Introduction
Business historians scarcely speak to their largest potential audience. In the United States, they appear to speak mainly to other historians, a relatively small and somewhat inward-looking community. This was once also so in the UK. Now UK business historians increasingly try to speak to other (social science–style) management academics (either directly or through editors and referees). This second community is also relatively small and inward looking, and it has further potential disadvantages deriving from fundamental differences between historical explanation and social science explanation.Footnote 1 One might have thought that there was a substantial potential audience in that part of the general public that reads history books. However, those individuals in America appear to be chiefly interested in the Civil War, World War II, and biographies—not the most encouraging signs. Business historians could, however—if we wanted to and if we approached the task thoughtfully—also speak powerfully to business people and, perhaps more to the point, to people studying to be business people. The numbers in this segment are large, and the individuals, already exercising responsibility or anxious to do so and to do so wisely, represent a much more likely prospective clientele.Footnote 2
This paper points a way to that outcome via consideration of the idea of action, an idea implicit if not often embedded directly in business history writing.Footnote 3 In a technical as well as a metaphorical sense, the robber barons were in a position to rob, so to speak, only because they had positioned themselves to hold up their customers.Footnote 4 The role of industrial statesman was only open to someone who could command negotiation. However, the analysis of these sorts of opportunities does not generally go much deeper than the facts that make them possible. Even in Flynn’s strategically quite perceptive biography of Rockefeller, a curtain descends when the question of the origin of the ideas comes up: There is at least speculation as to the origin of the South Improvement Company scheme, for example, but no sense of what other means to a similar end might have been contemplated or of the process through which that idea came into focus and came to be embraced.Footnote 5 Flynn’s Rockefeller thinks things through, basically keeping his own counsel, before acting. However, there is no real weighing of alternatives in Flynn’s account, only the observing of a status quo Rockefeller could change in his private interest and a description of the means he eventually chose of improving it. This is a history of a past leading up to a very particular present. In Chandler’s Strategy and Structure (1962), the actions are decisions of intrafirm decision-making bodies about what the company should do and how to organize control of and responsibility for it: both strategy and structure represent choices, whence, in the event, actions occur.Footnote 6 The administrative challenges are confronted and the formal steps of decision-making concerning what to do in the book’s four case studies of Du Pont, General Motors, Standard Oil, and Sears Roebuck are clearly laid out. That said, the text focuses intently on what was not working and on the proposals for what to do about that which were accepted. Its treatment of the range of possibilities that were or might have been considered and how those who had to make a choice between them might have thought about the alternatives is much more cursory.Footnote 7 In The Visible Hand (1977), Chandler develops the earlier book’s substantive themes on a much larger canvas, but there is a considerable air of Whig history about what is narrated.Footnote 8 In his subsequent international comparative study Scale and Scope (1990), the key decisions that emerge are investments in manufacturing, marketing and distribution, and management.Footnote 9 The focus is national patterns in these relatively overt and easily documented corporate actions.
Questions have been raised as to what the evidence invoked in Scale and Scope actually represents and whether that evidence is being interpreted properly.Footnote 10 These amount to what is called in other contexts questions of experimental design—whether in this book Chandler was, consciously or more likely unconsciously, not conducting a real test but rather, in effect, looking for evidence that would confirm what he thought to be true. However, the key aspect from the present perspective is not the epistemological status of the evidence presented but rather the conceptualization itself: Action was undertaking a particular investment program or not. Exposition of action might cover more than just an outcome—there might have been real discussion in board meetings about whether or not to proceed, debates about whether even to propose the question within upper management, and so forth. One could take this point even further. The genesis of the program idea itself appears to be of much less interest than what becomes of the program once it has taken on something like its final form. The question I want to raise here is whether such an analysis goes deep enough.
Going deeper requires confronting agency. Historians are, of course, familiar in at least a general way with the concept of agency. However, I will argue that it pays to dig down quite far into the concept and, once that is clear, determine what studying it in a business historical setting might entail. To exercise agency is, whatever else, to act. What it is to act is a more complicated matter than might initially be apparent. A long philosophical literature going back at least to Aristotle has struggled with the question.Footnote 11 One key strand argues that to be an action is to have particular kinds of causal antecedents. The version of this with which I will engage in this paper argues that reasons can be causes.Footnote 12 Another key strand does not deny that claim but argues that the difference between “what an agent does and what merely happens to him” lies not in the causal antecedents but rather in the presence or absence of the possibility of a kind of in-process readjustment known in the literature as “guidance.”Footnote 13 In the settings with which I will be concerned, guidance mechanisms involve choosing between alternative courses of action (one of which might well be doing nothing, that is, just continuing as before). Considerations such as these may strike the reader as very abstract; but as the paper proceeds, their salience to the writing of business history—to explaining, or at least shedding light on, why events taking place at an intraorganizational level proceeded as they did and not in some other way—will become evident.
The first section below considers how the idea of action is treated in the social sciences, the second in history writing. For the most part, the discussions considered in these sections concern either isolated individuals or large and not necessarily particularly organized groups. The third section draws out the significance of the fact that the subject matter of business history is, for the most part, behavior of or in a formal organization or in a group of such organizations. In the course of this section, questions about the relationship between structure and agency come into sharp focus. These questions are the subject of the fourth. The discussion of them is conducted at a relatively high level of abstraction. Making the best current response to the questions practically helpful to historians requires a deeper probing of the idea of agency in itself; this is the business of the fifth section. The probing exposes clear implications for the conduct of studies of the history of action in organizational life and so for business history more broadly. The sixth section concludes the paper with a statement of what these are.
