The design of appropriate regulatory frameworks for competition policy and merger control is a subject of considerable recent research and of topical political debate, particularly within the European Community but also in North America. This formed the subject of a Paris workshop on `Market Structure and Industrial Organization' on 12/13 May, organized by Patrick Rey, Director of CREST (ENSAE), Paris, a Research Fellow in CEPR's Applied Microeconomics programme. The workshop formed part of CEPR's research initiative on `Market Structure, Industrial Organization and Competition Policy', supported by a SPES grant from the Commission of the European Communities. It was hosted by the Ministère des Finances and co-sponsored by ENSAE, the Association pour la Défense de la Recherche en Economie et Statistiques and the Direction de la Prévision of the Ministère des Finances.

Alain Bayet (ENSAE, Paris) and Fabienne Rosenwald (INSEE, Paris) opened the workshop with `Vertical Restraints, Integration and Barriers to Entry', in which they compared the effects of entry at different levels of a vertical chain and studied the role of bargaining power. For a model with two incumbent firms, a producer and a distributor, each facing potential competition, they showed that the distributor can deter entry more easily by using long-term contracts. The relative bargaining power between the two incumbent firms also affects their ability to deter entry, and a producer has a greater incentive to integrate a distributor than vice versa.

Jean-Jacques Laffont (Université des Sciences Sociales, Toulouse) questioned the robustness of the authors' results to the modelling of competition and pointed out that commitments may not survive renegotiation. He called for a fuller analysis, embodying sunk investments or informational asymmetries as limits to renegotiation. He also asked for empirical evidence to support the authors' intuition.

Olivier Compte (Centre d'Enseignement et de Recherche en Analyse Socio-économique, Paris) presented `Robustness and History Dependence in Repeated Games with Perfect Monitoring', which examined tacit collusion. The framework relaxed the customary assumptions of perfect recall and signalling of publicly observed events. He considered players' deviations from equilibrium path behaviour that derive not from errors or `trembles' but rather from their imperfect observation or treatment of past information. This imposes restrictions on beliefs for strategies off the equilibrium path to which very few equilibria appear robust, and it restricts the set of available strategies to those that depend only on the most recent publicly observed actions. For example, in the prisoners' dilemma, only two equilibria that involve cooperation are robust.

Bruno Jullien (Centre d'Etudes Prospectives d'Economie Mathématique Appliquées à la Planification, Paris) stressed the strong assumption that players never consider whether they may have been mistaken in recall or signal processing. He noted that Compte's results seemed dependent on the ranking of various kinds of errors and asked for examples to indicate a better understanding of the modellers' choices. Finally, the restriction to strategies with limited memory would be useful empirically.

In `Firm Specific Demand Elasticities, Conjectural Variations, the Extent of Marginal Cost Pricing and Competition in Wholesale and Retail Gasoline Markets', joint with Melvyn Fuss, Leonard Waverman (University of Toronto) estimated firms' specific demand elasticities on Canadian gasoline markets during the 1970s. He reported high absolute values in the retail market. These suggest that the firms' behaviour is almost Bertrand competition and that the wholesale market is nearer to perfect contestable pricing than to oligopoly competition.

Bruno Crepon (INSEE, Paris) suggested estimating price and demand on the two markets simultaneously and noted that the assumptions ruled out strategic effects of intra-brand coordination on upstream competition, as highlighted by Rey and Stiglitz. Paul Geroski (London Business School) suggested that taking as constant a marginal cost which was in fact falling might underestimate elasticities, while the use of annual data for a volatile crude oil price posed a major aggregation problem.

In `Anti-trust Policy Towards Cooperative R&D Ventures', Paul Geroski first considered the externalities such ventures may entail in terms of the sharing of costs and risk, the existence of complementary skills or products, and the possible overlap of R&D projects. Technological spillovers may give rise to anti-competitive effects, but they are often found in fields such as electronics in which R&D projects are numerous, which contradicts the conventional wisdom; indeed firms may have incentives to invest in R&D to take advantage of spillovers. Such spillovers may also be overestimated if new technologies appear in different places at the same time, because engineers working in the same field often trade ideas.

Yannis Katsoulacos (Athens University of Economics and Business and CEPR) noted that authorizing cooperation in R&D but not in corresponding product markets creates disincentives. A recent EC study showed that few competitors undertake R&D joint ventures, and such cooperation is dictated mainly by financial considerations (for 90% of firms in the study). Henry Ergas (OECD) also stressed that there are many reasons why cooperation is not immediately efficient; it is paradoxical that EC programmes like Eureka encourage cooperation in R&D while competition authorities forbid it in product markets. Jean-Jacques Laffont suggested distinguishing firms that try to duplicate technologies from those that really invest in R&D.

Michael Riordan (Boston University) and David Salant (GTE Electronics) then presented `Exclusion and Integration in the Market for Video Entertainment', which sought to explain the regulatory system governing the chain that brings video products to the consumer, which involves `producers', `packagers' and `exhibitors'. This sector displays substantial vertical and horizontal integration and many exclusive, long-term contracts. They studied the effects of vertical integration to show that packagers' possession of all the bargaining power increases producers' acquisitions of films and the combined profits of the packager and exhibitor who integrate. This indicates that market foreclosure is more difficult to achieve in an integrated structure.

