Abstract
Organizations that have committed to decentralizing authority can face situations in which centralized decision making is useful or necessary. However, centralization risks undermining an organization’s perceived commitment to decentralization, which can threaten employees’ motivation and commitment, as well as organizational legitimacy. Leveraging a comparative case study of four organizations that had explicitly committed to decentralize authority and that subsequently centralized authority during the COVID-19 crisis, we identify a process by which organizations can deviate from decentralization commitments while keeping those commitments in place. This process, which we call “democratic deviation,” involves leaders and workers collectively authorizing centralization and then leaders enacting that authority with openness and transparency. In the three of our four cases that deviated democratically, decentralization commitments were upheld. Conversely, when centralization occurred implicitly through leader fiat and was then enacted opaquely by leaders, a process we label “monocratic deviation,” decentralization commitments were undermined. Through this study, we develop a theory of the structural antecedents and processes of durable commitments to decentralization, prompting re-evaluation of decentralization’s assumed fragility. More broadly, our theory suggests how organizations can uphold shared commitments even when they must deviate from them in the face of situational pressures.
Keywords
In response to mounting concerns that centralized hierarchies are ill-equipped to meet the complex social and economic demands that contemporary firms face (Adler, 2001; Battilana et al., 2018; Fjelstad et al., 2012), scholars have turned their attention to decentralized forms of organizing (Hamel & Zanini, 2020; Lee, 2024; Rajan & Wulf, 2006; Reineke et al., 2024; Reitzig, 2022; Turco, 2016). Decentralized alternatives range from empowering employees through greater delegation (Argyris, 1998) and removing hierarchical layers (Reitzig, 2022), to adopting self-management systems (Puranam, 2014; Lee & Edmondson, 2017) and expanding workers’ ownership (Shantz et al., 2019). Across this variety, approaches share a common feature: a broader distribution of authority than in centralized hierarchies.
Organizations are drawn to adopt decentralization by various aspirational aims: to foster innovation (Adler, 2001; Felin et al., 2009), align decision making with expertise (Barley, 1996), enhance employee engagement (Teece, 2003), or reduce workplace inequality (Adler, 2019; Young-Hyman et al., 2023). Yet, even in organizations that have embraced decentralization, situational pressures may prompt the centralization of authority (Abi-Esber et al., 2025). Leaders may centralize authority to assert control (Adler, 2001), establish strategic alignment (Child, 2019), expedite urgent decisions in times of crisis (Hart et al., 1993), or initiate changes in organizational routines (Knott, 2001; Valentine, 2017).
Shifts toward centralization, however, carry risks. They can undermine the credibility of a leader’s and, by extension, an organization’s commitment to decentralization. When organizations decentralize authority, they make a commitment to workers that authority will be allocated in a distributed, less-hierarchical fashion (Gibbons & Henderson, 2012; Stea et al., 2015). When leaders renege on these commitments, the negative consequences can be immediate and enduring. Workers may respond with disengagement, withholding effort due to the loss of authority (Aghion & Tirole, 1997; Baker et al., 1999; Stea et al., 2015). More broadly, backtracking on decentralization commitments can erode the legitimacy of and trust in leadership and weaken workers’ commitment to the organization (Rousseau & McLean Parks, 1993; Tyler, 2006).
Given the repercussions of reneging on decentralization commitments, research has concentrated on how leaders can resist pressures to centralize in order to preserve the credibility of their commitments (Foss, 2003; Stea et al., 2015). This focus rests on the assumption that maintaining credible decentralization commitments requires avoiding centralization altogether. But this assumption imposes a rigid constraint and binary choice on leaders: Reclaim authority and risk employee disillusionment, or preserve decentralization and risk losing control and alignment. Faced with this dilemma, leaders often revert to conventional hierarchical structures to accommodate their centralization needs (e.g., Burton et al., 2017; Foss, 2003; Turco, 2016). The challenge of centralizing decision making temporarily without disavowing decentralization permanently is a key reason decentralized structures are seen as more fragile than hierarchies (Adler, 2001; Foss & Klein, 2022; Freeland & Zuckerman Sivan, 2018; Gruenfeld & Tiedens, 2010; Jaumier, 2017; Michels, 1915). Conversely, if organizations could occasionally centralize decision making without undermining broader decentralization commitments, they would gain valuable flexibility, allowing them to respond to changing situations without reneging on core commitments. In this sense, the capacity to accommodate centralization may be essential to making decentralization more durable.
In this article, we examine how organizations can navigate pulls toward centralization in ways that uphold, rather than undermine, commitments to decentralization. To answer this question, we use a qualitative, multiple case study design (Yin, 2003), examining four extreme cases of decentralized organizations as they navigated the COVID-19 pandemic. Each faced a sharp negative shock to their business results and heightened uncertainty about their future. Consistent with existing theory most prominently elaborated in the threat rigidity thesis (Staw et al., 1981), each of the organizations responded by centralizing decision-making authority. However, despite undergoing similar degrees and types of centralization, the organizations diverged in how workers perceived and responded to these shifts: In three cases, workers believed and acted as if commitments to decentralization were upheld, while in one case, they believed and acted as if commitments were undermined. Analysis of qualitative data across our cases revealed that this divergence hinged on how centralization unfolded as a process and the presence or absence of supportive structural features that shaped this process. Drawing on these findings, we theorize a model that elaborates how firms can flexibly respond to situational demands to centralize while remaining steadfast in their commitments to decentralized authority.
Decentralization Commitments Versus the Need to Centralize Decision Making
Decentralization can be understood as the extent to which authority is distributed across interdependent actors relative to the total concentration of authority in a single actor. While any organization in which one person does not hold all authority could be described as decentralized, decentralization typically refers to organizations in which authority is distributed more widely than in centralized hierarchies. The extent of decentralization can vary considerably, ranging from incremental models that grant workers modest and occasional decision-making autonomy to radical arrangements that guarantee workers substantial discretion over their own work, and even influence over the organization’s governance (Lee & Edmondson, 2017; Puranam, 2014; Young-Hyman et al., 2023). In addition, while some organizations adopt decentralization to pursue functional goals such as creating knowledge (Adler, 2001; Fjelstad et al., 2012), increasing responsiveness (Aghion et al., 2021), and motivating employees (Freeland & Zuckerman Sivan, 2018; Teece, 2003), others pursue decentralization on normative or ideological grounds, seeing it as a way to uphold the dignity of workers (Hodson, 1996; Rothschild-Whitt, 1979), promote human-centric workplaces (Puranam, 2024), and restrain the inequities inherent within hierarchies (Ferreras, 2017; Ferreras et al., 2020).
Despite the varied motivations for decentralizing authority, centralization also offers a range of benefits. It facilitates coordination and control by ensuring accountability, resolving conflicts, reducing opportunism, and establishing common knowledge (Aghion & Tirole, 1997; Bunderson et al., 2016; Child, 2019; Freeland & Zuckerman Sivan, 2018). It also supports strategic decision making and the initiation of organizational change (Knott, 2001; Mintzberg, 1979; Valentine, 2017). Compared to decentralized decision making, centralized decision making more effectively filters out poor-quality opportunities (Csaszar, 2013) and can provide psychological comfort for employees in uncertain times (Friesen et al., 2014; Gruenfeld & Tiedens, 2010).
Given centralization’s many functions, even organizations deeply committed to decentralization may periodically face situational pressures to centralize decision making, whether to establish control (Foss, 2003), initiate strategic change (Bernstein et al., 2016), provide clarity and direction to workers (Turco, 2016), or respond to urgent threats. Defined as “an environmental event that has impending negative or harmful consequences for an entity,” threats have been shown to prompt organizational leaders to respond by narrowing the scope of information processing and by centralizing control (Staw et al., 1981, p. 502). These threats can range from simple information overload and slow-burning financial decline to sudden crises prompted by unexpected external events. Although this threat rigidity response is often characterized as maladaptive, limiting flexibility and reducing the uptake of novel ideas, it underscores the gravitational pull of centralization in moments of threat. Thus, whether centralization is a strategic adaptation to changes in organizational demands or a defensive reaction to environmental shifts, organizations committed to decentralized authority commonly face situations in which they may feel compelled to centralize decision making.
Centralization Threatens the Durability of Decentralization Commitments
The benefits of decentralization, from increased employee motivation and creativity to enhanced organizational agility and responsiveness, depend on workers believing that leaders and, by extension, the organization have made credible commitments to decentralize authority (Baker et al., 1999; Foss & Klein, 2022). Organizations signal their commitments to decentralization through both formal and informal means. Some codify them in formal documents, such as constitutions or bylaws (Lee & Edmondson, 2017; Young-Hyman et al., 2023) or through public declarations (Argyris, 1998). Others institutionalize authority-sharing through informal work practices and communications (Turco, 2016; Vallas, 2006). Over time, workers form expectations of wider authority distributions not only through formal policies but through their lived experiences. These expectations can be further reinforced by how organizations present themselves externally. When recruitment materials or external communications emphasize decentralized work practices, these practices become part of an organization’s identity and shape who chooses to join (Campbell, 2012; Hurst et al., 2024; Marchetti & Puranam, 2025; Puranam & Håkonsson, 2015). Signaling an organization’s commitment to decentralization through a combination of formal and informal practices, as well as through internal and external communications, forms a relational contract between leadership and workers that, in exchange for greater effort and commitment, workers will be granted greater authority and discretion (Anderson & Schalk, 1998; Gibbons & Henderson, 2012; Stea et al., 2015).
Moves to centralize place that contract at risk. Even if claimed to be temporary, such moves can cause workers to question whether decentralization is a durable commitment or a discretionary posture that is easily revoked (Stea et al., 2015). The counter-normative nature of decentralization (DiMaggio & Powell, 1983) amplifies this sensitivity; because hierarchical control remains the institutionalized default, workers are primed to interpret moves to centralize as reversions rather than exceptions. The consequences of perceived backtracking are manifold: It threatens to break the contract between leaders and workers, causing workers to withhold their effort and become disillusioned (Gibbons & Henderson, 2012; Stea et al., 2015). At the extreme, centralization can weaken the legitimacy of leadership, as workers come to see leaders as acting outside the shared norms, values, and practices that have guided the organization in the past (Zelditch, 2001). When wider and more egalitarian authority distributions become part of employees’ normative expectations, even isolated acts of centralization can erode trust, reduce commitment, and increase turnover (Anderson & Schalk, 1998; Robinson & Rousseau, 1994; Tyler, 2006).
