A cross-level process theory of trust development in interorganizational relationships Abstract
Most research on trust in inter-organizational relationships focuses on a single level of analysis, typically the individual or organizational level, and treats trust as a fairly static phenomenon.

To stimulate more cross-level research, we propose a theoretical model that explains how trust in inter-organizational relationships is related across various levels of analysis. At the same time, our model emphasizes the dynamic aspect of trust by examining how trust develops throughout consecutive relationship stages. Drawing from several programs of research, we identify the mechanisms that drive the progression of trust across levels as the inter-organizational relationship unfolds. Starting with the boundary spanner as the key individual at the beginning of a new collaboration, we specify how trust gradually becomes part of the fabric of organizational action. By integrating micro and macro approaches over time, the proposed model contributes to a better understanding of how trust evolves in inter-organizational relationships.
Keywords
Alliance dynamics, cross-level effects, institutionalization, inter-organizational relationships, multilevel theory, strategic alliances, trust
Introduction
In the development of the management of inter-organisational relationships, the issue of trust has increasingly become the main focus (MacDuffie, 2011; Zaheer and Harris, 2006).

As such relationships have a naturally uncertain dynamic (Lumineau and Quélin, 2012; McEvily et al., 2003a), in and of themselves, trust can be utilized as a key instrument in managing the level of uncertainty (Rousseau et al., 1998).

Although there has been an increase in interest amongst academics and scholars, towards the role trust plays in inter-organisations, the existing literature nevertheless is still very piecemeal (McEvily et al., 2003b). To be precise, previous research has examined trust either at the individual level, (e.g. Child and Möllering, 2003; Jap and Anderson, 2003) or at the organizational level (e.g. Das and Teng, 1998; Poppo et al., 2008). Up till now, there has been not much effort directed to integrate the varying levels of analysis or connect/incorporate them with the specific stages of trust development.

The aim of this paper is to work towards the development of a potential theoretical framework that would expound on the main paradigms and processes, at both the individual and organizational levels, which constitute the cross-level development of trust in inter-organizational relationships. At the heart of this model are the causal mechanisms that drive the transitions between levels of analysis over the course of a relationship. Starting with the boundary spanner as the key individual at the beginning of a new collaboration, we specify how trust gradually evolves and becomes part of the fabric of organizational-level actions.

This article contributes to the literature on trust in interorganizational relationships in several ways. First, it makes an integrative contribution by promoting conversation on the dynamics of trust development across levels of analysis in an effort to better understand how micro and macro approaches are interrelated. Second, it establishes individual–individual, individual–organization, and organization–organization trust as related yet distinct constructs whose antecedents and consequences require differentiated consideration. Third, it contributes to the nascent process perspective on trust in interorganizational relationships by elucidating how the nature of trust evolves across relationship stages. Fourth, by identifying key mechanisms underlying the emergence of interorganizational trust, it helps to substantiate the ontological status of organization–organization trust, which has notably been questioned. Fifth, it directs attention to intraorganizational processes as important interorganizational trust antecedents that have been largely unexamined in past work.
Conceptual background
Trust in interorganizational relationships
Definition of trust. Although the literature presents many different definitions of trust, most include an element of perceived risk and the vulnerability that comes with risk (Edmondson, 2004). Trust is said to reflect “the willingness of a party to be vulnerable to the actions of another party based on the expectation that the other will perform a particular action important to the party” (Mayer et al., 1995: 712). In this relational conceptualization of trust (Cook and Schilke, 2010), the term “party” may refer not only to persons but also to collective actors or corporations (Schoorman et al., 2007: 345). Thus, the definition is applicable to both individuals and organizations. At the individual– individual level, interpersonal trust pertains to the expectation of one individual that another individual will not take advantage of one’s vulnerabilities (e.g. McAllister, 1995: 25), while at the organization–organization level, interorganizational trust refers to one organization’s expectation that another organization will not take advantage of the trusting organization’s vulnerabilities (e.g. Gulati and Nickerson, 2008: 1).1
Levels of analysis involved in the trust relationship. In characterizing the nature of trust, a distinction is typically made between the respective parties involved in the trust relationship (Mayer et al., 1995)—that is, between a “trustor” and a “trustee.” The trustor is the trusting party that holds certain expectations, while the trustee is the party that is assessed by the trustor as trustworthy or not. In the context of interorganizational relationships, both the trustor and the trustee can be represented by different levels of analysis (Currall and Inkpen, 2002), either by an individual or by an organization.2
The individuals most relevant to the implementation and management of interorganizational relationships are “boundary spanners”—relationship managers who perform important functions with respect to communication and monitoring (Currall and Judge, 1995; Perrone et al., 2003). These boundary spanners are often project managers, department heads, or their representatives, who are primarily in charge of the relevant interorganizational relationships and tend to be more closely involved in these relationships than other members of the organization (Currall and Inkpen, 2003; Dodgson, 1993).
Focusing on the individual and organizational levels of analysis, the following three categorically distinct trust relations can be recognized as relevant to trust in interorganizational relationships:3 individual–individual (interpersonal), individual–organization (institutional), and organization–organization (interorganizational). These relations are the building blocks of the cross-level process model of trust we develop in this article.
Process theory
Our goal is to build a process theory of trust development, rather than a variance theory of trust correlates (see Langley, 2007; Poole et al., 2000, regarding the distinction between process and variance theories). A process theory is based on a narrative explaining the temporal sequence in which change occurs to produce a given outcome (e.g. first A and then B to get C). It identifies separable stages and transitions between these stages.4 Therefore, a process theory consists of explanations of how, why, and in what sequence a process unfolds over time.
