A The programmatic view of organizational capabilities

In neoclassical economics, a firm is understood as a production function that transforms a vector of inputs into a vector of outputs. The actual processes involved in turning factor market inputs into product market outputs are abstracted in these theories so that the internal processes and dynamics of the firm are treated as a ‘black box’, inside of which no description is offered.

Of course the similarity of the production function to a computer which takes inputs, process the input, and then return outputs, did not go unnoticed and opened attempts to describe the black box activity in terms of deterministic computer processing. The capabilities perspectives articulated by March and Simon (1993) and Nelson and Winter (1982) explicitly sought to explain how firms generate specific outputs from a particular set of factor market inputs by opening up the organizational black box and looking at the mechanics of production and organization.

The central idea that emerges in both of these approaches is that firm-level capabilities can be understood in terms of more basic elements that at their core are simple automata that execute deterministic programs when triggered by the appropriate stimulus. Action at the organizational level, we are told, is the result of a complex concatenation of rule-based elements, each operating by “matching of appropriate action to recognized situations” (March & Simon, 1993, p. 11).

March and Simon (1993) and Nelson and Winter (1982) each argued that organizational capabilities operate on the same principles as individual skills, borrowing heavily from the newly emerging paradigm of cognitive science that was arising at the time – a paradigm that Simon himself was instrumental in formulating as one of the early pioneers of cognitive psychology and artificial intelligence. March and Simon’s (1993) theory of organizational capabilities was deeply influenced by the parallel work Simon was pursuing in cognitive science on the nature of cognition; the theories in many ways were isomorphic.

As Simon himself noted, production systems “have an obvious affinity to the classical stimulus-response (S → R) connections in psychology” (1999a, p. 676). And in the same way that an individual’s complex behavior could ultimately be explained in terms of the operations of concatenated stimulus-response systems, March and Simon (1993) and Nelson and Winter (1982) argued that an organization’s capabilities ultimately constituted the programs that were the microfoundations which underwrote the organization’s actions as an productive entity (cf. Felin & Foss, 2005, 2009; Barney & Felin, 2013).

Accounting for how capabilities change and how new, de novo, capabilities are created has always been a weakness of the capabilities approach. March and Simon (1993) and Nelson and Winter (1982; see too Winter, 2012, 2013) have each provided an account based on what can broadly be considered evolutionary (though not necessarily Darwinian in a strict sense) processes. At the core of March and Simon’s (1993) approach is the construct of search. In novel, i.e. non-routinized, circumstances, whether they arise out of endogenously or exogenously driven changes, these routinized programs lie fallow. Absent the normal triggering flows of information, the capabilities are not exercised. The absence of a triggering stimuli doesn’t mean an absence of action, however. An aspiration of a certain level of organizational performance is itself an internal prompt to action of an ad hoc, improvisational kind (March & Simon, 1993). Aspiration alone can spur organizational search for effective novel action patterns which, if proven successful, are the genesis of novel organizational capabilities.

March and Simon (1993, pp. 198–200) elaborate on this process with a ‘warehouse’ model that, while metaphorical in parts, represents a fairly well developed process account. The ‘machinery’ of organizational action – the programs and subroutines that generate action – are stored in memory. When whole programs are triggered, things work smoothly and proceed in a routinized manner. When unmet aspirations spur a search for novel action patterns, the processes unfold differently. Instead of whole programs, it is subroutines and sub-subroutines that are invoked, in a process “characterized by a great deal of ‘randomness’” (1993, p. 199). And much like a rat in a Skinner Box, in the tradition of stimulus-response psychology from which the idea of Simon’s production systems derived, operant conditioning (the reinforcing of successful action via ex post reward) allows for the slow determination of new courses of action and ultimately the elaboration of new programs of immense complexity, March and Simon (1993) argue. Similar evolutionary models of the origin of novel capabilities are at the core of approach pioneered by Nelson and Winter (1982; see too Winter, 2012, 2013 and Aldrich & Ruef, 2006).

The Lingering Problems:

Such accounts of change, innovation, and the origin of novel capabilities based on random mutations that occur on the glacial pace of evolution are hard to square with accounts of entrepreneurship that take vision, imagination, and judgement seriously (Buchanan & Vanberg, 1991; Kirzner, 1999; Chiles, Bluedorn, & Gupta, 2007; Foss & Klein, 2012; Zenger, 2016). Accounts which highlight the decidedly non-random, rapid, purposeful pursuit of opportunities that are as much created as they are discovered (Alvarez & Barney, 2007), are hard to reconcile. In section V, we will show how the deontic architecture solves this problem.

