The primacy of economics: an explanation of the comparative success of economics in the social sciences.

Demsetz, Harold

I

The strong export surplus economics maintains in its trade in ideas and methods with the other social sciences is an important indicator of the success of economics. Not much has been said about the source of this success, but it has been attributed largely to advantages offered to other social sciences by the economics tool kit. Thus, Gary Becker, who has earned Commander-in-Chief ranking in the EEF (Economics Expeditionary Forces), attributes the interdisciplinary influence of economics to our relentless and unflinching application of "The combined assumptions of maximizing behavior, market equilibrium, and stable preferences..." (Becker [1976]). This view, but with a different emphasis, is shared in by Hirshleifer [1985], my distinguished colleague, who leads an active EEF group now reconnoitering sociobiology:

What gives economics its imperialist invasive power is that our analytical categories - scarcity, cost, preferences, opportunities, etc. - are truly universal in applicability. Even more important is our structured organization of these concepts into the distinct yet intertwined processes of optimization on the individual decision level and equilibrium on the social level of analysis. (p. 53)

The emphasis here is on the broad scope of phenomena that can be explained with our tool kit. Such breadth suggests the possibility of unifying the social sciences in an economics-led hegemony.

Brief statements such as these cannot be taken as adequate representations of full views, but there is a disposition among most economists to accept the interdisciplinary influence of economics as a measure of its success and to identify its tool kit as the source in this success. The most important exception is R. H. Coase [1977], who doubts that the interdisciplinary influence of economics will continue. He believes the economics tool kit is easily mastered by practitioners of other social sciences when they find it useful. As Coase judges the situation, a sustained influence from economics requires that economics and the other social sciences share interests in the same problems.

The nub of the disagreement between Coase and those, like Becker and Hirshleifer, who remain optimistic about a continuing strong influence from economics, turns on two issues that are not discussed explicitly. One relates to the definition of a "problem." Those who are optimistic seem to be distinguishing between problems according to whether their solutions do or do not depend importantly on scarcity and maximizing behavior. Interpreted this way, many problems across the social sciences are of the same type. Coase, on the other hand, distinguishes between problems according to differences in the factual knowledge needed to understand them. For Coase, understanding the firm is a different problem from understanding the marriage relationship because factual knowledge about one does not transfer to the other. For Becker, the essential similarity between these problems rests in utility maximization subject to constraints.

This difference in interpretation raises a second issue, one that is more substantive. Coase's view is that the tools of one social science discipline, in this case economics, are easily learned by practitioners of another, but that factual knowledge is not. Those optimistic about the interdisciplinary influence of economics implicitly have the view that mastering the tools of economics is more difficult than Coase presumes; repeated practice in their use is needed to become their master. It is difficult to judge which side has the best case. I am even puzzled as to how to define a researcher's discipline if he or she is trained in the tools of one social science and applies them to problems in another. One possibility is to determine whether the researcher's work appears in economics journals or in the journals of other social science disciplines. I am most familiar with interactive work between law and economics. Persons working in this interdisciplinary area but who were trained initially in economics seem to be publishing disproportionately in law journals. This would seem to mean that they are mastering the facts of rules of law and of the cases that help determine these; if true, this would seem to support the view of optimists rather than that of Coase. On the other hand, economists in law schools are being replaced rapidly by the increasing number of law professors who have been trained in the tools of economics, and this supports Coase's view.

The disagreement we have been discussing is relevant to the success of economics if this is measured by the interdisciplinary influence. The criterion of success adopted by the present paper differs from this. Economics may be judged the more successful social science because it has explained phenomena within its traditional boundaries better than the other social sciences have explained phenomena within their respective traditional boundaries. The primacy of economics may be established in this sense even if economics had never influenced other social sciences.

