SØK/ECON 535 Imperfect Competition and Strategic Interaction ENTRY AND EXIT Lecture notes 09.10.02 Introduction In the absence of entry barriers firms cannot make supernormal profits. Barriers to entry government regulations technological barriers economies of scale access to (efficient) production technologies, inputs etc. product differentiation, consumer switching costs, patents strategic barriers Bain (1956) blockaded entry deterred entry accommodated entry Contestable markets Potential (as opposed to real) competition (Baumol et al, 1986) sunk costs hit and run strategies Market characterisation Consider an example in which firms produce one homogenous products may be generalised to multi-product firms and differentiated products All firms have access to the same technology 0 when q = 0 C (q ) = > 0otherwise E Demand function: Q = Q ( p ) . Market configuration: {q1, q2 ,..., qn , p} A market configuration is feasible if Q ( p ) = ∑ i =1 qi n pqi − C ( qi ) ≥ 0, i = 1,2,..., n A market configuration is sustainable if ( ) e e e there does not exist p ≤ p and q ≤ Q p e ( ) e e e such that p q − C q ≥ 0 , e where p and q are the entrants price and quantity. A perfectly contestable market is market for which all market configurations are feasible and sustainable. Example: natural monopoly C ( q ) = f + cq, q > 0 figure (sustainable market configuration where the demand curve cuts the average-cost curve) The sustainable market configuration is cost efficient and second-best (in the Ramsey sense). Generalisation to multi-product markets total costs are minimised no firm earns positive profit no cross-subsidisation Ramsey conditions are fulfilled if the number of firms exceeds one price equals marginal cost Pareto optimality Note: distinguish between fixed and sunk costs barriers to entry are associated with sunk costs Hit and run Consider an example with a single firm in the market (natural monopoly). 2 A potential entrant can offer the market a contract at time 0 that specifies quantity q price p, and duration τ . The duration of contract should be thought of as the time it takes the established firm to react (i.e. change its price offer). Let β be the unit price of capital at time 0 α be the unit price of capital at time τ e The user cost of capital µ may be defined implicitly by β − αe − rτ τ = ∫ µ ee − rt dt ⇒ µ e = β r + r [β − α ] e rτ − 1 0 The established firm may consequently hold a price that exceeds its own user cost of capital ( β r ). Figure ( e ( C q, µ ( ) q ) and C ( q, β r ) q . pL implicitly defined by pLQ ( pL ) − C Q ( pL ) , µ e = 0 ; that is, where the demand curve cuts the average-cost curve of the potential firm). Conclusion: the market is perfectly contestable if entry is free firms face the same price of capital ( β ) firms have access to the same technology ( C , Q ) exit is costless [β −α ] e rτ − 1 = 0 Strategic entry barriers Credible entry deterrence An entry game between a (potential) entrant (e) and an incumbent firm (i): Stage 1: entrant decides whether or not to enter the market; Stage 2: incumbent (having observed the entrant’s choice) decides whether or not to ‘fight’ the entrant; 3 ( 0 0 payoffs: ( Π i , Π e ) if entry is met by ‘fight’; Π i , Π e ‘fight’; and ( Π m ,0 ) if there is no entry. ) if entry is not met by Results 0 the strategy ‘fight’ is credible only if Π i ≥ Π i ; ‘fight’ leads to deterrence only if Π e ≤ 0 ; 0 if Π e ≤ 0 and Π i < Π i , then the strategy combination {fight, stay out} is a Nash equilibrium but not a Subgame Perfect Equilibrium. Note: it is the ex post market situation that is relevant. ‘Limit-pricing’ ‘classic theory’: Bain, Sylos-Labini, Modigliani game-theoretic critique: Spence, Dixit asymmetric information and signalling: Milgrom and Roberts A strategic entry game We consider a new game in which the incumbent may take actions ex ante in order to deter entry; subsequently a game similar to the one above is played (this is summarised in the reduced-form profit functions): Stage 1: incumbent decides on level of entry barrier b ( b = 0 corresponds to no barrier); Stage 2: entrant decides whether or not to enter; payoffs: ( Π i ( b ) , Π e ( b ) ) if entry takes place and ( Π m ( b ) ,0 ) otherwise. Definitions entry is ‘blockaded’ if Π e ( 0 ) < 0 ; deterrence possible if there exists b such that Π e ( b ) < 0 ( ) ( ) * * deterrence profitable if Π m b ≥ Π i ( 0 ) , where Π e b = 0 ; ( ) * accommodation if Π m b < Π i ( 0 ) . Issues b must be irreversible (cannot be changed upon entry) what forms can b take?; how does different forms of position; b affect potential entrants’ competitive asymmetric information; oligopoly. 