Countervailing Power Hypothesis and Waterbed E ects

Transcription

1 Countervailing Power Hypothesis and Waterbed E ects Özlem Bedre-Defolie* Greg Sha er** *European School of Management and Technology ** University of Rochester IIOC, Vancouver 16 May, 2010 Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 1 / 15

2 Buyer power is a hot debate Buyer power is the bargaining strength of retailers vis-à-vis their suppliers. (Inderst and Mazzarotto 08) Detailed antitrust studies: The FTC reports (01,03) in the US, the EC report (99) in EU and the CC reports (00, 08) in the UK. The primitives of buyer power: size (Katz 87, She man&spiller 92, Snyder 96), suppliers production technology (Chipty&Snyder 99, Inderst&Wey 03), gatekeeping position (Inderst&Wey 07, Bedre-Defolie&Caprice 10). The e ects of an increase in buyer power on consumers and on other (weak/small) retailers while taking buyer power as given (Chen 03, this paper), relating it to size (Majumdar 06, Inderst&Valletti 09a, Inderst&Wey 03,07, Inderst 07). The latter is particularly important for policy makers, e.g. merger cases: Kesko/Tuko 97, Rewe/Meinl 99, Carrefour/Promodes 00 in Europe. Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 2 / 15

3 Countervailing power hypothesis Galbraith 52: The exercise of buyer power lowers purchasing costs of retailers, and thus lowers retail prices. Industry Structure Supply Contract Buyer power Bilateral monopoly linear # wholesale price. Downstream competition linear # wholesale price. Bilateral monopoly two-part tari no e ect (Stigler 54). Downstream competition two-part tari # unit price or xed fee? Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 3 / 15

4 Countervailing power hypothesis Galbraith 52: The exercise of buyer power lowers purchasing costs of retailers, and thus lowers retail prices. Industry Structure Supply Contract Buyer power Bilateral monopoly linear # wholesale price. Downstream competition linear # wholesale price. Bilateral monopoly two-part tari no e ect (Stigler 54). Downstream competition two-part tari # unit price or xed fee? More complex situation if downstream rms are asymmetric: possible "waterbed e ect". Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 3 / 15

5 Countervailing power hypothesis Galbraith 52: The exercise of buyer power lowers purchasing costs of retailers, and thus lowers retail prices. Industry Structure Supply Contract Buyer power Bilateral monopoly linear # wholesale price. Downstream competition linear # wholesale price. Bilateral monopoly two-part tari no e ect (Stigler 54). Downstream competition two-part tari # unit price or xed fee? More complex situation if downstream rms are asymmetric: possible "waterbed e ect". Few papers look at asymmetric downstream rms, and of those that do, contracts are assumed to be linear. (Exceptions: Chen 03, Bedre-Defolie and Caprice 08, 09). Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 3 / 15

6 This paper analyzes how buyer power a ects consumers and rival retailers in a setup developed on Chen 03. Results are di erent from Chen: Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 4 / 15

7 This paper analyzes how buyer power a ects consumers and rival retailers in a setup developed on Chen 03. Results are di erent from Chen: The wholesale price of the dominant retailer, in general, depend on the wholesale price received by the fringe rms. Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 4 / 15

8 This paper analyzes how buyer power a ects consumers and rival retailers in a setup developed on Chen 03. Results are di erent from Chen: The wholesale price of the dominant retailer, in general, depend on the wholesale price received by the fringe rms. The pass-through rate of the fringe rms wholesale price on the retail price is critical for the results. Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 4 / 15

9 This paper analyzes how buyer power a ects consumers and rival retailers in a setup developed on Chen 03. Results are di erent from Chen: The wholesale price of the dominant retailer, in general, depend on the wholesale price received by the fringe rms. The pass-through rate of the fringe rms wholesale price on the retail price is critical for the results. When the pass-through rate is between 0 and 1, an increase in the dominant retailer s buyer power is always good for consumers and leads to lower wholesale prices for the fringe. (parallel to Chen s ndings) Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 4 / 15

10 This paper analyzes how buyer power a ects consumers and rival retailers in a setup developed on Chen 03. Results are di erent from Chen: The wholesale price of the dominant retailer, in general, depend on the wholesale price received by the fringe rms. The pass-through rate of the fringe rms wholesale price on the retail price is critical for the results. When the pass-through rate is between 0 and 1, an increase in the dominant retailer s buyer power is always good for consumers and leads to lower wholesale prices for the fringe. (parallel to Chen s ndings) When the pass-through rate is above 1, an increase in buyer power is always bad for consumers and leads to higher wholesale prices for the fringe (waterbed e ects under some conditions). Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 4 / 15

11 The Framework Similar to Chen (2003): 1 The supplier makes a t.i.o.l.i. o er, (w f, F f ), to the fringe rms. If the fringe rms accept the o er, each pays F f. Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 5 / 15

