Notes on Exchange Theory and Rational Choice Theory

Arthur W. Frank

For Sociology 333

The first problem is that of names. Homans’ famous paper, "Social Behavior as Exchange" led sociologists to refer to his work as exchange theory. Homans never liked this term, preferring instead "social behaviorism". As important as exchange is within Homans’ work, he wanted to emphasize the behaviorist aspect (see p. 288b) of his work. That is, Homans wants theory to explain whether observable behaviors increase or decrease based on actors’ rewards and costs. Homans is a psychological reductionist (288a): his interest is the individual who then enters exchange relationships, in which social rewards and costs determine individual choices.

Blau on the other hand does offer an exchange theory. His interest is in how exchange as a form of social activity gives rise to different forms of association and different organizational forms. There is then a gap between Homans and Blau and the rational choice theorists, Coleman and Hechter. In rational choice theory the emphasis is on Coleman’s question, "Why do rational actors create obligations?" or any of the other things they create; Hechter’s interest is in why rational actors create forms of organizations that yield "goods". Coleman and Hechter’s "rational actor" has a good deal of resemblance to the actor in Homans’ social behaviorism, but rational choice theorists are more interested in the emergence and persistence of meso-scale organizations.

These comments are introductory. What counts is to recognize that among these

four theorists, there are at least three different forms of theory (social behaviorism, exchange, and rational choice) with three related but also different interests.

In my notes on interactionism I suggested that most theories have at least five components; let me review these with respect to social behaviorism, exchange theory, and rational choice theory (which I will now treat as one, with some disregard to what I have written above).

1 – Philosophical anthropology: these theories all emphasize that people are actors who then choose (rationally) on what terms to become members of society (in contrast, for example, to Schutz’s notion of society as "always already there" or Mead’s idea of the self being formed in social relations). Actors choose lines of action. Though Coleman (p. 302a) acknowledges that some choices may reflect internalized social norms, the emphasis is on "external rewards". Choice of action is privileged over response to the demands of membership in a collectivity. Actors are maximizing, seeking the most rewards at the least costs. Their rationality consists of evaluating which actions can be expected to maximize their interests.

2 – Basic questions are less about society as a whole ("social order") and more about why actors pursue courses of action and how those actions become aggregated into regularities of collective behavior (see Hechter’s definition of a social institution, p. 305b).

3 – Recurring concepts include exchange, obligation, reciprocity, and Coleman’s important idea of social capital. These concepts represent readily observable activities; the object is to say how and why actors engage in these activities, and what are the results of their aggregate behavior. Concepts are then used to generate propositions, which are predictive statements in the general form, x will increase (or decrease) in the presence (or absence) of condition y. Thus Homans writes that the rate at which the pigeon pecks will decrease as the pigeon becomes satiated on corn, or in its more general form, the rate of a behavior will decrease as the reward becomes progressively less valuable to the actor.

4 – The research program is to generate propositions that can be tested, or at least verified, by empirical studies. The empirical studies range from Homans’ laboratory experiments to anthropological and organizational studies (for example, Blau’s study of consultation in an office, described by both Homans and Blau). The question is: how does the theory explain the emergence and persistence of behavior. Homans was resolute in wanting propositions that would accumulate, progressively larger scale forms of social organization as they accumulated. Later theorists are more willing to begin at the "meso" (organizational, institutional) level. But the existence of the organization is never taken for granted. The research question always gets back to why it’s in the actor’s interest to create organizational forms and continue to participate in these organizations.

5 – Finally, rational choice theory is perhaps the strictest contemporary theory group concerning its boundaries: which studies are part of its program and which are not. In this sense it is perhaps the last of the 1970s style "theory groups." The key boundaries are, first, explaining from the perspective of the rational actor, or as Hechter calls it, the "rational egoist" (p. 309b). The second boundary limit is that theory should produce propositions that can be evaluated by whether they predict observable behavior.

Thus in terms of Alexander’s action/order matrix, both exchange and rational choice theories are resolutely in the individualist/rational cell. Certainly collective institutions are acknowledged, but they must be explained as arising from and persisting because they allow individuals to rationally maximize benefits through their involvement. And that again is the constant research question: why is such-and-such behavior a rational, maximizing, course of action?