Temporal order is particularly important for our investigation of interorganizational relationships. These relationships go through successive phases, each of which involves distinct activities, people, and—as we will argue—trust relations (i.e. individual–individual, individual–organization, or organization–organization trust). Therefore, we contend that specification of a process is key to understanding the nature of trust in these relationships.5
Toward a meso-level paradigm of trust
To study the links between different trust relations, we draw on the “meso” approach to theorizing in organizational research. House et al. (1995) describe meso-level research as concerned with the simultaneous study of at least two levels of analysis in which (a) one or more levels concern individual or group behavioral processes or variables, (b) one or more levels concern organizational processes or variables, and (c) the processes by which the levels of analysis are related are articulated. (p. 73)
In line with this meso approach, we provide a conceptual model that advances knowledge of the cross-level relationships involved in building trust within interorganizational relationships. That is, we specify and link the individual- and organizational-level trust relations within the same conceptual framework.
An important precondition for building such a meso framework is the determination of whether isomorphism exists among similar constructs at the different levels of analysis included in the model (Rousseau, 1985). Isomorphism is said to exist when the underlying structure of a construct (i.e. its properties and relations) is the same across levels (Rousseau and House, 1994).6 Organizational trust is fundamentally different from individual trust in its structure and its drivers (Ferrin et al., 2007; Lui et al., 2006; Rousseau, 2004).7 In a case study of six interfirm alliances in the biotech industry, for instance, Zaheer et al. (2002) find that informants were clearly able to distinguish between these two forms of trust with little effort. This finding supports the notion that trust in interorganizational relationships is not isomorphic across levels of analysis, and that, to avoid misspecification, a theory of trust in interorganizational relationships should explicitly address different levels and their interplay.
The next step in developing a meso theory is to specify the functional relationships between the constructs at different levels in an integrated theoretical model (Klein and Kozlowski, 2000). Initial support for a positive relationship between trust constructs across levels can be found in a study of 107 interfirm relationships between buyers and suppliers in the electrical equipment manufacturing industry. Zaheer et al. (1998) report a positive correlation between two forms of trust, namely, trust between individuals in two different firms and trust between an individual in one firm and the partner firm. However, the results do not reveal any causal direction in the relationship between these constructs, and the cross-sectional nature of the study did not provide insights into relevant causal mechanisms and temporal development stages. In the next section, we develop a process model that accounts for temporal dynamics and specifies the mechanisms that drive changes in the levels of analysis involved in the trust relationship.
The cross-level process of trust development in interorganizational relationships
Organizing scheme of the proposed model
Our cross-level framework of trust evolution in interorganizational relationships contains four consecutive stages. Each stage is characterized by a specific relation between a trustor and a trustee in which both parties (the trustor and the trustee) can be either an individual (the boundary spanner) or an organization, respectively. Consistent with a process theory approach, we propose relevant mechanisms that underlie the cross-level trust dynamics involved.
To simplify, we follow prior process research (e.g. Lewicki and Bunker, 1996; Worchel and Coutant, 2001) and initially describe the process model as it may apply generally. Importantly, we do not posit that the proposed process is completely deterministic or even applicable to every interorganizational relationship. Rather, following Poole et al.’s (2000) recommendation, we identify a baseline model that applies in general with high probability. As Staw (2005) puts it, “the temporal model represents a prototype around which individual cases will no doubt vary” (p. 228). We consider relevant contingencies that might be responsible for such variation in our discussion section.
In addition to specifying its central subject (trust in this case) and the type of change observed (i.e. changes in levels of analysis), a process theory should also articulate its logic for determining a beginning and an end, the sequence that connects the two, and its focal entities (Pentland, 1999). When describing our theoretical model, we take the viewpoint of an organization involved in the formation of a new interorganizational relationship (referred to as the focal organization), and we follow the process until the relationship with the partner organization is dissolved. Thus, we adopt a life cycle approach to process theorizing (Poole et al., 2000), with the typical progression of developmental stages of an interorganizational relationship serving as the anchor for the proposed sequence of trust development,8 structured around four common relationship stages (see Das and Teng, 2002b, for a review): relationship initiation, negotiation, formation, and operation. Potential partners are first identified, evaluated, and selected in the initiation stage. Second, negotiations among prospective partners take place. Third, the formation stage involves setting up the partnership by committing various types of resources. Fourth, partner organizations collaborate and implement the agreement in the operation stage.

As depicted in Figure 1, the suggested development of trust relations coevolves with these partnership stages and reflects the specific activities and parties involved. To briefly summarize the overall structure of the model, we argue that the focal organization’s boundary spanner is the starting point of the trust process. When evaluating a potential partner during the relationship initiation stage, the boundary spanner tries to predict the partner’s degree of trustworthiness. To this end, the boundary spanner gathers clues regarding the trustworthiness of the partner organization, including information regarding prior interorganizational interactions, the partner organization’s general reputation, and/or relevant categories to which it belongs. As a result of the information gained, a basis of individual–organization trust develops. In the second phase of the model, the boundary spanner communicates with his or her individual counterpart in the partner organization, engaging in negotiations. These interpersonal negotiations significantly shape the boundary spanner’s trust beliefs (i.e. individual–individual level trust). Third, as the partnership is formed, we propose that the boundary spanner “transfers” trust in his or her individual counterpart to the partner organization (initiating individual–organization level trust). Subsequently, in the fourth stage, when the relationship enters the operation stage, a common understanding regarding the trustworthiness of the partner organization develops and organizational alliance routines are established reflecting “how things are done” with this partner organization (i.e. the establishment and institutionalization of organization–organization trust occurs). Finally, in cases that involve a new relationship between organizations in which there is prior interorganizational experience, organizational-level trust feeds back on the boundary spanners’ trust beliefs pertaining to the partner organization, perhaps altering the level of individual–organization trust. A detailed discussion of each of these stages and their linking mechanisms follows.
