3.1 Rational control

The two theoretical frameworks building on principles of “rational” control are agency theories, focusing on “bureaucratic”, top down organizational designs to elicit cooperation, and social exchange theories, which inform research on collegial, bottom-up and emergent forms of control.

Agency (or Principle Agent) Theory is among the most influential frameworks used to model cooperation in organizations, and to design rational bureaucratic forms of control or governance. Being part of a broader class of Theories of the Firm (covering, among others, also Transaction Cost and Property Rights approaches) in neo-classical economics, its micro-foundation is the rational actor (Alchian & Demsetz, 1972). Key assumptions are that the interests of the owner of a firm (the principal) and of the managers who are hired to run it (the agents) diverge, since it is rational for both to maximize their own payoffs rather than the principal’s. What complicates matters further is the structural information asymmetry between both; since agents are closer to the work process and often also have a lot of operational discretion, they have an information advantage vis-a-vis the principal. This advantage can be exploited to their own benefit. For example, agents may misrepresent the causes of bad performance, attributing it to external circumstances beyond their influence, rather than to their own incompetence or negligence. In order to overcome the problems of goal incongruence and information asymmetry, principals can design governance structures that align their interests with the interests of their agents. The best way to do so, according to agency theory, is to provide incentives to agents, for example by linking their pay to their performance. Performance goals, in turn, require standards of evaluation and monitoring routines. Sub-standard performance or other deviations from the contractual obligation will be followed by negative sanctions (e.g., no or lower bonus), whereas meeting or exceeding the targets will be rewarded (e.g., with a promotion).

With regard to rational collegial control, social exchange plays an important role in organizational control processes, mainly through its impact on informal peer pressure. Social Exchange Theories (Cook et al., 2013; Cropanzano & Mitchell, 2005) come in many flavors. Most variants share a focus on the individual within the group, analyzing how the (inter-)dependencies resulting from exchange opportunities affect and are affected by someone’s position in the broader structure. Advantageous positions in the structure yield power and information advantages, benefiting individuals with more social capital, but hampering those who are less well endowed. Many social exchange theories build on rational choice reasoning, assuming that individuals want to improve their resource base, and also care for their status and prestige. Reputation management — individuals striving to acquire or maintain a good reputation or avoid getting a bad one — is not only assumed to be a key driver of such emergent, bottom-up dynamics, but also to be the main guarantor for getting and keeping cooperation going. The degree to which reputation management can be effective crucially depends on the (informal) web of relationships in an organization and its subunits, because gossip — the most important carrier of reputational information — travels through personal connections and contacts.

3.2 Normative control

The two institutional explanations rooted in the assumption that humans are inherently cooperative emphasize the importance of institutional designs that capitalize on the power of social norms. Stewardship theory focuses on designed, top-down bureaucratic arrangements, whereas bottom-up forms of collegial control take center stage in social identity theory.

Stewardship Theory (Davis et al., 1997; Van Slyke, 2006) is a prominent example for a model of normative bureaucratic control. It was developed as a reaction against the unrealistic propositions of economic Theories of the Firm, in particular the dominant agency model. Rather than taking conflicting interests between principals and agents as a point of departure, stewardship theory posits that managers will, in principle, be inclined to act as responsible stewards. Hence, it is possible to assume some degree of goal convergence between these and other stakeholders, because there are some shared collective interests resulting from the shared need to making the organization successful. This leaves room for arranging the relationship on the basis of trust, rather than on mistrust. The main challenge for an organization therefore is not the need to “align” the interests of management and workers, but to build and maintain a relationship of mutual trust. Self-governance is the key to achieve this, with autonomy and empowerment being its most effective tools.

Social Identity Theories (Ellemers et al., 2002; Spears et al., 1999) focus on the “group within the individual”, and provide an important framework informing analyses of normative collegial control. They assume that social categorization plays an important role in everyday human cognition, that individuals care (more) for the members of their (professional) in-group than for out-group members, and that the degree to which such differences are important to them depends on their level of identification with a group. Given the value individuals attach to their social identities, individuals are likely to actively “manage” the identities of the groups they belong to, as well as attach meaning to what it means being part of an out-group. The related processes of categorization, identification and evaluation can have a strong impact on cooperation. They can lead to preferential treatment of in-group members, and foster the willingness to socialize others into the culture of the in-group. But category membership also can come with heightened feelings of obligation towards and sacrifices for one’s in-group and the development or reproduction of negative out-group stereotypes.

