IV. MODELAREA MATEMATICĂ ŞI OPTIMIZAREA ÎN TRANSPORT LOCAL COURNOT–NASH EQUILIBRIA FOR DUOPOLY AND TRIOPOLY WITH LINEAR AND QUADRATIC COSTS V. GORBACHUK, B. CHUMAKOV Cybernetics Institute, National Academy of Sciences of Ukraine, Kyiv GorbachukVasyl@netscape.net Summary. The concept of local Cournot–Nash equilibrium has been introduced. All local Cournot–Nash equilibria for duopoly with general linear and quadratic costs are explicitly presented, and all local Cournot–Nash equilibria for triopoly with general linear costs are explicitly presented. Key words: Cournot, Nash, homogenous product, market, local equilibria, fixed costs, duopoly, triopoly. Let the market price of homogenous product is determined by (1) P = a − b Q, where Q is the quantity of a given product on the market, a , b are some positive parameters (of demand on the product). If the product is going to the market from firms 1 and 2 only, such a market is called a duopoly with Q = q1 + q 2, (2) q where i is the output of product by the firm i = 1, 2 ; if the product is going to the market from firms 1, 2, 3 only, then such a market is called a triopoly with (3) Q = q1 + q 2 + q3, where qi is the output of product by the firm i = 1, 2, 3. Each firm i is maximizing over qi ∈ [0, ∞) its profit (4) π i = P qi − C i ( qi ), where Ci (qi ) is the cost function of firm i . If this function is linear, then (5) Ci (qi ) = ci qi + Fi , where ci > 0 are average per unit variable costs, Fi > 0 are fixed costs of the firm i ; if this function is quadratic, then 233 Ci (qi ) = d i (qi ) 2 + ci qi + Fi , (6) where d i qi + ci are average per unit variable costs, Fi > 0 are fixed costs of the firm i , d i > 0. A duopoly (triopoly) is called heterogeneous if ci ≠ c j and / or Fi ≠ F j and / or d i ≠ d j for some firms i and j ≠ i of this duopoly (triopoly) [1]. Taking into account the participation constraint or entry constraint, the payoff Π i of firm i is equal to π i if π i ≥ 0 ; if π i < 0 and / or qi = 0 , then Π i = 0 (the firm i does not enter the market): π i ≥ 0; ⎧π , (7) Πi = ⎨ i ⎩ 0, (π i < 0) ∨ (qi = 0). A local Cournot–Nash equilibrium for functions Π i , i = 1, 2, 3, is such a C combination of q1 , q 2C , q3C ∈ [0, ∞) , that the inequalities Π1 (q1C , q 2C , q3C ) ≥ Π1 (q1 , q 2C , q3C ) ∀ q1 ∈ [q1C − ε, q1C + ε], Π 2 (q1C , q 2C , q3C ) ≥ Π 2 (q1C , q 2 , q3C ) ∀ q 2 ∈ [q 2C − ε, q 2C + ε], C C Π 3 (q1C , q 2C , q3C ) ≥ Π 2 (q1C , q 2C , q3 ) ∀ q3 ∈ [q3 − ε, q3 + ε]. (8) hold true for some ε > 0 . If the relationships (8) hold true ∀ ε > 0, then a local equilibrium is also the global equilibrium. C C C C Denote q = q1 + q 2 + q3 . If the functions π i , i = 1, 2, 3, are of the form (4), then a Cournot–Nash equilibrium is a