Strategic commitment Relaxing competition Pricing behavior Geir B. Asheim Market power Strategic commitment Relaxing competition Strategic commitment to affect future competition Another topic: Repeated interaction Product differentiation Capacity constraints (Cournot oligopoly model) Are two firms are sufficient for a perfectly competitive outcome? Competition can be relaxed by Pricing behavior by two (or more) firm with identical and constant unit costs (Bertrand oligopoly model) Pricing behavior by a monopoly firm To be covered under Market power & product differentiation ECON5200 Fall 2009 Department of Economics, University of Oslo Geir B. Asheim Applications of game theory 1 Market power & product differentiation Strategic commitment Relaxing competition Monopoly Duopoly Pricing behavior Geir B. Asheim Market power Strategic commitment Relaxing competition Pricing behavior Geir B. Asheim Geir B. Asheim Pricing behavior Market power Market power with equality if q m > 0 Other distortions; investment in cost reductions & quality choice. If q m > 0, then p(q m ) > c (q m ): Price under monopoly exceeds marginal cost; this leads to quantity distortion and welfare loss. p (q m )q m + p(q m ) ≤ c (q m ) First-order condition: q max p(q)q − c(q) p(·) = x −1 (·): inverse demand function; with p(0) = p̄. p max px(p) − c(x(p)) x(·): continuous and strictly decreasing demand function for all p with x(p) > 0; x(p) = 0 for p ≥ p̄ Pricing behavior by a profit-maximizing monopolist 4 November Seminar. 27 October Principal-agent problems. 20 October Adverse selection, signaling and screening. 13 October Market power & product differentiation. 3 lectures where game theory is applied: Introduction Strategic commitment Capacity competition Product differentiation Repeated interaction Relaxing competition Pricing behavior Geir B. Asheim Market power Strategic commitment Relaxing competition Monopoly Duopoly Not a static game, but repeated interaction Not homogeneous products, but product differentiation Not no capacity contraints, but capacity constraints (Cournot oligopoly model) Competition can be relaxed by Firms set prices. But the conclusion of the Bertrand model — that two firms are sufficient for perfectly competitive outcomes – is unrealistic. Relaxing competition There is a unique Nash equilibrium (p1∗ , p2∗ ) in the Bertrand duopoly model. In this equilibrium, both firms set their prices equal to cost: p1∗ = p2∗ = c. Proposition Strategic commitment Capacity competition Product differentiation Repeated interaction Relaxing competition Pricing behavior Geir B. Asheim Market power Strategic commitment Relaxing competition Monopoly Duopoly Pricing behavior Geir B. Asheim x(·): continuous and strictly decreasing demand function for all p with x(p) > 0; x(p) = 0 for p ≥ p̄. Firms 1 and 2 have constant unit cost c and set prices p1 and p2 simultaneously. Sales are given by: ⎧ ⎪ if pj < pk ⎨ x(pj ) 1 xj (pj , pk ) = if pj = pk 2 x(pj ) ⎪ ⎩ 0 if pj > pk Geir B. Asheim Pricing behavior Market power Structure: (1) Prices set (2) Demand determines quantities Market power Pricing behavior by profit-maximizing duopolists Bertrand model with homogeneous products (1) with equality if qj > 0 For each q̄k , let bj (q̄k ) denote j’s best response. p (qj + q̄k )qj + p(qj + q̄k ) ≤ c First-order condition: qj max p(qj + q̄k )qj − cqj Assume firms first choose quantities q1 and q2 and that these are sold at price p(q1 + q2 ) in