Market structure and competition policy

Market Structure and Competition Policy applies modernadvances in game-theory to the analysis of competition policyand develops some of the theoretical and policy concerns asso-ciated wi

Market Structure and Competition Policy applies modernadvances in game-theory to the analysis of competition policyand develops some of the theoretical and policy concerns asso-ciated with the pioneering work of Louis Phlips Containingcontributions by leading scholars from Europe and NorthAmerica, this book observes a common theme in the relation-ship between the regulatory regime and market structure Sincethe inception of the new industrial organisation, economistshave developed a better understanding of how real-world mar-kets operate These results have particular relevance to thedesign and application of anti-trust policy Analyses indicatethat picking the most competitive framework in the short runmay be detrimental to competition and welfare in the long run,concentrating the attention of policy makers on the impact onthe long-run market structure This book provides essentialreading for graduate students of industrial and managerialeconomics as well as researchers and policy makers.

ge o rg e n or ma n is the holder of the Cummings FamilyChair in Entrepreneurship and Business Economics at TuftsUniversity and is the Director of the Graduate Program inEconomics His publications include Economies of Scale,Transport Costs, and Location, The Economics of ImperfectCompetition: A Spatial Approach, and Industrial Organi-zation: Contemporary Theory and Practice (with D Richardsand L Pepall)

jac q u e s -f r a nc° oi s t h is s e is Professor of Economics inthe Center for Operations Research and Econometrics at theUniversite Catholique de Louvain He is the author of DoesEconomic Space Matter? and Discrete Choice Theory ofProduct Differentiation

Market Structure and Competition Policy Game-Theoretic Approaches

Edited by

GEORGE NORMAN

and

JACQUES-FRANCËOIS THISSE

The Pitt Building, Trumpington Street, Cambridge, United Kingdom

  

The Edinburgh Building, Cambridge CB2 2RU, UK

40 West 20th Street, New York, NY 10011-4211, USA

477 Williamstown Road, Port Melbourne, VIC 3207, Australia

Ruiz de Alarcón 13, 28014 Madrid, Spain

Dock House, The Waterfront, Cape Town 8001, South Africa

©

George Norman and Jacques-Franc°ois Thisse 1

1 Competition policy and game-theory: re¯ections

based on the cement industry case

Claude d'Aspremont, David Encaoua and

5 Should pricing policies be regulated when ®rms

may tacitly collude?

George Norman and Jacques-FrancËois Thisse 96

6 Tougher price competition or lower concentration:

a trade-off for anti-trust authorities?

Claude d'Aspremont and Massimo Motta 125

7 The strategic effects of supply guarantees:

the raincheck game

Jonathan H Hamilton 143

v

8 Product market competition policy and technological

10 Modelling the entry and exit process in dynamic

competition: an introduction to repeated-commitment

models

Jean-Pierre Ponssard 215

11 Coordination failures in the Cournot approach

to deregulated bank competition

Andre de Palma and Robert J Gary-Bobo 232

12 How the adoption of a new technology is affected

by the interaction between labour and product marketsXavier Wauthy and Yves Zenou 271

1.1 Cement consumption, 1970±1994 page 141.2 Maritime and land models of market structure 151.3 The effect of competition on pricing: Cournot

and Bertrand examples 243.1 The continuous Folk Theorem 624.1 AD investigations, by reporting country, 1987±1997 785.1 Comparison of price equilibria 1105.2 Welfare comparison of non-discriminatory versus

discriminatory pricing 112

6.2 Two-®rm Bertrand case 1396.3 Three-®rm Cournot case 1397.1 The six regions for the pro®t functions 1538.1 Exponential density function 1688.2 Equilibrium g ÿ p…ci† 1738.3 Expected time to discovery, monopoly and alternative

duopoly cooperation regimes 1748.4 Net social welfare, alternative R&D regimes 1758.5 Expected time to discovery, alternative spillover levels 1798.6 Net social welfare, non-cooperative R&D regimes,

alternative spillover levels 1808.7 Expected time to discovery, cooperative R&D regimes,

alternative spillover levels 1808.8 Net social welfare, cooperative R&D regimes, alternative

8A.1 Equilibrium g ÿ p…2qN†, g ÿ p…Q† 1889.1 The structure of the industry in two- and three-agent

10.1 The selection property 22510.2 The strategic values of incumbency 22811.1 Best response: risky, safe and supersafe 24711.2 Expected pro®t in the discontinuous case 25311.3 Equilibrium: risky, safe and supersafe 25411.4 Asymmetric equilibria 25711.5 Multiple symmetric equilibria 25811.6 Equilibrium correspondence 25912.1 A representative partition of the (a, g) space

according to the nature of SPE 281

1.1 Cost/pro®t structure for a production unit page 161.2 Estimation of transport costs 184.1 AD investigations, by reporting country, 1987±1997 775.1 Welfare comparisons 1137.1 Equilibrium types: comparative statics with respect

to demand uncertainty 1567.2 Pro®ts and prices in the four subgames 1588.1 Static monopoly market performance, alternative

investigation thresholds 1688.2 Static duopoly market performance, alternative

investigation thresholds 17012.1 The second-stage payoff matrix 278

ix

Claude d'Aspremont Universite Catholique de Louvain, Belgium

Gianni De Fraja University of York, England

Andre de Palma Universite de Cergy-Pontoise, France

David Encaoua Universite Paris I, France

James W Friedman University of North Carolina, USA

Robert J Gary-Bobo Universite de Cergy-Pontoise, France

Jonathan H Hamilton University of Florida, USA

Stephen Martin University of Copenhagen, Denmark

Massimo Motta European University Institute, Italy

Damien J Neven Universite de Lausanne, Switzerland

George Norman Tufts University, Massachusetts, USA

Jean-Pierre Ponssard Laboratoire d' EconomeÂtrie de l'Ecole Polytechnique,

FranceP.K Mathew Tharakan Faculteit Toegepaste Economisch Wetenschappen,

BelgiumJacques-FrancËois Thisse Universite Catholique de Louvain, BelgiumXavier Wauthy FaculteÂs Universitaires Saint-Louis, Belgium

Yves Zenou Ecole Nationale des Ponts et ChausseÂes, France

x

Louis Phlips: a brief biography

Louis Phlips was born in 1933 in Brussels He received a doctorate ineconomics at the Universite Catholique de Louvain and a doctorate inlaw at the Katholieke Universiteit Leuven After teaching for four years atFribourg University (Switzerland), he returned to his alma mater from

1966 until 1989 He then joined the European University Institute inFlorence (Italy) until 1997, when he retired from Academia He nowspends his time playing the piano and drawing

During his 23 years at the Universite Catholique de Louvain, Louiswas ®rst director of the Economic Analysis Unit (ANEC) and then amember of CORE (Center for Operations Research and Econometrics).Before Switzerland, he visited Nuf®eld College to work with P.S.W.Andrews on industrial pricing Between Switzerland and Louvain, hevisited the University of Chicago (Fall and Winter 1965) and HarvardUniversity (Spring and Summer 1966) to work with Henk Houthakker onhis econometric utility tree He wanted to ®nd out which empirical com-modity groupings correspond to the theoretical concept (if any) of an

no statistical industry data available for Europe, Louis volunteered tocompute concentration ratios from the new industrial census They formthe bulk of his North-Holland monograph Effects of Industrial Concen-tration: A Cross-Section Analysis for the Common Market (1971).His Ph D thesis had a yellow cover It was followed by the green North-Holland cover Having chosen blue for his elementary time series analysis(Analyse chronologique), the next cover couldn't but be red Louis used to

xi

say: `Mao has his red book, why shouldn't I have one too?' AppliedConsumption Analysis (1974, enlarged 1983) was the result Its exercisesmade a full generation of doctoral students suffer and grumble, but at theend of the day they had learned how to bridge the gap between demandtheory and econometrics.

