts and Organizational Capabilities. Organ. Sci. 23(5):1350– 1364. Greene WH (2014) Econometric Analysis. Seventh edition. Gulati R (1995) Does Familiarity Breed Trust? The Implications of Repeated Ties for Contractual Choice in Alliances. Acad. Manage. J. 38(1):85–112. Hart O (1989) An Economist’s Perspective on the Theory of the Firm. Columbia Law Rev. 89(7):1757–1774. Hart O (1995) Firms, Contracts, and Financial Structure (Clarendon Press). Hart O, Moore J (1990) Property Rights and the Nature of the Firm. J. Polit. Econ. 98(6):1119–1158. Henisz WJ, Dorobantu S, Nartey LJ (2014) Spinning gold: the financial returns to external stakeholder engagement. Strateg. Manag. J. 25(12):1727–1748. Hiatt SR, Grandy JB, Lee BH (2015) Organizational Responses to Public and Private Politics: An Analysis of Climate Change Activists and U.S. Oil and Gas Firms. Organ. Sci. 26(6):1769–1786. Hill CWL, Jones TM (1992) Stakeholder-Agency Theory. J. Manag. Stud. 29(2):131–154. Hillman AJ, Keim GD (2001) Shareholder value, stakeholder management, and social issues: What’s the bottom line? Strateg. Manag. J. 22(2):125–139. Jensen MC, Meckling WH (1979) Theory of the Firm: Managerial Behavior, Agency Costs, and Ownership Structure. Brunner K, ed. Econ. Soc. Inst. Rochester Studies in Economics and Policy Issues. (Springer Netherlands), 163–231. Kale P, Singh H, Perlmutter H (2000) Learning and protection of proprietary assets in strategic alliances: building relational capital. Strateg. Manag. J. 21(3):217–237. Kassinis G, Vafeas N (2006) Stakeholder Pressures And Environmental Performance. Acad. Manage. J. 49(1):145– 159. Keay I, Metcalf C (2011) Property Rights, Resource Access, and Long-Run Growth. J. Empir. Leg. Stud. 8(4):792– 839. King A (2007) Cooperation between corporations and environmental groups: A transaction cost perspective. Acad. Manage. Rev. 32(3):889–900. King G, Zeng L (2001) Logistic Regression in Rare Events Data. Polit. Anal. 9(2):137–163. Lieberman MB, Garcia-Castro R, Balasubramanian N (2016) Measuring value creation and appropriation in firms: The VCA model. Strateg. Manag. J.:n/a-n/a. Macaulay S (1963) Non-Contractual Relations in Business: A Preliminary Study. Am. Sociol. Rev. 28(1):55–67. Macher JT, Richman BD (2008) Transaction Cost Economics: An Assessment of Empirical Research in the Social Sciences. Bus. Polit. 10(1). MacNeil IR (1973) The Many Futures of Contracts. South. Calif. Law Rev. 47:691. Mayer KJ, Xing ZA (2017) Contractual design of alliances. Collab. Strategy Crit. Issues Alliances Netw. (Edward Elgar Pub, Cheltenham, Glos : Northampton, Massachusetts). McAdam D (1982) Political Process and the Development of Black Insurgency, 1930-1970 (The University Of Chicago Press, Chicago). Mitchell RK, Agle BR, Wood DJ (1997) Toward a theory of stakeholder identification and salience: Defining the principle of who and what really counts. Acad. Manage. Rev. 22(4):853–886. O’Faircheallaigh C (2013) Community development agreements in the mining industry: an emerging global phenomenon. Community Dev. 44(2):222–238. O’Faircheallaigh C (2015) Social Equity and Large Mining Projects: Voluntary Industry Initiatives, Public Regulation and Community Development Agreements. J. Bus. Ethics 132(1):91–103.