Alexander argues—and I agree—that this theoretical program is not free of contradictions. Specifically he claims that Homans eventually needs to introduce his notion of "sentiments" in order to explain why people—unlike pigeons—do things that don’t produce rewards, e.g., all forms of altruistic behavior. Homans answer is stated on 289a, bottom: "The problem is not, as it is often stated [by Parsons, with whom Homans is debating throughout this article], merely, what a man’s values are [Parsons’ normative orientation and latent pattern]…but how much of any one value his behavior is getting him now." The problem with this statement is what Parsons and other functionalists are often accused of: tautology, or proposing as an explanation that which you are trying to explain (see Blau, p. 316b). In Homans’ case, while the actor continues to do something, the observer can claim the actor is continuing to derive more value from that action. Then when the action stops, that’s when the actor no longer derives sufficient value. "Value" is the explanation that actors act because the action is "valuable". I hope that interpretation of exchange theory is not unfair; certainly exchange theorists would dispute it, but I’m not sure how. In fact there are very few exchange theorists left; they’ve all become rational choice theorists.

The problem of Homans’ social behaviorism is that his methodological commitments force him to treat "values" as if they were corn pellets. By introducing values he makes some space for non-rational behavior; by treating values as corn pellets he keeps his theory behaviorist. The problem is that values can’t be counted, only inferred. Thus when the behavior stops, the observer can hardly be wrong by claiming that the behavior is no longer valuable to the actor who quit doing it. The theory can’t be falsified.

I hardly want to trivialize Homans, who was one of the most intelligent sociologists of his generation. His formidable intelligence gave him a charisma of sorts, and his work is both lucid and persuasive (curiously, he is certainly sociology’s finest autobiographer). Just as Blumer taps into our self-recognition that objects have meaning for us, Homans taps into our sense that we act because we find it "rewarding" or we don’t act when that action would be too "costly". The problem is that Homans ties this recognition to a behaviorist research program that is out of fashion today, to say the least. The other problem is that while Homans had a genius for restating others’ research findings in his own theoretical vocabulary (reducing the behavior to a series of exchanges and then finding why these exchanges were valuable to the actors), the question is: is this the most useful and interesting way to explain the research? What Homans forces us to do, ultimately, is to suggest (in equally clear, persuasive language) why an alternative theoretical framework is "better".

Among the ideas Homans discusses in his paper—and this paper bears careful study—is that of "opportunity cost" (292b). Homans does not use this term, but he refers to the concept when he writes "that the cost of a particular course of action is equivalent to the foregone value of an alternative". In other words, whatever you do has among its costs all the other things you are not doing (or buying, or investing in, etc.). This idea leads him to a nice example of a more general proposition: "if we define profit as reward less cost, and if cost is value foregone [opportunity cost], I suggest that we have here some evidence for the proposition that change in behavior is greatest when perceived profit is least" (293a). In other words, people will tend to quit doing things the way they have—they will change—when the way they have done things yields least rewards. Or, it’s easy to get a student to change her or his study habits when she or he starts getting Ds; or, Franklin Roosevelt could persuade Americans to change their orientation to welfare during the Great Depression. Before 1929, when the economy was booming, change would not have been predicted (would you predict any significant change at the political level today? why or why not?).

If the actual content of propositions is not that startling (or "non-obvious" as Randall Collins claims sociology needs to be), that would not have bothered Homans too much. His objective was to build a verified set of cumulative propositions, in order to show how institutions and forms of social organization develop from individual behavior that is rational in its pursuit of rewards and avoidance of costs. As he says at the beginning of his article, "those massive structures called ‘classes,’ ‘firms,’ ‘communities,’ and ‘societies’ must ultimately be composed" of two or three persons who are "in a position to influence one another" (287b). In this social behaviorism of interpersonal influence, we are each constantly "conditioning" each other’s behavior by offering rewards and creating costs. Of course one major reward we all seek is other people’s "regard" (see 294b). But Homans’ point, against Parsons, is that the value of "regard" is not derived from the internalization of norms (on norms see 290a), but from the reward value of "regard". Thus there is a rational pursuit of non-rational rewards.

Blau’s version of exchange value makes a useful complement to Homans. Note that Blau has dropped the social behaviorism and psychological reductionism. His interest is how the "structure" of organizations is sustained by actors engaging in exchanges between themselves, and these exchanges are entered into (or not) based on calculations of rewards and risks. Yet among these readings Blau’s philosophic anthropology is perhaps the most open to the widest variety of rewards, see p. 315a. People in Blau’s world seem actually to enjoy each other, something I don’t see in Homans’ world.