3.3 The blind spot: vicious cycles of cooperation decay

All four approaches, or their combinations, are frequently used to explain cooperation in and between organizations, and to guide the design of organizational control structures. In most cases, there is no special emphasis on problems of cooperation sustainability. The implicit assumption seems to be that the conditions that get cooperation going are the same that also will keep it going. But quite some evidence shows that this assumption may be wrong; decline of cooperation has often been described and may be more common than one would expect based on these theoretical accounts. More generally, none of these approaches offers the analytical tools that would be necessary to explain why cooperation decays, how this decay evolves, and under which conditions it becomes likely. The next section will sketch, for each of the four design principles, how it may trigger a self-reinforcing process of cooperation decay, and how relational support structures may contribute to prevent or mitigate these governance traps.

The organizational literature has produced a series of in-depth accounts describing processes of cooperation decay. Though some of these descriptions received much attention when they were published, these studies did not spark the development of a strongly visible research field or many follow-up studies. The present section argues that the institutional arrangements and governance structures following from each of the four templates for organizational control bear the seeds for an endogenous process of decay or vicious cycle (Masuch, 1985). The remainder of this section will reconstruct these four vicious cycles, which we refer to as the incentive, empowerment, reputation and identity “traps” (see Table 1). This analysis will then be used to identify the missing links that might be necessary for a theory of sustainable cooperation in and between organizations.

4.1 Behavioral foundations: Goal framing

With regard to the behavioral micro-foundation, the stylized assumptions about human nature do more harm than good. We need to adopt a more flexible framework that is able to capture the fact that humans can have both selfish and altruistic motives, and tell us not only under which conditions one might supersede the other, but also that these motives can decay. Goal framing theory provides such a framework. It rests on four core assumptions (Lindenberg & Steg, 2007).

First, it posits that human decision making and behavior is goal directed, and that it is only one goal that can be salient at a given moment. This is also the goal that will provide the dominant frame for action.

Second, there are three overarching goal frames. In the hedonic goal frame, the salient goal is, “to feel better right now”; in the gain goal frame, it is “to guard and improve one’s resources”, and in the normative goal frame, the salient goal is “to act appropriately”.

Third, there is an a-priori hierarchy in the relative strength or salience of these three overarching goal frames, with the strongest being the hedonic, followed by gain, and the weakest being the normative one. This hierarchy has evolutionary reasons. Hedonic goals are related to the satisfaction of immediate needs, and are therefore the most basic ones. As such, hedonic goals need least “external” support in order to remain salient. In contrast, both gain and normative goals need extra support to remain salient. This can be institutions (like religious beliefs and the related moralization of behavior), but also social cues communicated through social disapproval for violating a norm.

Finally, whereas one of the three frames is salient in the cognitive foreground, the other two overarching goals will remain active in the cognitive background. Their changing strength can affect the salience and stability of the goal frame in the foreground. Where background and foreground goals are aligned, the background goal reinforces or strengthens the foreground goal, thereby contributing to its salience and robustness. For example, you may refrain from sharing information about inappropriate behavior of a team member with one of your closest colleagues, because doing so would both violate a remedial norm in your team (“we always resolve issues openly”), and put your own position at risk (e.g. sharing this information may ultimately increase your own workload). Conversely, in situations where background goals are at odds with the foreground goal, their increasing salience weakens the foreground goal, which may eventually lead to a frame switch in which the most salient background goal replaces the foreground goal. For example, the normative goal of complying to the rule of not talking behind a colleague’s back may come under pressure to the degree that not sharing the information in question is likely to damage you. The larger the personal price to pay, the more likely it will be that the goal of preventing damage for yourself (a gain goal) may override the normative goal prescribing backbiting against a team member.

In sum, which goal frame is salient in a given situation is highly context dependent, and previous research has pointed to a large variety of context conditions and their potential impact on goal frames. We focus on relational support structures, i.e. configurations of social or functional relations.

