LEECLEEL TRELLIS TELESIS TIT Tes The MIT Press Industrial Organization ‘Theory and Applications Oz Shy Cambridge, Massachusetts London, EnglandCopyright © 1905 Massachusetts Institute of Technology All rights reserved, No past of thie book may be reproduced in any form liy any electronic or mecianical means (including phetocopyiag, recording, oF information storage and ectrieval) without permission in writing froma the publisher. "This book was typonet by the author usig the BTPX document preparation software developed by Leslie Lamport, Camera-ready coptr was produced by "ype 2000, Mil Valley, California, and the book war printed and bound by ‘The Mple Veil Book Monufscturing Croup, Binghamton, New York in-Publication Data Library of Congress Cataloging Indaetrialorgsnization: theory and appliestions / Oz Shy. Inches bibliographical references and inde. ISBN 0.262 10966-8 (he + alk: paper). — ISBN 0.262-60170.6 (yh alk, paper) 1. ndusteil oreanization (Reanemie ‘heors) 2, Industria conganizstion -Case studies 1 title HID 2326.8565- 1996 . : 887-20 95.2647 cP First printing EREEEEESSSESESSISSDDIIIEREEEA For my mother, Hadass: Shy and in memory of my father, Asher ShyPEER ESE 4 y Contents List of Figures iit Preface xvii 1 Introduction 1 11 ‘The Study of fudustrial Organization 1 12° Law and Beonomica 5 13. Industrial Organization and Internationat Trade 7 1A Refecences 7 I Theoretical Background 9 Basie Concepts in Noncooperative Game Theory 11 24 Normal Form Games 12 22 Extensive Forma Games 22 23 Repeated Games 23 24 Appendix: Games with Mixed Actions 3 25 Appendix: Gaines with Imperfect lnformation 37 26 Exercises 40 27 References 42 3. Technology, Production Cost, and Demand 43 BL Technology and Cost 43, 32 The Demand Function 49 33 Appendix: Consumer Surplus: Quasi-Linear Uility 53, Ba Beercises 54CONTENTS TT Market Structures and Organization 57 4 Perfect Competition 63 41 NonIncroasings Returns to Seale 64 42 Increasing Returns to Seale 66 4.3 Marginal-Cost Pricing and Social Welfare 68 4A Exercises 6? 45° References 70 ‘The Monopoly Ti 5.1 The Monopoly's Profit-Maximization Problem 72 5.2 Monopoly ard Social Welfare 73 5.3. Discriminating Monopoly 75 5.4 The Cartel and the Multiplant Monopoly 78 5.8 Durable-Goods Monopolies 80 5.6 Appendix: The Legal Approach 80 5.7 Exercises 92 SR References 4 Markets for Homogeneous Products 97 6.1 Coumot Market Structure 98 6.2 Sequential Moves 104 G3 Bertrand Market Structure 107 64 Cournot vores Bertrand "112 6.5 Solf-Enforeing Collusion 115, 6.6 International Trade in Homogelicous Preucts 120 6.7 Appendix: Cournot with Heterogeneous Piems 126 68 Exercises 128 69 References 131 Markets for Differentiated Products 183 ‘Two Differentiated Products 135 Monopolistic Competition in Differentisted Products 143, “Location” Models 149 Appendix: Inverting Demand Systems 162 Appendix: Bguilibriumn in the Linear City 103 Exercises 164 References 166 PERERREDUSSTE E9381 CONTENTS ix ur 10 Concentration, Mergers, and Butry Barriers 109 an 82 83 84 85 86 a 88. 89 Concentration Measures 1 Mergers 173 try Barriers 182 Entry Detercence 186 Contestabld Markets 208 Appendix: Merger and Antitrust Law 200 Appendix: Entry Deterrence and Antitrust Law 212 Bxercises 213 References 214 ‘Technology and Market Structure 219 Research and Development 221 9a 92 93. of a5. 96 97 98 99 9.10 on Classifications of Process fanovation 222 Innovation Race 224 Cooperation in R&D 229 Patents 233 Licensing an Dunovation 239 Goveraments and International RleD Races 241 Appendix: Patent Law 244 Appendix: Legal Approach to R&ED Joint Ventures 247 Mathematical Appendix 248 ‘ Exercises 248 References: 250 ‘The Economics of Compatibility and Standards 253 104 wa 193 wa 105 ‘The Network Externalities Approach 256 "The Supporting Services Approach 263 ‘The Components Approach 208 Exercises 275 References 276x CONTENTS IV Marketing 279 1A Advertising 281 11.1 Persuasive Advertising 283 11.2 Informative Advertising 287 11.3 Targeted Advertising 290 11.4 Comparison Advertising 204 115 Other Issues Concerning Advertising 207 11.6 Appendix: Advertising Regulations 300 ILT Exercises 902 11.8 Relerénces 304 12 Quality, Durability, and Warranties 307 12.1 Personal Income and Quality Purchase 308 122 Quality as Vertical Praduet Differentiation 310 123 “Mncket Structure, Quality, and Durability 315 124 The Innovation-Durability Tradeot? 317 125 The Market for Lemons 922 12.6 Quality-Signaling Games 327 127 Wevranties 330 128 Appendix: The Legal Approach to Products Liability 395 129 Bxercises 337 1240 References 338 13 Pricing ‘Tneties: Two-Part Tariff and Peak-Load Pricing 341 13.1 Two-Part Tariff 342 132 Nomniform Pricing 396, 133 Peak-Load Pricing 348 184 Can Firms “Control” the Seasons? 352 13.5 Execcises 358 13.6 References 368 14 Marketing Tacti Dealerships 362 14.1 Bundling and Tying 962 M2. Killing Of Markets for Used Textbooks 376 14.3 Dealerships 380 M44 Appendix: The Legal Approach to Tying 388 145 Appendb: Legal Approach to Vertical Restraints 389 146 Exercises 900 14.7 References 392 : Bundling, Upgrading, and WEST CECCREET LE Stl y { CONTENTS V_ The Role of Information 393 15 Management, Compensation, and Regulation 305 15.1 The Principal-Agent Problem 296 152 Production with Teams 404 15.3 Competition and Managerial Compensaticn 407 15d Why Executives Are Paid More than Worvers 13, 15.6 Regulating a Firm under Unknown Cost 406 156 Bxeroises 419 15.7 References 420 16 Price Dispersion and Search Theory 421 16.1 Price Dispersion 421 1.2 Search Theory 426 15.3 Mathematical Appendix 432 164 Exercises 452 165 Reforonces 133 VI Selected Industries 435 17 Miscellaneous Industries 437 171 Restaurant Beonomies 438 17.2 The Aicline Industry 440 17.3 The Fishing Industry 448 174 Publie Roads and Congestion 452 175 Brexcises 450 ATS References 457 Index 459LEEESUEEESEDDSEDSSSESR SSS ESEb IT Figures 24 22 23 24 25 26 an 32 3a 34 35 36 m1 at 42 43 5 52 53 34 ‘Tho pilot and the tovrorist 23 ‘Two proper subgames 26 Best-response fumetions for the mized-action extended gare 37 AA game with imperfect information: Information sets 38 Game with imperfect information: Subgames 39 Battle of the Sexes in extensive form 42 Total, average, and marginal cost functions 46 Duality between the production and cost funetions 48 Inverse liner demand 49 Inverse constant-elasticity demand 50 Consumers’ surplus 52 Demand generated from a quasi-linear utility 54 Commonly assumed and used market structures 61 ‘Competitive equilibrium under constant returns to scale 65 Decreasing average cost technology 67 ‘Marginal-cost pricing and social welfare 69 ‘The monopoly’s profit maximizing output 73 Monopoly and social welfare TA Monopoly diseriminating between two markets 7 Durabie.good monopoly: the case of dowasard sloping demand 82 Durnble good monopoly: the case of discrate demand 86 ‘Two-period game of a durable-good monopoly facing discrete demand 87xiv FIGURES 6.2 Cournot test-response functions 100 6.2 Rageworth Cyeies: Bertrand et ‘constraints 112 5.3 Residual comand whon Gms have fixed inventories 114 64 IL's impott level under a uniform tariel 123 6.5 IL's import under the PTA 124 56 The welfare effects of the froe-trade agreement 125 upetition under eapaeity Approaches.to modeling differantiatod-products industries 134 Measuring the dogree of product differentiation 197 Best-response functions for quantity counpetition in diferentiated products 138) 74 Best-response functions for price competition in diferentiated produsts 139 7.5 CBS indifference curves for N= 2144 7.6 Decreasing averago-cost techmology 146 7.7 Hotelling’s linear city with two firms 150 7.8 Tho positicn of firms on the unit circle 158 7.9 Sequentialocation game 187 7.20 Discretelocation model 159 7.11 Undereutproof equilibrium for the discrete-location model 162 7.12 Existence of equilibrium in the linear city 163 8.1 Upstream factor suppliers and downstream producers U7? 82. Sunk costs and ontry barriers 184 83 Sunkoenst entry battiors with partial cost recovery 185 84 Incumbents profit levels and capacity thoiees fer different levels ‘of entry cost. 191 8.5 Capacity accumulation and marginal cost 193 8.6 Relaxing the Bain-Sylos postulate 193 8.7 Dest-response functions with fixed eapocity 194 88 Capital replacement and entry deterrenice 105 8.9 Judo economics: How an entrant secures entry accommoda- Hon 199 8.10 Two-period, signaling, entry-deterrence game 204 8.11 Contestable markets equilibrium 208 9.1 Glassification of process innovation 223 9.2 RED race botwoon twe firms 226 9.3 Gains and lessee due to patent protection 285 Wi Tat LIS A FIGURES, xv 10.1 Deriving the demand for tolecoramunication services 258 10.2 The PTT profit function in the presence of actwork externalities 259 10.3 Two-standard (incompatibility) equilibrium 26 104 Consumers’ distribution of tastes 264 10.5 Bauilibeium varity of brand-specific software 267 LLL Consumer surpins for a given persiasive-sdvertsing level 286 112 Equilibrium aumber of firms placing ads 289 11.3 Targeted advertising: Experienced versus inexperienced consumers 201 114 Informative versus persuasive advertising 204 115 Advertising-indused demand increase and falling prices 500 12.1 Horizontal versus vertical differentistioa $11 12.2 Vertical differentiation in a modified Hotelling model 312 33 Determination of the indifferent consumer ainong brands vertically diferentiated on the basis of quality 313 124 Innovation and durability 319 125 The market for lemons: Bad cars drive out the good cars 926 13.1 Quasi-linear uilty indifference curves 43 18.2 Puro two-part tariff club cbargos 345 18.3 Nonuniform pricing and price discrimination 346 134 Nonuniform price schedule 37 13.5 Seasonal demand structure and monopoly peak-bad pricing 349 13.6 Cost structure of a monopoly selling services iu two periods 255 13.7 Revenue functions for the vertical and horizontal dilferentiation cases 356 LM Bundling monopoly 362 14.2 Tervitorial dealerships in the linear city 385 15.1 Optimal contract under asymmetric information 403, 15.2 Manager's best-respnse function 410 16.1 Consumers with variable aearch oost searching foe Uhe lowest price 423 16.2 The detormination of the discount and expensive prices 425 16.3 Prices in a consumer-seareh model 427 16 Reservation-price strategy 43017.1 The equilibrium restanrant price 438 PIGURES 17.2 Pally connected (FCI and hub-and-spoke (HS) networks 442 17.3 Evaluation of aifate regulation 447 17-4 Equilibrium versus optimel highway congestion 455 UODRUEVITT Preface If we knew ohat it was we were doing, it would not be calf reseatch, would it? =A. Einstein Motivation for Writing This Book ‘The motivation for writing this book grew fom several years of eaching undergraduate and graduate industrial organization and interational trade courses at SUNY-Albany, Tel Aviv University, and the University fof Michigan, I felt that for both important Rlds in economies, no the- foretical book targeted advanced undergraduate ond beginning graduate stunts ‘Therefore, { was guided by my belief that there should no: be any necessary correlation between mathematical complexity and theoretical precision. ‘That is, the purpose of this book isto bring to the advanced ‘Student the basic and the latest developments in industrial organization in a very precise manner, but without resorting to advanced mathemati- ‘al techniques. By precise Lean that the various market structures and tequilibrin—and optimal allocations as well as the ruies by whch firms fand consumers actually behave—are always carefully defined. {eel that ‘a student of a theoretical course should be able to make precise defin- tions of what agents actually do, and that teaching the student how to precisely define the environment and market structures has nothing to ddo with getting mote mathematical raining. That is, [have attempted to precisely define the equilibria and the models despite the fact that the models are solved for specific examples with no mathernatical graccality. ‘The Level and Prerequisites My intention is to make this book readable to undergraduates who have some training in microeconomics using calculus. However, in sore invith Preface stances, this course cant be taught without using caleulus (see the list of toples in te next section). Before reading this book, the student should have seme experience in maximization techniques for one- and ‘wo-variables cptimnization problems, Occasionally, the student will have to have a very basic knowledge of whet probability is and how to caleus late the joint probability of two events. Nothing in this book requires methods more advanced than the ones ! have described. Students who id not have aay training in microeconomics using caleulvs may not be fable to handle several of the market structures, The reader questioning whether this book, fits his or her leval is advised to look at chapler 3, Which reviews the basic microeconoraics needed for a comprehensive skudy of industial organization Industrial Organization without Calculus Writers of good (extbooks should attempt to base most of their argu: ‘ments on simple logie rather than on long (or short} derivatives. In that respect, I admit that I failed to provide the reader with » completely free of calculus book for a very simple reason: inost of our research and Publications are based on calculus, and each time I attempted to avoid using ealeulus, {had to reprodiice the theory instead of using an existing ‘The following, however, is a list of topies that are analyzed without the use of calculus Basic Concepts in Game Theory: Chapter 2 Durable Goods Monopolies: Subscetion 5.5.2 Perfect Competition: Chapter 4 ~ Self-Enforcing Collusion: Section 6.5 Bertrand Price Competition: Section 6.3 Preferential Trade Agreements amor Sequential Entry to the Linear City: Subsection 7.3.3 Caleulus-tree Location Model: Subsection 7.3.4 Concentration Measures: Section 8.1 Butry Barriers: Section 83 Investment in Capital Replacement: Subsection 8.4.3, Countries: Subsection 6.6.2 TOCA Preface xke Credible Spatial Preemption: Subsection 8.4.5 Limit Pricing as Entry Deterrence: Subsection 8.46 Process Innovation: Section 9.1 Innovation Race: Section 9.2 Licensing an Innovation: Section 9.5" Ineers jonal Subsidies for New Product Development: Subsection 96.1 ‘The Econo: cluding subscetion 10.1.1) Advertising: Chapter 11 (excluding section 11.1) Quality, Durability, and Warranties: Chapter 12 (excluding section 122) Pi ng Tacties: Chapter 13 (excluding section 124) Bundling and ‘Lying: Section 14.1 (excluding subsection 14.1.6) Market Segmentation: Subsection 4.1.5 Killing Off Used Textbook Markets: Section 14.2 Territorial Dealership: Subsection 14.3.3 ‘The Principal-Ageat Problem: Section 15.1 Regulating a Firm under Unknown Cost: Section 18.5 Why Executives Are Paid More than Workers: Section 15.4 Search Theory: Section 16.2 Restaurant Economics: Section 7.1 Multiproduct Flems: Subsection 17.2.1 Price Regulation: Subsection 17.3 Law and Economics Appendixes: Most chapiers conclude with aon technical appendices discussing the major legal issues and laws ‘concerning the topics analyzed in the body of the chapterxx Preface ‘To the Instructor Since this hook grew out of lecture notes written for upper-division un- dergraduate and graduate courses, the instructor will ( hope) find thie book convenient to use, since almost all derivations are done in the book sel. Tf you are constrained to inetruct @ course without using calculus, then you can teach the list of topics given earlier. If you ean use some calculus, then the amount of material that you can cover depends on your preferences and the length of the course. ‘All the theoretical background the student, needs for s comprehen: sive study of this book is provided in the first part. In fact, not all the material covered im this part is needed co study this book, but it is brought up here for the gale of completeness, or for those readers who have either an insuTicent background in economics or none at all ‘Thorefore, the instructor is urged ta decide on how much time to devote to this preparation part only after having completed the entice plan for this cours. This theoretical preparation is composed of two chapters: Chiapter 2 provide all the necessary game theoretic tools needed for the study of this book and for understanding the literature on industrial Drganization. Background in game theory is not needed for reading this chapter, and no previous knowledge is assumed. ‘The main sections of chapter 2 must be taught before the instructor proceeds with the study of Industrial orgenization. Chapter 3 provides most of the besie microeco- homies hackground needed for the study of industrial organization, The materiel cavered in Uhis chapter is studied in most intermediate microe- feonomics and in some managerial economics courses atid can therefore bbe skipped. Two-semester course [A two-somester course can be logically divided into a more technically mmavket-structute-oriented semester, and an application-oriented semester. ‘Thus, the fist semester should stark with gamne theory (chapter 2), con- tinued by the sequence of three chapters dealing with market structure: perfect competition (chapter 4), monopoly (chapter 6), homogeneous products (chapter 6), and differentiated products (chapter 7). If time is Jelt, Use fitet semester may include mergers and entry (chapter 8) and research and development (chapter 9) For the second semaster, the instructor is froe to select from 9 wide voriety af mostly logically independent topies. A possible starting point could he the theory of network economics and standardization (chap- tes 10), continuing with selected topics from the remaining chapters: Preface oa advertising (chapter 11), ducabitty and quality (chapter 12), pricing tacties (chapter 18), mazhoting tates (chapter 14), mansgornsit and i formation (chapter 15), price dpersion ad seatch theory (eapter 16), and the special industries (chapter 1 One-semester course A common mistake (at least my mi ‘course would be to treat it as the first semester of a two-semester course. ‘When this happens, the student is left with the weong impression Haat industrial organization deals only with the teemical formulation of mar- ket structures, yet without the knowledge that industrial organization hag alot to say about product design, marketing techniques, and chan- nels {chapters 11, 12, 13, 14, 15, and 17). These chapters have many less technically oriented sections, with direct applications. Sorse sections rely on tho knowledge of Cournot, Bertrand, and somotime Hlotelling’s ‘ark structures, and for this reaion, in a one-semester couse, Ladvise the instructor to carafully plan the logical path for this course. Finally, fhe mete on merch theory (chapter 16} can bo coveted wih 0 i salty Lot mo summarize then: the two-semester course fits the structure and the depth of the coverage of this book, The insteucter of a one semester course using this book should study the list of topics covernd in the lator chapters, and then, working backwards, should determine ‘hot is the minimal knowledge of markt structures that students need to acquire in order to be able to understand the lator chaptes New Material Almost by definition, a textbook is not intended for presen:ing newly developed material and ongoing research. However, during the course of simplifying I was forced to modify or to develop Some new concepts or example, I felt thot it is important to include a location model without using calculus for thove courses that do not require the use of calculus. However, as the reader will find, a Nash-Bertrand equilibrium for the diserte location model simply does not exist. For thi reason, I was forced to develop the wndercutproof equilibrium concept described in subsection 7:34 on page 158. ‘Three other topics are also new: (a) the concept of e-forelocure developed in subsection 14.1.4 oF page 365, (b) endogenous peat-load pricing theory (section 13.4 0n page 352) that ‘emphasizes the role ofthe frm in determining which period would be the peak and which would be the off-peak, and (¢) toneted and eraparison advertising theory (sections {1.3 on page 290 and 11.4 on page 294),xxii Preface ‘Typesetting and Acknowledgments The book was typeset by the author using the ISTEX document prepa- ation software by Leslie Lamport (a special version of Donald Knuth'e ‘TX program) duriug the months from June 1993 to July 1904 (Tel Aviv University) and from August 1994 to August 1995 (University of Michigan), All the figures ate also drawn in WTGX using public do- ‘ain software called TeXcad developed by Georg Horn, which can be downloaded from various mainframes. ‘The reader will notice that this book does not have any footnotes ‘Writing a book with no footnotes ivopases a significant constraint on the writer, because footnotes enable the integration of quasi-related topics into a text. However, I felt that footnotes Impose a great inconvenience 0 tho rondor because they tend to disturb the natural low of reading For this reason, I decided to eliminate them. As boring as it may sound, the following cliché is the whole truth ‘ond nothing but the truth: Without the help of the people listed below, 1 would not have been able to complete writing this book! Therefore, E thank: igal Mendel (Princeton), who was the fist person to read the very first, draft of several chapters; Val Lambson (Brigham Young), who was the frst to test this manuscript in an undergraduate industrial organtea: lion class at BYU and the was the first to report a success with teaching, this material to undergraduates in the United States; Tomer Bloomin (e doctoral student at ‘Tet Aviv), for reading the mamiscript several times ‘nd providing many comments and many suggestions throughout that year; Henrik Hom (Stockhoim University), for a great many comments land suggestions and for Lesting the manuscript in a short undergraduate Sougata Poddar (a doctoral student at ; Stephen Salgat (Michigan) for a great maay comments and ilhi- ninating discuss.ons; Yossi Spiegel (Tel Aviv), five anonymous reviewers for The MIT Press, and my undergraduate industrial organization and International trado students st ‘Tel Aviv and Michigan. { thank Mike Meurer (SUNY-Buffalo), Christopher Proulx (Michigan), Ennio Stee. chetti (Michigan), and Abi Schwarte (Tel Aviv), for providing me with comments on seiscted topics. Needless to say, T am the only one re. sponsible for al the remaining errors. Talso would like to thank Mattin Osborne (McMaster) and Hal Varian (Michigan) for theie most helpful advice and Tianlat Shy for all her help. During the preparation of the manuseript, } was very fortwnate in working with Ana Sochi of The MIT Press, to whom I oie many thanks for managing, the project ia the most efficient way. ‘Ann Arbor, Michigan (August 1905) ‘oashyteceg. tau.ac. <2 and ozshy@econ,tau.ac.i1 TTT Chapter 1 i Introduction ‘The purpose of an economic theory is to analyze, explain, Predict, and evaluate Gathered! from Joe Bain, industrial Orgenizasion 1.1 The Study of Industrial Organization 1.1.1 Major observations Our approach to analyzing industry behavior is based ow four stylized facts: Conceniration: Many industries are composed of few fem. Product sharucteristies: lems in some industries produce homog ‘neous or almost identical products, whereas frin in otters distin. guish themselves from the competing Erm by alin dffereutisted rand, Costly actinties: Firms ia aa industry ace engaged in repeated costly ‘activities targeted for the purpose of enhancing the sales of thee brands, In some industries, these activities constitute the major ost ofthe frm and may exceed the cast of producing the product. itself: These costly asivties may include advertising, quality com twol, product differentiation costs, marketing and dealership costs Research ond development: Firms allocate resources for inventing cost reducing production technologies as well as new products, Those ‘resource allocations also include large investments in imitations of ‘echnologies invented by rival firms (reverse engingering),2 Introduction It is often thought that these four observations are interrelated. Most of the coclier empstical studies in Industrial organizntion focused on rn~ ig regressions of variables such as profit morgias, firms’ size, adverts. ig expenditure, and research and development. (R&D) expenditure on cncentration (cee Goldsctmid, Mann, and Weston 1974 for » summary Of these works). ‘The purpose of this book is to provide theoretical Tinkage of the factors that affect concentration, and how concentration affects the strategic behavior of firms. The reason why we think of con- centration az major issue of industrial organization theory follows from the failure of the eeynpetitive market structure to explain why industries fre cotnposed of a few Iarge firins instesd of snany emall firms. ‘Thus, the theory of competitive market structure, although easy to solve for if fan cquiitziam exists, is riost enses cannot explain the composition ancl Dbehavior of firms in the industry. Given the noneompetitive behavior of firms, markets are also inf encod by buyers? zeactiors to Firms’ attempts to maxiinize profits, Tn this rospect, our analysis hese will have to fully characterize how eon fsumere determine which brands to buy, how much to buy, and bow to search and select the lowest priced brand that fits thelr specific prefer~ cnens, For this reason, the approach we take is mostly a stratenic one, fearing that bath firme and consumers learn the market structure and choose an action that maximizes profit (for the firms) and utility (for the consumers). In addition, given the complexity of decisions made by Shaategie (voncompetitive) firms, the iseve of the internal organization Of firms becemes an important factor affecting their behavior. Ths, wwe brieRy address the isue of how management structure under conic Hone of imperfect information affects the performance of the firm is the market Finally, we extebsively analyze the role of the regulator, Flrst from theoretical point of view we ask whether intervention can increase social ‘yolfare under various nnrkot. structures and firms’ activites. Second, whe dencribe and analyze the legal system affecting our industries. 