1. The Idea of Action in the Social Sciences
Much theoretical discussion of action, both in philosophy and social theory, has concerned isolated individuals.Footnote 14 Any analysis is inevitably more complicated when groups of individuals are the subject. There is a natural temptation to think of groups as nothing more than a collection—the sum, so to speak—of the participating individuals. However, the common, impulsive equivocation on the notion of sum (as above) betrays an unavoidable sense that groups may be something larger and more complex. This idea has roots in the literature of sociology down to the earliest days of the discipline.Footnote 15 But even consideration of how the ancient Greek hoplite armies were different—in the field, in their phalanxes—from their members in their armor and with their weapons, on their way from their homes to their first muster makes one wonder.
The subject of the social sciences as a general matter is human behavior.Footnote 16 Most of psychology aside, however, all the social sciences are ultimately concerned with group-level rather than individual-level phenomena.Footnote 17 Action is a profound problem for social scientific students of isolated individuals. It is even more of a problem, I would argue, for students of groups and of individuals embedded in groups in some essential way. The difficulty is that individuals embedded in groups and cultures behave differently from the way they might behave if not embedded. These differences are a suitable subject for study. That said, it would be an extreme view to think that the exercise of individual agency is somehow absolutely incompatible with the idea of an ambient culture or other sorts of structures conditioning or otherwise constraining agency’s free play.
It is worth returning briefly to the example of conventional economics. Economics represents an extreme instance among the social sciences in having a single, widely accepted theory as its basic organizing principle. However, that may make it a helpful example from which to start. Economic analysis, particularly in the past forty-odd years, has aimed to provide an internally consistent, thoroughgoingly theorized account of large swathes of human activity.Footnote 18 The usual interpretation by students and their teachers (if not perhaps the theorists) is that the account of individual action and its counterpart concerning firms are there principally to provide a clear, relatively unambiguous, and easily manipulated foundation for the main action of analysis of market-level supply and demand.
The other striking, deeply rooted, and completely explicit feature of almost all of the presently admired corpus of economic analysis is that it is focused on and, more broadly, oriented around systems’ equilibrium states. This would not miss out on much if equilibrium tended to be unique and stable (or if equilibria were multiple but all were stable) and if adjustment processes were such that perturbations pushing the system away from equilibrium died out quickly and relatively costlessly. However, many economic systems do not have unique equilibria.Footnote 19 Multiple equilibria are typically not all stable, and in any case, it is readily apparent that the welfare properties of the individual equilibria might differ. That means that being part of a system stuck in a bad equilibrium might be a very bad thing; and there is no particular reason, extreme abstractions aside, to imagine that after a perturbation, the return to a local equilibrium is swift. (Consider the course of the Great Depression of the 1930s.) The fact that much of quantitative social science has a close counterpart to the economists’ focus on equilibrium states is a little disquieting.
Economists often argue that the thoroughgoing optimization on the part of individual consumers and firms—the latter term encompassing both actual and potential entrants—drives the system to equilibrium and makes the simple equilibria stable. However, thoroughgoing optimization is a very extreme assumption. It portrays firms as completely transparent to themselves, masters of relevant information, and effective machines for computation and analysis. In all of these respects, this portrait does not sound much like the overwhelming majority of actually existing firms. Consumers, in a similar sort of portrayal, are assumed to know their own preferences; be effortlessly effective in figuring out what courses of consumption would make them happiest; and, without any sort of weakness of the will, relentlessly pursue those courses. This, too, will seem unfamiliar. No one thinks these assumptions are realistic, of course. These assumptions are put forward as mere, helpful abstractions—“helpful” meaning, at least in the first instance, something like “tractable.”
Tractability in a theory is good, all else being equal. However, is all else equal? One must ask at what cost the tractability is purchased. This is not just a matter of the modeling of the individual decision-making units with the issues itemized above: Methodological individualism is routinely regarded as another costless assumption in this discourse, sweeping away nothing of any great significance. From that, one also gets the notion that there is nothing to be considered but the isolated individual production and consumption units, each so small relative to the relevant “market” that its own actions have no first-order effect on market supply or demand, and so no direct effect on any other unit’s choices or outcomes. There is no consideration of other ways in which individual units might affect or be affected by others. As I suggested above, the other ideas being invoked—the efficiency of market processes—are also not at all obviously innocuous theoretical commitments. The potential for serious distortion, particularly at the system level, should be obvious; and indeed there are large facts that this approach has some difficulty explicating without seriously ad hoc adjustment.Footnote 20
Stepping away from economics toward less deeply theorized social sciences, it will be obvious that individuals in modern society live their lives in groups and collective organizations small and large. These groups sometimes have explicit objectives and codes of behavior; but there is almost always some sense of norms and routine. (As Sidney Winter remarks, there are very few organizations of which one could never say, “That sort of thing would never happen here.”) The most casual reflection will suggest that such norms and routines might have some influence on how individuals decide to behave. Even taking such elements of structure as given exogenously and immutable, the extent to which they actually do influence, never mind fully determine, the actions of individuals appears to be an empirical question rather than a matter for general theory.
