The Literature on Institutions

Philip E. Agre
Department of Information Studies
University of California, Los Angeles
Los Angeles, California 90095-1520

This is a draft. Please do not cite it or quote from it.
Version of 9 October 2000.

9500 words.


Note: I keep telling you that I take an institutional perspective on the place of information technology in people's lives. But I've never explained in detail what I mean by the word "institution". The text I've enclosed is something I wrote mostly for myself to digest the sociological, political, and (to some extent) the economic literature on institutions. I've tried to write it in reasonably plain English, just because I don't feel like I understand anything unless I can say it in plain English. I've certainly cut some corners on theoretical accuracy when I couldn't find a plain-English way of saying things. The writing isn't the best either. But I think the ideas are useful, so I hope you get something out of it.

1 The need for a theory of institutions

So far I have been using "institution" in an uncritical way. Now it is time to define the word "institution", drawing on a variety of powerful literatures. "Institution" is a complicated word, and any definition should be treated only as a placeholder for subsequent analysis. That analysis should provide some valuable services:

* A common language for gathering a large and diverse body of useful ideas about institutions in a coherent way. These ideas come from many disciplines, and they build on experiences in many areas of social life.

* Equal justice for a wide range of phenomena. Although I am obviously interested in those institutions that support information technology, I am also concerned with the infinitely more numerous institutions in which information plays an important role -- that is, nearly all of them.

* A "meso-level" theoretical concept. Theories of society often get stuck in a dichotomy between "macro" and "micro" theories that cannot be persuasively connected to one another (e.g., O'Neil 1998: 10), and the concept of an institution provides something of a "trading zone" (Galison 19xx) for these very different ways of talking about the social world.

* A conceptual bridge between technology and the social world. Institutions have formal, structural aspects that relate directly to the workings of computers, and they also have substantive, process aspects that relate to the much larger social processes within which computers are embedded. A good theory of institutions can thus support some heavy intellectual traffic between the technical world and the social sciences.

* Protection against overgeneralizing about activities in different settings. Attempts to generalize about people's motives, cognition, interactions, and relationships are often futile, for the simple reason that people think and interact in different ways within the frameworks that different institutions provide. Interactions within a stock market, for example, are entirely different from interactions within a family. By making the institutional context of activity explicit, we can respect these differences without making them sound random.

* The ability to evaluate institutions. In making theories of institutions, I am not endorsing them. Some institutions are efficient and fair, others are pathological, and others have simply outlived their usefulness. In any case, a period of rapid technological change will bring choices, and we need a vocabulary for discussing them.

* A conceptual framework to support design. Information technology is increasingly intertwined with activity, and it mediates human relationships of increasing complexity. A theory of institutions may thus provide a robust analytical framework for drawing three-dimensional pictures of the context in which a device will be used. It may also help differentiate settings that should be treated differently for design purposes; as the example of Notes suggested, a system that works well in one institutional setting may cut across the grain of another.

* An intellectual basis for the design of new institutions. Institutions are changing, and the design of important new information technologies will inevitably be part and parcel of institutional design. The very concept of institutional design will turn out to be almost a contradiction in terms, but perhaps an understanding of the difficulties can prevent some errors.

The word "institution" already has several vernacular definitions:

(1) Carceral organizations such as prisons and mental hospitals. In this sense, to be "institutionalized" is to be removed from the regular social world and subjected to a dehumanizing regime of total control (Goffman 19xx).

(2) Organizations that, in Selznick's (1957: 17) words, are "infused with value". For example, a university is an organization because it has a certain legal status, hires employees, issues policies, and so on. But it is also an institution because many people have invested it with an emotional set of values, such as teaching, scholarship, and public service. Even a restaurant can call itself an institution if it is sufficiently embedded in the local culture.

(3) Rigid, inflexible patterns of activity. This is Kling and Iacono's (1989) definition of the word, and it has a long history going back to Veblen's (19xx) theory of organizational routines and habits as a mechanism for transmitting practical knowledge in industry.

These vernacular definitions of the word "institution" might be contrasted in various ways. Whereas the first two meanings of the word identify institutions as particular types of organizations, patterns of activity need not be coextensive with the boundaries of a single organization. But more importantly, observe the ambivalence in these definitions: the first definition is extremely negative, the second is extremely positive, and the third appears negative at first ("rigid, inflexible") but upon reflection reveals more positive aspects ("mechanism for transmitting practical knowledge"). Although more recent definitions of the word "institution" have become much more general, this ambivalence remains crucial for reasons that will become clear.

Vernacular usage also tends to portray institutions as remote: they happen in big buildings, they are official and perhaps even secretive, they issue rules, and so on. So far as social theory is concerned, however, this intuition is almost precisely wrong. Banks project a forbidding image of formality and permanence, but the institution of banking is something that lives in the everyday lives of normal people. It is centrally about the habits of thought that people acquire in a society of bank accounts. It is also about the routines of interaction between bankers and their customers, and about the artifacts (checkbooks, ATM machines, bank tellers' windows, and so on) that support these interactions. The institution of banking is also much larger than any particular bank. Banks come and go, but the institution of banking remains much the same. Banking is founded on a complex legal framework, and on the patterned outcomes of negotiations between banks and the firms that borrow money from them. It is shaped by labor law, by accounting standards, by cultural norms of fairness, and by customers' long-standing expectations about what a bank should be. The institution of banking is this larger complex of social relationships that is spread all around into every corner of society, not the particular organizations that happen to serve as banks at any given time (North 1990: 4; Offe 1996: 203). That is the intuition.