Given the costs of backtracking from decentralization commitments, research has focused on how organizations can avoid centralization. Scholars have argued that organizations can raise the reputational costs of centralization, for example, by formalizing promises to decentralize or by increasing managers’ informational distance so that they become less aware of potential reasons to intervene in front-line dynamics (Stea et al., 2015). Organizations can also reduce the likelihood of managerial intervention by structuring the organization to minimize coordination demands across units and by establishing robust measurement systems that allow leaders to monitor performance from a distance (Foss, 2003; Stea et al., 2015). These prescriptions rest on the assumption that any move to centralize will be perceived as a breach of decentralization commitments and, therefore, is to be avoided at all costs (Aghion et al., 2021; Baker et al., 1999).
This assumption points to a unique, asymmetrical challenge of decentralized organizing relative to hierarchical organizing: While leaders of centralized hierarchies can flexibly decentralize without great concern about breaking foundational commitments, leaders of decentralized firms are less able to centralize with the same ease. This asymmetry constrains decentralized organizations, limiting their flexibility to respond to changing demands. The difficulty of centralizing without unraveling broader commitments to decentralization leads to situations in which organizations renounce decentralization commitments altogether and permanently revert to traditional hierarchical models (Burton et al., 2017; Foss, 2003; Turco, 2016). This dynamic helps to explain why decentralized structures are often seen as less durable than their centralized counterparts (Adler, 2001; Foss & Klein, 2022; Gruenfeld & Tiedens, 2010). Summarizing this view, Robert Freeland and Ezra Zuckerman Sivan (2018, p. 156) wrote,
The key practical problem is that insofar as the firm is inherently hierarchical, the project of suppressing fiat is necessarily quite precarious—what is in the background can spring quickly and painfully into the foreground. Accordingly, although it is true that some firms have clearly succeeded at this “high road” project, this is a very difficult road indeed, and some of these difficulties can be traced to the difficulty of suppressing hierarchy in a consistent and credible manner.
Recent work has begun to peer beyond this dominant assumption, considering whether the manner in which centralization unfolds could mitigate the potential loss of employee motivation (Asmussen et al., 2019). Yet, we have little understanding of what such a centralization process could look like, the factors that make it possible, and the mechanisms linking the process to the durability of decentralization commitments. We advance this line of scholarship by examining how organizations can navigate moments of centralization without undermining, and possibly even while reinforcing, their commitment to decentralization.
Methods
To explore this question, we followed multiple organizations with radically decentralized structures as they responded to the onset of the COVID-19 pandemic. Our research design and sample selection strategy reflected the following principles. First, we chose organizations that had radically decentralized authority, because their commitments to decentralization and the threat of centralization were more likely to be visible and salient to their members (Eisenhardt, 1989). In addition, if in these extreme cases, centralization could occur while the organization upheld its commitment to decentralize, we felt the findings would also be applicable to less-extreme cases. Second, given that prior literature predicts that organizations will centralize in moments of crisis, we viewed the onset of the pandemic, a global public health crisis that triggered a global recession, millions of layoffs, supply chain disruptions, and the nearly complete shutdown of in-person work, as an opportunity to observe organizations in a situation that typically compels organizations to centralize decision making. Third, we selected a multiple case study design in order to increase the chances of observing variance in centralization processes and to enhance the generalizability of the findings (Eisenhardt, 2021).
To identify and recruit our cases, we created a broad list of radically decentralized organizations from publicly available sources like the U.S. Federation of Worker Cooperatives and the Holacracy One Network. To limit extraneous variation, we sought firms of between 15 and 300 employees, which are large enough to have multi-layered hierarchies but small enough to effectively capture organization-level dynamics. We also limited our sample to firms in knowledge-intensive industries, in which decentralization is considered particularly valuable (Adler, 2019; Young-Hyman et al., 2023), and firms with established commitments to decentralization, operationalized as those that had adopted decentralized structures at least two years prior to the beginning of the pandemic. Following these criteria, we identified 25 organizations and were able to hold initial interviews with leaders at 14 of them. From these 14 potential cases, we pursued data collection at six, where leaders offered rich accounts of their efforts to navigate the crisis. Two of the six organizations became unresponsive during the data-collection process, leaving four cases for which we collected extensive data. 1
Table 1 presents the four cases that we selected. All four organizations are headquartered in the United States. Each organization was substantively impacted by the COVID-19 pandemic and centralized decision making in response to the crisis. The table also describes the nature of decentralization at each firm, which we elaborate in our findings.
Summary of Cases and Crisis Effects
Data Collection
Our primary source of data was employee interviews. Drawing on employee directories provided by the companies, we sought interviews with individuals across levels of formal authority and occupational areas in the organizations. In addition, during interviews, we asked for recommendations of other individuals who would have thoughtful perspectives on the organization’s response to the crisis.
Interviews were semi-structured and examined employees’ histories at the company, their experiences during COVID, their impressions of the organization’s crisis response, and their perceptions of the company’s alignment with its principles and values. Each interview was conducted by video conference or phone, recorded, and transcribed. At multiple points, we paused data collection to review what we had learned and to identify gaps in our understanding of each company’s crisis response. Armed with this information, we conducted additional interviews to fill in gaps and ensure that we developed a clear, consistent view of the company’s response. We also conducted additional interviews to ensure that we captured a broad range of employee perspectives, including potentially divergent views. Finally, we conducted a set of interviews at each organization after the pandemic had ended, to gather longitudinal perspective on the cases. In total, we conducted 64 interviews across our cases, 54 between June 2020 and January 2022 during the height of the pandemic, and 10 between August and October 2023, after the pandemic had ended. Table 2 displays how the interviews were distributed across case, time, and type of informant.
Summary of Interview Data Across Cases
Given Evolve’s smaller size, only its founder was classified here as a leader.
Finally, throughout the data-collection process, we collected various archival documents: company bylaws, governance documents, slide decks, internal meeting notes, and other documents that workers referenced during our interviews. We used archival data to triangulate and contextualize insights we gained from the interviews.
Data Analysis
Our data analysis process followed a grounded, inductive approach. For each case, we created an extended case summary by organizing data into a narrative structure and including all relevant quotes. The resulting case summaries were each 45 to 55 single-spaced pages and served as an initial way to understand the organizations’ efforts to navigate the crisis. Each co-author then independently conducted a round of focused coding of the case summaries. Our initial codes were action-oriented and descriptive, grounded closely in the data (Charmaz, 2006). We then merged our separate codes into a single document and discussed them to develop shared understanding. This step in the analytic process sensitized us to the key dynamics of each case: the areas of centralization, the overall tenor of workers’ perceptions of the centralization, and the factors that shaped workers’ perceptions. In this phase, we discovered notable variation in how workers perceived centralization across the cases, with workers perceiving that commitments were undermined at Plant and preserved at Energy, Enterprise, and Evolve.
In the next phase, we focused our analysis on the factors that influenced workers’ perceptions of centralization, seeking to understand why centralization was viewed as consistent with decentralization commitments in some of our cases but not others. To do so, we wrote new analytical memos for each case that were organized around distinct centralization events and workers’ perceptions of these events. We coded these data, identifying factors such as leaders’ behaviors or contextual features that influenced workers’ perceptions of centralization events. When evidence of a particular factor came only from a single source, we conducted additional interviews to assess whether other individuals would confirm the importance of these factors. Ultimately, within each memo, we constructed a table that laid out each centralization event, factors that shaped perceptions of that event, and whether it supported or undermined perceptions of the organization’s commitment to decentralization.
In the next substantive stage of analysis, we iteratively compared factors within and across cases (Charmaz, 2006; Glaser & Strauss, 1967), which allowed clear categories to emerge. One category of factors related to the existence (or non-existence) of explicit rules for centralizing authority. For instance, positive perceptions of centralization at Enterprise, Energy, and Evolve involved instances in which centralization followed a set of formal rules, while negative perceptions of centralization at Plant coincided with instances in which no clear set of rules for centralization existed. A second category of factors related to leaders’ behaviors that signaled commitment to decentralized authority. In particular, these behaviors included open and transparent communication.
One limitation of this emergent set of findings, however, was its static nature. We had limited understanding of how and when the rules and leaders’ behaviors were leveraged during centralization events to shape workers’ perceptions of centralization. 2 Therefore, we returned to the data and recoded them with a focus on their temporality. In doing so, we realized that the two factors we had previously identified were relevant during distinct phases of centralization. Formal rules, when present, were salient prior to leaders’ enactments of centralization, when leaders were seeking to define and gain consent for centralization, which we came to call the authorization phase. By contrast, the leaders’ behaviors we identified as important took place during the execution of centralized authority, which we came to call the enactment phase. We analyzed the differences in each phase between our cases, contrasting, for example, the explicit, collective, rules-based authorization process at Energy, Enterprise, and Evolve and the implicit, fiat-based authorization process at Plant. We also analyzed the distinction between the open and transparent enactments of centralized authority at Energy, Enterprise, and Evolve and the opaque enactments at Plant. This processual perspective also allowed us to see how the nature of the authorization process impacted the way that centralization was subsequently enacted by leaders, linking what were, until then, separate components of our findings. Finally, this processual perspective focused our attention on workers’ behaviors after centralization, which provided further evidence of whether decentralization commitments were still in place. Collectively, this step in our analysis resulted in the development of a dynamic process model of centralization that upholds rather than undermines commitments to decentralization. In the following sections, we present the cases and our findings in detail.
Prior Decentralization and Crisis-Induced Centralization
Prior to the pandemic, all of our case organizations were organized in an explicitly decentralized fashion, with some variation in structure. First, workers held greater decision-making authority than in conventional hierarchies. Enterprise and Evolve both operated a system of radical decentralization called Holacracy, which formally granted employees the authority to self-manage in their roles without needing to seek approval from a manager. At both of these organizations, quasi-managerial roles could set priorities and allocate resources but could not direct subordinates’ work or hire, fire, or promote them. Design rights were also decentralized, enabling workers to propose changes to their own and others’ roles, with group consent required for adoption.