After 10 years of applied econometric work, Louis felt relieved toreturn to his old love, industrial economics His 1983 CambridgeUniversity Press book of The Economics of Price Discrimination washis answer to a remark (made by a so-called `pure' theorist) that pricediscrimination is a market failure Louis made the point that if pricediscrimination is a failure, then everything business men do is a failuretoo: price discrimination is indeed present everywhere, in pricing over thespace and time domain as well as over the quality and income domain.Being surrounded by world-famous theorists, Louis had to try his hand

at applying some (elementary) game-theory to his favourite topics Hencehis Economics of Imperfect Information (1988) which got a policy ¯avour

in Competition Policy: A Game-Theoretic Perspective (1995), both fromCambridge University Press

The ®nal bouquet was ± at the request of Cambridge University Press ±

a selection of published articles which he found worth turning intorequired reading on Applied Industrial Economics (1998)

Louis was a visiting professor at the Universities of MontreÂal,Pennsylvania, Cornell, Leuven and Bielefeld One of the founders ofthe European Economic Association, he was its Secretary until 1989and its President for 1995 He is a member of the Academia Europaeaand a fellow of the Econometric Society

George Norman and Jacques-Fran Ëcois Thisse

In his doctoral thesis published in 1962, Louis Phlips argued that European

®rms in the cement industry attempted to coordinate their actions by usingbasing-point pricing systems and more or less formal agreements aboutgeographical markets At the time that Louis was formulating his ideas,European competition policy was still in its infancy It is perhaps nosurprise that those who were formulating policies at that time paid littleattention to the work of a doctoral student It is somewhat ironic that thesehave come to centre stage at the end of Louis' distinguished academiccareer It is also amusing to note that after a long and productive detourthrough consumption analysis, applied econometrics and industrial eco-nomics, Louis himself has chosen to return to his original love as shown byhis Competition Policy: A Game-Theoretic Perspective

Game-theoretic methods are now indispensable in the design, tion and testing of competition policy in Europe and anti-trust policy inthe United States Until very recently, the connection was from marketstructure through market behaviour, as explained by game-theoretic tools,

formula-to competition policy We can see this timeline, for example, in theformulation of merger policy and policies with respect to cartels What isnew is the realisation that this is a two-way street Just as market structureaffects competition policy, competition policy equally affects marketstructure As European competition policy is becoming more active, ithas become increasingly endogenised in the strategic decisions of the ®rmswhose behaviour the policy is intended to affect It is dangerous for policymakers to ignore this change in behaviour For example, we are now awarethat in some circumstances making a market more competitive is notnecessarily bene®cial to consumers Rather, the additional competitionmay increase market concentration and may facilitate tacit or even explicitcoordination among the surviving ®rms This connection from competitionpolicy to market structure and the welfare effects of policy is a recurrenttheme of this book

1

Louis' early interest in basing-point pricing extended to spatial pricepolicy when he wrote a report for the European Commission in 1976 Thisculminated in his book on The Economics of Price Discrimination that had

a signi®cant in¯uence on scholars and policy makers alike An essentialpreliminary to any discussion of price discrimination is that we should beable to de®ne what we mean by `discriminatory prices' The conventionalde®nition prior to Louis' analysis was that price discrimination exists whenthe same product is sold to different consumers at different prices but this isunsatisfactory, for at least two reasons First, such a de®nition might lead

us to conclude that price discrimination exists when a company sells itsproduct in two different cities ± say, New York and London ± at differentprices Clearly this conclusion would be wrong since it ignores the differentcosts of supplying these two cities Secondly, we might conclude that there

is no price discrimination if the ®rm sells its product in London andNew York at the same prices This is equally wrong since the prices now

do not re¯ect the different costs of supplying these two cities Louis wasable to circumvent these problems by providing us with the followingde®nition:

Price discrimination should be de®ned as implying that two varieties of a modity are sold (by the same seller) to two buyers at different net prices, the netprice being the price (paid by the buyer) corrected for the cost associated with theproduct differentiation (Phlips, 1983, p 6, emphasis in the original)

com-Applying this to our example, price discrimination exists if the differencebetween the London and the New York prices is not equal to the difference

in the seller's marginal costs of supplying London and New York.Starting from this de®nition, Louis was one of the ®rst to point out thatprice discrimination is a pervasive marketing practice that survives despitethe attempts by regulators to limit or eliminate its use This might come as

no surprise if we were to consider only situations where ®rms are able toexercise considerable market power since price discrimination provides the

®rm with a remarkably ef®cient means by which consumer surplus can beconverted into pro®t What was more surprising and in¯uential was Louis'clear demonstration that price discrimination is widespread in oligopolisticand more generally imperfectly competitive markets Moreover, he showedthrough both theory and evidence that the degree of price discriminationpresent in such markets is, if anything, stronger than would characterise amonopolist in the same markets This analysis set an agenda that remainscurrent and active today

European cement manufacture provides a classic case study of many

of Louis' ideas The price and competition policies of the major facturers are under scrutiny by the European Commission Chapter 1, by

manu-d'Aspremont, Encaoua and Ponssard shows how the questions that vated Louis Phlips in his doctoral dissertation can be revisited usingmodern game-theoretic techniques In particular, these authors discuss therelationship between spatial pricing policies and market behaviour andperformance in an industry characterised by high transport costs Theiranalysis provides an important illustration of the connection noted abovebetween competition policy and market structure Denying cement ®rmsthe use of, for example, basing-point pricing, has increased price competi-tion but has also been associated with a dramatic increase in marketconcentration.

moti-There is a related issue that also recurs in a number of chapters in thisbook: the role of information d'Aspremont, Encaoua and Ponssard discussthe impact on prices of facilitating practices such as most-favouredcustomer clauses or meet-the-competition promises Recent analyses sug-gest that this kind of information exchange between ®rms changes theresulting market equilibria from Bertrand to Cournot, with the surprisingresult that consumers lose out These authors show that this is a short-runeffect only that ignores the connection between the competitive environ-ment and long-run market structure The idea behind this is in fact verysimple and general If ®rms expect tough competition (e.g aÁ la Bertrand) weare likely to see greater industry concentration and higher prices than ifthey anticipate soft competition (e.g aÁ la Cournot)

Competition policy is still evolving in the European Union, perhapsbecause such a policy is relatively young in Europe by historical standards.This is in sharp contrast with the long history of anti-trust policy in theUnited States Neven in chapter 2 correctly points out that European policymakers could bene®t from applying some of the lessons that have beenlearned in the United States There are some common elements Forexample, on both sides of the Atlantic, the principle is emerging that thepossession and exercise of market power is not of itself evidence of violation

of competition or anti-trust rules Rather the appropriate courts have to

®nd evidence of explicit coordination when there are several ®rms involved,

or evidence of attempts to extend market power when the market iseffectively monopolised Microsoft was not being investigated because ithas an effective monopoly of operating systems It was being investigated tosee whether it has tried to use its operating system monopoly to extend itsmarket power into browser markets By contrast, there are some sharpdistinctions between United States and European policies Neven points totwo of these First, it is reasonably common practice in the United States totake the existence of market power as an indication of the possibility thatthere is coordination between ®rms Secondly, the United States anti-trustauthorities tend to take the existence of facilitating practices as a

presumption of coordination Neither principle is yet established inEurope.