From the individual to the organization: gathering clues about trustworthiness
Research on swift trust indicates that, prior to their first encounter with another party, individuals tend to construct initial trust beliefs regarding that party (Meyerson et al., 1996). In the context of establishing a new interorganizational relationship, the formation of preliminary trust beliefs is particularly relevant during the relationship initiation stage. We propose that boundary spanners start the trust-development process by gathering clues about the trustworthiness of the potential partner organization. Building on the organizational embeddedness approach (Uzzi, 1996, 1997), relevant clues about the trustworthiness of a future partner organization may come from three sources: (1) from the organizations’ relevant prior interactions, (2) from public information (general reputation), and/or (3) from institutional categories to which the partner organization belongs.
Prior interactions. Managers can certainly draw inferences about the trustworthiness of the partner organization from previous interactions between the organizations. These interactions may include not only prior interorganizational relationships but also arm’s length relationships or even encounters during mergers and acquisitions. The idea of trust emerging from prior contact is based on the premise that the organizations have had previous opportunities to learn about each other (Burt and Knez, 1996; Uzzi, 1997). Information regarding previous interactions is stored within the organization and can be accessed by its members, even if they were not personally involved in those interactions. Past experiences transform an “undersocialized relationship” into one in which the history of prior relations and interactions forms an informal social institution capable of supporting trust building.

Proposition 1A. Prior interactions provide a basis for individual–organization trust during the initiation stage.
Parkhe (1993) provides some empirical support for the idea that prior exchange supports the building of trust beliefs in new strategic alliances. In a survey study, he observes that a history of interactions between two firms limits managers’ expectations of opportunistic behavior on the part of the partner firm in new alliances.
General reputation. If the boundary spanner has no priors based on previous interactions with the focal organization, the boundary spanner’s initial trust perceptions of that organization are likely to be based on an assessment of the organization’s general reputation in the marketplace. Research in negotiations shows that, especially in the absence of direct prior experience with a specific party, one will prepare for interactions with that party by gathering reputational information (Glick and Croson, 2001). The influence of reputation is particularly salient in an uncertain environment characterized by incomplete and ambiguous information (Lui and Ngo, 2005), such as in the early stages of negotiations, when information about the other partner—particularly about the managers involved—is incomplete. In the context of interorganizational relationships, relevant information available to the boundary spanner pertains to the partner organization’s general reputation in the marketplace (i.e. market knowledge of that organization’s typical behavior (Uzzi, 1997)). This information can be obtained from the press, as well as from specialized business information services.
Proposition 1B. General reputation provides a basis for individual–organization trust during the initiation stage.
Institutional categories. Especially in the absence of information derived from prior interactions and reliable public information in the form of reputation, categorization processes may play a critical role in boundary spanners’ early trust assessments. According to social categorization theory (Tajfel and Turner, 1986; Turner, 1985), categories help make sense of the world by facilitating the grouping of social actors and the evaluation of them based on abstract prototypes in addition to recalled exemplars of actors viewed as typical of the category. Category classification enables one to “impute” missing information about an actor by applying knowledge and inferences about a category instead of having to rely on the direct perception of the specific actor’s attributes (Jonsson et al., 2009).
Research on swift trust suggests that, when relevant information is scarce, institutional categories strongly affect trust beliefs (Gambetta and Hamill, 2005; Meyerson et al., 1996). If the trustor considers the institutional category to which the trustee belongs to indicate trustworthiness, the trustor will perceive the trustee to be trustworthy as well. In cases where the trustee is an individual, relevant institutional categories include family background, national origin, and age (Zucker, 1986). If the trustee is an organization, analogous categories are industry, geographic location, and organizational age.
Partner organizations that are similar to the focal organization in terms of industry and geographic location are more likely to be assessed by the boundary spanner as being trustworthy. Similarity in these categories supports the assumption of a common background, thus conveying the perception of common values and shared background expectations. In terms of organizational age, the concept of liability of newness (Hannan and Freeman, 1984; Stinchcombe, 1965) suggests that younger organizations are known to have less well-established business practices, making them less predictable and thus decreasing trust in their competence. For these reasons, the boundary spanner’s initial trust belief may be influenced by a classification of the partner organization into the relevant institutional categories.
Proposition 1C. Institutional categories provide a basis for individual–organization trust during the initiation stage.
From individual to individual: becoming acquainted during interpersonal interactions
While previous ties, organizational reputations, and/or institutional categories can provide a basis for preliminary individual–organization trust,9 the nature of the trust relationship changes as the relationship process advances and negotiations unfold. The complexity of negotiating interorganizational relationships usually requires risk-taking and intense interactions between managers of both organizations (Ring and Van De Ven, 1994). Firsthand experiences resulting from the ongoing interactions between managers in the partnership begin to outweigh initial assessments of the organization as the basis for trust. Boundary spanners, thus, use their personal encounters to test the validity of their preliminary trust judgments toward the partner organization.
In addition to evaluating information actively provided to them, they also recognize patterns in the stream of experiences that they have with the managers of the partner organization. Based on these active and passive transfers of information, boundary spanners make informed judgments about the trustworthiness of these managers. As the managers from both organizations become acquainted, the boundary spanner’s initial, calculative-based trust assessment typically evolves into a more intense, experience-based, and personal trust relationship.
Proposition 2. Interpersonal interactions (between boundary spanners of both organizations) trigger the progression from initial individual–organization trust to individual–individual trust during the negotiation stage.
Empirical support for this evolution comes from marketing research. Investigating three industrial buyer–seller relationships in mature industrial markets, Narayandas and Rangan (2004) observe that an individual–organization trust platform is often a prerequisite for engagement in person-to-person negotiations. Whether this platform is strengthened, however, depends on how members of the dyad engage with and gain knowledge of one another. Boundary spanners in the three analyzed relationships replaced the individual–organization trust platform when engaging in personal interactions. The researchers observed intense negotiations between individual managers over an extended period of time and found that such negotiations often yielded strong psychological contracts regarding the need for cooperation between their respective organizations. According to Narayandas and Rangan (2004), these boundary spanners eventually strongly identified with their counterparts and even felt obliged to become their partner’s spokesperson inside their own firms.