At Plant, the principle of decision making by experts led workers to have significant autonomy. One of Plant’s founders described their approach:
We have a lot more autonomy as a base foundational thing than a typical hierarchy or bureaucratic company. The principle that I think I like best is, experts with the knowledge make decisions. I think that it sounds like, “Oh yeah whatever, everybody should do that,” but the reality is most companies don’t do anything like that. It usually ends up being a job title or something else that winds up having the authority to make decisions. We try to get experts to do that.
Plant maintained an accountability hierarchy whereby some individuals oversaw the execution of specific workstreams, but these individuals did not serve people management functions. Instead, these functions were overseen by the company’s “agent” and “professional development partner” roles. Individuals in these roles helped employees navigate the organization, resolve conflicts, and pursue professional development but had no formal authority. Hiring and promotion decisions were made by volunteer committees rather than centralized leadership.
At Energy, a worker cooperative, workers’ expansive authorities were formalized at the governance level. A worker-dominated board, in which four of seven members were elected by other worker-owners, held authority over governance matters. At the operational level, Energy had a conventional hierarchical structure composed of a CEO, a leadership team (called the company circle), and middle management layers, with decision making formally allocated through this hierarchy. However, in practice, managers ensured that decision making was highly participatory. The CEO described their approach:
In a transparent culture, we try to, what we call, “work out loud.” And working out loud, having an open conversation with the CEO and the leader of a team can be a pretty juicy conversation. And you end up with, six people participating, and maybe ten people can hear it, and it propagates, kind of organically when you’re able to do that in person.
Another cultural principle at Energy that supported decentralization was referred to as “bringing people along.” A manager described this principle as “hav[ing] a group conversation . . . allow[ing] people to have a chance to get familiar with it, get comfortable with it, challenge it, and allow it as an opportunity to potentially evolve as a result.”
In addition to workers having significant decision-making authority, a strong commitment to transparency existed across the four cases. Financial and strategic documents were openly shared with employees, ensuring that workers had access to relevant information and that leadership decision making was visible. At Energy and Plant, even board and leadership team meetings were fully transparent and could be attended by anyone in the company.
Employees were attracted to these organizations because of the explicitness and centrality of each organization’s commitment to decentralization. The Enterprise founder described how being decentralized impacted recruiting:
We found it is a clear both attractor and detractor for the type of people we want in the organization . . . There are certain people who are looking for a career ladder and career path that is predefined or preordained, and for them, it’s about a ladder and a title. Well, because of what we do, when we explain it, it’s a huge detractor, which is great, because we weren’t great about ladders and titles before Holacracy. That’s just not how we’re wired.
Thus, across our four cases, organizations had made strong and explicit commitments to decentralization prior to the crisis. Workers were drawn to join these organizations because of these commitments, highlighting the central place the commitments occupied in the relationship between workers and the organization and the risks that centralization posed to this relationship. Table 3 provides additional evidence of the nature of decentralization at each case.
Additional Evidence of Pre-Crisis Decentralization
Despite each organization’s explicit commitment to decentralization prior to the crisis, decision making became more centralized once the pandemic began. This shift occurred in two domains: worker protocols and resource allocation. In the domain of worker protocols, the crisis prompted centralized decisions related to worker safety, remote work, and returning to the office. In total, three out of the four cases instituted new worker protocols in a centralized manner. At Energy, the leadership team unilaterally decided to require employees involved in solar installations to drive separate cars and remain outside of customers’ homes during installations. At Enterprise, leadership instituted a policy that employees were not permitted to travel to client sites and, later, directed employees to take all video calls with their cameras on. At Plant, leadership set remote work policies, established protocols for essential workers like lab researchers coming to the office, and later in the pandemic, made the decision to bring employees back to the office. At Evolve, centralization of worker protocol decisions did not occur, partly because their work was primarily remote even before the crisis. 3
Resource-allocation decisions constituted the more contentious domain of centralization. Across our cases, prompted by the financial strain caused by the crisis, companies made significant reductions in spending and compensation, and three of the four implemented layoffs. In every case, these decisions were made centrally by founders or leadership teams. At Evolve, for example, the founder unilaterally decided to reduce worker compensation to preserve cash. In addition, unlike the company’s normal participatory process for firing workers, layoff decisions were concentrated in the founder’s hands. At Enterprise, the CEO, CFO, and COO jointly made decisions to ensure the company’s viability during the crisis. They suspended loan payments related to a 2015 employee stock ownership plan (ESOP) conversion, deferred all commissions and bonuses, asked for volunteers to work reduced schedules or take furloughs, and instituted layoffs. While firing decisions were normally made in a decentralized fashion, during the crisis top executives assumed full control of layoff decisions. At Plant, the crisis prompted the leadership team to impose new spending restrictions that, under normal circumstances, would have been made locally by relevant work groups. The crisis also prompted an expansion and formal activation of the company leadership team’s decision-making domain, which prior to the pandemic had remained informal and more constrained due to the company’s commitment to decentralization. At Energy, in contrast to the normal participatory mode of decision making, the CEO gained unilateral authority for cost-cutting and staffing-level decisions, setting staff reduction targets, which the leadership team implemented. Leadership team meetings, which were normally open to all employees, were restricted so that discussions about layoffs could take place in private. Finally, the leadership team unilaterally decided to seek private capital funding, a decision that under normal circumstances would have required broader approval. Each instance of centralization represents a centralization event. Table 4 presents the centralization events that occurred across our four cases.
Centralization Events by Company and Type
While centralization occurred in similar domains and degrees across the four cases, there was variation in how workers perceived it and, subsequently, how they behaved following centralization. Workers at Enterprise, Evolve, and Energy continued to believe and act as if the organization was committed to decentralization, while at Plant, workers believed and acted as if the company had deviated in fundamental ways from its commitments. To explain this variation, we next describe the different processes of centralization that prompted divergent perceptions and behavioral responses. We organize our findings by phase of centralization.
Authorizing Centralization
The initial phase of the centralization process that shaped variation in outcomes concerned how centralization was authorized. Because centralization represented a shift in authority that potentially threatened each organization’s shared commitment to decentralization, the way it was authorized proved crucial. We observed stark differences in how centralization was authorized across our cases.
Explicitly and Collectively Authorizing Centralization
In the three cases in which the shared commitment to decentralization was upheld, organizations maintained explicit rules governing the distribution of authority. These rules operated at two levels: operational rules, which specified who had authority in day-to-day decisions, and collective choice rules, which governed the process for changing the operational rules. 4 The rules were codified in formal documents and could be changed only through participatory processes that involved workers’ input and consent.
At Energy, a document called the “decision zone chart” specified decision rights held by different roles across a spectrum of domains, specifying, for instance, that “Team Leads” held authority to “Determine Vendors of Specific Product and Service Offerings,” and “Individuals” had authority to “Determine Daily Work Hours.” The document also delineated processes for changing these rules, which involved different degrees of workers’ consent and approval, from consultation to direct votes. Similarly, at Evolve and Enterprise, a document called the “Holacracy constitution” granted each individual full authority to take actions in their role without needing approval. Furthermore, a set of detailed roles published on a digital platform called Glassfrog specified the authority associated with each role and the individuals holding them. Finally, the constitution established a participatory process by which the roles and their accompanying authorities could be changed. Any worker could propose a change, but changes required approval from all role-holders in the relevant group.
When rules governing the distribution of authority were present, leaders and workers relied on them to authorize centralization, either by activating dormant authority specified in operational rules or by using collective choice rules to modify authority.
Leveraging operational rules to collectively authorize centralization
At our case organizations in which decentralization commitments were upheld, leaders needing to centralize decision making first consulted existing operational rules to justify the shift. In some instances, the rules outlined areas of authority that had been formally allocated to leaders but remained dormant in stable times, either unused or informally delegated to workers. To centralize decision making, leaders activated these dormant authorities and referenced the operational rules to justify their actions.
At Energy, for example, the leadership team quickly realized the need to secure a new loan to maintain cash flow when the pandemic hit. The decision zone chart granted them the formal right to solicit board approval to take on new debt without consulting the broader membership. Typically, decision making at Energy was far more participative, with leaders inviting broad input. However, the leadership team bypassed the normal process and unilaterally sought board approval for the loan. Some workers were bothered by the lack of consultation, but leaders justified their actions by referencing the decision zone chart. One leader explained,
[We were] really worried about cash. . . . And we have what’s called a decision zone chart. . . . and in the decision zone chart, the decision to raise capital lies with the board of directors. So we went to the board and said, “Hey, we’d like to see if we can raise some money in the face of the pandemic because of these debt capital needs.” And there were a few owners who were like, “You should’ve asked the ownership before we went further in debt.” . . . We told the people who complained, “Look, we use the decision zone chart . . . and we feel the need to do it, and we’re not going to stop and ask and try to coordinate a vote. So I’m sorry.” Having enough of it codified, so you can refer to it when you get criticism, was really helpful.
Despite the rules granting leaders authority to unilaterally request a capital raise, leaders exercising this authority was a deviation from the status quo. By consulting and referencing the rules in their interactions, leaders justified their actions and mitigated workers’ concerns. One worker observed, “I think [the decision zone chart] is really important to give leaders legitimacy and to keep the confidence of the ownership . . . People know it exists and I think as long as you just note that you’re following it, no one’s going to be like, ‘Oh, but are you sure?’”
At Enterprise, formal authority to hire and fire was vested in the “People” role, held by the CEO, CFO, and COO. Historically, the People role-holders had exercised their authority only in cases of underperformance or lack of cultural fit, and they typically consulted relevant team leads before making decisions. However, according to the roles defined in the Glassfrog platform, the People role had explicit authority for all hiring and firing decisions. When the pandemic hit, the People role-holders activated their dormant authority to make unilateral layoff decisions. They met to develop an initial list of people to be laid off, intentionally excluding team leads from the process. One of the People role-holders explained,
So the reason we did it this way, which is not the normal way we do it—it would normally be a very open kind of process—is we decided that we had to act quickly. So within a couple of days’ time we had decided what a first and second wave [of layoffs] would look like, and we acted.
Despite the uncharacteristically centralized process, the shared understanding that the People role granted leaders the authority to make these decisions mitigated potential concerns about the legitimacy of their actions. The founder of Enterprise remarked,
We didn’t need any new governance or roles or accountabilities. We felt like what we had in place [with the People role] served us well, And what we’d done when we implemented holacracy, we made very explicit was that we have a separate role called People, which is accountable and has domain over hiring and firing for the entire company.