4.2 Institutional foundations: Relational support structures

With regard to the group or organization level, we need a more refined understanding of those forms of governance that may contribute to sustaining cooperation and those that may undermine it. Goal framing theory has successfully been applied to explain cooperation in a wide variety of contexts and social dilemma situations, but it is only recently that its implications for institutional design and organizational governance have been explicitly theorized (Lindenberg & Foss, 2011; Silitonga et al., 2019). In their paper ‘Managing Joint Production Motivation,’ Lindenberg & Foss (2011) argue that the key to (sustainable) cooperation in and between organizations is a salient normative goal frame supporting joint production motivation, i.e., “a special kind of motivation that is particularly geared to the fact that organizational members need to engage in collaborative activities such that organizations that tap into it would gain a performance advantage” (Lindenberg & Foss, 2011), with joint production being defined as “productive activities involving heterogeneous but complementary resources and a high degree of task interdependence” (Lindenberg & Foss, 2011). In their theoretical paper, the authors discuss a variety of formal governance practices (i.e., knowledge-based authority design, recognition based reward structures, cognitive and symbolic management, and integrated task and team design) that may be conducive to triggering and sustaining a salient normative goal frame supporting joint production motivation. Whereas their paper focusses on what one could call deliberately designed top-down bureaucratic support structures, it pays only scant attention to the role of emergent, bottom-up relational support structures (Lazega, 2001). This is the focus of the present study.

The important role that the “embeddedness” of economic transactions in personal ties and informal social networks plays for cooperation is well documented (Smith-Doerr & Powell, 2005; Schweitzer et al., 2009). A considerable variety of network structures at the individual (e.g., social capital through structural holes) and group level (e.g., close-knit structures) have been linked to a wide range of phenomena (e.g., performance, inequality), invoking dozens of different theoretical mechanisms (Contractor et al., 2006). The picture concerning systematic associations between specific network forms as enablers or barriers of (sustainable) cooperation emerging from this research is far from clear-cut. The fact that the same types of social network structures were found to support cooperation in some settings or conditions but to undermine it in others (Wittek et al., 2003), points to the strong context dependence of network effects. The social rationality framework adopted in this essay elaborates on this contingency argument. It suggests that the social networks foster sustainable cooperation if they create a relational support structure for eliciting and backing the joint production motivation of those involved.

5.1 Gossip failure

Reputations need to be shared, which implies that positive and negative evaluations about third parties need to be communicated. Consequently, there are two basic requirements for reputation mechanisms to sustain cooperation. First, there need to be interaction opportunities enabling communication — which is where personal relations and the structure of informal networks come in. Second, opportunity structures alone won’t work without agency; those involved need to be willing to share evaluative information about third parties. This is commonly referred to as gossip (Wittek & Wielers, 1998). It is therefore no surprise that gossip has been hailed as one of the most important ingredients for sustaining cooperation without the presence or intervention of formal authorities, both within or outside the group. Since gossip is considered to be almost costless for those engaging in it (Coleman, 1990), and in addition attributed with a plethora of other positive side-effects, it should be ubiquitous and effective. This enthusiasm for the almost magical powers of gossip, which is reflected in a literal multidisciplinary explosion of research on gossip during the past decade (for an overview, see Giardini & Wittek, 2019), seems to have obfuscated that there may be many situations in which individuals are likely to refrain from spreading gossip even if they possess relevant and important evaluative information about somebody. With the main pillar of reputation management not working, it is likely that cooperation problems will increase because information about opportunistic behavior will not spread. Goal-Framing Theory has been used to specify the conditions under which individuals are likely to withhold rather than share sensitive new information (Giardini & Wittek, 2019). Most gossip scholars have stressed the inherent pleasure that individuals experience when engaging in gossip episodes. In fact, evolutionary scholars consider the related release of endorphins (Dunbar, 2004) as a key trigger behind the widespread incidence of gossiping. According to GFT, this hedonic mindset may indeed provide a strong and constant inclination to share third party information. But GFT also points to the conditions under which this inclination may be tempered or replaced by gain or normative background goals. This is particularly likely when functional or affective (inter-)dependence between one or more dyads in the gossip triad pushes other concerns into the foreground. For example, competition resulting from strong negative functional interdependence between the potential gossip sender and receiver may create a situation in which sensitive third party information can yield one-sided benefits for the sender, whereas sharing this information may neutralize this advantage. In this case, a gain goal frame that is not aligned with the hedonic goal frame may become salient, with the result of suppressing the hedonic motives that otherwise would trigger a gossip episode. But also normative concerns may have similar effects. For example, solidarity norms may keep individuals from sharing negative information about their friends, i.e., others with whom one is linked through a strong cognitive affective tie. For a more detailed elaboration of GFT predictions concerning conditions inhibiting gossiping, see Giardini & Wittek (2019).

