American Economic Association

Oligopoly Theory, Communication, and Information Author(s): Martin Shubik Source: The American Economic Review, Vol. 65, No. 2, Papers and Proceedings of the Eighty-seventh Annual Meeting of the American Economic Association (May, 1975), pp. 280-283 Published by: American Economic Association Stable URL: https://www.jstor.org/stable/1818865 Accessed: 13-03-2019 06:01 UTC

REFERENCES Linked references are available on JSTOR for this article: https://www.jstor.org/stable/1818865?seq=1&cid=pdf-reference#references_tab_contents You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at https://about.jstor.org/terms

American Economic Association is collaborating with JSTOR to digitize, preserve and extend access to The American Economic Review

This content downloaded from 117.240.50.232 on Wed, 13 Mar 2019 06:01:49 UTC All use subject to https://about.jstor.org/terms Oligopoly Theory, Communication, and Information

By MARTIN SHUBIK*

The stage is set for the development of (5) the roles and goals of economic a mathematical institutional economics actors. which is simultaneously more abstract and These items primarily involve the descrip- more institutional than much of economic tion of the structure of a firm, industry, or theory to date. The need for this develop- market. Considerations of structure (espe- ment can be seen clearly by examining cially in reference to information and com- both the successes and failures of oligopoly munication) cannot easily be separated theory. from considerations of behavior. Thus we Oligopoly theory has been the skeleton may need to reconsider the nature of the in every economic theorist's closet. Vir- decision making of the individual we tually every course in theory has a few choose to regard as the economic decision hours spent in telling the students about maker. the problems posed and showing them a few classical "solutions," after which the I. Different Approaches to subject is dropped. The reason for this Oligopoly Theory treatment is because we do not have a There are four broad approaches to oli- generally accepted adequate theory of gopoly theory which merit distinguishing. oligopoly. Not only are they complementary with The reasons why we do not have a satis- each other, it is my belief that in the new factory theory are central to many eco- theory that is emerging they will come to- nomic problems. They exist for micro- gether. The approaches are: economic models of competitive markets (1) mathematical models of oligopolis- and they are hidden in the assumptions tic competition, made in macroeconomic models. However, (2) institutional studies and industrial it is possible to gloss over the basic diffi- organization, culties in these because, although they are (3) "neoclassical" oligopoly theory, and (4) behavioral models of oligopolv, the still there, they do not appear to play as "'new industrial organization' and central a role as they do in oligopoly gaming experimentation. theory. The key items are: Ever since Cournot, oligopoly theory has (1) the number and "size" of economic provided a rich source of models for the agents, mathematically inclined. The works of (2) the communication system, A. A. Cournot, J. Bertrand, F. Y. Edge- (3) the state of information, worth, H. Hotelling, T. Bowley and others (4) the specifics of structure of both in- attest to this as do the more recent works dustrial and financial institutions, and using game theory including R. J. Au- mann, R. Selten, Shubik, B. Shitovitz, * Yale University. The research was supported by the L. Telser, and others. Office of Naval Research and also partially funded by a grant from the Ford Foundation. Most of the mathematical models have