Orts EW, Strudler A (2002) The Ethical and Environmental Limits of Stakeholder Theory. Bus. Ethics Q. 12(2):215–233. Ostrom E (1990) Governing the Commons: The Evolution of Institutions for Collective Action (Cambridge University Press, Cambridge ; New York). Ostrom E (2010) Beyond Markets and States: Polycentric Governance of Complex Economic Systems. Am. Econ. Rev. 100(3):641–672. Ostrom E, Gardner R, Walker J (1994) Rules, Games, and Common-Pool Resources (University of Michigan Press, Ann Arbor). Parks V, Warren D (2009) The Politics and Practice of Economic Justice: Community Benefits Agreements as Tactic of the New Accountable Development Movement. J. Community Pract. 17(1–2):88–106. Penrose ET (1959) The Theory of the Growth of the Firm (John Wiley & Sons, New York). Pfeffer J, Salancik GR (1978) The external control of organizations: A resource dependence perspective (Harper & Row, New York, NY). Pigou AC (1932) The Economics of Welfare: Volume I (Cosimo Classics, New York). Poppo L, Zenger T (2002) Do Formal Contracts and Relational Governance Function as Substitutes or Complements? Strateg. Manag. J. 23(8):707–725. Porter ME, Kramer MR (2011) Creating shared value. Harv. Bus. Rev. (January). Ring PS, Van De Ven AH (1994) Developmental Processes of Cooperative Interorganizational Relationships. Acad. Manage. Rev. 19(1):90–118. Rowley TJ (2016) MOVING BEYOND DYADIC TIES : A NETWORK THEORY OF STAKEHOLDER INFLUENCES. 22(4):887–910. Rowley TJ, Moldoveanu M (2003) When will stakeholder groups act? An interest- and identity-based model of stakeholder group mobilization. Acad. Manage. Rev. 28(2):204–219. Ryall MD, Sampson RC (2009) Formal Contracts in the Presence of Relational Enforcement Mechanisms: Evidence from Technology Development Projects. Manag. Sci. 55(6):906–925. Salkin P, Lavine A (2008) Negotiating for social justice and the promise of community benefits agreements: Case studies of current and developing agreements. J Afford. Hous Cmty Dev L 17(1–2):113–144. Schepker DJ, Oh WY, Martynov A, Poppo L (2014) The Many Futures of Contracts: Moving Beyond Structure and Safeguarding to Coordination and Adaptation. J. Manag. 40(1):193–225. Selznick P (1949) TVA and the Grass Roots: A Study in the Sociology of Formal Organization (University of California Press, Berkeley, CA). Sosa I, Keenan K (2001) Impact benefit agreements between aboriginal communities and mining companies: Their use in Canada (Canadian Environmental Law Association, Ottawa). Stigler GJ (1989) Two Notes on the Coase Theorem. Yale Law J. 99:631. Tantalo C, Priem RL (2016) Value creation through stakeholder synergy. Strateg. Manag. J. 37(2):314–329. Wilkes R, Corrigall-Brown C, Myers DJ (2010) Packaging Protest: Media Coverage of Indigenous People’s Collective Action. Can. Rev. Sociol. Can. Sociol. 47(4):327–357. Williamson OE (1985) The Economic Institutions of Capitalism (Free Press, New York; London). Williamson OE (1999) Strategy Research: Governance and Competence Perspectives. Strateg. Manag. J. 20(12):1087–1108. Wilson K, Peters EJ (2005) “You can make a place for it”: Remapping urban First Nations spaces of identity. Environ. Plan. Soc. Space 23(3):395–413. Yue LQ, Rao H, Ingram P (2013) Information Spillovers from Protests against Corporations: A Tale of Walmart and Target. Adm. Sci. Q. 58(4):669–701. Zald MN, McCarthy J (1987) Social Movements in an Organizational Society Reprint ed. (Transaction Publishers, New Brunswick, U.S.A.).