That said, let us consider some definitions of "institutions" << Footnote: Commons (1959: 69) describes some of the recurring ambiguities in uses of "institution". DiMaggio and Powell (1991: 7-10) review various definitions of the term. >>. One classic definition by Walton Hamilton (1932: 84, quoted by Hodgson 1999: 143) describes an institution as "a way of thought or action of some prevalence and permanence, which is embedded in the habits of a group or the customs of a people ... Institutions fix the confines of and impose form upon the activities of human beings". Where Hamilton emphasizes patterns of activity in general, more recent definitions focus specifically on patterns of interaction among people. The economic historian Douglass North (1990: 3) has suggested that institutions are "the rules of the game in a society or, more formally, are the humanly devised constraints that shape human interaction", and elsewhere (1991: 97) that they are "the humanly devised constraints that structure political, economic and social interactions. They consist of both informal constraints (sanctions, taboos, customs, traditions, and codes of conduct) and formal rules (constitutions, laws, property rights)". The language of rules and constraints suggests the inherent ambivalence of institutions. And the political scientist Robert Goodin (1996: 19) defines institutions as "organized patterns of socially constructed norms and roles, and socially prescribed behaviors expected of occupants of those roles, which are created and re-created over time". The language of roles suggests the role of institutions in shaping individual identities. For present purposes, then, let us provisionally define institutions simply as "the persistent structures of social relationships", with the understanding that these structures include both public and private, formal and informal, legal and cultural, large and small -- the entire range of possibilities.

Examples of institutions include intellectual property law, technical standards, church service, taxicabs, the English language, form contracts, the research university, the nuclear family, parliamentary procedure, holiday customs, venture capital, hospitality, and the rules and conventions of driving on the highway. These examples will recur as the next section enumerates the properties of institutions in more detail.

2 Existing theories

Several schools in the social sciences have developed theories of institutions, and my strategy will be to synthesize them. This will not be hard; despite their surface antagonisms, they share a fundamentally consistent view of the nature of institutions, even though their theories of institutional change are all over the map. The theories appear diverse because they tend to generalize from particular favorite examples (Goodin 1996: 1), emphasizing some themes at the expense of others. Putting these themes together, however, one obtains a whole theory. In presenting this theory, I will not try to compare and contrast the details of the authors' views; that would be a valuable project, but not here.

Here, briefly, are the institutional theorists and the gists of their theories.

(1) John Commons was the mechanic philosopher of the New Deal. A printer turned professor of public administration, he developed a theory of social institutions modeled on his experience of collective bargaining over work rules. He built a theory of institutions starting from the basic unit of the "transaction", whose rules emerge through the common law or processes like it.

(2) Paul David is a British economic historian who has drawn attention to the reciprocal relationship between economic institutions and technical standards. David's work is relevant here because of the role that standards play in the long-term stability of institutions.

(3) Paul DiMaggio and Walter Powell are American sociologists whose theory developed from their studies of nonprofit arts organizations. Their theory describes the transfer of loosely integrated packages of practices from one organization to another within a given organizational field, and they emphasize particularly the "isomorphism" that can result. Their New Institutionalism -- one of several theories by that name -- downplays economic rationality but is nonetheless strongly cognitive in orientation.

(4) John Dryzek is an American political scientist who describes the "informal logic" of institutions that can be found in discourse. No institution will remain stable for long unless its basic categories are inscribed in the taken-for-granted language that its participants speak, and Dryzek uses empirical discourse-analysis methods to recover the informal logics that can be found operating in American political culture.

(5) Michel Foucault was a French philosopher and historian who presented a morbidly exaggerated, if suggestive and compelling, theory of the power of institutions to define people and structure their activities. His concept of discourse, though serving a similar purpose to Dryzek's, is much broader and includes the full range of artifacts and embodied practices by which an individual is knitted into an institutional order.

(6) Harold Garfinkel is an American sociologist who founded a school known as ethnomethodology. Although it developed from the phenomenological tradition of Husserl and Schutz, ethnomethodology is concerned not with the structure of inward experience but with the public displays by which the participants in social settings make ongoing sense of their shared situation. Because it locates the various categories of social order in this improvised sense-making activity, ethnomethodology is chronically at odds with the mainstream sociological tradition that treats society as an objective structure that can be studied with conventional scientific tools.

(7) Anthony Giddens is a British social theorist whose theory of structuration is the most widely accepted solution to the serious and longstanding problem of reconciling the free will of human agents with the determining structures of society. As director of the London School of Economics, Giddens has also played a central role in the intellectual development of political centrism by the "New Labour" government.

(8) Friedrich Hayek was an Austrian economist who synthesized and defended the European tradition of liberalism against socialist prescriptions for economic planning. He regards engineering as a catastrophic model for social policy because of its orientation to centralized control, and he also developed a theory of knowledge as intrinsically local and thus resistant to centralization. In particular, his understanding of the economy as an institutional framework for the production of knowledge contrasts with the mainstream economic focus on the allocation of scarce resources.

(9) Geoffrey Hodgson is a British economic theorist who has synthesized a historical tradition of research on market institutions. He observes that economics is currently dominated by a neoclassical tradition that analyzes supply and demand in formal terms, with almost no concern for the actual concrete workings of markets as institutions of social life, and he tries to reverse this trend by systematically analyzing the institutional aspects of action, contracts, property, firms, knowledge, and individualism.

(10) James March is an American management scholar and Johan Olsen is a Norwegian political scientist. Together they have developed theories of the political life of institutions (March and Olsen 1989) and the institutional life of politics (March and Olsen 1995). Their conception of social life is resolutely non-rational, and they attempt to recover a positive sense of politics as the search for majorities by finite actors whose interests undergo reflection and revision in the course of assembling alliances.

(11) Douglass North is an American economic historian. He is the most sophisticated of the many authors who have built upon Coase's (19xx [1937]) theory of transaction costs as the economic basis of the firm. He recounts the growth of economic and political institutions that have enabled some societies to reduce the costs of economic exchange and thus capture the gains from trade. Because societies have varied dramatically in their ability to construct these institutions, he has tried persistently, although without much success, to "endogenize" ideology as an element of economic theory.

(12) Claus Offe is a German political scientist who has analyzed the social welfare states of Western Europe and the post-1989 political and economic transitions in the East. His impressively lucid work on the transitions in Eastern Europe, which I will draw upon heavily here, is concerned with the delicate problem of institutional transition in a society whose culture, values, norms, and ideology are still largely aligned with the discredited and unworkable old order.

3 Properties of institutions

Despite their diversity, institutions share numerous attributes. I will consider these separately, citing the authors who have described them and providing examples.

a. ontology

Every institution comes with an ontology -- it determines what the world is made of. The institution of banking, for example, proposes that the world includes accounts, loans, transactions, balances, currencies, among other things. If the institution did not exist then accounts would not exist, or at least they would mean something quite different. Church services propose that the world includes altars, prayers, and sermons. The English language proposes that the world includes nouns, verbs, and adjectives, all of them subdivided into categories that serve different grammatical functions. The rules and conventions of highway driving propose that the world includes vehicles, intersections, lanes, and one-way streets.