B Roles and role hierarchies

March and Simon (1993) also talk a good deal about hierarchy. And while this discussion is certainly congruent with their discussions of organizational capabilities, it should not be taken as identical. These discussions cover different territories. The theory of capabilities that March and Simon put forward attempts to explain how the macro-level phenomenon of routinized and programmatic organizational action is accomplished via an architecture of lower-level elements. It is a project in microfoundations (Felin & Foss, 2005, 2009; Barney & Felin, 2013; Winter, 2012). The discussion of hierarchies and roles (March & Simon, 1993; March, 1994) picks out a surface-level feature of organizations – the observation that individual members of an organization typically have relatively well-defined sets of responsibilities and tasks within the overall economy of the firm and that the individuals within the firm typically have a formally defined set of interactions and authority relationship with other members of the organization – and attempts to provide a functional explanation for why this should be so.

March and Simon (1993) suggest that hierarchies highlight three different aspects of the work done by organizations. First, they argue that most of the work done in an organization is accomplished by relatively small teams or groups of individuals who may work closely together in an interdependent manner requiring a fair amount of planning and close coordination but whose work is relatively independent and only loosely coupled with the work done by other individuals, teams, and groups within the economy of the organization. Simon described this kind of organization as ‘nearly decomposable’ (1999b). While the information flows and task interdependencies comprise a dense network within teams or groups, the flow of information and the task interdependencies between divisions is typically much lower and the interactions subject to more standardization and formalization.

Second, March and Simon (1993) note that there is a rather well-elaborated system of authority relationships, with individuals reporting to other individuals who are ‘higher up’ in authority – where authority is usually understood to involve both i) oversight and control functions, where individuals are evaluated and rewarded (or punished) for their efforts and contributions by a higher ranking individual, and ii) fiat, where an individual is agreed to forgo some measure of autonomy within the organization and accept that their activities (to a certain degree) can be directed by the a higher ranking individual (Barnard, 1938; Nelson & Winter, 1982).

Third, March and Simon (1993) argue, to a first approximation at least, that the orchestration of efforts within an organization reflects something of a means-end type relationship, where overarching goals are factored into subgoals and further sub-subgoals (and so on) and individuals are assigned to teams and groups responsible for the execution of specific deliverables within the plan represented by the means-ends decomposition of work. This is a reflection of what Chandler (1962) described as “structure following strategy.”

One explanation that March and Simon (1993) provide for the hierarchical nature of organizational work is that this way of orchestrating efforts is efficient (in a loosely understood way) given the bounded rationality of the individuals who comprise the organization. Individuals have limited knowledge, limited skills, and limited cognitive powers; there are limits to what they know and what they can accomplish, and factorization of work enables individuals to specialize and improve their effectiveness within particular task-domains. A second explanation is that this kind of means-ends organization of work parallels the more basic processes of problem-solving that are a basic component of human cognitive faculties. Organizations are arranged hierarchically, this argument asserts, because human problem solving, in general, is accomplished via means-ends decomposition of large complex problems into sub-problems, each of which can be tackled relatively independently.

March (1994) develops a complementary account of roles and role hierarchies in his discussion of the ‘logic of appropriateness’ in the context of organizational decision making. For March, a logic of appropriateness, where individuals make decisions on the basis of what is expected of them given their roles within an organizational hierarchy, is contrasted with a ‘logic of consequences’, where decisions are made on estimations of the outcomes of alternative lines of action as these impact preferences and/or utilities. March (1994) argues that much of the decision making actually observed in organizational settings is better explained by a logic of appropriateness than it is by a logic of consequences. This behavioral account is in contradiction to the traditional account of decision making in economics and decision science and while the processes involved in a logic of appropriateness requires “thought judgement, imagination, and care…. [this is] quite different from the processes of rational analysis (1994, p. 61).

The basic premise of March’s account of role-based decision making is that individuals within an organization make decisions based on the identities that they assume and enact within that context. And for March, the core of such identities can be expressed in the set of rules that the role-holder uses to make decisions. “When individuals fulfill … identities” he notes, “they follow rules or procedures that they see as appropriate to the situation” (1994, p. 57). And what are these rules? Not surprisingly, they turn out to be the same kind of stimulus-response elements that played a central role in March and Simon’s (1993) account of organizational capabilities: “The reasoning process is one of … matching rules to recognized situations” (1994, p. 58).

Roles matter, according to March’s (1994) account, because they are the basic element of hierarchical order. The features of hierarchy that March and Simon (1993) outline, in other words, are best understood in terms that are cashed out by a formal network of roles, where each role is assigned certain tasks within the overall economy of the organization and where each role fits into a complex of authority relationships and information flows that may be formal or informal and more or less temporary and ad hoc or permanent and routinized. March notes:

The Lingering Problems:

The weakness in the theory of role and role hierarchies is the same that we noted for the theory of capabilities – namely, the difficulty of providing an account of novelty, innovation, and the origin of novel roles and role hierarchies. March (1994) does suggest three different primary processes by which new roles arise and existing roles change. He suggests that: i) individuals can learn from experience and incorporate what they learn into future contexts, essentially rewriting the triggers and/or the programmatic responses that are encoded in the production systems that constitute the rules that comprise the roles they play in the organization, ii) that rules and roles will evolve in something like the processes of natural selection, where feedback from the environment impacts which rules and roles are carried forward into the future based on how they perform in past situations, and iii) that individuals envision and write roles for others as they plan for the future, in what constitutes a type of programming exercise.