At Evolve, the founder held an existing role called the “Compensation Architect,” which was authorized to make decisions regarding “designing, implementing and evolving Evolve’s compensation system.” This role had been used to make permanent salary adjustments, not for temporary pay cuts. However, the founder determined that the role authorized him to implement a temporary 20 percent across-the-board salary reduction. Workers referenced the rule system as justification for his centralized authority. A worker noted,
I think that there’s something in our contract that allows those adjustments to be made, and then Ryan [the founder] holds the Compensation Architect role . . . So, it must have been the Compensation Architect role. Let me look it up, I think he has the authority to make those adjustments.
As depicted here, workers were aware of the operational rules that authorized centralized decision making, and when needed, reviewed the rules to confirm their understanding.
Leveraging collective choice rules to collectively authorize centralization
In some instances, leaders sought to centralize decision making beyond the authority granted by the operational rules. In these situations, leaders at Enterprise, Evolve, and Energy turned to collective choice rules—rules to change operational rules—to formally shift the distribution of authority. The collective choice rules, which were explicitly specified, required formal proposals to be submitted for workers’ input and approval.
As the crisis unfolded at Energy, the leadership team sought to further centralize authority in two ways: empowering the CEO to make layoff decisions and reducing the transparency of leadership team meetings so that discussions of layoffs could be made in private. The decision zone chart specified that under normal circumstances, the CEO could not lay off workers without consulting team leads and obtaining HR approval and that all co-owners could attend board or leadership team meetings. However, the decision zone chart also specified that leaders could petition the worker-dominated board to change these rules to centralize layoff authority and suspend the open-door policy for leadership meetings, which they called “delayed transparency.” A leadership team member explained the rationale for delayed transparency:
We were going to have to have conversations that we didn’t want to have in real time in front of 150 people and that included, what are we going to do about layoffs? . . . Because inevitably, what happens is somebody might call in for one meeting, or only part of a meeting and only hear one conversation about a layoff. What they might not know is that we’re talking about all the possible outcomes. And that might have been something that we were weighing and taking off the table. So it can inflict more fear and cause people to get really scared and nervous.
The leaders followed the process specified by the decision zone chart and secured board approval for both CEO empowerment and delayed transparency. Recognizing that centralization was uncomfortable given the organization’s commitments, leaders emphasized the authorization process they followed:
We had an all-company meeting and before we made any decisions, we said, “Here’s how this is going to work. Here’s the team. Here’s the task force we’ve put together. Here’s how we’re going to make our decisions. And we’ve requested delayed transparency, approval from the board. And they’ve given us a thumbs up because this is an emergency situation . . .” It was a little uncomfortable for a lot of people, because they wanted to be in the room knowing what was happening. But they also accepted it and understood.
Workers accepted the centralization both because leaders followed the authorization process and because the process involved approval from the worker-dominated board. One worker explained,
I mean, it certainly didn’t make it feel any better, but there’s at least a little bit of comfort in, there was some due diligence . . . I think if that information had not been included, people would have lost their minds. They would’ve demanded to understand more about the process . . . I think it helps at least assuage concerns about the legitimacy of the process and about how decisions are being made. It doesn’t completely allay them—people still have concerns and there’s still this tension floating around—but at least there was oversight. It wasn’t just a unilateral decision. The board had a role and that mitigated some of those concerns.
Workers were uneasy about the power shift, but leaders’ adherence to the formal, participatory authorization process helped to “assuage concerns about the legitimacy” of centralization.
A similar dynamic occurred at Evolve, where the founder sought to centralize authority to make layoff decisions. Under normal circumstances, firing decisions occurred through a lengthy, participatory process designed to address performance issues. However, the founder felt this process was too slow to adequately respond to the crisis. He explained,
We realized that we might need to make that [layoff] decision, and our normal process for evaluating when to let someone go wasn’t designed for this. This wasn’t a case of somebody not performing. It was a much more strategic decision of how much do we burn through cash reserves to get through this versus how much do we reduce our burn? So what we did in this case is create a new role specifically to do that, and we gave that role the authority to make that decision and cut people from our staff.
Leaders and workers followed the process set forth in the collective choice rules to create a new role called Cost Cutter, which gave the founder unilateral authority over layoffs. The leader proposed the role, and group members were invited to object or suggest amendments. Despite their concerns about this centralization, workers felt reassured by the process. One worker noted,
So in terms of power granted to a specific role, yes, the new role of Cost Cutter had more centralized authority, but the process to put that role in place was exactly the same as putting any other role in place. It was adapted to the needs of the organization and the present condition of the world, so in that sense it happened exactly how it should have. And even though these conversations were much harder than other business conversations, the tone of them and the way they kind of fleshed out—one partner states an intention to do something, and someone says, “Whoa, bad plan please stop”—that space was available.
Although the discussion was “much harder than other business conversations,” workers emphasized feeling reassured because the standard role-creation process was followed and space for objections remained intact. One worker did object that the initial proposal granted the founder too much unrestrained authority. In response, the group amended the proposal to require the founder to explicitly address the concerns of more experienced partners, and then passed the revised proposal.
At Enterprise, leaders believed that the existing operational rules had already authorized the centralization needed. Still, rather than relying solely on pre-existing operational rules, they leveraged the collective choice rules to create a new “Crisis Management” role, which gave the founder broad authority to respond to the crisis. The head of the HR function proposed the new role, and it was approved through the participatory process specified by collective choice rules. Explaining her rationale, she stated, “My impetus was just for clarity’s sake . . . just to call it out specifically . . . like let’s create clarity for the organization that there is somebody who is looking after how we steward the organization in the midst of crisis.” The episode illustrates how employees valued explicit rule-based authorization enough to go through the pains of proposing a new role and viewed collective choice rules as a valuable tool for authorizing centralization while preserving worker confidence.
Collectively authorizing new centralization allayed workers’ concerns by both requiring their consent and imposing boundaries on centralization. Leaders gained centralized authority, but this new authority was limited in duration and scope. At Energy, when leaders proposed restricting transparency in leadership team meetings, they specified what was to be centralized (access to leadership team meetings) and for how long (30 days). At the end of that period, members knew that transparency would return to normal. A worker on the board of directors described the importance of the temporal limit:
It was kind of like, yeah, this all makes sense. And this is how we can be supportive of the company and hold them accountable by having an end date on this shifting of transparency. And so that’s where I felt the comfort that the trust was there. And I recognize my role as the gatekeeper for that date of like, “Okay, this is when it ends.”
At Evolve, a worker expressed comfort with the Cost Cutter role being temporally and topically bounded:
Yeah, there was a certain discrete set of functions that got temporarily centralized . . . the centralization itself was not . . . It’s not like a war-time powers act or something where the government may or may not give that authority back after you give it to them. It was more like in a traditional organization, you might make a committee for something and give them some focused powers. It didn’t feel like that authority was just being given away forever. It was being put in stewardship for a particular thing so decisions could be made quickly.
That this worker described the centralized authority as something that was “put in stewardship” by workers suggests that even as authority became centralized, it remained a collective resource governed by the workers.
Implicitly Authorizing Centralization Through Fiat
The centralization process at Plant differed markedly from those at Energy, Evolve, and Enterprise. At the heart of this difference was the absence of an explicit or collective authorization process. Prior to the crisis, decentralization was established through principles such as “experts make decisions” and informal practices such as volunteer committees that decided on hiring and promotions and the ability to join meetings in areas beyond one’s immediate work domain. The precise allocation of authority and rules for changing authority were never explicitly codified, and leadership decision making had historically been reserved informally for the two co-founders. As the company had scaled, a “portfolio team” emerged, composed of the co-founders and senior functional leaders. While the portfolio team’s original mandate centered around making product investment decisions, it had begun to make decisions outside of this charter. One employee remarked,
So originally that group of people was called the portfolio team . . . That team was acting in a management capacity as well. So they would talk about the products a little bit, but then they were also making decisions about, for example, what positions we’re going to hire for next, and there were a couple of compensation things that came up.
The crisis hit within this context, triggering greater centralization of decision making in the hands of the portfolio team. With the onset of the pandemic, the portfolio team started to meet almost daily, assuming control over cost-cutting and other resource-allocation decisions that, under normal circumstances, would have been distributed to various teams. For example, a worker sought approval to purchase new equipment essential for advancing core scientific work. She drafted a 15-page investment justification, expecting that her group lead would make the decision. However, the portfolio team intervened and rejected her proposal, citing cost concerns. She described her frustration at the outcome:
Well, I was expecting because experts drive decisions, right, that because this is a purchase within our group, for our group, that [the leader of our group] would be able to make that decision as the leader of the group, right? If we need to make budget cuts, it seems to me that you should say to each team, like each pillar, each program, we need 50% reduction in something. [But] do it the way that makes most sense for you.
The portfolio team’s decision violated expectations that such decisions would be distributed to experts. Unlike at Energy, Evolve, and Enterprise, where authorization occurred in a collective fashion according to clear rules, authorization at Plant occurred through leader fiat, according to no rules, with no oversight or transparency, and without clear boundaries.
The lack of explicit, collective authorization for centralization meant that workers were often unaware of shifting decision-making authority until they were directly affected. The same worker whose purchase was denied described learning about centralization only after her proposal was rejected. When asked whether it was explicit who makes these decisions, she responded, “Now they do, it seems, because that’s where it ended up.” She further described how the change in budgetary decision making had been neither announced nor discussed, explaining, “It was revealed to me.” Highlighting the lack of explicit, collective authorization of the centralization, leaders hesitated to formalize it, despite discussions among some workers that the portfolio team was really acting as a leadership team. One worker explained,
There was a lot of discussion, conversations on the side here, and they’re like, “Why are you guys calling yourself the portfolio team? It’s really more of a leadership team.” But there was kind of a hesitancy to call it a leadership team because of our culture. In this nonhierarchical organization, there’s often a lot of hesitancy that comes with certain words. “Management” has been kind of a bad word for years.
These responses highlight the absence of an explicit authorization process that could have made centralization visible and legitimate to workers. Instead, authorization was obscured from view, dictated by leaders without any worker feedback or consent.
Enacting Centralization
We describe enactment as the second phase of centralization, but it was in fact the defining moment of the process—the point at which centralization manifested in organizational decision making. The ways in which leaders enacted centralized authority differed across our cases, and these differences shaped workers’ perceptions of centralization and their subsequent behaviors. Importantly, enactment was deeply intertwined with authorization. Whether organizations first paused to authorize centralization explicitly and collectively or, instead, moved directly to enactment through leaders’ fiat shaped the character of enactment itself.