In what sense do bank accounts, sermons, verbs, and intersections exist? This turns out to be a hard question. In the case of a bank account, it seems reasonably clear that the "thing" -- the account -- exists because the banking system says it does. Even though it exists in various physical forms, such as bank statements printed on paper and received in the mail, it is still an abstraction. Of course, it is not a fiction: if your bank account is empty then you cannot eat. But it is only real because certain people agree to continue treating it as real. Why do they continue treating it as real? That is what a theory of the institution should explain. Sermons too are real: an actual person is standing in a certain place in a church, speaking in a certain fashion, while other people listen according to their customs. The speaker is uttering certain words whose structure and meaning conform to the grammar of some language. But what makes these utterances a sermon? Austin (1962) observes that many speech acts require certain institutional conditions in order to be "felicitous". If a stranger were to stand up during a prayer and speak those same words, the words might be intelligible but they would not be a sermon. Like the bank account, the sermon must be arranged appropriately in physical terms, and it must also be recognized as a sermon by certain people. Likewise with verbs and intersections -- a spoken sound and a physical arrangement of asphalt whose significance for social life depends on the institutions that specifies what they are. In each case, the ontology straddles the boundary between "subjective" (existing solely in the heads of the institution's participants) and "objective" (perfectly discernible to a non-participant). The ontology is a conceptual framework made material.

This is still an oversimplified account of institutionalized ontologies, and it needs to be complicated in at least two directions. In formal terms, every institution comes with a discourse, that is, roughly speaking, a vocabulary for talking about the world. Bankers and their customers talk about bank accounts, for example when establishing a new account or withdrawing money from it, and they need a common vocabulary to do so. A religious institution will provide ways of talking about sermons, for example their parts and their citations to religious texts. Natural languages provide their speakers with a tremendous array of grammatical forms for talking about the language itself (Silverstein 19xx), for example when reporting what other people have said. Highway traffic comes with an elaborate discourse that drivers are expected to master. Each discourse has a complex structure, and Dryzek (1996: 109) observes that:

"... any politically interesting discourse contains: (1) An ontology, or set of entities whose existence is recognized. (2) Ascription of agency to some entities: these entities, be they individuals, groups, institutions, wood nymphs, or social classes can act; other entities can only be acted upon. (3) For agents, some ascription of motive, and concomitant denial of other motives. (4) Taken-for-granted relationships (especially hierarchies) across agents and other entities."
But the discourse itself does not bring the social world into being. People do that, and they do it through a complicated combination of talking, writing, pointing, interacting, making things, and so on. Detailed study of particular examples of social interaction by practitioners of ethnomethodology (Garfinkel 1984 [1967]) have made clear how complicated these processes of "social construction" really are. For example, a bank teller and customer may feel that they are talking about the same bank account, but they can always discover that they were actually talking about different accounts. That sort of agreement is always provisional, always open to subsequent revision. In many ways the ongoing agreements that underwrite the social world are fragile, tenuous, and contingent. But in other ways they are remarkably robust, given the massive array of methods that people employ to maintain a workable consensus about the nature of the situations they face. This complex interactional consensus-building work is normally invisible to its participants and to sociologists alike, not least because pointing it out would make it even harder.

This is a difficult idea, since it would seem to question whether things like bank accounts, sermons, verbs, and intersections really exist. But as institutional facts they assuredly exist and have consequences for everyone involved, and they way in which everyone provides for an ongoing shared sense of their existence has to be discovered by going and looking. It is often objected that, if things "only" exist as institutional facts, then the participants in a social setting could simply conspire in an arbitrary way to change them, or to invent wholly absurd things. But that objection wrongly presupposes an unlikely answer to a hard empirical question: under what conditions people really can organize a shared institutional reality. The institutional world operates differently way from the internal workings of a computer, and yet computers are densely embedded in the institutional world nonetheless.

b. roles

Part of every institution's ontology is a set of roles to which people can be assigned. Banking defines roles such as customer, teller, branch manager, and banking regulator. Church services define roles such as preacher and congregation member. The English language defines roles such as speaker and hearer -- what Goffman (1981: 124-159) calls "footings". Traffic rules define roles such as driver, passenger, and pedestrian. These assigned roles set the stage for all of the interactions of institutional life, and the interactions themselves often begin by ascertaining who is a customer, who belongs to the congregation, who is allowed to hear what the speaker is saying, and who the drivers were. Having established these institutional facts, every competent participant in the institution will be able to enact -- and expect others to enact -- a long list of rights, duties, and protocols of interaction. Failure to make oneself intelligible as a proper occupant of an institutional role, whether as a customer in a bank, a congregation member in a church service, a hearer in a conversation, or a driver on the highway will bring hints, requests for clarification, requests to behave oneself, criticism, ostracism, or worse. In each case, the other participants in the setting will try to repair the situation by rendering the miscreant intelligible: as confused, as a newcomer, as a scientist conduct a perverse experiment, as a troublemaker, or if all else fails, as insane.

The participants in most institutions take their roles seriously. They do not simply regard them as fictions or conventions, but instead take them to heart as ways of understanding themselves. Doctors, for example, generally believe that they really are doctors, and people who drive cars generally regard themselves as drivers. But this is not always true. New graduate students often suffer from "impostor syndrome": the delusional belief that one was acccepted to graduate school by mistake, and that the error will soon be revealed. Meanwhile they try to "pass", simulating the interactional conventions of graduate school as best they can. And in some sense these students take their roles more seriously than anyone, given the great effort they feel obligated to invest in their simulation.