The problem with these accounts is that, where they are forward-looking (as opposed to contexts where rules and roles change because of evolutionary processes) they fall into trouble trying to provide an account of how a new rule can be specified in the absence of a well-defined triggering condition or action pattern. If a production system can be specified, there is no problem. A new rule is just a new action program attached to a specified input trigger. But when either or both of those elements of a production system are vague or ill-specified, the account fails to provide a way to explain how a new rule comes to be specified. Despite this, new organizations – with new roles and role hierarchies – come into being all the time. And many of these involve organizations pursuing novel opportunities (Buchanan & Vanberg, 1991; Kirzner, 1999, Alvarez & Barney, 2007; Chiles et al., 2007; Foss & Klein, 2012) where it is not easy to see how the entrepreneurial action entailed in the pursuit of these new opportunities can be easily squared with production systems rooted in older production system specifications in any straightforward way (contra Winter, 2012) – especially when the pursuit involves the invention of new technologies, products, business models, or value chains.

How does this model attempt to account for the creation of de novo roles and role hierarchies? Even in cases where the production systems themselves would, in principle, be fairly standard – so that it is safe to assume that there were existing social models that could be used in the specification of the production system rules – it is not necessarily clear that the model outlined by March (1994) can provide the needed speed and flexibility to account for the fact that firms are quite adept at creating new roles ‘on the fly’. New teams and new startups can be launched quickly; there isn’t necessarily an enormous lag between the time when a group of individuals commits to a venture and when the individuals begin to operate within a set of structured roles and authority relationships – even in an ad hoc and ephemeral situations (see Barnard, 1938).

Evolutionary processes, where new roles evolve out of older models that individuals bring with them from prior experience, work too slowly to provide the kind of rapid deployment of roles and hierarchy that we see in these contexts. And the same is true for models based on experiential learning and socialization (Berger & Luckmann, 1966). What is still needed is an account of how the kind of rapid deployment of new roles and new rules in de novo contexts that March (1994) suggests takes place actually occurs.

C The firm as a nexus of contracts

Prior to Alchian and Demsetz (1972) and Jensen and Meckling (1976), firms were anomalous entities in a sea of well-structured microeconomic theory which described things like the coordination of resources in markets and the effects of competition on supply, demand, and profitability. The internal workings of firms – as the exceptions – were treated as idiosyncratic black boxes within which microeconomic theory did not reach. The internals of firms were instead described by management theory which was grounded in a view of the firm as a structured hierarchy of power relations. Management organized the factors of production, dominated the bureaucracies via authority, and justified its authority on the basis of organizational expertise, specialized knowledge (Bratton, 1989b), and entrepreneurial vision (Knight, 1921). Such an account is provided by deontic architecture, as we will describe in section V.

The nexus of contracts approach suggests that the firm is essentially nothing more than a set of interlocking contracts distinguished from the broader markets in which it is embedded only in terms of the density of interlocks rather than by some important qualitative distinction (Jensen & Meckling, 1976). This move away from the management-centered approach accomplished a number of important feats. First, for a century prior, corporate law had tilted between theories of the state-derived legitimacy and authority (where firms were seen as state-facilitated entities, serving the interests of the state); theories of hierarchy and managerial authority; and theories of contracts. The nexus of contracts view pushed things solidly toward contracts (Bratton, 1989a) as the primary ordering principle.

Second, it gave neoclassical economics analytical purchase inside the black box of firms where, it was argued, the very same principles of economics could be applied as in the market. The rational actors inside a firm were postulated to act just like the rational actors outside a firm, pursuing their self-interests in a utility-maximizing way.

Third, as the microeconomic wave moved “into” the firm it de-emphasized traditional conceptions of hierarchical power relations and authority structures. Instead, management was seen as an ongoing process of contract negotiation and formation with employees who choose what conditions they would associate with a firm and what conditions they would more or less cede decision authority to a manager (Barnard, 1938). From this perspective, an employee is not bound to the authority of higher-level management as much as they are a part of a freely designed, voluntary exchange agreement, a contract. Alchian and Demsetz (1972) were so comfortable rejecting managerial authority that they said the firm,

Fourth, the firm as a productive entity is itself de-emphasized as an important object in the economic landscape. The individual agent is the preeminent point of causation and theoretical discussion by virtue, primarily, of the agent’s contract-making which, again, provides traction on the firm as a standard subject of microeconomic analysis. While production was traditionally understood as the special provenance of the firm, nexus of contract thinking blurred the lines between the firm and the market, arguing that production was primarily orchestrated via contracts and that the firm was no special boundary for the these means of coordination.