Table 1: Summary Statistics

Obs Mean SD Min Max CBA within firm-community dyad 38,556 0.022 0.148 0 1 Property rights, continuous (H1) 38,556 0.361 0.483 0 2 No property rights (H1) 38,556 0.641 0.480 0 1 Use rights (H1) 38,556 0.358 0.479 0 1 Full property rights (H1) 38,556 0.001 0.038 0 1 Externalities (H2) 38,556 0.142 0.349 0 1 Social conflict (H3) 38,556 0.000 0.020 0 1 Mine-community geodesic distance 38,556 2.930 1.289 0 4.999 Legal conflict 38,556 0.000 0.013 0 1 Social and legal conflict history 38,556 1.102 2.125 0 16 Community's past CBAs 38,556 0.783 1.892 0 15 Proximate community CBAs 38,556 3.621 3.849 0 19 Community population 38,556 6.746 0.873 3.689 9.361 Employment rate 38,556 0.434 0.118 0.111 0.833 Indigenous language speakers 38,556 0.275 0.273 0 1.000 Mine value to the firm 38,556 0.113 1.029 0 23.718 Mine phase 38,556 2.936 1.666 0 7 Firm size 38,556 18.671 2.849 13.325 26.084 Other conflict against firm 38,556 0.105 0.472 0 3 Firm's past CBAs 38,556 0.998 1.610 0 10 Media attention to indigenous issues 38,556 17.342 6.508 6.392 26.787 CBAs in province 38,556 38.330 33.906 0 111

Table 2: Correlations (N=38,556)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 1 CBA within firm-community dyad 1 2 Property rights, continuous (H1) 0.150 1 3 Externalities (H2) 0.244 0.216 1 4 Social conflict (H3) 0.068 0.018 0.026 1 5 Mine-community geodesic distance -0.206 -0.287 -0.490 -0.014 1 6 Legal conflict 0.063 0.006 0.033 0.000 -0.026 1 7 Social and legal conflict history 0.052 0.182 -0.015 0.028 0.056 0.003 1 8 Community's past CBAs 0.257 0.035 0.035 0.048 -0.064 0.002 0.020 1 9 Proximate community CBAs 0.071 0.007 -0.002 -0.005 -0.098 0.029 0.011 0.328 1 10 Community population -0.013 -0.237 -0.069 0.010 0.121 -0.001 0.162 -0.065 -0.120 1 11 Employment rate 0.034 0.064 0.019 -0.005 -0.049 -0.001 -0.009 0.121 0.113 -0.303 1 12 Indigenous language speakers 0.048 -0.167 -0.043 -0.011 0.132 -0.002 0.049 0.187 0.074 0.175 -0.194 1 13 Mine value to the firm 0.000 0.066 0.012 -0.002 -0.017 0.000 0.057 -0.010 -0.026 -0.005 -0.038 -0.037 1 14 Mine phase 0.083 0.167 0.003 0.009 -0.075 0.005 0.094 0.072 0.142 -0.006 0.022 -0.070 -0.034 1 15 Firm size 0.012 0.000 0.011 -0.005 -0.029 -0.004 -0.022 -0.060 -0.096 0.050 -0.071 -0.009 -0.099 0.312 1 16 Other conflict against firm -0.002 -0.052 -0.031 0.076 0.016 0.009 0.044 0.007 -0.011 0.040 0.009 -0.031 -0.013 0.019 0.205 1 17 Firm's past CBAs 0.111 -0.176 0.005 -0.002 -0.020 -0.001 -0.004 0.102 0.150 0.154 -0.086 0.060 -0.038 0.190 0.233 0.283 1 18 Media attention to indigenous issues 0.076 -0.010 -0.014 0.002 -0.009 0.013 0.090 0.181 0.426 0.052 0.009 -0.002 0.008 0.251 -0.094 0.018 0.283 1 19 CBAs in province 0.034 -0.137 -0.070 0.001 0.025 0.020 0.030 0.253 0.615 0.005 0.129 0.038 -0.045 0.218 -0.115 -0.005 0.242 0.559 1