the market. Each firm j’s problem: Structure: (1) Quantities set (2) Demand determines price Quantity choices by profit-maximizing duopolists Cournot duopoly model (1) Conclusion: If J firms have constant unit cost c, then all firms set price equal to c, provided J ≥ 2. Part 2: (p1 , p2 ) is not a Nash equil. if (p1 , p2 ) = (c, c). Case 1: min{p1 , p2 } < c. At least one firm j has negative profit. Zero profit by setting pj ≥ 0. Case 2: pk > pj = c. Firm j can increase its profit by setting pj = 12 (pk + c). Case 3: pk ≥ pj > c. Firm k can increase its profit by setting pk = pj − ε, for ε > 0 sufficiently small. Part 1: (p1∗ , p2∗ ) = (c, c) is a Nash equilibrium. Assume that pk = c. If firm j chooses pj = c, then profit equals 0. If firm j chooses pj < c, then profit is negative. If firm j chooses pj > c, then profit is 0. Therefore, pj = c is best response to pk = c. Proof. Pricing behavior by profit-maximizing duopolists Bertrand model with homogeneous products (2) Strategic commitment Capacity competition Product differentiation Repeated interaction Relaxing competition Pricing behavior Geir B. Asheim Market power Strategic commitment Capacity competition Product differentiation Repeated interaction Relaxing competition Pricing behavior Geir B. Asheim Market power + + q2∗ )q1∗ q2∗ )q2∗ + + p(q1∗ p(q1∗ + + q2∗ ) q2∗ ) with equality if with equality if ≤c ≤c q1∗ q2∗ >0 >0 Proportional rationing: Any consumer willing to pay a good offered at p has an equal change of buying at p. Efficient rationing: A consumer with the highest willingness-to-pay is served first. With efficient rationing, the game considered by Kreps & Scheinkman (1983) leads to the Cournot outcome. Proposition 2 1 How are consumers rationed? In the subgames starting at stage (3), all possible profiles of capacities and prices must be considered. Structure: (1) Capacities set (2) Prices set (3) Demand determines quantities Capacity choices followed by price competition Kreps & Scheinkman (1983) In a Nash equilibrium of the Cournot duopoly model with p(0) > c > 0 and p (q) < 0 for all q > 0, the market price is greater than c (the competitive price) and smaller than p(q m ). Proposition If p(0) > c, then (q1∗ , q2∗ ) 0, implying that ∗ ∗ q +q p (q1∗ + q2∗ ) 1 2 2 + p(q1∗ + q2∗ ) = c p (q1∗ p (q1∗ (q1∗ , q2∗ ) is a Nash equilibrium if and only if for each j, qj∗ ∈ bj (qk∗ ). Hence, if (q1∗ , q2∗ ) is a Nash equilibrium, then Quantity choices by profit-maximizing duopolists Cournot duopoly model (2) Strategic commitment Capacity competition Product differentiation Repeated interaction Relaxing competition Pricing behavior Geir B. Asheim Market power Strategic commitment Capacity competition Product differentiation Repeated interaction Relaxing competition Pricing behavior Geir B. Asheim Market power Not covered How do firms product differentiate to relax price competition? Bertrand model where demanded quantity is a continuous function of the price profile Horizontal differentiation How does product differentiation affect price competition? Two issues Part 2: p(q1∗ + q2∗ ) < p(q m ) in a Nash equilibrium. Since p (q) < 0 for all q > 0, this is equivalent to q1∗ + q2∗ > q m . Subpart (i): q1∗ + q2∗ ≥ q m . Suppose q1∗ + q2∗ < q m . If one firm j increases quantity to q m − qk∗ , then total profit is increased and the profit of the other firm is decreased. Hence, firm j can do better. Subpart (ii): q1∗ + q2∗ = q m . Suppose q1∗ + q2∗ = q m . The FOCs for monopoly and Cournot duopoly cannot both be satisfied. Part 1: p(q1∗ + q2∗ ) > c in a Nash equilibrium. Follows from the FOC and the assumption that p (q) < 0. Proof. Quantity choices by profit-maximizing duopolists Cournot duopoly model (3) Strategic commitment Capacity competition Product differentiation Repeated interaction Relaxing competition Pricing behavior Geir B. Asheim Market power Strategic commitment Capacity competition Product differentiation Repeated interaction Relaxing competition Pricing behavior Geir B. Asheim Market power ⎪ ⎩ and 0 (t+pk −pj )M 2t ⎧ ⎪ ⎨M t + p2 − p1 2t xj (pj , pk ) = ẑ = t + p1 − p2 2t if pj ∈ (pk − t, pk + t) if pj ≥ pk + t if pj ≤ pk − t 1 − ẑ = Assume that M consumers are distributed uniformly along the unit inverval: [0, 1]. Also, assume that firm 1 is located at 0 and firm 2 is located at 1. For a consumer located at z (∈ [0, 1]) the cost to buy from firm 1 is p1 + tz and the cost to buy from firm 2 is p2 + t(1 − z). Finally, assume that every consumer buys one (and only one) unit from a firm with lowest cost. Indifferent consumer ẑ: p1 + tẑ = p2 + t(1 − ẑ), which implies Horizontal product differentiation Linear city with linear transportation costs (1) If x1 (p1 , p2 ) > 0 when p1 ≤ min{c, p2 } and x2 (p2 , p1 ) > 0 when p2 ≤ min{c, p1 }, then in any Nash equilibrium (p1∗ , p2∗ ) we have that (p1∗ , p2∗ ) (c, c). Proposition pj max(pj − c)x(pj , pk ) Firms 1 and 2 have constant unit cost c and set prices p1 and p2 simultaneously. Sales are given by xj (pj , pk ), which is the continuous demand function for firm j. Structure: (1) Prices set (2) Demand determines quantities Pricing behavior by profit-maximizing duopolists Bertrand model with differentiated products (1) Strategic commitment Capacity competition Product differentiation Repeated interaction Relaxing competition Pricing behavior Geir B. Asheim Market power Strategic commitment Capacity competition Product differentiation Repeated interaction Relaxing competition Pricing behavior Geir B. Asheim Market power (t + pk − pj )M 2t What product differentiation maximizes welfare? Do firms maximize product differentiate to relax price competition? Problematic to discuss choice of product differentiation with linear transportation costs. Why? Unique Nash equilibrium, (p1∗ , p2∗ ) satisfies: p1∗ = p2∗ = c + t. Hence, price exceeds c. FOC if pj ∈ (pk − t, pk + t): t + pk + c − 2pj = 0 ⎧ ⎪ if pk ≥ c + 3t ⎨ p̄k − t t+pk +c bj (p̄k ) = if p̄k ∈ (c − t, c + 3t) 2 ⎪ ⎩ p̄k + t if p̄k ≤ c − t (pj − c) pj ∈[pk −t,pk +t] max Horizontal product differentiation Linear city with linear transportation costs (2) Case 2: min{p1 , p2 } = c. If pj = c and pk ≥ c, then by assumption xj (pj , pk ) > 0, and firm j’s profit is zero. Since xj (·, ·) is continuous, firm j can attain positive profit by setting pj = c + ε for ε > 0 sufficiently small. Case 1: min{p1 , p2 } < c. At least one firm j has negative profit. Non-negative profit by setting pj ≥ 0. To be shown: (p1 , p2 ) is not a Nash equilibrium if min{p1 , p2 } ≤ c. Proof. Pricing behavior by profit-maximizing duopolists Bertrand model with differentiated products (2) Strategic commitment Relaxing competition Pricing behavior Assume sufficient differentiability. Assume unique Nash equilibrium: (s1∗ (k), s2∗ (k)). Best response functions: b1 (s2 , k) and b2 (s1 ). Payoffs: π1 (s1 , s2 , k) and π2 (s1 , s2 ) Actions are “aggressive”: ∂π1 (s1 , s2 , k)/∂s2 < 0 and ∂π2 (s1 , s2 )/∂s1 < 0. Strategic commitment Relaxing competition Pricing behavior Geir B. Asheim Structure: (1) Firm 1 makes strategic investment k. Observable. (2) Firms 1 and 2 play some oligopoly game, choosing s1 and s2 Geir B. Asheim Strategic commitment Capacity competition Product differentiation Repeated interaction Relaxing competition Pricing behavior Geir B. Asheim Market power Market power Strategic commitment to affect future competition (1) The simple strategy profile σ((p(0)t ), (p(1)t ), (p(2)t )) is a subgame-perfect equilibrium if and only if δ ≥ 12 . Proposition (p(2)t ) = (c, c), (c, c), . . . (p(1)t ) = (c, c), (c, c), . . . (p(0) ) = (p, p), (p, p), . . . t Consider the following paths, where p ∈ (c, p m ]: Firms 1 and 2 have constant unit cost c. In each stage, firms choose prices simultaneously. The chosen prices determine quantities in this stage and becomes observable for both firms before the next stage. Infinite number of stages. Payoff is profits discounted at constant discount factor δ ∈ (0, 1). Market power Strategic commitment Capacity competition Product differentiation Repeated interaction Relaxing competition Pricing behavior Geir B. Asheim Market power Repeated Bertrand duopoly (1) 1 2 1 (p − c)x(p) Payoff without deviation 2 or, equivalently, δ ≥ 12 . ≤ db2 = ds1 1 db2 ∂b1 dk ds1 ∂k ∂b1 ∂k 2 ∂b1 − db ds1 ∂s2 ds2∗ = ∂b1 ∗ ∂b1 dk ds + ∂s2 2 ∂k db2 ds2∗ = dk ds1 db2 ∂b1 1− ds1 ∂s2 ds2∗ Identity: s2∗ = b2 (b1 (s2∗ , k)) Strategic commitment to affect future competition (2) What if there are more than 2 firms? What if prices are not perfectly observable? Is a unilateral one-period deviation from (c, c) profitable? Payoff with deviation is non-positive. Payoff without deviation equals zero. Not profitable if 1 − δ ≤ Supremum of payoff with deviation (1 − δ)(p − c)x(p) Is a unilateral one-period deviation from (p, p) profitable? Enough to consider unilateral one-period deviations. Why? Proof. Repeated Bertrand duopoly (2) Strategic commitment Relaxing competition Pricing behavior Geir B. Asheim Market power Strategic commitment Relaxing competition Pricing behavior Geir B. Asheim Market power = <0 ∂b1 (·) ∂k ∂b1 ∂k db2 ∂b1 1− ds ∂s 1 2 >0 ∂b1 (·) ∂k db2 ds1 <0 >0 ds2∗ (k) dk >0 <0 ∂b1 (·) ∂k ∂b1 (·) ∂k date Entry db2 (·) ds1 >0 > 0 Δk < 0 Puppy Dog ds2∗ (k) dk db2 (·) ds1 > 0 Δk < 0 ds2∗ (k) dk < 0 Δk > 0 Fat Cat Mean & Hungry Look small & firm to look fat & mellow to look tough & aggressive soft & inoffensive ds2∗ (k) dk big & strong to look small & weak to look tough & aggressive soft & inoffensive < 0 Δk > 0 Top Dog <0 Strategic substitutes: ds2∗ (k) dk To Accommo- <0 >0 Strategic complements: ds2∗ (k) dk ds2∗ (k) dk >0 db2 (·) ds1 db2 (·) ds1 <0 Strategic complements: Strategic substitutes: ds2∗ (k) dk Strategic commitment to affect future competition (5) ds2∗ dk Strategic commitment to affect future competition (3) Strategic commitment Relaxing competition Pricing behavior Geir B. Asheim Market power ∂b1 (·) ∂k ∂b1 (·) ∂k Entry <0 >0 To Deter <0 >0 > 0 Δk > 0 Top Dog ds2∗ (k) dk db2 (·) ds1 Strategic complements: > 0 Δk < 0 ds2∗ (k) dk < 0 Δk < 0 Mean & Hungry Look Mean & Hungry Look small & firm to look small & firm to look tough & aggressive tough & aggressive ds2∗ (k) dk big & strong to look big & strong to look tough & aggressive tough & aggressive < 0 Δk > 0 Top Dog ds2∗ (k) dk db2 (·) ds1 Strategic substitutes: Strategic commitment to affect future competition (4)