If everything is a contractual agreement, explicit or otherwise, why bother arranging these contracts in a firm when they could be formed in the markets directly without a firm? The answer, as Coase (1937) noted, is that the formation and monitoring of such contracts is not without cost. One chooses a firm when the transaction costs of establishing such agreements in the free markets are greater than the costs of doing so inside a firm. But cost considerations endogenously vary with scope, scale, and technology, learning, and co-specialization among other factors. Firm boundaries, from this perspective, then, are the result of impinging forces and considerations rather than constituting some causal feature of the economic landscape on their own.

The Lingering Problems:

The contract ‘unit’ is not an especially tractable construct with which to theorize. Almost every contract is different. There’s little hope for comparing contracts without some more fine-grained unit of analysis. Understanding contacts as a formalization of relations between individuals that are engaged in utility-maximization makes it difficult to see how the goals of the firm find their place.

This brief overview above, though, is enough to appreciate the wide overlap between the nexus of contracts view and the deontic architecture view which can solve these problems. Contracts are, of course, essentially a collection of rights and duties that are described explicitly in formal contracts and implicitly in informal contracts. What contracts specify is precisely the obligations and rights of each party. A deontic architecture view makes possible a more fine-grained analysis of contracts because the unit of analysis, instead of being the contract, are the deontic components that in aggregate comprise the contract. This approach sees the contract itself subsumed into the broader network of deontics that structure and order the action of the individuals coordinating their efforts in pursuit of organization-defined goals. At the same time, deontic architecture provides for a reconstruction of what it means to order the collective efforts of individual efforts in ways that illuminate how authority is constructed and wielded within organizations. On one view, deontic architecture can be seen as a powerful extension to transaction cost economics. How exactly this works shall be explained in section V.

D Dynamic capabilities of the firm and organizational transformation

Organizations transform themselves in response to exogenous changes in the competitive environment and in response to endogenous decisions to pursue entrepreneurial opportunities. Nelson and Winter’s (1982) inquiry into the nature of routinized action suggested that this process of change and transformation – in some circumstances – could itself be routinized and institutionalized as an organizational capability of its own (see too Eisenhardt & Martin, 2000; Winter, 2003; Teece, 2012). In Winter’s (2003) formalization, the organization’s current routines and capabilities – the capabilities that it exercises when exploiting its current opportunity set (March, 1991) – are termed ‘zero-level’ capabilities. Dynamic capabilities are then understood to be the organizational capabilities that modify and transform existing ‘zero-level’ organizational capabilities or create de novo ‘zero-level’ capabilities.

Teece (2012) has suggested that it can be a mistake to think about dynamic capabilities in exactly the same way that we think about ordinary ‘zero-level’ capabilities. In particular, in contradiction to ‘zero-level’ capabilities, where the microfoundations are typically understood in terms of the knowledge embedded in actions of a large number of individuals in the form of the production systems or ‘roles and rules’ that dictate what is to be done in various circumstances, a firm’s dynamic capabilities are often tied to the idiosyncratic, ad hoc actions, beliefs, and visions of the individuals occupying positions of senior leadership within the organization. What makes for a dynamic capability in this perspective is not that it is itself embedded in a capability – as is the case for Nelson and Winter (1982) and Winter (2003) – but what it does, i.e., that it is able to generate change in response to exogenous factors or endogenous decisions to pursue new opportunities. In this respect, dynamic capabilities can be understood as the process account underlying organizational transformation more generally, in the sense outlined by Weick and Quinn (1999).

The Lingering Problems:

A now-familiar question presents itself here. How is the organization capable of instituting these broad transformational changes in its underlying ‘zero-level’ capabilities? In Weick and Quinn’s (1999) terminology, how is the organization able to ‘transition’ or ‘rebalance’ the focus of its operational capabilities in either episodic or continuous contexts? What are the mechanisms by which intention and vision are transformed into new action patterns in contexts where the change occurs in time frames that make it hard to conceptualize the process in terms of evolutionary or experiential learning processes? The dynamic capabilities literature makes the case that these kinds of transformational change can and do happen on a regular basis. What is needed is an account of the mechanisms that makes this possible. Deontic architecture provides it.

Deontic-mediated imaginative decision-making processes

How should one understand the role of deontics in decision-making? We have suggested that they play a role in choices that is somewhat analogous to the role of a ‘key premise’, but this needs some elaboration.

March (1994) argues, as described earlier, that there are two modes of decision-making. The first he terms the ‘logic of consequences’. In this model, decisions are understood as choices made on the basis of the anticipated outcomes of alternative lines of possible action, where the choice between the alternatives is made by examining how the anticipated outcomes affect the agent’s preferences or desires. As Nanay (2016) notes, this model is definitive of rational decision-making in the belief-desire psychology paradigm.