In line with this theoretical account, some related empirical evidence indeed suggests that gossip failures may be more common than one would assume based on standard accounts. According to the sociologist Ronald Burt (2001), rather than being a tool for reputation management, much gossiping is actually caught in the cage of conversational conventions. These conventions or etiquette push gossip senders and receivers into reinforcing their pre-existing beliefs about a third party, rather than fundamentally altering the evaluation of the third party’s behavior itself. This so-called echo effect is particularly likely in close-knit social network structures and illustrates how normative concerns — in this case not violating conversational rules — may temper one’s inclination to share new sensitive information. It has far reaching implications for the functioning of reputation management, because it suggests that the participants in a gossip conversation may both hold different opinions or experiences concerning a third party, but these differences will remain unspoken; etiquette will push them to share only information which one believes is already known to the other, and consistent with the other’s evaluation of the third party. Burt (2001) finds evidence for this echo effect in three organizational populations (senior managers in a leading manufacturer of electronic components, staff officers in two financial companies, and investment bankers in a large financial company). The implications of these findings are not trivial, because they suggest that closed social network structures do not improve the flow of evaluative information (see also the findings reported in Wittek & Wielers, 1998), as the standard reputation theory would assume, but rather reinforce what one already knows, i.e., they produce “echo, not accuracy,” thereby intensifying the opinion (e.g., about somebody’s trustworthiness) that one held before.

Another source of gossip failure results from the sharing of negative third-party information being driven not so much by reporting “objective” violations of group norms, but by the gossiper’s intention to strengthen the bond with the receiver at the expense of the third party. This may be realized partly through strategic misrepresentation of the third party’s actual behavior or characteristics, for example by magnifying his or her misbehavior, or by downplaying or not mentioning positive information that does not fit into this picture. From a goal framing perspective, this kind of gossiping is fueled by the gossip sender’s gain and hedonic goal frame reinforcing each other; sharing selective third party information serves the objective to increase or consolidate the social capital of the gossiper, rather than being driven by the intention to transmit accurate information concerning a third party’s reputation. At the group level, this gossip effect may eventually result in a segregated structure, because it facilitates and reinforces the emergence of coalition structures (Wittek & Wielers, 1998; Shaw et al., 2011) in which the bond between gossip sender and gossip receiver becomes stronger, and their ties to the third parties become weaker. Such coalition structures, in which the other party is considered an ally, often come with normative expectations in which “siding” with the despised third party would be considered as a betrayal of the solidarity obligations that define the dyadic alliance behind the coalition.

The evidence concerning the echo and the coalition effects shows that normative concerns may remain so strong that they can either actually discourage the sharing of reputation damaging information (Giardini & Wittek, 2019), or they can result in the strategic distortion of third party information for the purpose of bonding, rather than facilitating the transmission of objective information about the third parties’ “type”. In both cases, a salient normative goal frame oriented to preserve in-group solidarity undermines the fundamental diagnostic value that gossip is supposed to have according to the standard reputation models.

5.2 Signaling failure

If individuals care for their reputation and know that others might share evaluative information about them, then they will anticipate on how their own actions and those of others will affect their reputation. They will therefore have an interest in signaling traits and intentions that may help building a positive reputation, and to suppress or hide those aspects that may damage it. And they will have an interest in getting an accurate idea of relevant traits and intentions of their (potential) exchange partners.

Hence, individuals are likely to actively engage in attempts to manage their reputations. One potential consequence of these attempts may be the emergence of dysfunctional status hierarchies. An ethnographic study of the group processes as they unfolded over the period of three years within the management team of a German paper factory illustrates the kind of vicious cycle that may emerge when status considerations become the dominant motive in a setting where professional reputations matter (Wittek, 1999; Wittek et al., 2003). See Figure 1 for a summary.

The management of the paper factory, while relying on a clear hierarchical structure of authority, strongly encourages and lives what they refer to as a “culture of trust”. The latter is based on joint problem-solving and mutual adjustment. An important process in the team involved informal control. This took two forms. One consisted of collegial advice related to identifying, preventing and mitigating technical problems at the machines, related to human error or unforeseen contingencies. The other form of control consisted of nudges related to problems of a more behavioral nature, including bad time management, not sticking to promises (or being late with doing so), but also free-riding.