Early sociological writers such as Tönnies, Simmel, and Durkheim were concerned with changes in the character of life and society in Europe that had, by the late nineteenth century, as industrialization and urbanization progressed, disrupted many earlier social (and indeed political) formations.Footnote 21 It is not as if individuals are absent from their discussions—quite the contrary. Individuals are inevitably part of the story, in one way or another, of what can be observed of aggregates of individuals. The deeper point is that these authors seem, at least when read in retrospect, to be starting from top-down rather than bottom-up images of collective life.Footnote 22 So, although one could say that there is an interest in action (generally viewed as embedded in a social context) at the heart of this first real flowering of sociology, there is retrospective projection: For these earliest writers, the social came first.Footnote 23 Methodological individualism as a project in itself in the social sciences came rather later and initially through a different community, though the idea certainly did in due course get a major push from Max Weber.Footnote 24 Talcott Parsons began his own long career of theorizing about social action with a monographic analysis of individual action. His emphasis in that was on what he called the “voluntaristic” aspect (in which “voluntaristic” should be read as meaning something like “not strictly or mechanically determined by external conditions”).Footnote 25 But one way of understanding Parsons’s thinking as his career progressed is to observe that the scope of his ambition declined. Structural functionalism, for which he is chiefly now remembered, is what took the place of a social theory with aspirations to complete microfoundations. Structural functionalism is fundamentally an explanation in terms of an equilibrium-oriented account of the macro-level relationships between institutions. In the end, it represents a shift of focus to a higher level of relationships and their dynamics and away from individual actors.Footnote 26 The alternative perspective on Parsons is that he never really had thoroughgoing microfoundation to begin with: Even in The Structure of Social Action, individuals are to an important if incomplete extent subservient to the collective via the internalization of social values (“norms”). Both views are ultimately unsatisfactory because both are ultimately question-begging.Footnote 27
The idea that these apparent dead ends derived from a defect in the underlying conceptualization is now most widely associated with the writings of Anthony Giddens and those who have engaged with him. I discuss both in more detail in section 4 below. For present purposes, I need only say that the argument Giddens and the others make is that structure and agency—the latter meaning roughly the ability of individuals to act, at least as it might strike them, of their own volition—do not exist independently of each other. Structures change over time, and in particular, the accumulation of individual actions can reform, over time, taken-for-granted elements that might appear to be exogenous context in any moment. This may appear to be a destructive sort of conclusion, but in the context of what I have been writing here, it represents something more like an invitation. This analysis suggests that students of social action in organizations should focus on what happens and how whatever is done comes about. They should, while doing that, try hard not to lose track of individuals making sense of what they might do and what the consequences of those alternative courses of action might be, or of how those individuals’ sense of what is normal and expected might change over time. This brings us to history.
2. The Idea of Action in History Writing
The objectives of the social sciences and of history differ in fundamental ways. The most central is that social scientists are in search of relatively robust, broadly relevant generalizations, whereas historians seek to understand why some specific sequence of events, or group of sequences of events, developed as they did.Footnote 28 One looks outward from singular instances while the other looks, so to speak, inward. Historians are pleased when their interpretations shed light on situations beyond the ones they were studying, but this is absolutely not key to their sense of success in accomplishing their primary objective.
These considerations have implications for the matter under investigation here. The social scientist needs evidence that is in some relatively strict way comparable across the cases being studied. In quantitative social science, this almost always requires measures external to the actors. Such measures can be more or less close to what a genuinely actor-level explanation would call for examining, and are often, in fact, rather far from it. The measured variables are referred to as proxies or constructs when the distance is conspicuous. There is at best only rarely any extended analysis of the potential for errors-in-variables problems in the statistical analysis that is the heart of these papers.
Most historian are not primarily interested in central tendencies at all.Footnote 29 Most historians are interested in a regression’s epsilons. Although all the social scientists care about is that these “error” terms have the statistical properties required for meaningful interpretation of the calculations, historians care about why these errors fall where they do, why some particular ones are large and others small, how these errors might develop in nonrandom (i.e., path-dependent) ways over time, etc.Footnote 30 This sort of impulse is what leads the historian to want to understand all the concrete details of particular settings, particular actions, particular actors, and the world they regarded. That leads directly to a concern with actors’ intentions, and that in turn to a concern with their understandings in the moment of both what the facts are and what the consequences of potential courses of action on their own part might be. This, in turn, leads researchers to whatever sources might shed light on such considerations and, because of the complex mechanisms and motives potentially at work in what gets recorded and in what of that gets preserved, onward to practices such as source criticism, triangulation, contemplation of research subjects’ backgrounds and evolving worldviews, and the like in which professionally trained historians are instructed from their first day in graduate school onward.Footnote 31
These are problems for historians in general. Historians of business enterprises face them in a notably acute way. Most firms face resource constraints far more crisp than any politician, author, or other public figure. If firms run out of money (including credit, of course), they have to fold their tents. Although I will argue below that the day-to-day life of firms is mainly a matter of routine, particularly in the operating rather than the executive parts of the organization, it is nonetheless the case that there are usually executives within firms who are explicitly responsible for certain sorts of decisions. They have to either act or refrain. They can fail to recognize alternatives or even fail to grasp that there might be any.Footnote 32 But making decisions or refraining from them is their job; and their judgment on such matters is often subject to scrutiny in a way that the judgment of members of the organization further down the hierarchy is not.
It is tempting, once one’s natural sympathy with one’s protagonists begins to sink in, to see analyses of such judgments through the eyes of these individuals, perhaps even in retrospective self-congratulation. The great difficulty, as one gets down into those weeds, is to distinguish structure, at least in the moment, from agency in order to assess, in a concrete case, how much of the outcome owes to the conscious intentional actions of individuals that might have been other than as they were and how much was an essentially unalterable consequence of circumstance and external constraint. Any sort of evaluation of the role the actors played is inevitably dependent on some assessment of what they did not control or even know. Giddens could well be right despite it still being the case that in the moment, not everything is possible (or even imaginable).Footnote 33 As suggested above, solid evidence about what was, for practical purposes, imaginable can also be quite illuminating.