The second mode of decision-making in March’s (1994) framework is what he terms ‘the logic of appropriateness’. In this mode, agents are understood to make choices not on the basis of anticipated outcomes of future lines of actions, but rather, on the basis of how any particular line of action fits with the agent’s identity; in other words, on the ‘correctness’ of the action under consideration for the agent’s role.

Deontics shape an agent’s choices and actions in organizational contexts by outlining for the agent how their socially instantiated role is to be enacted in terms of specific sets of rights, permissions, duties, and obligations. It is in this sense that we talk about deontics as premises within decision-making processes. The deontics lay out for the agent a very specific set of obligations to perform certain kinds of actions in certain contexts and they provide for the agent a set of resources (in terms of rights, permissions, etc.) that the agent can avail themselves to as they seek to accomplish the objectives of the deontically determined lines of action that define their organizational role.

In the model we put forward here, deontic-based decision processes are not determinative in the same way as the kind of production systems (Lewis, 1999; Simon, 1999a) that underlie March and Simon’s (1993) model of organizational capabilities are postulated to operate. The core theoretical mechanism underlying the capabilities model is instead closely related to the traditional stimulus-response arc of behaviorist psychology (Simon, 1999a). The S-R process, even when understood to reflect the probability of a response (rather than the necessity of a response) given an input, is essentially a deterministic system. The decision-process model we adopt here, on the other hand, is based on a framework of imaginative decision making (Craik, 1943; Nanay, 2016).

In the imaginative model of decision-making, the agent is understood to simulate future alternative lines of actions and results that arise under the various envisioned scenarios and use those results to choose between the alternatives. As Nanay says, “when we decide between two possible actions, we imagine ourselves in the situation that we imagine to be the outcome of these two actions and then compare these two imaginings” (2016, p. 134). These simulations don’t compel action in anything like the way that an S-R type production system is postulated to determinatively evokes action sequences in response to input conditions. They don’t really compel the agent to behave in any way at all, really.

Our agents are rational and evaluative – in a boundedly rational sense – but free in the sense of having the ability to decide for themselves whether or not to engage in action on the basis of their strategic priorities (Swidler, 1986). This is so even when their action is bound, shaped, and channeled by deontics that are imposed or voluntary (McBride, 2015).

Deontics matter in agent decision-making because they provide the agent a set of guides for what is expected of them and a set of tools for helping them think through how to accomplish their directives in a given contexts. This is what we mean when we suggest that deontics both shape and channel an agent’s action. Deontics, in the form of duties (responsibilities, prohibitions, and obligations) and rights (the usable resources, authorizations, and permissions) avail themselves to the agent as critical ‘guardrails’ that determine available courses of action.

Deontics have granularity and specificity. Even individuals working in similar positions (defined by a functional role) can be subject to quite different specific deontic packages, differentiating the work they do from other agents within the organization. The specific ‘articulation’ of content matters.

The way that deontics impact decision-making, then, is by providing constraints, resources, and evaluative ideals that the agent can flexibly deploy in their goal-oriented reasoning. In this way, our model of the role of deontics parallels the role played by cultural elements, more generally, in Swidler’s (1986) model of strategic action. “Culture is not a unified system that pushes action in consistent direction” Swidler argues, but “rather, it is more like a ‘tool kit’ or repertoire … from which actors select differing pieces for constructing lines of action (1986, p. 277).

Deontics, then, play the role akin to premises in decision-making because they delineate how an action should be performed or what outcomes the agent should be targeting (March & Simon, 1993) while simultaneously articulating the means and resources that the agent can use as they go about their work. And they do all this by shaping and constraining the imaginative simulations that the agent uses to decide how to act in both proximate and longer-term contexts of action – in the ways outlined by Craik (1943) and Nanay (2016).

One final note, here, is that deontics shape reasoning in ways congruent with both modes of decision-making highlighted by March (1994). Deontics, in other words, define appropriate action for an agent occupying a specific role by articulating how an ‘ideal’ agent should behave and what outcomes the agent should strive to produce. At the same time, deontics can be understood within a logics of consequences as articulating the evaluative rubrics that the agent uses when deciding what line of action to pursue. That is, they sketch out for the agent an idealized template of the consequences of action that the organization seeks from the agent qua holder of a specific role within the overall economy of the organizations means-ends directed strivings.

How novel deontic architectures are created

It is not enough to be able to provide an account explaining how an existing deontic framework shapes and channels the strategic action of agents acting in organizational contexts. Organizations operate in a constant state of flux and transformation (Nelson & Winter, 1982; Weick & Quinn, 1999; Tsoukas & Chia, 2002) and new organizations are created in the pursuit of entrepreneurial opportunities (Foss & Klein, 2012). We must also be able to explain how new deontic architectures are created and how existing deontic architectures are transformed.