Reputation indeed was an important driver in this process, both with regard to collective reputation (the firm in the paper industry) and with regard to individual prestige as a professional, in this case a good “paper maker” or engineer. The reputation motive worked well during the first year of my observation, 1995. During this phase, the team showed an exceptional performance that was a textbook case of sustainable cooperation. Much of their energy and joint action went into the planning and realization of a large infrastructural project, consisting of the construction of a second production hall, and the dismantling, transfer, rebuilding, and activation of a third paper machine in this new production hall. All of this happened while regular production on the two available paper machines had to continue. During this phase, advice was shared freely, both solicited and unsolicited. Good advice was greeted with respect and admiration, contributing to the professional standing of the respective colleague — but importantly, it was seen as a welcome and necessary contribution to the collective good. This changed during the second phase, in 1996. With the collective good being less salient, striving for and consolidating one’s prestige acquired some toxic side effects. This reputation management involved not wanting to share advice anymore, because one feared that the other party would “steal” the idea — in the sense of not recognizing the source. And it involved not accepting or actively discouraging colleagues from giving advice, out of fear that this would improve the advice giver’s professional prestige at the expense of one’s own. The resulting vicious cycle reinforced a behavioral pattern in which team members refrained from providing unsolicited advice in situations where this would have been helpful to solve technical problems more quickly — out of fear of being rejected, but also out of fear that the colleague would hide the true source of the solution to fellow team mates. They refrained from accepting advice (or actively rejected it) from a colleague, out of fear of losing status. What had happened? How was it possible that a highly cooperative team, most of whose members successfully worked together for more than two decades, suddenly imploded, losing its self-correcting abilities? According to the ethnographic account, several seemingly small changes ultimately resulted in the erosion of joint production motivation and its underlying normative base. With the big project being successfully completed, an important and salient joint endeavor and the related need for frequent coordination and communication had disappeared. However, this in itself was not sufficient to undermine the cooperative orientation of the team members, who, like all employees of this small factory, strongly identified with the production site, which was located in a small village of 600 inhabitants. But with the completion of the project, technical problems with the third machine started and would trouble the team and the production process for months after its activation. Given the complexity of the technical problems, all senior paper engineers were given the assignment to work out a solution, with responsibilities for who was in the lead not being explicated or specified in more detail. This created uncertainty about the allocation of rewards and professional prestige. This coincided with a major drop in trust towards the site manager and the mother company, caused by the sudden and unexpected announcement, shortly after the project had been completed, of restructuring plans that put at risk also the positions of members of the management team. The resulting uncertainty about how one’s own actions (like solving a technical problem or helping a colleague to solve it) would affect the reputation of oneself and of one’s colleagues fundamentally changed the signaling value of knowledge sharing. During the first phase, joint production motivation was strongly supported by salient solidarity norms, according to which proactive knowledge sharing was key to solve the manifold problems that the team and the factory were facing. Giving and receiving professional advice represented positive relational signals among team members, i.e., they were cues reinforcing the mutual perception that everyone was still committed to the project and to helping each other. The exchange of advice was perceived as an unambiguous and necessary contribution to the collective good. With the growing concern for one’s own professional status and employment opportunities, the related uncertainties in the organizational context and increased general distrust in the intentions of management and one’s peers, simple acts of collegial advice giving, for example with regard to suggesting a solution to a technical problem with a paper machine, were now viewed with suspicion. The suggestion could be useful, but the recipients started to question the underlying motive behind the advice — was it to enhance the advisor’s status at the expense of the receiver’s professional status? In line with this reasoning, would asking for advice be perceived as a sign of incompetence, and eventually be broadcasted like this within the team? Finally, would giving a good advice be acknowledged as thus, or would the recipient “forget” to mention who had originally suggested this solution? Behavior that once was a positive relational signal suddenly had acquired an ambivalent character. This example shows how emergent group dynamics related to the attempt to manage professional reputations can actually undermine and weaken the normative base underlying the once strong joint production motive of the team members.