solution of the variational inequality 3 ⎡ ⎤ ∂ Ci (qiC ) ∂ P (q C ) C − qi − P (q C )⎥ (qi − qiC ) ≥ 0 ∀ qi ∈ [0, ∞). (9) ⎢ ∂ qi ⎥⎦ i =1 ⎢ ⎣ ∂ qi The relationships (1), (2), (4), (5) imply the value of profit function for firm i = 1, 2 in the duopoly with linear costs S i [a b(q1 q2 )] qi ci qi Fi [a b (q1 q2 ) ci ] qi Fi ; (10) ∑ The relationships (1), (3)–(5) imply the value of profit function for firm i = 1, 2, 3 in the triopoly with linear costs (11) π i = [a − b (q1 + q 2 + q3 ) − ci ] qi − Fi ; the relationships (1), (2), (4), (6) imply the value of profit function for firm i = 1, 2 in the duopoly with quadratic costs (12) π i = [a − b (q1 + q 2 ) − ci ] qi − d i (qi ) 2 − Fi ; 234 the relationships (1), (3), (4), (6) imply the value of profit function for firm i = 1, 2, 3 in the triooply with quadratic costs S i [a b(q1 q 2 q3 )] qi d i (qi ) 2 ci qi Fi . (13) Note the variational inequality (9) does not take into account the fixed costs Fi and therefore does not take into account the participation constraints (7) [2]. Besides, the variational inequality (9) does not take into account the non-negativity condition for price P, determined by (1). Theorem 1. Let the duopoly conditions (1), (2), (4), (5) hold true, and the profit of firm i = 1, 2 is determined by the relationship (10). Assume a − 2 c2 + c1 > 0. a − 2 c1 + c2 > 0 , Then the firm 1, maximizing over q1 ≥ 0 its profit S1 [a b(q1 q 2 )] q1 c1q1 F1 [a b (q1 q 2 ) c1 ] q1 F1, May achieve the only following local maxima of its payoff Π1 : 1a) 2 a − 2 c1 + c2 (a − 2 c1 + c2 ) 2 , if (a − 2 c1 + c2 ) ≥ F1 , − F1 under q1 = 3b 9b 9b (a − 2 c2 + c1 ) 2 ≥ F2 ; 9b 2 a − c1 (a − c1 ) (a − c1 ) 2 1b) , if q 2 = 0 , − F1 under q1 = ≥ F1 ; 2b 4b 4b 1c) 0, if q1 = 0, including the case ⎛ a − c1 − 2 b F1 a − c1 + 2 b F1 q2 ∈ ⎜ , ⎜ b b ⎝ ⎞ ⎟ ∩ [0, ∞) . ⎟ ⎠ The similar takes place for the firm 2. The condition 1c) shows that fixed costs are barriers to entry. Corollary 1. The local maximum of Π 1 over q1 ≥ 0 equals to 0: In the case 1a), if (a − 2 c1 + c2 ) 2 = F1 ; 9b in the case 1b), if (a − c1 ) 2 = F1 ; 4b in the case q1 = 0 . Theorem 2. Under the conditions of theorem 1, all the Cournot–Nash equilibria are given by 4 situations: 235 a − 2 c1 + c2 a − 2 c2 + c1 , q2 = ; 3b 3b a − c1 (a − c1 ) 2 , q 2 = 0 , if 2L2) in the case 1b1) q1 = ≥ F1 ; 2b 4b 2 a − c2 (a − c2 ) ≥ F2 ; 2L3) in the case 1b2) q 2 = , q1 = 0 , if 2b 4b 2L4) in the case q1 = 0 , q 2 = 0 . Theorem 3. Assume that under the conditions of theorem 1 we have a − 2 c1 + c2 ≤ 0 . Then firm 1 maximizing over q1 ≥ 0 its profit S1 [a b(q1 q 2 )] q1 c1q1 F1 [a b (q1 q 2 ) c1 ] q1 F1, May achieve only the local maxima 1b), 1c) of its payoff Π1 . The similar takes place for firm 