From the individual to the organization: transferring trust in an individual to an organization
During the formation stage, the boundary spanner is responsible for overseeing the commitment of various types of resources to the interorganizational relationship. At this point, interdependencies with the partner organization begin to develop, prompting a reexamination of the trustworthiness of the partner organization. We suggest that this reexamination will be significantly influenced by the boundary spanner’s previous interpersonal interactions, and, at this point, the trust that was established in particular employees of the partner organization during the negotiation stage will be transferred to the partner organization as a whole.
The premise behind the concept of trust transfer is that trust is not necessarily based on direct experience with a particular trusted entity. It can also be extended from a third party related to that entity. Strub and Priest (1976) describe how trust is transferred from the better-known third party to a closely associated, but less well-known, entity. In this way, trust that is placed in the third party is used as a proxy (Krackhardt, 1992), and the third party assumes the role of an intermediary of trust (Coleman, 1990; Shapiro, 1987).
This logic of third-party trust transfer can be applied not only to trust relations that involve individuals but also to situations in which the trustee is a larger social entity or collective actor, such as an organization. Research on institutional trust suggests that a third-party individual can serve as an intermediary who brokers trust in the organization in which the individual is employed (Zucker, 1986). The sentiments and characteristics displayed by a manager, for example, are assumed to reflect the values and attitudes of the organization in general. With increasing interpersonal interactions, the behavior of a manager is attributed more generally to the organization. As a result, the manager assumes the role of the third party from which trust is transferred to the organization itself as the eventual trustee. In this way, individuals can generalize their personal trust in individuals to larger organizations consisting of those with whom they may have no direct experience (Lewicki and Benedict-Bunker, 1995; Maguire and Phillips, 2008).
Applying this line of thinking to interorganizational relationships, trust in the partner organization is inferred on the basis of the boundary spanner’s perceptions of the trustworthiness of the boundary spanner in the partner organization, especially when the focal organization’s boundary spanner has limited experience with other members of the partner organization. The counterpart’s trustworthiness, as observed in previous interpersonal interactions, acts as a referent or signal regarding trustworthiness of the partner organization. Having been assigned the specific role of boundary spanner representing the partner organization in the interorganizational relationship, the counterpart is perceived as acting in line with the entire partner organization. Accordingly, the focal organization’s boundary spanner’s trust in his or her individual counterpart is transferred, or generalized, to the partner organization as a whole. This process is likely to happen in many such settings, though clearly, it is more complicated and may take more time when the organizations involved are large and more complex.
This conception of trust transfer from individuals to larger social entities is notably consistent with Giddens’ (1990) idea of system trust. For Giddens, the system (in our case, the partner organization) is attributed trusted characteristics through “disembedding”—the “lifting out” of concrete social relations from local contexts of interaction and their integration into larger entities. Trust is disembedded from face-to-face interactions with particular managers to become a more general stance toward the partner organization as a whole.
Proposition 3. Trust transfer (from the partner boundary spanner to the partner organization) triggers the progression from individual–individual trust to individual–organization trust during the formation stage.
From organization to organization: institutionalizing trust
Drawing from institutional theory, we suggest that the institutionalization of trust at the organization–organization level occurs as an increasing number of employees of the focal organization become involved in the relationship during its operation stage. We differentiate between two subprocesses of institutionalization: objectification and habitualization specifically identified by Tolbert and Zucker (1996).
Objectification of a common understanding. Objectification refers to the development of general, shared social meanings by a set of social actors (Tolbert and Zucker, 1996). Research on shared social meanings in the context of organizations suggests that individuals working closely together strive to create a common understanding to ensure that the actions they select are consistent and coordinated with those of their colleagues (Mathieu et al., 2000; Tolbert, 1985). This common understanding allows social actors to predict and explain the behavior of those around them, to recognize and remember relationships among components of the environment, and to construct expectations for what is likely to occur next (Cannon and Edmondson, 2001; Rouse and Morris, 1986), all important elements during the operation stage of an interorganizational relationship.
Dialogue and workplace interaction drive the development of a common understanding. A “collective mind” (Weick and Roberts, 1993) develops through continuing conversations among organizational members. During various conversations, organizational members convey their expectations, values, beliefs, and assumptions to other members, resulting in a diffusion of interpretations throughout the organization. Through dialogue, employees create shared mental models of various aspects of their work environment (Tolbert and Zucker, 1996). Specific “stories” told by individuals evolve into a repository of information that forms part of the collective memory of the organization (Weick and Roberts, 1993). Besides communicating certain beliefs consciously, this diffusion of beliefs can also occur unconsciously over the course of workplace interactions. By observing how a certain individual behaves, fellow organizational members “make sense” of this behavior (Weick, 1995) and deduce the specific values and beliefs of that individual which, in turn, affect their own attitudes.
Building on this notion of objectification, we argue that boundary spanners build a common understanding with their fellow organizational members regarding the trustworthiness of the partner organization. Trust beliefs are thus spread beyond the boundary spanners who embodied them at more nascent stages of the relationship, as these boundary spanners disseminate their trust beliefs to fellow organizational members. Individual boundary spanners typically have multiple affiliations within their organizations (Currall and Inkpen, 2003). These connections and the social networks that they constitute expose them to a variety of interactions and serve as important channels for the dissemination of beliefs about the trustworthiness of the partner organization. Moreover, not all dissemination of trust-related information necessarily occurs actively (i.e. by communicating the information directly to fellow employees). It may also occur passively and even unknowingly, as trust spreads within the organization by inference from the boundary spanner’s actions in relation to the partner organization.
The theoretical notion that common understandings emerge within organizations and can be used to explain organizational trust development is supported by the work of organizational learning theorists. Notably, Crossan et al. (1999) and Nonaka (1994) explain how knowledge created by individuals is transformed into organizational knowledge through common understanding. They view knowledge creation as an upward spiral, a process that starts at the individual level and subsequently moves up to the organizational level.