9.1 Escaping reputation traps: Relational signaling

Reputation traps can be avoided if the setting minimizes the likelihood of ambiguous relational signals (Lindenberg, 2000; Wittek et al., 2003). Relational signaling is key to sustainable cooperation. Unlike broadcast signals, these more subtle, repeated cues indicate that the sender is still in a normative goal frame, with solidarity norms being salient (Bliege Bird et al., 2018).

The importance of reciprocal knowledge sharing as a precondition for sustainable cooperation also became evident in a longitudinal 4-wave sociometric study of advice relations among 59 employees in a Dutch housing corporation (Agneessens & Wittek, 2012). This study compares two models of knowledge sharing, both based on the overarching principle of social exchange. The status model (Blau, 1963) argues that advice seeking has to be repaid by deference to the advice giver. Since individuals try to use social exchanges to increase or at least not diminish their status, an advice giver will refrain asking advice from his or her advisee. The result is that advice relations are likely to be asymmetrical, which produces hierarchical informal structures of knowledge sharing at the group level. In contrast, according to the social capital explanation, advice giving is subject to reciprocity norms. As a consequence, advice relationships are likely to become symmetrical, i.e., individuals will tend to ask advice from those who have asked them for advice before. At the group level, this mechanism is more likely to result in close-knit, non-hierarchical informal structures. The empirical study revealed an interesting pattern, since it provides an important refinement of the original status hypothesis as formulated by Blau (1963). Normative considerations, motivated by reciprocity obligations rather than status enhancement, were found to operate at the dyadic level. Status dynamics were important beyond the dyadic level, but these were restricted to one specific case: very active advice givers are unlikely to ask advice from very active advice seekers. Overall, the patterns of advice relations at the level of the organization suggest that status considerations do not inhibit the process of knowledge sharing. Put differently: a normative goal frame (in this case reciprocity concerns), trumps gain seeking (in this case the drive to increase one’s own professional prestige at the expense of one’s colleagues).

9.2 Escaping incentive traps: Advisory trust

A reanalysis of one of the first longitudinal sociometric intra-organizational studies (Wittek, 2001) illustrates how an emergent relational support structure mitigates the potentially disrupting forces of a highly competitive incentive system. This ethnographic study was conducted in 1954 among the 25 salesmen in a Canadian furniture store (French, 1963). The setting is a textbook example for incentive based governance structures. Salesmen had a baseline salary, which could be complemented by a 3% commission for every item sold, and a bonus based on performing better than one’s colleagues. The payment system resulted in a considerable divergence of wages, with individual salaries ranging from $6,000 to $10,000 per year. A major source of securing it was regular clients, with whom salesmen had built a personal relationship. The problem was that these regular clients could also show up when the salesman who knew them was not present in the showroom, but in the back office, where he was waiting on his turn based on an elaborate rotating scheme. In such cases, it was up to the salesman on duty to call his colleague in case a regular customer would ask for a specific salesman by name. Situations like these were frequent, and introduced a major source of uncertainty into the system, since compliance to this rule was difficult to control. In fact, questions concerning the violation of the group norm not to steal regular clients were a constantly salient and major concern for the salesmen. The setting confronted the salesmen with a repeated trust problem, because those who complied and sent through a regular customer could not be sure whether this colleague would do the same. Solving this trust problem and mitigating the underlying uncertainty is challenging. A “standard” solution proposed by economic and organizational sociologists to manage critical interdependencies is “social embedding” (Granovetter, 1985; Lazega, 2001), i.e., building and maintaining personal relations to those on whom one depends (friends do not cheat on each other). But building friendship ties simultaneously with 24 colleagues of course won’t work. The ethnographic evidence shows that one of the possible straightforward alternatives, based on shared ethno-religious identities also did not work out. Half of the salesmen were Jews. But whereas during the first observation period there was a significant tendency to nominate someone from the same ethno-religious group as a friend, this effect became non-significant in later phases. Hence, sharing the same identity category does not protect from being cheated upon. My reanalysis of the dynamics of the friendship network suggests that in this particular setting, a hitherto understudied relational support structure emerged, which may have contributed to a stable solution to the trust problem. At the behavioral level, this solution is based on advisory trust, i.e., individuals having the tendency to maintain interpersonal trust relations with alters who proved to be trustworthy and who dispose of as well as are willing to disclose reliable information about the trustworthiness of other actors. At the group level, this principle results in a structure with four positions, based on the similarity of the sociometric pattern of in- and outgoing friendship choices. In this structure, salesmen occupy either the position of a truster, a trustee, isolate, or advisor. The latter play a fundamental role to keep this system going: they both receive and send friendship nominations. The analysis shows that the most robust inter-position choices in this structure indeed consist in the trustors choosing the advisors, and the advisors choosing the trustees. This system also converged towards a clear separation between a component of salesmen connected by friendship ties, and an internally and externally almost completely disconnected group of isolates.