The relevance of this second notion of success can be illustrated by referring again to the question of interdisciplinary influence. There is undeniable truth in Becker's notion that we economists are much more prone to use our tools to reformulate and investigate problems in other disciplines than practitioners in these disciplines are prone to extend the use of their tools to economics, but why are we more confident than other social scientists? The answer, I believe, is that economists have been more successful in dealing with the traditional problems of their own discipline than other social scientists have been with the problems in their own disciplines, and, so, we are led to a source of primacy different from interdisciplinary influence - success within the traditional boundaries of a discipline. Success breeds confidence, and sometimes over-confidence.

A similar question can be raised about Hirshleifer's explanation of the success of economics - that our analytical categories, scarcity, cost, preferences, opportunities, are truly universal in applicability. If this is so, why have these categories emerged from the study of economic behavior and not from the study of behavior in the other social sciences? Political science need not have waited for Buchanan and Tullock's Calculus of Consent [1962] before applying individualistic utility maximization to political problems. Becker's Treatise on the Family [1981] need not have preceded the use of costs and benefits by sociologists to study marriage. Legal scholars, immersed as they have been in the study and resolution of conflict, could have seen the relevance of scarcity constraints without waiting for Tullock's Logic of the Law [1971] and Posner's Economic Analysis of Law [1972] to instruct them. And why did Posner not write his influential text before being exposed to economics?(1) That economics has sparked these important works indicates that the tools of economics, even if they are universally applicable, are not universally easy to develop. They emerge from the study of the economic system, and this brings us back again to a notion of primacy based on success within the traditional boundaries of economics.

Even Coase's view forces us to consider this view of success. The various social sciences may work on different problems, as Coase claims, but why has economics been more successful with respect to its problems than the other social sciences have been with respect to theirs?

II

The source of the success of economics, I shall argue, lies in the particular characteristics of the problem that has been central to its inquiry during the last two centuries. These characteristics are not equally present in the problems faced in other social sciences, and it is this difference that accounts for the greater success of economics. The problem I refer to is the search for an understanding and an evaluative judgment of resource allocation in a complex, decentralized system. Because its solution rested on the role of prices in guiding producers, investors, and households in the supply of and demand for resources in the production of goods for use by those who played little or no part in their production, it was the specialized and commercial aspects of decentralization that were especially important, and this is what I shall mean by decentralization.

This was not the only topic that interested economists, as the scope of topics indexed in Smith's Wealth of Nations and Marshall's Principles of Economics clearly reveals, but resolving the coordination problem of a decentralized economic system was the primary goal of both classical and neoclassical economists. Smith's Wealth of Nations, Ricardo's Political Economy and Taxation, Mills' Principles of Political Economy, Marshall's Principles of Economics, and Pigou's Economics of Welfare all have the analysis of decentralization as their main purpose.

The solution to the decentralization puzzle, hammered out during the first 200 years of economics, is a remarkable achievement. Three aspects of the inquiry have been instrumental in its solution and in causing this solution to be so much a part of the success of economics. These are (1) the large degree to which our understanding and evaluation of the economic system is significantly affected by knowledge of how decentralization works, (2) certain aspects of commercial activity that facilitated the solution of the decentralization problem, and (3) the successful modeling of the inquiry by our classical and neoclassical predecessors.

As to the importance of the inquiry, it is clear that most positive and normative questions about the economy, whether they are asked by economists, media persons, or policy makers, can be answered only if one understands how commercial activities are coordinated in a decentralized economic system. These questions include a familiar litany of topics - tariffs and international capital flows, taxes and growth, occupational choice and wage differentials, entry barriers and price-cost margins, agricultural programs and supplies of food stuffs, free agency and professional baseball, shopping malls and externalities, and so on. An understanding of decentralized commercial activity is required if one is to deal sensibly with a large fraction of the economic phenomena in which we and others are interested. It is not enough to understand how an isolated individual, a Robinson Crusoe before Friday, allocates the resources at his command so as to maximize his utility. Competition, price formation, and decentralization itself form no part of the solution to the Crusoe problem. Because of this, its solution is not very important.