There is a major dif®culty confronting the Commission in its pursuit ofcoordinating practices that is well articulated by Friedman in his insightfuldiscussion of the Folk Theorem in chapter 3 This can be simply stated.Once ®rms recognise that they interact repeatedly, then it is possible forthem to settle on a non-cooperative dynamic equilibrium that looks verylike a market outcome that would emerge from explicit coordination This

is an example of what Louis Phlips has referred to as the ability problem' The theory of repeated games suggests that ®rms can formnon-cooperative strategies that support collusive outcomes These strate-gies always involve some credible threats to punish deviations It is dif®cult

`indistinguish-to see how these threats can be made credible without their being municated between the relevant ®rms since in principle they are neveractually observed The act of communication is in violation of competitionpolicy, but is remarkably dif®cult to observe

com-An equally dif®cult issue facing both the Commission and the national trading community is the design and implementation of effectiveanti-dumping (AD) legislation These problems are eloquently addressed

inter-by Tharakan in chapter 4 and draw together two important themes ofLouis' work: price discrimination and the design of competition policy, inthis case at the supra-national level A particularly interesting feature of theuse of AD measures is the dramatic proliferation in the number of countriesinitiating such measures In 1990 four groups launched around 82 per cent

of AD investigations: Australia, Canada, the European Union and theUnited States By 1997 this proportion had fallen to less than 49 per centwith AD actions being actively used by a number of developing and NewlyIndustrialising Countries (NICs) There is a danger that the strategic use of

AD measures will seriously undermine movement towards multilateraltrade liberalisation Indeed, there is the real-risk that these measures willlead to the escalation of protectionism under the guise of measurespurported to ensure some kind of `level playing ®eld' in international trade.Tharakan points out that the welfare effects of AD legislation are atbest ambiguous ± a conclusion that applies equally to legislation intended

to prevent price discrimination Indeed, most of the analysis that has beenconducted has concluded that AD legislation actually imposes largewelfare losses on both the exporting country and the importing countrythat initiates the AD investigation The solution that is suggested to correctthe detrimental strategic and welfare consequences of AD actions is tochange the regulations developed by the World Trade Organisation (WTO)and individual nation states on AD legislation, restricting their application

to cases of predatory price dumping This type of dumping does have

detrimental welfare effects and needs to be corrected Identifying suchdumping suffers from many of the same problems that confront the anti-trust authorities in trying to prove predatory pricing within a country:another application of Louis' `indistinguishability problem' Tharakanpoints out, however, that a number of new methods have been developedfor detecting attempted predation One particularly useful such testinvolves a `two-tier approach' First, assess the market power of thesupposed predator: only if such power exists is predatory power eitherfeasible or likely For those cases that `pass' the ®rst test, consider price±cost and other factors It is a relatively simple matter to extend this type oftest to the international arena If this had been done, Tharakan notes thatits impact would have been to reduce signi®cantly the number of ADcomplaints that reach the second-tier test.

We noted above that there is an important link from competition policy

to market structure that has been neglected by policy makers, both inEurope and in the United States The next group of chapters focuses on thislink from different perspectives Norman and Thisse in chapter 5 argue thatthe naive application of the idea that competition is always and everywheredesirable may have unforeseen and harmful effects Policies that create tootough a competitive environment may be detrimental to consumers andsocial welfare through their impact on ®rms' medium- and long-rundecisions The stronger are the structural effects of competition policy,the more likely is it that blind adherence by the anti-trust authorities to thebene®ts of competition is misguided In particular, these authors show thatconsumers are likely to lose from price deregulation in markets char-acterised before deregulation by high levels of concentration This suggests

a role for regulators that has not been considered, despite the fact that it lies

at the heart of the Folk Theorem of repeated games: the regulator shouldimpose a minimum period of time over which prices cannot be changed.Such a slowing in the speed of response undermines the effectiveness of thepunishment that supports the tacitly collusive outcome

The same trade-off between tough competition and concentratedmarket is also at the centre of d'Aspremont and Motta's work and con-cerned Phlips in the introduction to Applied Industrial Economics Inchapter 6, they develop a similar set of policy conclusions, using a differentsetting Speci®cally, they consider a situation in which anti-trust authoritiesattempt to break down price coordination to create an environment inwhich prices are set competitively In so doing, the variety of products isreduced, prices are increased and consumers are worse off

Hamilton in chapter 7 considers a related but somewhat different set ofideas In the United States, the Federal Trade Commission (FTC) regulatesadvertising In particular, it has developed policies to prevent the use of

`bait-and switch' tactics If a ®rm advertises a low price, it must also be able

to show that it has suf®cient inventory to meet anticipated demand Thiscan be a signi®cant constraint on the ®rm but it can be circumvented if the

®rm offers a `raincheck' This is a promise to supply at the sale price oncenew inventory has been received What Hamilton shows is that the require-ment that rainchecks be offered deters vigorous competition Again, apolicy designed to protect consumers may actually harm them

The idea that competition policy may drastically affect market structure

is illustrated in chapter 8 by Martin in a yet different context Suppose that

®rms can undertake R&D that leads to process innovations The ®rms mayalso collude and the competition authorities take market performance as asignal of the potential existence of collusion Martin shows that a strongercompetition policy reduces both pre-innovation and post-innovationpro®ts, but the latter relatively less than the former Consequently, toughercompetition policy induces additional R&D spending The additionalR&D, by reducing costs, also reduces the probability of investigation by theauthorities This is not necessarily bene®cial to the collectivity There is aninverted U-shaped relationship between competition policy and expectednet welfare Once again, a moderately strict competition policy improveswelfare; excessively strict competition does not

Regulation remains one major dimension of competition policy, cularly in its application to the behaviour of previously state-ownedmonopolies that have been privatised, or the creation of new naturalmonopolies such as cable television The interesting issues are, ®rst, thedesign of regulatory policy itself and, secondly, whether it is possible tocreate a competitive environment in some industries De Fraja in chapter 9discusses some of the main issues that arise in the design of regulatoryregimes Recent analyses have discussed how competition and regulationaffect industry performance and how the interaction between the regulatorand the regulated affects industry structure in ways that are determined

parti-by the regulatory rules What De Fraja shows is that a wide variety ofoutcomes can arise, leading to the need for a case-by-case approach to themodelling of the interplay between the regulator and the regulated.The ®nal three chapters of this book open new avenues for research inwhich competition policy, while not yet developed, will undoubtedly have

an important role to play It is fair to say that time is at best implicit in manygame-theoretic contributions to the design of competition policy Yet,entry and exit of ®rms arise in real-time and seems to exhibit some robuststylised facts: (1) entry is frequent and relatively easy; (2) entry tends to beassociated with innovation; (3) entrants suffer a high failure rate and(4) exit follows successful entry Ponssard in chapter 10 develops a dynamicmodel of competition that has the potential of exhibiting many of these

features The main message to be drawn from his analysis is that theoutcome of the entry/exit dynamics will be determined by the interplaybetween competitive advantage that tends to favour entrants and mobilitybarriers that tend to favour incumbents.