At the core of our approach is the idea that deontics are fundamentally cognitive and that they are created – and in some sense constituted – discursively. The discursive foundation of deontics has two aspects. First, agents understand what they are supposed to do and how they are supposed to act in any given organizational context in light of the set of responsibilities, obligations, rights, permissions, etc. that define the specific role(s) that they are to enact. These deontic ‘rule sets’ are not constituted by stimulus-response (S-R) type production systems of the kind suggested by Simon (1999a; see also Lewis, 1999). Rather, they are best understood as something that is formulated and transmitted primarily through language, conversation, narrative, and other symbolic or semiotic modes (Searle, 1969, 2010; see too Ford & Ford, 1995; Ford, 1999; Phillips, Lawrence, & Hardy, 2004; Weick et al., 2005; Lawrence & Suddaby, 2006).

Second, arising in conversations, deontics emerge as the settled formulation of what the agent needs to get done and what social resources the agent can avail themselves of as they go about their coordinated organizational work. As such, they are better understood as vague, flexible, and fluid, constantly adapting to the changing contexts of the work and the needs of the actors to get work done and figure out what needs to get done (Weick et al., 2005; Felin & Foss, 2009; Salvato, 2009). But even more importantly, they should be understood as literally talked into being by the agents themselves as they go about their work in organizational contexts.

Deontic architectures – and the roles and rules that constitute them – are not in any simple or direct sense, then, to be understood as determinate action grammars (Pentland & Rueter, 1994), production system-style programs (March & Simon, 1993), procedural memories (Cohen & Bacdayan, 1994), or action generators (Starbuck, 1983). They are, instead, best thought of as something like a web of stories – told and retold, woven and rewoven over time by the agents themselves – about how to act in the context of their organizationally defined pursuits. Stories that are told as ways of shaping the actions of the agents as the agents autonomously and strategically decide on what needs to be done and how they will attempt to go about accomplishing what they are pursuing (once they have, indeed, chosen to pursue a collective goal (Barnard, 1938; Nelson & Winter, 1982).

Unifying theoretical accounts of the firm

The deontic architecture framework introduced in this paper provides a unified account of the firm. In this section, we show how the framework can provide an account of the core features of: a) organizational capabilities; b) roles and role hierarchies; c) nexus of contracts approaches; and d) dynamic capabilities. We also show how the deontic architecture framework can provide insight into those aspects of these four areas that have not been well accounted for in existing theoretical treatments. And we emphasize throughout this section, in particular, how the deontic architecture framework provides a more natural account of dynamism, transformation, and origin – aspects that have sometimes been less well theorized in existing treatments of these fundamental organizational phenomena.

A The programmatic view of organizational capabilities

Deontics shape and channel the behavior of an individual by providing them with a set of resources that they can use to reason through situations. They form, as Swidler (1986) suggested, something of a flexible toolkit the agent can draw upon when faced with the problem of determining what exactly is going on at any point of time and deciding upon what line of strategic action they need pursue if they are to simultaneously meet their socially determined obligations and achieve their collectively determined and mandated goals. The deontic bundle specific to an agent, in this sense then, articulates the key premises for a role-defined logic of appropriate action. At the same time, by specifying specific, particular, and concrete (granular) role-specific outcome desiderata, this same deontic bundle supplies the delineation of goal-states that the agents strive for as they engage in their organizationally defined roles. The deontic bundle shapes the agent’s logic of consequences, channeling action towards organizationally defined outcomes. The deontic bundle, in this way, motivates and directs the behavior in both modes of decision making – the logic of appropriateness and the logic of consequences. And it does so at simultaneously.

As noted, one of the weakness of the capabilities approach, as developed by March and Simon (1993) and Nelson and Winter (1982) has been finding an account of both routinized and improvisational modes of action in organizational contexts. In March and Simon’s account, for instance, routine organizational action is ultimately cashed out in terms of performance programs built on something like a S-R production system architecture while improvisational and innovative action is explained by recourse to the processes of action search within an economy of exogenous consequences that guide learning at the level of the individual and the group. Not only are these two modes of action predicated on what amounts to two quite different kinds of cognitive processes, but the account suggests that improvisation and innovation are essentially random. While acknowledging that organizational action can sometimes be myopic, it is important to remember that it is often purposeful too (Wiltbank, Dew, Read, & Sarasvathy, 2006).

The deontic architecture framework provides a simple account for both modes of action. In both contexts, an agent’s actions are guided by the specific content of their deontic bundle – by the specific articulations of just what exactly the agent’s obligations are in particular contexts and what exactly are their responsibilities. While often vague and sometimes even ambiguous to varying degrees, these outcome desiderata and processual norms help the individual decide how exactly they should and could act as they pursue their role-specific objectives (cf., Ouchi, 1979; Simons, 1994). These deontic bundles anchor the agent’s reasoning, suggesting strongly that certain ways of acting or certain outcomes are valued and/or even mandated by the organization in particular contexts. In this way, the deontic bundle actively shapes the agent’s understandings and expectations about just what it is that is required of them if they are to be successful in an the role that they are living (cf. King et al., 2010). And it is these understandings and expectations – as articulated in the agent’s deontic bundle and unpacked by the agent in the social contexts of ongoing organizational action – that we argue guides the agent’s actual choices and actions in both more routinized and more improvisational and innovative contexts.