The aspects of the commercial decentralization problem that facilitated its solution are two. Consider that whether a firm is purchasing inputs or selling outputs, whether a household is allocating income among goods or between consumption and saving, or whether the investor is allocating wealth across firms and time, these commercial activities offer the economist commensurate money value measures of activity. Theoretical recognition of and empirical support for broad generalizations are considerably more difficult if commensurate measurement is absent. The inquiry into decentralization is advantaged in this respect because commercial activity relies on monetary measurement and because markets reveal the prices upon which such measurement rests. The second facilitating aspect of commercial activity is that it is guided to a large extent by wealth considerations. The importance of this to the success of economics is discussed later.

Of course, the importance of a task signifies little for the success of a discipline if the task remains unaccomplished. The third reason for the success of economics is that the adroit modeling of its central inquiry by our classical and neoclassical predecessors allowed it to be substantially and successfully completed before World War II began.(2) The classical economists framed the problem in terms of resource allocation, and they demonstrated the importance of rational behavior guided by self-interest and constrained by competition. Neoclassical economists fine-tuned and clarified the analysis and made it more powerful by adding marginalist logic.

The modeling avoided complications that if attended to would have delayed discovery of the principles of decentralized coordination. The avoided complications are those we associate with planning, a legal system, crime, information cost, and non-commercial activity. No serious attempt was made to deal with the state, the firm, or the household. The characterizations of these institutional arrangements in neoclassical theory are correctly described as "black boxes," but they served a purpose by allowing economists to ignore planning, even of the limited sort that takes place within firms and households, and this enabled them to focus instead on coordination achieved through impersonally determined prices. This focus is different from one that seeks to understand the personal psychology of persons and the inner workings of the institutions through which personal interactions take place. Hence, tastes were taken as given and the organizations, such as firms, through which persons interacted on a personal basis were slighted.

Collusion was a mere footnote. And this is as it should have been, since collusion substitutes managed coordination, or planning, for decentralization. Crime and ignorance were also slighted. And this is as it should have been. The key distinction between decentralization and centralization is not the degree to which ignorance and crime are present but the degree to which individuals are able to control others. The closest approaches to the information problem came late in the neoclassical period when Mises [1936] and Hayek [1945] wrote noteworthy, but very general, comparisons of the quality of information available through socialist and market organizations. The nature of private ownership and of the legal system that underlies it were slighted because the components of these that are essential to a discussion of the price system relate primarily to only two features - the exclusivity and exchangeability of private rights.

These premises restricted the inquiry to that of understanding coordination in a private propertied, competitive, unplanned commercial system characterized by specialized activities and populated by well-informed, law-abiding, self-interested, rational persons who are strangers to each other but who exchange goods and services.

This conceptualization has been repeatedly criticized for excessively abstracting from reality. The criticism has been answered by defending abstraction (Friedman [1953]). Two notions underlie the defense. One is that a model's power to predict and not its realism is the arbiter of its usefulness. The second is that abstraction from reality is essential if a theory is to offer generalizable predictions.

I have no great disagreements about this, but my topic forces me to demur in some respects. Predictive power is an important arbiter of a theory's success, but this does not answer the substantive question. Is predictive power enhanced or impaired by abstraction? Abstraction is necessary for generalizations, but the more useful generalizations derive from the slighting of aspects of the problem thought to be unimportant and not from those thought to be very important, so what matters is the manner in which abstracting is, and is not, practiced. To be useful, a theory must make concessions to realism, especially with respect to variables thought to play more important roles in explaining the phenomena under study. Complete abstraction yields pure mathematics, not empirically relevant science. Complete accurate representation yields descriptive classification, not generalization. The truly useful theory is a mixture of realistic and abstract representation. I state this to emphasize my belief that the neoclassical conceptualization, contrary to what its critics say, is a realistic portrayal of those characteristics that are important in an inquiry into the commercial activities of large, decentralized economic systems. The most frequent criticisms of this portrayal are that people are not motivated exclusively by narrow self-interest and that they are not always rational in pursuing their goals. Each of these is considered next.