2. Theorem 4. Under the conditions of theorem 3 all the Cournot–Nash equilibria are given by 3 situations 2L2)–2L4). Lemma 1. c2 < c1 under the conditions of theorem 3. Therefore, if a > c1 , then the both situations 2L2), 2L3) are feasible. If a ≤ c1 , then the situation 2L2) is not feasible. If a ≤ c2 , then the only Cournot–Nash equilibrium is the situation 2L4). Theorem 5. Assume that under the conditions of theorem 1 we have a − 2 c1 + c2 ≤ 0, a − 2 c2 + c1 ≤ 0 . Then the only Cournot–Nash equilibrium is the trivial situation 2L4). Theorem 6. Let the triopoly conditions (1), (3)–(5) hold true, and the profit of firm i = 1, 2, 3 is determined by the relationship (11). Suppose a − 3 c2 + c1 + c3 > 0 , a − 3 c1 + c2 + c3 > 0 , 2L1) in the case 1a) q1 = a − 2 c1 + c2 > 0 , a − 3 c3 + c1 + c2 > 0 , a − 2 c2 + c1 > 0, a − 2 c 2 + c3 > 0 , a − 2 c3 + c1 > 0 , Then the firm 1, maximizing over q1 ≥ 0 its profit S1 [a b(q1 q 2 q3 )] q1 c1q1 F1 a − 2 c1 + c3 > 0 , a − 2 c3 + c 2 > 0 . [a b (q1 q 2 q3 ) c1 ] q1 F1 , May achieve only the following local maxima of its payoff Π1 : 2 6a) (a 3 c1 c2 c3 ) F1 under q1 16 b 236 a 3 c1 c2 c3 , if 4b (a 3 c2 c1 c3 ) 2 (a 3 c1 c 2 c3 ) 2 t F2 , t F1 , 16 b 16 b (a 3 c3 c1 c 2 ) 2 t F3 ; 16 b 6b) a − c1 (a − c1 ) 2 (a − c1 ) 2 , if q 2 = 0 = q3 , − F1 under q1 = ≥ F1; 2b 4b 4b 2 2 a − 2 c1 + c2 , if (a − 2 c1 + c2 ) ≥ F1 , 6c) (a − 2 c1 + c2 ) − F1 under q1 = 3b 9b 9b (a − 2 c2 + c1 ) 2 ≥ F2 , q3 = 0; 9b 2 a − 2 c1 + c3 (a − 2 c1 + c3 ) 2 6d) , if (a − 2 c1 + c3 ) ≥ F1, − F1 under q1 = 3b 9b 9b 2 (a − 2 c3 + c1 ) ≥ F3, q 2 = 0 ; 9b 6e) 0, if q1 = 0 , including ⎛ a − c1 − 2 b F1 a − c1 + 2 b F1 ⎞ ⎟ ∩ [0, ∞). q 2 + q3 ∈ ⎜ , ⎜ ⎟ b b ⎝ ⎠ The similar takes place for firms 2 and 3. Remark 1. Note that case 6e) presumes a duopoly of firms 2 and 3 and therefore applies the theorems 1–5. Therefore, in the case similar to 1a) we have a − 2 c 2 + c3 a − 2 c3 + c 2 q 2 + q3 = + = 3b 3b = 2 a − c2 − c3 ⎛⎜ a − c1 − 2 b F1 a − c1 + 2 b F1 ∈ , ⎜ b b 3b ⎝ ⎞ ⎟ ∩ [0, ∞) . ⎟ ⎠ wherefrom 3 a − 3 c1 − 6 b F1 < 2 a − c2 − c3 < 3 a − 3 c1 + 6 b F1 under a − c1 − 2 b F1 > 0 , 0 ≤ 2 a − c2 − c3 < 3a − 3c1 + 6 b F1 under a − c1 − 2 b F1 ≤ 0 . It means that 237 − 6 b F1 < a − 3 c1 + c2 + c3 < 6 b F1 under a − c1 > 2 b F1 , − 6 b F1 ≤ a − 3 c1 + c2 + c3 under a − c1 ≤ 2 b F1. As the conditions of theorem 6 imply the latter automatically, then under 6e) in the case similar to 1a) we obtain a − 3 c1 + c2 + c3 < 6 b F1 . Under 6c) in the case similar to 1b1) we have q 2 = 0, a − c3 ⎛ a − c1 − 2 b F1 a − c1 + 2 b F1 ⎞ ⎟ ∩ [0, ∞), q3 = ∈⎜ , 2b ⎜ ⎟ b b wherefrom ⎝ ⎠ 2 a − 2c1 − 4 b F1 < a − c3 < 2 a − 2 c1 + 4 b F1 under a − c1 − 2 b F1 > 0, 0 ≤ a − c3 < 2 a − 2 c1 + 4 b F1 under a − c1 − 2 b F1 ≤ 0 . It means that − 4 b F1 < a − 2 c1 + c3 < 4 b F1 under a − c1 > 2 b F1 , − 4 b F1 ≤ a − 2 c1 + c3 under a − c1 ≤ 2 b F1 . As the conditions of theorem 6 imply the latter automatically, then under 6e) in the case similar to 1b1) we have a − 2 c1 + c3 < 4 b F1 . Under 6e) in the case similar to 1b2) we have a − 2 c1 + c 2 < 4 b F1 . Theorem 7. Under the conditions of theorem 6 all the Cournot–Nash equilibria are given by 8 situations: a − 3 c1 + c2 + c3 a − 3 c2 + c1 + c3 7L1) in the case 6a) q1 = , q2 = , 4 b 4 b a − 3 c3 + c1 + c2 q3 = ; 4b a − c1 (a − c1 ) 2 , q 2 = 0 = q3 , if 7L2) in the case 6b1) q1 = ≥ F1; 2b 4b 7L3) in the case 6b2) q 2 = (a − c2 ) 2 a − c2 , q1 = 0 = q3 , if ≥ F2 ; 4b 2b 7L4) in the case 6b3) q3 = a − c3 ( a − c3 ) 2 , q1 = 0 = q 2 , if ≥ F3; 2b 4b 238 7L5) in the case 6c) q1 = a − 2 c1 + c2 a − 2 c2 + c1 , q2 = , q3 = 0 , if 3b 3b (a − 2 c2 + c1 ) 2 (a − 2 c1 + c2 ) 2 ≥ F2; ≥ F1 , 9b 9b 7L6) in the case 6d) q1 = a − 2 c1 + c3 a − 2 c3 + c1 , q3 = , q 2 = 0 , if 3b 3b (a − 2 c3 + c1 ) 2 (a − 2 c1 + c3 ) 2 ≥ F3 ; , ≥ F1 9b 9b 7L7) in the case 6e) q 2 = a − 2 c 2 + c3 a − 2 c3 + c 2 , q3 = , q1 = 0 , if 3b 3b ( a − 2 c3 + c 2 ) 2 ( a − 2 c 2 + c3 ) 2 ≥ F3 ; ≥ F2 , 9b 9b 7L8) in the case q1 = 0 , q 2 = 0 , q3 = 0 . Theorem 8. Assume that under the conditions of theorem 6 we have a − 3 c1 + c2 + c3 ≤ 0 . Then the firm 1 maximizing over q1 ≥ 0 its profit π1 = [a − b (q1 + q 2 + q3 ) − c1 ] q1 − F1 , may achieve only the local maxima 6b)–6e) of its payoff Π1. The similar takes place for firms 2 and 3. Theorem 9. Under the conditions of theorem 8 all the Cournot–Nash equilibria are given by 7 situations 7L2)–7L8). Lemma 2. c3 < c1 , c2 < c1 under the conditions of theorem 8. Therefore, if a > c1, then the situations 7L2)–7L4) are feasible. If a ≤ c1 , then the situation 7L2) is not feasible. If a ≤ c3 , then the situations 7L2), 7L4) are not feasible. Lemma 3. Under the conditions of theorem 6, the inequalities a − 3 c1 + c2 + c3 ≤ 0 , a − 3 c 2 + c1 + c3 ≤ 0 are inconsistent. Proof. The first inequality gives 0 < a − 2 c1 + c3 ≤ c1 − c2 , and the second inequality – 239 0 < a − 2 c2 + c3 ≤ c2 − c1 . Remark 2. For n firms, 7L1) has the form a − n qi + n ∑ q j , i = 1,..., n . j =1, j ≠i qi = (n + 1) b Theorem 11. Let the duopoly conditions (1), (2), (4), (6) hold true, and the profit of firm i = 1, 2 is determined by the relationship (12). Suppose that A12 { 2 (b d 2 )(a c1 ) b (a c2 ) ! 0 , A21 { 2 (b d1 )(a c2 ) b ( a c1 ) ! 