BIICL The Seventh Annual TransAtlantic Antitrust Dialogue Empirical
BIICL: The Seventh Annual Trans-Atlantic Antitrust Dialogue Empirical Evaluation of Coordinated Effects Dr Peter Davis 1 May 2007
CC Coordinated Effect Cases § Since the Enterprise Act came into effect (June 2003) the CC has considered coordinated effects in three merger inquiries § DS Smith/LINPAC Containers § James Budgett Sugars/Napier Brown Food § Wienerberger Finance Service BV/Baggeridge Brick plc (live case – provisional findings just issued) § Another case, Wiseman/SMD, was referred by the OFT to the CC but was abandoned
Framework for Analysis
Empirical Tests of Collusion versus Competition § Can we use data to test whether market participants are competing or colluding? § Bresnahan (1982, Economics Letters) says YES § But requires a model of behaviour under competition which you test against one under coordination § Idea: Use information on exogenous ‘rotations’ of demand § Subsequently a large economic empirical literature tests whether data were generated from competitive environments or via coordination. § Eg. , (1) Differentiated products and (2) Auction data
Framework for Evaluating Coordination 1. AGREEMENT § Tacit colluders must know what it means to coordinate § Eg hundreds of products and hence prices may be complex to coordinate and require a system for simplification, Eg. , price books or per-mile pricing 2. MONITORING § Aware of behaviour of competitors § Able to spot deviations from prevailing behaviour 3. ENFORCEMENT § Internal stability – main focus of current work § External stability – barriers to supply expansion from non-coordinating firms (Eg. , entry barriers)
Ability to Support Tacit Collusion § § § The theoretical explanation of (tacit) collusion is based on theory of non-coordinated repeated games (dynamic oligopoly models) § Firms meet repeatedly § Firms individual maximise stream of profits § But firms reach a tacit understanding of a mutually beneficial conduct Many models but generic one says a firm is willing to collude if NPV of payoffs to tacit collusion > payoffs to cheating + NPV of payoffs under ‘punishment’ + NPV of payoffs under competition Collusion and Competition payoffs normally calculated in a unilateral effects simulation model
Internal Stability § To sustain tacit collusion, each firm must find it: 1. sufficiently rewarding to tacitly collude 2. insufficiently rewarding to cheat when others are coordinating 3. sufficiently costly to cheat and then get caught § We can model each element of the incentive to collude 1. Return to collusion 2. Return to cheating 3. Costs of cheating § And putting them all together will tell us about one element of the ability to collude: § A collection of firms will be able to sustain collusion only when they each find it worthwhile
The Return to, or incentive for, Tacit Collusion (1) Unilateral effect of merger A to B. Is B>A? (2) Coordinated effect of merger • Strengthening of coordination: C to D • Newly coordinate: A to D Equilibrium Outcome A Competitive outcome A = pre-merger B = post-merger B C (prices D Monopoly (perfect cartel) outcome or profits)
Incentives For Coordination - If you can coordinate, what does it do for you? § Impact of merger § If strengthening coordination = D–C § If results in newly coordinating = D–A § Profit Incentive to Newly Coordinate § Pre-merger = C–A § Post-merger = D–B § Notice we currently sometimes get A and B from unilateral effects simulation models
Ability to Collude Results from simulation models • Some mergers will reduce or even entirely remove firms ability to ‘fully’ collude • These tend to be mergers that generate additional asymmetry • If partial collusion is possible then less dramatic effect – continuum between competition and collusive outcomes using the maximal sustainable profit level
Two different 3 to 2 mergers with 6 products, eg (4, 1, 1) to (5, 1) merger D (5, 1) B (5, 1) C A (4, 1, 1) An Example of Incentives to Collude. NB: discount factor is in each of the ‘NPV’ calculations
Conclusions § Mergers that introduce asymmetries in market structures can hinder coordination § Pre-merger Coordination can be explicitly tested for against Competition § We can attempt to measure the impact of a merger on: 1. The three elements that drive the internal incentives to coordinate 2. The ability of a market structure to support coordination § However, as with unilateral effects simulation, formal empirical analysis requires that we settle on a particular model of competition and also a particular model of collusion for analysis
- Slides: 12