III

The realism of the narrow self-interest motivation cannot be judged without attention to the context within which decisions are made. Between close friends and family members altruistic motivations assume an importance not found in dealings between strangers. The purchaser of wheat does not know those whose efforts went into its production and delivery, and he or she seldom thinks sympathetically about these suppliers when deciding on purchases. Most activity in the large, decentralized, commercial economy is between strangers who are represented not by familiar faces but by impersonal prices. Realism is served, not sacrificed, if narrow self-interest is relied upon to interpret behavior in the commercial economy. Abstraction from reality in the decentralized economy would be extreme indeed if, instead of placing reliance on self-interest, the model were to assume that altruism plays a large role in decision-making; this is so not because persons in business differ from other persons but because the circumstances of choice differ.(3)

The relevance of circumstances is attested to by modern sociobiology. The propensity to behave altruistically has been shown to depend on the degree of genetic relatedness between the interacting parties. Many economists, especially those who are involved in interdisciplinary work, see this finding as supportive of the notion that economics ties its findings too tightly to "baser features" of man's nature. They are wrong in this. This sociobiological finding implies that altruistic behavior is unimportant if dealings are between persons unrelated to each other, and it is dealings of this sort, and not those between relatives and close friends, that are of paramount importance in the large, decentralized commercial system.

Economists need not look to sociobiology for this distinction. They have the writings of Adam Smith. In his Wealth of Nations, Smith characterizes behavior as steadfast in its pursuit of self-interest. In his Theory of Moral Sentiments, he characterizes behavior as often sympathetic to the feelings of others. These different descriptions of human behavior are not inconsistent with each other. They describe behavior under different circumstances. Interactions in the Wealth of Nations take place in the world of commercial activity. They are between persons who bear no kinship or friendship to each other and who, most often, are not even known to each other. The sympathetic feelings described in the Theory of Moral Sentiments are between persons who are related to each other or who are close friends. This arena is contained within the household or the small community. Smith even observes that the intensity of sympathetic feelings is correlated with the degree of relatedness between the interacting persons. These views are completely in agreement with modern sociobiology; the one element missing from Smith's works is a reference to the selfish gene, but, then again, Smith's works have not been searched with an eye toward uncovering this reference.

The argument just made merits repeating with respect to irrational behavior. Rational calculation may be undermined by emotional feelings. There is little doubt that some decisions have the appearance of irrationality, and that emotional considerations account for this appearance. "Appearance" rather than "fact" because emotions are not ad hoc. They are incorporated into our decision-making mental machinery through evolutionary or heavenly processes, and, in subtle ways, they presumably enhance survival or serve other useful goals. To describe their influence as deleterious is therefore to rush to judgment, but let us accept this judgment for the sake of argument. What types of decisions would be most subject to influence by emotions? It is difficult to make convincing argument that emotions play much of a role if dealings are between persons who do not know each other and who generally are reacting not to persons but to impersonally determined prices. Emotions will have more influence on interactions between family members, friends, and enemies. As between strangers, cool calculations of a rational sort are the order of the day.

In sum, the neoclassical conceptualization abstracts from reality when persons are interacting with fellow workers, friends, neighbors, and family, and it brings reality into the analysis when persons interact with others through the commercial world of the large, decentralized market economy. To assure yourself of this, ask and answer the following questions:

On the relevance of narrowly conceived self-interest:

* What fraction of commercial activity turns importantly on friendship? What fraction is significantly influenced by altruistic motives? Only a small fraction. (With apologies to economists who suffer from a desire to be sociologists.)

On the relevance of rationality:

* What fraction of business decisions are made on the basis of emotion? What fraction on the basis of systematic calculation? Answers: small to the first question, large to the second.

On the relevance of competition:

* How many goods are sold at prices high enough to yield significantly abnormal profit for long periods of time? Not many. (With apologies to the Justice Department's Antitrust Division.)

* For how long a period is tactical behavior by incumbent firms, whether or not based on asymmetrical information, able to forestall the competition of efficient potential entrants? Not long. (With apologies to practitioners of single-play games.)