1.1.2 Schools of thought and methodolony ‘The standard approach to the study of industrial organization, as laid ‘out by Joe Dain, decomposes a market. into structure, cmduct, and per Jormance of the market. Siructure moans how sellers interact with other Zatlers, with buyers, ané with potential entrants. Markot structure also ‘ofines the product in terms of the potential number of variants in which the product ean be produced, Market conduct refers to the bebavior of the firms ina given masket structure, that is, how firms determine their price poliey, sales, and promotion, Finally, performance refers to the PEDEEELESESESESEEESESTETE TTT LA The Study of Industrial Organization a welfare aspect of the market intersction. That is, to determine per- formance we measure whether the interaction in the market leads to a Adsiced outoome, oF whether a failure occuzs that requires tion of the regulator. ‘Many aspects of performance are discussed ints book. First, is the technology efficient in the sense of whether itis operated on an optlinal (cost-minimizing} scale? Second, doos the tndustzy produce a socially primal number of brands corresponding to consumers’ preferences end the heterogeneity of the consumers? Thied, are the fms dynamically cficient-—do they invest a proper amount of resources in developing new technologies for current and future generations? All these eficieney requirements age generally suramarized by a particular social welfare function that can combine the trade-off among the diferent aficiency criteria, For example, the welfare of consumers who bave preferences fbr vaciety increases with the number of brands produced in an industry. However, if each brand is produced by a different factory where each factory is constructed with a high Bxed-cost investment, then itis clear that from a technical point of view, the number of branes produced in fn industry should be restricted. Hence, there will always be a trade fff between technical efficiency and consumer welfare that will require Udefising a welfare function to determine the optimal balance between ‘consumer welfare and efficent production patterns ‘in 1939, Edward Mason published a very iluential aticle emphasiz- ing the importdnce of understanding the macket-specific causes of uo competitive behavior. In that article, Mason discussed the methodology for studying the various markets: 3 interven It goes without saying that realistic treatment of tbeee ‘questions necessitates the use of analytical tools which are amenable to empicical application. ‘The problem, as T see it, is to reduce the voluminous data conceraing industrial or ganization to some sort of order through a classification of morket structures, Diferences in, market structure are tl timately explicable in terms of technological factors. ‘The economic problero, however, is to explain, through ant ex aminatioa of the structure of markets and the organization ‘of firms, difecences in competitive practices including pviee, production, and investment policies, “Thus, Mason argued that to be able to understand different cegrees of competition in different mackets, the researcher would have to analyze the different markets using diferent assumed market structures. The reader will appreciate this methodology afver reading this book, whe: wre tey to Bt an appropriate market structure to the studied specific4 Introduetion trket, where the vaticty of market structures are defined and developed in part I, J his article, Mason emphasized the importance of understanding Sources of market power (“market contro!” in his language) in order te imderstand how prices are determined in these markets ("price policy n his language) A firin may have a price policy by reason of the existence of rivals of whse action it must take account, ofthe desirability ‘of considering the effect of present upon future price, af the possibility of competing in other ways than by price, and for many other seasons, ‘Mason continues and hints at how the degree of industry concentration 's correlated with noncompetitive behavior ‘The su of fm inuenes i competitive plies in mm. ber of ways The sae of te puchaes and slr vate the total volun of transitions. the abate of Rees simcity meee o lm ef alo tle to price sad prhucion pce. Selig jee ees atthe lisp of the large frm may be bese te seu os amar competitors. The sae Ufa fie owes inlloenes trenton to gre mathe station Analysis of industrial organization after Mason continued mostly to use & deseriptive language, but, later ones used price theory (sometimes ‘efersed to as the Chicago School). The Chicago price-theory approach conceded that monopoly is poasible but contended that its presence is Jwnich more often alleged than confirmed. When alleged monepalica are semuine, they are usually transitory, with freedom of enury working to liminate their influence on price and quantities within a fairly short time period (see Reder 1982). Thus, the so-called Chicago School wes ety supportive of the persistent-market-power approach that con sultuted Buin's major theory of entry barriers. ‘The fast development of game theory in the 1970s gave a push to Lhe strateytc approach to industrial onganieation and later to strategic trade analysis. Unite the eompetitive-markets approach, the strategic approach models Ube firms on the assumption that they and ‘ther firms can afer, the market outcome consisting of prices, quantities, and the number of brands. In addition, game theory previded the tools for swalyzing dynamic scenstios such as how established Grms roast ta f threat of entry by potential competitors, Our approach dees not atterpt to represent any porticular school of thought, In fact, the main purpoce of this book is to demonstrate Tit TET if 1.2 Law and Economics 5 that there is no general methodology for solving problems, hence each ‘observation may have to be worked out in a different model, Thus, each lime we address a new observation, we generally eonstmuct a special ‘ad hoe model, where the term ‘ad boc" should not be given a negative ‘connotation. To the contrary, the ad hoc modeling methodology frees the researcher fromm constraining the theory to temporary “fashions” which are given a priority in the scientific Kteratuze and allows the scientist to concentrate on the muri of the modelitself, where merit means how well the theory or the model explains the specific observation that the scientist scoks to explain. Nevertheless, the ceader will discover that the strategic game-theoretie approach is the dominant one in this book. 1.2, Law and Economics ‘he logal structure governing the monitoring of the industry is called antitrust law. The word “trust” reflects the spirit of the laws aiming at any form of organization, trust, communication, and contract araoag firms that would impede competition. In this book we confine the discussion of the legal aspects of the industry mainly to U.S. law. I chose to deal with US. law since itis perhaps the most advanced in terms of schieving conpetition and the restraints of monopoly power. Although not the oldest, the US. an- tltrust system seems to be the most experienced one i terms of famous court cases that put the legal system into effect, For example, the Re- sirictive Trade Practices Act, which isthe British equivalent of she 1890 Sherman Act regarding eartél prohibition, was enacte! a very long time fter the Sherman Act, in 1966 to be previse. In other words, the U.S. was and remalas a leader in antitrust legislation It is interesting to note that in the United States rel prices of prod- ucts tend to be the lowest in the world. However, the United States also has the most restrictive antitrust regulation structure in the wold. Hence, although itis commonly argued that market intervention ia the forma of elation results in higher consumer prices, hee we observe that antiteust regulation is probably the cause for low consimer prices in the United States. For this reason, tho study of the US. antitrust systems is an integral part of the study of industrial organization, especially for those stndants from countries with less competitive markets, Several chapters in this book conclude with appendixes discussing the legal matters related to the topies analyzed in the :heo the chapter. In these appendixes, itself and to its historical origio. Court cases are not discussed in this book, since they are analyzed in a large aumber of law-and-economics textbooks, for example Asch 1983, Gellhora 1986, anc Posner 19.6 Introduction 1.21 The development of the antitrust legal system 11 is not surprising that when the Sherman Antitrust Act. was passed in 1890, economists vere elniost unanimously epposed to it, on the basis ‘hat “trust basting” would iavolve a Jess of the eficieney advantages of combinations or trusts (Wiest 1987). Interestingly, ster a decade of Fret enforcenient of the older merger's guidelines issued by the Federal Trade Commission, the newer 1984 guidelines have brought back the affcieney argument 08 an argument for merger in medium concentrated Industries. ‘The reader interested in learning the development of the antitrust laws should not miss reading Bork 1978. According to Bork, the major development (and the entire set of disputes and theoretical ccnjectures) were all formed daring the period from 1800 (Sherman Act) to 1014 (Clayton Act and the Federal Trae Commission Act) "The Sherman Act of 1890 was intended to strike at cartes, horizontal mezgers of monopolistic natare, and predatory business activities. Sec- tion 1 of this act stated that "Every contract, combination in the form ‘of trust or otherwise,.in restrhink of trade or commeree”..1s hereby de: ‘lated to be illegal.” Baler court interpretations followed section 1 of the act previely as stated but, soon began to adhere to the “rule of ren- fou’ in which not every act oF anerger was considered as a restraint of trade, The courts began identifying which restraints were reasonable and ‘which were no. In 1911 a major ruling based on the Sherman Ack was handed down, wherein some cf Standard Oil's activities were found to be illegal, leading to the dissolution of this giant into thirty companies. In thot peciod, American Tobacca also broke up. A large-scale dissolution ‘centred again in 1982, when AT&T responded to pressure to break 1p {nto the seven “aby” Bell companics and ATS&T. The ATRT brenkup ‘vas effected by consent decree and not by litigation. "The search for which restcaints of trade are reasonable led to a more refined legislation, the Clayton Act of 1914, in which price diseriminn- {lon exclusive deating, and eorporate stock acquisition that may lend to reduced competition were declared illegal, ‘The Federal ‘Tvade Commis. on Act of 1914 mandated the FTC. to categorize and identify whnt constitute unfair methods of competition 1.2.2 The “Per Se” vorsus the “Rul of Reason” approaches {nll the aw-and-economaies appendixes, we make a vse of two methods of court suling in antitrust cases: the per se rule, and the rule of reason. Dork (1978) defines the rule of reason as a set of general cate- aries that are given content by ideas about the proper goals of the law, ceonomnies, and the requirement of the judicial process. In other words, ‘court rulings consist of two major categories: (a) business behavior that 1.8 Industcial Organization and International ‘Trade 1 is legal per se, and (b) business behavior that is judged by stancaeds fof the party's intent or the effect the behavior Is Iikely to have. Zor our purposes, we will wefer to the rule of reason as category (b) Bork (1978) regards the per se cule as containing a degree of arbie trariness. ‘The per se rule implies that the judgment is handed down ‘on the basis of the inherent effect of the act committed by the accused patty. That is, to have a particular behavior declared illegal per se, the Dlaintif aoeds only to prove that it occurred. The per se cule jus Fed ia cases where the gains associated from the impesition of tke rule sll far outweigh the losses since significant administrative costs can be faved. ‘That is, the advantage of the per se rule is that the particular ‘cave noed not be identified, since the act itself is assumed to be iliegal 1.3 Industrial Organization and International Trade In this book the render will ind a wide variety of international issues, for the simple reason that international markets should not be very different froma national markets. Ths, one might expect that concentraticn would chaeacterize international markets as well as national marks. AS a Tosult of this (cather late) recognition that international trare can be Characterized by oligopolistie market structures, & tremendous amount of literature emerged curing the 1980s (see Krugman 1989) ‘Once this newer trade theory picked up, a broad new array of issues bad to be analyzed. The frst was, bow can international trade n differ entisted peoduets be explained by a monopolistic competitio: market Structure? ‘Then, what are the implications of ollgopoistic international racket structures for the gains from the imposition of trade barciers? ‘Whereas easier writers got excited by learning that countries have a lot to gain when imposing trade restrictions or allowing subsidization of in~ ddusteies competing in internationally oligopolistic markets, later writors have managed to calm down this new wave of protectionism by demoa~ strating that any trade policy tesommended under a particular market Structure may not be recommended under a different market structure ‘Thus, since itis hard to estimate what the ongoing market structur ig and the form of competition of a particular marie, it may be bet tee shat goveroments refrain from intervention at all. These later pape: Ihave somewhat mitigated the strong policy actions recommended by th arly steategio trae Iterabure. 1.4 References ‘Asch, P. 1983. Industrial Organization and Antitrst Polley. Mew York ohn Wiley &e SousIntroduction Bain, 4.1968, Industrial Organization. 2d ed. New York: John Wiley & Bork, RL 1978, The Antitrust Paradox. New York: Basie Books Gelihorn, B. 1986. Antitrust Law and Eeonomice in a Nutshell. St. Pau, ‘Minn. West Pablihing Goldschmid, H., H. Mann, and 3. Weston. 1974, Industrial Concentration: The New Learning. Boston: Littie, Brown. Krugmsn, P. 1989. “Industrial Organizetion and International Tyade" In Hansook of Industrial Oreanszation, edited by R. Schinslensee and Wiig, Avnstecdam: North-Holland, ‘Mason, B. 1999. “Price and Production Policies of Large-Scale Enterprise.” Arverican Beonomse Review 29, pt. 2 61-74 Posner, R. 1977. Bomomie Analysis of Law. Boston: Little, Brown, oder, M. 1889. “Chicago Beonomies: Performance and Change.” Journal of Beonomie Luerature 201-38, West, B. 1987, "Moropels." In The New Palymive Dictionary of Economics, ‘cdited ty J. Estwell, M, Milgate, and’ P. Newman, New York. "The Stockton Pres, Peete rt hea Part T Theoretical Background Game Theory and MicroeconomicsLEYRESEDUDDDERDSEEESEDOSDOUGSOSEN Chapter 2 Basic Concepts in Noncooperative Game Theory IE you know the enemy and kuow yourself, you need aot feat the result of a bundred batsles. If you kaow yourself but hot the enemy, for every vietory gained you will also suflec & efeat, If you know neither the enemy aoc yourself, yor will suooumb in every battle. ‘AIL con cam see these tneties whereby T conquer, bub wha: hone can seo is the strategy out of which victory is evolved. jun Tau, The Art of War (490 B.C.) Game theory (sometimes referred to as “Sntecactive Decision Tteory”) ig a collection of tools for predicting outeomes for a group of intarscting gents, where an action of a single agent directly afects the payos (relia oF profits) of other participating agents. Te term game theory gems from the esemblance these tools to sports games (0. football, Soecer, ping-pong, and tennis), as well as to “social” games (eg, Chess, ‘ards, checkers, and Diplomacy) ‘Game theory 1s expecially useful when the number of imberactive ‘agents is small, in which case the oction of each agent may haw a sig- fideant eflct on the pay of other players. For this reason, she Ba bf tools and the reasoaing supplied by game theory have been epplied to a wide variety of fields, including economics, political scieure, ani- thal behavior, military studies, psychology, and many more. The goal Of a game-thtoretic rode it to prodict the outcomes (@ list of actions2 Basic Concepts in Game Theory ‘erdopted by each participant), given the assumed incentives of the par Licipating agents, Thus, gamue theory is extremely helpful in analyzing industzies consisting of & small member of competing firme, sce any ace tion of each frm, whether price choice, quantity produced, research and lovelopment, or marketing techniques, has strong effects on the profit levels of the competing firms. As the title ofthis chapter suggests, our analyses focus only on non cooperative games. We generally distinguish between two types of gaane representations: nermal form games (analyzed in section 2.1), and e- tensive form games (analyzed in section 2.2). Roughly speaking, we ean say that in normal form games all players choose all theit actions si ‘multaneously, wheteas in extensive form games agents may choose their actions in different time periods. In addition, we distinguish between ‘wo types of actions that players can take: a pure action, whece a player plays a single action from the player's set of availabe actions, and a ‘mized action, where a player assigns a probability for playing each ne: tion (say by flipping a coin). Our entire analysis in this book is confined to pure actions. However, for the sake of completeness, mixed actions fare analyzed in an appendix (section 2.4), Finally, informetion plays a Key role in game theory (as well as in ‘eal life). ‘The most important thing that we assume is that the players that we model are at least as intelligent as economists are, That i, the players het we mode) have the same knowledge about the strueture, Ube rules, and the rayofls of the,game as the economist that models the ame does. Also important, our analysis in this chapter is confined to ames with perfect information. Roughly, this means thst in perfect information gomes, each player has all the information concerning the actions taken by other playors carlior in the game that affect the place's decision about which action to choose at a particular time, Games under |mperfoct information are not used in this book; however, we introduce them in an appendix (section 2.5) for the sake of completeness, 2.1 Normal Form Game Our frst encounter with games will be with normal form games. In normal form games all the players are assumed to make their moves at tse same time, 2.1.1 What is a game? ‘The following definition provides three eleiments that constitute what wwe call a game. Each time we model an econoinie environment in a ssame-theoretic framework, we should make sure that the following three Pete oUt tet eee 2.1 Normal Form Games 8 cements are clearly stipulated Derwrrtow 2.1 A normal form game is described by the following 4. A set of N players whose nomes ave listed in the et I Nj. 2 Back player %,¢ € IF, has an action vet At uhich is the set of all actions available to player t. Eet a € At denote a particular action taken by player i. Thus, player 1’ action sect a hist of all actions available to player é and fence, A! = (0,04... .04,} wihere ky is the nuinter of actions available to players : Let a = (a'a?,...,0',...,a) be a Hist of the actions chosen by each player.’ We eull this Kt of actions chosen by each pl outcome of the game 9 Bach player { has a payoff function, w, which essigns « real nur. ter, a*(a), to every outcome of the game. Formally, each payoff function « maps an N-dimensional vector, a= (a, ..,a®) (the faction of chosen by enc player), and assigns i a real muinber (a). A few important remarks on the deSaition of a gaine follow 1. Its very important to distinguish between an artion set AY, which is the set of all actions available to a particulor player 4, and an ‘outcome a, which isa list of the particular actions chosen by ali the players, 2 Part 2 of Definition 2.1 assumes that the each player lat @ nite fouriber of actions, that is, that player i has &, actions in the action set At. However, infinite action sets are commonly used in indusérfal organization. For example, often, we will assume that firms choose prices from the sot of aonnegative real aurabers 3, We use the notation (ts? of elements} to denote a set where a (eg. an sction set] contains elemeats in vhich tite order of listing is of uo consequence, In contrast, we use the notation (is) to denote a vector where tho arder does matter, For example, af outcome isa list of actions whore the frst action on the lst isthe action chosen by player 1, the second by player 2, and so on, ‘The literature uses the term action profile to éescribe the Ist of ‘actions chosen by all players, which is what weeall an outcome, For ‘ne purposes there is no harm in using the terra eutcome (instead of the term action profile) for deseribing this list of ations. However,M Basic Concepts in Game ‘Theory st games Invelve some uncertainty to some players, these two terms Should be distinguished since under uncertainty an action profile Thay lead ta several outcomes (see for example mixed actions games Tieseribed in the appendix [Section 2.4). 5. In the literature one often uses the verm strategy instead of the term action (and therefore stratgy set instead of action set), since fn wormal form game, there is no distinction between the two terran, However, when we proceed to nnelyze extensive form games {Cection 2.2), the term strategy is given a differont meaning than the term action, “The best way to test whether Definition 2.1 is clear bo the render is to apply it te a simple example. A simple way to describe the data that tlefne 8 particular gome is todlsplay them in a matrix form. Consider the following game described in Table 2.1. We now argue that Table 2.1 Country 2 WAR PEACE country 1 WAR [YT TSO peace [O32 “Table 2.1: Peace-War game ‘containg all the data needed for properly defining @ geme according to Sheknitina 2.1. First, we hinvetwo players, 1V = 2, called country tans 2 Second, the two players happen to,have the same action sets: A? i ={WATL, PEACE). There ace exactly four outcomes fr this sme: (WAT, WAR), (WAR, PEACE), (PEACE, WAR), (PEACE, PEACE) Tain the enisies ofthe matrix (ie, the four squares) contain the payolls te slayer T (on the left-hand side) and to player 2 (om the right-hand sie), corresponding to the relevant outcome of the game. For example, the doteame «= (WAR, PBACE) specifies that player 1 opens a war while player 2 plays peace. ‘The payofl to player 1 from this outcome Feria) = 1(WAR, PEAC3) = 3. Similarly, the payoff to player 2 i 7(a) = #2(WAR, PEAGE) ~ 0 since country 2 does not defend itself The story bebind this game Is as foliows. If both countries engoat ina wat, then ench country gains a utility of 1. If both countries play PEACE. then each country ening 6 utility of 2. Tf one countey plays WAR while the other plays PEACE, then the aggressive county reaches the highest possible utiliy, since i "sing" a war against the nonviolent Ceumtry with ne effort. Unier this outcome the ntiity of the “pacifist Country” should be the Fowest (equal to aero in our example). PEPDEREEEESEDESDES EH = 2.1 Normal Form Games 6 Im the literature, the game described in Table 2.1 is commonly re ferro to as the Prisoners’ Dilemma game. Instend of having two couse ies Bighting a war, consider two prisoners suspected of having :ommit- Tor which the police lack sufficient evidence to convisseithor suspect. ‘The tivo prisoners are pit in two different isolated cells and ace blfered a lower punishment (or a higher payolt) if they confess of haw- ing jointly committed this crime. Tf we replace WAR with CCNPESS, and PEACE with NOT CONFESS, we' obtain the so-called Prisoners! Dilemma game. Ta the peecent analysis we refrain from raising the question whether the game described in Table 21 is observed in reality or not, or whether the game is a good description of the world. Instead, we ask a differ- cent set of questions, aamely, given that countries in she world behaves like those described in Table 2.1, can we (the economists or politics! scientists) predict whether the world will end up in countcies declaring war of declaring peace. In order to periorm this task, we need to define equilibrem concepts, 2.1.2 Equilib m concepts (Once tte game is properly defined, we can cealze that games may hare many outcomes, ‘Therefore, by simply postulating all tbe posible out Gomes (four outcomes in the game described in Table 2.1), we cannot make any prediction of how the game is going to ond. For example, can soa predict bow a game like the one described In Table 2.1 vould end tip? Will there be wat, or wil peace prevall? Note shat formulating ‘a zae without having the ability to predict implies that the game Is (of Ltele value to the researcher. In oeder to make predictions, we need to develop methods and define algorithins fbr narrowing down the set fof all outcomes to a smaller set that we call epuilirium outcomes. We ‘also eaust specify properties that we Sind desirable for an equtibriu to fulfil. Ideally, we would like to find a method that would select only fone outcome. If this bappens, we say that the equilibrum = unique However, as we show below, the equilibrium concepts developed here of ten fail to be unique. Moreover, the opposite extreme may oozur where fs particular equilibrium may not exist at all. A game that cannot be rin is of less interest to ns since ao real-life prediction Bolore we proceed to defining our frst equilibrium concept, we need to define ove additional piece of notation. Rocall that an outcome of the game a= (a4,...,a8,...,a%) is alist of what the A’ players ere doing {playing). Now, pick a certain player, whom we will cal player 4, (2 ean be player Lor 89 or NV, or any player). Remove from the outcome16 Basie Concepts in Game Theory a the action rlaved by player i himself. Then, we ate left with the lst «of what all players are playing except player i, which we denote by a Formally, (e', "Note that after this minor surgical operation is peeformed, we can still express an outcom? as & union of what action player f playe and all the other players’ ections. That is, an onteome a can be expressed ns we) Equilibrium in dominant actions Our frst equilibrium concept, called equilibrium in dominant eteategies, isa highly desirable equilibrium, in the sense that