Two more points are worth adding. One concerns the phenomenon of sample selection bias, that is, the disappearance from the evidence of the body of survivors of what happened to firms that withdrew or otherwise failed. A picture of what could have and did happen, what factors influenced that outcome, and to what degree they did, which considers only the survivors will certainly get the proportions of effects wrong and may well get any notion of how reliable the estimates are wrong too. We are conditioned by efficiency arguments, whose assumptions are not spelled out very clearly and are sometimes wrong in positively misleading ways. There is an exact counterpart in historical inference. A less statistical way of putting this point might be to say that the conditions for survival and even prosperity, at the end, may not have been the conditions all along. To understand a race and not just learn who won but how they won it, you need to watch it from the start, not just scroll backward from the finish until you lose interest.
This point has a deeper interpretation. There is a great deal to be learned from failure. Certainly, knowledge of possible failure modes is helpful in understanding industry dynamics. However, there may be a deeper connection still. Insiders, in my experience, know far better than outsiders how things might go wrong. (They may not always canvass potentially promising possibilities sufficiently broadly. However, they often have very thoughtful views on unpromising ones.) Such knowledge may condition, through its categories and through the sense of the likelihoods of various contingencies, what possible courses of action even get considered and which ones do not.
I should summarize the discussion so far. Action is treated rather differently by philosophers, many social scientists, and many historians. The philosophers are on to something. Reasons can be causes, and if conscious reasons can be causes, unconscious (or understood if only half-articulated) reasons can be causes too. Process can also matter. The sharp edge of the argument of this section is that understanding decisions is only in rare cases strictly a matter of understanding the explicit evaluation of the merits of specific and concrete alternative plans.Footnote 34 Additionally, there is sometimes as much interest to be found in what does not rise to the level of demanding a decision and in what does not get considered when a decision does have to be made as there is in the mechanics of decision-making among explicitly articulated alternatives. Furthermore, none of this speaks to the obviously very great importance of the question of where new ideas, particularly new ones that take over whole industries and technologies, come from and how they get into consideration and get implemented.
It is not difficult in specific instances to identify the institutions and other even more abstract constraining forms, the routines, and the individual habits that are a proximate cause of this sort of thing. Few social sciences go as far as economics in taking these as exogenous, more or less strictly parametric to decision-making processes. However, even if they are exogenous in the moment, there is clearly often change in their content over time. The relationship between individual action (the exercise of agency in many usages including the one discussed at some length below) and structure (construed in one way or another) and the ways that relationship and the contents of each of these elements might change over time is of obvious interest in many fields and has attracted the attention of some profound thinkers. Some of this, I will argue, is useful to business historians.
3. Businesses as Organizations
To see how this might be, one needs to take seriously that the subject matter of business history usually involves in some way an organization or a group of them. One might imagine that this could introduce additional considerations into the analysis of action.Footnote 35 What then might these be?
One place to begin exploring the significance of viewing businesses as organizations is the literature of organizational theory. The antecedents of this literature date back into the nineteenth century. However, the managerially oriented among these precursors were not focused on organizations as such.Footnote 36 The subject only truly emerged as large-scale organizations became not just a conspicuous feature of the social and economic landscape but an accepted and naturalized one, say in the middle third of the twentieth century.Footnote 37 Theorizing about why organizations formed and developed as they did began not from some hypothesized microfoundations but rather with what there was to observe.Footnote 38
In this context, it can hardly be surprising that one of the earliest modern analytical schemes was contingency theory, an argument that complementarities between particular institutional arrangements often made for higher levels of overall performance. Product market competition, or perhaps intrafirm optimization (itself perhaps owing ultimately to pressure from shareholders), were of less importance. The early empirical work in this vein was based on observation and surveys of existing firms rather than theorizing from explicit assumptions.Footnote 39
If one does not start out with universal efficiency as an assumption, one looks to evidence on the ground. Globally efficient outcomes do not emerge from that process as seeming very likely. Early empirical findings along these lines prompted a school of organization theorists to explore what might be true of firms in settings in which thoroughgoing optimization was not the beginning and end of explanations.Footnote 40 It had been well known for years, for example, that firms did not behave as profit maximizers in pricing, perhaps the single most obvious place to look for optimizing behavior.Footnote 41 Closer examination suggested a far broader pattern and one far from the neoclassical vision of the firm. This is not just a matter of careful incentive design and monitoring coupled with continuous decision-making failing to be the solution to all problems. In many large organizations, affirmative decisions are in fact relatively rare, and alternatives are not often systematically canvassed, still less explored, in the course of decision-making. Conventional patterns of behavior, not necessarily even explicitly articulated, and conventional senses of what alternative sets of action might be possible if anomalies in some situation suggest not proceeding along entirely conventional lines are often far more powerful influences. The main pattern for firm action is that firms and smaller units within them continue to do what they have done previously. Sometimes startlingly novel situations do present themselves, of course, and sometimes the outcomes of routine behavior create conditions and crises that threaten the ongoing existence of the enterprise. However, these are exceptional. Routineness, in the loose everyday sense of the word, is the predominant theme and the predominant mechanism of change is feedback (i.e., from the environment) rather than foresight.Footnote 42
If there is a great deal of this essentially routine behavior at all levels of complex organization, if change is in itself costly to embedded actors, and particularly if decision-making typically devolves away from general executives toward operators, then the form and content routines could themselves exercise a powerful influence on how the day-to-day actions and life of the collective develop.Footnote 43 Indeed, the sometimes slow process of change in routines at the most micro level might well be a reason why firms’ behavior at more macro levels appears to be something other than a font of optimization.
This is a view that sees agency, at least in principle, deep in the bowels of the organization. It mainly confines agency to people, with some suggestion that routines of interaction and even technology present a set of affordances narrower and more constrained than might be available in principle.