280

This content downloaded from 117.240.50.232 on Wed, 13 Mar 2019 06:01:49 UTC All use subject to https://about.jstor.org/terms VOL. 65 NO. 2 FIRMS IN OLIGOPOLISTIC MARKETS 281 been static. They have divided more or for instance) there is little in the way of less naturally into those stressing some micro detail on- the differences in the com- variant of a noncooperative equilibrium as munication and information patterns in a solution and those considering coopera- different industries or on critical techno- tive solutions such as a contract curve or logical factors which limit the scope and the core. timing of moves. The research and teaching on industrial This type of oligopoly theorizing is suf- organization has scarcely intersected with ficiently spiritually close to the style of the mathematical work, although this is Marshall's partial equilibrium analysis and now beginning to change. Nevertheless, comparative statics that it supplies the until recently the painstaking detailed major source for the two or three lectures studies of individual industries and firms, on oligopoly that are thrown into the such as those of A. F. Kahn and M. G. microeconomics courses. DeChazeau, J. Peck, F. M. Scherer, J. B. The fourth approach to oligopoly theory Dirlam, and others, have been rarely if has come about through a basic dissatis- ever read by those concerned with mathe- faction with the other three. The key matical models. Being institutionally ori- characterizing feature of this approach is a ented, this work may be read by lawyers central concern with the explicit descrip- before other economists. (See, for example, tion of process. It is explicitly concerned H. Packer.) with dynamics and with disequilibrium. Neoclassical oligopoly theory is scarcely There are possibly three different modes of well defined, but I would lump under this development which can be discerned. They title E. H. Chamberlin, W. Fellner, are: H. Brems, F. Machlup, G. Stigler, H. von (1) behavioral models of the firm and Stackelberg, P. Sylos-Labini, J. J. Bain, simulations, and several others. They are all less mathe- (2) experimental oligopoly investiga- matical than the first group and less insti- tions, and tutional than the second. Chamberlin's (3) mathematical models of search and other economic activity under in- contribution, for example, was mathe- complete information. matically less rigorous than that of Cour- not, but contained much more relevant These approaches cannot be easily com- economic modeling. Bain's contribution on partmentalized as they overlap. The first the study of entry is closer to an institu- is best characterized by the work of R. M. tional study than the others in this group. Cyert and J. G. March, R. Nelson, The key characterizing features of what S. Winters, and several others. It is spiritu- I have called "neoclassical oligopoly the- ally linked to the large business games, ory" are that the analysis is in an open or such as the Carnegie Tech game, as well partial equilibrium context; much of the as to operations research work in market- discussion is cast in dynamic or quasi- ing, production scheduling, finance, and dynamic terms; information implicitly other special functions. plays a role; money explicitly plays a role An important link among economic (trade is in goods for money and vice theory, operations research, and computer versa) and the solution sought is some sort simulation of behavior comes in the at- of noncooperative or quasi-cooperative tempts which have been made to build equilibrium. models of the behavior of specific firms or In general, in this type of theorizing parts of firms (see A. M. Schrieber). Much (with some exceptions, Bain and Brems, of this work is done without academic

This content downloaded from 117.240.50.232 on Wed, 13 Mar 2019 06:01:49 UTC All use subject to https://about.jstor.org/terms 282 AMERICAN ECONOMIC ASSOCIATION MAY 1975 blessing; many studies done by and for ker "trust" and "cooperativeness" in the private firms do not see the light of aca- work of Hoggart, T. C. Liu, and others. demic day and probably many do not de- Furthermore, different approaches to co- serve to be taken too seriously. Neverthe- operation and to decision making are less, the efforts expended in attempting to evident in the work of Siegel and Fouraker, construct detailed simulations are begin- D. Stern, Shubik and G. Wolf, and others. ning to provide a connection between in- The third of the new approaches covers stitutional and theoretical studies. The explicit models of search and behavior simulations are frequently rich in detail, where information is scarce and costly. yet at the same time spell out process. This includes work by P. Diamond, Experimental oligopoly research is small H. Leland, J. McCall, D. McFadden, relative to other scholarly activities, but M. Sobel, M. Spence, and others.2 since 1948 it has been growing and includes II. A Reconciliation of General Equilibrium work by Chamberlin, T. Dolbear, L. Fou- and Oligopoly Disequilibrium raker, J. Friedman, A. Hoggatt, Selten, Shubik, S. Siegel, V. Smith, and others.' General equilibrium theory and partial There are several messages that this work equilibrium offer us the concept of the effi- has to deliver, the major one being that cient anonymous market price system. the specific details of communication, in- Oligopoly theories, in contrast, suggest the formation, and the mechanisms of the mar- noncooperative equilibrium: quasi-cooper- ket have considerable influence on the play ative solutions involving leaders and fol- when numbers are few. lowers, "kinked oligopoly reaction func- Most of the experiments have employed tions" cooperative agreements, or a students or individuals who are not senior variety of behavioral mechanisms. businessmen; the time of play has been When general equilibrium theory is con- relatively brief and the stakes have been trasted with work in oligopoly theory, it for the most part small or nonexistent. appears to be so pristine, general, mathe- Thus what we are learning about is the be- matical, and noninstitutional that it makes havior of students playing in business or the latter appear in utter disarray. This economics games with low financial incen- could be because oligopoly theorizing is not tives and little if any institutional identi- abstract enough or because there may be fication. Even so several useful observa- within the presentation of the general equi- tions can be made. Numbers play an librium system a false generality which important role. Without face-to-face com- conceals many gaps in basic modeling. I munication the numbers in an experi- believe the latter is true. mental game appear to be 1, 2, 3-6, and There are at least eight desiderata for "many." Information overload and com- any robust economic model of markets putational limitations are also clear in the which any student of oligopoly will recog- behavior of business game players. The nize. A reasonable model should: coding and information processing step (a) depend explicitly on the number of going from a matrix game to one with participants; balance sheets is large. In all of the work (b) contain an explicit description of the the importance of sociopsychological vari- functioning of the market mecha- ables is evident. These include "bargaining nism; ability" in the work of Siegel and Foura- (c) be able to handle nonsymmetric in-