ifit exists, it desertoy the most intuitively plausible pradiction of what players would actually ds, The following detnition applies fora single player in the sense that it "lassifies actions in « player's action set acconding to a certain eritecion, Depinrtion 2.2 A particular action a € At is said to be @ dominant action for player i if no matter what all other players axe playing, lening & aitays merinizes ployer i's payog.. Formally, for every choloe of uctions by all plagers except i, a4 al 19) 2 wala), for every at € A or example, Claim 2.1 In the gene described in Table 8.1, the action @! = WAR fs « dominent action for player 1 Proof. Ie hias to be shown that no matter what player 2 does, player ie aloays better off by starting a war. ‘Thus, we have to scan over all the pessiblo netions that can be played by player 2. If player 2 playa a? = WAR, then 71(WAR, WAR) = 1 > 0= s(PEACE, WAR), ‘so, if player 2 plays a? = PEACE, then "(WAR, PEACE) (PEACE, PEACE), 2.1 Normal Form Games a Similarly, since the game is symmetric (meaning that renaining player 1 fs player 2 and vice versa, does not change players’ payois}, the condor ‘con establish thet a? = WAR is a domiaant action be player 2, We now tum to defining our fst equilibrium eoncept. An equilib ‘um io dominant actions is simply an outcome whove each play playa 8 dominant ation. Formally, Deeiimion 2.3 An outcome (3! " ery = 1,2,...,.V) 48 said to be an equill fs @ dominant action for each player Guere a & AY for eo jum in dominant actions Cleaely, since WAR is a dominant action far exch player in the game described in Table 2,1, Definition 2.3 implies that {al 1°) = (WAR, WAR) 15 an equilibeium iv dominant actions Although an equilibrium in dominant actions constitutes very ce sonable prediction of how piayacs may interact ia the real world, untor. tunately, this equilibrium does not exist for most genes of interest to us. To demonstrate this point, let us analyze the fllewing Butile of the Sexes game deseribed in Table 2.2. ‘The intuition bebind this (eather Rachel OPERA (4) EDOTBALL (6) Jacob _OPPRA (u) To a FOOTBALL (3 z ‘Table 2.2: Battle of the Sexes romantic) Battle of the Sexes yam is that it is relatively important foe Javob ant Rachel to be tayethor. ‘Thali, assiuning that the payolts to the players io Table 2.2 represent ubilities to each player unde each out come, each player gains the lowest possible utility when the player sous alone to ove of these entertainment events. Both of them gain higher uulity if they go together to oue of these events. However, comparing the to outcomes where the players ate “together,” we oan ohowrve thet Jacob prefers the OPERA, whereas Rachel prefers FOOTBALL, Thus the Bittle ofthe Sexes is sometimes referred to as 4 carthnation yam ‘The Baitle of the Sexes game exhibited in Table 2.2 daveribos some teal life situations, For example, in chapter 10 we analyze economies ia wich ‘oducts operate on diferent standards (suck as differant TV systems} The Battle of the Sexes gaune happens to be an ideal theoratical frame. work to moxie! two firms with two available actions: cloose standart oF standard 2. Failure to have both firms chousiag the same standard ‘may result in aving consumers reject the product, theeeby leuving the ‘00 firms with zero profits8 Basie Concepts in Game Theory ‘After formulating the Battle of the Sexes game, we naw seek to find some predictions for this gare. However, the reader will probably be isappointed to find out that: Cinim 2.2. There does not exist an equiibrinm én dominant actions for the tile of the Sexes game. Proof, This sufficient to show that one of the players docs not ave fa dominant action, In this case, there cannot be an equilibrium in Gosninant actions since one player will not have a dominant action to lav. "Therefore, itis sufficient to Jook at Jacob: If Rachel chooses o! {then Jacch would choose w because Pleo) 2>0= 8") However, when Rachel goes to & football game, a = ¢,.then Jacob wotild choose 6 because (4,4) = 12 Oe Wland)) So, we have shown that one player does not have a dominant ection, and tie snfices to conelude thot, Definition 2:3 cannot be applied ences there does not exist an equiltrium in dominant actions for the Battle ‘of the Sexes game. " Nash epuiorivm (NE) So far we have filed to develep an equilibrium coocept that would select fan cutennie that. would be a “renconable” prediction for this model. In 41951, John Nash provided an existence proof for an equilibrium concept {earlier uted by Cournot shen studying, duopoly) shat bas become the most commonly ised equilibriuin concept in analyzing games. Derinmmion 2.4 An ontooms & = (@,d%,...,8%) (uhere a! € At for fewer i = Ip2yeaN) #8 said to be a Nash equilibrium (NE) if ne player would find it beneficial te deviate promided that al other players do ret deviate from their strateyies played at the Nash outcome. Formally, for every player é, F= 12-90 ailal,a™) for every at € At waa ‘The general methodology for searching which outoomes constitute a NE js to check whether players beneBt from a unilateral deviation from a Certain outcome. ‘Thot is te rule out. an outcome as a NE we nood only HHH PDEDERDERELERSERDTEDES: 2.1 Normal Form Games cy demonstrate that one of the players can increase the payo by deviating. toa ulferent action than the one played in thie specific outcome, asurn- ing that all other players do not deviate, Once we find an outoome in which no player can beneht from any cevietion feom the action played ia that outcome, we ean ascert that we found a NE outcome. ‘We continue our discussion of the NE with the investigation of the relationship between Nash equilibrium and equiltbrium in dominant ac- tions. To demonstrate the rolationship between the two equidbcium concepts, we fist seasch for the NB outcomes for the game described ‘Table 2, Recall that we have already found that (WAR, WA) is an ‘equilibrium in dominant actions, but can this fact help us in searching for a NE for this game? Not surpeisingly, yes, it can! Since an equi librinen in dominant actions means that each player plays.s dominant action, ao player would find it beneficial to deviate no matter tow the ‘others play. In particular, no player would deviate if the other players stick to their dominant actions. Hence, Proposition 2.1 An euitbrium in dominant actions outcome is also NE, However, a NE outcome need not be an equilibrium in dminant Alcogether, we haweit that (WAR, WAR) isa NE for the game describ in Table 2.1, We lanve it to the ceuier to verify that no other cutcome in this game is « NE. ‘Therefore, this equilibrium is called uniqne, The second part of Proposition 2.1 follow from the Battle of the Sexes game, where there exist two NE, but there does not exist an equilibrium in dogsinant actions, Multiple Nash equlibrin ‘We now demonstrate that 2 Nach oquilbriun need not be unique, For example, applying Definition 2.4 to the Bottle of the Sexes game yields Claim 2.2 The Bottle of the Soass game described in Table 2.2 has to Nash equilibrium outcomes: (OPERA, OPBRA) ond (FOOTBALL, FOOTBALL). Proof. Vo prove that (w, ) if a NB, we have to show that no player ‘would benefit from deviation, given that the other does not deviata. In this game with two players, we have to show that, given that « = w, player J would play a” =; and that given that a? =, player ® would play a =u, These two conditions follow from & 0=r du) (a) 2 0= nw, d)20 Basle Concepts in Game Theory Using the same procedure it can be easily shown that the outcome (¢, 6) is aleo e NE. Finally, we need to show tliat the other two outeomes, wd) ‘and (6, ©) are not NE. However, this follows immediately from (2.1). i Noneistence of e Nash equilibrium, So far we have seen examples whore there ie ane or more NB. That is, as in the Battle of the Sexes game displayed in Table 2.2, itis always poss\- Die to find games with multiple NE. JF the equilibrivin isnot unique, the ‘model has & low prediction power. In contrast, Table 2.3 desmonstrates ® ome where a Nesh equilibrium does not exist. Therefore, consider ‘Ue variant of the Battle of the Sexes game after thirty years of mas- The intuition behind the geane descriied in Table 2.3 is that fier Rachel OPERA («)_ FOOTBALL (6) Jacob __ OPERA (w) [FOTO 2 FOOTBALL (@) [0 Tt ay ‘Table 2.3. Nonexistence of » NE (in pure actions) thirty years of marsinge, Rachel's desire for being entertained together with Joseph has faded; however, Joseph's romantic attitude remained fs before, and he weuld always gain a higher utility from being together with Rachel rather than alone, Proposition 2.2 The game described ia Table 8.9 does not hove a NB. Proof. We must prove that each entcome Is not a NW, That is, im each Of Ue four ontteomes, at least one of the player would find it beneficial to deviate (1) Por the (wy) oxteome, 2! (0,9) =2> Hence, Rachel would deviate to a” = 4 (2) For the (6,12) outcome, (us. Hence, Joseph would deviate to a? (3) For the (9,9) eutoame, =(2, 0) ad). Hence, Rachel would deviate to a” = w, (8) For the (w, 4) outcome, 1(¢, 4) © 1 > 0-= x(a) Hience, Joseph wonld deviate ta a’ = ¢. . (oy) 2>0= 6 (u). “> Using “test-response” functions to solve for NE We now develop a tool called “hest-response” fonctions i the search for NE. 2.1 Normal Form Games a1 Derinitioy 2.5 i: In a two-player game, the bost-response function of player i 48 the function HMw), that for every given acon w! of player j assigns an action a! = K'(a?) thet mazimizes player i's puyol? #(a,0') 2 More generally, m an 1¥-player game, the best-response function of player is the function R'(a™), that for given actions a-* of players 12,09 1it 1,41, assign on action a! = fea~) that mazimizes player i's payoff n(a!,a™ Let us now construct the best-response functions for Jacob and Rachel scribed in the Battle of the Sexos game given ia Table 22. Te is straightforward to conclude that arama {s Fenr sand Me?) {¢imze ex ‘That is if Rachel plays w, Jacob's “best response” is to play wand if Rachel plays 6, Jacob's “best response” is to play 6, and so oa, ‘Now, the importance of learning how to construc: bast-respouse func tons becomes clear in the following proposition: Proposition 2.3 [fi iy a Nash eulibrivin outeome, thea for every player i Ra) Proof. By Definition 2.4, in a NE outcome each player does not benelit fom deviating from the strategy played in a NE outvome (given that all othe players co not deviate), Hence, by Delnitios 25, each player is on her best-response function, . Thot is, ina NE outcome, each player chooses an acton that is & best response to the actions chosen by other players in & NE. Propasitivu 23 's extremely useful in solving for NE in a wide variety of games and will be used extensively ‘The provedure for inding a NE is now very simple First, we caleu- late the best-response function of each player. Second, we check which outcomes lie on the best-response functions of all players. ‘Those itt comes that we find to be on the best-response functions of all players constitute the NE outcomes. For example, in the Battle of the Sexes game, (2.2) implies that outcomes (u,w) and (6,4) each satisfy bot players’ best-response functions and therefore constitule NE ouecomes22 Basic Concepts in Game Theory 2.1.8 Welfare comparisons among outcomes So for, our analysis has concentrated on defining equilibrium concepts that cnnble te to select equilitxium outcomes for predicting how play~ fers would end up acting when facing similar games in the teal world. However, we have not discussed whether the propased equilibria yield ficient outcomes, That is, we wish to define an efficiency concept that ‘would enable us to compare outcomes from 8 welfare point of view. In particular, using the Pareto eficiency criterion, we wish to investigate ‘whether there are outcomee that yiekd higher payoff levels to some play~ fers without reducing the payots of all other players. Por example, in the Peace-War game of Table 21, the outcome (PEACE, PEACE) yields higher payos to both players comparod with the ontcame (WAR, WAR), In this case, we say that the oxicome (PEACE, PEACE) Pareto domi- nates the outcome (WAR, WAR), Formally, Derinition 2.6 1. The outcome @ Pareto dominates the outcome a if (@) For every player t, 1) > 0'(@), and (A) there exiets at least one player j for whom x(a) > mt(a). 2, An outcome ® is called Pareto efficient (also called Pareto op- timal) if there does not exist any outcome which Pareto dominates the outcome at 8. Outcomes & end dave called Pareto noncomparable if for some player i, (@) > 2(a); bu for sore other player j, (a) < 79(@). or example, in the Peace-War game, the outcomes (WAR, PEACE} and (PEACE, WAR] are Pareto noncomparable. In the Battle of the Sexes ome of Table 2.2, the outcomes (OPERA, FOOTBALL) and (FOOT- BALL, OPERA) ate Pareto dominated by eack of the other two out- comes. ‘The outcomes (OPERA, OPERA) and (FOOTBALL, FOOT- BALL) are Pareto efficient and are also Pareto noncomparable. 2.2, Extensive Form Games ‘Our analysis to far as concertrated on normal form games where the players are restricted to choosing an action at the same time. In this sec- lion we analyze games in which players can move at different times and ‘more than once, Such games ae called extensive form games. Extensive form gotnes enable us to introduce timing, into the model. 2.2 Extensive Form Games 23 Befoce going to the formal treatment, let us consider the following example, A tertoris: boards a fight from Minneapolis to New York. AF. ter thirty minutes, after reaching a cruising altitude of thirty thouserds feet, the terrorist approaches the pilot and whispers to the pilot that she will explode a bonab if the pilot does not fy to Cuba. Figure 2.1 de. seribes the Pilot-Terrorist game. One player is the pilot and the other is 1 (Pt moves) FLY TO CUBA,/A\ PUY TO NEW YORK, JS Hho oroit os) BX not soma “ re Figure 2.1: The pilot aad the terrorist the terrorist, The gas i repented by a Fre, witha tasting decision node (point I), other decision nodes (Ili and Ic), and tersinal uodes {cad points). Note that in some literature, the (erm verter (vertices) is used in place of the term node/s). Te branches connecting decision todas, and decision nodes to terminal nodes describe actions available to the relevant player on a particular decision node. In this Pilot-Terorist game, after hearing the terors's shea, the pilot gels to be the player to choces aa action at the stating node. [At the staring node, the pilot's action set is given by AP = (NY, Cuba}. Dopeniing on what action is chosen by te pilot, the terrorist tas her turn to move at node Ig or Il. The terrors ation set is Ape" = (B, NB) at the node Tic, and Afre"™" = (B, NB) at the ode [fy tn this simple game, the terorst's action seta happen to be the same at both nodes, bt this need not always b We caa tow give a formal definition to extensive form games with perfect information. Extensive form games with imperfect information fre defined ia Definition 2.17 on page 28 Dewirion 2.7 An extensive form game is: 1. A game tree containing a starting nade, other derision nodes, ter ‘minal nodes, and branches linking ench decision node to successor nodes.cy Basic Concepts in Game Theory & A list of N 21 players, indeed by i, i= 1,2,...,¥. 4% For ench decision nade, the name of the player entilled to choose fn eetion, 4. For each player i, a specification of is action set at ench node that ployer i is entitled to choose an action. 5. A specification of the payoff to each player at each terminal node 2.2.1 Defining strategies and outcomes in extensive form games Our profiminary dissassion of extensive form games emphasized that a player may be eallec to choose an ection more than once and that eaeh time & player chooses en action, the player has to choose an action from the ection set available at that particular node ‘Therefore, we need to define the following term, Drrinitton 2.8 A strategy for player i (denoted by s') is a complete plan (list) of actions, one action for each decision node that the player is entiled to choose an action, ‘Ths, it is importan! to note that a strategy is not what a player does at a single specific node but i list of what the player does at every node where the player is entitled to choose an action. ‘What are the stravegies available to the terrorist in the Pilot-Teeeor sine deseribed in F gure 2.17 Since the terrorist may end up in either nade Ho oF Ly, strategy for the terrorist word be a specification of ‘the precise action she will be taking at each node, ‘That is, althengh itis clear that the Lerrorist will reach either aoe JJe or Hy but unt both, 8 slrategy for this payer must specify what she will da at each of the ‘wo nodes. Therefore the terrorist hae four possible strategies given by (B, B), (8, NB), (NB, B), (NB, ND), where the first component refers {to the (errorist’s action in node 27, and the second component refers (0 her action at node Fy Since the pilot is restricted to making 1 move anly at node J, and since his action set as two possible actions, this game has eight out comes given by (NY, (B, B)), (NY, (B, NB)), (NY, (NB, B)), (YY, (NB, NB), (C,{B,B)), (C, (B, NB)), (C, (NB, B), (C, (NB, NB) 2.2.2 A normal form representation for extensive form games Now that the game is well defined, we geek to find ome predictions ‘The first step would be to search for a Nash equilibria. Recalling our ofnition of Nash equilibrium (Definition 2.4), in extensive foren games AEC a 2.2 Extensive Form Games 25 we ook for a Nash equilibrium in stratagies, wise each player cannot Increase the payotf by unilaterally deviating from the strategy played at the NE outcome. It cnens out that in many instanees transforming an extensive form ‘game into « normal form makes it easier to Gnd the Nash equilibria. Ta. bic 2.4 provides the normal form cepreseutation for she Pilot-Terrorist game deseribed in Pigure 2.1. Table 24 shows that there are three Nash rrorist (B.B) _(B,NB)(NB,B) (NB, NB) sy Gate yaa] cups [Tr It —iyt T Pitot ‘Table 2.4: Normal form repeesentation of the Pilot Terrorist game ‘equilibrium outcomes for this game: (NY, (NB, NB}), (VY, (B, NB)) and (CUBA, (NB, B)). Note that here, as in the Battle of the Ses game, multiple NE greatly ceduce our ability to generate predictions from this game. For this ceasou, we now turn to defining an equilib- rium concept that would narrow down the set of NE outoouies into a smaller set of oxtcomes. In the ltersture, an equilibrium concept that selects @ smaller number of NE outcomes is called a refinement of Nasi, equilibrium, which i the subject of the following subsection, 2.2.3 Subgames and subgame perfect equi In this subsection we define an equilibeiwin concept that satisfies all che sequireme (sve Dutinition 21) und has voue additional cesiie- tions. his equilibrium concept may be helpful in selecting a smallee ot of outcomes from the set ot NE outeames, by aliainat.ng some uncesie- able NE outcomes, Before we proceed to the formal pact, let us go tack to the Pilot ‘Terrorist game and look at the three NE Gutcomes for this game. Cont- Paring the three NE outcomes, do you consider aay equilibrium outeomes to be unseasonable? What would you suggest if the pilot were to hire you as her strategic advisor? Woll, you would probably tell the pilot to fly to New York. Why? By lookiig at the verrorist's payolls at che terminal nodes in Figure 2.1 we can see that if the pilot Bies to NEV YORK, the terrorist will NOT BOMB (a payoit of x = 0 compared with af = Lif she does), and she pilot will gain a payoll of x? = 2 compared with a payolf of r® = 1 for fying to Cubs. In other words, after the pilot ies to any destination (New York, or Cuba) the terror. 'si's payoff is maximized by choosing the NOT BOMB action. Fron26 this we conclude that the Hinitrtion of the NE concept is that it cannot capture the pilot's ability Yo predict that the terrorist will not have the incentive ta explode the bomb once the plane arrives in New York {in te Ciba), More precisely, under the NE outcomes (CUBA, (NB, B)} and (NY. (B, NB}) the terrorist roems to be polling what game theorists call fan snevedile threas, since the verrerst’s payatis ab. the terminal nodes Indicate that once reaching either node ZTe-ot Ti, the terrorist will not explode the bomb, ‘We now want to formalize aa equilibrium concept that would exclude the unrensonable Nash equilibria. In particular, we look for an equilit~ umn concept that would exclie outcomes where the terrorist commits hreesef tn the BOMB action, sitce such an action is incredible, Moreover, swe seek to define an equilibrium concept where the player who mraves Feet (the pilot in our case) wuld calculate and take into account how tnlsequctt, players (the terrorist in the present case) would respond to the moves of the players who move carlir in the grime. Hence, having computed how eubsequent pleyers would respond, the first player ean Gptimize by narrowing down the fet of actions yielding bigher payor. Th the Pilot-Terrorist example, we wish to find an equilibrium concept that would generate a unique outcome where the pilot fies to Now York ‘We first define a subgame of the game. Dprintrion 2.9 A subgame i ¢ decision node from the original game long with the derision nodes and terminal nodes drecty following this fhode, A subgarne is eolled a proper subgame if it differs from the original gare Ciearly, the Pilol-Terrorist getne has three subgames: One is the gatne jeeaf whereas the abher two ar proper subgames with nodes 1c and Jy. rte otorting nodes. ‘The two proper subgames are iustrated in Figure 2.2. He (Terrorist moves) 1 tom getvony bom" gat-bomb é ~ a ela oe wlan er vot Hel fad Figure 2.2: ‘Two proper subgames In 4065, Rheintintd Selten proposed a refinement of the NE concept fied as follows 2.2 Extensive Form Games Deswrrtox 2.10 An outcome is said to be a subgame perfect equi librium (SPE) if it induces @ Nash amuiibranm in every suigome of the original game. Definition 2.10 states that a SPE outcome is alist of strategies, one for each player, consisting of players’ actions that constituies a NE at every subgame, In pacticular, » SPE outcome must be a NE for the original ‘game since the original game is a subgame of itself. Note that in each subgame, the action NB is a NE. ‘We now seek to apply Definition 2.10 in order to salve for a SPE of the Pilot-Terrorist game, Claim 2.4 The outcome (NY, (WB, NB)) constitutes a unique SPE for the Pilot. Tervorist game. Proof, Since a SPE is also» NE for the original game, itis suffi- cient to look at the three NE outcomes of the original game given by (NY, (B,NB)), (Cubs, (NB, B)) and (NY, (NE, NB)). Next, each proper subgame has oniy one NB, namely, the terrorist chooses NB. Hence, igiven that a SPE oateome must be a NE for every subgame, we con clude that the outcomes (NY, (B, NB), (Cuba, (NE, B)) ace not SPE. Finally, the outcome (NY, (NB, NB)) is a SPE since it is 2 NE for the original game, and the outcome (action) NB is a unique NE for every proper subgame, . ‘Thus, we have shown that using the SPE refines the NE In the sense of excluding some outcomes which we may consider unzeasonable ‘We conclude this discussion of the SPE by describing the mthodolo ies commonly used for finding SPE outcomes. The general methocology {or finding the SPE outcomes isto use backward duction, mecning that ‘ve start searching for NE in the subgames leading to the terminal nodes. ‘Then, we look for NE for the subgames leading the subgames leading to the terminal nodes, taking as given the NE actions to be played in the las: evbgames before tho terminal nodes. Then, continuing to solve backwards, we reach the starting node and iook for the action shat max: itmizes player 1's payolf, given the NE of ali the proper subgames. Note ‘hae the backward induction methodology is particularly useiul when the game tree is long, Finally, another common methodology iste fiat find the NE outcomes for the game, say by transforming the extensive form representation into a normal form representation (see subsection 2.2.2), ‘Then, once we have the set of all NE outcomes, we are let to select those outcomes that aze also NE for all subgames. ‘This can be done by trial and ereor, or, a we do in the proof of Claim 2.4, by ruling ont the NE outcomes of the original game that are not NE for sone proper subgazes.28 Basie Concepts in Game Theory 2.3 Repeated Games Repeated eames are used only once in this book, in section 6.5, where ‘we analyze collusior among firms tinder iaaperfect competition A repeated game is a one-shot game that Is identically repeated more ‘han once. The importance of analyzing repeated games ie thet under certain circumstances cooperative outcomes, which are not equilibrium outcomes under a cne-shot game, can emerge as aquilibrivin outcomes In repeated, noncoaperative gatne A repented gain: is a special “kind” of an extensive forin game in ‘which each period, all players move simultaneously and each player's ction set does not vary with time; in a more general extensive forin ame, actions sets may dilfor from one time period to another. More precisely, a repeated game is a one-shot game (see Definition 2.1) that is repented for several periods, where the repeated gamie is played once in cack period, Each period, after the game is played, the players move to ‘the next period. In a subsequent period, the playere obseeve the actions chosen by all players (including their own) in all previous periods, and ‘only (en simultaneously choose theit actions for the new game, ‘Thus, the important thing tp remembor is that players enn perfectly monitor all the actions chosen in earlier periods prior to choosing au action in ‘8 subsequent perce. The data collected by perfectly monitoring the ‘actions piayed in each period is called a history at « period ‘To define the players? strategies under @ repeated game, we now wish to modify Definition 2.8 to repeated gatnes: Dewnrrion 2.11, 1A period 1 history of w game, Hy, is the lst of outcomes played in all pevieds (2 1,2,...47 = 1 2A strategy of a player in a game repeated T times, is a list of ‘ctions that the player takes in each period t,t = 1,2,.