There remains the question of the meaning of action from what we might loosely call an explanatory perspective. I have argued that action within and by organizations is generally dependent upon individual-level understandings of routines. From this, it is clear that explaining organizations’ development will in most cases require some exploration of those routines and how they changed over time. All conscious action is not necessarily conscious decision. The explanation of action must sometimes probe into the shadows of what is understood, what is assumed, and what is taken for granted.
Organizations only rarely have formal records—board minutes and the like—that explicitly reference either such routines or changes in them.Footnote 44 Even more rare are records that go beyond merely referencing those things and instead discuss them. This is so even of organizations that clearly underwent major changes in routines. More visible elements may well, in particular cases, be sufficient to explain the evolution of a firm’s capabilities and its outcomes. However, they may well not. Informal evidence from the inside may be illuminating, even decisive.
4. Structure and Agency in Relation to Each Other
When looking at a biological specimen—say, a fish or a mouse—the notion that its skeleton is a constraint on the size and location of everything else within its skin is a natural one. It is not outlandish to imagine that there are counterparts to this in social life, and we often speak casually as if structure has some sort of natural priority. The idea of structure has indeed long played a leading role in the development of sociology. The idea might even precede urgent professional desires to have something definite to study: consider the psychotherapist’s commonplace that people do not find themselves in situations so much as they put themselves in situations. This truism even denies the intuitive notion in an interesting way: It asserts that however influential structure may be, agency—the ability of individuals to gather their own impressions, make up their own minds, and take action accordingly—matters as well. It is no great step from this to the thought that the same might be true in societies. Understanding the relationship between structure and agency has certainly been a running theme in modern sociological discourse.Footnote 45
Two main approaches to the treatment of structure are immediately apparent. One, which could be thought of as deriving from Marx, sees structure almost entirely in materialist terms: The course of a society’s development over time derives from its material resources at particular moments and what he calls its “social relations of production.”Footnote 46 Marx argued that there is an inevitable aspect to such development. The works of the “structuralist” anthropologist Levi-Strauss lie at the other extreme. Levi-Strauss saw binary conceptual oppositions—most famously, the raw and the cooked—as deep determinants of social life, though he did not ever specify how these deep structures came to manifest themselves and control lived experience.Footnote 47
Because the sense of personal agency is a familiar one to readers of such literature (as indeed it is to almost everyone else), it is only natural that a literature sprang up attempting to reconcile the two approaches. In our time, the leading figure for such thinking is surely Anthony Giddens. Giddens has written on a wide variety of topics but is best known for the notion he calls “structuration.” The basic idea is that structure and agency reciprocally determine each other in a roughly symbiotic relationship, neither existing in a recognizable form without the other. Knowledgeable individuals succeed in acting because they can mobilize structures. The structures exist in part due to such mobilizations.
Giddens’s position has changed over time, but the core notion can be discerned and considered.Footnote 48 There are both strengths and weaknesses to his formulation. Giddens thinks of structures as being virtual (i.e., without any necessary explicit existence). He also thinks of structure as having two components: rules and resources. Rules are certainly virtual and at least some sorts of resources clearly have virtual components. That said, it is difficult to understand how resources could be entirely virtual. The reformulation in Sewell (1992) is much more helpful.Footnote 49 Sewell renames rules “schema” but continues to think of them as virtual. He thinks of resources as essentially material. In Sewell’s view, resources both “empower and constrain” social action, and the distribution of resources is reproduced by social action. There is a profound lack of determinacy to all of this, however. Structures are multiple, schema are transposable, and resource accumulation is subject to external forces and is often unpredictable. Resources are susceptible to multiple interpretations. (They to some extent embody schema and so are potentially quite ambiguous.) Structures intersect and overlap. The exercise of individual agency can involve, in effect, a choice of exactly which structure is to be enacted. This makes for a much more natural (and, in principle, researchable) account of how change in dominant structures happens.
One can see structure and agency, at least in outline, in all three settings in which I have been considering how action might be explained. It is dealt with a bit elliptically in the philosophy of mind discussions of how to reconcile physical causation with mental explanation. There is not much explicit reference to the influence of external context and ideas on mental events, Frankfurt (1978)’s dancing spider notwithstanding. However, at least as far back as the Nichomachean Ethics of Aristotle cited above, there has been a tradition of acknowledging that the origins of actions may lie outside our bodies and indeed even in a sense precede them. (Character, for example, may inform action.) But this does not matter for questions of free will and agency so long as the causal chains pass through us in the right sort of way, a way implicating consciousness of our own beliefs and desires. The focus of the literature is, as noted above, resolute on this last point.
If that focus represents a sort of maneuvered triumph of agency over structure, the bulk of modern quantitative social science reaches out in the opposite direction. In the most extreme presentations, the evidence at hand is presumed to be a random sample of some wished-for population and the project of research is to find patterns in that evidence.Footnote 50 There may be background assumptions as to the logic of action on the part of those whose actions are represented by this evidence, in which case the exercise could be thought of in part as a weak test of those assumptions.Footnote 51 But in some versions of this way of working, there is no such assuming at all: The object of the enterprise is to expose, or at least to provide some incremental evidence about, what that logic was. In a curious way there is no agency at all in this approach. Globalizing principles—presented as theories—are the objective. There are parametric facts and mechanical rules. No embedded actor has to make sense of complex or ambiguous evidence or otherwise to construe. There might be calculations a machine could make. But no one would mistake such computations for fully human cognition in the wild.