The participants in an institution generally live their roles in at least five ways: (1) they learn characteristic patterns of thought and language that encode an ontology and view of the world (North 1990: 17, 20, 40), (2) they invest emotional values, whether good or bad, in the institution and their role within it (Parsons 1937: 11, Selznick 1957; cf. DiMaggio and Powell 1991: 16), (3) they accept the role as part of their personal identity, (4) through their participation in the institution they become embedded in a network of relationships, both with people in the same role and with people in complementary roles, and (5) their activities as occupants of the role expose them to certain kinds of information about the world, while shielding them from other information. Of course, this list is a maximal scenario whose full generally does not apply to simple roles like taking a taxi. Simple institutions do entail roles, but the roles are not so all-consuming as being a doctor. Furthermore, every element of the list admits exceptions: an unconscious person can be a medical patient, people who take unpleasant jobs can make a special point of remaining emotionally indifferent to them, people who are assigned stigmatizing identities by others can refuse subjectively to be defined by them, and the occupant of a particular role can make a point of investigating the perspectives of others. But the institution can generally survive these minor forms of dissidence, at least until they grow into social movements.

The general point then, is that "individuals are changed and reconstituted by the institutional context" (Hodgson 1999: 115, emphasis deleted). People spend their lives practicing institutional roles. They develop skills and habits, acquire objects, wear appropriate clothing, and build appropriate social networks. They may even modify their bodies through exercise, and may immerse themselves in a whole way of life that the institution and its participants organize. Most likely they participate in numerous institutions, large and small, to various extents, and it is through this participation that they become who they are as social beings. Indeed, Laudon (1985: 732) defines an institution as:

"... a set of widely shared values and interests pertaining to areas of strategic and social importance. These values and interests are served by specific organizations through the allocation of status and roles, and they are internalized by individuals through lengthy socialization carried out by organizations".
The word "internalization" can be misleading. It does not mean that people build models of the institutional world in their heads, although that may be true as well. Instead, it means that they master the discourses, practices, and interactional protocols of their new roles, and that these patterns of social activity become part of their cognitive equipment. This is Vygotsky's (19xx) theory of cognition as the internalized form of social interactions.

A vivid example of these phenomena can be found in Silicon Valley, whose entrepreneurs intend to become rich by selling venture capitalists on their business plans. Venture capitalists are exceedingly rigorous analytical thinkers, however, and they spend much of their time providing feedback to entrepreneurs whose business plans have holes in them. An elaborate system of rituals has developed for this purpose, such as events in which entrepreneurs present their plans to an audience of venture capitalists, who then publicly tear them to shreads. Silicon Valley culture, however, does not interpret these shredding sessions as rebukes, because everyone understands that frank criticism is the most efficient way to internalize the analytical thinking of the people with the money. Having deepened one's capacity to think like a venture capitalist, one can return to the drawing board and prepare a more robust proposal. As Hodgson (1988: 179) observes, "[t]he manner in which people are affected by the market as an institution is not merely that it provides information, or merely constraints, but that it structures the process of cognition of the agents involved and can actually affect their preferences or beliefs".

The view of the human person that emerges from these theories differs from any kind of atomistic individualism. As Commons (1959: 74) puts it,

"Instead of individuals the participants [in an institution] are citizens of a going concern. Instead of forces of nature they are forces of human nature. Instead of the mechanical uniformities of desire of the hedonistic economists, they are highly variable personalities. Instead of isolated individuals in a state of nature they are always participants in transactions, members of a concern in which they come and go, citizens of an institution that lived before them and will live after them".
Sandel (1984) expresses the point (in the context of a theory of political culture) by pointing to the disadvantages of an "unencumbered self" whose cognition and fate are not bound together with those of others by shared participation in an institution -- in his case, a nation and its collective identity. "What is denied to the unencumbered self is the possibility of membership in any community bound by moral ties antecedent to choice" (Sandel 1984: 87).

Yet one should also be ambivalent, since the price of this membership in a community is an imperfect understanding of its nature. Participation in an institution is a large and complex phenomenon, and nobody who occupies an institutional role completely understands its nature, much less the full consequences of it. The institution's discourses will have implications and subtexts that are not obvious to its speakers. Its ways of speaking and thinking will edit reality by making some possibilities much easier to conceive than others. And nobody can learn the sheer mass of detail of any complex modern institution -- considerable division of labor is required. One's institutionally organized views of the world will be reinforced by association with others through the institution. This partial awareness is important for explanations of institutional life, but it is also an exceedingly complex phenomenon. Its boundaries are hard to describe, degrees of awareness vary among individuals and over time, and countermovements can arise to present their own understandings of the situation. Nonetheless, the participants in an institution really are "programmed" to a certain extent -- turned into gears in a social machine that they partially understand, largely accept, and mostly collaborate with.

Opinions differ about how oppressive it is to be defined so completely by an institution (Goodin 1996: 6). Foucault takes the matter to extremes. In his view, people are completely determined by the institutions that constitute them, and cannot stand outside of them, choose among them, or act back upon them. Although he speaks of resistance to institutions, the actual nature and scope of the potentials for this resistance are always vague. Another view is that many institutions oppress people much more completely by excluding them. What is worse, to be subjected to the programming and partial awareness of a doctor or lawyer, or to sit on a street corner and observe these institutions from afar? The choice, fortunately, is not binary: institutions can be organized in many different ways, some of which are less oppressive in these ways than others.

c. rules

Roles come with rules, most of them taken for granted by the institutional participants to whom they apply (DiMaggio and Powell 1991: 9, North 1990: 3). An immigrant may need instruction on how to stand in a bank line or fill out a check, but a long-time bank customer will do these things without thinking. Church-goers know when to sit and stand, when to talk and when to be quiet. Speakers of English conform to its grammatical rules without reflection. And drivers think about a hundred subjects while fluently obeying a thick book of traffic laws. The rules may well vanish from consciousness, assuming that they were ever conscious in the first place. The rules may be either formal and informal, and patterns of activity can become institutional rules simply by happening often enough. Thus Meyer and Rowan (1977: [x]) define institutionalization as "the processes by which social processes, obligations, or actualities come to take on a rulelike status in social thought and action". Commons (1970 [1950]: 27) refers to these "working rules" as "the duties imposed on individuals by the collective action of all together", and his use of the word "collective" is important. It is a word with a bad reputation, having gotten narrowed through the sort of compulsory collectivization practiced in the Soviet Union. But while the collective farm is certainly an institution, the notion of collective action is much broader. Culture is a collective phenomenon, for example, because it is shared among its members, handed down across generations, and presupposed as a common coin of thought and action. Institutions are also collective, and Hodgson (1988: 178) even observes that "the market has ineradicable social and 'collectivist' aspects as well".