Bernard Bensaid (Banque de France) suggested extending the model to consider the case where vertical integration need not depend on the packager's bargaining power. Jorge Padilla (Centro de Estudios Monetarios y Financieros, Madrid, and CEPR) argued that the packager's bargaining power will depend on the whether the exhibitor is an entrant or an incumbent and will change once entry is observed.

The second day focused on the theoretical background to and empirical investigations of merger policies. In his joint paper with Paul Klemperer, `Do Firms' Product Lines Include Too Many Varieties, and Do Shops Open Too Many Days?', Jorge Padilla presented a theoretical model to account for the excessive length of product lines. When purchasers concentrate their business on a single supplier because of `switching costs', firms face incentives to extend their product lines beyond the socially optimal length: a firm that offers an extra product captures consumers' business for other products as well, through a `business stealing effect'. This may provide incentives to merge without introducing new products, provided the merger extends the resulting firm's product line. This may damage social welfare both through the risk of foreclosure (if small firms that cannot pay the sunk costs of extending their product lines are forced out of the market), and because prices may rise in less competitive markets.

Anne Perrot (Université de Paris I) stressed the difficulty of measuring the various partial effects in evaluating such practices. Patrick Rey suggested looking at dynamic effects, since there may be further mergers after the first.

In `An Empirical Analysis of Canadian Merger Policy', joint with Daniel Shapiro, Shyam Khemani (University of British Columbia) investigated explicitly whether market shares are the only determinants of merger decisions. He reported estimations of an ordered probit model which suggested that concentration is the most important, but not the only factor considered; import competition and barriers to entry seem to be the most important additional factors.

Kai-Uwe Kühn (Princeton University and CEPR) pointed to some weaknesses of the analysis which resulted essentially from the small sample size. Paul Seabright (Churchill College, Cambridge, and CEPR) suggested testing a two-step decision model, in which market shares are the sole determinants of the decision to block or delay a merger, and other factors could explain the Merger Authorities' final decision.

Damien Neven (Université de Liège, INSEAD and CEPR) and Paul Seabright provided an appraisal of `The Economics and Politics of Merger Control in the European Community' based on their forthcoming book (see pages ???). Their theoretical analysis of merger policies allowed an assessment of the procedures used by the European Community and a comparison with the US Merger Guidelines. The authors identified four areas in which EC merger control practices are inconsistent: EC and member countries' jurisdictions overlap and may even contradict in the cases of joint ventures for which control (single or joint) determines jurisdiction; market definition is mainly based on demand and takes no account of supply substitution; the law governing `efficiency defence', whereby mergers raise productive efficiency or concentration, is unclear and case-histories reveal no clear pattern; and there is no explicit means of monitoring the cases ex post. The authors concluded that the current EC Merger Regulation unduly favours firms that are seeking to merge.

Jean-Louis Colson (Merger Task Force, Commission of the European Communities) stressed the need to take account of the historical background when considering the EC position on issues such as supply substitution and maintained that merger policies will evolve. He also defended EC practices about efficiency defence, while acknowledging one case in which efficiency may not have been taken into account. Concerning ex post control, he argued that each case was unique, and that any remedies would be peculiar to it.

Frédéric Jenny (Conseil de la Concurrence, Paris) opened a Roundtable discussion of `Merger Policies in Europe, North America and Japan' by noting that the French Minister of Finance has absolute discretion to challenge mergers, submit cases to the Conseil de la Concurrence, and finally to follow its recommendations or not, which results in widespread mistrust of merger policy. It is also difficult to identify the basis on which to evaluate a merger's potentially damaging effects, including for example whether a merger should be blocked if one of the firms is going into bankruptcy.

Janusz Ordover (New York University) then discussed the new US Merger Guidelines. Cases in which concentration remains low are not scrutinized, and other factors are taken into account when it is found to be high. The authorities distinguish between the supply and demand sides of the market and seek to categorize participants accordingly.

Shyam Khemani noted that foreign trade and merger policies are linked for the Canadian economy, because it is relatively small. Merger Authorities therefore tend to allow a higher concentration of Canadian firms on the domestic market to enhance their capacity to compete internationally.

Emil Paulis (Merger Task Force, Commission of the European Communities) then presented the EC merger policy, which is not based on published guidelines as found in the US or Canada; rather it evolves by mixing jurisprudence with best practice taken from other developed systems. The Merger Task Force has recently studied oligopolistic dominance, on the grounds that competition may be threatened equally by one or several firms. He stressed the need to identify the competitive relationship between members of an oligopoly and entry barriers, since efficiency gains may warrant a dominant position that deters entry; market definition is also critical in reconciling merger policies with the enforcement of the European single market.

Hiroyuki Odagiri (Tsukuba University) gave an overview of Japanese practice, in which mergers are controlled by a Competition Council, which takes account of the market share and the rank of the post-merger company in deciding whether to scrutinize a merger. The rising number of mergers in the late 1980s raised the Council's profile, but it settled many cases by compromise with the companies involved.

Finally, Henry Ergas highlighted some features of Australian merger policy, for which the relevant legislation is fairly recent (1991), so it is still too early to assess its effects. Some principles nevertheless emerge, such as its focus on markets in non-traded goods and industries with a history of monopolistic competition. Ergas concluded by advocating increased international cooperation by merger authorities.