Most history writing has more in common with the philosophers. Although, on the surface, most historical writing lays out narratives, most historical writing is, if not explicitly then at least implicitly, about contingency. It is about how things might have been one way or another. It is fundamentally about choices and therefore inevitably about agency. Historians sometimes say that they aim to complicate the picture. (Some social scientists are eager to concur, albeit without flattery as an intention.) What this remark describes is the gathering and consideration of elements that other accounts of the past might abstract away. However, the purpose of bringing in such detail is not an interest in detail as such: It is usually to develop the possibility that familiar facts and outcomes might have been other than as we have observed them.
The problems with attempts to explain action in all three of these settings is that they ignore the aspect of the relationship between structure and agency on which Giddens focused; the two are to some extent at least constitutive of each other in some respects. Actors do not make sense and construe ex nihilo, and structures do not persist if they are not enacted and replicated. The philosophers take a part for the whole. (We do, in fact, hold people morally accountable for their character.) In the sort of social scientific analysis I was describing, the image of the individual borders on the unrecognizable. Also, in the (perhaps somewhat idiosyncratic) account of historical writing I gave, it would be almost willfully naive not to imagine larger forces creating and constraining opportunities. What each of these perspectives lack, to one degree or another, is the sense of structure and agency in some sort of ongoing tension with each other, a tension that might itself stretch one way and tighten in another, interestingly and consequentially, over time.
This observation begins to expose questions of what it is to be a truly social science. The initial aspiration was indeed to be a sort of physics of social relations But many of the giants of the sociological tradition in the late nineteenth century worked very far from the ideas of globalized theory, the valorization of testable (and tested) hypotheses, and actually testing them. Central aspects of that early tradition live on, quite vigorously, in a number of fields that are usually considered social sciences (parts of sociology, most of anthropology, much of political science, etc.). The turn toward positivism in academics' research is an artifact of the twentieth century, in major part due, one supposes, to the widespread availability of increasingly inexpensive computing power.Footnote 52 To a scholar with a hammer, perhaps the whole world really can look like a nail.
The philosopher Wittgenstein started out as a sort of positivist as well. His Tractatus Logico-Philosophicus of 1921 begins “Die Welt is alles, was der Fall ist. Die Welt is die gesamtheit der tatsachen, nicht der dinge.”Footnote 53 Sentences picture the world. Sentences are true or false and can be tested.Footnote 54 Yet after deep and, one gathers, anguished reflection, he eventually changed his position. By the time the notes for what is now known in English as his Philosophical Investigations (1953) were written, he had come to the view that far from bearing a sort of one-to-one relationship to the world, language was a set of conventions and understandings among speakers, a kind of game everyone who knew how to communicate knew how to play.Footnote 55 Although traditional Western philosophy begins from the experience of the isolated individual (consider Locke and Hume on sense data), the later Wittgenstein, with his emphasis on rule-governed practices, privileges the social.Footnote 56 This gives a very different sense of the basic subject of social science—not so much focused on isolated individuals as on individuals precisely and directly as they interact with one another.Footnote 57 This yields an account that is open to change in what it assesses as structure and agency over time, as circumstances, practice, and purposes evolve.Footnote 58
5. Agency More Deeply Probed and Implications Drawn
Apropos agency, I have proceeded thus far in a casual way, invoking only a loose notion of what agency represents and taking a meaning for agency out of little more than the contrast to structure (i.e., to a relatively strict determinism). The latter approach is clearly too broad and sweeping to be very useful to a working historian of any sort today. The inadequacies of the former may also be worth some attention.
There are two distinct reasons the loose and derivative way is not good enough. Many readers of this text will have been exposed to some version of the economic theory of agency and may well carry it around in the back of their minds as representing what the word means.Footnote 59 Economists affirmatively prefer definitions with an absolutely minimal level of detail for their usual purpose of seeking out general conclusions about their usual subjects—considered as abstractly as possible—of resource allocation and efficiency. Our subject, business history, is a kind of analysis that takes an interest in generalization but is oriented to consequences of the responses to particulars. Detail and the distinctions it suggests might well matter. Furthermore, if this is so at the level of action by an isolated individual, it is even more likely to be true at the level of action in the context of an organization—consider the embeddedness of the individual and the coordination issues that such a context raises.
A more elaborated and circumspect definition of agency offers the possibility of more illumination. The sociologists Mustafa Emirbayer and Ann Mische have proposed a more detailed and subtle definition of agency, one that has, given the concerns of this paper, many virtues.Footnote 60 The virtues are also useful in overcoming the principle defect of the Sewell view discussed above, namely that it sees agency mainly as an activity of transposition of already existing elements and schema into new settings. This seems untrue to felt experience as well as to observation. This section explicates Emirbayer and Mische’s core ideas and sketches some implications for the working historian, in particular the working historian of business organizations.
Emirbayer and Mische first observe that agency would be meaningless in the absence of time passing. The first element of their definition, then, is the claim that agency is a process inherently embedded in time. Why is this so? Consider the conceptualization of agency conventional among economists. Economists, in standard microeconomic analysis, want to study choice in an isolated present and to view choice as being a consequence of individuals’ preference orderings. These are, in the moment at least, static, well-known to the individuals, and defined over given and transparent specific alternatives. The entire process of evaluation and decision-making is complete and whole in that moment and intervals of time are just a sequence of present moments.Footnote 61 This approach is self-consciously stripped-down, with enough complexity to present the problems and patterns the economists want to analyze but nothing in it that would create inessential contingency or complication. This is so for a reason. The economists are after tight, logical deduction from minimal assumptions—elegant severity is their friend, and realistic detail is their invitation to clutter and ambiguity. Emirbayer and Mische are not hostile to abstraction. However, their notion of what is too important to be left out starts with realistic observation, not the drive toward the simplest possible cases. The latter can certainly be helpful in the analysis of concepts. The former may well be more helpful in contexts ripe for a more hermeneutic inquiry.