Most institutions organize collective sanctions against rule-breakers. This is obvious enough in the case of criminal law, but applies equally well to a wide variety of other religious, economic, or state settings, each of which has its own formal or informal enforcement mechanisms ranging anywhere from dirty looks to vigilantism (Ellickson 1991). "Each kind of collective action is a government" (Commons 1970 [1950]: 40). Because the health of an institution depends on people following its rules, third parties with a stake in the institution often help to punish rule-breakers (David 1995: 22 [?], Offe 1996: 204-205). In an economic setting, rules need to be enforced because the parties to a deal may not have complete information about the nature of the goods, or about the other party's willingness or ability to perform their contractual duties (North 1990: 57).

d. defines a terrain of activity

The participants in an institution can see before them an elaborate terrain of institutionally defined people, places, and things, and looking out on this terrain they can imagine an elaborate space of institutionally defined actions. The institution will define states of affairs such as an account being overdrawn, the singing of a certain religious song, a sentence needing a verb, or a fire truck having the right of way. Against this background every situation will present a terrain of options, and it is upon this terrain that each participant will pursue his or her own strategies. Many of these strategies are themselves part of the institution, having been well-developed by previous occupants of one's role. And in being socialized into an institution's ways, one learns these strategies along with everything else. Career strategies, for example, are on display in the comportment of one's fellow workers and in the stories one hears about the more successful ones. The strategies are not something separate from the institution; they are part and parcel of its ways, institutionalized as what Bourdieu (1977 [1972]; cf. DiMaggio and Powell 1991: 25-28) calls the "habitus". "[F]orms of collective action are not something different from what people do. The organization of activity is simply the more stabilized aspects of activity. The form is part of the process" (Commons 1970 [1950]: 21).

This talk of institutionally defined terrains of activity must be interpreted carefully, however. Simon (1945) initiated a tradition of "behavioralist" research by observing that the real action in an organization may bear little relationship to its official organization chart and published procedures. (Danziger et al's (1982) emphasis on the primacy of organizational politics represents an especially (and excessively) strong version of this view.) The rules of an institution are thus not the same as official job descriptions in an organization. At the same time, the relation between them is most unclear. Where exactly is the dividing line between official stories that are pure rationalization and formal structures that define the individuals and their actions? The answer will vary, and it has to be sought empirically in each setting. At one extreme is the world of routinized cynicism that Vaclav Havel (1985) described in communist Czechoslovakia, where everyone talks the Party line but nobody -- including the Party -- believed it. Somewhere in the middle is Kunda's (1992) equally disturbing picture of "corporate culture" at a prominent high-technology firm -- a discourse that encourages deep personal commitment to work, and which succeeds despite the endlessly ironic use that the employees make of it.

e. structures interaction

It should already be clear that institutions structure the interactions among their participants (North 1990: 4). The participants may sometimes violate the rules, but the rules nonetheless provide a moment-to-moment system of expectations that confer meaning on whatever they do (Sacks 19xx). Some of the protocols of interaction, such as those of cashing a check or participating in a religious ritual, are relatively rigid, whereas others, such as driving on city streets or conversing in English, provide an infinite world of options. These structures of interaction help to make other people predictable and to reduce uncertainty (Hodgson 1988: 191-192, North 1990: 25). They can be heavily formalized, or they can simply be settled patterns. If nothing else, the institution structures interaction simply by bringing the people together in a context already saturated with meaning. This is the terrain of institutionalized activity, and it makes interaction productive and economical with a minimal need for cumbersome negotiations.

In this way, institutions solve the deep theoretical problem of agency and structure (Goodin 1996: 17). People have free will, but they are also densely constrained by institutions -- how to reconcile this seeming tension? If we overemphasize free will, then it will be hard to explain how society hangs together at all. But if we overemphasize the determining effect of institutions then we will never explain innovation, resistance, irony, or the juggling of multiple commitments (cf Hodgson 1999: 130-131). One approach is to view society as a set of free agreements among individuals, each perfectly free but choosing to constrain themselves pairwise. This is progress, but it is too weak to explain relationships between strangers. It predicts that we will see a lot more laborious negotiating and agreeing than we actually see. Instead people rely overwhelmingly on conventionalized "agreements" called institutions. Free will supposes self-consciousness, but people are only partially conscious of their institutions. They derive their personal identities from institutional roles, they take rules for granted, they engage in routinized strategies, they have limited information, and they have limited capacity to reason with the information they do have. North (1990: 107) says that "[w]e cannot see, feel, touch, or even measure institutions; they are constructs of the mind", but this supposes that institutions are available in the mind for conscious reflection. But institutions happen first and foremost in the habitual categories of thought and routines of interaction. People do exercise their free will, but only against the background of this embedding in ongoing institutional life. Institutional theory thus draws a complex line between free will and its limits.

The interactions that institutions organize are not necessarily cooperative; indeed, an important role of institutions is to provide a structure for conflict (Commons 1970: 23-35, Offe 1996: 204). This is obvious enough in cases like collective bargaining and debates in legislatures, but it is also true in the fine details of everyday life. For example, many institutions set rules to reduce conflict:

"Many of these institutions regulate the social use of space and/or time, as is the case with parks, market squares, vacation trips, carnival, bank holidays, birthday parties, sports stadia, political party conventions, bars, music halls, and many others. They have in common that they evoke and legitimate certain themes, routines, and orientations and thus facilitate the interaction (or economize on the transaction costs) of those involved, and restrict this interaction to a particular range of themes, practices, entitlements, and premises" (Offe 1996: 205).
Offe (1996: 205) strikingly refers to these institutions as "an exoskeleton of social life".

f. simplifies life

Institutions reduce uncertainty in everyday life (North 1990: 3). They divide labor and eliminate unnecessary negotiation, thereby allowing people to focus their limited attention in a few areas of their lives. They may not be perfect, but they are usually better than reinventing the wheel every day:

"A well-functioning institution unburdens actors from purposive and strategic considerations, as an institutionally prescribed course of action can be trusted to yield beneficial or at least tolerable outcomes. Thus institutions allow for instrumental concerns being displaced by a healthy dose of ritualism and conservatism. Once established and widely supported, institutions, as it were, fly by themselves due to the invisible operation of an autopilot (Offe 1996: 200)."
In this sense they provide "a supraintentional framework" (Offe 1996: 201) that coherently structures and coordinates human activities well before anyone conceives a specific plan.