-.,Tj where ‘each period t wction, af € A‘, ie based on the period ! history of ‘the game (ise, of maps 6 history H, lo an action in the set A") Teuce, a strategy of a player in a repeated game is list of actions to be played in esch period +, where encls period r action of player i is Doased ou te observed list of actions played by all players in all periods 1,2,...,7 1 summarized by the history H,. ‘Therefore, an outesme of a repeated game would be a list of actions each player is taking in every period, whereas the period r payoff to ench player is a function of le actions played by the players in period 1. Consider our Peace-War game described in Table 21, and suppose that this game is repeated T times, in periods 1,2,...,77, where T is JAE 2.8 Repeated Games 29 fn integer number satisfying 1 0. In this case, Wwe can assume that the economy's real interest r would adjust so that r= Yp~lorp=1fll +r) si We now make the following assumption regarding the players’ payolts in a repeatod game ASSUMPtON 2.1 Let af denote the action taken by ayer in period E12, C5 12.047. Also, let m(at,of) be the period t payoff to Payer i, 4 = 12, where x! is given in Table 21. Then, the payoff to player i when the game is repeated T times is defined by mm Spt taflel.a2) { ila} a) + eni(al, a3) +e si(ob, ah) YT 1/2. Proof. Let ws look at a representative period, call i period r, and Suppose that country 2 has aot deviated in periods 1,...,7. Then, if ouizy 1 deviates and plays a = WAR, Table 2.1 shows thot nf = 3. However, given shat country 1 deviates, country 2 would deviate in all subsequent periods and play af = WAR for every > + +1, since country 2 plays a tigger strategy Hence, from petiod r+ L aad o0, country Learns a pay of Leach period. ‘Therefore, the period “+1 eum of discounted payots to country 1 forall periods tr +Lis 1/2 Since no unilateral deviation i bene ficial to nay country at any subgame starting at an arbitrary period +, we conclude that no unilateral is beneficial toa country at any period t So for, we have showed that when both countries play the trigger stzategy no country has the incentive to unilaterally deviate from playing PEACE. In the language of game theorists, we showod that deviation fora the equilibrium path is not beneficial to eny country. However, to prove that the trigger strategies constitute a SPE: we need to show that one conntiy deviates and plays WAR, the other country would adhere to Its trigger strategy and would play WAR forever. In the language of ane theorists, lo prove SPE we need ta prove that no player has the Incentive to deviate from the playod strategy even if the game proceeds ‘off the equilibrium poth. To prove that, note that if conntry 1 deviates from PEACE in period 7, then Definition 2.12 implies that country 1 will play WAR forever since Definition 2.12 states that any deviation (by country 1 or country 2) would trigger country 1 to play WAN forever, Hence, comtry 2 woud punish country 1 by playing WAR forever since WAR yields payoff to country 2 of 1 each period (compared with payoft of 0 if country 2 continues playing PEACE). Altogether, the trigger sirategies defined in Definition 2.12 constitute a SPE for the infinitely repeated Peace-War game 7 Proposition 2.5 demonstrates the relationship between the players! rate of time preference, given by p, and their incentive to deviate from the cooperative action. That is, when players have & low rate of time preference (say, p Is lose to 26r0), the players do not cate rouch about future payorts. "Hence, cooperation cannot be a SPE since the players ‘ish to maximize only their frst period profit. However, when pis large (p> 1/2 in our cass) players do not heavily discount future payee, 0 cooperation booomes more beneficial fo the players since the punishment ‘on deviation becomes significant. because the discounted flow of payoffs under cooperation (2 per period) is higher than the short-run gain from TUT Ta 24 Appendix: Games with Mixed Actions 38 deviation (a payoff of 3 for oue period and | thereafter). This discussion leads to the following corollary Corollary 2.1 in an infinitely repeated game eoogeration is easier to sustain when players have a higher rate of time preference, 2.3.3 A discussion of repeated games and cooperation In this section we have shown that a orie-shot game with @ unique uon- cooperative Nash equilibrium can have a eaoperative SPE when it is repeated infinitely. However, note that ia the repeated game, this SPE is not unique. For example, itis easy to show that the noncooperative outcome where each country plays WAR. in every petiod constitutes a SPE also. Moreover, the Falk Theorem (Folk, becaus> it was well known to game theorists long before it was formalized) states that for a suf ficiently large rate of time preference, a Ineye suanber of outcomes in the repeated game can be supported asa SPE, Tih, the fact that we ‘merely show that cooperation is a SPE is insuiiciert to conclude that | game of this type will always end up with cooperation. All that we ‘managed to show is that cooperation i a possible SPE in an infiaitely repeated game, Finally, let us lool at an experiment Robert Axelrod conducted ia which ke invited people to write courputer programs that play the Pris oners' Dilemma game against other computer prograns a large number of times. The winner was the progratamer who mavaged to scon largest sum over all the games played against ail other programs, ‘The important result of this tournament was that the program that used ‘8 strategy called Tit-for-Tat won the highest score, ‘The Tit-for-Tat strategy is different frou the triggor steategy defined in Definition 2.12 because it contains a less severe punishment in ease of deviation. In the Tit-for-‘Tat strategy, a player would play in period ¢ what the oppoaeut played in period t~ 1. ‘Thus, even if deviation accurted, once the oppo nent resumes cooperation, the players would switch t9 cooperation in a subsequent period. Under the trigger strategy, once one of the players dovintes, the game enters 2 noncooperative phase forever. 24 Appendix: Games with Mixed Actions ‘The tools developed in this appendix are not implemented in this book, and are brought up here anly for the sake af completeness. ‘Thus, this appendix is not necessary to study this book successfully, and the be- Binning readers are urged to skip this appendix, Games with mixed actions are those in which the players random ine over the actions available in their action sets. Oiten, it is hard toa4 Basie Concepts in Game Theory motivate games with mixed actions in economics modeling. This is not ‘because we think that players do not choose actions randomly in real ie (On the contrary, the render ean probably recall many instances in-which hear she decided {o randomize actions. The major reason why games With mixed actions ate hard to Interpret is that it is nob alwavs clear ‘why the players benefit from randomizing, among their pure nctions, "The eteractive feature of games with mixed actions is that a Nash cqnilibeium (in mixed actions) always exists. Reval tat Proposition 2.2 demonstrates that a Nash equilibrium in pure actions noed not always Tn whint follows, our analysis will concentrate on the Tep-Bottom- Loft-Right given in Table 2.5, The reason for focusing om the game in Ms. 0 ett) Right) Ms, a __ Titop) [0 OTO 1] Bibotiom) [Tet] ‘Doble 2.5: NB in mixed actions how that a Nach equilibrium ip mixod actions exists 1m in pure actions does not (the yin pure actions ‘Table 2.5 is that despite the fact uhat a Nach equilib tender i urged to verify that indeed « Nash equilit does not exist) ‘We ntow wish to modify a game with pure strategies to a game where the players choose probabilities of taking actions from their action sets. Recall thal by Definition 2.1, we need to specify three elements: (a) tis lst of players (already defined), (b) the action set available to each player, and (¢) the payoff to each player at each possible outcame (the DEFINITION 2.13, 4, A mixed action of player a is probobility distribution over play ing 0 = T and playing a® = B. Formally, o mixed action of layer 0 18-0 probably 7, (0.< 7 <1) such thet player o plays T with probability 7 and pleys B with probability 1 — 7. 2. A mised action of player 6 is a probability X, (0 < X <1) such that player A plays E wrth probolsty © ana plays Me with probability 1 J. An action profile of ¢ mite actions game és 0 fst (7,3) (oe the list of the mized ection chosen by each player). 2.4 Appendix: Games with Mixed Actions 35 4 Anoutcome of a game with mized actions is the list of the real ization of the actions played by each player. Dion 2.13 impli eat the sisson set of eah pape ese ner] whew pager pits a7 Dy and lager Bacio ee (ih The center has pot iol sat Defoe 21 trees new term ation prof wl places the tem outcome usd seat foe gus, Defeton 21. The vowot fr intodeany nie han at na gene wt med actions, the payers cee ony probable fr plying tir sagen ao the automo tl stanton lo ene with pve actos he tr aston poe andthe Len ute ee rise ection theta ection prof ad to Soest Ine Brabniltydntbattons over saon chosen Uy eh pla oe the term elcome species ist facts pape by seh sae ne the uncertainty is resolved. Here eteeeee Our dion of ir "nied extension of the gate b norte unless we specify the payoff to each player under all possbie action profiles. Derintrion 2.14 payoff function ofa player in the mixed-action game 4 the expected value of the payoffs of the player in the game uith the pure actions. Formally, for any given action profile (A,r), the expected Payoff 0 player 4,1 = 0,8, is given by Berd) = rAs'(T,L) trl —d)a(T,R) + (L=7)Aa'(B,0)+ (= eB, R) According to Definition 2.1 our game is now well defined, since we spect ified the action sets and the payoll functions defined over all possible action profiles of the mixed actions game. Applying the NE concept, defined in Definition 24, to our mixed: actions game, we can state the following definition: Dernsrriow 2.15 An action profile (F,4) (here 4,4 € (0,1), is sad to te a Nash equilibrium in mixed ‘actions if no player would find 1€ beneficial to deviate from her or his mized action, given that the other Player does not deviate from her or his mized action. Formalin, Bee A)& Bx%(r,3) for every 7 € (0, 1} (es) BRR 2 BRA) for every N€ [0,1]36 Basic Concepts in Game Theory ‘We now turn to slving for the Nash equilibrium of the mixed-actions ‘extension game of the game described in Table 2.5, Substituting the payoffs associated with the “pure” outcomes of the game in Table 2.5 into the “mixed” payolf functions given in Definition 2,14 yields Ex(s,A) = rAxO47{1—A) x0 eo) + G=Mx Te (1 90=a)x (2) = -n)aa—1) (A) me PAS Oe r=) x (=1) en + (Lara 04 (1-7) =A) x3 = (1 2)(9— 45). Restating Definition 2.15, we look fore pate of probabilities (#,4) that satisfy two conditions: (a) for & given 4, # maximizes Ex9(r,.) given ‘n (2.6), and (b) for a given #, 3 maximizes Ex°(F, \) given in (2.7) It is easy (0 check that the players’ payoffs (2.6) anid (2.7) yield best-response functioas (see Definition 2.5) given by 1 Hace 0 itrea/a RO)=f [041] 3/2 and RE) = 4 Joy itt w BA oo isa 1 itr 3/4 (28) ‘That is, when player d plays K with a high probability (I-A > 1/2), player 0° best tesporse i to play T° with probability I(r = 1) in ovder ‘0 asnimize the probability of getting m payelf of —1. However, when player 8 plays Z with a high probability (A > 1/2}, player are best espe i bo play J with prokubilty 1 (7 = 0) jn order to tonainize he probability of getting & payoff of +. Similar explanation applies to ihe hestrespone function of player ‘The bet-response fonctions of each player are drawn in Figure 2.3 Equotions (2.8) and Fgure 2.3 show that wien @ plays A= 1/2, player is inciferent to the choice among all her actions. That is, when d= 1/2, the payoff of player asthe same (ero) for every mined action + € [1] In particular, player as inlifferent tothe choice between playing a pure strategy (meaning that + 0 or 7 = 1) and playing any other mixed fictions (0 < + < 1), Sinlarly, player 6 Is Indliferent to the choice ‘among. all her mived setions A €[0,1), when player a plays r = 3/4, Aldiough a NE in pure actions does not exist forthe game described in Table 25, the folowing propestion sh: Proposition 2.6 There exists © wnigue NE in mized actions for the sgome described in Table 8.5. In this equilibrium, 7 = 3/4 and A= 1/2. 2.5 Appendix: Games with Imperfect Information 37 T RSA) A Ry ea) 2 1 1 7 RO) a 7 A a ol ee Figure 2.8: Best-response functions for the mixed-action extended game ‘The proposition follows directly from the right-hand side of Figure 2.8 tat shows that the two best-response functions given in (2.8) hav a unique intersection Finally, the best-response functions given in (2.8) have a property of being composed of horizontal or vertical line segments. Since the vie Xbrium occurs when the two curves intersect in their “‘midale” sections, we have it that under the NE mixed outcome, each player is indifferent {0 the choice among all other probabilities that can be played assim, ing that the other player does not deviate from the mixed action. "This esult makes the intuitive interpretation of a mixed-action game rather dlficult, because there is no particular reason wity each player would stick to the mixed action played under the NE. 2.5 Appendi: Games with Imperfect Information Games with imperfect information are brought up bere only for the sake of completion, and the beginning readers are urged to skip this ‘appendix. Games with imperfect information describe situations where some players do not always observe the action taka by another player earlier in the game, thereby making the player unsure which ade nee bbeen reached. For example, Figure 2.4 describes a variant of the Pilot, ‘Terrorist game given in Figure-2.1, I Figure 24 we suppose that the terrorist cannot monitor tie direction in which the plot is fying, say because the terrorist cannot read a compass or because the pilot dissbleg some ofthe navigation equipment. The broken line connecting nades Ip and Ly describes an information set for the terrorist. ‘The informatiog sot tells us that in this game, the terrorist cannot distinguish whether node Ho ot Ly has been reached. Thus, when the serrorist has her turn to make a move, she has to choose an action without knowing tne precise node sit is ou. Formally,38 Basie Concepts in Game Theory J (Pile mows) muy 10 aia, Puy TO New York (Torrorit moves) Ho 7 SHU (Peerorist moves) Bons mj), then Ihe assumes that ji honest and it bying. Henes, in this ease, the airline manager will pay ny ~ 2 to trnvelor 3, and m2 to traveler j. Thus the manager penalizes the craveler asswmed to be lying and rewards the 2.6 Exercises a one assumed to be honest. (6) IF ng = ny, then the manager assumes that both travelers are hon and paye thers the declared value ofthe antiques. Letting ny and nz be the artions of the players answer the following questions. {2} Under Definition 2.6, which outeoenes ave Pareto Optienal? (b) Under Definition 2.4, which outcomes cons om for this game, a Nash xpi 4. Consider a normal form game between thre ino car produces. C.F, fad G. Bach producer can produce either large ears, or small eats bout aot both, That i, ehe action sa of each producer i, 4 = C,F.C. is AY = (SM, LG), We denote by a! the action chosoa by playec i, and by x'(a%,a",°) the profit to fim {Assume that the profit function of each player fis defined by 1 Wa! = £0, fora j=0,R6 1 fel =5M, forall j=CRG @ ifn! = 2G," anda = SM for ats 46 @ ifa'=SAf, anda) = LG for al #4 B fat = a 210, and at = SU, Ak AE B iat mat = 9M, andat = LG) 2 EE Answer the following quostious. (9) Does there exist a Nash equilibrium when a > 0 > > 07 Prove your aaswer! (8) Does there exist 3 Nash equilibrium when a >> 8 > 0? Prove your answer! 5, Figure 2.6 dasribos an extensive form version of the Battle of the Sexes ‘ome given initially in Table 2.2. Work through the billowing probiens, () How many subgames are inthis game? Descrion and pla all the subgarnes. (b) Find all ehe Nash equilibria in each suk-ame, Prove your snsse! {) Find all she subgame perfect equilibria for thia game, () Before Ractol makas her more, she hears Jacob shouting that he intends to go to the opera (Le, play a). Would such a statement change the subgame perfect equilibrium outexmes? Prove and explant 5. (This problem refers to mieud actions games studiel la the appendix, section 2.4) Consider the Battle of the Sexes gam describes in Ts" ble 22, {a} Denote by 0 the probability tat Jacsb goes t che OPERA, and by p the probability that Rachel goes to the CPERA, Formulate the expected payeif of each playera Basic Concepts in Game Theory (is) Deaw the bos-responte function foreach plaver[(p) and (0) (6) Who is the NE in mixed actions for this eame? (2) Calculate the expected payoll to aack player in this NE, {e) How many times do the to best-response functions intersect? Explain the difarence inthe member of intersections between this nme aud the best response functione Uiustéated in Figure 23. och x A Ur \e dco ? te 4 NX < 2 wag Figure 2.6: Battle of the Sexes in extensive form 2.7 References ‘Aumann, Ro 1987. “Gaine Theory” In The New Palgrave Dictionary of ‘Bernomics, ited by J. Eatwell, M. Migate, and P. Newman. New ‘Yook: The Stockton Press Axelrod, RL 1984, The Evoluhon of Cooperation. New York: Basle Books Binnore, K. 1992. Fon ond Garnes. Lexington, Mass: D.C. Heath, Fretiinn, 3. 1986. Come Theory with Applications to Boonomics. New ‘York: Oaford University Pres. Fudenbergy Dy and J. Title, 1001. Game Theory. Cambridge Moss: MIT Press. ons, R 1992. Game Theory for Applied Beonomists. Princeton, Nl Princeton Univesity Pras. MeMitla, J. 1992. Games, Sirsegien, ond Managers, New York: Oxford University Press, Movin, H. 1882. Gore Theory for Ure Soetal Sciences, Now Yorks New ‘York University Press Osborne, M., and A. Rubinstein, 1994, A Course én Came Theory Cam bridge, Mass: MIT Pros. Rasmusen, E1989, Games and Fyormation: An Intraduction to Came ‘Theory. Oxford: Blacewel HHH HEH Chapter 3 Technology, Production Cost, and Demand Large increases in cost with questionable increase in_ per formance can be tolerated only for race horses and faney [apouses), Lord Kelvin 1824-1907 (President of the Royal Society) ‘This chapter reviews basic concepts of microeconomic theory. Ser tion 3.1 (Technology and Cost) introduces the single-produet production function and the cost function. Section 3.2 analyzes the basic properties ‘of demand functions. The reader who is fareliar with these consapts and properties can skip this chapter ond proceed with the study of in- dustrial organization, The student reader should note that this chapter reflects the maximum degree of technicality needed to grasp the material in this book. ‘Thus ifthe reader finds tbe material in this chapter to be ‘comprehensible, then the student should feel technically well peepsted for this course 3.1 Technology and Cost ‘The production function reflects the know-how of a certain entity that we refer to as the firm, This know-how enables the firm to transform factors of production into what we call final goods. In general, we refrain from ‘addressing the philosophical question of where technological know-how coines from. However, in chapter 9 (Research and Development) ve do analyze some factors that affect the advance of technological know bow.“4 ‘Technology and Demand 5.1.1 The produetion function We ass ne that twa inpats are noeded to produce the single Snel good We call these inputs lator and capital. Note that we restrict our die. cuseion to production technologies for producing one and only one type of output. In reality, many production processes yield more than one type of output. Far example, an oll refinery yields a variety of oil and plastic products from the same input of crude oil. We pastpone the liscussion of multiproduct production activities to our analysis of the airline industey given in section 17.2 The production function represents » mapping from the amount of labor (denoted by 1) and the amount of capital (denoted by h) emplayed in the production process to the number of units of output prochiced We represent this zelationship by a function f, where the mumaber af Units of ovtput ie given by @ = fC, 4), Assuming that the function fis twice continuously differentiable (with respect to both arguments), we define the maryinal product of labor function (Af, (t,)) as the amount of output increase ossociated with a sinall inerense in the amount of labor. Formally, we define the ‘marginal product of labor and capital functions by arts) . afk) MPLOR = and MP Rok) = LOY 1) For example, the marginal-product functions associated with the clase of production functons Q = (I + k)%, where «3 > 0 are given by MPL(Lk) = atl +h YPMUPF and MPR (0,8) = Bo( 4 8O)P° KO Its important to note that the marginal product of a factor isa function {not necessarily # constant) of the amount of labor and capital nved in the production process. Jn our example, limyig MPs (i,K) = roo, meaning that in this production process, the ruarginal product of labor sets larger and larger as the amount of labor becomes scaree, So far, we have ot discussed the relationship between the two face tors, We therefore make the following definition, Deewtriow 3.1 4. Labor and capital are called supporting factors in a perticular Production process ifthe increase in the employment of ene factor raises the morzinal product of the other factor. Formally, if OMPLDK) _ OMPellAK) . 4 ea 3.1 Technology and Cost 45 2 Labor and capital are called substitute factors in a particular ‘rodction process, ifthe increase in the employment of one factor decreases the marginal product of the other product. Farmaly, if MPU) _ OM Feb) 5 a or In our exainple, the reader can verify that labor and capital ere sup- Porting factors if 4 > 1, and substitute factors if 8 < 1 ‘We conclude the discussion of the production function by looking at the effect of input expansion on the amount of production. Formally, Derinttion 3.2 Let A be any number greater than . Then, a produce tion technology Q = f(l,) is said to exhibit 4, Increasing returns to scale (IRS) if f(A, Ak) > Af(U,4). That is, if expanding the employment of labor aad capital by the factor of A will increase the ouput by more than a factor of N 2 Decreasing returns to seale (DRS) sf f(Al,M) < Af(I,4) That %, expanding the employment of labor ard capital by the Jactor of will increase the output by less than a factor of 3. Constant returns to scale (CRS) if /(A, Mk) = A/(I,k). That 4s, i expanding the employment of labor and capital by the factor of X will increase the output by exaclly « factor oj In our example, the production toch exhibits (RS if touy Q = FL) = (Ie RY? (00% + 2K)°y7 > Ale LAS? if and only if 98 > 1 8.1.2 The cost funetion ‘Tho cost function is a mapping from the rental prices af the factors of productiou aud the production level to the total produstion cost. The ost function isa technological relationship that can be cerived froin the production function. Let W denote wage rate, and R the rental price for one unit of capital ‘Tne cost function is denoted by the function TC(W, RiQ) measures the total production cost of producing @ units of output, when factor prices are W (for labor) and R (For capital). We define the average cost function by the ratio of the total peo- duction cost to output level. Formally, the average cost function (the46 ‘Technology and Demand cost per tinit of outpat) at an cutput level @ is defined by AC(Q TCIQMQ. We define the marginal ensi fnction ns the change in total eost resulting from a "small increase in ovtput level. Formally, the marginal ‘cont function at an output level Q is defined by MC(Q) = “7490 ‘As an example, consider the total cost. function given by TC(Q) = F + 6Q?, FC 20. This cost ftnction is illustrated on the left, part of Figuce 31, We for to F as the fred cost parameter, since the fixe 0007 J ZL cig) TC) ve se, and mnarginal cost functions Pigure 3.1: Total, aw cost is independent. of the ontput level Tis straightforward to calealate that AC(Q) = F/Q + cQ and that MC(Q) = 209. The average axd marginal cost functions are drawn on She gt pat of Fg 51. The AEC (Q) con i in and sing iy , and has a slope of Je. The AC(Q) curve i falling with Q as long ns the output level is suicently small {Q < yF/e), and is rising with Q for higher output lovels (Q > VFTe). Thus, in this example the cost ‘per unit of output reaches n minimum at an output level Q = Fe. We now demonstrate an “easy” method for finding the output level that minimizes the average cost Proposition 8-1 If the average cost fsction roaches @ minimums at fs ctnclly povitive output feel then at that particular oxtrut level the leverage east equals the marginal cnet. Formally, if Q™* > 0 minimizes (ACIQ), then AC(Q™) = MEQ™) Proof. Al the output level Q, the slope of the AC(Q) fmction imust be zero. Hence, aacrqn) _ 27") aQ 3 worgnsjann — Tala") rm 3.1 Technology and Cost a Hence, Mog? = FOI) soya, a . “To.demmonstrate how useful Proposition 3.1 could be, we now setuen to our example illustrated in Figure 3.1, where TO(Q) = F + o@# Proposition 3.1 states that tn order to figd the nutput level that mini- ries the cost per unit all that we need to do is extract Q™ from the equation AC(Q™*) = MC(Q™). In our exaznple, ACIQ™) = Bie + cM = 20" = MCIQ™). one Henoe, Qu" = V/Fie, and AC(Q™*) Mo(Qn) =F. 3.1.8 Duality between production and cost fun We sow provide simpla illustration of the relationship between pro- duction and cost functions, for the case of a single-input production fonction. Suppose that only labor is required for producing the final good, and let the production technology be given by Q = fi = I", 7 > B, This production function is ilustrated fa the upper pact of Fig: ure 4.2, for three parameter cases where <7 < 1,7 = 1, and > 1 In what follows, we show how the cost function ean be derived from the production function. Let W denote the wage rate. Now, by inverting the production function we obtain {= Q'/T. The total cost is the ‘wage rate multiplied by the amount of labor employed in the produetion proves, Hence TC = Wi = WQ', which is illustrated in the lower part of Figure 3.2, again for the throe parameter eases whore <7 < 1, yel,and y>L ‘We conciude this discussion by looking at the relationship betwora the production and cost function regarding the expansion of the produc tion activity. More provisely, applying Definition 3.2 to the production function Q=17, we have it that (ANT > AM if and only if y> 1 Hence, this production exhibits IRS when 7 > 1, ORS when y= 1, aod DRS when 7 <1 Isis important to realize that since the total cost function is derived from the production function, we should be able to infor from the shape of the average cost function whether the production process exhibits IRS, ORS, or DRS. When 7 > I, there are TRS. The case of IRS is4s ‘Technology and Demand / TE) = 20? Te = 19h gee eereereee. AC = wQ' AC = wQ's* 2 @ Figure 3.2: Duality between the production and cost. functions ‘tlustrated on the sight side of Figure 3.2. Under IRS, te average cast cines with the output level, reflecting the fact that under IRS the eost ‘ver unit declines with a larger scale of production, say, because of the ‘uloption of asseanbly tine technology. Under CRS, the east por wait is constant, reflecting m tecmelogy where an inerense i the output level {docs not alter the per unit production cost. ‘Phe left side of Figure 3.2 rofiecis a DRS technology, where an increase in the output level rises the per unit production cost : Finally, recall ovr two-input example where Q = (194 k*)9, We showed that this production technology exhibits IRS iff > 1 and DRS if @B <1. Deriving the cost fanction of this production technology would take us beyond the level of this book. However, forthe sake of i lustration we state that the cost function assoctater! with this technology is given by TCUW, HQ) = 6Q™ whore ¢ is & nonnegative function of W and R. Now, in this esse AC(Q) = AQ, ‘Then, AC(Q) is declining with @ if 1/{o8) ~1 <0, ar af > 1, which is the condition under Which the technology exhibits IRS, In contrast, AC(Q) is vising with Q 3.2 The Demand Function 49 if 1/(a) 1 > 0, oF a < 1, which is the condition under which the technology exhibits DRS, 3.2 The Demand Function ‘We ddenote by Q(p) the (aggregate) demand functoa fra single product, vrhere Q denotes the quantity demanded and p denotes the unit price ‘Formally, a demand function shows the ‘maximum amount consumers ate willing and able to purchase at a givin market pre. For exam wwe take the finear demand function given by Q(p) = §~ 1p where « and are stritly positive constants to be estimated by the econametrcian, Alternatively, we often use the inverve demand function p{Q), which expresses the maximum price consumers are willing and able to pay for a given quantity purchased. Inverting the linear demard function yields P(Q) = a ~ 1G, which is drawa in Figure 33, Nove that pact of the > elastic: up| > anit elasticity: [al Pee TA _ inelastic: Ing <1 \ ; _21@) = AR1Q) =~ ¢ aN $ at R(Q) Figure 8.3: Inverse tinear demand emand is not drawn in the Sgure. That is, for p > a the (inverse) demand becomes vertical at Q = 0, 20 the demand coincides witha the vertical axis, and for Q > a/b, it coincides with the horizontal axis ‘An example of nonlinear demand function is the constant elasticity demand function given by Q{p) = ap~* oF (Q) = aVSQ-HE, whieh is drawn in Figure 3, This class of functioas has sone nice features, which we diseuss below. 