If Newton saw time as a series of isolable moments, it must be said that a more observer- or agent-based perspective is also possible. This is hardly unknown to the discourse of philosophers: Bergson’s flow of events and experience and George Herbert Mead’s situated but continuously moving frame of past, present, and future would be examples.Footnote 62 In this sort of perspective, human experience is always at least implicitly contextualized by some sense of a past and a future. If that is so, then acts of agency at whatever moment cannot be thoroughly understood without reference to that moment’s and the actor’s past, the particulars of the present and of the actor in the present, and the moment’s and the actor’s future.Footnote 63 The demands and other realizations of the present are continuously measured against that past. Before there is action, there is possible action, and before there is any set of possible actions, there is some projecting—straight up and with variations to be considered—of the present onto the future. The present is always, in this view, experienced as a sort of interregnum between a recognized past and a still underdetermined future.Footnote 64 All this is at once fixed in the moment and continually moving forward in chronological time. There is no timeless agency.
Though Emirbayer and Mische think that all three of these distinguishable activities, which they call “engagement,” “projection,” and “adjustment,” can go on at any moment, they think this need not be so and that even when all are going on, the different elements can be of unequal salience and force. The mixture of them—and the balance between them—may change from moment to moment. One can easily see that such changes might well affect actors’ responses to the constraints, uncertainties, and ambiguities presented by (i.e., in) any particular moment.Footnote 65 This lays the basis for a rich and dynamic account of the ability of actors to influence structure over time.Footnote 66 This relationship between structure and agency, both in the moment and over time, is a much more complicated one than the reading of Sewell given above.
The second main element of the Emirbayer and Mische view is that agency is usually socially embedded as well. Clearly, there could be degenerate cases—the economist’s canonical example of an absentee principal providing a set of rules and incentives is one. However, degenerate forms are always nested in larger spaces, and even in this case, some fairly specific social arrangements (for example, property rights and enforceable contracts) must be present in the background for the roles of principal and agent to be what the economist assumes them to be. Emirbayer and Mische think the common case is, in fact, the nondegenerate one, the one in which the rules and incentives are at least partly tacit and are, stronger still, ones the agent does not need a concerned principal to teach them about.Footnote 67
Organizational routines have precisely this character. They represent distinctive and well-recognized ways the organization has of responding to particular situations, both in terms of what is exigent and what is desirable. However, knowledge is also part of what they represent: They are why one can say with confidence, as one can of most organizations, “That is how things are done here,” or, as noted above, “That is the sort of thing that would never happen here.” Organizational routines are not necessarily (and are certainly not generally) entirely articulated or codified, and what is not entirely articulated or codified should be regarded as tacit knowledge. They are passed on through something like apprenticeship, can pass away in similar fashion, alter in response to experience, and alter very slowly (situations of suddenly revealed and potentially disastrous dysfunctionality aside).Footnote 68 All these time-related attributes ultimately derive from the fact that routines exist in a sort of collective space, an essentially social rather than individual setting.
The final element of the Emirbayer and Mische notion of agency is its connection to action along three specific lines. Each of these lines is rooted in human cognition, and each is complex enough to call for a little exposition.Footnote 69
The first idea is that agency involves cognition informed by the past, specifically in the iterative or habitual aspect of that remembered past. Established ways of doing and being are always consciously present to mind in this view, individual as well as collective. “Habit,” in this Deweyan sense, is more like an available repertoire than an inevitable response to a specific stimulus.Footnote 70 So the first idea is that agency necessarily involves a kind of knowledge of the past, including past action patterns themselves. Past actions are actions—performances, individual or collective—that might be re-performed.
There is a more general air of American pragmatism about this view of actors’ relationship to the past. It begins by seeing the totality of the past at any moment as being so vast and detailed as to be overwhelming. One might say that humans consciously trying to engage with the past seek not a total past but a usable one. One might see individual habit, as Dewey does, as playing a central if not uncomplicated role in both daily individual life and in the collective life of institutions. Habit, in this view, involves understanding what response is appropriate to any particular, and potentially ambiguous, situation. Habit can also have an active element.Footnote 71 How is this to be understood? Emirbayer and Mische see three main components: selective attention, recognition of types, and categorical location.Footnote 72 Selective attention matters because only by not attending to everything is there any hope of recognizing order in the initially chaotic and ambiguous. Meaning arises from grasping patterns as such and situating patterns in a context.
Second, agency also involves some sort of consciousness of the future. Because the future is, more or less by definition, not yet known, the consciousness in question is inevitably complex. For the notion of choice to be meaningful, it must be possible to imagine alternatives. For the notion of active choice to be meaningful, it must be possible in some sense or other to evaluate imagined alternatives. For many possible affirmative choices of interest, this is not easily reduced to a single expected payoff, either given exogenously or inferred via some expected value calculation. Much of economic life, at the organizational level as well as at the individual one, looks like a cooperative game because of the constrained but nonetheless ultimately underdetermined structure its outcome set usually has. With that said, much of economic life appears to its participants as being much more full of uncertainty than risk.Footnote 73 In either case, the key feature of particular possible outcomes is often not their specificity so much as that they point toward certain sets of more far-off possibilities, perhaps less than precisely defined or elaborated, rather than toward others. In short, projection can be an open-ended affair. Imagining alternatives (i.e., recognizing ones not immediately presenting themselves and perhaps even dreaming up some new ones) is fully as central an aspect of choice in such a setting as is evaluating (deliberating, as Dewey would have it). Dewey would be further pleased to note that it is difficult to believe that emotional aspects of possible alternatives, not least the comfort of the known, play no role. It is surely often the case that the emotional plays a role when the presenting choice is the passive one of doing nothing affirmative—just continuing on as before with no changes—rather than doing something new.