The banking industry has often been considered as a prototype of acompetitive market However, the large number of mergers observed inrecent years suggests that it is now more appropriate to see these markets asbeing oligopolistic Although the banking sector has been and is still veryregulated, very little attention has been paid to the process of competitionbetween banks In order to develop appropriate tools in competitionpolicy, one must develop a better understanding of the future working

of this sector because of its new more concentrated structure In thisperspective, de Palma and Gary-Bobo in chapter 11 present one of the ®rstmodellings of oligopolistic competition of the banking sector They showthat the behaviour of banks is potentially unstable in that a small change inthe underlying parameters can induce a sharp change in equilibria ± forexample, from safe to risky Their contribution thus sheds light on theimportance of determining the role of the central bank as a regulator ofcompetition in this sector

The spirit of chapter 12 by Wauthy and Zenou is similar in that it invites

us to think of other institutions as possible actors in the design ofcompetition policy It draws our attention toward the interaction betweenthe product and labour markets By affecting the product market, the anti-trust authorities may in¯uence the choice of technologies and, therefore,the need for skilled or unskilled workers One is not accustomed to think

in these terms but their contribution leads us to think of the possibleimplications for workers of competition policy as well as of the connectionsbetween competition policy and training

1 Competition policy and

game-theory: re¯ections based on the

cement industry case

Claude d'Aspremont, David Encaoua and

Jean-Pierre Ponssard

1 Introduction

Is the main objective of competition policy the maintenance of competitionper se or the promotion of economic ef®ciency? These two goals do notnecessarily have the same basis or the same implications.1 The goal ofmaintaining competition per se can be justi®ed morally, politically andlegally by the wish to protect individual freedom and rights, and bylimiting the power of agents This faith in the democratic virtues ofinteracting competitive forces is grounded in a political philosophywhich sees regulatory mechanisms resulting from impersonal marketforces as a guarantee against the arbitrariness of authority, whetherpublic or private In this sense, competition is a right which warrantsprotection Economically, competition is not considered as an end in itselfbut rather as a mechanism for allocating resources which in many, if notall cases, promotes economic ef®ciency The question the economist hasthen to answer is whether or not, depending on the circumstances, com-petition promotes the reduction of costs, the selection of the most ef®cientbusinesses, the welfare of consumers, the creation of new products, theentry of new enterprises, the development of technological progress andinnovation and so on

To what extent do these two goals of competition policy overlap?Before setting out our framework to formulate an answer to this question,let us introduce the basic issues

Clearly, if competition policy adopted an exclusively normativeapproach, consisting of the decentralised inducement of an ef®cient

9

An initial version of this chapter was presented at the conference `Economic Analysis and Antitrust Issues in Concentrated Sectors: The Case of the Cement Industry', Paris, Carre des Sciences (15 January 1996) We wish to thank Louis Phlips, Herve Tanguy, Jacques- FrancËois Thisse and the other participants at the conference for their comments and suggestions.

1 See, for example, Jenny (1993); Encaoua (1997).

allocation of resources, based on the perfectly competitive behaviour of

®rms, the convergence between the above two goals would be total,according to the First Welfare Theorem Such an approach means, how-ever, that each business would be obliged to comply with the rule ofmaximising pro®ts by taking the environment in which it operates as ®xed

± an outrageous requirement We know that that is not how competitionpolicy functions Rather than decreeing rules a priori, free competitionlimits itself to prohibiting certain types of behaviour judged to be repre-hensible in so far as they hinder the free play of market forces However,the interpretation of this notion is tricky since no precise system ofreference exists for judging deviant behaviour

Thus, in many oligopolistic sectors the reference to `perfect tion' is totally unrealistic Market forces are not impersonal and thelimited number of actors naturally leads ®rms to adopt strategic behav-iour in which they anticipate their competitors' reactions We have thus toascertain which rules would need to prevail on these markets in order toensure that the discrepancy was not too great between the principle ofmaintaining rivalry, implicit in the free play of market forces, on the onehand, and the concern to enhance economic ef®ciency and the socialoptimum, on the other

competi-The independent behaviour of the different actors is one of the guidingprinciples of all competition policies; they defend this rule by opposinganything which may indirectly facilitate collusion between ®rms (agree-ments or information exchange concerning prices, quantities produced orcapacities, etc.) However, this type of approach is soon limited without

an appropriate conceptual model to analyse imperfect competition assuch It results, for example, in only explicit agreements being condemnedwhile tacit collusion becomes acceptable, the latter being seen as anexpression of rational behaviour between independent agents with acommon perception of their environment.2

With the formalisation of imperfect competition by means of theory, another step forward can be taken The ambiguous notion ofparallel behaviour is replaced by the more precise one of non-cooperativeequilibrium It then becomes possible to re¯ect on the interaction between

game-2 Wood pulp is a case in point The alignment of prices among about 50 wood pulp producers was judged by the European Commission to be an expression of a concerted practice The European Court of Justice (ECJ), however, regarded wood pulp to be an homogeneous product for which the market is perfectly transparent It considered that the

®rms may have reacted identically to modi®cations in their environment without any formal agreement For the European Commission decision (19 December 1984), see the Ocial Journal of the European Communities, L851, and for the ECJ judgement in the appeal case (31 March 1993), see Recueil de la Jurisprudence de la Cour de Justice et du Tribunal de PremieÁre Instance, I, 1993, 3 For a case study, see Phlips (1995, pp 131±6).

certain rules of the game and the degree of economic inef®ciency of thenon-cooperative equilibrium which may result from it Some rules maythen appear to be less effective than others and be condemned as such,whereas others will be encouraged This approach thus provides a morepowerful frame for examining competition policy.