B Roles and role hierarchies

In the deontic architecture framework, a role is understood as essentially a bundle of deontics that attach to an individual agent in an organizational context. This means that roles are created when a collection of specific, particular, and granular deontics are assembled as a collective bundle that is assignable to an individual. In principle, this means that roles exist independently of the individual agents that inhabit them. A role can exist even if it is never filled. Likewise, a role is not necessarily somehow diminished, altered, or transformed just because an agent fails to uphold any specific deontic element or acts in contradiction to a specific deontic. A deontic role, in other words, exists independently of any individual agent who might hold that role or the agent’s success or failure in that role.

Deontics define roles by defining an interlocking set of rights, responsibilities, obligations, permissions, etc. that the role occupier is expected and intended to further in the context of the ongoing activities of the organization. As the specific content of the deontics change, the role itself changes. The actual change can vary from incremental, or adaptive, that may not even be recognized, to fairly radical and transformational changes to the basic charter, causing significant rewrites of the deontic roles of related agents as a result.

The deontic architecture framework suggests that role hierarchy creation can be best understood as the chartering of an interrelated collection of roles, each of which is created via the discursive articulation of specific deontic bundles. This chartering is sometimes primarily a top-down affair, where the senior leadership delineates deontic bundles in response to the need for a division of labor and a network of authority relations within the organization. But deontics are often established from the ‘bottom-up’ and from the ‘outside-in’ as an agent goes about establishing how they will go about their organizationally relevant pursuits in coordination and interdependence with other agents in the organization. Nelson and Winter (1982) described this feature of role articulation as a ‘truce’ while Barnard (1938) spoke of ‘zones of indifference’.

By cashing out roles and role hierarchies into bundles of deontics assignable to agents, the deontic architecture framework suggests that ‘organizational becoming’ (Tsoukas & Chia, 2002) and sensemaking (Weick, 1995b; Weick et al., 2005) are both reflections of the same underlying exercise to some extent. In each case, the goal is to articulate a set of guides for action that the agent can use when determining their strategies. By framing this process as a discursive articulation of a bundle of specific, particular, granular, and role-specific deontics, the deontic architecture framework begins to cash out what it means to create a new role or role hierarchy or to transform existing roles and role hierarchies in a concrete and actionable manner.

Adam Smith opens The Wealth of Nations (1776) applauding the radical improvement in productivity that role specialization provides: “The greatest improvement in the productive powers of labour … seem to have been the effects of the division of labour.” We now have the ability to say what exactly division of labor is. A role is a specific collection of rights and duties. A division of labor is a division of roles. Productivity occurs when a diversity of rights and duties that might otherwise occur in the same person are divided into multiple roles across different individuals enabling gains from the learning attendant to specialization. The productivity of each role is enhanced by specialization and when such roles are combined under a single firm to an appropriate aggregate the result is radically enhanced productivity. A role just is a specific collection of rights and duties and a role specialization is, then, the unique assignment of rights and duties to determine unique roles.

C Nexus of contracts

The nexus of contracts theory suggests that a firm differs little from the markets in which it is embedded in that both are essentially collections of contracts and both are analyzable through microeconomics. The arrival of the theory had important implications for law, economics, management, and strategy (Bratton, 1989a).

The nexus of contracts view is not without its criticisms, of course. A number of problems, for example, arise from that fact that contracts can be both imperfectly executed – one party may “shirk” its responsibilities (Alchian & Demsetz, 1972) – and the fact that one can never cover every eventuality in a contract which results in ambiguous areas in the contract and also “residual rights” that remain outside of the purview of that (or possibly any) firm contract. Every contract is incomplete with regard to some concern.

Deontics are, as suggested, the very “atoms” of any contract. Rights, responsibilities, obligations, and permissions are exactly what are made explicit in a formal contract in order to coordinate both parties’ behavior and expectations. And they are exactly what remains implicit in an implicit contract. Deontic architecture, then, is not so much an alternative as it is means of providing greater granularity and traction for the nexus of contracts view.

This more fine-grained granularity means that some long-standing problems in the nexus of contracts view can be addressed by deontic architecture. One problem is the inability of the nexus of contracts view to denote many of the agent-guiding structures that exist outside of a contract between two parties. There is a vast panoply of expectations – implicit obligations, promises, duties, or expectations of past behavior – that fall outside of the realm of contracts proper, even informal contracts. Insofar as deontics exist at the root of such structures, they can be individuated, described, and made tractable.