"... in a world of uncertainty, ... institutions play a functional role in providing a basis for decision-making, expectation, and belief. Without these 'rigidities', without social routine and habit to reproduce them, and without institutionally conditioned conceptual frameworks, an uncertain world would present a chaos of sense data in which it would be impossible for the agent to make sensible decisions and to act" (Hodgson 1988: 204).
g. constrains and enables

Institutions constrain, but they also enable. Indeed, they enable by constraining. "Money facilitates transactions because none of the parties can affect its value; democracy only transfers power because minorities are protected" (Offe 1996: 207). Institutional constraints are pervasive (North 1990: 5), but a well-designed institution compensates for these constraints by enabling a vast space of potential actions:

"On the one side, institutions such as the market, the university, the party system, general elections, or the business firm make an almost unlimited range of choices available which can be selected by combining and recombining the resources and interactions possible within them, thus allowing for the maximization of utilities or the refinement of styles and tastes. On the other hand, this option-generating arrangement itself must be relatively immune from choice" (Offe 1996: 206).
For Commons, this is the central point.

"... collective action is the general and dominating fact of social life. Human beings are born into this process of collective action and become individualized by the rules of collective action. Thus an institution is collective action in control, liberation, and expansion of individual action" (Commons 1970 [1950]: 21).

"Collective action means more than mere 'control' of individual action. It means liberation and expansion of individual action; thus, collective action is literally the means to liberty" (Commons 1970 [1950]: 35).

Commons' formulation here is controversial in several ways. He generalizes over institutions, pointedly making no distinction between state institutions and others. His prototype is collective bargaining in industry, and his overgeneralization of this prototype led him to an approving comparison between American interest group politics and Mussolini's corporate state (1970 [1950]: 33). Hayek (1960: 17), by contrast, draws a strong distinction between state and non-state institutions, on the grounds that only the former can legally engage in coercion (in the sense in which he uses the term). Hayek also attributes to Commons a conflation between liberty and power -- specifically, the power to act without any constraint. Hayek's is thus a negative conception of liberty -- freedom from coercion -- whereas Commons' is a positive conception -- the wherewithal to do something in particular.

But Commons and Hayek agree that institutions greatly narrow a recurring conflict over the nature of markets. "Free markets" and "constraining institutions" are not opposites (Hodgson 1988: 178); indeed, markets can only function when they are provided with a robust institutional framework (Hayek 1960: 220-222). This is also the central premise of North's work.

h. has both formal and substantive aspects

Some theories of institutions, especially in economics, emphasize their formal aspects: the ones that can be specified with rules, concepts, finances, and other finite structures. Thus North (1990), for example, explains the evolution of economic institutions almost entirely in terms of the costs of economic exchange. Other theories of institutions, especially in sociology, emphasize their substantive aspects: organizational routines, collective memory, and other aspects of the institution as a social organism. Thus Hodgson (1999: 60) paints this deliberately un-economic picture of the same phenomena:

"Socio-economic systems are essentially and unavoidably built up of historically layered and densely entangled institutions and routines. The more advanced the society, then the more complex the institutions and the more dense the entanglement. These institutions store and support both tacit and explicit knowledge. In customs and traditions, the knowledge of the past is accreted."
(Political scientists are divided into warring camps, with some rationalists on the formalist side and others emphasizing substance.) Both sides are half right, and no institution is half comprehended until its formal and substantive sides have both been elucidated (Offe 1996: 202-203). Fortunately, an increasing number of syntheses have reached across this disciplinary divide (Swedberg 1990).

i. aligns with both instrumental and cultural norms

Institutional theorists contend that people participate in an institution for two somewhat independent reasons: because they believe in it and because it delivers practical benefits. Thus Offe (1996: 201) speaks of "internal socialization and external effectiveness", "doing things 'the right way' and 'getting things done'", and "inculcat[ing] duties and generat[ing] outcomes" (cf March and Olsen 1989: 23). In disciplinary terms this distinction tends to align with the previous one, with formalists focusing on the instrumental aspects of institutions and substantivists focusing on norms. Both sides are necessary, and Dryzek (1996: 104) and Offe (1996: 218) both suggest that institutional norms and discourses can be understood as "software" -- a kind of "cultural infrastructure" (Offe 1996: 218) -- and that "[i]nstitutional hardware exists in the form of rules, rights, operating procedures, customs, and principles (Dryzek 1996: 104)".

"But in the case of institutions, unlike that of computers, the software on which they depend for their operation is not easily exchanged or replaced, as it is generated by the hardware itself in the process in which people 'get used to' and 'make sense of' or 'cope with' the institutions, thus adopting a set of standards, obligations, and expectations that in Weber's terminology is referred to as the 'spirit' (Geist) of institutions. This includes, in addition to moral commitments, familiarity with the institutions, codes of appropriate conduct, a reasonable measure of trust in their proper functioning, and the like" (Offe 1996: 218).
Institutions live mainly not on paper but in people's lives. It is in the details of everyday life that people will cooperatively enact the institutions and hold one another to the expectations they create. If people do not comprehend or believe in an institution then it will eventually fall apart (Offe 1996: 219). One cannot impose an institution on people unless they hold beliefs and values that accord with it. And one purpose of the institution is to instill and reproduce the values that it requires.

"... institutions embody normative intuitions or principles of those who live in or under the institutions in question. The relationship between institutions and social norms is, however, not unilateral, but reciprocal and cyclical. Social actors generate, support, and enact institutions, and these institutions, in turn, generate social agents capable of observing social norms (Offe 1996: 199)."
The instrumental success of the institution will reconfirm those values, but the institution will have no chance to succeed instrumentally unless people already believe in it. Because of this reciprocal relation between institutions and norms, attempts to design new institutions must confront a chicken-and-egg problem.

j. persists

Institutions can persist largely unchanged for centuries, even when faced with well-organized campaigns to overthrow them. They "have a stable and inert quality, and tend to sustain and thus 'pass on' their important characteristics through time" (Hodgson 1999: 143). This obstinacy needs explaining. One problem is simply that, because the institution has shaped the thinking of its participants, and because of their partial awareness of what the institution actually consists of, it is hard to imagine alternatives (DiMaggio and Powell 1991: 11).