The present chapter develops this type of analysis in relation to thecement industry It considers several rules concerning price policy, theexchange of information and external growth operations (mergers andacquisitions), with particular reference to models derived from game-theory

The cement industry is a typical example of an oligopolistic sector.Cement is an homogeneous good for which the price elasticity of demand

is weak, production requires heavy investments and distribution involveshigh transport costs Consequently, there are often few local competitors.They are, however, subject to competitive pressure from the outside, fromdistant ®rms which try to sell at marginal costs

The sector has a rich history of anti-trust cases in the United States,Europe and Japan, which have provided subject-matter for an extensiveliterature on the various standpoints taken In the present chapter wedraw essentially on the cases referenced in the historical analysis byDumez and JeunemãÃtre (2000) In some of these cases there is clearproof of agreement while in many others the questions concern practiceswith far less obvious effects ± e.g the choice of price regulation (theuse of points of parity, for instance), the role of information exchangebetween competitors and the choice of the relevant market for analysingconcentration

We shall consider these questions of principle in the light of severaltheoretical developments which are particularly relevant to a study of thecement industry

First, what is the impact of a pricing system, in relation to its degree ofdiscrimination, in a context of horizontal differentiation? Numerous stu-dies have focused on this question since the ®rst articles by Spence (1976)and Salop (1979) Most reached the classical conclusion that more com-petitive pricing had a positive impact on welfare (Phlips, 1983) Normanand Thisse (1996) examined the same question by considering the role ofthe irreversibility of investments They show that highly competitive pric-ing may lead to greater market concentration and ultimately to a loss ofwelfare for the economy as a whole

The second question concerns information exchange ± or, more erally, trade practices which shape competition How are they justi®ed andwhat is their impact? Information exchange usually concerns commitments

gen-to align prices on advance noti®cation But there are other facilitating

practices Holt and Scheffman (1987) showed that such practices couldin¯uence the intensity of competition ± for example, by causing it tochange from Bertrand to Cournot competition This conclusion is used

by d'Aspremont and Motta (2000) in a context of horizontal tion They show that more intense competition may lead to greaterconcentration

differentia-These theories, in terms of both pricing and facilitating practices,provide arguments in favour of the maintenance of rules tending tomoderate competition in the short term and thereby limit concentration

in the sector Of course, in these models it is always assumed that ®rms'behaviour remains non-cooperative The question of whether a particularrule promotes agreements between ®rms remains relevant However,empirical studies by Sutton (1991) reinforce the general assumptionthat the intensity of competitive pricing can have a retroactive effect onconcentration The value of theoretical analyses is then to specify themechanisms which may favour this retroaction to a greater or lesserdegree

Lastly, we examine a point which has received relatively little attention

in the literature but is directly relevant to the empirical analysis of thecement industry When we study this sector over periods of about 10years, we are struck by the considerable importance of the buying andselling of assets ± production units, here ± for purposes of restructuring(Tanguy, 1987) The indivisibility of investments, the stagnation ofdemand in most developed countries and the increase in the minimaleconomic size of production investments are all factors which makecompetition in the cement business resemble a game of Go In thisgame, some positions which are still pro®table do not seem viable inthe long term; the company then tries to sell them at a pro®t to a rival

in a better position who has anticipated the situation more accurately.This process of restructuring the industry, favoured by a degree of ®nan-cial concentration, seems to play a major part in the strategy ofcement ®rms (Collomb and Ponssard, 1984) From a theoretical point

of view, we are then led to question the relationship between short-termcompetition rules and ®rms' capacities to engage in this process of long-term ef®ciency

A natural starting point for the study of this question is the modelling

of competition in a dynamic context with free entry Now, it has alreadybeen shown that in this type of context strong potential competitionwhich facilitates entry does not necessarily lead to greater economicef®ciency but may, on the contrary, lead to a waste of capital (Eatonand Lipsey, 1980; Maskin and Tirole, 1988) This results from the factthat incumbent ®rms may be induced to create entry barriers arti®cially

by means of defensive measures involving heavy costs (advertising, tical integration, renewal of assets before this is due, etc.) rather thanlowering their prices This analysis, developed in the absence of compe-titive advantages between ®rms, has been completed so as to take intoaccount possibilities of asymmetry (Gromb, Ponssard and Sevy, 1997).The authors show that an effective process of selection will be initiated,

ver-in which a more ef®cient entrant will replace a less ef®cient ver-incumbent.However, this selection depends on a mechanism of rational expectationswhich presumes that ®rms are able to assess their respective structuralpositions

The rest of this chapter is organised as follows In section 2 the acteristics of the cement industry are analysed in detail Section 3 developstheoretical considerations and explains the results mentioned above Inthe conclusion (section 4) we summarise the lessons drawn from theproposed approach and suggest some general ideas in terms of competi-tion policy for an oligopolistic sector such as the cement industry

char-2 Characteristics of the cement industry

In this section we present the basic economic characteristics of the cementindustry by following the classical approach which consists of successivelyexamining demand, supply and market structure On the basis of thesecharacteristics we are then able to de®ne the main economic stakes in thesector Our presentation concerns the industrialised countries and, morespeci®cally, Europe We have drawn upon the French case for many ofour examples

Demand

Demand in the cement industry is typically that of an activity which ismature, cyclical and with low price elasticity It is also characterised by ahigh degree of horizontal differentiation in terms of location and a lowdegree of vertical differentiation in terms of quality

Let us look at each of these points Cement is an homogeneous uct Most of its sales concern about half a dozen commercial varieties, ofwhich Portland cement is by far the leader No brand name exists, so thatone supplier's products can easily be substituted for another Cement is,however, an experience good; its quality is guaranteed by standards withwhich the supplier has to comply These standards are often national but

prod-in most cases the products of one country can easily be approved prod-inneighbouring countries Standards therefore do not constitute tradebarriers as such, even if they may hinder trade

Although cement is one of the main ingredients used in the constructionindustry, it accounts for only 2 per cent of the costs The price of cementtherefore has little impact on ®nal demand which is essentially the result

of macroeconomic conditions (economic growth rate, interest rates, icy of infrastructure development, etc.) By contrast, intermediaries such

pol-as producers of precpol-ast concrete or prefabricated material are stronglyaffected by prices, with the result that pressure is constantly exerted onsuppliers to lower prices This pressure will be particularly strong whenthe sector is concentrated downstream

Figure 1.1 represents the consumption of cement in France over theperiod 1970±95 (trade syndicate data) This demand, typical of industrial-ised countries, appears to be cyclical with a downward trend after peaking

in 1974 (this peak occurred a little earlier in the United Kingdom andGermany and more recently in Spain and Italy) This demand curve doesnot encourage the entry of new competitors

Let us now consider horizontal differentiation in this sector Thedemand for cement is geographically widely dispersed and correspondsroughly to population density Although cement is an upstream industry,

it differs from other basic industries such as aluminium, steel or glass, forwhich demand is concentrated both geographically and in terms of thenumber of customers In the cement industry demand is, by contrast,dispersed in multiple zones of consumption, each of which comprisesnumerous customers Geographical factors thus determine the structure

of the market For example, in areas with high levels of consumption,accessible by waterway (such as London, Marseilles or Barcelona) themarket stakes differ from those of more isolated areas (such as Berne,Grenoble or Madrid)

Figure 1.2, adapted from Tanguy (1987), illustrates this phenomenon.