3.2.1 The elasticity function "Tho elasticity function is derived from the demand function and maps ‘he quantity purchased to a certain very useful umber which we call‘Technology and Demand p W(Q) = a'eqrie _ Figure 3.4; Inverse constantclasticily demand the elasticity at « point on the demand. The elasticity tieasures how fast quantity demanded adjusts to 0.small change in price, Formally, we dlefine the demand price elasticity by 220) 2 (32) wl SG ) DerIserION 3.3 At a ginen quamtity level Q, Ube demand is called 1, elastic f14(Q) < 1 (or bf Qi > Bs 2 inelastic if -3 < np(Q) <0 (ors np(C@It <1 3. and hes a unit elasticity if mp(Q) = —1 (ors pl @)i= 1 For example, In the linear case, 7p(Q) = 1 = o/(0Q)- Hence, the demand he @ wait clacicity when Q = o/(2h). Therefore, the demand is elastic when Q e/(21). Figure 3.3 iilustrates the elasticity regions for the lineat demand case Tar the cossnt-elasty donand function QUp) = op * we hae i that mp = al 2p 'p/(ap*) = —e. once, the elasticity is constant aka poner ofthe rie vale demand fancion Hee =, {is demand fonction bas # snl elasticity at all usput levels 3.2.2. ‘The marginal revenue function "The inverse demand function shows the macimam, amount a consumer is ‘willing to pay per unit of consumption at a given quantity of purchase Tho tatol-evenne function shows the amount of revenue collected by sellers, associated with each price-quantity combination, Formally, we (pot wo hve it. 3.2 The Demand Function BL FQ)Q. Fo te ine ease, TR(Q) ~'00 = 80" a ithe courant ec Soman T(Q) = 0'"Q!—U Note thas fore sale sme fo ie seven futon pou be teal the ttt egret ice we ata ert onmuner ened tae thas peice creme, "Tb, Ogu epee nocd mt eal due’ tren, br example, when tse we Pred on consi: ‘Tha, ths ttl sve oncton mre how tk Stnsomes open at vey tive mart pe a ot nee the ‘enn nee by producers The morgiaevoue function (ogi, wove spot tse th “maga expen) shows teat by wich oa creme Stroner hen tho conse lily ier the nanan tay buy onal we ice th aint romnae faction by BR) = SES Foc the na demand ete we cate tin ling Proposition 3.2 If the demand function is linear, then the maryinal- revenue function is also Hinear, has the same interorpt at the derand, but has twice the (negative) slope. Formally, MR(Q) =a ~ 250. Proof ATR) _, dloQ- 1G") RQ) dQ dQ = 289. . ‘The marginal-revenue function for the linear case is draw in Fig- ture 33. Tbe marginal-revenue curve hits 2ero at an output level of Q = a/(2). Note that a monopoly, studied in chapter 5, will never pro- duce an output level larger than Q = @/(28) whore the marginal cerene is negative, since in this ease, revenue could be raised with a docresse in ‘outpat sold to consumers For the constant-elasticity demand we do not draw the corresponding ‘marginal-tevenue function. However, wo consider one special case where = 1. In this case, p= aQ~!, and TR(Q) = a, which is a constant, Hence, ME(Q) = 0. You have probably already noticed that the demand elasticity and the marginal-revenue functions are telaied. That is, Figure 3.3 shows that MM2(Q) = 0 when np(Q) = 1, and AFR(Q) > 0 when [np(Q)] > 1 ‘The complete relationship is given in the following proposition. Proposition 3.3 MQ) = 00) [t+ 5]82 ‘Technology and Demand Prong. EPRQ) _ ap(Q)Q] M UR(Q) ae =? b Tae . 3.2.3 Consumer surplus We conclude our discussion of the demand structure by a grass approx: Imation of consumen' welfare associated with trade. We define a mea. sure thot approximates the utility gained by consumers when they are allowed to buy a preduct at the ongeing market price. That is, suppose hat initially, consunecs are prokibited from buying a certain product, Suppose next that the consumers are allowed to buy the product at the ongoing market price. The welfare measute that approximates the welfare gain associated with the opening of this market is what we call consumer surplus anil we denote it by CS. Jin what fotlows we discuss a common procedure used to approximate Consumers’ goin from buying by focusing the analysis on linear demand functions, Additional motivation for the concept develope in this sec- tion is given in the appendix (section 3.8). Pigure 3.5 illastrates how to caleulate the consumer surplus, assuming that the matket price ls p. Qe) = 8S F gure 34: Consumers’ surplus For a given market price p, the consumer surplus is defined by the area beneath te demand curve above the market price. Formally, de- nating by C5(p) the consumers surplus when the market price is 7, We ose) (= 99% 2 (33) 3.8 Appendix: Consumer Surplus: Quasi-Linear Utility 52 Note that C'S(p) must always incresce when the merlot price is reluced, reflecting the fact that consumers’ welfare increases when the market ples falls Jn industrial organization theory, and in most partial equilibrive ‘analyses in economies, it is common to use the constmers’ surplus as a measure for the consumers’ gain from trade, that is, te measure the gains from buying the quantity demanded at a given matkot price compared with not buying at all. However, the séader should bear in mind that this measure is only an approximation and holds true only if consumers hhave the sowalled quasi-linear utility function analyzed in the appendix (seotion 3.3). 3.3. Appendix: Consumer Surplus: The Quasi-Linear Utility Case ‘The analysis performed in this appendix is brought up here only for the sake of completeness; quasi-linear utility is used only once in this book, in section 19.1, where we analyze two-part tarifs, We therefore advise the beginning student to skip this appendix. In this appendix, we demoustrae that-when consumer preferences are characterized by a class of utility functions called zuasi-inear willy Junetion, the measure of consumer surplus defined ia subsection 3.2.3, ‘equals exactly the total utility consumers gain from tuying in the mar- vet. Consider & consumer who has preferences for two tems: money (m) ‘and the consumption level (@) of a certain product, which he can buy at a price of p per unit. Specifically let the consumes utility Funetion be given by U(Q.m) = VO+m. (3.4) Now, suppose that the consumer is endowed with a fixed income of f to be spent on the product or to be kept by the consumer. ‘Thea, if the consumer buys Q units of this product, he spends pi) on the product and retains an amount of money equals tom = ~pQ. Substituting into (34), our consumer wishes to choose a product-eansumption level Q to maximize VO+1 99. (as) ‘The first-order condition is given by 0 = 8U/8Q = 1/(2V@) ~p, and the secoad order by °U/8Q? = ~1/(1Q"%?) <0, which constitutes a suicient condition for & maximum. ‘The first-order condition fora quasilinear wtlity maximization yields the inverse demand function derived from this utility function, which is mgxU(Q.1 ~ 70)54 ‘Technology and Demand given by 1” a-e (6) wa 2 “Thus, the demand derived from a qusilinen uty function sa n= ‘ran laste dono Fveton utente carer in Figure 4, 90 mo drawn In Figure 36 Q) [igure 2.6: Inverse demand generated from a quasilinear utility function ‘te shaded area in Figure 86 corresponds to what we call consumer serplas in subsection 3.2.3. The purpose of this appendix is to demon strate the following proposition r function ts generatad from 0 quasi-linear Proposition 8.4 Ifo demand function is generat ily function, then the ars rarked by C(p) in Figure 8.6 mencucs Ccoclly the utlty the consumer gnina from consuming Q” units of the produet at @ market price p°. Pron, The area CS) in Fie 3.6 ie ealeulted by ae ga z es = f° (55g) 40-9 (a7) = V@— pA" = UAT WP} constant 3.4 Exercises 1. Consider the Cob-Dauglns production funetion given by Q = °K, where a, > 0 3.4 Exercises 55 (2) For which values of the parameters a and does this production technology exhibie IRS, CRS, and DRS? (9) Using Definition 3.1, infer whether labor and capital are support Ing or substitute factors of production, Consider the production funetion given by Q =U +4", where > 0 (2) For which values of dos this production technology exhibit IRS, (CRS, and DRS? - () Using Definition 3.1, infor whether Isbor and capital are suppor'= ing ot substitute factors of production, Does the production function given by Q == 1-4 VE exhibit IR3, CRS, ‘or DRS? Prove your answer! Consider the cost funtion TC(Q) = F +69, where Fe > 0. (a) Calculate and plot the 7C(Q), AC(Q) and M1C(Q), {b) At what output level i the average cost minimized? (6) afer whether vis tchnoleny exhibits IRS, CRS, o¢ DRS. Expl CConsier the demand function Q-= 99. (a) At wht ots love dows the esti gant 2? {b) At what ouput lve dos bw aeticity oqaal 1? (6) Catculate and draw the marginal-povenue function associat with this demand. (a) At wat output level does the marginal evens equal zero? 6. Consider the constantelsticity demand function Q = Ap™* where Ave 0. (©) Calculate the consumers? surplus when p = 3 and () Solve forthe iaverce demand funcsion p(Q). (0) Using (8.2), eatoulate the demand price elasticity, (e) For what values of¢ isthe domand elastic? Por what values is the demand inolastic? (2) Using Proposition 3.3, show thas te catio ofthe matsinal-yevenue function tothe iavorce dem function, 9(@)/R(Q).s indepen. dent of she output level QPart II Market Structures and OrganizationMarket Structures and Organization 50 We define market structure as.» description of the firms’ behavior in. a given industry of market. The list of items defining firms’ behavior include precise specifications of (1) The actions availabie to each firm, (ea. choosing a price, setting quantity produced, setting ptortuetion npucity or location, etc.), (2} The number of firms im the fedustry ind whether this number is fixed or whether free entry of new Sems is Sliowed. (3) Firms’ expectation about the actions avaliable vo competing firms, and how the competing firms will respond to each frm’s action (4) Firms’ expectation about ths umber of firms and potential entry "Thus, specifying a market, structure is simila to specifying the rules of the gume or rules for interaction among existing or potentially entering farms. Ia many cases, specifying a market structure is similar to sling a game ancording to Definition 2.1 on page 13) igure TL on page 61 lists most of the market structures used in this book, The top of the tree in Figure [1.1 shows that market structures tre classified into two categatics: competitive and imperfectly competi- five. The competitive market structure studied in chapter 4 (ard which You have probabiy studied in your intermediate microoconomie clas) as James that each Brwn’s action sct is its production quantity, while each fra tokes the market price a8 given, where the market price is deter- fined by the intersection of the wnarkot demand curve and the industry's taggregate supply curve. Competitive market structures can be solved forby assuming either a fixed number of fms (sometimes roferced to 8s short-run equilibrium) or free entry (sometimes eeferred to as a long-run euilbrinan), "Among the imperfectly competitive market structures, the reader is probably most familiar with the monopoly market structure, which js studied in chapter 5. Under this market structure, ther? is only fone seller, who can choose any price-output combination on the con- Cumers) aggregate demand curve, Given the one-to-one relationship etwoen price and quantity implied by the market demand curve, the monopoly is restricted to choosing a price or a quantity produced bu not oth, Monopoly market structures can be classified as static, where the monopoly sells its product only once, oF dynamic, where the monopoly ells (durable or nondurable goods) over more thas 9ne period. Monopoly aarket structures ace then classified into discriminating and nondiscri {nating monopolies, A discriminating monopoly can earn a higher proft thaa e sondiscriminating one by selling the product to different consumers at diferent prices. ‘The duopoly (two sellers) and the oligopoly (more than vo sellers) ‘market structures are classified as cooperative and noncooperative. Co- Sperative behavior is defined by Arms’ colluding by agreeing '0 produce cr otal the monopoly’s profit-maximizing outpat level or to charge the0 Market Structures and Organization movopely’s price. A noncooperative behavior can be modeled either using one-shot games where all firms choose their strategie variables cumantity produced cr price) once and at the same time, or dynaraically where the firms move in sequence. Whether firms move simultaneously ‘or whether they move in sequence, firms choose either prices (Bertrand) ‘or quantity produced (Coumet) Finally, one marlet structure that economists tend to focus on ase ssuine that firms are engaged in a repeated interaction of a simultaneous. ‘move oligopoly game, ‘That is, in each petiod, each fina chooses its action from the same action set after observing what actions have been ‘choven in earlier periods. The upward arrow in Figure IL hints that a fascinating possible cutcome of an infinitely repeated oligepoly gasne is where firms chooee to play their collusive (cooperative) actions (output Jevel oF price) Market Structures and Organ 61 — [PERFECT comPaTTTION| (price taking behavior) Decisions quatity produced eee x A DUOPOLY, [MONOPOLY] Ce | (acorn (Stal demand) sal dora) Some Sure ce at aN (ess Dungy) - \ [Sranic] [DYNAMIC ZN. “N t5i (RONCOOPERATIVE) [CocPaRNaWVE : ORR Ue i (cles 6 price or austy tN Pte ee erat ae (lesder-ellower models) Dersson: price or quantity Decision: pri or quantity aN LN Figure (11; Commoaly assumed and used market structures (ote: Dadiseriminating, ND=noodiscrimizating)HED Chapter 4 Perfect Competition In perfect markets, whether monopolistic oc competitive, price i hardly a matter of judgment and where there is n> judgment there is no policy. —Baward $. Mason, "Price and Production Polisies of Lange Seale Enterprise” ‘This chapter describes perfectly competitive markets. We first reed to define what do we mean by the term competitive market, of equivae lently, a perfectly competitive market. We define a competitive market (or perfect competition) as a market where agents (buyers and sellers) behave competitively. But, what do we mean by competitive betavioc? In economics, the following definition is commonly used for competitive behavior. Dertnrriow 4.1 A buyer or a seller (agent in what folloms) is said 0 be competitive (or alternatively, fo behave competitively) if the agent ‘assumes or believes that the market price is given and that the agent's factions do not influence the market price. ‘Thus, the assumption of competitive behavior relates only to what agents believe about the consequences of their actions. That is, competitive behavior implies that agents think that their actions (say, quantity- produced) will not have any effect on the market price tis important to note that the assumption of competitive behavior is iodepeodent of how many firms or consumers there are in the market it telates only to beliefs. More precisely, assuming competitive behavior ddoes aot imply that the number of sellers is large. In fact, one of the ex- excites accompanying this discussion asks you to define and soive fr the ‘competitive equilibrium price wben there is only one seller in t3e mar-64 Perfect Competition het. ‘Thus, as Jong as she agents behave competitively, the competitive ‘qilibrium price can be solved for any muruber of buyers and sellers. The common mix-up between the assumption of competitive beligy- ior ad the assumption that the muraber of sellers must be large stems from two reasons: Fire, the assumption of price-taking beievior seems ‘more reasonable when the mumber of firms is large, and each firm selle & sinall amount relative o the aggregate industry sales. Second, the equi= ‘ibrium price solutions for some innperfectly competitive market struc Lures converge on (ges closer to) the competitive price when the mimber ‘of firms increases. ‘Therefore, when there ig a large mumber of sellers, ‘quilibrinm price under various masket structures gets closer ta the price solved by competitive Sehavior; nevertheless, the definition of competi live behavior is completely independent of the nutmber of firms, Snppose that our coususners demand a homogeneous prodvet, De noting the price of the product by p and the (aggregate) quantity de manded by Q, we sesime that consumers’ aggregate inverse demand funetion is Tinear and i given by Q) 2-80, where a,b > 0, (ay 4.1 Non-Increasing Returns to Scale Suppose that there sre two firms (named firm 1 and érta 2) producing, this homogeneous preduct. We denote by g the quantity produced by finn 3, and by 7'C;(q;) the total cost fonction af firm i, ¢ > 1,2. To be ‘more specific, let us assutne that the firms have constant returns o scale (cchnologies summarized by linear cast functions given by Ta) =a 12 1,2, where @ > 01 20. « ‘The linear cost functions have the property that the marginal cost (the increment in cost duo to @ small increase in the produetiou level) enuals the average cost (cost per unit of production), Formaally, ey and ep are called constant unst extts of production if ¢, satishies PEsla) a. ACA(qi), for every gy, 1 = 1,2 (43) Jn goneral, constant anit costs are associated with constant-returns-to scale (CRS) production functions since CRS production functions rep- resent technologies where doubling the inputs would double the output ‘constant unit costs mean that doubling the output will exactly double Ue total enst of production. Observe that in equation (4.2) we assumed with ne los of generality Hal fitm 2 hes a higher unit production cost than firm 1 (or an equal 4.1 Non-increasing Returns to Scale 65 ove). Figure 4.1 illustra ‘output space. the demand and the unit costs in the price- Figure 4.1: Competitive equilibrium under constant returns to seale [Now that the economy is well defined, we define a competitive eq librium as a vector of quantities produced and a price sich that (1) each firm chooses its profit-maximizing output at the given equilibrium price, and (2) ab the equilibrium price, aggregate quantity demanded equals aggregate quantity supplied, Formally, Deeisrrioy 4.2 The triplet [p*,af, qf} is culled a compet librium if 1 given pt; af salves mala) = pa -~TCa), 61,2 Ot + a8): Now that we have defined competitive equilibrium, we seek to solve for ‘this equilibrium for the industry desecibed in (4.1) and (4.2). The first step would be to calculate the supply functions of the two firms, which fare found from the prof-maximization procedure deined in part 1 of Definition 4.2. Lemma 4.1 The supply fimetions are given by oo fp>a a= 4 (G00) fra 4) 9 Ypce66 Perfect Compet Proof, Since each firm i (teats p et a constant, the firm’s profit margin Sefned by pc, # constant, Hence, p~ cy Is treated by the firm as the Constant persunit profit (loss if negative). Therefore if p~ ci > 0, the firm would produce g) = a0, and if p— cj <0, the frm would produce 0, whereas if p— 6 = 0, the firm makes a zero profil at every level Ur production implying that the eutput Tevel is indeterminste, . ‘We search for the enuilibeinm orice that would satisfy Definition 4.2 However, observing (4:4) cam tall us which prices cannot constitute an ‘equiliirium. More specifically, aay price above the unit cost of frm 1 {Gr > ex) eannot be an equiibciuim price, since (4.4) tells us that if p> ¢1s tr = 26: however, the demand function (4.1) tells us that the qventi remanded (at ny’ yrice) ie alwys finite, Hence, for p> cy the quantity ‘supplied exceeds the quantity demanded, thereby violating past 2 of Definition 4.2. ‘Tins, if a competitive equilibrium exist, it most be that 7° Sex However, if p < ¢1 © cs, the suppy Functions (4.4) imply thot qx = 2 ~ ‘f, and since the quantity demanded is greater than zero, it exceeds the ‘quantity supplied, thereby violating part 2 of Definition 4.2, Bence, Proposition 41 Ifa > & price ts f= ey, and cay the unique competitive equlibrivin 1. son > €1.45 = 0 (firm 2 i not producing) and gf = 5 2 ifea = cy then QO = af + gh = SG and ange 2 0. That fs the aggregate industry letel is determined, but the division of the Industry output between the frm ts indeterminate Finally, let us make three remarks: (a) Observe that fe < o1, (ane ing that the demand is ks}, then neither firm would produce. (b) This tnedel ean be easily extended to any number of firms. Clearly in equi ‘only the firin(a) with the lowest nit cost would produce. (c} Definition 4.2 allows us to impose the competitive morket structure even It there is only one firm, For exainple, if there is only one firm with @ funit cost 2 0, then p* =e and g* = Q? = SPF constitute a unique competitive equilibrium. 