Leaders Enacting Centralization with Transparency and Openness
In cases that upheld their commitment to decentralization, leaders not only followed formal rules to collectively authorize centralization; they also enacted their centralized authority with openness and transparency. While the formal rules did not dictate specifically how leaders should enact this authority, rule-based collective authorization shaped enactment in two ways. First, the requirement to gather workers’ input and consent during authorization established a relationship of accountability between leaders and workers, which prompted leaders to be thoughtful and attentive to workers’ concerns during enactment. Second, explicit authorization from workers granted legitimacy to centralization, infusing leaders with confidence to enact their authority with openness and transparency.
The requirement to collectively authorize centralization at Energy, Evolve, and Enterprise established a relationship of accountability between leaders and workers. Accountability refers to “the implicit or explicit expectation that one may be called on to justify one's beliefs, feelings, and actions to others” (Lerner & Tetlock, 1999, p. 255). Leaders could not act unilaterally to centralize; they had to engage with workers, justify the need for centralization, and gain consent.
5
This accountability relationship slowed leaders down and prompted thoughtful engagement with workers. A worker at Evolve noted the constraints on leaders’ behavior:
I think about organizations that were maybe six months into their [Holacracy] practice when the pandemic hit. If I were in those organizations, I would’ve been like “Screw this Holacracy thing, I’m the CEO. I’m arbitrarily making decisions and maybe I’ll ask for your opinion, but maybe not if I think what I’m doing is right.” Ryan didn’t have that option . . . He wanted to save his company that he founded, but he didn’t have the authority to make the decisions he made without putting governance in place, stating his intention, integrating objections when necessary, and compiling that with deep commitment to getting feedback and others’ thoughts.
The requirement to gain workers’ consent to authorize centralization not only constrained leaders, but it also gave them confidence that they had workers’ support. The combined sense of accountability to workers and confidence in their own legitimacy supported leaders in enacting centralized authority with openness and transparency. A leader from Energy explained,
Four of our seven [board] seats are elected from our cooperative, so they’re co-owners of Energy. So there’s this kind of check and balance. You’ve got representation at the board level from other members of the cooperative . . . [and] if you’re confident in Energy and the way we do business, you look at this decision . . . you have to have some level of trust that the way we’re structured has allowed for something like this to happen . . . [But] for some, when they lose friends on their team . . . a knee jerk reaction is, “What the hell are we doing?”. . . And so there is a lot of time spent on the why—how did we get to that decision, where do we go from here—all the things people want to know, you spend time talking that through.
Here, the leader describes how the authorization process produced accountability to workers through the board’s “check[s] and balance[s],” as well as confidence that workers had “trust” in the legitimacy of leaders’ newly centralized authority, which made transparency a natural mode of enactment. Leaders anticipated that workers would have concerns and felt accountable enough to “spend time talking [their decisions] through.” In addition, their openness reflected confidence that they had nothing to hide; they were making decisions that workers had already authorized them to make.
The effort and confidence required to communicate and invite input during enactment was evident at all three of our high-commitment cases. A leader at Enterprise described their conscious decision to over-communicate and invite workers’ input during enactment:
[We] made a really conscious decision, and I would credit Rick with this quite a bit, to share with people, “Here’s what we’re thinking about. Here’s the things we are not considering right now. We’ve thought about them, but they are things we are not willing to do right now, and here are the things that are probably pretty likely” . . . Knowing that people fill in the blanks with the worst case, we gave them that level of detail of information, and we left the door open for people to bring concerns.
At Evolve, even though the Cost Cutter role granted the founder unilateral authority over layoffs, he actively solicited alternative perspectives to challenge his decisions. He described his process:
There were several people I called and told them my draft plan: “I’m thinking about these people, what do you think?” That also helped me not let go of the person I almost let go of . . . Then finally when I had narrowed it down to the two people I did let go of, I called somebody that I knew was a big fan of the full-time woman I let go of, and I said, “Look, I know you’re a big fan. I want you to try to talk me out of this.”
The transparent and open enactments of centralized authority thus continued the dynamics created by the rules-based collective authorization. Guided by their sense of accountability to inform workers about the decision process and confident that workers had authorized them to act, leaders shared openly and sought alternative views during enactment.
Leaders’ transparent and open enactments positively shaped workers’ perceptions of centralization. By openly deliberating and inviting workers’ input, they reassured workers that the organization remained committed to its principles. A worker at Evolve described how the founder’s openness during layoffs eased her concerns:
Someone has to make that decision, we don’t have a lot of time to figure it out . . . But that being said, Ryan was very transparent and open through the whole thing, and there was a lot of discussion, and everyone was involved in the discussion. And as sucky of a discussion as it was . . . we also recognized we have to do something, and somebody is going to get hurt in the process. But I felt like it was very open and honest.
At Energy, a worker similarly noted how leaders’ openness to questions and feedback about layoffs mitigated concerns about centralization: “It wasn’t discomforting because I knew if I saw a name on the list that I had a question about, I could ask someone.” Thus, while workers lacked formal authority over decisions, the way that leaders enacted their authority made workers believe they still had a voice in the process.
While most workers at Energy, Evolve, and Enterprise positively evaluated how leaders enacted their authority, not all felt that leaders were fully open. One worker said that he did not understand “the rationale” behind layoffs, while another said that cost-cutting decisions “didn’t feel very transparent.” However, even in these cases, workers were able to contextualize their negative perceptions of how leaders enacted their authority within the broader frame that their authority had been collectively authorized. A worker at Energy, frustrated by how leaders communicated layoffs, still acknowledged they had the right to make those decisions:
They were well within the authorized powers of their station and position in the company to make these decisions . . . they have the authority to do that . . . [but] I don’t think they handled that decision or displayed great leadership in how they communicated with that, how they managed the emotional impact of that with the organization.
The ability to separate concerns about enactment from concerns about authorization reflects the stabilizing role of the rules-based collective authorization process. Workers could critique how decisions were carried out without questioning the legitimacy of centralization itself.
Leaders Enacting Centralization with Little Transparency and Openness
The enactment of centralized authority at Plant looked radically different. As opposed to the open and transparent enactments of authority that sustained workers’ confidence at Energy, Evolve, and Enterprise, at Plant leaders enacted their authority in a closed and opaque manner. Similar to the other cases, the mode of authorization at Plant shaped the quality of enactment.
At Plant, the implicit, fiat-based authorization process meant that leaders faced no formal requirement to stop and obtain workers’ consent before proceeding with centralization. This contributed to leaders feeling little sense of accountability to workers in how they approached their decision making. A member of the portfolio team, in this instance referring to it as the leadership team, described their decision process regarding cost cutting:
I would say philosophically, people on the leadership team don’t want to make these [detailed cost-cutting] decisions . . . It was an interesting experience because we had this kind of crisis situation where we need to do some austerity measures, we need to manage our burn, and that immediately got into a pretty detailed exercise around stuff, and we just were kind of caught up in it . . . But I think within a few weeks of that experience, at least personally, I thought we would’ve had better outcomes if we had just gone to each major group in the company and said, “We don’t care how, but you guys need to trim 20%” or whatever it is, and just left it at that. And it was just one of these things that never got to the table. We didn’t pause long enough and ask what would’ve been a better way to do this. So we kind of went with it and then got busy doing other stuff.
Here, the Plant leader acknowledges that centralization went against the principles of decentralization and that leaders were “caught up in” the process without sufficient deliberation. Unlike the leaders in the other cases, leaders at Plant “didn’t pause long enough” to consider how to best enact their centralized authority. Implicit in this reflection is the fact that leaders were not accountable to workers; they did not have to obtain consent to authorize centralization, and when it came time to enact their authority, they did not feel compelled to stop and communicate with workers.
The absence of explicit, collective authorization, while leaving leaders unconstrained, also left them tentative about the legitimacy of centralization. For example, the pandemic had led to a decline in the activity of the volunteer HR committee, and instead, the HR leader began to make decisions unilaterally. However, the HR leader described how she was hesitant to make certain decisions despite having the bandwidth and desire to do so:
There’s a particular policy we really need to clean up . . . I had a lot of time on my hands and I could have done a whole lot of work, but it just wasn’t quite the time to be rolling out a bunch of new people initiatives. There just wasn’t an appetite and there was almost a little bit of suspicion that it was nefarious. . . . Like, why is the company rolling something out now? How is this going to impact me? And surely this is something to their benefit and not mine.
Perhaps the clearest sign that leaders were tentative about the legitimacy of centralization was the portfolio team’s hesitation to acknowledge that they were now acting as the company’s de facto leadership team, even though workers perceived it as such. Instead, they tried to keep centralization under the radar, hoping to avoid workers’ concerns. One worker explained,
I know, in talking to some members of the leadership team, they felt like as soon as they call themselves the leadership team, and this is sort of their hesitancy in calling themselves that, people would assume that they need you to go to them for every decision.
Even when the portfolio team, months later, eventually changed their name to the “leadership team,” leaders did not formally announce the change. Instead, they simply changed the name of the company-wide email distribution list.
The opaque authorization and enactment of centralization, intended to preserve decentralization, only fueled doubts about leaders’ commitments to it. One worker explained,
One of the things that was a frustration—It was hard to understand how some of these decisions were being made, right? It was hard to understand what the portfolio team was actually doing. What are their motivations? We’re a very transparent company, which I think is good, but when you’re a super transparent company and something doesn’t feel transparent or understood, that can raise some red alert flags, and I think that can happen even when that’s not intended to happen. And so to me, just the lack of understanding how some of these decisions were being made where things didn’t feel aligned—When you have a belief and behavior that says experts make decisions, but it seems sometimes there would be an asterisk. Experts make decisions, asterisk, except in times of financial difficulty . . . that can be a little bit frustrating for people. There’s really a lot of asking people to keep faith that everything is going to come through.
The lack of both collective authorization and transparent enactment of centralization left the boundaries around centralization ambiguous and made workers question whether decentralization was truly a core commitment or merely an ideal that the organization abandoned when convenient. Instead of receiving clear confirmation, workers were asked “to keep faith.”