Consistent with this view, we propose that the evolution of trust also involves a spiral process, as various members of the focal organization interact during the operation stage of the interorganizational relationship and information is aggregated and diffused through various mechanisms. Workplace interactions, including gossip (Burt and Knez, 1996), serve as the primary mechanism by which trust beliefs are disseminated within the organization. Once disseminated, trust in the other organization attains a “firmness,” or type of external validity, and thus can no longer be readily changed by any single individual. At this point, we may speak of organization–organization trust.
Proposition 4A. The development of a common understanding (among members of the focal organization) triggers the progression from individual–organization trust to organization– organization trust during the operation stage.
Habitualization of alliance routines. Habitualization describes the development of patterned problem-solving behaviors in response to particular stimuli (Tolbert and Zucker, 1996). As individual interpretive processes come together around a common understanding, corresponding collective behavior is implemented. That is, based on their common understanding, organizational members mutually adjust and take coordinated action (Simons, 1991). Over time, those actions deemed to be in line with the organization’s common understanding of a stimulus are repeated. Initially, this judgment about what actions should be repeated is made informally; soon, however, rules and procedures are established that govern individuals’ behavior (Crossan et al., 1999). Tasks are formally defined, actions specified, and organizational mechanisms implemented to make sure actions are coherent throughout the organization. The process of habitualization occurs as particular behaviors become embedded in the organization, particularly in its routines (Crossan et al., 1999; Tolbert, 1985). Habitualized patterns of actions cease to be tied to specific individuals (Zucker, 1987), becoming both objective (repeatable by other members of the organization without changing the meaning) and exterior (intersubjectively defined within the organization; Zucker, 1977). Through repeated application and formalization, these behavioral patterns become taken for granted, and macro categories emerge (Powell and Colyvas, 2008).
Reflecting this general habitualization process in the context of interorganizational relationships, we propose that employees of the focal organization strive for coordinated interaction with the partner organization by adjusting their activities so that they are consistent with the focal organization’s common understanding regarding the trustworthiness of the partner organization. Repeated action patterns of “how things are done with this partner organization” that account for the partner’s trustworthiness become increasingly taken for granted. Partner-specific alliance routines—rule-based behavioral patterns for organizational actions related to the interorganizational relationship—emerge (Ireland et al., 2002). These routines reflect the logic that governs alliance-related decision making throughout the focal organization and determine the organization’s approach to handling important activities relevant to the inter-organizational relationship, such as inter-organizational coordination, learning, or adaptation (Schilke and Goerzen, 2010). Ways of dealing with and perceiving the partner organization thus become independent of specific individuals. A particular view of the partner’s trustworthiness has formed that is now present throughout the focal organization and habitualized in the form of alliance routines. Since trust then resides within the roles and routines of the organization, it has become a “distinctive behavioral signature”—a coherent, relatively time- and situation-independent pattern of choices (King et al., 2010; Mischel and Morf, 2003).
Individuals belonging to the focal organization accept the joint commitment to uphold certain trust beliefs. Thus, even when their preference would suggest an alternative course of action, organizational members are committed to the routines reflecting the focal organization’s trust beliefs. Thus, even when their personal preference would suggest an alternative course of action, organizational members are committed to the routines reflecting the focal organization’s trust beliefs in the partner organization. As more people get involved in the inter-organizational relationship, the existing social formation is transmitted to them, and they become socialized into the prevalent behavioural expectations and norms. Similarly, when particular individuals leave the organization, the nature of the focal organization’s trust in the partner organization does not necessarily change. Trust perceptions have become embedded in the systems and routines of the organization, making them largely independent of individual actors. Thus, trust perceptions within an organization eventually converge into a set of institutionalized organizational rules of action that are relatively resistant to change initiated by specific individuals.

Proposition 4B. The habitualization of alliance routines (within the focal organization) triggers the progression from individual–organization trust to organization–organization trust during the operation stage.
From the individual to the organization: getting feedback from the organizational level
So far, we have described the model in a largely linear and sequential way. However, it is important to note that trust perceptions established at an organizational level will feed back to the individual level, thus affecting how boundary spanners act and think. As mentioned in the earlier section on prior interactions, when a new relationship is formed between organizations with prior inter-organizational experience, trust that has been established previously at an organization–organization level has a substantial effect on the boundary spanner’s initial trust beliefs pertaining to the partner organization.

Such a feedback process, however, may create certain tensions. For example, an institutionalized trust belief may prevent individuals from updating and testing the validity of their trust judgments toward the partner organization. Those individuals who do question organizational-level trust perceptions will have to go through what Schumpeter (1959) calls “creative destruction”— destroying or setting aside the institutional order to enact variations that allow for a re-evaluation of the trustworthiness of the partner organization. Trust attains a strong level of firmness once advanced to the organizational level, where routines can act as “core rigidities” (Leonard-Barton, 1992) that decrease organizational flexibility in responding to environmental changes. In short, organizational-level trust creates a particularly strong foundation for a new inter-organizational relationship, and individuals will find it more difficult to re-engage the processes of cross-level trust formation outlined previously in an effort to influence organizational-level trust.
Discussion
As common in process theorizing, the previous sections described the process model as it may apply in general. In the following discussion, we contextualize the model in two ways: (1) by varying the degree of trust, previously assumed to be high, and (2) by considering the role of several factors that may potentially impede the proposed trust-development process at various stages.
Degree of trust
A distinction can be made between the nature of trust development across levels of analysis and the degree of trust (low to high) that develops. Adopting a transformational approach to trust development (Lewicki et al., 2006), our model mainly focuses on how the nature of trust changes during the relationship development process. Thus, we did not seek primarily to answer the question: “how much does a party trust?” (i.e. the degree of trust), but rather the questions, “what unit of analysis trusts and what unit of analysis is trusted, and how does this change over time?” (i.e. the nature of trust). In our model, we assumed a fairly simple relationship characterized by a relatively high degree of trust. Let us now consider how the nature of trust might evolve when the degree of trust is low.