On the left, producers compete on a major market; on the right, eachproducer is relatively isolated on its natural market These two extremecases ± called the maritime and the land model, respectively, by Dumezand JeunemãÃtre (2000), as well as all the possible intermediate forms,constitute the playing ®eld of the cement industry The traditional playing

®eld is the land model, but the maritime model takes over when nication over vast distances becomes possible (the Great Lakes andMississippi networks in the United States, the North Sea network, theEuro±Mediterranean network, and so on)

commu-Supply

Two economic considerations are important a priori in structuring supply

in a market characterised by strong horizontal differentiation:

. The trade-off between ®xed costs and transport costs which, ing on the economic size of the factories, gives an initial idea of thedensity of the network of production units covering the territory, inrelation to the density of demand

depend-. The level of investment costs and the life-span of facilities whichdetermine the rigidity and the duration of the network

We shall begin our discussion of supply by giving a rough idea of themain expense items and the pro®tability factors of a production unit, and

by simplifying the transport question

(a)

Figure 1.2 Maritime and land models of market structure (a) Maritime model(b) Land model

Factory costs and key factors of pro®tability

The following data (table 1.1), drawn from interviews with professionals

in the sector in France, give a breakdown of expenses for a productionunit which has a capacity of 1 million tons per year and costs 1 billionfrancs in investments This size is representative both of current technicalcapacities and of the economic stakes in most industrialised countries.For high-growth urban markets or for on-shore plants intended for anessentially maritime environment, larger dimensions would be moreappropriate

The main items in table 1.1 may be grouped together as variableexpenses, which change in proportion to production, and as ®xedexpenses which are reduced to the ton but remain constant, irrespective

of production (By contrast, ®xed costs may vary in relation to capacity;

we shall return to this point below.) With regard to variable expenses, theitem `market access' represents transport costs for an average geographi-cal dispersion For a production of 1 million ton/year, variable expensesare 150 Fr/ton and ®xed expenses are 180 Fr/ton, a total cost (excludingeconomic depreciation) of 330 Fr/ton

In 1995 in France the average customer price including transportwas roughly 450 Fr/ton This type of factory therefore has a pro®tbefore tax of 450 ÿ 330 ˆ 120, or 80 Fr/ton after tax (for a tax rate of33.3 per cent)

To evaluate the operating pro®t after depreciation and taxes, one has

to subtract the capital charges for investment (taken here to be equal to

Table 1.1 Cost/pro®t structure for a production unit with a capacity of

1 million tons per year, running at full capacity

Capacity: 1 million ton/year Fr/ton (F/T) Total

8 per cent), taking into account the delayed deductibility of this expenseowing to tax depreciation By considering a life-span of about 20 yearswhile tax depreciation is over a shorter time period, one can obtain arough estimate in proceeding as follows First compute the tax shieldassociated with depreciation (given local ®scal rules, in our example thisamount would be 250 Fr/ton); secondly, after subtracting this amountfrom the investment cost (to obtain 1000 ÿ 250 ˆ 750 Fr/ton), computethe economic depreciation associated with this capital investment of

750 Fr/ton on 20 years With a unit capital cost of 8 per cent per yearthat gives approximately 76 Fr/ton in our case The economic rent gen-erated by this production unit would then be 4 Fr/ton This cost structureimplies that the economic rent is quite sensitive to price variation and toutilisation rate This sensitivity is typical of a capital-intensive processindustry in which the ®xed costs (after tax) together account for over 65per cent of the total cost

It is generally considered that ®xed factory costs and investments arelargely determined by capacity When the latter increases from 800 k ton/year to 1,500 k ton/year, they may be reduced by a factor of about 35per cent This calculation makes it possible to determine the part of ®xedcosts which is truly ®xed The corresponding economy explains why itmay be advantageous to build large plants, provided that transport costs

to the market are not too high

The preponderance of transport costs

Transport costs depend on several factors: the means of transport used,the quantity transported and the distance covered The three main means

of transport are: road for short distances (less than 200 km), rail forlonger distances (200±600 km) and ®nally water, essentially maritime

In the latter case, the cost is generally not considered to depend on thedistance

Each means of transport is economical not only for certain distancesbut also in relation to a minimum quantity which ranges from 25 ton for alorry to 1,300 ton for a train and about 10,000 ton for a boat or ship This

is explained primarily by the loading and unloading costs involved Boatsare usually loaded directly from an on-shore plant whereas unloadingcosts require expensive facilities

It is therefore possible to draw up a comparative table of transportcosts (see table 1.2) All the corresponding data are drawn from interviewswith professionals in the industry

In an analysis of competition transport costs, which may easily amount

to 100±150 Fr/ton, rapidly account for a signi®cant fraction of the factorycost Greater ef®ciency in production costs is thus soon lost in relation to

a competitor who may be better placed on the market On the other hand,the discrepancy between the price and the variable cost clearly constitutes

a strong incentive to marginal-cost selling Given the destabilising nature

of this type of selling, it is likely to develop over large distances to avoidpossible retaliation In such cases harbour areas will be structurally morevulnerable to imports than inland areas

If we wanted to use managerial stylisation, we could say that able transport conditions will tend to multiply the areas of contactbetween a large number of competitors, since the market will resemble

favour-a commodity mfavour-arket By contrfavour-ast, the exclusive use of rofavour-ad trfavour-ansport inareas of moderate consumption will bring together only a small number

of competitors since the market will resemble a juxtaposition of ised activities This is another way of distinguishing between a maritimeand a land model

special-The network and its rigidity

By taking the geographical distribution of demand, ®xed factory costsand transport costs, it is possible to determine an effective networking of agiven territory Applied to the territory of France, and excluding allimports and exports of cement, we ®nd about a dozen productionunits with capacities of between 800 k ton/year and several million ton/year (We note, however, that a capacity of several million ton/year is notrealistic because of competitive vulnerability.)

Although theoretical, this calculation helps us to locate the real work In 1995 France had about 20 production sites, whereas therehad been about 50 in the early 1980s for a market which, admittedly,was 50 per cent bigger Thus, the size of the plants has increased, whichhas enabled them to bene®t from economies of scale in a shrinkingmarket

net-Table 1.2 Estimation of transport costs

Transportation mode Road(0±200 km) Railway(200±600 km) Seaex: Greece±USA

Per km 0.35 Fr/ton  km 0.25 Fr/ton  km 70 Fr/ton

Total/ton 18±88 Fr/ton 85±185 Fr/ton 150 Fr/ton

This type of calculation is, however, too static and overlooks someimportant dimensions The historical analysis in the French context illus-trates the extremely rigid nature of cement production For example, nonew plant was built between 1980 and 1995 Three factors explain thisphenomenon: ®rst, the life-span of factories is very long ± about20±30 years; secondly, it is relatively more economical to renovate oldfactories than to build new ones; and, lastly, environmental constraints(notably for the opening of quarries) make the creation of new units moreand more dif®cult In a context of stagnating demand peculiar to indus-trialised countries, these three factors generate a very stable industrialstructure.