Blau takes the exchange itself as the unit of analysis, though actors are using exchanges for their own purposes. These purposes are longer-term and subtler than behaviorism is able to consider. Actors seek to build networks of obligation; they "establish partnerships" (318b, though read the context). Blau is particularly interested in how exchanges can produce "both social bonds between peers", that is relations of equality, and at the same time, "differentiation of status" (317b-318a). His eventual explanation is: "the contingency [the particular social circumstances] that determines whether social exchanges lead to friendships between peers [equality relations] or superordination and subordination [hierarchy] is whether benefits received are reciprocated or not" (p. 319a). The question of whether obligations are reciprocated also figures significantly for Coleman and Hechter. For all of them, in different ways, societies are networks of exchange, and the problem (of order, in Alexander’s terms) is whether all actors fulfill their reciprocal obligations (see Hechter’s discussion of the free-rider problem, discussed below, as one variation on this theme).

James Coleman’s excerpt on social capital is one of the most important readings in the entire collection, and the study of social capital is one of the most active contemporary research interests. If Homans’ behaviorism, and even Blau’s exchange theory, are pretty much gone by the wayside, social capital is near the top of the sociological agenda with the most prestigious journals featuring whole sections debating studies that utilize this concept.

The idea of social capital is clearly stated in 297a-b. Social capital is always a relation, and it is a process. "Social capital…is created when the relations among persons change in ways that facilitate action" (297). This sounds simple, but as the rest of the section shows, it has a variety of implications. On 298 Coleman writes that the "value of [social capital as a concept] lies primarily in the fact that it identifies certain aspects of social structure by their function." Unlike Homans, Coleman has no interest in distancing himself from Parsons; he was self-consciously revising Parsons’ early theory of the unit act (the actor seeking an end, his choice of means determined by the normative orientation). Nor is Coleman afraid of talking about social structure. Yet social structure is understood by what he later calls "the micro-to-macro transition." Like the other three, Coleman is committed to building any sense of social structure up from the individual to the collective. What, then, does he mean by function, which in Parsons’s usage implies a macro-to-micro transition, the antithesis of rational choice theory?

Coleman writes: "The function identified by the concept ‘social capital’ is the value of those aspects of social structure to actors, as resources that can be used by actors to realize their interests" (298a). So again, "function" now works micro to macro. As I understand Coleman, his argument is that actors enter into relations (interpersonal obligations through organizations) because they can, in these relations, achieve something they could not achieve by themselves. Having entered into these relationships, actors then (without necessarily intending it) "create" institutions; thus social capital itself is often, as Coleman writes, "a by-product of activities engaged in for other purposes" (303a; cf. Merton’s idea of unanticipated consequences). Public goods, which it would not be in actors’ self-interest to create, are often generated as "by-products" of relationships entered into because they do have direct rewards (302a).

To get a sense of how powerful Coleman’s argument is, I suggest careful study of his examples. My favorites are, first, his explanation of why rational individuals incur obligations (300b-301a). The favor that is owed is a kind of currency, and Coleman shows that the network allows actors to do favors when it costs them little, and then claim a return favor when they truly need it. He doesn’t credit Simmel, but the argument is straight from Simmel on exchange: exchanges can add value, Simmel argues, because the objects are each more valuable to the one who receives them than they are to the person trading them off (think of a garage sale, in which the seller’s old junk is just what the buyer needed; thus value is added). Of course the other person can claim a favor when you have least resources, and it will be costly to honor your obligation, but Coleman, like Goffman, sees the risk in social life.

My other favorite example is the medical one on 300a-b. Coleman starts with a practical problem: why are there so many malpractice suits in U.S. medicine? He argues that in the old days, physicians were crucially important and people did not perceive themselves paying their doctors adequately in proportion to the value of the medical service. What he calls the patient’s "felt obligation" became "a form of social capital". How? We need to ask what action was facilitated; in this case, medical practice could get on without being belabored by malpractice suits. Then times change, and Coleman suggests four changes: the loosening of physicians’ monopoly on expert knowledge, fewer personal relationships between physicians and patients, high incomes for physicians, and increase of liability insurance (thus the patient knows it’s the insurance company that will have to pay the claim, not the physician him or herself). The "social capital that protected physicians" is thus reduced, and suits follow.

My question is whether the logic of social capital doesn’t require that given these new relations, some new "action" is facilitated. One new action is more malpractice suits, which may be said to depend on the new social capital of increased distance of patients from physicians, a sense that physicians are paid plenty for their services, etc. Another new action is the new class of non-physician medical administrators, as part of the "medical-industrial complex". What I’m wondering is why Coleman doesn’t explore how the "new" social capital creates these possibilities, even as it withdraws "protection" from the old relationships. My point is that social capital is inherently neutral, and the "changes" it facilitates may be, to my values or yours, negative as well as positive. (Apparently when Coleman was dying and in hospital, he wrote a paper analyzing the medical relationships he observed around him; I’d love to see it.)