Highlighting the extent of the communication void from leaders, workers turned to gossip. One worker described how rumors and hearsay filled the information void:
Some of this stuff I hate to even repeat because it’s more like hearsay as opposed to there was an official announcement, but my friends here would be like, “Well did you hear what the portfolio team decided?” and I’d say, “No, because I’m not on that team.” And there were some situations where decisions were made . . . for example, somebody wanted to buy something recently and they were told no. The portfolio team was the one saying, “Well we’re not going to be buying this stuff,” and it’s like okay, why would the portfolio team say that? So I think some of it was just hearsay, people being like, “Well the portfolio team blah blah blah . . .”
Scholars have found that rumors and gossip emerge “when people lack verified information on issues that they deem important or when they mistrust their sources of formal communication” (Sobering, 2019, p. 414). At Plant, the lack of explicit authorization and transparent enactment of centralization created an informational vacuum that workers filled with speculation and hearsay.
Unlike at Energy, Evolve, and Enterprise, where workers could contextualize perceptions of leaders’ enactments within a frame that centralization was collectively authorized, workers at Plant lacked this frame. Instead, their perceptions of centralization—its implicit fiat-based authorization and its opaque enactment—fused together and told the same story: that centralization was unbounded and the organization less committed to decentralization. A worker captured this sentiment:
I do think it is a little bit more closed than it used to be. So people feel like they are not as much a part of any of the big decisions, or at least a lot of those discussions that are going on. They’re not hearing all of that information, or at least they feel like they’re not.
Ironically, leaders obscured centralization in an effort to safeguard the firm’s decentralization commitments, fearing that naming centralization would trigger concern. Yet, this very opacity produced the opposite effect: It left workers in the dark and deepened skepticism about whether decentralization still held.
Outcomes of Centralization
The way that centralization unfolded in each of our four cases not only shaped how workers perceived their organization’s commitment to decentralization; it also drove distinct behavioral outcomes. As centralization continued, workers at each organization began to act in ways that either reflected confidence in decentralization’s continued importance or signaled its waning significance.
At Energy, Enterprise, and Evolve, workers’ behaviors reflected a belief that authority remained fundamentally decentralized. One behavioral manifestation of this belief was that workers continued to exercise authority within their roles. At Evolve, where operational rules granted workers autonomy to fulfill their roles as they saw fit, workers continued to exercise that role authority even after the creation of the Cost Cutter role. One worker explained, “All of the business decisions to some extent stayed the same, they were still done by the local role holder who makes these decisions.” In this way, operational rules supported workers in exercising their retained authority.
In some instances, centralization events directly challenged workers’ authority granted by the operational rules. At Enterprise, leaders directed employees to turn on their webcams during virtual calls. While existing operational rules authorized leaders to make this request, the same rules also affirmed that workers had autonomy in how they fulfilled their roles. Despite this seeming conflict, workers used the rules to interpret this directive in a way that preserved their discretion. One worker described her response to the directive:
At first, with everything else going on, [the webcam directive] just felt too much. But I saw some of the reward of it. I kind of played with it a little bit, what it helped and what it didn't help, and then came back to the realization that nobody can tell you exactly how you have to do your job. And so it did feel like a very uncomfortable nudge but then again, it kicked off some, how can I do my job better? If it doesn’t help me do my job better, including taking better care of the customer, then maybe that scenario didn’t need it. And so I usually have taken my lead from customers. If they’ve got their camera on, I’ll turn my camera on. If they don’t have it on, and I don’t have to see them and I am not in a scenario where I need to see their facial reaction, then we didn’t use it.
While the worker initially perceived the directive as an infringement on her role authority, she ultimately reminded herself of the rule that “nobody can tell you exactly how you have to do your job.” Drawing on this rule, she interpreted the directive not as a strict requirement but as an “uncomfortable nudge.” By using the operational rules as a resource to enact their retained authority, workers demonstrated their confidence that decentralization remained in force.
By contrast, at Plant, the erosion of workers’ belief that decentralization remained a commitment led them to increasingly defer to leadership, even in areas leaders had not intended to centralize. The implicit, opaque nature of the centralization process left workers uncertain about the boundaries of their authority, prompting them to seek leadership approval for decisions that had once been made autonomously. A worker remarked on the emergent conservatism:
There were a couple of people on that team that always wanted to default to, “We should ask the leadership team.” And I would have to keep saying, “No, this is our decision. We’re the ones leading this” . . .There’s going to be certainly some uncertainty about what level that needs to be or where some of those decisions need to fall, and maybe some people feel less empowered to make some decisions. But that is not the intention [of the leadership team] at all, I know that. There’s something about human nature and hierarchy that they feel like that’s the way it should be or the way it is because we have the leadership team here now.
The proliferating deference of workers to the leadership team, an unintentional by-product of the way that centralization occurred at Plant, expanded centralization beyond the scope of what leaders had intended and compounded a sense that the organization had lost an important aspect of its decentralized character. One worker expressed concern:
One of the most appealing things to me about Plant that was slightly crazy and it drove me crazy on some levels was that people were so empowered to do things. And I feel like some of the empowerment has been lost . . . People are becoming more and more conservative . . . I think that, if I have a tinge of sadness, it’s that Plant was so unique, and I thought functioned so well, and I loved the way we did things even though it took massive amounts of energy, and perseverance. I worry slightly that we lose that.
Another behavioral outcome of the centralization process was how workers and leaders chose to end centralization—or not. When workers and leaders collectively authorized centralization and leaders enacted it transparently, they were likely to end centralization in an explicit and formal manner. For example, at Evolve, workers and leaders eliminated the Cost Cutter role by using the same process used to create it, both marking the formal end of centralization and reaffirming the organization’s commitments. One worker noted, “[T]he Cost Cutter role has been removed from governance, and just having that removed from governance feels better.” At Energy, workers and leaders took similar steps to formally mark the end of centralization. The board chair, a worker-owner, described how he took the time to publicly announce the end of delayed transparency:
There was a big moment at the subsequent board meeting, in that public forum, because co-owners can listen into the board meetings. [We announced that] this is the end date [for delayed transparency], that that is no longer in place. Just so everybody knows . . . At that point, [delayed transparency] wasn’t really being used . . . so I had that added . . . I have always felt really strongly about doing things like that. Symbols and rituals are important.
Even though leaders had already ceased to use their centralized authority in practice, they felt it was still important symbolically to publicly and explicitly remove the centralization. 6
At Plant, however, no such formal end to centralization occurred. Just as centralization had been initiated in an implicit and opaque fashion, its conclusion was left undefined. Perhaps not surprisingly, workers perceived the absence of a clear ending as further evidence that the shift toward centralization was permanent. One worker observed,
I was in a conversation—this was prior to the deal going through, and we were talking about another much smaller deal, and somebody asked me, “Well do you think this needs approval from the leadership team?” And you know, it was an interesting question. I hadn’t even thought about it, it was so not consequential in the big picture . . . And so it suddenly dawned on me, people are starting to ask that question. That’s a huge change . . . It’s become this much more formal structure that was not really in place before the pandemic . . . I’ll be shocked if that disbands post-pandemic.
Across our cases, centralization processes shaped not only perceptions of commitment but also behavioral responses, including whether workers continued to enact their retained authority and whether centralization was formally removed. In two tables below, we present additional evidence of the distinct centralization processes and their outcomes in our cases of upheld commitment (Table 5a) and in our case of undermined commitment (Table 5b).
Evidence of Centralization Processes Upholding Commitment
Evidence of Centralization Processes Undermining Commitment
Considering Alternative Explanations
Our analysis indicates differences in the authorization and enactment of centralization as key to understanding differences in the strength of resulting decentralization commitments across our cases. Nonetheless, we briefly consider three alternative explanations for the variation we observed.
One possible alternative explanation relates to differences in ownership structure. Two of the three cases in which commitments were upheld were employee owned: Energy was a worker cooperative, and Enterprise was fully owned by employees through an employee stock ownership plan (ESOP). Given this ownership structure, workers may have viewed centralization as less of a threat to the organization’s governance principles because they had the legal right to revoke it. However, in only one of the three high-commitment cases, Energy, did ownership provide a legal means for workers to protect decentralized authority. Evolve was not employee owned, and the ESOP at Enterprise did not provide workers a means to sanction or alter the authority of leadership. 7 Hence, variation in ownership structure cannot fully explain the differences in workers’ perceptions that we observed.
A second alternative explanation is that Plant had greater centralization prior to the crisis, as evidenced by the fact that the group of leaders called the “portfolio team” had begun to informally take on more decisions prior to the crisis. However, two factors undermine this alternative explanation. First, a worker at Plant noted that the leadership team “was not really in place before the pandemic” and that “as the pandemic came through, it crystalized [it] into place.” Second, similar concerns about pre-crisis centralization existed in other cases, most notably at Energy, where a new leadership team had emerged two years prior to the pandemic. Yet, unlike at Plant, where leaders refused to acknowledge the emergence of the leadership team, Energy’s leaders not only acknowledged the team’s emergence but also codified its charter in the decision zone chart and submitted it for workers’ approval. Thus, what distinguished Plant was not the presence of pre-crisis centralization, which appeared elsewhere, but the refusal to acknowledge and legitimate it through collective authorization.
A third alternative explanation is that firms that operated face-to-face before the crisis, and thus experienced greater disruptions to their routines, faced greater challenges in preserving commitment to decentralization during the crisis. As a biotech firm with research laboratories, Plant engaged in substantial in-person work, and the shift to remote work may have made workers feel less connected to the company, feeding mistrust of leadership actions during the pandemic. However, Energy also operated face-to-face prior to the crisis and yet was able to uphold perceptions of commitment to decentralization to a far greater degree than Plant was able to do. In short, while the qualitative design of our study and small sample size do not allow us to formally test or exclude alternative explanations, several of the most obvious alternative explanations do not appear to explain the pattern of variation we observe.
Theoretical Model of Democratic Deviations from Commitments
Building on the findings presented above, we developed a model, shown in Figure 1, of how organizations can navigate pulls toward centralization in ways that uphold rather than undermine their commitments to decentralization. Our model depicts two distinct pathways of navigating centralization, one that reinforces decentralization commitments, which we call “democratic deviation,” and the other that weakens them, which we call “monocratic deviation.” For each pathway, we theorize the leader and worker behaviors during the centralization process that shape workers’ perceptions and behavioral responses, as well as the structural features whose presence or absence support this process.