These elements enable us to complete the description of the spatialplaying ®eld by introducing a time dimension, and then examine thestrategies used by competitors on this time±space playing ®eld

Market structure and the implications of restructuring

The time±space playing ®eld has to be analysed in light of the fact thatthe vast majority of ®rms in the sector have several plants Many ofthem are part of major multinationals active in several countries.Surprisingly, in view of its regional character, the cement sector is highlyinternationalised

In France, for example, there were about 30 factories in 1995 but onlyfour rival ®rms (Holderbank, Lafarge, Ciments FrancËais and Heidelberg).These ®rms were, moreover, well established in other European countries,North America and, in some cases, Latin America and Asia Similarly, inthe United Kingdom there are three dominant cement groups, and thistype of concentration is also apparent in Spain and Germany, even if inthose countries many independent single-plant ®rms remain operational.This highly concentrated multiplant structure results far more from aprocess of acquisitions than from one of internal development Therigidity of supply explains why

Concentration has two main objectives The ®rst is the wish to stabilisecompetition in a context of a tit-for-tat-type strategy Numerous acquisi-tions in Europe thus followed the setting up of the Single Market and therapid increase in uncontrolled exports Several large companies acquiredpositions in Greece or Italy, for example, in an attempt to exert pressure

on national manufacturers

The second factor seems equally important Financial concentrationmakes it possible to bene®t from industrial rationalisation campaignsthrough the renovation and/or closure of several plants in the samearea On the one hand ®nancial concentration enables ®rms to raise

funds, which is essential in such a capital-intensive industry and, on theother, the existence of several plants close together makes it possible toreorganise ¯ows without becoming involved in trade wars.

Let us consider two examples of this type of process concerningthe border area between France and Belgium In the early 1990s the comp-any Ciments FrancËais bought out the Belgian ®rm CCB Followingthis acquisition, it closed down two of its own plants in the region.Conversely, Holderbank bought out the company Cedest and closedone of the newly acquired plants Thus, within a few years a ®nancialand industrial restructuring had taken place in an area which for a longtime had had an overcapacity This occurred without any price war for theselection of the best plants The questions are: on what was the selectionprocess based, and how effective was it?

We suggest the following interpretation In the process of acquisitionand restructuring, it was as if the ®rms practised a form of indirectcompetition on the physical assets market (either to acquire existingfactories or to sell them) rather than on the product market Consider

a given playing ®eld Some production units seem to be doomed (e.g.problems of quarries being too small to warrant heavy but essentialinvestments), although still able temporarily to defend a natural market.Moreover, for various reasons other more modern factories in the vicinityhave an overcapacity There thus exist opportunities for value creationderived from industrial restructuring The ®rms will prepare this type ofset up by means of purely ®nancial acquisitions and/or overinvestments inexisting sites to discourage investment in other sites This amounts to asort of game of Go in which the status of a production unit may switchfrom one side to another without this being immediately foreseeable Thefact that there are now four cement ®rms in France while in the 1960sthere were close to 40 accounts for the size of the phenomenon of ®nancialconcentration and industrial restructuring

The cost structure is at the origin of this process It explains why aplant, even an old one which is less ef®cient as regards variable costs and

®xed factory costs, yet no longer has depreciation charges, remains ginally pro®table unless the selling price drops by at least 40 per cent.However, this type of price decrease would by nature be extremely costlyfor all the actors involved In the cement industry, the selection process

mar-by price war is hardly credible and easily back®res on those who initiate it

A production unit is consequently a long-term strategic asset It allows

a ®rm either to acquire plants close by in order to improve the ef®ciency

of the area, or to realise a capital gain on sales by trying to recover asigni®cant part of the value derived by the acquirer from this enhancedef®ciency

3 Theoretical analysis of some relevant competition rules for

the cement sector

In this section we examine the theoretical implications of the abovecharacteristics of the competition process, as regards both the priceregime and information exchange (or others facilitating practices).When transport costs account for a signi®cant proportion of all costs,competition must be analysed on two levels The ®rst is the establishment

of the way in which transport costs are incorporated into prices Multiplepricing systems are possible At this level coordination between ®rms mayalready appear, in so far as they may agree on a particular pricing systemfor transport costs The second is that of the establishment of price levels

as such, incorporating transport costs in relation to the pricing systemadopted at the ®rst level At this second level (which is the only one toconsider when transport costs are not very high) coordination, or mutualunderstanding on the basis of information exchange between ®rms, plays

an essential part in the establishment of a mode of competition

The role of the pricing system

In a context of geographical differentiation, competition is extremelyintense locally (although limited to a small number of neighbouringcompetitors) By nature it is scarcely affected by changes in distantareas In these conditions, the direct threat regulating prices is theentry of a competitor This may either be a direct entry through theconstruction of a new production unit or, more probably, in the cementsector, an entry linked to the construction of a terminal allowing for massdeliveries from an existing but distant plant Entry on the market is thus amajor strategic decision in which the reaction of local competitors cannot

be overlooked

One of the ®rst articles to study this question was that of Eaton andWooders (1985) The authors showed, in particular, that spatial competi-tion is `®erce' but `local' The same point is examined by MacLeod,Norman and Thisse (1988)

To study this question, two systems of pricing are usually chosen: thesystem of uniform FOB pricing (or mill pricing), and the system of non-uniform discriminatory delivered pricing (discriminatory pricing withabsorption of transport costs) It is these two systems which are of interest

to us here, although we cannot entirely overlook other pricing systemswhich have been used and analysed For example, a system at the origin ofmany discussions is that of points of parity, where the delivered price isequal to a base price associated with a point in space (the point of parity,

agreed in advance) to which are added (shadow) transport costs lated from this point and not from the point where the seller is located(unless this corresponds to the point of parity) Phlips (1983) presentsthis pricing system in detail It was prohibited in the United States,particularly in the case of cement (see Areeda and Kaplow, 1948) InEurope, in the case of steel, it was adopted by article 60 of the ECCStreaty It was even considered during discussions (between 1981 and 1994)between the German, Belgian and Dutch cement industries and theEuropean Commission This system is generally considered to favourprice collusion (all the producers agree on a single rate for transport)and to be globally inef®cient in terms of location Since transport costspaid by buyers do not correspond to real costs, cross-hauling will gen-erally occur Another price system, also used in the past by the cementindustry in the Benelux countries, is that of uniform delivered prices perzone This system poses similar problems when there is wide geographicaldispersion; it is generally applied in areas with a strong concentration ofbuyers (in cities).

calcu-Let us revert to a theoretical comparison between the two most mon price systems ± mill pricing and discriminatory pricing

com-For a long time most economists considered that non-uniform minatory pricing was preferable because it provided an incentive for morevigorous short-term competition for established ®rms Moreover, undersome symmetry assumptions, this regime results in collectively optimallocations If buyers' reservation prices are high enough for demand to becovered completely, we then have an ef®cient solution Yet this result alsodepends on an implicit assumption of relocation without costs, following

discri-an entry When this assumption is not veri®ed, as is the case in a sectorsuch as the cement industry, it is possible to show that the absence ofdiscrimination becomes socially preferable The reason for this resultderives from the fact that an entry penalises incumbent ®rms far more

in a system of mill pricing and that they have therefore to protect selves by a less concentrated market structure (more ®rms) and, ®nally,lower prices (Norman and Thisse, 1996)

them-The impact of information exchange on prices

In so far as it may lead to collusion, or to agreements between ®rms likely

to limit competition, or to the abuse of a dominant position, the exchange

of information on prices is one of the main targets of anti-trustauthorities The Sherman Act (1890) in the United States served aboveall to prohibit price collusion, as did the application, in Europe, of articles