Starting with the individual–organization trust relation, the process model might begin with an initial negative reputation or other types of cues that tend to invoke low trust in the potential partner organization. This initial basis of low trust subsequently informs the boundary spanner’s interactions with individual employees of that organization. One might expect that the initial low degree of trust at the individual–organization level will be transferred to the individual–individual level, where the key boundary spanner develops a perception of low trustworthiness on the part of the other organization’s boundary spanner. This low trust perception will be incorporated into the transference process, causing a low degree of trust to be placed in the partner organization as a whole. Subsequently, a boundary spanner perceiving the partner organization as untrustworthy will share negative perceptions and experiences with colleagues. As this now-shared perception of low trustworthiness becomes engrained in appropriate alliance routines (e.g. constant monitoring, nondisclosure of confidential information etc.), organizational trust in the partner organization will be quite low.
While we submit that the linking mechanisms proposed in our framework thus do hold for different degrees of trust, there is reason to believe that the degree may affect both the speed and the intensity with which they operate. Particularly when it comes to sharing trust beliefs with colleagues to construct a shared understanding, the boundary spanner will be very motivated to disseminate those trust beliefs within the focal organization if trust is high. In this situation, identification-based trust permits the individual to act as the other party’s agent and to protect its reputation (Deutsch, 1949). Narayandas and Rangan (2004) even report that boundary spanners may feel obliged to become the partner’s spokesperson inside their own firm, as we have noted. Managers in high-trust relationships may gradually become embedded in both the personal and work-related relationships involved and, thus, also become firmly invested in the ongoing social and economic exchanges. Because they want to ensure that their individual integrity is not called into question, these managers have a strong incentive to make sure that the partner organization is trusted and treated with respect by communicating their own trustworthiness assessments to other members of their organization.10
Conversely, if trust is not high, such identification-based trust is unlikely to evolve (Lewicki and Bunker, 1996). Nonetheless, in cases of significant trust breaches, managers will most likely feel obliged to share their negative experiences with colleagues, warning them in a timely manner about the low trustworthiness of the partner organization. This expectation fits game-theoretical predictions made by Rabin (1993), suggesting that people wish to help those they trust and hurt those they do not trust.
In fact, the diffusion of information concerning low trust actors may be similarly fairly swift indicating a curvilinear relationship between the degree of trust and the diffusion of trust information, implying that diffusion is faster at both extremes (high and low trust) and less so in the middle of the trust spectrum. Overall, we thus expect the motivation to share trust beliefs to be comparatively weakest when the degree of trust is moderate. In this case, the boundary spanner feels neither bonded to the partner organization nor compelled to warn other members of the focal organization about the potential dangers of cooperating with the partner organization.
A related question regarding the interplay between the degree and the nature of trust is whether the degree of trust will be transferred to the next phase proposed in our model. Previously, we assumed implicitly that the degree of trust at one level of analysis would have a substantial impact on the degree of trust at subsequent levels of analysis. Recent work on trust spirals (Ferrin et al., 2008), which finds that the degree of trust is typically reinforced as a relationship unfolds. Previous degrees of trust influence the subsequent development of the degree of trust as actors mutually evaluate their counterparts’ trustworthiness, and trust perceptions become reciprocated over time. Therefore, the degree to which boundary spanners trust their counterparts during the initial stages leaves a strong imprint on the development of the degree of trust in later stages of the interorganizational relationship.
There are, however, a variety of reasons why the degree of trust may change over the course of the relationship, either from low to high or from high to low (cf. Dirks et al., 2009; Vlaar et al., 2007). We expect such a change in the degree of trust to be more likely in earlier than in later stages of the trust-development process in our model. In the early stages, trust beliefs are still preliminary and are limited to the domain of the individual boundary spanner. Especially when the boundary spanner is not yet heavily invested in the relationship with the managers of the partner organization, trust perceptions can easily be adjusted when relevant new information is obtained about the partner organization’s reputation or when interactions with the managers show that the initial high trust placed in the organization and its members was not justified.
With increasing duration of the interorganizational relationship, however, perceptions concerning trust obtain a certain degree of firmness. For example, if perceived trustworthiness decreases only after the boundary spanner has constructed a strong trust belief regarding his or her counterpart and has already transferred this perception to the partner organization as a whole through the mechanisms described in detail earlier, he or she might be more reluctant to change his or her attitude. Since the high trust placed in the other party has served to reduce the undesired uncertainty associated with the interorganizational relationship through increased commitment to that party (Cook et al., 2005; Yamagishi et al., 1998), the boundary spanner might be motivated to maintain the relationship and to overlook occasional untrustworthy behavior. Also, close personal ties with the counterpart developed over the course of repeated meetings might lead the boundary spanner to ignore new information about a breach of trust, an occurrence well documented in the popular literature on infidelity between trusted parties.
At an even more advanced stage of trust development, we assume the firmness of the degree of trust to be strongest. Existing cognitive schemas and scripts can lead a decision maker to resist new evidence (DiMaggio, 1997; DiMaggio and Powell, 1991). Even if a boundary spanner is willing to adjust his or her individual trust perceptions when encountering untrustworthy behavior of the partner, trust can no longer be readily changed by him or her since the level of the trustor has shifted to the organization. We acknowledge that our discussion of the interactions between the nature and degree of trust is only preliminary, and more fine-grained analyses (particularly in terms of the specific circumstances surrounding a breach of trust) are required. Detailed examination of these issues is beyond the scope of this article but should clearly be the focus of future theoretical and empirical development.