On the knowledgeability of commercial transactors:

* What fraction of the typical consumer's and typical businessman's commercial purchases turn out to be significantly different from the quality they expected? A very small fraction.

* What is the ratio of this fraction to an analogous fraction that tracks dealings with the state? Very, very small.

* Are expectations realized more often in commercial transactions than in noncommercial transactions? Divorce rates now over 50 percent say, you bet they are.

This correspondence between conceptualization and commercial reality enabled neoclassical economists to draw conclusions that have been sustained empirically for a wide range of situations. The gathering of evidence was aided substantially by the commensurate measurability of commercial activity. Useful theory and supporting evidence contribute to the solution of a problem that in its nature is of great importance to both economics and policy. These account for the success of economics. This three-part explanation of the success of economics should make us less apologetic than we seem to be about Marshall's definition of economics. Marshall begins his Principles of Economics [1890] by defining economics as the

...study of mankind in the ordinary business of life; it examines that part of individual and social action which is most closely connected with the attainment and with the use of the material requisites of well-being.(4)

Interpreted as describing what economists should do or as what they do today, the definition is obviously too circumscribing, especially to those who value the application of economics beyond its traditional boundaries. For them, it seems terribly narrow, dull, and bourgeois. But it is a good representation of what neoclassical economists actually studied when Marshall wrote, and it is a representation that is focused on those studies of economics that most account for its success within its own boundaries. I include among these the studies, such as those of Pigou, that were most concerned with the question of social optimality. That neoclassical economics could reach defendable and even persuasive conclusions about social optimality is rather remarkable and, on reflection, undermining of the notion that the economics encompassed by Marshall's notion was a dull affair.

I have said little about rational behavior in the contexts set forward by Knight [1921] and Simon [1947]. Knight emphasized the absence of rationality under the condition he named "uncertainty." He describes this condition as one in which there is so little experience with phenomena related to the decision at hand that it is impossible to calculate empirically based probabilities necessary for rational action. The emphasis is on the absence of experience. Decision-making cannot be guided by standards of utility or profit maximization if we assume the existence of this condition, and so, in a rather casual and brief way, Knight substitutes emotion for rational calculation. His arm-chair psychoanalyzing leads him to describe businessmen as "optimistic" about outcomes under uncertainty, and he alleges that this leads them to over-invest and, on average, to earn negative returns. Simon takes a similar position, but he criticizes the economist's use of rational calculation by supposing that many business problems are so complex that coping with them would require mental capacity (rather than, as in Knight, require experience) beyond that possessed by mere mortals. Because this "bounded rationality" makes rational calculation impossible, Simon puts forward a decision criterion different from maximization of utility or profit. His substitute is "satisfying behavior." Satisfying behavior reflects a willingness to settle for outcomes that fall short of being optimal but that exceed a subjective standard of minimal acceptable satisfaction. What this subjective standard is surely depends on psychological disposition, for it involves acceptance in the absence of the capability to judge what is best in terms of benefits and costs.

Both concepts are rather useless in the analysis of the decentralized commercial system, and this for three reasons. First, it is never the case that experience is completely lacking or that problems are so complex as to render rational thought processes useless. The pure cases put forward by Knight and Simon are virtually (I would say entirely) empty of empirical relevance. Second, no attention is given by Knight and Simon to the conditions that inform us as to when actual decision problems approach these pure cases sufficiently closely to make the psychology of the decision-makers, rather than the rational interpretation of "objective facts," the dominant force that guides decisions. Because no operational guidelines for uncertainty and bounded rationality are provided, it is not possible to assess the empirical relevance of these guidelines. Third, the emotion-based decision procedures put forward by Knight and Simon as substitutes for rational calculation are so inadequately described that it is impossible to assess whether they are operative in a given decision situation. For these reasons, they offer no practical alternative to presuming action is based on rational considerations.