4.2 Increasing Returns to Scale “The analysis in subsection 41 i valid only if firme’ tachnologies exhibit decreasing or constant returns to scale technologies. Suppose now that firme have increasing-returns-t-scale (IRS) tecimologies. To simplify ‘ve assuine that there is only ene firm whose total cost of production Jp oompased of «fixed cost (independent of the production level) snd 8 4.2 Increasing Returns to Seale er constant marginal cost. Formally, the total cost of producing utits of output is given by rom={5°* (45) Figure 4.2 ustratee the marginal and average-total cost functions asso- ciated with this technology, showing that the average cost decreases and ‘approaches the (constant) marginal enét as the output level inczeases {ince the average fixed cost approaches zero}. MC(a) =e —- ee ee Cg Figure 4.2: Decreasing average cost technology Our main result is given in the following proposition: Proposition 4.2 Let a > c. If firms’ technologies ezhibit increasing returns to scale (decreasing average cost), 4 competitive equilibrium does rot esse, Proof. Bya way of contradiction, suppose that = competitive equilibrium exists, Then, from Figure 4.2, the equilibrium price has to satisy one of the following: p* <¢, or p? > & a) Suppose that 2° = pf Sc Then, pf < F/q+o= AT Cig) for every q > 0. That is, the equilibrium price is below that average cost Tor all sirctly positive output levels. Hence, the firm would produce g° =0. But, g° = 0 cannot be an equibrinm, since at this price range, the quantity demanded is strictly positive, and excess demand violates part 2 of Definition 4.2. {(b) Now, suppose that pe = ph > ¢ Then, ph > F/g-+e = ATCC) for q exceeding a cectain level. That is, the equilibrium price is above the average east for sufficiently large output levels. Moreover, the per- tunit profit measured by 2 ~ ATC(q) increases with q, implying that ‘the competitive Sem produces q° = +00. But g* = +00 cannot be an ‘equilibrium; the quantity demanded is always finite, and excess suppl violates part 2 of Definition 4.2 768 Perfect Competition 4.3 Marginal-Cost Pricing and Social Welfare In this section we detnonstrate & very jmportant feature ofthe competitive: equilibrium outcome More precisely, in this section we demonstrate that the peefectly competitive market structure yields 6 market outcome that maximizes social welare, to be defined below. We frst wish to define a social welfare function for our economy. Ja subsection 3.2.3 on page 52 we defined the concept of consumer surplus, denoted by CS(p), and showed that this measure approximates ‘con. sumers' utility level at a given market price. In order to fully capture he economy's welfare, we also need to take into consideration the fact ‘at firms are owned by our consumers, and therefare we defined social welfare hy the sum of consumer surplue and firms’ profits. Formally, Devwmrnion 4.3 Let the market price be given by p, and suppose that there are N > I firms im the industry, We define social welfare by wo) CS(P) + D> mi). Im what follows we show that the perfectly competitive market steue- tune yields a market price that maximizes social welfare az defined in Definition 4.8. Indeed, we are going to prove something more general an that, We will show chat when Lhe market price equals the marginal ‘ast of producing firms, thea the qnantity produced and eousuesed max imizes social welfare, Now, given that competitive equilibrium results in moaginal-cost pricing, itis Clear the competitive outeome maximizes welfare, Pigure 4.3 illustrates the welfare level for every given market pie. Figure 4.3 illustrates three important areas under the inverse demand ‘curve if we assume Uist the market price is pp > 0, ‘The consuiner sur- plus, defined in subsection 3.2.3, is given by C'S{pp) = a.. The industry profit is the distance betwoen price and unit cost multiplied by the quan ‘ity sold, and is therefore given by }r{pq) = B. By definition, the total welfare is given by W= a+ 8. Figure 4.3 shows that the area marked by + is not part of measur ing welfare, Indeed, the area measured hy 7 Is oansidered to be the sleoduweight loss associated with higher-than-margival-cost pricing. The inwwition behind the definition of the deadweight loss area js that since the demand function shpes downward, a bigher-thaa-marginal-cast price would reduce the quastity demanded, "The eonsumier surplus loss assoc. ated with @ lower consumption level eanot be fully eaphured by a higher profit level (if any) astociated with a higher price 4.4 Exercises 60 Figure 4.3: Marginal-cost pricing and social welfaw: CS(p) = a, Ex) = 8, Wp) =a+8 Figue 4.3 shows that when the market price is reduced feos py «6 the deadweight-Ioss area mergos into the consumer surphis. In addition, the reduction in industry profi is offet by the merease in consumer surplus. Altogether, social welfare increases with a price reduction as long as price exceeds marginal costs. Finally, notice that wwe do not discuss cases where market prices ate below unit costs (p < «) since when the price is reduced below marginal cost, the increase in farms’ loss exceeds the increase in consumer surplus 44° Exercises The market demand curve for a certain product given by Q(p) = 120 p, where p is the mncket price and Q denotes the quantity purchased by the coamuisers, Suppose that the product i prodursd with a single factor of production called labor, denoted by £. Assume that gach frm i ean hire say amount of labor at a fixed given wage eate, denoted by w > D, The production function ofeach Kem # is siven by q where £y ths amount of labor employed by frm 1. Suppose that there is only one firm producing this product; call t Sew 1 Solve the fim’ profit maximization problem and peeve that tho Rema supply curve is given by py ven by aa 20 2, Suppose now that w = 1. Using Definition 4.2, solve for the coupestive equilibrium price and quantity for this singlefrm indsstry 3. Calculate the profit of this firm in « competitive equim70 Perfect Competition “4 Now, eupote that there are two firms, whese xia levels are denoted Peet tl ge, Save for the competitive equleium price and quantities rode by each fim. Compare the morkot price and aggregate production when the eompet- {aioe watlibrivin is eolved for s single frm and when i is solved for @ ‘wore industry 6. Draw the supply curve of each frm, and then plot the agerexete industry ‘Sunvly curse, Label production on the horizontal axis and price on te Silica ari, Then, draw the ivdistry"* demand carve, and graphically Solve for the ecmpetitive-equilbrinm price. 4.5 References Mason B. 1909, “Price ond Production Policies of Large-Scale Enterprise.” Mimerican Economde Fevicw 23, pt. 2: 61-74 VUDEEEEEEEEEEEEEEE EEE HEH Chapter 5 The Monopoly Every person who shall monopolize, or attempt to monopo- lize, oc combine ond conspire with any other person oF per~ sons, to monopolize any part of the trade or commerce -.., shall be deemed guilty of a felony. “Sherman Antitrust Act of 1890. In this chapter we develop a theory of a single seller facing competitive (price-taking) consumers in one or several markets, over one of sev- ‘eral periods. It is importaat to fully understand the extreme momopoly Case since when few rms compete, the firms can always exercise some monopoly power. In addition, for the sake of simplicity several argu tents i this book are demonstrated only for the monopoly case rather than for some other forms of market structnes. ih single sellor is facing a downward sloping demand curve. ‘Thus, since consumers are always on thoir demand curve, the monopoly can determine either the price for the product or the quantity supplied. That is. decision about price implies a decision about quantity produced and ‘vice versa, since quantity and price are related via the demand curv. For this zeason, the monopoly needs to devote resources to the careful study of the demand curve facing ite product, that is, the moxopoly thas to familiarize itself with all the demand properties discussed in sec tion 3.2. Aftor estimating the demand curve, the monopoly bas tostudly the market demand to determine its proft-maximizing outpat. ‘Section 5-1 presents the fariliar monopoly profit-maximization prob- lem for a single market, Section 5.2 (Monopoly and Welfaze) reviews the Sstandaed” welfare argument demonstrating the welfare loss associated swith a lower-ban-optimal production level. Section 5.3 (Discriminating Monopoly) departs froma the single-markst assumption and anaiyaes @ profit maximizing monopoly that can charge diferent prices in diferent2 ‘The Monopoly matiets. Section f.4 (The Cartel and the Multiplaat’ Monopoly) ana 120s two forms of collusive contractual arrangements among alte Remo Producing in the industry that, tagether, behave as a monopoly prof ‘uvximizing entity. Section 5.8 (Durable Good Monopolies) analyees the manepaly's weliaver over a period of time, where the monopaly calle © good that provides services for more than one period. ‘The appendie {sertion 5.6) discusses the egal antitrust approach to the monopoly and o price discrimination 5.1 The Monopoly’s Profit-Maximization Problem The technology of the firm is summarized by its east function, which relates the quantity produced to the cost of producing this quantity Let PC(Q) denote the total cost function of the monopoly. Denoting by (2) the monopolys profit level when producing Q units of output, the Imionopoly ehiooses Q™ to mgex(Q) = TR(Q)- TC(Q) A necessery (but ust uficent) condition for Q”™ > 0 to be the manopaly's profi-masimizing output is HO") ATAQ™) AFC) rom aa a IG ig = MRQ™) —NCQ™). (6.1) Notice that (8.1) is only necessary condition, sneaning that if the profit- maximising ontput i strictly positive, then it has to satisfy (5.1). Hom «ver, especially i he monopoly has te pay high fixe costs, tis possible that the monopoly’s profit-maximizing output lovel is 0. AR together, (5.1) implies that if a profl-maxiimiting monopoly produces 8 sitiely positive ouput level Q”, then the profit output lovel must satisy the condition 4FR(Q™) = MO(Q™). ‘Thus, the easiest method for nding the monopoly’s profit-maximizing Dutput level is first ta solve for Q” from (5.1), and then to substitute to the total profit function to check whether m(Q") is greater than F equal to zero, If it is not, then the monopoly sets Q™ = 0, and iC profit ie nonnegative, then the output Jevel solved from (8.1) ie the rofit-maximizing output level Alter finding the monopoly’s profit-maximizing output, the price reed by tho monopoly can be foud by substituting Q™ inte the Seman fonction, Figure 6.1 iustrates the monopoly solution for the ‘ase where TC(Q) = F + 02, and a linear demand function given by MQ) = a ~ bQ. Figure 5.1 (left) shows the case where the demand 4s high enowgh (or the fixed cost is low enough) so that Lhe monopoly UCU 5.2 Monopoly and Social Welfare 3 MEQ) w "RQ ee VE atRQ) Pigure 5.1: The monopoly’s profit maximizing output would produce Q" > 0, and hence would charge a price of p™. Fig. ture 5.t (right) illustrates a case whore the demend is 40 low that tbe monopdly’s price cannot cover the average cost. Hence, Q” = ‘To solve it explicitly, aote that by Proposition 42 we have it that MR(Q) = @ = 259. Hienco, if Q™ > 0, then by (51) Q™ solves a — 26Q" = 209, implying that an) m ym a(0-+ 20) "= aig tnd bene ta’ Ta Consequently, (Q") = TR(Q)-TC(Q") ab+2) pif a \?_ at Tere -* (seq) = agieg Altogether, the monopoly’s profit-maximizing output is given by { wry FFsay ° otherwise on 5.2 Monopoly and Social Welfare ‘The US. legal system discourages monopolies (see the appendix, sub- section 5.6.1). In what follows, we provide two arguments for why mo- nopolies are discouraged. 5.21 The conventional argument against a monopoly Figure 6.2 illustrates the conventional argument against monopolies ‘The monopoly equilibrium (p",Q") is ilustzated in the left side of74 ‘The Monopoly ura & Figure 5.2: Monopoly and social welfare Fignce 6.2, where the aten CS measures the consumers’ surplus (see subsection 3.2.3). We define the total welfare, W, as the suin of indus try profit and consumers’ simplus. Fottally, Ws + GS, which is measured by the entire shaded aren of the left side of Figure 6.2 ‘The right side of Figure 5.2 illustrates a welfare-improving case in- volving marginal-cost pricing associated with perfectly competitive mar- bets (see section 4.3 on page 68)- Compating the monopoly outsome with the marginal-cost pricing outcome reveals that whereas the industry's profit is lower under marginslcost pricing (possibly zero), the C5 is flearly much larger under marginal-cost pricing. That is, the gain to total welfare when the macket outcome changes fcom monopoly to per fect competition i precisely the dendweight-loss area (marked by DL) sscociated with the monopoly market structure. 5.2.2 The social cost of « monopoly argued that the cost to the society associated with the ‘existence of a monopoly is rch higher than the deaditeightoloss area iarked by DD in Fignze 5.2. That is, following Tullock (1967), he argued that the pursuit of monopoly teats is itself a competitive activity, and fone thst consuites resources. This activity was given the term rent serking by Krueger (1974). More precisely, Tullock and Posner srsued tht the social cost of having & monopoly should also inchude the costs of deterring competition that are analyzed in section 8.3 and in Section 8 4 ‘The point is that fms, wishing to obtain 2 monopoly status or wishing to maintaining a monopoly position, must allocate resources for that onl, ‘These resouress may or may not be counted as a waste to the 5.3 Discriminating Monopoly 5 ‘economy. Resources allocated to establishing or maintaining monopoly power that should not be considered as reducing welfate include: 1. R&D leading to 8 patent monopoly right for seventeen yrs (see seetion 9.4) since the RED improves technologies and results in ew products 2, Bribes to politicians oF civil servants for the purpose o” getting exclusive business rights (since this‘ eoastitutes only a transfer of wealth) Now, resources allocated to the establishment of monopoly power that may count as eocial waste include: 1. Persuasive advertising (sce section 11.1), needed to convince con- ssumers tbat alternative brands are inferior 2. Resources needed to preempt potential entrants (rom entering the industry. Also, excessive production or investment in capital for the purpose of making entry unprofitable for potential competitors (ee section 8.3) 3, Lobbying costs, nooded to convince the legislators that a particular monopoly is not harmful (provided that these casts divert resources fiom productive activities) 4, Excessive R&D resulting from a potent race 5.3 Discriminating Monopoly (Our analysis so far bas focused on monopolies charging a single, unt form price to all customers. A firm can, however, increase 3 profit by charging different prices to consumers with different characteristics. ‘That is, a firm may be able to differentiate among consumers according to tastes, income, age, and location in order to charge consumers with different characteristics different prices. Note, howover, that in order to be able to charge consumers dilfereot prices, a firm must possess the means for making arbitrage (buying low for the purpose of reeling at ‘ high price) impossible. In other words, price discrimination 's impos- sible when those consumers who are able to purchase at a low 2rice can make a profit by eeeelling the product to the consumers who buy at high plows. ‘Thus, fms resort to various marketing techniques te prevent arbitrage from taking place. For example, 1. Firms can charge different prices at different locations. In this case, inorder for price discrimination to be sustained, the market76 ‘The Monopoly should be isolated by geography, by prohibitive taxes (auch as tar- ifs), or Uy prohibitive transportation costs such as those resulting from product spoilage while being transported from one location to another : 2, Firms that provide services (auch as transportation companies, restaurants, and places of entertainment) charge senior citizens lower prices than they charge younger consumers, In this case, for the price discrimination to be sustained, the firtn must demand ‘hat sonior citizens presont their ID cards, 3. Firms can sell discount tickets to students. In this case, the seller will ask for a student ID card from those consumers seeking to purchase at » discount. 4, Book publishers manage to charge institutions higher prices than ‘hey charge individuals by selling hardeovers to institutions and softcovers to individuals In what follows, we do not analyze how the monopoly manages to segment the markets so thst no arbitrage can take place between two ‘markets with different market prices. The exemples given above pro- vide some explanations of how a firm can prevent arbitrage between ‘wo markets. In addition, subsection 14.1.5 demonstrates that a firm can prevent arbitrage by tying the basic product to some service for service-demanding consumers while salling it without service to other conkumers, Here, we meraly assume that arbitrage cannot take place Consider a moncpoly selling in two different markets. We assume tat the two markets ate isolated in the sense that the monopoly ean charge different prices, and the consumers cannct perform arbitrage by buying in the low-price market and selling in the high-prico market. We ow seek to investigate how a monopoly determiues the output level (hence, the price) iz each market. Figure 5.3 illustrates the demand schahules in the two markets (market 1 snd market 2). ‘The lef side of Figure 6.3 illustra‘es the demand function and the derved-marginal- revenue funiction in narket 1. The middle Sgure illustrates the demand ‘and marginalsevenne functions in market 2. ‘The right side of Figure 5.3 ilustrates the oggregate demand facing the monopoly (Dy + Ds), and tive horizontal sum of the marginal-revenie functions (5° MR). ‘Tho monopoly choases the output levels soll in each market, gf and oP, that solve waxroive) = TA(a) + TRale)~TC(g +a). (62) 5.3 Discriminating Monopoly ita cn PD Cun % Fur, oe Sue Figure 5.3: Monopoly discriminating between tro markets If the monopoly sells » strictly positive amount in each market, then the following two first-order conditions are satistod: Oatar, 06 af) Wha? ~MC(G +g) for ewh i= 1,2. (5.3) Hence, the discriminating monopoly equates MR (af = MRa(af) MC(af* + af), when it sells the proft-maximlzing outyutlevls ie cock market ‘The intuition behind this condition is as follows: “F the monopoly chooses af" and gf such that M/Ry(qi") > MRy(@P), then itis cleat that the moaopoly should transfer owe wat from market 2%0 market 1. In thio case the reduction in revenue in market 2 is analler than the increase in revenue in mathet 1 ‘To solve for the proit-maximizing output lovely gf? and of, we ueed to solve two equations with the two variables givea i (¢3). Instead, we provide a three-step graphical illustration for bow to solve this problem First, uote that Figure 5.3 illusteates how the total praduction level ts determined by the intersection of S°MR with the MC(al" = qf) to determine the aggregate production level Qr* = gf 420" Secoad, to seometrically nd the output level sold in each marke, daw &horlenntal line fom the intersection of EMR, = MC(Q™) to the MR, and MR functions, ‘This determines the amouat of output sold ia each mackee (af" and 9). Third, to find the price charge in each market, note that consumers are “always on their demand curves,” hence, extead vertical lines from ganda to the corresponding demand curves to locate pi" and pp. Finally to find the relationship between the price charged in each ‘atket and the demand elasticities, Proposition 2.3 and equation (5.3)16 ‘The Monopoly imply that pi? Af) = PE CL+ Ate) Hee, > if eh Pe fis rualing that easy iso nagative umes), Hemet Proposition 6.1 A discriminating monopoly selling & strictly poste eerie tach market wll chorge © higher pric of the market with the Teas elastic demand 5.4. The Cartel and the Multiplant Monopoly “the cartel and the multiplaat monopoly are forms of organizations and a eae agrements amon, pats, as, ot countries. For exampiey peice tie oil-producing countries as plants, the earl i an enka swe contract with the countries on how much exch would produce Me orccem what would be the workd price, Other examples of Or aie nee cthe IATA, (International Ait Transport Association), which rer Tato airfares, and bar asocintions, which regulate attorneys se plant monopot i very similar to the cartel except that all On pinnts re pu under 8 singe owners. Mltiplant monopoly at the Pir cover fms ithe industry merge together into e mele gee ergontsl merger), of when a onepoly fir opens several Plone producing the same product Ane ike the cartel, the mltiplant monopoly bas the power Yo decide wether to shut doen some of ita plants (or whether te open sete Moor). A earl generally doesnot st down pots or cruntrieg sere le logal reason that te cartel doesnot cwn the plant, end (er Float would jan the eacel Knowing that it could be shut dow We osume a linear aggregate demand given by p= a—b@. We now “defn the thology’ ench plant. We assume thet there are Y plans dere uy Gx LdvesnwN} Let qu denote the ovtpot level of lant in Oy at each plant ins the technology summarized by the tls] test fonction given by TOlaEP HAG), Foro a) “Thus wo assure that al lauts ave identical cost functions wc He TAP ont as a xed (ootpot independent) cost of F. The plant's Seer Fe and marginal cost functions are given by ATs) = 7/6 cl we Me alga) = 2e. Figure 5.1 on page 73 iustrate tis ost structure, ‘which is common to all plants 5.4.1 The cartel “te cartel ongnnies all the Y plant by direckng exch plan to prods aoe car ambent ‘The osjective is to masimize the sum of the profs \EEDEDUDDEDESORDESEOESOSEDSDEDOEH 5.4 The Cartel and the Multiplant Monopoly 70 of all the NV plants, Let x(q) denote the prot of plant i, and et the suegrogate cartel output be denoted by Q, = Tfarq- The objective Of the cattel i to choose die fay +N #9 ee : max H(gugay-ssan) = Lynda 63) non asl (5s) -Sr04e) “The cartel has to solve for NV quantities, so, after some manipulations, the N first-order conditions are given by Fe 4 an 0m BE 2b Yap MEsa) = MR(QI-MOa), F=22 al. Go) ‘Thos, Proposition 5.2. The cartel’s projt-masimizing output produced by each plants found by equting the maryinal-evenus function (derived fom The marke! demand curve, evaluated at the aggregate cartel-outpst level) to the maryinal-east function of cach plant, Since all plants have ideatical cost functions, we search for a sym metric equilibrium where the cactel directs each plant to produce the fame output lovel, That i,q) = z=... =n =4- Hence, { @—%Ng= eq implying that a= sayy ‘ jq= oq implying that a= aR OT) “The total cartels output and the market price are given by Ma On +26 =Nq= SA and p= (ON +20) eaten aay Meme nor (OM) Notice that when N = 1, the cartel’s output and prioe coincide with the pure monopoly levels. It can be easily veifed that as the aunber fof femas in the cartel increases (iV increases), both the output level of teach firm and the market price fall (q and p decrease). Hence, the total fevenve aad profit of each firm must fall with an increase in the umber Of carvel members. For this reason, many professional orgaxizations, Goch os those of lawyers and accountants, impose restrictions on new Candidates who wish to practice in their profession.80 ‘The Monopoly 5.4.2 ‘The multiplant monopoly The nnitipiant monopoly is very similar to the carta, except thot it has tive anihority (ownership) to shut down some plants, shereby “saving” variable and fixed costs associated with maintaining the plant. This, If we suppose that the moltiplant monopoly can choose the nnber of pants, that, N ix choice variable by the multiplant monopoly owner, eo the question is What is the proft-maximizing mizaber of plants ‘operated by the mutiplant monopoly? The answer is vary simple: given thet the mulkplant monopoly can ald or discard plans, the monopoly would seek to adjust the mamsber f plants to minimize the cost por unit of production. In other words tive mosiplant monopoly will adjust Une mumber of plants to minnie ATGlys) for every plant In operation. In order to domenstrate how the nomber of plana is determined, we approximate the menber of fis by a real (continuous) number Father than by an integer number. Like the eartal, the multiplant mnonepely would equnte MR(Q) = MCy(a) for every operating plant (yieing ‘output levels given in (57) equal to q: = gifyay) and in addition, vil adjust V so chat ench opersting plant would operate at miniusn ATCA), sven by gs = Fe. Hence, equating JF ]e = wees a0 solving for N yields that the profi-masinsizing numberof pants N= ‘Thus, the multiplast monopoly’s profi-maximising number of plants increases with an increase in the demand parameter a, and decreases with the fixed cost parameler of each plant F. 5.5 Durable-Goods Monopolies (Our analysis so far has focused on one type of goods called flow goods. By flow goods we mean goods that are purchased repeatedly and that perish after usage, for exanple, food products such as apples and bananas, and many plastic and peper single-use products. In contrast, durable goods ‘are bought only once in a long time and can be used for long time, for example, ears, housts, and laud. Clearly, with the exception of land, all goods eventually perish, so these two concepts ate relative to a certain Le hotizon that is relevant to consumers ‘Coase (1972) hyst pointed out that a monopoly elling a durable good will behave differenty from the (Iamiliar) monopoly selling a perishable ood analyzed earlie in this chapter. Coase considered the extreme case ‘of 8 person who owns all the land in the world, and wants to sell tat the 5.5 Durable-Goods Monopoties aL Jargest discounted profit. Clearly, Coase chose to analyze land because it is definitly a good example of a durable good. if land were perishable, then our analysis implies that the monopoly wont nat sell all the land. That is, the monopoly would restrict output (land) and raise Uke price high enough so that not all the land would be sold. Now, suppose that the monopoly charges the monopoly price and cls haif of its land by the end of this year. Let us try to predict sha will happen next year. Well, the mondpoly still owas the remainder ‘of the world’s land, and there is no reason why the monopoly will not offer that land for sale next year. However, itis clear that noxt year (if population is not growing vary fast) the deanand for land will be lower than the demand for land this year. Thus, the monopoly land price next _year wil! be lower than the monopoly prive this year. Given that the monopoly’s next-yenr price will be (substantially) lower than the monopoly land priee this year, it ie clear that those consumers who do not discount time too heavily would postpone buying land untif next year. Hence, the current demand facing the monopoly. falls, implying that the monopoly will charge a lower price than what a monopoly selling a perishable would charge Cosse's discussion of durable goods monopolies was formalized in Bognoli, Salant, and Swiernbinski 1989; Bulow 1982 1986; Gul, Son- fnschein, and Wilson 1986; and Stokey 1981. In what follows, we provide two siaple but rigorous analyses of durable-goods monopolies, Subsection $.5.1 demonstrates Coase's conjecture in an example for a downward sloping demand curve, Subsection 5.5.2 provides an example for a discrete demand in which there is @ faite aumber of eonsumecs, ach buys, at most, one unit of a durable good, and cermonstrates that Coase's analysis is false under this demand structure 5.5.1 Durable-good monopoly facing a downmard sloping domand Suppese there is a continuum of consumers baving diferent valuations for the annual services of a car that are summarized by the familiar downward sloping demand curve, Suppose that cousumers live for two Deriods denoted by t ¢ = 1,2, and that a monopoly sells a durable product ehat lasts for two periods. Thus, if a conaumer purchases the product, she will have it for her entire life, and sbe will not have to replace it ever again. ‘The consumers have different valuations for the product summarized by the aggregate period ¢ = 1 inverse demand funetion for one period of service given by p = 100 ~ @ and illustrated in Figure 54 (left). Figure 5.4 assumes that in period 1 there is a contimmm of consumers, each having a diferent valuatioa for purchasing82 ‘The Monopoly Ph Dy ® wink\ 100-4 7 MRa() Figure 5: Durable good monopoly: the ease of downward sloping de and fone unit of the product, Akogether, t demand illustrated in Figure 6 (lft). Ta the following two subsubsoctions, we compare the monopoly" profit ander two types of ccmmercial transactions: selling and rent- Ing. ‘Ta formally distinguish between selling nd renting we state the following definition. ey form & downward sloping Dewinion 8.1 1. By selling 0 product to a consumer, for a price of p®, the firm tronsfers all rights of oumership Jor using the product and getting the product bock from the consumer, from the time of purchose extended indefinitely. 