Coleman’s paper is worth your most careful study. His was truly a sociological imagination. I’d draw a comparison to Harvey Sacks’ lecture, which is as far from Coleman’s rational choice as you can get in one sense. But in both articles you can see "social structure" emerging from everyday actions such as answering the telephone or joining the PTA (which is, incidentally, the Parent-Teachers Association). My only regret is that Coleman’s Foundations of Social Theory from which your reading was taken is about 1,000 pages, which is a long night’s read.

Finally we have Michael Hechter’s article, which you should spend some time with, but it would not have been my choice for inclusion. It fits the collection because Hechter is interested in why it’s rational for actors to enter into the formation of certain types of institutions. His notion of "cooperative institutions" is particularly indicative of rational choice theorizing. Such institutions provide a "jointly maximizing strategy" in which each can get more together than either could get alone (306b).

Hechter’s article is also important (besides Hechter being perhaps the most important living rational choice theorist) because the "free-rider" problem is one of the oldest chestnuts in this form of theorizing. Start at the end, 310b, with Mancur Olsen’s discussion of unions, which I read when I was a graduate student (it’s that old). Some employees do the work of forming a union, pay dues, negotiate, maybe lose wages while on strike, and eventually secure a better contract. The contract, however, applies to all employees, including some (the free-riding swine) who never joined the union. You have a free-rider problem whenever you have a public good—something inherently available to an unrestricted group—that some people are paying for and others are able to avoid paying for. I believe the term goes back to roads, which are there for everyone to use whether or not each has paid the taxes that financed the road. When I drive on Montana roads, I’m a free rider, and vice versa when they drive up here. In this case it’s assumed that we will allow be free riders as often as our roads are ridden upon, and we don’t worry. Toll roads occur when there is worry that too many people are using the road without paying the taxes (or that tax money should not go for something only a segment of the public will use). Tolls are an example of what Hechter means by "control" that organizations have to develop to protect themselves from free riders.

Hechter’s problem is why people enter into those forms of regular collective behavior that he calls institutions (305b). If action is rational, everyone will try to be a free-rider, and so the only way institutions can form is to control, somehow, the problem of free riders. Hechter’s three tiers are the problems that an institution must solve in order to come into existence.

Now to be honest, I don’t understand everything Hechter says in this article, nor do I expect you to. In particular I don’t understand exactly how his explanation differs from Mancur Olsen’s. I was supposed to be on a panel with Michael Hechter this spring at the Pacific Sociological Association meetings, and I could have asked him about all this. Unfortunately I was given the wrong dates for the meeting, had a conflict with another meeting, and so rational choice theory will remain a mystery until the next opportunity. For your reading, know what the free-rider problem is (cf. Coleman, 302b; it’s everywhere), and read what he says about different forms of institutions on 305-307. What matters most in filling in your idea of rational choice theory is Hechter’s idea that "individual preferences must be aggregated into a collective design" (308a) in order to generate public goods in particular, and I’d say in order to sustain social life as anything worth that name. Going back to Parsons’ Hobbesian question of order, the rational choice response seems to be generated here: aggregations of individual preferences sustain order because they provide, more than occasionally, for "the greatest amount of good at the least (private) cost" (308a).

In conclusion, note the politics of all this. Homans is candid in writing that he, as an affluent and socially prominent Bostonian, wanted to create a sociology that would answer the 1930s Marxists whom he perceived (probably correctly) as attacking him personally. He could understand that they wanted some good part of what he had, and he was willing to discuss the division of the spoils so long as everyone agreed they were basically pursuing their own self-interest. "If we could only meet as honest men—or honest rationalizers—we might divide up the take without fighting. It was the intellectual guff talked about by the alleged leaders of the proletariat that put one’s back up and got in the way of a settlement" (Homans in Sentiments and Activities, 1962, p. 4).

Rational choice theory remains an inherently conservative project, though those involved might prefer Homans’ designation that it’s simply "honest". I tend to believe that one person’s honesty is another’s politics, and rational choice theorists would probably agree if I went on to describe politics as another form of the pursuit of self-interest. My critique is not that rational choice theorists are wrong in what they say, but that there’s a great deal they do not say much about (the disempowered, the suffering, those who lack the resources to exchange and thus can do little to maximize their positions, and those who care for them), and maybe cannot say much about. Sometimes a theory is best known by its silences, and rational choice theory is silent about much of what I find most significant in social life.