;草草RAND Joumal of E∞noml国tAVo\. 20, No. 1, Spnng 1989 革37在ajpv33主fJF4The theory of business strategy 吨di--户wfJ:17C6a4r2l :S"hwapJiVroO’J" 1. Introduction 一;: The field of industrial organization has been transformed during the past twenty years. In the 1950s and 1960s, . was predominantly an empirical field with little theory to guide either industry analysis or cross-section regression studies. During the 1980s, by contrast, 飞JLZFftthere has been an intense flurry ofactivity in . devoted to the development ofnew theory. This new wave of research consists almost exclusively of game-theoretic studies of behavior and performance in imperfectly competitive markets. 泣In his companion piece Franklin Fisher argues that the game-theoretic approach to 'rrufJapindustrial organization has been unsuccessful. My aim here is to provide one participant’s view of what industrial organization economists have learned from the recent theoretical 气research and where the field of industrial organization should go during the 1990s. hF4字2. The theory of business strategy Industrial organization economists are both blessed and cursed: our field encomp臼ses1:1a wide range ofbusiness behavior that is a rich arena in which to apply economic principles, 一主山贯穿南兑遥司空嘎明明but the very richness of business strategy defies simple and general theories. In analyzing behavior in concentrated markets, one must face questions like these: What is the timing of investment and pricing decisions? Wh ch costs are recoverable and which are sunk? What information does each firm have about its rivals’ actions or market conditions, and when does the firm acqu re this information? Game theory has emerged as the predominant methodology for analyzing business 明明strategy. Much of the work in the new , involves specifying a game among competing Atz;firms and solving that game in extensive form using the noncooperative solution con臼ptofNash equilibrium or one ofits refinements. Using extensive-form games to model strategic interactions has the virtue of forcing the analyst to think carefully and to be quite precise about the specific nature of competition. At this time, game theory provides the only coherent way of logically analyzing strategic behavior. A metatheme that emerges from this research is that the predictions of the model conceming the character of the equi1ibrium in the game tend to be quite sensitive to the exact specification of the firms’ strategies and the timing of actions. This is not a new Prin四tonUniversity. 1 thank Avinash Dixit, Joe Farrell, Gene Grossman, Alvin K1evorick, Barry Nalebuff, John Vicke罚，andthe Editorial Board for valuab!e comments on an ear1ier draft 125
126 THE RAND JOURNAL OF ECONOMICS notion; it has been appreciated for at least 100 years that the Coumot ( 1838) equilibrium (Nash equi1ibrium in quantities) and the Bertrand (1883) equilibrium (Nash equilibrium in prices) generate very different predictions, given the same cost and demand conditions. Sequential moves, la von Stackelberg ( 1934), yield yet another outcome. Critics such部Fishersee this sensitivity of equilibrium behavior to the specif cation of the extensive form of the game as evidence that the game-theoretic approach has failed, especially since the "correct" specif cation may be hard to discem from available industry information. This complaint is reminiscent of the old lament that "there is no theory of oligopoly" (since several competing theories have yet to be rejected). The fact of the matter is that competiti ve strategy in practice encomp臼sesa wide variety of strateg:ic and tactical decisionmaking, from the pricing of products to investment in production and distribution facilities to contracting practices with customers and input suppliers to research and development expenditures. There is no reason to expect or strive 叹气，for a single unif ed oligopoly theory that would deliver unique predictions to armchair ;{ititheorists, independent of the particul缸"sofhow∞mpetition is played out in a given ind山位 diversity of predictions in different game-theoretic models reflects our broadening understanding of business strategy. With our new game-theoretic tools, we can carefully 骂-yir-analyze a much wider range of competitive strategy than was previously possible, and we know better what to look for in doing case studies and industry analysis. The theory tells us the conditions under which different outcomes occur and what factors are most 币critical 22in shaping behavior and performance in concentrated industries. The new theories also help 问叫us understand what we observe happening in concentrated industries, although rather detailed 142information is n四essaryto know which piece oftheory best applies in any particular setting. 1 like to think of our emerg:ing theory of oligopolistic behavior as analogous to the 吁吁λtheory of evolution. Crudely put, the basic hypothesis in the theory of evolution is that 主FZJliving species adapt to their environment to survive and prosper. It is a general the。可thatis meant to apply throughout the biolog:ical world. Applying the theoηof evolution to 叫ιJUindividual species, however, reveals a stunning diversity of ;J古【ffz飞、且}湾续将提荔句号ddffI Of course, the more applied field of corporate strategy has a long and rich history. Economists have much to leam from scholars who have studied corporate behavior and corporate strategy in detail. 1 hope由eleaming will be mutual，ωeconomists undertake more case studies to test their theories and as corporate strategy experts integrate game-theoretic insights into their work.