12. Byreuting « product '9 a consumer, for e price of P®, the renter maintains ownership of Ue product, but contracts with the con- “wmer to allow the consumer to derive services from the product {for 0 vinen period specified in the renting contract. ‘Thus, selling means charging a single price for an indefinite period, whereas renting means charging @ price for using the product for a spe- Ce, limited time period, Tr should be emphasized that Definition 6.1 {does not imply that by selling, the manufacturer always transfers all rights on the product sold. For example, even when a product is sold (rather than rented) the new owner does not have the rights to produce ‘identical or similar products If the prociuct is untler patent protection, 6.5 Durable-Goods Monopolies 83 A renting monopoly [Assume that each period the monopoly sents a durable produet for one period only. For example, 2 cominon practice of firms in several indus: hes, in particular in the car industry, is to lease a car for a given time period cather than sell the car. Although there could be several expia- Petions (baKes, ete.) why such a trade benefits fraas and consumers, in this subsection we prove thet leasing would yield higher profs than felling. Suppose that in each of the two periods the monopoly fares the demand Grawa in Figure 5.4 (left) ‘Assuming zero production cost, we recall from section $.1 that the monopoly would eeat an amouunt determined by the condition M (Qs) = = 0 = MC(Qq), implying that Q = 50 and pf = 50, and 2. Hence, the life-time sum of profits of the renting, 00. monopoly is given by x® A seller monopoly ‘A seller monopoly knows that thee consumers who purchase the durable good in t= 1 will not repurchase in period t= 2. That is, in t= 2 the Fonopoly will face a demand for its product that is lower than the period 1 demand by exactly the amount it sold in ¢ = 1, ‘Therefore, in Period 2 the monopoly will have to sell at a lower price resulting from a lover demand, caused by its own earlier sales. Formally, we define this two-period game as follows: The payoil to he monopoly is the total revenue generated by period 1 and period 2 Gales, The strategies of the seller aro the prices set in period L, pry fand the price set in period 2 as a function of the amount purchased fn period 1. pa(di)- ‘The stzatogies of the buyers are to buy ct not to buy as a function of fist period price, and to buy or not to buy as a function of second period price. We look for a SPE for this simple game {eae Defaition 2.9 on page 26). The methodology for solving this finite horizon game isto solve it backwards—to determine how the monopolist woud behave in period 2 for ench possible set of buyers remaining thea. ‘The second period Figure 54 (cight) shows the (cesidual) demand facing the mozopoly in petiod 2 after it has Sold; units in period 1, givon by gy = 100d ~ 7a be ps = 100 — 4, ~ qx Since production was assumed to be costes, in the second-peciod the monopoly sets ‘Mf i%(¢z) = 100 ~ dh ~ 2a implying thet 2 = 50 ~ qi/2. Hence, the second period price and profit levels are given by pa = 100 ~ Ge ~ (50 ~ 4/2) = 80 — 7/2, and ma = pats = (50 = 0/2" .i ‘The Monopoly The fst period Suppose thatthe monopolist sells inthe Birt pevid vo g buyers with the highest eservtion pce. Then, the marginal buger, with resereaion price 100 dy, wil be indir betwoen purchasing ia the Rest prod (gaining utility of 2(100 ~ G1) ~ py) and buying in the second period (euinng uty of (00 ~ 1) p= (100-45) ~ (80 ~Gy/2). Thaw 2(400 ~ gy} = pi = (100 ~ Gu) ~ (60 44/2). (5.9) Solving (5.9) for py yielde (510) Let us mote that equation (6.10) can also be derived by observing that. {he first-period price should Include the secondperiod price in addition lo pricing the firstpetiod services because buying in the first period yields services for the two periods, hence, the praduet ean be resold in ‘the second period for a price of pz. ‘Thezefore, Pe = 100 ~ a +p = 300g +50— B ‘whic is identical to (5.10) In a SPE the seliug monopoly chooses a Brst-period output level i ‘that solves mater) = (150 - aa +( yielding a first-order condition given by ert) _ joy ye 100-1 A = 09g, — = 100 - Denoting the solutin values by e superscript 5, we have that of = 5040/2 = 50, pf = 50~ 40/2— 30 and pf = 100-40 Hence, ns = 4,500 < 5,000 = Hal + aha5 Therefore, Proposition §.8 A monopoly aeling © duralle goods earns a lower profit than a renting monopoly. ‘The intuition behind Proposition 5.8 is that rational consumers are able to calculate that a selling-durable good monopoly would lower future 5.5 Durable-Goods Monopolies 85 Prices due to future fll im tie demand resulting frema having some cot sumers purchasing the durable product in eusliee perils, ‘This calen tation reduces the willingness of consumers w par high prices in the first period the monopoly offers the product for ssle. In other words since the monopoly cannot commit itself not to caduce fatute prices, tins ‘monopoly is induced to lower its frst-period price. An argusnent such as Proposition 5.3 led some 2couomists to claien that monopolies have the incentives to"produce less than an optimal level of durability (e.., light bulbs that buma very fast), We disoes the (ia)validity of this argument in section 12.3. 5.5.2 Durable-good monopoly facing a discrete demand ‘The analysis of subsection 5.5.1 has confined itself to « demand eueve ith a continuum of monatomic buyers. Following Hagnoli, Salant, and Swierabinski (1989), we now provide an example vhich demoustrates that Coase's Conjecture is false whon the number of sousuumers is finite Let us consider an economy with two consumers living only for two Periods. Both consumers desire car services for the lwo peclody of thew lives, however, the consumers difer in their willingness to pay for car Services, The maadium amount a consumer denoted by Hs willing to pay for one period of car service is V, and the maximum amoust 2 ennsumer denoted by LZ is willing to pay for one pariod of car service 's VE. We assume that the consumers’ willingness te pay per period of car service are substantially different. ASSUMPTION 5.1 Type If consumers are willing to pay more than tuize as much for a period of car service ax type [ consumers. Formally, VE > ove sy, Figure 6.5 (let) illustrates the aggrogate inverse demand function for ‘ove period of survice facing the monopoly each period, Because the product is durable, consuaners buy it oace in thelr fe sither at ¢ = Lor ¢ = 2 The utility functions for consumers type i= HyL that yield the demand structure illustrated in Figure 5.5 ooo sven by V'—p) if he buys a car in period 2 Gx) 2V4— py if be buys ear in period 1 o if he does not buy a car in any period. ‘Thus, if consumer i, HY, tuys a car in the Brot geriod; he gains a Deuefé of 21" since the car provides services for two peiods, and he pays Whatever the monopoly charges in ¢ In contrast, ifthe conser86 ‘The Monopoly ver n ” 1 a i Figure 6.5: Durable-good monopoly: the case of discrete demand waits and purchases the ear in # = 2, he gains only one period of utility fof V¥ minus the price charged in period 2. ‘On the production side, we assume that there is only one frm pro ducing eats, at zero cost. Like the consumers, the monopoly fiem lives for two periods and moximizes the sum of profits from the sales during the two periods. We denote by q the amount produced and sold by the ‘monopoly, and by pe the periad 1 price of @ car set hy the anonopoly in period f, £= 1,2 "The monopely chooses py and p2 to maximize the sum of revenue from two periods worth of sakes given by x = pigs +t pada. Note that we have implicitly assumed that buyers and the monopoly do not discount fnture utility and profit, since assuming otherwise would not have a ‘qualitative effect on the resus, A renting monopoly ‘Suppose now that the monopoly firm does not sell cars, but Instead rents cart for one period only, Thes, each consumer who rents @ cat in ¢ = 1 hha to retuen the car at the end of the fast period and rent it again in the second petiod. We denote hy pf the rental pre for one period of renting in period 4 Since car rentals last for ane period only it is sufcient to calculate the price for ench period sepavately, Since the renting fin is 8 monopoly, it has two options: (1) setting pf = V#, which by (5.12) induces only ‘consumer H to rent a car esch period, while consumer J will nt rent; (2) setting pft = V", which induces both consumers to rent a car each period. In the first ease, the two-period profit is x” = 2V", and in 5.5 Durable-Goods Monopolies sr the second case, x Hence, AV, However, by Assumption 6.1, V > 2V¢, Proposition 5.4 A renting monopoly would rent cars only to the high yaluation consumer by setting a rental price equal to pf = V¥, t= 1,2; anu it will ear a two-perind profit of x = 2 A seller monopoly Nom, sppote thatthe monopoly sls the cars to coniumars. We denote the sling prices by pf, ¢ = 1,2. By Dofsition 51, the period 1 selling price, pf, means thatthe consumer pays fortwo periods of using the eat (compared with the renting price p{f that entitles the consus the ent for period 1 ony) The second period ‘The effect of selling in the frst period on the second period domand is ‘strated in Figure 5.5 (cight). If consumer If parchases in period 1, only consumer L demands a car in the second period. If consumer Hf does not purchase in the frst period, then the socond poriod domand is the given rental demand curve (Figure 5.5 (left)). The lower part of Figure 5.6 illustrates the subgames associated with consumer H's decision whether to purchase in the first perio. wave Ut =pu% = av vty ‘t= 1: Consumers decide whether to purchase] Consumer H buys _ Hf dos not buy t gave’ ppavies, Saw" svt faa ut aut=o ut aot Figure 5.6: Two-period game of a durable-good monopoly facing discrete domand, Figure 6.6 ilustrates that when consumer H buys in the fist period, the monopoly will maximize second period profit by setting pf = V"88 ‘The Monopoly and will earn a second period profit of x = V¥ (the monopoly will extract all surplus from convener £.). H consumer H does not buy in period 1, then in the secon period ‘he snonopoly faces the entire demand, hence, by Assumption 5.1, the monopoly: charges pf = V#" (culing only to consumer H) yielding a secand paciod profit of xz = Vit The first persod In the first period, the monopoly sets pf, and consumers deckde whether 2V4, Hence, if the seller seta p$ = 2 both consumers would purchase Clearly, the monopoly will not set pf > 2V* because this price now check whether pf = 2V/ is the profit maximizing first period price ‘that consumer H earns a utility of zero (U = 0) whether or not he buys response for consumer H to the first period icy pf = Ve ‘Thon ine Proposition 5.5 A durable.good selling monopoly facing a discrete de mara will, 1. charge a first period selling price that ie equal to the sum of the sper period rental prices, pf = 2plts 2 carn a higher profit than the renting monopoly, that is, w= BVH EVE > VM a ‘Tins, in the case of discrete demand, a selling monopoly can extract, f higher surplus from consumers than the renting monopoly. Cosse conjectured that the ability of a durable-good monopoly to extract coi sumer surplus is reduced when the monopoly is forend to sell rather nnn rent. Here, we demonstrated the opposite ease, where selling en- ables the monopoly t> price discriminate among different consumers by 5.8 Appendix: The Logal Approach 59 setting prices which would induce different consumers to purchase at diffrent time periods. 5.6 Appendix: The Legal Approach to Monopoly and Price Discrimination 5.6.1 Antitrust law and the monopoly Suction 2 of the Sherman Act of 1880 states that Every person who shall monopolize, or attempt to monapo- lize, oF combine and conspire with any other person oF pe sons, to monopolize any part of the trade or commerce among, several States, or with foreign nations, shall be ceomed guilty of & felony. At frst glancs, it seoms that section > makes it clear “hat 8 monopoly i ‘egal, but a cioser look reveals that the act does not provide the cours ‘ith any guidelines that define what degree of market power or merket concentration constitutes monopoly. Therefore, in practic, comnts tend to focus on abuses of monopoly power in a toucedtrated marist and on the intent of the monopoly to kivp its postion; monopoly stains alone isnot legal. Anticorpetitive activities such as predatory priciag_ have to be ex tablished to turn s monopoly into an illegal practice, "To establish tr legal activities, the court fist defines the proiuct ad the googrepiic market. Sacond, the court considers the market share of the accecol firm. ‘Third, the court considers the ease of entry, ealablity of soc ndand and new substitutes, and whether the accuse has the abil to tae prices. Defining the product & basically dectiag which por, ucts soul be considered as close substitutes. Defining the geogrophic ‘market should consider the magnitude of transportation costs, wien ig ‘aay cases are insignificant, thereby lending the cout to define the tire nation as the geographic market. When these tts are unclear, the court resorts toa hypothetical question: [na particular goographic tae kt, an the accused firm raise the pric without attracting competition? UT the answer is positive, then the matket is well defined, During the ears courts have added a refusal to den}, when a marufacturer teases to sll to dealers for the purpose of establishing a monopoly power oe all distnbution channels, as an abuse of monopoly poner, 5.6.2 Antitrust law and cartels Cartels may involve price fixi 1g, Output controls, bid rigging, allocation of consumers, allocation of 5 ales by product or vervitory, establishanent80 ‘The Monopoly ‘of trade practices, or common enles agencies (Welss 1987). Cartels have txisted ns guilds in the Europe of the Middle Ages, and were comtoon in most European countries tiroughout the nineteen century and the fit third of the present century, The Sherman Act of 1890 minde eartel illegal. Exceptions were made during the Great Depression, and for some special quasi-public industries such as agriculeure, coal, civil aviation, and oil refining Section 1 of tke Sherman Act 1890 states that Every contract, combination in the form of a trast or other. ‘wise, or conspiracy, in restraint of trade or commerce ainoog, ‘the rovera! Sater, of with foreign nations, is deelazed to be ‘Moga. CGiearly, the moet severe and most common cartel contract i a price. fixing contract, Firms that are found guilty of price fxing ace subject to treble-damage penalties. Recently, several authors raised the question of whether tzeble-damage penalty woud result in market price reduction o ‘market-price increase. Salant (1987) shoved that tzebledamage penalty ‘can increase the market price above the price that would be charged by a cartel without che enforcement of this antitrust law. Barlier court eases interpreted section 1 to mean that every contract constituted a restraint of trade, thereby leading courts to rule on a per se basis, defined in subsection 1.2.2 on page 6 of this book. That is, every price fixing wae illegal. In sore Inter eases, courts considered some price fixing arrangements under tb» rule of reason. However, courts began Jearning that sny judgment unver the rule of reason involves tremendous drainistrative costs, since it ie not clear what a reasonable price is, and its bard to measure marginal-cost functions to determine whether the price is fixed with a high markup. Tb was also clear that prices should ‘often fiiclunte with cost variations, something that may not eccur in the presence of price fixing. Hence, courte began judging price fixing under the pee se rule. The losic was that if price-fixing agroements do rot bnve ai effect on prices then these agrecinents would not be formed ‘Thus, price-fixing agreements should be illegal per se. The per se rule bras also applied to other forms of contracts, such as market allocations, Finally, one advantage of the per e rule is that it warns the firms in advance about the consequences (generally, treble damages) associated with price-fixing agreements, sherens the rile of reason may leave sore doubts whether, with a good defense, a eartel ean survive section 1 in a laws, ‘To summarize, we can say that the major effect of section } of the Sherman Act is rather noticeable: The act indeod eliminated major cartels from American markets, Meet noticeable cartels nowndays for 5.6 Appendix: The Legal Approach on exemple, OPEC and IATA, are international, and cannot be challenged for rather visible price-ixing agreements, 5.6.3 Antitrust law and price discrimination Section 2 of the Clayton Act of 1914 amended by the Rebinson-Patman Act of 1936, states that It sholl be unlawfal for any person engaged in commerce, in the course of such commerce, either directly or indivectly, 1» discriminate in price between different purchasers of cm- rmodities of ke gracie and quality... where the effect of such discrimination may be substantially to lessen competition or tend to create a monopoly in any line of commerce, o to injure, destroy, or prevent competition with any person who cither grants or knowingly receives the benefit of such cis ‘rmination, or with the consumers of either of them: Pro- vided, That nothing herein contained shall prevent differs tala who make only due allowance for differences in the cost of manufacture, sale, or delivery. ‘Thus, section 2 explicitly states that price discrimination should aos be considered illegal (a) unless price discrimination substantially decreases ‘competition, and (b) if price differences result from dilferences in pro- ‘duction of delivery costs, Thus, the coupoas appearing in the Sunday newspapers offering a price reduction upon the presentation cf a pice ‘of paper Is a good example of price discrimination between people with ‘high value on time and a low value on timo; however, there is nothing, ‘legal In using coupons for providing discounts, ‘The Robinson-Patoaan Act of 1936 came during the Great Depres- sion and was intended to strike against large, chain grocery stores that ‘engaged in local price cutting to deter competition. Note that at that time the legislators were not concerned whether prie discrimination and price cutting are efficient. In fect, Varian (1989) shows conditions un- dor which the act of price discrimination is welfare improving, compared ‘with uniform price mechanism. Also, note that the GATT (General Agreement on Tiff and Trade) enacted a rule similar to the one ex ‘acted in Robinson-Patman stating that damping (selling below cost in 1 foreign country) is illegal. However, it has never been theoretically ‘established that dumping reeiuces welfare, and it is passible to demon- strate thot dumping ean actually iinprove social welface. AAltogetber, its not clear whether price discrimination has arything to do with anticompetitive behavior, and in fact, price discrimination can actually he procompetitive. Bork (1978) warns against possible92 ‘The Monopoly damages inflicted by this act by conjecturing that there may be hundceds (of thousauds of pricing decisions every year thet are altered through fear of the Robinson-Patman Act, meaning that lnmdveds of thousand of quantity discounts and promotional discornte are foregone at the expense of having consumers paying higber prices. However, during the years following thes act, the FTC rarely enforced this law, thereby raking price differences more observable. 5.7 Exercises 1. Consider a monopoly selling at a single market where the demand is tiven by Q(p) = ap", € > 1. Suppose that the oot fonction of this monopoly is givea by TC(Q) = €Q,e > 0 (=) Csloulate the demand elasticity, and, using Proposition 3.3, waite dom the marxinal-reveae function as « function of price. (b) Using your shove calculation, find the price charged ty the monopoly as function of ‘Wet heppess to the monopoly's price when ¢ increase. Interpret your rest (a) What happens to the mouopolys price as «+ —17 Explain! (¢) Calculate te total-revenue function TR(Q) and the marginsk revenue Function MR(Q), ‘What isthe monopoly's profit-maximiing output? a (Consider the market for the G-Jeant (the fates! foshion among people in Uheic lave thirties). G-eans are eold by s single re that cerioe ‘ve patent for the design. On the demand aie, there aren!” > 0 hale come consumers who are willing to pay a masinwin amount of H!" for ‘pit of GaJeacs, and n! > O low-income consumers who are willing to pay a maximune amount of V" for a pair of G-Jeans. Assume that V'> V4 > 0, auc that cach consumer buys only one pale of jeans, Suppose thet the ScJaans monopoly cannot price discriminate and Uherofore constrainod to set.» uniforte market prion (8) Drow the macket aggregate-demand curve facing the monopoly: (8) Find the profi-maximiing price set by G-Jeans, considering all possible parameter values of" .n®,V", and V# ‘Suppose thi a monopoly can price discriminate bobween two markets market, where the demand eurve fe given by qi = 2~ pa, ord micket 22 ‘where the demand curve is givon by gy = 4px. Suppese that ance the producti sold it cannot be resold in the other market. That I aime ‘hot arbitrage is smooasible, say, du to strict cuistorn inspections om the border between the two markets, Assume that the monopoly produces edi unit at a cost ote et 5.7 Exercises 93 (@) Calculate the proft-maximizing output love. that the monopoly sellin eah market, Caleulaea the price chaiged ia each marke. (©) Calculate the mouopalys profit level (© Suppose that markets 1 and 2 are aow opeo, and al! consumers face fre to trade and to transfer the good cxstesly between the ‘markets. Thus, the monopoly caa no loager price discriminate and bas to charge a uniform price denoted by p, p= p1— pa, Pind the profitsmaximaing value ofp.” 4A disctiminating wonopoly sells in two markets. Assume that ao a2 bieaye is passible. The demand curve in market is given by 100 ~ y,/2. The demand curve in market 2 is given by pr u 10D 9 We denote the monopoly’s ageregate production by Q wbere Q = noi ‘The moaopoiy’s cost function depends on total profuction and i give by TO(Q) =Q*. Answer the following questions (2) Formulate the monopoly’ profit function as» function of gr and @ (b) Caleuiate eke monopoly’s profit-manizaizing quantity oid in mar- et Land marke: 2, (©) Calculate the prof level ofthe diseritainating monopoly (4) Suppose now that a new management sssunee control ofthis rm ‘The young CEO decides to dacompave the motopely plant into two plants, where plant 1 slls in market 1 only and plant 2 sells ia ‘marlat 2 only. Caleuste the profit-masimizing output level sold by each plat le) Calculate che sum of profits ofthe two plants (8) Conclude whether this plant decomposition ivereases or decreases profit. Explain your assuer by investigating whother the above technology exhibits increasing or dacreasing seturas to scale (Com. sult Desnition 2.2 on page 45) 5, The demand elasticity in market 1 is mensusid to be «1 = —2. The demand elasticity in market 2 is measured to be €2 = ~4. Suppase that ‘8 monopoly that can price discriminate between tie markets sets the rice pia market 1, and pe ia market 2. Prove whether the following Statemeat i right or wrong: “The price in masket 1, pa, will be 150% Ligher than the price in mazket 2, pa.” 5. In s tworperiod lived economy, ane comiumer wishss to buy a TV oot In period 1. The consumer lives for two periods, and is willing to pay ‘8 maximum price of $100 per period of TV usage. [a period 2. exo ‘consumers (who live in period ? only) are born. ath of the newly bora ‘coasumers is willing to pay a maximum of fifty dollars for sing & TV ia period 2. Suppose that in this market ther is only oa fit producing ‘TY sets, that ‘TV wots are durable, and thet production is cals.4 ‘The Monopoly (a) Coaloulate the prices the monopoly charges for TV sets in peviods 1 nd 2 (b) Answer the previous question assuming that in the frst pesiod, ‘commer who lives te periods is wig Lo pay Wo more than thoonty dolla pee peiod for TV usage, 1. A monopely is facing » cownward sloping linear dem curve given by pa ~ @. The monopoly’s unit production cost is given by ¢ > 0 Now, suppose that the government imposes specific tax of f dollars per unit on each unit af exept sol to consumers (a) Show that this tax position would raise the pre paid by com fmers by less than £ Ttints One nay ta fd the monopo}y's prof tasiwiing outpot level to rose the equation AFR(Q) = e+ 6 fn then to solve fr conmumer and producer price (b) Would your answer change ifthe mathe ineersedemand) curve has a constants elasticity and is given by p= Q7?? 5.8 References ‘ogni, M.,8.Salont, and J.Sviersbineli. 1680. “Durable-Goods Monopoly ‘with Discrete Demand.® Journal of Political Beonomy 97: 1459-1478, Bork, R. 1978. The Antitrust Parador. New Yaris Basic Books. Bulow, 11082. “Durable Goris Mangpolists” Journel of Pola! Beonomy Yb: 314-92, Bulow, J. 1986, "An Economic Theory of Planned Obsolescence” Quortariy “Fournal of Beonemses 51: 128-748 Coase, R. 1972. “Durable Goods Monopaists” Journal of Law ond Beo Tomice 1: 143-180, Getthorn, E. 1986. Anditrust Law and Bonomi ina Nutshell. St. Pail ‘Minn West Publishing Gul, F..M. Somensehein, and R. Wilson. 