0 , B12 { 4 (b d1 )(b d 2 ) b 2. Then firm 1, maximizing over q1 ≥ 0 its profit S1 [a b(q1 q 2 )] q1 d1 (q1 ) 2 c1q1 F1 = [a − b (q1 + q 2 ) − c1 ] q1 − d1 (q1 ) 2 − F1 , may achieve only the following local maxima of its payoff Π1 : 11a) (b d1 )( A12 ) 2 ( B12 ) 2 (b d 2 )( A21 ) 2 ( B12 ) 2 F1 under q1 A12 , B12 if (b d1 )( A12 ) 2 ( B12 ) 2 t F1 , t F2 ; 2 a − c1 (a − c1 ) 2 b 11b) (a − c1 ) b − F1 under q1 = , if q 2 = 0 , ≥ F1; 2 (b + d1 ) 4 (b + d1 ) 2 4 (b + d1 ) 2 11c) 0, if q1 = 0 , including the case ⎛ a − c1 − 2 (b + d1 ) F1 a − c1 + 2 (b + d1 ) F1 q2 ∈ ⎜ , ⎜ b b ⎝ In the case 11a) we obtain S1 F1 ⎞ ⎟ ∩ [0, ∞) . ⎟ ⎠ F1 ( a c1 ) q1 b q1q 2 (b d1 )(q1 ) 2 (a c1 ) A12 B12 b A12 A21 (b d1 )( A12 ) 2 ( B12 ) 2 240 A12 {( a c1 ) B12 b A21} (b d1 )( A12 ) 2 . [4 (b d1 )(b d 2 ) b 2 ]2 Here the nominator is (a c1 ) B12 b A21 F1 4 (a c1 )(b d1 )(b d 2 ) b 2 (a c1 ) 2 b (b d1 )(a c2 ) b 2 ( a c1 ) wherefrom 2 (b d1 )[2 (b d 2 )(a c1 ) b (a c2 )] 2 (b d1 ) A12 , (b d1 )( A12 ) 2 . ( B12 ) 2 The theorem 1 is a particular case of the theorem 11 under d1 = 0 = d 2 . Lemma 3. (a c1 ) B12 b A21 2 (b d1 ) A12 under the conditions of theorem 11. S1 F1 In particular, under d1 = 0 = d 2 we have: B12 A12 2 b (a c1 ) b (a c2 ) , (a c1 ) B12 b A21 A21 3b 2 , 2 b (a c2 ) b (a c1 ), 3 b 2 (a c1 ) 2 b 2 (a c 2 ) b 2 (a c1 ) 4 b 2 (a c1 ) 2 b 2 (a c2 ) 2 b [2 b (a c1 ) b (a c2 )] 2 b A12. The statement, similar to the theorem 11, takes place for firm 2. Theorem 12. Under the conditions of theorem 11, all the Cournot–Nash equilibria are given by 4 situations: A21 A12 , q2 ; 12Q1) in the case 11a) q1 B12 B12 a − c1 (a − c1 ) 2 b 12Q2) in the case 11b1) q1 = , q 2 = 0 , if ≥ F1 ; 2 (b + d1 ) 4 (b + d ) 2 1 a − c2 (a − c2 ) 2 b 12Q3) in the case 11b2) q 2 = , q1 = 0 , if ≥ F2 ; 2 (b + d 2 ) 4 (b + d 2 ) 2 12Q4) in the case q1 = 0 , q 2 = 0 . Theorem 13. Assume that under the conditions of theorem 11 we have A12 d 0 . Then the firm 1, maximizing over q1 ≥ 0 its profit 2 π = [a b (q1 q 2 ) c1 ] q1 d1 (q1 ) F1 , 1 241 may achieve only local maxima 11b), 11c) of its payoff Π1 . The similar takes place for the firm 2. Theorem 14. Under the conditions of theorem 13, all the Cournot–Nash equilibria are given by 3 situations 12Q2)–12Q4). Lemma 4. 2 d 2 (a − c1 ) − 2 d1 (a − c2 ) < b (c1 − c2 ) under the conditions of theorem 13. Therefore, if d1 = d 2 = d , then 0 (b 2 d )(c1 c2 ) , under a > c1 the both situations 12Q2), 12Q3) are feasible. If a ≤ c1 , then the situationт 12Q2) is not feasible. If a ≤ c2 , then the only Cournot–Nash equilibrium is the situation 12Q4). References 1. Gorbachuk V. Cournot–Nash equilibria and Bertrand–Nash equilibria for a heterogeneous duopoly of differentiated products // Cybernetics and systems analysis. – 2010. – Vol. 46. – № 1. – P. 25–33. 2. Gorbachuk V. Monotone operators for Cournot–Nash equilibria search / Generalized convexity and monotonicity. – Cluj-Napoca, Romania: Babes-Bolyai University, 2011. – P. 54–55. 242