Other sources of the belief that rational behavior is too strong an assumption are experiments that purportedly uncover decision errors that, because they are systematic, are judged to be inconsistent with rational solutions to the problems put before the subjects of these experiments. The findings in these experiments are difficult to take very seriously. The results of a handful of experiments cannot offset the evidence from a multitude of situations every day in. which we see that less is bought if price increases and in which we see resources flowing out of low rate of return uses and into high rate of return uses. Moreover, the results of these experiments are difficult to interpret because they confront their subjects with choices that are not straightforward, that are unfamiliar to them, and that involve extremely small payoffs. Rather than ask a subject if he or she prefers more of a good to less, the subject is confronted with unfamiliar probability-based lotteries in a situation in which the subject is never quite sure what rating scale is really being applied to his or her decisions.(5)

In assessing these attempts to fault the assumption that behavior is motivated by narrowly conceived self-interest and is guided by the rational pursuit of this, it is important to remember that the object of the central inquiry of economics is behavior in the large, decentralized commercial economic system. In this arena, few decisions, if any, are made under conditions of uncertainty, of bounded-rationality, or of extreme unfamiliarity. Buying and selling, producing and consuming, and saving and investing are activities that are frequently repeated, often by specialists, and are undertaken under circumstances that if not entirely familiar are far from unusual. Even if the criticisms of the rational behavior assumption that we have been discussing are given full weight in circumstances likely to bring them into play, these circumstances are not those that describe the overwhelming number of decision problems faced in the commercial world. Remembering this brings us back to the proposition that the modeling of this world by our classical and neoclassical predecessors was quite realistic.

IV

There remains a question of why the other social sciences do not seem to have progressed as much within their traditional boundaries as economics has within its own. Practitioners in these other social sciences are as clever..., well, almost as clever, and in the case of lawyers perhaps more clever, as are economists. They are quite capable of devising conceptual models appropriate to the problems they normally confront in their sciences.

One explanation for their more limited progress is that their sciences do not contain a problem that is as important in the totality of their disciplines as decentralization is in economics. I suspect this is true but do not know it for a fact. If true, practitioners of other social sciences need to solve many different problems before their work can impact their disciplines as much as the study of decentralization has impacted economics.

Another part of the explanation is that although practitioners of these sciences have methods for making measurements, these measurements generally are specific to the problem being studied and are not commensurate across different problems. This makes useful generalization difficult. There are technical problems in devising a commensurate unit of measures in these sciences but the chief problem is a different one. It is the reluctance by practitioners in these disciplines, and by the general population also, to credit wealth with having great explanatory power for the noncommercial behavior studied in these disciplines. This defendable reluctance accounts for the popularity of the notion that "economists know the price of everything and the value of nothing."

Less poetic but more correct in substance is the notion that the wealth consequences of one's actions explain much of what happens in the world of commerce but much less of what happens in the non-commercial world. There exists a high correlation between personal wealth and personal utility when it comes to decisions concerning commercial goods, and this makes the wealth consequences of decisions a very important predictive index for commercial behavior. Practitioners of sociology and other social sciences certainly do not deny the relevance of wealth consequences to the behavior they study, but they believe these consequences fail to explain this behavior to the degree that they explain commercial behavior. Decisions pertaining to marriage and to the having and raising of children for example are influenced by wealth consequences, but not as much as are mergers between firms and investments in new subsidiaries. Similarly, voting in national elections is difficult to explain solely by appealing to wealth consequences, for one person's vote has no perceptible wealth consequences. The weakening of the explanatory power of wealth consequences in regard to non-commercial behavior makes it difficult to apply a single analytical framework to the set of problems on which other social scientists work, and it raises a major obstacle to the development of a single commensurate unit of measure.