Since these insights were based on a reanalysis of an existing dataset, and more detailed context information on the link between friendship ties and behavior (i.e., sending or not sending through a regular customer) is not available, the true effectiveness of this relational support structure and its underlying behavioral principle cannot be judged from this data, and would have to be assessed in future research. Nevertheless, this particular case illustrates which kind of relational support structure structure may emerge in an institutional environment that strongly relies on monetary incentives, and it suggests that these endogenous dynamics might at least contribute to mitigate the potential incentives trap that such governance structures may cause.

9.3 Escaping empowerment traps: Autocalibrating relational support structures

Empowerment Traps are not inevitable. The iconic case study uncovering the conditions under which they can be avoided is Union Democracy: The Internal Politics of the International Typographical Union (Lipset et al., 1956). The authors discuss several reasons why the iron law of oligarchy did not hold at ITU. One was its factionalized structure, which created a system of mutual vigilance in which norm violations and wrongdoings like corrupt behavior by members of one faction would be readily exposed by the other faction. These factions were rooted in the importance that the local unions who had founded ITU traditionally attached to their local autonomy.

An informative example comes from an empowerment intervention study in two different organizations (Silver et al., 2006). Whereas the intervention — consisting of a bundle of measures ranging from enhancing information sharing, the development of clear boundaries, and the introduction of self-managed teamwork — could be sustained in one of the organizations, it faded in the other. In the former setting, the key mechanism sustaining empowerment values and practices can be described as an autocalibrating relational support structure. Its core consists of a feedback loop between on the one hand, a leader who constantly reinforces the existing joint production culture present in the senior team and the rest of the organization and, on the other hand, team members having and using the opportunities to participate in decision making for constructive voice. This positive feedback loop is further strengthened by the continuous use of external training and feedback from outside coaches. In contrast, in the other setting under investigation, these ingredients were lacking.

9.4 Escaping identity traps: Dyadic toolkits and structural folds

Relational support structures also have the potential to prevent or mitigate identity traps. Several mechanisms and have been described. First, DiBenigno & Kellogg’s (2014) comparative study of two department in a hospital discusses the importance of what they call dyadic toolkits that are enabled by cross-cutting demographics and shared non-occupational identities. Dyadic toolkits allow the use of alternative status rules, emotional scripts, meanings and expertise on which the members of different occupational groups can draw to negotiate tasks, understand each other’s actions, and share knowledge.

Second, in their seminal study of team performance in the gaming industry, De Vaan et al. (2015) describe the importance of structural folds within multidisciplinary project based teams of experts. An important task of these teams is to produce new games that are not only innovative and technically cutting edge, but that also appeal to a critical audience of reviewers and users. The researchers show that such game-changing video games are more likely to be invented by teams containing at least one member who has previously collaborated with several other previously unrelated members in the team. Being familiar with the qualities of several team members, these structural folds can then vouch for the trustworthiness and expertise of the other team members, assuring them about each other’s qualities. This allows them to bridge the often quite large cognitive distance that exists between the different professions in such teams, helping them to create a common language and to recombine insights from a large variety of backgrounds.

Finally, the deliberate creation of workflow interdependencies, for example through the creation of transitory project teams, is another instrument to overcome identity traps as they result from the homophilous tendency to form ties with others having a similar socio-demographic background (Yakubovich & Burg, 2018). This study resonates with earlier findings showing that in organizational settings, functional interdependencies trump identities as the driving force in group dynamics (Wittek, 1999, 2001). The argument is that workflow interdependence creates a shared interest in regulating each other’s work-related behaviors, which increases the chances for communication and the emergence of normative expectations and their monitoring (i.e., embedding the work relation in a social relation). The resulting shared experiences further lay the foundation for value congruence. Formal interdependence also creates an interest in shooting down disagreements and conflicts, thereby increasing the chances for intensification and persistence of the personal relationship.