Model of How Organizations Can Centralize While Upholding Their Commitment to Decentralization*
The democratic deviation pathway, observed at Energy, Evolve, and Enterprise, depicts how organizations can centralize while upholding decentralization commitments. This pathway is supported by democratic rules of authority, which we conceptualize as operational and collective choice rules about who holds authority, created and changed through processes involving workers’ participation and consent. Leaders and workers leverage these rules to explicitly and collectively authorize centralization, which can reinforce decentralization commitments by setting clear boundaries on centralization and signaling that deviations from decentralization are governed through collective oversight—implicit oversight when drawing on operational rules and explicit oversight when engaging collective choice rules.
Collective authorization not only directly reinforces workers’ perceptions that commitments are upheld but also shapes how leaders subsequently enact centralized authority. The requirement to secure explicit and collective authorization for centralization binds leaders into a relationship of accountability with workers, prompting intentionality in their exercise of authority, and confers legitimacy to their authority. As a result, leaders are likely to feel both obligated and empowered to enact their authority with transparency and openness to contestation, supporting workers in feeling they have influence in the process and reinforcing their sense that decentralization commitments remain intact.
Beyond shaping workers’ perceptions, our model indicates how democratic deviation from decentralization commitments—the interconnected processes of collective authorization and open and transparent leader enactments—can also drive concrete behavioral outcomes. First, when workers see decentralization commitments as intact and centralization as clearly bounded, they are more likely to actively exercise their retained authority rather than to passively concede it. Second, because workers continue to perceive decentralization as legitimate and supported by collective choice rules, both leaders and workers are able and likely to use these rules to explicitly end centralization once it is no longer necessary. These behavioral outcomes further reinforce perceptions that centralization is temporary and decentralization commitments remain in place.
By contrast, the monocratic deviation pathway, exemplified by the Plant case, depicts the process of centralizing in a way that undermines a shared commitment to decentralization. In this pathway, centralization is not explicitly or collectively authorized but, rather, occurs implicitly through leader fiat, leaving workers with the perception that centralization is unbounded and lacks oversight, and undermining trust in decentralization commitments. In turn, fiat-based authorization neither binds leaders into a relationship of accountability with workers nor instills confidence in the legitimacy of their new authority. As a result, leaders are prone to enact their centralized authority in a tentative, reactive, and opaque manner. At Plant, leaders implemented cost-cutting decisions without much thought or deliberation and hesitated to acknowledge centralization until weeks after decisions had been made. Opaque leader enactments of authority can reinforce the sense that workers have no influence over the process, undermining their beliefs that decentralization commitments remain in place.
Ultimately, like the pathway of democratic deviation, the pathway of monocratic deviation also leads to distinct perceptual and behavioral outcomes. Workers are likely to perceive that decentralization commitments have weakened, leading them to defer to leadership beyond the intended scope of centralization. This behavioral shift can further reinforce workers’ perceptions of weakened commitment, creating a self-reinforcing cycle in which centralization grows and endures. Notably, in the pathway of monocratic deviation, an explicit end to centralization is unlikely. Even if leaders choose to end it through fiat, the absence of democratic rules for doing so means that termination is likely to remain implicit and open to misinterpretation.
The model’s two contrasting pathways highlight the critical importance of the authorization phase of centralization. Collective authorization slows down leaders, prompting them to proceed thoughtfully and in a participatory way in deciding both whether to centralize and how to enact it. By contrast, implicit authorization through fiat allows leaders to move quickly. However, without democratic constraints to slow them down, leaders are more likely to act in a reactive and unilateral manner, enacting centralization in ways that are opaque and erode decentralization commitments.
The link between the authorization and enactment phases also illustrates a core feature of our model: its partial path dependence. While collective authorization makes transparent and open enactments of centralized authority more likely, and fiat-based authorization makes opaque enactment more likely, these relationships are best understood as probabilistic rather than deterministic. While we did not observe other patterns of authorization and enactment, we can consider hypothetical alternatives. For example, leaders could centralize by fiat but then enact their authority transparently. Conversely, centralization could be collectively authorized, but leaders might enact their authority opaquely. Even in our cases, we observed minor instances in which leaders deviated from expected patterns, highlighting the enduring role of leadership in shaping the centralization process.
Finally, our model spotlights the critical role of democratic rules, both operational and collective choice, in supporting the democratic deviation pathway. The rules can clarify when deviations from the commitments have occurred, establish that such deviations require collective authorization to be legitimized, and offer a structured mechanism for both initiating and halting centralization. In doing so, the democratic rules link leaders and workers through structured processes of communication and accountability, allowing them to establish explicit agreements about the beginning of centralization, its scope and duration, and its end. Furthermore, the rules can provide workers with clarity on the spaces of decision making in which they retain authority, supporting their continued enactment of authority in the midst of centralization.
However, while it is a critical enabling condition, the presence of democratic rules is insufficient to preserve commitments to decentralization. Workers do not evaluate decentralization commitments based on whether rules exist in principle but on whether and how leaders engage with these rules in practice. The model visually depicts this indirect relationship: The democratic rules impact workers’ perceptions only through leaders’ and workers’ actions. While in our cases of upheld commitment, leaders and workers actively referenced and leveraged these rules throughout the centralization process, an organization could have ceremonial democratic rules that are neither followed nor enforced. Our model suggests that in such cases, decentralization commitments would not be preserved.
Our model thus captures the dynamic interplay between democratic rules and their use in democratic deviation. This interplay reflects the duality of structure and action consistent with theories of structuration and practice theory (Feldman & Orlikowski, 2011; Giddens, 1984). Our explanation of how decentralization commitments can be upheld even during centralization is therefore both structural and processual.
Discussion
An enduring assumption in organization design research is that decentralized authority structures are fragile. This lack of durability stems in large part from the difficulty of meeting situational demands for centralization. In broad terms, any urgent situation involving heightened demands for coordination or a sense of survival risk may prompt organizational leaders to consider centralization. Because decentralization is counter-normative, even temporary or well-justified interventions by leaders are likely to trigger workers’ fears about wavering commitments (Baker et al., 1999; Stea et al., 2015). As a result, extant theory has emphasized resisting centralization—even occasional or temporary—as the primary way to sustain credible decentralization commitments (Foss, 2003; Freeland & Zuckerman Sivan, 2018; Stea et al., 2015). However, this leaves leaders of such organizations with a stark choice: avoid any centralization or disavow their commitments to decentralization completely. Our article offers a potential way out of this bind by identifying how organizations can centralize in ways that uphold rather than undermine their decentralization commitments. In doing so, our findings show how decentralization can be resilient to centralization pressures rather than fragile to such demands. Given the mutually constitutive relationship between structure and action reflected in our model, our findings build theory on both the processes and structures of durable decentralization.
Processes Supporting Durable Decentralization Commitments
This article offers three key theoretical insights about the process by which decentralization can endure moments of centralization. First, preserving commitments to decentralization does not require avoiding centralization, as prior theory has presumed. Instead, what matters is the process by which centralization unfolds. Centralization, if carried out in a democratic fashion, can be experienced as consistent with, and even reinforcing, decentralization commitments, whereas centralization imposed monocratically erodes them. Thus, structurally similar acts of centralization can produce divergent effects depending on how they are carried out.
Second, this article highlights two distinct but interrelated phases of the centralization process, authorization and enactment, that together shape how centralization is received. While prior research has proposed that transparent enactment of centralization improves workers’ responses to it (Asmussen et al., 2019), our research both supports this proposition and goes beyond it. Specifically, our findings highlight the critical role of the authorization phase that precedes enactment. Existing accounts often assume that centralization is authorized through leader fiat (e.g., Freeland & Zuckerman Sivan, 2018), but our findings show that authorization can occur through processes of democratic consent. Such authorization transforms the meaning of centralization from a breach of commitment to a legitimate, bounded, and commitment-reinforcing act. In this sense, the character of authorization emerges as a central lever that shapes whether centralization upholds or undermines commitments.
Third, our findings illuminate the mechanisms through which authorization influences the durability of decentralization commitments. Collective authorization constrains centralization’s scope and duration, subjects it to oversight, and instills in leaders a sense of accountability and legitimacy that inclines them to enact authority openly and transparently. In contrast, authorization by fiat leaves centralization unbounded and leaders unchecked, weakening their own sense of accountability and legitimacy and making opaque enactments more likely. These mechanisms reveal how the form of authorization shapes the subsequent beliefs and behaviors of both leaders and workers, ultimately determining whether commitments are sustained or undone. Notably, while collective authorization is a more complex, and therefore slower, process than centralization by fiat, it serves as a generative pause, one that not only secures workers’ acceptance and leaders’ legitimacy in the moment but also recalibrates the ongoing interactions through which decentralization is sustained.
The Structures Supporting Durable Decentralization Commitments
While our model is processual, it offers three structural insights into how decentralization can be made more durable. First, our findings highlight the existence and importance of decentralization at multiple levels of authority. Existing theories have primarily focused on decentralization at what Elinor Ostrom terms the operational level, where decisions that directly shape work, resources, and strategy are determined (1995). However, Ostrom also theorized a deeper collective choice level, where rules about who decides are made. Our study suggests that the durability of decentralization depends not only on how authority is distributed operationally but also on whether collective choice authority is decentralized. In cases of upheld commitment, collective choice-level authority was decentralized such that the decision to centralize operational authority was made through a collective process. In the case of undermined commitment, collective choice authority remained centralized in leaders’ hands, enabling unilateral centralization of operational authority. Our findings suggest that the durability of decentralized structures depends less on maintaining a specific distribution of authority at the operational level and more on a collective commitment to making those distributional decisions together.
Extant theory in organization design has lacked this multi-level perspective on decentralization, which explains the seeming paradox of our findings. Existing conceptions of decentralization have focused almost exclusively on the distribution of operational-level authority; only in research on organizational democracy, largely in sociology and industrial relations, have scholars been attentive to the decentralization of collective choice authority (Ben-Ner & Jones, 1995; Chen & Chen, 2021; Ferreras et al., 2024; Russell, 1985). Yet, even in this literature, scholars often theorize organizational democracy as an uncompromising avoidance of hierarchy and have not explored the design flexibility that this form of decentralization might afford. Decentralization at deeper levels allows for a temporary decoupling of the distribution of operational and collective choice authority such that centralization can occur at the operational level while decentralization remains at a deeper level.