Let me be quite clear about this. When economists analyze the consumption behavior of households, the employment choices of workers, and the investments of capitalists, their conclusions are largely drawn from the wealth consequences that flow from alternative decisions. We do not have much to say about tastes and how these may differ across persons and situations, but, in principle, variations in tastes also explain variations in behavior. Our focus, not exclusively but most often, is on wages, prices, rates of return, and budget constraints. This works quite well in practice if most tastes change only slowly. In effect, we identify commercial activities by the fact that wealth consequences are important for them. Differences in wealth consequences track differences in utility levels well for these activities. If quantities did not respond much to price changes, if investments did not respond much to rate of return changes, if work patterns did not respond much to wage changes, the price system would be of no importance to resource allocation. The sensitivity of behavior to price or, more accurately, to wealth consequences, is high for goods that we identify as commercial goods, and neoclassical economics has had much to say that is useful because it dealt mainly with these goods. Arguably, the sensitivity of non-commercial behavior to wealth consequences is not as high, and sensitivity to other considerations, such as religion, culture, upbringing, is higher. The other social sciences therefore have a task more difficult than that which faced neoclassical economics. A heterogeneous mixture of several important considerations must be related to different taste patterns and to resulting differences in behavior. This is not to deny the relevance of the logic of economics to most elements in this mixture. Rather, it is to say that the difficulty in explaining behavior even through economic logic is greater than for commercial activity because the important channels of causation are more numerous and are less correlated with a single unit of measurement. Obviously, the severity of this problem varies. Criminal activity is motivated strongly by narrow self-interests; these interests are measured well by wealth and they are pursued rationally. Here, the problems to which I allude are minimal. They are stronger for problems of passion, marriage, child-bearing, and patriotic fervor.

Marshall senses this toward the end of Appendix C of his Principles, titled "The Scope and Method of Economics." He sees advantages and disadvantages in extending the scope of economics, so it is incorrect to view his position to be one of discouraging interdisciplinary work. Rather, he points to a cost of extrapolating economics beyond its normal boundaries. He makes the following point:

Economics has made greater advances than any other branch of the social sciences, because it is more definite and exact than any other. But every widening of its scope involves some loss of this scientific precision; and the question of whether the loss is greater than the gain resulting from its greater breadth of outlook, is not to be decided by any hard and fast rule.

Within the scope of the few pages surrounding this quotation, Marshall gives no supporting argument for his claim that inexactness accompanies the application of economics outside its normal boundaries, but his emphasis on the commercial context of economics leaves no doubt that he realized the advantages offered to economists by the presence of prices. More broadly interpreting prices as wealth, we can translate Marshall's work as claiming that a major source of inexactness in the other social sciences derives from the diminished ability of wealth to explain behavior outside the commercial context. This interpretation is consistent with Marshall's definitional claim that economics "...examines that part of individual and social action which is most closely connected with the attainment and with the use of the material requisites of well-being."

Marshall, being so expert on the relation between margins and totals, surely would agree that inexactness is a cost worth bearing if doing more interdisciplinary work brings the marginal returns from this work closer to equality with work within the traditional boundaries of economics. The investment of much effort in interdisciplinary work would have been difficult to justify on grounds of marginal equalities before the decentralization problem had been solved, and, in fact, not much such investment took place then. The decentralization inquiry was in its essence completed by World War II, and it was not until the decade of the 1960s that work by economists began to extend seriously into other disciplines. The extension has continued at a more significant pace to the present.

But it is not only in regard to interdisciplinary work that the efforts of economists have been extended. Side by side with their explorations into other disciplines, economists have also looked more closely at aspects of commercial behavior that were neglected by our predecessors when they searched for a solution to the decentralization problem. I refer to those "black boxes" in neoclassical economics. Because we have begun to look into these boxes, we have progress in our understanding of the problems resolved by the firm, by different private ownership arrangements, and by the special provisions that frequently attach to market agreements. There remains much about these that we do not yet understand. As has been made apparent from the lack of useful advice from Western economists to policy makers in Eastern Europe attempting to convert economies from communism to capitalism, we know much less about our institutions, or, at least, much less about creating them, than our predecessors presumed. This work, like interdisciplinary work, waited upon completion of the central inquiry of economics.(6) But unlike much interdisciplinary work, the study of the firm and of contractual agreements continues the neoclassical tradition of focusing on behavior for which wealth consequences remain the dominant consideration. However, neither this work nor interdisciplinary work presently offer a problem of paradigmatic significance equal to the decentralization puzzle. Perhaps none is needed, but one wonders if the next 200 years of economics will be as successful as its first 200 years. Only time will tell. In the meantime, we can be grateful to find our professional plates filled with interesting food for thought. The sensible thing is to come to the table and join the feast!