4 The Spence-Dixit model b = k is capacity. Homogenous product, linear demand: p = 1 − Q . Costs ex ante (before investment): C ( q, k ) = ck + f , q ≤ k ex post (after investment): C ( q, k ) = 0, q ≤ k . After entry has taken place, firm i choose quantity simultaneously with firm e choosing both quantity and capacity. The ex post duopoly game (given entry and k i ): clearly not profitable for firm e to hold excess capacity, hence qe = ke ; { } * * * * reaction functions: qi = min ki , 1 − ke 2 and ke = 1 − qi − c 2 * firm e’s profit: 1 − qi − c 2 4−f . Note that deterrence * requires qi ≥ 1 − c − 2 f credible if ki ≥ 1 − c − 2 f and [1 + c ] 3 ≥ 1 − c − 2 f , i.e. c ≥ 21 1 − 3 f . The ex post subgame if firm i remains a monopolist: { } max qi [1 − qi ] qi qi ≤ ki ; * at equilibrium we must have qi = ki since marginal revenue is greater for a monopolist than for a duopolist and excess capacity cannot be optimal. Consider then the first-stage choice of firm i and suppose first that the choice involves deterrence: { max [1 − ki − c ] k i − f k i ≥ 1 − c − 2 f } ⇒k i { } = max [1 − c ] 2,1 − c − 2 f ; blockaded entry if [1 − c ] 2 ≥ 1 − c − 2 f ,or c ≥ 1 − 4 f ; deterrence if [1 − c ] 2 < 1 − c − 2 f , yielding profits 2 f 1 − c − 2 f − f . Consider next the first-stage choice in the case of accommodation: excess capacity clearly unprofitable; { max 1 − ki − k3 ( ki ) − c ki − f } where k e ( ki ) = [1 − ki − c ] 2 5 result is k i = [1 − c ] 2 and profits [1 − c ] 8 − f 2 It follows that firm i chooses deterrence rather than accommodation if 2 2 f 1 − c − 2 f > [1 − c ] 8 , or c ≤ 1 − 4 f 2 − 2 , and vice versa. Numerical example: f = 0.01: entry blockaded if c ≥ 0.6 ; deterrence credible if c ≥ 0.35 ; deterrence blockaded if c ≤ 0.77 . Conclusion entry deterrence by over-investment in capacity; no excess capacity; strategic incentive for over-investment also in the case of accommodation (cf Stackelberg); price is low due to large capacity, but no ‘limit pricing’. Variations Price competition at the post-entry stage. Multiple incumbents deterrence is a ‘public good’; price > marginal costs; Gilbert and Vives (1986) show that second effect dominates and hence deterrence is even more likely with many incumbents. Strategic types and types of strategies Consider the case in which the incumbent accommodates entry Profit functions incumbent: Π i ( qi , qe , k i ) entrant: Π e ( qi , qe , ke ) First-order condition for incumbent’s ex ante decision 0= d Π i ∂Π i dqe dqi ∂π i = + ∂qe dqi dk i ∂k i dki 6 The first element on the right-hand side constitutes the ‘strategic effect’, and so (assuming sign ( ∂Π i ∂qe ) = sign ( ∂Π e ∂qi ) ) ∂Π e dqi dqe sign ( strat .eff .) = sign sign ∂qi dk i dqi 2 2 Around the ‘open loop’ point ( ∂Π i ∂k i = 0 ) we have ∂ Π i ∂ki < 0 . Consequently, a negative (resp. positive) strategic effect will lead to overinvestment (resp. under-investment). Definitions ∂Π e dqi Firm i is tough (resp. soft) if sign < 0 (resp. > 0 ); ∂qi dk i Strategies are strategic substitutes (resp. complements) if reaction dq functions are increasing (resp. decreasing), or sign e > 0 (resp. < 0 ). dqi Ex ante investment makes the incumbent firm Strategic complements Strategic substitutes Though Soft Puppy Dog Fat Cat Top Dog Lean and Hungry Lool Entry deterrence: only toughness/softness determines strategy (effect on competitor’s profits). Entry accommodation: type of strategies matters also (effect on own profit). Quantity competition: Top Dog in both cases. Price competition: Top Dog if deterrence, Puppy Dog if accommodation. Examples: generic advertising and price competition: excessive advertising when accommodating entry (Fat Cat); brand advertising and quantity competition: excessive to deter entry (Top Dog); location (product differentiation): The Principle of Differentiation (Puppy Dog); learning-by-doing: excessive production to deter entry (Top Dog); 7 most-favoured-customer-clause with price competition (Puppy Dog) Exit Industries in decline fight to survive. War of attrition Two symmetric firms supplying a homogenous good. Production cost per unit of time: C ( q ) = f + cq Bertrand competition: p1 = p2 = c m m Monopoly gross profits (not including fixed costs): Π , Π > f At any point in time, each firm may choose to leave the market (receives 0 profits afterwards – cannot re-enter). We are looking for a symmetric equilibrium (in mixed strategies). Let x ( t ) be the probability that a firm leaves the market at time t. Then the other firm will be indifferent between staying in an leaving over the next period of time dt if −f d t + x d t Πm − f fr =0 ⇒ x = m . r Π −f Conclusion before exist happens, both firms run deficits (over a period with stochastic length); expected profits are zero (ex post the remaining firm earns positive profits); welfare loss both before (due to duplication of fixed costs) and after exit (because of monopoly pricing); in the a case of price collusion (maybe tacit) the exit rate would be lower, and hence welfare would be lower also (due to both higher price and more duplication ex ante) Will small firms exist before large firms? If contraction is not an option, large firms will leave first (Ghemawat and Nalebuff, 1985). 8 If contraction is possible, large firms will contract before exit takes place (Whinston, 1986, and Ghemawat and Nalebuff, 1987). 9