Factors impeding the trust-development process
Evolution of many organizational processes is contingent on the particular context in which these processes are embedded. Thus, we turn our attention to the question of which factors may cause the trust-development process depicted in Figure 1 to be terminated before reaching its final stage. This is an important issue since more advanced forms of trust are not relevant in all relationships (Lewicki and Bunker, 1996). In the context of our model, we posit that termination of the trust development may occur for at least six reasons.
The first, and most obvious, reason is that the interorganizational relationship does not materialize or is ended prematurely under a variety of circumstances (one of which would be a breach of trust). In this case, there is simply not enough time for the entire trust-development process to take its course (or in the case of a breach it ends abruptly). The second reason for termination is that the scope of the interorganizational relationship is not comprehensive enough to make complex forms of trust relevant. This is particularly likely when interorganizational relationships involve a low degree of task integration and very few employees in the relevant organizations are involved in or even aware of the partnership. Under such circumstances, it is unlikely that trust will develop at the organizational level. Because task integration is typically lower in marketing alliances than in R&D alliances (Casciaro, 2003), for example, the probability of organization–organization trust emerging may be lower in marketing than in R&D alliances.
Third, when a large number of contractual safeguards (such as lawsuit provisions or arbitration clauses) are put in place during the negotiation stage to forestall self-interested behavior and reduce the risk associated with the relationship, especially when a great deal is at stake, there is little room for further trust development during the subsequent formation and operation stages (Cook et al., 2005; Costa and Bijlsma-Frankema, 2007).
The fourth reason is that the perceived proto-typicality of the partner organization’s boundary spanner (i.e. the extent to which this individual is perceived as representative of the collective identity of the partner organization) is low. Some individuals are perceived to resemble the prototype of a collective and represent its standards, values, and norms more than others (Hogg et al., 2004). If the perceived proto-typicality of the partner organization’s boundary spanner is low, then the trust transfer from this individual to the partner organization as a whole is less likely.
The fifth reason is that the social influence of the boundary spanner within the focal organization might be too low. Social influence is a function of visibility and salience (Friedkin, 1993); thus, in order for trust beliefs to disseminate through common understanding at the individual level to the organizational level, the opinions of the boundary spanner must be visible to the organization and the organization must value those opinions. Both visibility and salience are affected by three features of the interpersonal communication network in which the boundary spanner is embedded: structural cohesion, similarity, and centrality (Friedkin, 1993; Marsden and Friedkin, 1993). If these factors are low, the boundary spanner’s social influence within the organization will be low and the diffusion of trust beliefs to the organizational level will be hindered.
The sixth reason is that the key boundary spanner responsible for the interorganizational relationship is displaced by a new employee of the focal organization. Even in cases in which organizational trust in the partner organization has already developed, the new boundary spanner may not yet be familiar with the common understandings and alliance routines pertaining to the partner organization. Newcomers to an institution need to undergo socialization, which then leads to the internalization of organizational perceptions and procedures. However, the new boundary spanner interacts with the counterpart in the partner organization on a daily basis, and these interactions may determine the preliminary trust beliefs of the boundary spanner regarding the partner organization. A newcomer may therefore continue to go through the initial stages of trust development until he or she becomes more embedded in and aware of the routines of the focal organization.

Contributions, implications, and conclusion
The current investigation contributes to the literature in several ways. First, research on trust at different levels has so far largely existed in “intellectual silo,” with studies focusing on a single level of analysis—primarily either the individual or the organization (cf. Fulmer and Gelfand, 2012; Schoorman et al., 2007). This focus on within-level research, while significant for many purposes, has left some important questions regarding between-level dynamics unanswered. As in many areas of organizational studies, in order to more fully grasp how trust operates, we must understand how different levels of analysis interact (Klein and Kozlowski, 2000). Our first and most fundamental contribution is thus to facilitate integration in the field of trust research by offering a cross-level perspective of trust development, showing that the micro and macro levels are not isolated but highly intertwined. Our work thus redirects current thinking by suggesting that trust at any one level does not occur in a vacuum but needs to be considered in the context of trust at other levels. This premise provides a conceptual foundation for a new line of theoretical and empirical investigation into how trust at one level affects or is affected by trust at other levels.
Second, while trust processes at different levels of analysis are clearly related, the distinct causal mechanisms identified in our proposed model strongly suggest that trust at different levels is not isomorphic; that is, the underlying properties of the construct and its relations to other important constructs are not the same across levels. Our theoretical model thus implies that, in order to avoid misspecification, more research is needed to investigate the relevant antecedents and consequences of trust at multiple levels simultaneously. Such research could build on similar efforts in the field of efficacy studies. For example, Feltz and Lirgg (1998) found that group-level (i.e. collective efficacy) perceptions were related to team success, while individual-level (i.e. self-efficacy) perceptions were not. In this vein, future research on trust in interorganizational relationships needs to determine whether all, some, or none of the three trust relations analyzed here (individual–individual, individual–organization, and organization–organization) affect relationship performance and which moderating factors may influence their relative effects.
A related research effort could focus on relationships between trust-development processes and the choice of governance mechanisms in interorganizational relationships. Gulati and Nickerson (2008) report that preexisting interorganizational trust increased the probability that a less formal mode of governance would be chosen over a more formal one. Based on the more nuanced perspective developed in our model, future research could investigate whether choices of governance mode vary in the three types of trust relations we analyze. Furthermore, adopting a dynamic approach, researchers may study at what stage in the relationship the willingness to rely on informal governance mechanisms emerges and which processes support such willingness. Adopting a longitudinal approach to studying how the nature of trust (across the proposed levels) and the nature of the interorganizational relationship (formal or informal) coevolve would be particularly fruitful in this respect.