85 and 86 of the Rome Treaty (1957) We note, however, that information

of this nature is not always exchanged in the same way It may be direct orindirect, organised through announcements or contracts For example, itmay be agreed that a competitor may make public advance announce-ments on price changes (which are not binding), which the other compe-titors may or may not follow.3 Similarly, sales contracts may includeparticular clauses such as the most-favoured customer, which excludesdiscrimination between consumers, or meet-or-release, which guaranteesthe customer the best price compared to other competitors It has beenshown that such practices generate more coordination between ®rms,resulting in less competition This reduction in competition depends to

a large degree on the speci®c practices adopted Thus, if all the practicesjust cited are adopted, and if adjustments as compared to announcedprices are possible at the time of the sale (by granting discounts), then thesolution observed should be a Cournot-type solution The role of dis-counts is to enable ®rms to defend their territory (Holt and Scheffman,1987)

Yet reduced competition and welfare losses, resulting from certaintypes of coordination, may be a short-term effect only The long-termeffect, if we take structural adjustment in the industry into account, may

be positive for the collective welfare by allowing a less concentratedstructure to be maintained

The following example (inspired by the Hotelling model), described in

®gure 1.3, illustrates this possibility (d'Aspremont and Motta, 2000)

We presume that the consumers are uniformly distributed in a straightline Each of them buys a unit of the good if, and only if, the price is lessthan a given price (the reservation price, presumed to be the same foreveryone) There are three possible (equidistant) locations and a ®xed set-

up cost for three potential producers In the ®rst stage, the producersdecide whether or not to set up The transport pricing scheme is presumed

to be that of FOB prices where the customer must pay the factory priceplus the transport cost, with the latter taken to be proportional to thedistance In the second stage no producer may increase its pro®ts byunilaterally changing its price If competition is of the Cournot type(more coordinated), this change is envisaged by considering that thecompetitors will adjust their factory prices to retain their customers

If competition is of the Bertrand type (less coordinated), competitorsare supposed to maintain their factory price at a ®xed level Ex ante, aproducer sets up only if it anticipates a positive pro®t after set-up costs

3 This is standard practice in the cement industry In the United States the industry was sued on this point by the anti-trust authorities but won its cases (cf Dumez and JeunemãÃtre, 2000).

In the example (®gure 1.3), two producers will set up if competition is ofthe Bertrand type (®gure 1.3a) and three will set up if it is of the Cournottype (®gure 1.3b) Moreover, the consumer surplus, measured by the dif-ference between the purchase price and the reservation price (the shadedarea on both ®gures), is greater in the Cournot than in the Bertrand case(owing to the presence of an additional competitor) The same relation-ship is veri®ed for the total surplus (consumer surplus plus sum of pro®ts).

More short-term competition but greater concentration in the long runThe general ideal conveyed by the above results is that it may be prefer-able to have a less concentrated structure (more ®rms) with less intenseprice competition, rather than more intense price competition resulting

in a more concentrated structure (with fewer ®rms) The trade-offbetween the level of short-term competition and the level of concentrationmay then become an issue Now, it is accepted that a high level of

price

Firms’ and consumers’ location axis

Firms’ and consumers’ location axis

Firm A Firm C Firm B

concentration often goes hand in hand with the exercise of increasedmonopoly This idea is corroborated by the empirical studies of Sutton(1991) in the agri-food sector.

As far as the cement industry is concerned, we can certainly interpretthe construction of the European Union as a period of intensi®ed com-petition in the short term, challenging national oligopolies The extent ofmergers and acquisitions in Europe can then be considered as the naturalconsequence of this intensi®cation of competition

The basic question is whether this increase in concentration willnot eventually result in less intense competition We may also wonderwhether a policy limiting competition in the short term by allowing anti-dumping within the Union, for example, would not have had the effect oflimiting purely defensive acquisitions, without hindering the restructuringprocess

In order to consider the terms of this question in more depth, we need

a frame of analysis enabling us to understand the possible forms of theacquisition±restructuring process, and to investigate corresponding gainsand the impact of the competition dynamics in the materialisation of suchgains

Dynamic ef®ciency and selection process in the long run

Let us assume that potential competitors are constantly trying to enter amarket on which only a limited number of ®rms can make a sustainablepro®t Let us consider two questions Does the entry/exit process selectthe most ef®cient ®rms? Does potential competition generate ef®ciency inthe incumbent ®rms? The ®rst question concerns productive ef®ciencywhile the second concerns allocative ef®ciency

Intuitively it may seem that the less rigid the market, the closer toperfect competition the competitive process, the more the answers to theabove questions are likely to be af®rmative This idea is more particu-larly linked to the notion of contestable markets (Baumol, Panzar andWillig, 1982) It is then advisable to encourage everything which maypromote market ¯exibility by eliminating all forms of rigidity However,this type of reasoning is particularly misleading, for two reasons.First, this reasoning is theoretically inaccurate because strong potentialcompetition can lead to high levels of waste This point was initiallydemonstrated by Eaton and Lipsey (1980), who were also the ®rst topropose a formalisation of dynamic competition with free entry In theirmodel, ®rms use their capital investments as entry barriers Once paid for,

a facility has a de®nite life-span known to all, and its use involves zeromarginal costs It is therefore in an incumbent ®rm's interests to renew

its capital well before the facility reaches the end of its life, so that a pricewar in the event of a competitor's entry will be credible Even if theincumbent's pro®ts are whittled away by competitive pressure, the con-sumer does not bene®t from this pressure which is merely a waste ofcapital Maskin and Tirole (1988) considered this question by looking atthe role of capacity as a barrier to entry They show that potentialcompetition not only disciplines the incumbent ®rm but also induces it

to act ef®ciently Steinmetz (1998) shows that when ®rms can choosebetween anticipated renewal resulting in wasted capital on the onehand and reduced prices on the other, it is in their interests to choosewastage These different approaches show that any judgement of produc-tive ef®ciency will strongly depend on the form of competition

Let us illustrate this type of wastage in the cement industry If wecompare the market zones in which a given factory delivers, we notice thatthese areas vary considerably from one country to the next ± for example,they are geographically limited in France but spread out in the UnitedStates As an explanation we may imagine that efforts towards extreme

¯exibility, required by strong short-term competition in the United States,may lead to overinvestment in logistics, including in numerous terminals.Steinmetz' results suggest that this form of capital waste in the UnitedStates is favoured by ®rms, to the detriment of price reductions, whereas

in France weaker short-term competition results in greater logisticef®ciency

Let us now consider the second reason for which a policy aimed atmaximal contestability may be illusory The underlying reasoning may beill-suited to reality, for the concrete nature of rigidity may be partlystructural and its elimination is not always possible In these conditions,

it is certainly preferable to develop a theory of imperfect competitionwhich takes into account the existence of mobility barriers limiting entriesand exits on a market By simplifying, we could then imagine that thestrategy of incumbent ®rms consists of acting on these barriers in order tochoose between immediate pro®ts, of whatever size, on the one hand andrisks of entry, of whatever degree, on the other As for potential competi-tors, one may imagine that their strategies consist of seeking a competi-tive advantage (innovation, enhanced ef®ciency, etc.) enabling them toenter at a lower cost and with greater chances of success.4 It is thenworthwhile to explore those factors which favour the substitution ofmore ef®cient ®rms for less ef®cient ones and, ®nally, promote technicalprogress

4 Scherer (1992) and Geroski (1995) provide a large number of empirical studies to justify this formulation of the problem.