Table 3: Logistic Regression of Probability of CBA within Firm-Community Dyad

Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Independent variables Property rights, continuous (H1) 1.659*** 1.546*** (0.497) (0.443) Use rights (H1) 1.604** (0.620) Full property rights (H1) 3.451* (1.437) Externalities (H2) 1.147** 1.022** (0.356) (0.357) Social conflict (H3) 10.56*** 10.06*** (1.169) (1.094) Dyad controls Mine-community geodesic distance -2.784*** -2.554*** -2.552*** -2.422*** -2.791*** -2.266*** (0.357) (0.338) (0.339) (0.355) (0.361) (0.341) Legal conflict 1.769 1.766 1.760 1.500 1.788 1.508 (1.102) (1.197) (1.202) (1.159) (1.133) (1.283) Community controls Social and legal conflict history 0.0597 0.0562 0.0570 0.0524 0.0452 0.0334 (0.062) (0.060) (0.060) (0.065) (0.064) (0.064) Community's past CBAs 0.383*** 0.383*** 0.382*** 0.404*** 0.382*** 0.401*** (0.059) (0.059) (0.060) (0.063) (0.059) (0.063) Proximate community CBAs 0.0558 0.0589 0.0588 0.0601 0.0611 0.0684+ (0.038) (0.042) (0.042) (0.038) (0.038) (0.041) Community population -0.0564 -0.0891 -0.0889 -0.0395 -0.0656 -0.0875 (0.190) (0.200) (0.200) (0.183) (0.191) (0.195) Employment rate 0.612 0.293 0.290 0.716 0.660 0.476 (1.153) (1.097) (1.099) (1.151) (1.158) (1.128) Indigenous language speakers 0.0194 0.109 0.117 -0.0557 0.161 0.202 (0.823) (0.780) (0.789) (0.795) (0.828) (0.780) Mine controls Mine value to the firm -0.0655 -0.118 -0.122 -0.0758 -0.0652 -0.140 (0.102) (0.111) (0.110) (0.103) (0.101) (0.121) Mine phase 0.231 0.238 0.235 0.226 0.226 0.230 (0.156) (0.153) (0.155) (0.153) (0.156) (0.148) Firm controls Firm size -0.0889 -0.0872 -0.0857 -0.0870 -0.0867 -0.0855 (0.073) (0.073) (0.076) (0.073) (0.073) (0.073) Other conflict against firm -0.964* -0.927* -0.929* -0.842+ -1.046** -0.916* (0.392) (0.372) (0.371) (0.461) (0.377) (0.409) Firm's past CBAs -0.112 -0.0850 -0.0853 -0.111 -0.113 -0.0848 (0.073) (0.073) (0.073) (0.073) (0.072) (0.074) Institutional controls Media attention to indigenous issues 0.266*** 0.254*** 0.255*** 0.270*** 0.272*** 0.266*** (0.069) (0.068) (0.069) (0.070) (0.068) (0.069) CBAs in province -0.0163+ -0.0128 -0.0129 -0.0167+ -0.0174+ -0.0147+ (0.010) (0.009) (0.009) (0.009) (0.009) (0.009) Fixed effects Year, firm Year, firm Year, firm Year, firm Year, firm Year, firm N 38556 38556 38556 38556 38556 38556 ll -1353.6 -1299.2 -1299.1 -1317.8 -1339.5 -1258.4 Robust standard errors clustered at firm-level in parentheses. +p<0.1; ∗p<0.05; ∗∗ p<0.01; ∗∗∗ p<0.001.

Figure 1: Predicted Probabilities of CBA within Firm-Community Dyad

Panel 1 Panel 2 Predicted probabilities of CBA by property rights Predicted probabilities of CBA by externalities

B .

. No rights Use rights Full rights No externalities Externalities Property rights (H1) Externalities (H2)

Panel 3 Predicted probabilities of CBA by social conflict

i Note. Figures plot predicted probabilities from a logistic

4 regression model with year and firm fixed effects (Model 6)

b .

o r predicting a CBA within a firm-community dyad as a function of

P the community’s property rights over the mine (Panel 1), whether

2 . the community is subject to environmental externalities (Panel 2), and whether the relationship has past conflict (Panel 3). All other

0 variables held at their sample means. No social conflict Social conflict Social conflict (H3)