Third, most extant research tends to examine trust in interorganizational relationships as a static phenomenon (cf. Inkpen and Currall, 2004; Koza and Lewin, 1998). The model we propose highlights how the nature of trust, including the relevant trust drivers and consequences, changes quite substantially over the course of an interorganizational relationship. Our work suggests that examinations of trust need to theorize, or at least control for, the relationship stage. Each relationship stage is associated with specific actors and activities, which in turn affect the nature and correlates of trust. Failing to consider trust as evolving and changing over time, as common in some investigations, thus means losing explanatory power.
Fourth, our research also contributes to the ongoing debate concerning whether organizations are “able” to trust (e.g. Blois, 1999; Möllering et al., 2004; Scheer, 2012). Since the study of trust has its roots in psychology and social psychology, trust has traditionally been thought of primarily as a psychological disposition that only individuals have. Unfortunately, little theorizing has been devoted to extending the scope of trust from individuals to larger social entities and collective actors. Organizational-level trust has mostly been treated as taken for granted in macro-level trust research while simultaneously being dismissed by many micro-level researchers. Our conceptual model aims to address this issue directly. Building on a definition of trust that is applicable to individuals as well as organizations and other collectivities, we specify the theoretical mechanisms by which trust develops into a collective, organization-level property. Specifically, we argued that trust can become independent of the individual actors involved through the processes of objectification and habitualization embedded in the organizational context. A by-product of this work is that we provide a much-needed ontological basis for further research on organizational-level trust and interorganizational trust relations, more specifically.
Finally, prior research on the drivers of interorganizational trust has predominantly examined relational, organizational, and environmental attributes (Zaheer and Harris, 2006). While such factors are clearly important, our approach directs attention to a domain of interorganizational trust antecedents—intraorganizational processes—that has been largely missing. Emphasizing the importance of various processes internal to the trusting organizations, including clue gathering, trust transfer, and institutionalization, our proposed model identifies novel factors worthy of further exploration in future empirical research.
In conclusion, the notion that trust in interorganizational relationships exists at various levels that are systematically linked over time holds great promise for developing a more complete, integrative perspective on the development of trust in interorganizational relationships. Our process model proposes a sequence that relates trust across different levels of analysis as the interorganizational relationship matures, identifies theoretical mechanisms underlying those shifts in level, discusses relevant contingencies that may affect the proposed processes, and suggests theoretical and empirical directions for future research on multi-level trust.

Acknowledgements
The authors are thankful for the insightful comments provided by the editor, Ann Langley, and by three anonymous reviewers. The authors also thank Mark de Rond, Gianvito Lanzolla, Roger C. Mayer, Cecilia L. Ridgeway, Andrew H. Van de Ven, and Lynne G. Zucker for helpful comments on earlier versions of this article.
Funding
The first author gratefully acknowledges grants from the German Research Foundation (DFG) and the Alexander von Humboldt-Foundation (DFG: SCHI 869/5-1; Alexander von Humboldt: 3.1-DEU/1123954).
Notes
1. In this article, we follow King et al. (2010), Scott (1998: 6), Stinchcombe (1965: 142), and others who view organizations as social actors that possess their own agentic capabilities above and beyond the individuals who constitute them. Based on this view, we will develop a more detailed argument for trust at the organizational level in a subsequent section of this article.
2. To limit the complexity of the analysis, we restrict our argumentation to two levels of analysis (individual and organization). While a third level, that of groups within organizations (such as departments), could also be considered (Currall and Inkpen, 2002, 2006; Kozlowski and Klein, 2000), most of the extant literature on trust in interorganizational relationships deals either with individuals or organizations as the trusting and trusted parties.
3. We have simplified the analysis in that we only focus on bilateral arrangements. In reality, however, multilateral interorganizational relationships between more than two partners are becoming increasingly ubiquitous (Das and Teng, 2002a), and future research should aim to improve our understanding of how trust operates within these more complex constellations.
4. Although single stages might overlap in reality, we agree with Ring and Van De Ven (1994: 97) that it is useful to separate these stages for analytical purposes.
5. This is not to say, however, that we consider variance theories inadequate per se for studying trust dynamics. Our choice of a process approach rather reflects this article’s objective of penetrating the logic behind temporal progressions in trust relations by identifying how different states and causal mechanisms come into play at distinct stages of an interorganizational relationship. We believe a process approach is most useful for this purpose. Still, a variance theory’s focus on specifying quantifiable relations between independent and dependent variables makes it an attractive complementary strategy, as noted in the section on future research.
6. Based on this definition, the term isomorphism has a different meaning in the literature on meso-level theorizing than, for example, it has in institutional theory, where it refers to organizations in the same field becoming more similar (DiMaggio and Powell, 1983).
7. While we propose that the structure of trust varies among different levels of analysis and that trust is therefore not isomorphic, the function of trust is suggested to be similar (Ferrin et al., 2007)—that is, the functional definition of trust as “the willingness of a party to be vulnerable to the actions of another party” holds across different levels (see the section on Definition of Trust).
8. The life cycle approach reflects our focus on trust as a situational (rather than a personality-related) variable (see Furr, 2008, regarding the personality–situation debate). A key assumption here is that trust is not entirely stable but can change with the social situation (Boon and Holmes, 1991).
9. In the language of social perception theory, previous ties, organizational reputation, and institutional categories provide the basis for inferences in terms of tentative hypotheses that are adjusted in light of new information (Hamilton and Sherman, 1996).
10. The same is likely to also apply in reverse to the boundary spanner from the partner organization, thus aligning the trustworthiness of the boundary spanner from the partner organization with that of the partner organization as a whole.

References
Blois, K. J. (1999) “Trust in Business to Business Relationships: An Evaluation of Its Status,” Journal of Management Studies 36(2): 197–215.
Boon, S. D. and Holmes, J. G. (1991) “The Dynamics of Interpersonal Trust: Resolving Uncertainty in the Face of Risk,” in R. A. Hinde and J. Groebel (eds) Cooperation and Prosocial Behaviour, pp. 190–211. Cambridge: Cambridge University Press.