1986, "Poundations of Dynazaie Monopoly and the Coase Conjecture.” Journal of Economic Theory 89 156-190. Kmeger, A. 1874. “The Poitical Beonomy of the Rent-Sooking Society" “American Econotic Resiew 64: 281-308, Posner, R. 1975, "The Socin Coots of Monopoly and Regulation” Journal ‘of Political Beonormy 83: 807-827 Selonk, 5.1987. "Treble Damage Awards in Private Lawsuits for Price Fix- ing” Journal of Poitial Beonomy 95; 1926-1386. Stokey, N. 198). “Rational Expectations and Durable Goods Pricin “Jotrnal of Beomortes 1212-128. Dullock, G. 1967, “The Weare Costs of Tals, Monopolies, and “Theft.” Western Econgmite Fowrnal 5: 224-232, Bett 5.8 References 95 ‘Varian, H. 1989, “Peice Discrimination.” In Handbook of Industria Orga tion, edited by R. Schmalensea, and R. Willig, Amsterdam: Nort: Holland, Weiss, b. 1987, “Cartel* In The New Palgrave Dictionary of Eemomics, Talited by 3. Batwell, M. Milgate, and P. Newman. New Yera: ‘The Stockton ProseChapter 6 Markets for Homogeneous Products Only theory can separate the competitive from the anticom- petitive. “Robert Bork, The Antitrust Paradoz In this chapter we analyze the behavior of firmas and consumer welfare un der several oligopolistic mackes structures. The aain sssunplion in this ‘chapter is that the products are homogeneous, meaning that consumers cannot differentiate among brands or distinguish among the producers when purchasing a specific product, More precisely, sonsumers cannot (or just do not bother) to read the label with the producer's namie on the product they buy. For exainple, nou-brand-naie products sold it most ssupermarkets—bulk fruit, vegetables, containers of grain—are generally purchased without having consumers learning the producer's naiue, In what follows, we assume that consumers are always price tak- ers (henceforth, competitive) and have a well-defined ageregate-demaad function. However, firms behave according to the assumed market struc- tures analyzed below. Our oligopoly analysis starts with section 6.1 (Cournot), which as- sumes that firms set their output levels simultaneously, believing that the output levels of their rival rms remain unchanged. Historically, as wwe discuss below, Cournot was the first to provide tais modern teat- ‘ment of oligopoly equilibrium. Section 6.2 (Sequential Moves) modilies ‘the static Cournot setup, by assuming that rms move in sequence, and analyzes whether a firm benefits by setting its outpu: level before any ther one does. Pollowing Bertrand’s eriticism of the use of quantity produced as the actions chosen by firms, section 6.3 (Bertrand) analyzes98 Markets for Homogeneous Products fa market structure where firms set their prices by assuming that the prices of their rival firme remain tnchanged, We then discuss how the extreme result of price garmes lending to competitive pricee obtained un der the Bertrand competit ruitigated by introducing eapecity constraints Section 6.4 (Cournat Versus Bertrand) analyzes the rela tionship betwesn the Cournoy and the Bertrand markot siructures. Sec tion 6.5 (SelfEnforcing Collusion) analyzes the conditions under which Firms can maintain higher prioes and lower output Jevols compared with the Cournot levels, assuming that the firms interact infinitely many Limes. Section 6.6 (International Trade) analyzes international merkets in homogenous products, G.1 Cournot Market Structure Noncooperative cligapoly theory started with Antoine Augustin Conmnot's hook, Researches into the Mathematical Principles of the Theory of Wesith, published in France in 1838. In that book, Cournot proposed fn oligopoly-analysis method that we today view as identical to finding ‘a Nach equilidrium in # game where firms use their production levels at strategies Conmat eared hs doctorate in science in 1821, with a main thesis in mechanics aud astrcoomy. Cournot’s writings extended beyond ‘economies to mathematics and philosophy of science and philosophy of history (spe Shurbik 1987). ‘Cournot wna central to the founding of mederm mathematical eco nomics, For the case of monepoly, the familiar condition where marginal ue equals marginal cost come direetly fom Cournot's work (Siaubik 1987}. In chapter 7 of hie book, Cournot emplaye the inverse-demand function to construct « eysiem’ of firms’ marginal-revenue functions, which could be then solved for what we will call the Cournot ouput levels. Then, he introduced firms’ cost fanctione and the system of first ‘oxder conditions to be solved. Cournot did not consider the possibility that Gres with sufficiently high cost may not be product, in this eq Sibsiomo, In what follows, we develop the Cournot oligopoly model where fitms soll identical products. In this model, firms are not price takers. Instead, ccoch firm is fully aware that changing its output level will affect the racket price 6.1.1 Two-seller game Jt us consider a two-fim industry summarized by the ccst function of cach frm # (producing g units) given by TCs(a) = 0, 1,2, where ener 20, (6.1) 6.1 Cournot Market Structure 99 and the market-demand function given by P(Q)= 4-22, 0,6>0,0>6, whereQ=atm (62) In contrast to chapter 4, where we solve for a competitive equilibrium foe this industey, here we solve for & Courzot oligopoly equllbrium. We first have to define a two-firm game that corresponds to a definition of a game given in Definition 2. Lat each firm's action be defined as Choosing ito production level, and astume that both fms choose shir fictions simultaceously. "Thus, ench frm i chooses q € A, = (0,00), TOL 2 Also, lt the payoff function of each fw {be its profit function Getined by mas 2) = (ai Haas = TCs). Now, the nme is properly defined since the players, ther sction aes, and their payof fanctions age ‘explicitly defined. All that i eft todo now is to define the equlibsiuma concent. Dexiwirion 6.1 The triplet (p%qf,9§) i o Cournot-Nash equilib sum if Gy wot sec D)4t = Pen-ba Pah 2 4 1. (a) given qa = Gia solves max, (48) apie + af)a ~TC(n) = fe = Kaelin ~ om (0) given qs = afi solves maxy, *{ fs) eyes + alae TCa(ae) = a — Ha + eee ~ ae Mata), Pag 20.4 2 pF ‘That is, according to Definition 6.1, a Coumnot equilibrium is a list of output levels produced by each frm and the resulting macket price 90 that no fein could increase ita profit by changing its output level, elven that other firms produced the Cournot output levels, ‘Thus, Cournot quilibrium output levels constitute a Nash equilibrium in s game where firms choose output levels. ‘Now that the equilibrium concept is well defined, w: are left to calculate the Cournot equilibrium for this industry. Firm 1's profit- maximization problem yields the first-order condition givea by 1(a3, 48) a, 0 bar ~ 0 which yields the famiiar proft-maximizing condition in which each firm (Gem Lin this equation) sets its marginal revenue (MR(q) = a2 ~ fgg) equal to marginal cost (c,). Tae second-order conction guaran- tesing a global maximum is satisfied siace $itjy = —2) < 0 for every 1g; aad gz. Solving for q a8 a fimction of gy yields the best-response100, Markets for Homogeneous Products Junetion (also commonly known as reaction function) of firm 1, which ‘we denote by Ryle). Hence, i a= Ralee) = (63) Similarly, we can guess that fim 2's best-response function is given by tom Relay) = 952 5 (4) ‘Phe best-response functions of the two firme are draven in Figure 6.1 in the (nga) space, a Reals) : ey we ete ” a ae ona Figure 6.1: Cournot best-response functions (Uhe ease for ¢2 > 1) The two best eaporse functions ae downward slong implying tha cach fifth as utp le nto, te el owe utp level The tution i that sone fm rae te outpot Yh price would drop, and hence in order to maintain & high price the other frm wonld fd proftaeta decease is oatpet evel A paps more intiive oxlanatton fry afr bet ceqpone fenton ie downmard Sloping i that an incese in thas outpat shits te residual demand facing Brn roar Hence, when afm acs lower demand it would produce «stale amount ow, the Counnct euiibriom output levels canbe calesbated by solving he ta bettesponge functions (6) ad (rt), which orespond tothe inervetion of the cnrwes seated in Figure 6. Thy fa ten : STE ant og 2ep ter (6s) 6.1 Cournot Market Structure ion Hence, the agsregate industry-output level is Q* = qf ~ 0§ = faad she Cournot equilibrium price is atate, pean tg = tS (68) Tis easy to coufiem from (6.5) that the output of the high-cost fem is owes thaa the output level of the low-cost firm, That i, ey 2 cy implies that gy > a: “Altogether, the Cournot profit (payoff) level of frm i, a# « fanetion of the unit costs for firms ¢ and j, i #7, is given by = (oot) = (E348 - (67) We conclude this section with some comparative staic analysis Sup- pase than fm [ivenes 2 new production proces that reduces unit Position cost Rome, tof, where &) < er. Tho type of R&D lead Presto coo reduction ealed “process innovation,” to which we wil sae Chapter 3. Beation (85) tales that terases wie 0 {oeaers. Th ig also shown in Figure 6-1, where adzrease in iN ‘Ria) tothe righ, thereby mvwosing the egos gf while deseas se to (80) mpi that desreaze in o (9 9) woul dorwase BE dita price and (67) ples Chat a cereae i ex would inerese the prot of fem 1 while lowering the profit of firm 2 6.1.2 N-seller game Suppose now the industry consists of NV firms, N 2 1. We analyze two types of such industries: (a) NV identical firms, all having the same cost function, or (b) heterogenous firms, where some fics bave cost functions diferent from others, Since solving tbe general case of fimns vith different cost functions would require solving WV Brst-order coud tions (intersecting NV bast-response functions), we fist solve the snode! by assuming that all fms have identical technologies. That is, ci = ¢ for every 1 = 1,2,...,.¥. In the appendix (section 67) we introduce a procedure thet tales colving the heterogeneous-rms case easy. Since all firms have the same cost structure, the first step would be to pick up one firm and caleulate its output level as a function of the output levels of all other Brms, In other words, we would like 10 ‘calculate the best-response function of a representat.ve firm. With ao toss of generality, we derive the best-respoase functicn of fmt 1. Thus,102 Markets for Homogeneous Products firm 1 chooses a1 to ingens = piQ)au ou | The first-order condition is given by an 7 ye nhac ence, the best-response function of frm 1 as a function of the output Tews of firms gay 435+ 4a i given by Ralanvaa sn) = Le (6.8) In the general ense, where firme Ino liflorent cost functions, we ‘would lnve to derive the best-rsponse fiction for exch of the A firms However, since all firme are icentesl, we can guess that. in a Cournot ‘euilibsium, the firms wouid produce the same output level: we guess (and later verify) that gf = 0 =... = ofp. Thus, we denote te com non output level by q, where = 9; for every #. Note that a common rnistake among students isto sibstithte q for q before the best-response functions are derived. ‘This poceduze e obviously leading to the wrong solution, since it implis that esch frm “controls the output level af all fens. ‘Therefore, here we substitute the common g only into the frend derived becttesponse functions. The use of symmetry here is purely technical and i dome to facilitate solving V equations with NT unknowns, From (6.8), we have it that ¢= 43° ~ (N ~ 1)q- Hence, eopfais o™= (S2) (gH). 9 ‘Tho equilibrium price and the profit level of each firm are given by atNe ee re? and a f= 0 HO =H)? - (6.10) Varying the mune of frm We now ask how would the Cournot price, quantity produced, and profit levels change when we change the number of firms in the industry? Firs, note that substitinting N = 1 mto (6.9) and (6.10) yields the monopoly solution deserihed in section 5.1. Second, substituting IV = 2 yields the duopoly solution described in (6.5), (6.6), and (6.7). 6.1 Cournot Market Structure 103. Now, we let the aumber of firms grow sth no bounds, (NV — oo). “Them, we have it that gt =6 ot ter= (255) (ya) = C52) Ne Wei Hence, the Cournot equilibrium price approaches the competitive price thst equals the unit production cost of a firm (see Proposition 4.1) "These results often cstse some confusion among students, leading therm to believe that competitive behavior occurs only when thers are many (or infinitely many) firms. However, as we pointed out in chapter 4, ‘we can assume a competitive market structure for any given number of rms, and even soive for a competitive equilibrium for the case where IN = 1. What equations (6.11) and (6.12) say i that the Cournot market structure yields approximately the same price'and industry output ss the competitive market structure when the number of firms s large. Aine = a eT (6.12) 6.1.3 Cournot equilibrium and welfare Since our analysis state with given demand functions (rather than the onvumers silty factions), we cannot measute the social welfare by Cleulating consumor’ equlibrium-wiity levels. Instead, ve approxi tate social welae by adding consumer siplus and Bens pros (ste ubeection 3.2.3 on poge 2 fora justification of this procedure of wet fare approximation), Note that profit shouldbe pact of ta economy's smlfce becuse the Rens are owned by the consumers, who collec the Pros via firme’ distributions of dividends Subrtiuting the Couraot equilibrium price (6.10) iat (3.) on page 52 we tain the consumers suis a8 Rnetioa ofthe nur of is, ie-28 Clearly, 25 > 0, meanitg that con ues! surplus rises with ta entry of more fire, due tothe rection in price andthe Increase in the quantity consumed. "We define socal welfaze a the sur of consumers surphs ps the industey aggregate profit (se section 4.3 on page 63 fora definition). Thus, i we reall (6.10), We(N) = CSW) + NaeN) (0.13)104 Markets for Homogeneous Products (3%) Th) = SS, Also, note that 2482 > 0, epee, although the industry profit declines ‘with am increase i the mumber of firms, the increase in consumers’ surplus dominates the reduction in the industry profi. "Thus, in this economy, free entry is welfare improving! 6.2 Sequential Moves Jn the previous sectien, we analyzed industries where firme strategically choose their output levels, All those games were static in the sense that players simultaneously choose their quantity produced. In this section, ‘we assume that the firms move in sequence. For example, in a two Firm, sequentiakinoves game, firm 1 will choose its output level before firm 2 does. Then, frm 2, after observing the output level chosen by firm 1, will choose its output level, and only then will output be sold and profits collected by the two firms. This type of market structure is often referred to as Leader-Follower om the basis of von Stackelborg's ‘work (2934) (s00 Konow 1904 for von Stackelberg’s biogrepby). ‘This type of behavior defies an extensive form game studied in section 2.2 In this seetion we do not raise the important question of what deter- tines the order of meves, that is, why one firm gets to choose its output level before another. We return to this question in chapter 8, where ‘we distinguish among established firms (called inevmbent firme) and potential entrants. Fere, we assume that the order of moves is given, in) we develop the tools for solving an industry equiibrivm under a predetermined order of moves We aunlyze a two-stage game, where firm 1 (the leader) chooses the ‘quantity produced in the first stage. ‘The quantity chosen in the frst stage is irreversible and eannot be adjusted in the second stage, In the second stage, only finm 2 (the follower) chooses how much to produce afver observing the output level chosen by firm 1 in the fet stage, Here, te game ends after the second stage, and each firm collects its profit (Our main questions ave (a) Is there any advantage for moving in the fst ‘stage rather than the second! and (b) Has would the equilibriuin merket price and production levels compare to the static Cournot equilibrium price and output levels? Fallowing Definitisn 2.9 on page 26, this game has # continuum of ssubgames indexes! by the output level chosen by firm 1 in the first stage. A finite-horizon dynamic game is generally solved backwards. We look for a eubgame perfect equilibriven (Definition 2.10 on page 27) for this gaimte, Hence, we frst analyze the players’ (firm 2 in our ense) action in VOCUS C OLS ESUea ' ’ 6.2 Sequential Moves 105 the last period, assuming that the actions played in previous period are given, Then, we go one period beckwards, and analyze firm I's action, given the strategy (soe Definition 2.8 on page 4) of haw Bran 2 chooses its Dutpat level based on the first-period action. ‘To simplify the exposition, let all firms have identical unit cost, 6 ‘The secondperiod subgames tn the second period only frm 2 moves and chvoses gy to maximize ts prot, taking frm 1'S quantity produced, qu, as given. As you probably noticed, we have already solved this problem before, since the second petiod problem of frm 2 is identical to the problem firm 2 solves in Cournot markst structure. ‘This masximization results in the best- response function of fitm 2 given ia (6.4). Hence, Ro(e1) = 8395 ~ as Note thatthe funtion Ma(q) eonsticutes ra 2s strategy for tis gatne, since it specifies its ation for overy posible action chen by fra 1. The frst-perod game In period 1, firm 1 calculates Ra(ox) in uke same way as firme 2. Thos, firm 1 is able to calculate how firm 2 wi best reply to its choice of output level. Knowing that, firm 1 chooses 42 to pert = ola + alan om = [a4 ( + Jaen (6.4) We leave it to the reader to derive the Brst- and second order coultions. ‘Thus, the quantity produced by the leader is Bes of : an Si Hat (6.15) Hace, under the sequential-moves masket structure, th leader produces «higher level of output than the Cournot market strusture. Substitut ‘ng (6.15) into Ra(qy) yields the followers’ equilibriumroutput level Seu : Bape (6.16) ‘implying thot the folower's output level fills compared with the Cournot output feel. Ths, the leader's gaia in output expansion comes partly from the raduction inthe fllower’s output fev The aquilibrurn peice and oggrogate output levels ae given by arte at Mare), ao e and Qt106 Markets for Homogeneous Products Proposition 6.1 4 sequentiokmones quentity game yields a higher ag grepate industry-outpu Fevel asd a lower market price then the state Cournot market structure ius, the equilibrium market outcome under @ sequential-woves game is mare competitive then the Cournot. equiliteiur outcome in the sense thot this outcome is somembere in between the competitive equilib rium outcome derived in chapter 4 and the Cournot outcome derived in section 6.1. The inttition behind Proposition 6.1 is as follow: Un- der the Cournot market structure, firm 1 perceives the output produced by fem 2s given, However, tinder sequential-moves market structure, firm 1 knows firm 2's best-reeponse function and therefore calculates that firm 2 will reduce ts ona level in response to iés increase in output level, Hence, when firm 1 expands output, it expects the price ta fall faster under Cournot than under esquential-moves matket struc- tire, ‘Therefore, in order maintain high price, firm 1 will produce more under the sequential game thaa it will under Cournot, Now, (6.15) and (6.16) demoustrate that the increase in aggregate output stems from the fact that the follower does nat find it profitable to cut its output level Thy the same eanount as the inareare in the lender's ontput leved. This happens because tive reacticm funetions are sloped relatively fat (shops is negative but exceeds -1), implying that a firm reduces its output level by los them the increase in the output level of the rival firm. ‘We now compare firms’ profit Jevels under sequential moves to the Cournot profit levels. We leave it to the reader to verify that the leader's profit increases hile the follover's declines. Ths is, (6.18) where af anit § are given in (6.7). Note that we coukd have concluded ‘even wilhout going into the precise calculations that the leader's profit tinder the sequential-game equilibrium wil be higher than under the ‘Cournot. How? It is very simple! Since firm 2 reacts in a "Nash fash jon.” firin 1 could just choose to produce the Cournot output level 45 Im this ease, firm J would earn exactly the Cournot profit. However, ince in the sequential game frm 1 chooses to produce # different out- Dut level, 1 mist be increasing its profit compared with the Cousnot profit level. ‘The kind of reasoning we jnst deseribed is called a revoaled proftakibty argument, and the reader is urged to learn to use this kind ‘whenever possible because perforting calculations to in- vestigate economic effets does not generate an intuitive explanation for these effects, Tn contrast, logical deduction often provides the necessary intuition for understanding ceamomie phenomena, 6.3 Bertrand Market Structure 107 Finally, we can logically deduce how industry profit under sequential ‘moves compare with Indusicy profit under Cournot. Equations (6.17) show that the market price tnder sequentiat moves is lower than is is under Cournot, Since the Cournot morket price is lower than the monopoly’s price, and since monopoly makes the highest possible profit, itis cleae that industry profit must drop when wo further recuce the price helow the monopoly’s price. Hence, whenever cy = e2, industry profit must be lower under sequential roves. In a more general enti- Tonment, this argument may not holds when the industry profit is not a concave function ofp 6.3. Bertrand Market Structure Ina Cournot market structure firms were assumed to choose theit out put levels, where the market price adjusted to clear the masket and ‘was found by substituting the quantity produced into consumers! de- rmaad function. in conteast, in a Bertrand market structure firms set prices rather than output levels. The attractive feature of the Bertrand Setup, compared with the Cournot market structure, stems from: the fact that Gems are able to change prices faster and at less cost than to set ‘quantities, because changing quantities will require an adjustment of Sentories, which may necessitate a change in frms’ capacity to produce, ‘Thus, in the short run, quantity changes may not be feasible, or may be too costly to the seller, However, changing prices is a relatively low-cost action that may require only a change in the labels displayed on the shelves in the store. Let us turn to the Bertrand market structure. {In 188% Joseph Bértrand published a review of Cournot's book (1838) harshly critical of Courpot's modeling. It seems, however, that Bertrand vas dissatisfied with the general modeling of oligopoly rather than with the specific model derived by Cournot. Today, most economists believe ‘that quantity and price oligopoly games ace both needed to understand @ variety of markets. Thats, for some markets, an assumption thak firme set quantities tay yield the observed market price and quantity pro- duved, whereas for others, a price-setting game may yield the observed ‘market outeomes. Our job as economists would then be to decde which market structure yields a better approximation of the abserved price and quantity sold in each specific market. ‘We now analyze the two-firm industry defined in (6.1) and (9.2) and look for a Nash equilibrium (se2 Definition 2.4) in a game where the two firms use their prices as their actions. First, note that sv far, our analysis has concentrated on a single market price determined by our tesumptioa that consumers are alveays on their demand curve. How- fever, in a Bertrand game we have to consider outcomes where each fran108 Markets for Homogeneous Products ots a diflerent price for its product. ‘Therefore, we now make two ex plicit assumptions about. consumers” behaviar under all possible prices announced by oth firms 3, Consumers.chways purchase from the cheapest seller. 2, If two sellers charge the same price, half of the consumers purchase from firm 1 and the other half purchase from firm 2 Formally, we modify the demand given in (6.2) to capture the quantity demand faced by each firm i, 1= 1,2. Thereiore, we assume that © ifpaa p> vs tp.=p, ify; < min{a,p5) LAiA; G19) iquation (6.19) isthe quantity demand facing firus ¢ at any given px ‘nd pa and incorperates what is commonly called rationing rule, which tells us how the market. demand is divided between two firms selling & Domogeneous product. ‘Thus, if firm i charges a higher price than frm j, then no consumer would purchase the product from firm i. In contrast, iffy < py, then all the consumers will purchase only from firm é, and none will purchasefrom frm j. In this ease, the quantity demanded feom firm iis calculated directly from (6.2). Finally, if both firms charge the sane prices, chen the quantity demand detariained in (6.2) is equally split between the two firme, Dernarvton 6.2 The quadruple (ota. a8) #8 @ Rertram equilibrium if Nash 1 given pa = pl, pf maziizes maaip, 3 (pnso8) = (mi ~ ea)on given py = ph, pf marimizes mary, r2(G2, Ps) = (Pa ~ eae 5 ty and qe are determined in (6.19). Definition 62 states that in a Bertrand-Nash equilibrium, no frm can increase its profit by unilaterally changing its price In the next two subsections we apply Definition 6.2 to two types of markets: the first, where firms do uot have eapacity constraints end ean produce any amount they wish under the assumed cost structure; and the second, where we assume that firms’ capacities ere limited and ‘herefore, in the short run, they are unable to expand production. 6.3 Bertrand Market Structure 109 6.3.1 Solving for Bertrand equilibrium Before we characterize the Berteand cquilibris, it ie important to um derstand the discontinuity feature of this game. In the Cournot gaune, the payof (profit) functions ace coutinuous with respact to the steategic variables (quantities); in the Bertrand price game, by couttast, ex Hon (6.19) exhibits © discontinuity of the payolf functions at all the outcomes where pr = po, That is, If one firm sells at a price that is ‘one cent higher than the other firm, it would have a zero market share, However, a two-cont price reduction by this firm woeld givo this firma a fone 100 percent: market share. The sction of a firta to sligauly reduce the price below that of its competitor is called wnderculting. Since un- ercutting involves setting a price slightly lower than the competitor's ‘we need to examine the types of currencies used in cider to determine ‘the smallest possible undercutting actions available tc firms. ‘Theretore, we make the following definition: DEFINITION 6.8 Let ¢ be the smallest possible monetary denomination (Gmailest legal tender). The medium of exchange (money) is satd to be continuous ife=0, and discrete sf > 0, Exainples of discrete smallest legal tenders are: in China, ¢ 1 Benji Israel, €= 5 Agorot; and in the US, e= | vent. The following proposition characteriees Bertrand equilibsia. Proposition 6.2 1. Uf the medium of exchange is continuous and if the firms have the same cost structure, (ey = 1 = c), then a Bertrsnd euibrinm is pha ot = c, and at = a = (a ~0)/(20). 2, Let the medium of exchange be discrete, and assume that cx iy de- ‘ominated in the medium of exchange. Tht ty, t4 =e, where AZ Lis an snteyer Also let be suffctetly smal. that sate ing (62 ~e~e1) (#922) 5 (eg ~e4) (8554). Then, fore, > ¢, the unique Bertrand squtriun is py 6d =0, andgeeanait oie ey PL ‘Thus, if firms have equal unit costs, the Bertrand equiibrium prive and aggregate output are the same as for the competitive equilibria, In other words, undereutting reduces the prices to matyisal cost. In cases ‘where firm 1 bas a lower unit eost than fin 2, firm 1 undercats rin 2 by charging the highest possible price that is lower than ¢,, which is siven by ps = 03 —6