Second, our findings suggest the need to revise theories on the centrality of ownership in supporting durable decentralization. Research in organization design has assumed that in firms where ownership is concentrated, decentralization is invariably fragile because ownership rights grant fiat authority to owners, and as a result, delegated decision rights can be loaned but never owned (Foss, 2003; Freeland & Zuckerman Sivan, 2018; Williamson, 1981). We find, however, that durable forms of decentralization can exist even when ownership is concentrated. Evolve and Enterprise decentralized authority at the collective choice level even though, by legally enforced property rights, leaders still held fiat authority. In this sense, the constraints on leaders’ ability to centralize by fiat in these cases were not endorsed by the sanctioning power of the law, but they were nevertheless upheld as authoritative by the collective agreement of organizational members. Breaking the rules, while legally possible, was nevertheless normatively costly.
Thus, we theorize that distributing ownership is not the only way to deepen decentralization beyond the operational level and that it is possible to establish de facto constraints on leader fiat, even when such constraints are not de jure sanctioned. While ownership may be necessary to ensure the durability of decentralization under more extreme circumstances, our findings suggest that decision rights can be effectively loaned to and owned by workers, even as ownership rights in the firm remain concentrated.
Third, our findings spotlight the value of formal rules in supporting more flexible and durable forms of decentralization. Traditionally, scholars have associated decentralization with informality, viewing informality as helpful in reducing the rigidity of centralized bureaucracy (Burns & Stalker, 1961; Kellogg et al., 2006; Rothschild-Whitt, 1979). However, in our study, formal, democratic rules governing authority were critical for supporting democratic deviations. Such rules discouraged leaders from centralizing by fiat, clarified when a deviation from commitments was occurring, and provided a shared process for authorizing and ending the deviation. The democratic, formal rules not only lent legitimacy but also improved efficiency, streamlining deliberation and agreement. While existing theory has largely dismissed collective decision processes as too costly in times of crisis (Hansmann, 1988), our findings indicate that formal rules can make them reasonably efficient. Thus, our study suggests that the fragility of decentralized structures results not from the fact that the structure is too decentralized but, rather, from the fact that the decentralization is insufficiently structured (Freeman, 2013). As a result, we propose that decentralized structures that are formally institutionalized through democratic rules governing authority are more flexible and durable.
Sustaining Organizational Commitments Through Democratic Deviation
Finally, our findings point to a broader proposition about the role of democracy in helping organizations to sustain their foundational commitments: When organizations are compelled to situationally diverge from core principles, deviations that occur democratically will be perceived as an expression of the commitment rather than as a breach. This proposition speaks to several literatures concerned with how organizations can adapt to environmental demands while maintaining shared commitments, values, and principles, and it contributes to emerging scholarly interest in the role of democracy in organizations (Battilana et al., 2025).
First, the proposition has relevance for research on corporate purpose. Although purpose has long been a foundational concept in organizational theory (Barnard, 1968; Selznick, 1957), it has gained renewed attention as a means for firms to look beyond profit-seeking and to galvanize stakeholders around the pursuit of societally transformative goals (Durand & Huynh, 2024; Gartenberg & Zenger, 2023; Gulati, 2022; Valdés et al., 2025). An ongoing question for corporate purpose scholars concerns how corporations can maintain long-term commitments to purpose in the face of short-term market pressures. Scholars have suggested that democratic governance may help but have provided limited insight into the mechanisms or processes by which democracy can help sustain corporate purpose (Davis, 2021; Kaplan, 2023). Our model of democratic deviation helps to fill this gap, suggesting that democratic rules can support corporations in navigating pressures to deviate from one’s purpose while keeping the purpose intact. Furthermore, beyond abstract ideals, our model points to concrete democratic processes, such as collective authorization and transparent enactment, that can contribute to making purpose resilient in the face of competing pressures.
Second, our model of democratic deviation contributes to a broad set of literatures exploring how organizations preserve complex organizing practices in the midst of disruptive change (Fitzsimons et al., 2024; Gibbons & Henderson, 2012; Weick, 1993). Past studies have emphasized how disruptive change can narrow leaders’ attention, restrict information flows, and provoke anxiety, leading to defensive and fragmented responses that undermine collective action and lead to breakdowns in organizing practices (Fitzsimons et al., 2024; Staw et al., 1981; Weick, 1993). Recent research has suggested that principle-based relational contracts may help organizations weather such shocks because of their inherent flexibility (Gibbons et al., 2023). Our findings refine this view by showing that flexibility without procedural clarity can itself exacerbate fragmentation: When principles lack shared rules for how they should be applied or amended, they invite divergent interpretations at precisely the moment when common meaning is most essential. In our positive cases, rules that specified both current expectations and the procedures for altering them maintained cohesion under threat. Democratic rules requiring leaders to deliberate, consult, and interpret collectively transform heightened anxiety into coordinated, rather than fragmented, action. In this way, democratic deviation functions as a form of collective emotional regulation and sensemaking (Weick & Roberts, 1993), preserving shared commitments and enabling value-consistent responses under pressure.
Finally, across these literatures, much has been made about the role of leadership in sustaining organizational commitments. Scholars have indicated the need for “courageous leaders who can deal with the discomfort that comes with assuring a purpose-driven organization” (Kaplan, 2023, p. 297) and “leaders who can interrogate the emergence of anxiety . . . instead of treating it as a shameful personal frailty” (Fitzsimons et al., 2024, p. 22). Our findings do not rule out the importance of leadership, but they do offer evidence of a distinct and separate perspective: that democratic systems may foster the kind of leadership that can sustain commitments. In our cases, leaders acted with the aforementioned qualities not simply because of innate disposition but because democratic rules demanded and supported it. Democratic rules constrain leaders from deviating too far from commitments, but they also empower them to act. They impose accountability on leaders, which prompts intentionality, engagement, communication, and ultimately confidence. The very constraints that democratic rules impose also generate the legitimacy and grounded confidence that enable leaders to act with clarity and transparency. This suggests that sustaining organizational commitments in the face of turbulence may stem less from exceptional leaders than from democratic systems that bring leaders into closer relationship with their followers, systems in which leaders are both explicitly accountable to and authorized by those they lead. In doing so, democratic organizing may help to produce leaders who are courageous rather than tentative, engaged rather than isolated, and open rather than defensive.
Generalizability to Other Contexts Prompting Centralization
Any study conducted in a context as singular as the COVID-19 pandemic inevitably raises questions about generalizability. Given the life-threatening nature of the pandemic, one may question whether our findings apply to other situations that prompt centralization. The severity of COVID-19 may have led workers to broadly grant leaders goodwill and deference out of a desire for safety and security. Indeed, across our four cases, workers acknowledged that the urgency and unprecedented nature of the crisis justified leadership exercising greater control. However, this reality only makes the variation in responses to centralization that we observed more notable. We might have expected uniformly high deference from workers; yet, we observed meaningful differences in both perceptions of and behavioral responses to centralization, which were shaped by the process through which centralization was authorized and enacted. In less extreme contexts, where workers may be less forgiving of leaders, the processes we theorize might become even more important in preserving workers’ confidence in their organizations’ core commitments. We hope that future scholarship will examine how our theory extends to other contexts.
Conclusion
Organizations that embrace decentralization continually face the risk that situational pressures will compel shifts toward centralization, threatening both their commitments and their legitimacy. However, with the right structures and processes, organizations can flexibly and temporarily deviate from their decentralization commitments without undermining them. By limiting leaders’ ability to centralize unilaterally, decentralized organizations enhance their capacity to centralize strategically. Collective adherence to a set of democratic rules makes decentralization more flexible and, ultimately, more durable.
Supplemental Material
sj-pdf-1-asq-10.1177_00018392261421927 – Supplemental material for Democratic Deviations: How Organizations Sustain Decentralization Commitments in the Face of Centralization Pressures
Supplemental material, sj-pdf-1-asq-10.1177_00018392261421927 for Democratic Deviations: How Organizations Sustain Decentralization Commitments in the Face of Centralization Pressures by Michael Y. Lee and Trevor Young-Hyman in Administrative Science Quarterly
Footnotes
Acknowledgements
This manuscript benefited from the thoughtful and generative guidance of Siobhan O’Mahony and three anonymous reviewers. We also received valuable perspectives from Robert Freeland, Arvind Karunakaran, Gianpiero Petriglieri, Jennifer Petriglieri, Phanish Puranam, and members of the INSEAD inductive brown bag. Finally, we would like to thank the members of Enterprise, Evolve, Energy, and Plant for participating in this study and allowing us to learn from their experience.
1
Both of these cases that dropped were worker cooperatives and were otherwise similar to the other four cases in size. Despite the organizations becoming unresponsive, early data collection at these two cases revealed that both had centralized decision making during the crisis.
2
We thank the reviewers for encouraging us to develop our theorization in this direction.
3
Workers at Enterprise and Evolve already worked remotely prior to the pandemic, whereas workers at Plant and Energy worked face to face.
4
We draw this distinction between operational and collective choice rules from Elinor Ostrom’s institutional analysis and development (IAD) framework (Ostrom, 1995). Within Ostrom’s IAD framework, institutional systems are composed of multiple levels of action and decision making. Rules at each level define who will participate in a decision and what potential actions they can take at that level. Two levels of rules are relevant for our purposes: operational rules define the participants and potential actions in “day-to-day decisions,” and collective-choice rules are “the specific rules to be used in changing the operational rules” (Ostrom, 1995, p. 58).
5
When leaders leveraged operational rules, the engagement with workers was more implicit in that leaders needed to engage with the rule system that workers and leaders had previously established. When leaders leveraged collective choice, the engagement with workers was explicit, as a common feature of these rules was an oversight process involving the workers’ consent.
6
Among our cases of upheld commitment, an explicit end to centralization was least visible at Enterprise, where most of the centralization had occurred through leveraging operational rules to activate dormant centralization. As a result, there were no significant centralized rules or roles to eliminate. Yet, even here, workers noted the end of centralized operational authority, for example, remarking, “The cameras on directive didn’t last.”
7
The ESOP at Enterprise is best understood as a benefit plan whereby employees have voting rights only on major corporate transactions, including the sale of the company or merger with another company. The plan is administered by a trustee whose legal responsibility is to maximize share financial value and who is not directly elected by the employees. In turn, that trustee selects a board of directors (Rosen et al., 2011).