Table 4: Sample & Estimation Robustness of Probability of CBA

Matched Sample 300km Sample Original Sample Estimation Logistic LPM LPM LPM Logistic Rare events Model 7 Model 8 Model 9 Model 10 Model 11 Model 12 Independent variables Property rights, continuous (H1) 1.614*** 0.0128* 0.0128+ 0.0345* 1.459*** 1.734*** (0.452) (0.006) (0.007) (0.017) (0.397) (0.442) Externalities (H2) 1.621*** 0.0461*** 0.0461*** 0.0341*** 1.000** 0.999** (0.435) (0.009) (0.011) (0.010) (0.369) (0.348) Social conflict (H3) 3.593*** 0.330*** 0.330*** 0.484*** 10.62*** 9.525*** (0.593) (0.054) (0.021) (0.113) (1.160) (1.062) Dyad controls Mine-community geodesic distance -1.967*** -0.00722*** -0.00722*** -0.0546*** -2.442*** -1.805*** (0.362) (0.002) (0.002) (0.008) (0.341) (0.410) Legal conflict 0.997 0.323 0.323 0.395 1.471 2.791+ (1.171) (0.259) (0.260) (0.264) (1.439) (1.666) Community controls Social and legal conflict history 0.0830 0.00235 0.00235 0.00553 0.0316 0.0575 (0.059) (0.002) (0.002) (0.004) (0.066) (0.061) Community's past CBAs 0.344*** 0.0137*** 0.0137*** 0.0322*** 0.419*** 0.383*** (0.060) (0.002) (0.003) (0.005) (0.062) (0.058) Proximate community CBAs 0.122*** 0.00157 0.00157 0.00280 0.0849* 0.0472 (0.035) (0.001) (0.001) (0.002) (0.039) (0.042) Community population -0.0767 0.000853 0.000853 0.00566 -0.161 -0.0232 (0.214) (0.002) (0.002) (0.005) (0.199) (0.195) Employment rate 0.368 -0.0247* -0.0247+ -0.0215 0.0493 1.988 (1.286) (0.012) (0.014) (0.031) (1.176) (1.343) Indigenous language speakers 0.109 -0.0155** -0.0155+ -0.0586** -0.510 1.760+ (0.572) (0.005) (0.008) (0.019) (0.693) (1.039) Mine controls Mine value to the firm -0.423 0.000215* 0.000215+ -0.0000389 -0.0614 -0.0646 (0.422) (0.000) (0.000) (0.001) (0.075) (0.141) Mine phase 0.162 0.000126 0.000126 -0.00190 0.249* 0.127 (0.103) (0.002) (0.002) (0.005) (0.111) (0.148) Firm controls Firm size -0.170 -0.00170 -0.00170 -0.00348 -0.0234 -0.000148 (0.135) (0.001) (0.001) (0.003) (0.072) (0.069) Other conflict against firm -0.633* -0.00236 -0.00236 -0.0236*** -1.119** -0.818+ (0.303) (0.002) (0.003) (0.006) (0.368) (0.461) Firm's past CBAs -0.0527 0.000787 0.000787 0.00226 -0.116 -0.0741 (0.076) (0.001) (0.001) (0.002) (0.078) (0.065) Institutional controls Media attention to indigenous issues 0.164** 0.00332 0.00332 -0.000650 0.311*** 0.295*** (0.060) (0.002) (0.002) (0.003) (0.079) (0.065) CBAs in province -0.00257 -0.000361+ -0.000361+ -0.000877+ -0.0217* -0.0179* (0.008) (0.000) (0.000) (0.000) (0.008) (0.008) Fixed effects Year, firm Year, firm Year, firm Year, firm Year, firm, Year, firm province N 45284 55346 55346 19127 38556 38556 ll -1155.2 -1186.6 Robust standard errors clustered at firm-level in parentheses, except Model 9 where standard errors clustered on firm and community. Models 7 to 9 employ the CEM matched sample of 174 mines. Models 10 to 12 employ the original 101 mine sample. +p<0.1; ∗p<0.05; ∗∗ p<0.01; ∗∗∗ p<0.001.