"... the strong, mutually reinforcing interaction between social institutions and individual cognition provides some significant stability in socio-economic systems, partly by buffering and constraining the diverse and variable actions of many agents. Institutions become cumulatively 'locked in' to relatively stable and constrained paths of development" (Hodgson 1999: 144).
So long as an institution remains functionally viable, the incentive to change it may be lacking (Offe 1996: 208). When power among the different groups in an institution is not balanced, reinforcement politics continually steers new computin resources toward the interests of the most powerful (Danziger et al 1982).

Valid as they are, these explanations are relatively simple. At a deeper level, institutions persist because they are dispersed, in several ways:

(1) Individuals hold one another to the customs of the institution, and even if they could conceptualize different customs, it would be exceedingly difficult for them to coordinate a shift to them.

(2) Many individuals have placed bets on the institution by investing in career strategies that presuppose the institution's continued existence. These include investments in social networks that locate the individual strategically and help to establish local power centers. It would be nearly impossible to negotiate a change in the institution that would preserve the value of the investments they entail.

(3) The ideas, practices, and artifacts that make up the day-to-day activities of enacting the institution are numerous; they are acquired and comprehended somewhat independently of one another; their underlying unity is never drawn out and represented to anyone; and they tend to reinforce one another. As a result, questioning or replacing a few of them is impractical because the others will tend to reinstate them

Moreover, these modes of persistence reinforce one another.

Of course, one can overemphasize institutional rigidity, and institutions can fail. They might become so rigid that they cannot adapt to changes in their environments. Their internal social dynamics might blow them apart, for example by destroying their ecosystems or immiserating some of their participants. New applications of technology might undermine their former workings. Their participants might learn about another institution that they find more attractive.

k. depends on the persistence of other institutions

To a society saturated with the assumptions of modernism, these qualities of inflexibility, rigidity, and unchanging persistence will sound like bad things. So it is important to appreciate just how profoundly even ordinary activities depend on the persistence of institutions, and the extent to which institutions themselves depend on one another.

"Even the most simple economic activities rely on a taken-for- granted network of institutional supports. Ludwig Wittgenstein used the example of signing a cheque. Such an act depends upon the prior existence of many institutions, routines and conventions -- banks, credit, law -- that are the practical antecedents and frameworks of socio-economic actions and interaction. Without such institutions the activity would be hopeless. Similar remarks apply to other everyday activities, such as posting a letter or waiting for a bus. In every case, we habitually and unthinkingly depend upon a dense network of established institutions and routines" (Hodgson 1999: 66).
In many ways, we value institutions because of their stability (Goodin 1996: 22-24, Hodgson 1988: 190). Their persistence rewards investments in skill and learning, and they help form expectations of the future (Hodgson 1988: 191). Innovation, on the other hand, means more rapid depreciation of learning (Casson 1994: 152). Markets in particular depend on a tremendous complex of institutions. Describing the case of buying or selling a house, for example, North (1990: 63) observes that:

"The particular institutions matrix of this housing market consists first of all of a hierarchy of legal rules derived from provisions of the US constitution and the powers delegated to the states. State laws defining the conveyance characteristics of real property, zoning laws restricting which rights can be transferred, common and statute law undergirding, defining, or restricting a host of voluntary organizations -- all of these influence transaction costs. Realtors, title insurance, credit bureaus, and savings and loan associations that affect the mortgage market all will be influenced. The efficiency of these organizations is a function of the structure of property rights and enforcement and of the capital market (including voluntary as well as governmental guarantees and subsidies and other instruments that exist in the capital market). Equally important are the informal constraints that broadly supplement and reinforce the formal rules. They range from conventions of neighborhood conduct to ethical norms defining degrees of honesty in information exchange between the parties involved."
A stable institutional background also provides the conditions for economic growth and technological innovation.

"A 'good' institution is one which creates stable conditions for growth while at the same time not obstructing change. There is a real dilemma here. Since it is a major function of institutions to create stability they will inevitably have a tendency to produce inefficiency from time to time relative to the forces of production. On the other hand, total flexibility of institutions does not make much sense. If institutions are overly flexible they will be changing all the time which, of course, creates nothing but insecurity and instability. It is this duality in the function of institutions which is so fascinating" (Gunnarsson 1991: 64-65).
As a result, the limitations of an institution can be crucial to its success. Consider the example of a legislature, which creates institutions by passing laws. Most of those laws will remain stable for years before the legislature revisits the issue again, simply because the legislature is only capable of visiting a handful of major issues in a year. A law, once passed, settles the issue for a while, and everyone, whether they are happy with the outcome or not, has a reasonably predictable framework for action for the foreseeable future. A vastly more efficient legislature might be more productive in a narrow sense, but it might also disrupt the society beyond repair. All interest groups would regard every issue as permanently open, and nobody could make plans.

Other paradoxes of institutional life are even more troubling. Even if it functions poorly, a rigid institution -- an economic system, perhaps -- is at least reliable, and this very reliability can contribute to its legitimacy. The institution is rigid partly because its participants are only partially aware of its workings, and because they cannot imagine alternatives. Yet the institution, through its rigidity, makes a wide range of actions possible. One resists the logical implication here, that freedom requires incomplete awareness, and the further implication that criticizing rigid institutions, by increasing awareness of them, can undermine them and thereby decrease freedom. It is thus crucial, whenever the opportunity arises, to design one's institutions correctly the first time.

l. path-dependent

Institutions take form in different ways, whether suddenly or through evolution. But however they arise, institutional theorists of every political strip agree that they reflect a particular pattern of social relationships. North, for example, once held that new economic rules emerged to make markets more efficient; he has since abandoned that view (1990: 7), and now contends that "[i]nstitutions ... are created to serve the interests of those with the bargaining power to devise new rules" (1990: 16). Writing in a political context, Goodin (1996: 10) likewise characterizes institutions as "ossified past practices and the power imbalances and bargaining asymmetries embodied in them". Because institutions persist, they do not continually reconfigure themselves to adapt to new circumstances. Instead, they become "locked in". North (1990: 7-8), for example, speaks of "the lock-in that comes from the symbiotic relationship between institutions and the organizations that have evolved as a consequence of the incentive structure provided by those institutions". Even the English language continues to reflect ancient social relationships among the Anglo-Saxons (who contributed words relating to towns and farming), Norse (sailing and fighting), and French (religion and bureaucracy).