I For a detailed bibliography, see M. Shubik. 2 A recent article by M. Rothschild provides a survey.

This content downloaded from 117.240.50.232 on Wed, 13 Mar 2019 06:01:49 UTC All use subject to https://about.jstor.org/terms VOL. 65 NO. 2 FIRMS IN OLIGOPOLISTIC MARKETS 283

formation conditions; cesses which make up economic life. There (d) be defined for all positions of dis- is an underlying structure to these eco- equilibrium as well as equilibrium; nomic processes but in order to appreciate (e) be able to handle markets with few it we must be simultaneously more ab- traders and other markets with stract and more concrete or institutional many traders together; than most of current microeconomic the- (f) have assets and capital structure ory. The theory of games, the new work on play an important role; search and decision making under uncer- (g) have money and finance play an im- tainty are beginning to provide the basis portant role; and for the former; process models, simula- (h) be consistent with behavioral mod- tions, and direct empirical work are pro- els of economic process. viding the latter.

REFERENCES The general equilibrium model as pre- H. Packer, The State of Research in Anti- sented by K. J. Arrow, Gerard Debreu, trust Law, Walter E. Meyer Research In- and others fails to satisfy any of these stitute of Law, 1963. criteria. The noncooperative behavioral M. Rothschild, "Models of Market Organiza- model first suggested by Cournot as a solu- tion with Imperfect Information," J. of tion to oligopolistic competition can be Pol. Econ., 1973, 81, 1283-1308. modified so that it satisfies all eight.3 The A. M. Schrieber, ed., Corporate Simulation work of Chamberlin, Stackelberg, Fellner, AModels, Seattle 1970. and many others and the recent work in M. Shubik, The Uses and Methods of Gam- game theory as applied to economics can ing, New York 1975, ch. 10. be interpreted as variations on this theme. , "A Theory of Money and Financial The variety and difficulties manifested Institutions, Part XVI: Mathematical in oligopoly theory are a reflection of the Models for a Theory of Money and Finan- cial Institutions," C.F.D.P. No. 377, to maze of institutions and complex of pro- appear in the Proc. of the Symposium on Adaptive Econ., Math. Res. Center, Univ. 9 For a discussion of this somewhat cryptic remark, see Shubik. of Wisconsin.

This content downloaded from 117.240.50.232 on Wed, 13 Mar 2019 06:01:49 UTC All use subject to https://about.jstor.org/terms