The second problem is the tension between the nexus view and the classic theory of management structured around authority and control hierarchies. A full discussion of power and control hierarchies from a deontic perspective is beyond the scope here⁵ but the relevant implications flow quickly from one basic fact about deontics: one person’s right is another person’s duty. The CEO has the right to restructure the R&D division which means that the employees of the R&D division are under obligation to move their equipment, cancel research projects, eliminate costs, or hire hundreds of more employees to redouble efforts on a project, at the discretion of the CEO. In general, as one moves up the hierarchy the employees have more rights and as you move down the employees have fewer rights. And in general (in a hierarchical organization), obligations aim upwards to top-level management and the board of directors, and rights aim downwards. It is easy to run through exercises analyzing the rights and duties of any given role in an organization. Hierarchies are essentially composed of rights and duties that shape the structure of the hierarchy, even if the formal contracts between employees and the firm specifies no hierarchy. Deontics allows you to specify the managerial authority hierarchy in a way that neither nexus of contracts nor classical economics does.

The third problem is the inability of the nexus view to explain the transition from an entrepreneurial venture to a firm. The interesting thing about this transition is that the structure of the deontics moves from one where there are agreements interconnected and distributed among the founders to one where all the internal agreements instead now point from the individual agents to the firm (McBride & Westgren, 2018). There are at least three reasons why this phenomenon is difficult to approach from a nexus of contracts view. (a) The deontics – the obligations, duties, rights, etc. – that structure the relations between founders are often not contractualized. (b) Many of the deontics (and in ventures with one founder, all of the deontics) are self-imposed. Unless the nexus view is willing to entertain the possibility of contracting with oneself it cannot explain the self-imposed duties that characterize motivated founders and the overwhelming bulk of their behavior. What’s a self-imposed duty? Steve Jobs committing himself to building the first consumer personal computer (and then enrolling others into that vision), or Elon Musk committing himself to building an electric car company. To commit oneself is, of course, to impose a duty upon oneself.

Fourth, there is a gap in the discussions of the nexus view and residual control rights. Returns that are not contractually pre-determined and fall outside of established contracts (with, e.g., debt holders) fall to the owners of the firm (Stout, 2012). But the owners also get discretionary control over the deployment of the firm’s assets and human capital (Foss & Klein, 2012). These two aspects of ownership do not, however, exhaust the rights that accrue to the owners of the firm and management teams to which they delegate day-to-day control. Ownership also entails the ability to determine deontics. Insofar as deontics are the most important means of structuring the interconnected flow and shaping human behavior, control rights over the firm’s deontics is arguably one of, if not the most, important components of the firm’s modes of behavioral control.

One overall advantage of this deontic architecture approach is that it provides a theory with all the advantages of the nexus of contracts, but with a more fine-grained set of explanatory elements and so is essentially an explanatory “super-set” relative to the nexus view. The rights are the “atoms” for the molecules-as-contracts.

D Dynamic capabilities of the firm and organizational transformation

Dynamic capabilities can be understood as the mechanisms by which the organization deliberately and purposefully transforms its underlying functional architecture – understood in terms of capabilities, roles, and role hierarchies. What distinguishes the kinds of transformation that are commonly understood as the result of dynamic capabilities versus those more incremental and local transformations that are characteristic of adaptation and change (cf. Felin & Foss, 2009; Salvato, 2009) is arguably the scale, coherence, and simultaneity of the changes. Local, adaptive, and evolutionary change can arise from problem-solving by individual agents in non-central leadership roles in response to exogenous circumstances or purposive attempts to ‘do better’ or ‘do differently’ – but these rarely cascade far enough across the organization to constitute a global reconfiguration of activities or purpose. What distinguishes a dynamic capability from these more local and adaptive forms of change is that the former are wrought by the organization’s senior leadership in a deliberate attempt to make just these kinds of ramifying structural reconfigurations of activity and purpose.

The deontic architecture framework suggests that these kinds of broad-based structural changes can be made, effectively rewriting some aspects of the deontics common to large collections of roles within an organization en masse via the discursive efforts of the organization’s senior leadership (cf. Ford & Ford, 1995; Ford, 1999; Lawrence & Suddaby, 2006). The capabilities of the organization at any point in time – its ‘zero order’ capabilities (Winter, 2003) – constitute an organization’s production set (Nelson & Winter, 1982). Rewriting key elements of the deontic bundles held in common by organization as a whole (or significant subsets of the organization) provides senior leadership a tool for ‘deforming’ the production set in its entirety. Different ‘rule sets’, which is to say different deontics, entail different capabilities. Deliberate efforts to change components or elements of the overarching, general obligations, responsibilities, duties, and rights, etc. held in common across the organization or important subsets of the organization, then, is a means for enacting the kinds of transformational changes that are definitive of dynamic capabilities.