The relationship of actors to the future raises different issues than actors’ relationship to the past. Emirbayer and Mische see three different main components here. I find it helpful to consider these in a different order than theirs. In the first, actors take elements of meaning or intelligible patterns apart and combine the constituents of these in different and even potentially novel ways. This Emirbayer and Mische call, somewhat awkwardly, “symbolic recomposition.”Footnote 74 In the second, actors situate future possibilities in more or less coherent temporal sequences— not necessarily a strictly causal sequence, as I suggested above, but certainly not a causally incoherent one. This Emirbayer and Mische call, clearly but perhaps somewhat simplistically, “narrative construction.”Footnote 75 The remaining element they call “hypothetical resolution,” coming to an assessment of where these processes of succession might be going.
The third idea about cognition is that it involves an orientation toward the present. This takes up some of the spirit of my discussion of construing possible futures. Concerning the present, it is the ability to see in the moment what need not be routine, what opportunities there might be for doing something differently. This is, of all three, the one most deeply within the domain of the counterfactual conditional. It invokes the familiar but also the evaluation of whether that represents the only possible response to present conditions. To the extent that such an evaluation is partly a matter of affect, as often will be the case, it implicates the emotional aspects of cognition as well. In an organizational context, it, of course, also implicates a sophisticated awareness of institutional (and institutional-political) opportunities and constraints and all that these bring in their train.
“Practical-evaluative behavior” is the name Emirbayer and Mische give to the adjustment of cognizable actions in the present to context and possibility. The first element they see under this heading is the recognition of what aspects, if any, of present activities and plans are, under present circumstances and expectations, problematic. If something is, there is then the matter of coming to a conclusion about what to do—one way or another, and not necessarily through “rational” abstracted evaluation processes. Then there is execution—perhaps just continuing on as before, perhaps intervening and doing something different. In either of these cases, execution may not be simple. If the circumstances of the moment are changing, even the continuance of routines may require projection of familiar activities onto the not necessarily entirely familiar circumstances of the moment.
6. Conclusion
Where does this get us as business historians? If agency indeed implicates all the ideas of the preceding section, that fact in itself has clear implications for history writing. The notion of agency being embedded in time means the history of agentic behavior has to seriously consider the agentic actors’ (more precisely, the potentially agentic actors’) understanding of their pasts, presents, and possible futures. It must also consider whether those actors are to be understood as individuals, as individuals embedded in organizational contexts, or as organizations themselves. For history writing in general, the notion of agency as socially embedded means that the subject can never just be the man on a white horse. How he came to be dressed as he was; how the horse, and that horse, came to be under him in the first place; what he was pointing to with his outstretched hand; and for whose consumption he was striking the pose—all these may, so to speak, be part of the subject. For writing the history of organizations, this makes an organization’s whole range of activities and resources potentially an intrinsic part of what must be studied. Agency being informed by the iterative or habitual aspects of the remembered past means that potential actors’ sense of their own experience must be reported as part of their sense of what they could do. For writing the history of organizations, it means that the content and forms of organizational routines are an important part of the background of any account of challenging circumstances and possible action. The projective aspect of agency means, for history writing in general, that an important part of an account of any actual decision is the paths contemplated but not chosen, and an important part of an account of any situation threatening enough to even suggest decisions or actions is the actors’ sense of what might be possible. For writing the history of organizations, this is only more so, both in terms of what might be conceived and the politics of selling alternatives internally. The practical-evaluative aspect of agency means, for history writing in general, that careful attention should be paid to dysfunction—real or imagined—as it can be an important part of potential actors’ understanding of the present situation, of what they understand the alternatives might be, and of the more ordinary subject of execution. For writing the history of organizations, these suggest in particular a perspective less oriented to outcomes than is usual and more to actors, processes, and the progress of sense-making. The rhetoric and politics of persuasion within both the groups that actually come to decisions and those that might, or might frustrate the decisions of others but do not, could also be part of such inquiries.
One virtue of this conception, taken all together, is that it advances the discussion well beyond where the Giddens and Sewell views had left it. The sorts of structure that they chiefly had in mind were on a socially larger scale than that which are found in the analyses I have been talking about.Footnote 76 This conception takes agency down to the level of the individual potential actor or organizational collective. Such a devolution has two interesting aspects. The first is that it advances the discussion of the preceding section at a conceptual level: As noted above, if agentic behavior at any moment or in any circumstances might involve a different balance of the three underlying elements, then it might, in the particulars of that balance, represent different visions of what might happen, and thus, when viewed from the outside, represent different actual opportunities for change. Part of what is interesting about the whole subject, after all, is the potential for interplay between structure and agency, both in understanding the past and, in some settings, in thinking about how best to organize for the future. The second is that this subtler analysis of the elements and aspects of agency suggests a very detailed agenda of matters to be investigated: It offers the analyst of agency in specific embedded historical settings a rich menu of topics for investigation and consideration.
It would certainly be fair to say that pursuing this calls for the examination of some unfamiliar aspects of organizational life and its history. These are frequently accessible in long defunct firms, certainly accessible in ongoing firms, and obviously accessible firms whose records and important figures are still available for interviews and probing. It would also be fair to say that it points to a notion of action and the explanation of action that is less determinate than might otherwise be imagined. That does not mean that this considered notion of action is somehow a less useful tool, however. The usual narratives are incomplete. The notion of action advanced here is a tool that exposes the incompleteness fact in a way with which we could do something.