In economic terms, institutions are path-dependent (Offe 1996: 208; cf Hodgson 1999: 139) because of "increasing returns": the tendency of institutions to produce effects that further entrench their hold over society (Arthur 1989, North 1990: 92-95). Once individuals learn the ropes of a given institution, that institution will have an advantage over its potential competitors, and so they will go ahead and learn even more, thus reinforcing the advantage. Likewise, as a critical mass of individuals commits to a given institution, the benefits of compatibility will attract others to that same institution. North (1990: 101) observes that the "tenacious survival of institutional constraints in the face of radical alterations in the formal rules of the game is the best evidence of the increasing-returns characteristics of an institutional framework". Thus in order to understand an institution one must understand its history. An understanding of its current-day environment and functions will not suffice. At the same time, these history-dependent effects are controversial because their magnitude is intrinsically hard to measure. We can point at seemingly arbitrary or dysfunctional features of an institution, and we can demonstrate that they are holdovers from an earlier situation, but how can we really know that the institution might have turned out importantly different?

m. nested

Institutions are often nested inside of other institutions (Goodin 1996: 23). For example, a contract between two parties is an institution, but it is typically nested inside a larger institution, such as a type of contract that has become conventional in an industry. That nesting is convenient because the conventional contract has probably acquired a body of case law in the courts, thus making it more predictable than a contract that the parties might try to write from scratch. The conventional contract, in turn, is nested within the institution of contract law as such. Similarly, a specialized disciplinary language such as "hacker jargon" (Raymond 19xx) is an institution that is nested inside of the English language, relying on its rules and resources but also extending them. A large church can provide the institutional framework for a hundred specialized groups: bible study, day care, singles, political action, and so on. Local traffic laws are nested inside of state laws.

Nesting provides a convenient mechanism by which an institution can adapt itself to new circumstances (Offe 1996: 209). Perhaps the most famous example is the English common law, whose rules emerge inductively from the disputes that people bring into the courts. Thus Commons (1970 [1950]: 21-22) observes that "property relations are not something fixed and permanent. They are undergoing change all the time within the processes of collective action". Legislatures and universities, likewise, have their own internal rules, which include metarules specifying how they can be revised. As a general matter, the "outer" layers of institutions are more stable than the institutions that are nested within them, and the rights of minorities are typically protected by requiring supermajorities before the more fundamental institutions can be changed, for example amending the Constitution.

References (partial list)

W. Brian Arthur, Competing technologies, increasing returns, and lock-in by historical events, The Economic Journal 99, 1989, pages 116-131.

Mark Casson, Economic perspectives on business information, in Lisa Bud-Frierman, ed, Information Acumen: The Understanding and Use of Knowledge in Modern Business, London: Routledge. 1994.

John R. Commons, The Economics of Collective Action, Madison: University of Wisconsin Press, 1970. Originally published in 1950.

James N. Danziger, William H. Dutton, Rob Kling, and Kenneth L. Kraemer, Computers and Politics: High Technology in American Local Governments, New York: Columbia University Press, 1982.

Paul A. David, Standardization policies for network technologies: The flux between freedom and order revisited, in Richard Hawkins, Robin Mansell, and Jim Skea, eds, Standards, Innovation and Competitiveness: The Politics and Economics of Standards in Natural and Technical Environments, Edward Elgar, 1995.

Paul J. DiMaggio and Walter W. Powell, The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields, in Walter W. Powell and Paul J. DiMaggio, eds, The New Institutionalism in Organizational Analysis, Chicago: University of Chicago Press, 1991.

John S. Dryzek, The informal logic of institutional design, in Robert E. Goodin, ed, The Theory of Institutional Design, Cambridge University Press, 1996.

Robert C. Ellickson, Order without Law: How Neighbors Settle Disputes, Cambridge: Harvard University Press, 1991.

Harold Garfinkel, Studies in Ethnomethodology, Polity Press, 1984. Originally published in 1967.

Robert E. Goodin, ed, The Theory of Institutional Design, Cambridge University Press, 1996.

Christer Gunnarsson, What is new and what is institutional in the New Institutional Economics? An essay on old and new institutionalism and the role of the state in developing countries, Scandinavian Economic History Review 39(1), 1991, pages 43-67.

Friedrich A. Hayek, The Constitution of Liberty, Chicago: University of Chicago Press, 1960.

Geoffrey M. Hodgson, Economics and Institutions: A Manifesto for a Modern Institutional Economics, Cambridge, UK: Polity Press, 1988.

Geoffrey M. Hodgson, Economics and Utopia: Why the Learning Economy Is Not the End of History, London: Routledge, 1999.

James G. March and Johan P. Olsen, Rediscovering Institutions: The Organizational Basis of Politics, New York: Free Press, 1989.

James G. March and Johan P. Olsen, Democratic Governance, Free Press, 1995.

Douglass C. North, Institutions, Institutional Change, and Economic Performance, Cambridge: Cambridge University Press, 1990.

Patrick H. O'Neil, Revolution from Within: The Hungarian Socialist Workers' Party and the Collapse of Communism, Northampton, MA: Elgar, 1998.

Claus Offe, Designing institutions in East European transitions, in Robert E. Goodin, ed, The Theory of Institutional Design, Cambridge University Press, 1996.

Michael J. Sandel, The procedural republic and the unencumbered self, Political Theory 12(1), 1984, pages 81-96.

Richard Swedberg, Economics and Sociology: Redefining Their Boundaries: Conversations With Economists and Sociologists, Princeton: Princeton University Press, 1990.

Oliver E. Williamson, The Economic Institutions of Capitalism: Firms, Markets, Relational Contracting, New York: Free Press, 1985.