1. Posner's text was written a few years after he joined the University of Chicago Law School. Although he had a proclivity for economic reasoning before. coming to Chicago, as revealed by some of his work in the law, he developed this quite extensively through interaction with economists at Chicago. However, his serious education in economics probably began before this at Stanford where, as law school professor, he met Aaron Director who had taken up residence at Stanford after retiring from Chicago and from editorship of the Journal of Law and Economics.

2. The major exceptions to this have been the inadequacy of neoclassical treatments of oligopoly, aggregate employment, and economic development. Oligopoly, of course, strains the decentralization presumption by allowing decisions of individual firms to have an impact. Fluctuations in employment and differences in rates of economic development are time-dependent topics whose solutions are resistant to analyses based on a decentralization model whose essential nature is static.

3. Even strangers may react altruistically on those rare occasions when a community is struck by some catastrophe. For a short period of time, stranger treats stranger as if in the same family. But the study of decentralization cannot be based on the rare event. Its objective is to understand resource allocation under normal circumstances, and so, again, narrow self-interest is the proper motivation to use.

4. In the first two paragraphs of later editions of Marshall's Principles he modifies the exposition to stress even more the importance to be attached to economics as a study of man in the context of commercial activity:

For the business by which a person earns his livelihood generally fills his thoughts during by far the greater part of those hours in which his mind is at its best; during them his character is being formed by the way in which he uses his faculties in his work, by the thoughts and the feelings which it suggests and by his relations to his associates in work, his employers or his employees.

5. For a detailed discussion of some of these experiments, and also of other criticisms of the rational behavior assumption, including the desire to substitute Darwinian notions of natural selection for rational behavior, see the Fourth Commentary, "Profit Maximization and Rational Behavior," in Demsetz [1995].

6. This is virtually true, but not literally. Knight undertook serious discussion of the firm in 1921, and Coase wrote his path-breaking article on the firm and long-term contracting in 1937. However, both works were neglected until the 1960s. It was only then that modern work on the firm and on contractual arrangements became an important concern of economists.

REFERENCES

Becker, Gary S. The Economic Approach to Human Behavior. Chicago: University of Chicago Press, 1976.

-----. Treatise on the Family. Cambridge, Mass.: Harvard University Press, 1981.

Buchanan, James M., and Gordon Tullock. The Calculus of Consent. Ann Arbor: University of Michigan Press, 1962.

Coase, Ronald H. "The Nature of the Firm." Economica, 4, 1937, 386-405.

-----. "Economics and Contiguous Disciplines." Journal of Legal Studies, June 1978, 201-12.

Demsetz, Harold. The Economics of the Firm: Seven Critical Commentaries. Cambridge: Cambridge University Press, 1995.

Friedman, Milton. Essays in Positive Economics. Chicago: University of Chicago Press, 1953.

Hayek, Friedrich A. von. "The Uses of Knowledge in Society." American Economic Review, September 1945, 519-30.

Hirshleifer, Jack. "The Expanding Domain of Economics." American Economic Review, December 1985, 53-68.

Knight, Frank H. Risk, Uncertainty, and Profit. Hart, Schaffner, and Marx, 1921; republished New York: Harper & Row, 1965.

Marshall, Alfred. Principles of Economics, 8th ed. London: Macmillan Publishing Co, 1948; 1st edition, 1890.

Mises, Ludwig von. Socialism. (First English edition published by Jonathan Cape, London, in 1936; Liberty Classics, 1979.

Posner, Richard A. The Economic Analysis of Law, 2nd ed. Boston: Little, Brown and Co., 1977.

Simon, Herbert. Administrative Behavior. New York: Macmillan, 1947.

Tullock, Gordon. The Logic of the Law. New York: Basic Books, 1971.