12 The Framework Similar to Chen (2003): 1 The supplier makes a t.i.o.l.i. o er, (w f, F f ), to the fringe rms. If the fringe rms accept the o er, each pays F f. 2 The supplier and the dominant retailer negotiate (w d, F d ). Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 5 / 15

13 The Framework Similar to Chen (2003): 1 The supplier makes a t.i.o.l.i. o er, (w f, F f ), to the fringe rms. If the fringe rms accept the o er, each pays F f. 2 The supplier and the dominant retailer negotiate (w d, F d ). 3 The dominant retailer sets p and the fringe supplies the market at p. Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 5 / 15

14 The Framework Similar to Chen (2003): 1 The supplier makes a t.i.o.l.i. o er, (w f, F f ), to the fringe rms. If the fringe rms accept the o er, each pays F f. 2 The supplier and the dominant retailer negotiate (w d, F d ). 3 The dominant retailer sets p and the fringe supplies the market at p. Market clearing condition for a fringe rm: p = MC (q f ) + w f Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 5 / 15

15 The Framework Similar to Chen (2003): 1 The supplier makes a t.i.o.l.i. o er, (w f, F f ), to the fringe rms. If the fringe rms accept the o er, each pays F f. 2 The supplier and the dominant retailer negotiate (w d, F d ). 3 The dominant retailer sets p and the fringe supplies the market at p. Market clearing condition for a fringe rm: p = MC (q f ) + w f The supply of a fringe rm: s(p w f ) q f = MC 1 (p w f ). Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 5 / 15

16 Assumption to satisfy the SOCs Chen (2003) assumes that the demand and supply functions satisfy the following conditions: C1. D 0 (p) + (p c)d 00 (p) < 0, C2. s 0 (p w f ) + (p w f c)s 00 (p w f ) > 0. We instead assume the necessary and su cient condition: A1. 2D 0 (p) + (p c)d 00 (p) [2s 0 (p w f ) + (p w f c)s 00 (p w f )] < 0. We relax either C1 or C2 while satisfying A1. Remark: Chen s results would be the same if he assumed A1 and C1 (or A1 and C2). Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 6 / 15

17 Negotiation between the supplier and dominant retailer "the deal-me-out bargaining outcome" (Binmore, Shaked and Sutton 89): The outside options do not a ect the sharing unless they are binding, i.e., the agents share their anticipated bilateral pro t not the gains from trade. The anticipated bilateral pro t is (p c) [D(p) ns(p w f )] + nw f s(p w f ) Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 7 / 15

18 Negotiation between the supplier and dominant retailer "the deal-me-out bargaining outcome" (Binmore, Shaked and Sutton 89): The outside options do not a ect the sharing unless they are binding, i.e., the agents share their anticipated bilateral pro t not the gains from trade. The anticipated bilateral pro t is (p c) [D(p) ns(p w f )] + nw f s(p w f ) Chen ignored the supplier s pro ts from the fringe: nw f s(p w f ) Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 7 / 15

19 Negotiation between the supplier and dominant retailer "the deal-me-out bargaining outcome" (Binmore, Shaked and Sutton 89): The outside options do not a ect the sharing unless they are binding, i.e., the agents share their anticipated bilateral pro t not the gains from trade. The anticipated bilateral pro t is (p c) [D(p) ns(p w f )] + nw f s(p w f ) Chen ignored the supplier s pro ts from the fringe: nw f s(p w f ) Let γ 2 (0, 1) denote the dominant retailer s share over the anticipated bilateral pro ts. Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 7 / 15

20 Negotiation between the supplier and dominant retailer "the deal-me-out bargaining outcome" (Binmore, Shaked and Sutton 89): The outside options do not a ect the sharing unless they are binding, i.e., the agents share their anticipated bilateral pro t not the gains from trade. The anticipated bilateral pro t is (p c) [D(p) ns(p w f )] + nw f s(p w f ) Chen ignored the supplier s pro ts from the fringe: nw f s(p w f ) Let γ 2 (0, 1) denote the dominant retailer s share over the anticipated bilateral pro ts. The supplier s outside option: Sell only to the competitive fringe, where Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 7 / 15

21 Negotiation between the supplier and dominant retailer "the deal-me-out bargaining outcome" (Binmore, Shaked and Sutton 89): The outside options do not a ect the sharing unless they are binding, i.e., the agents share their anticipated bilateral pro t not the gains from trade. The anticipated bilateral pro t is (p c) [D(p) ns(p w f )] + nw f s(p w f ) Chen ignored the supplier s pro ts from the fringe: nw f s(p w f ) Let γ 2 (0, 1) denote the dominant retailer s share over the anticipated bilateral pro ts. The supplier s outside option: Sell only to the competitive fringe, where The retail price is p o : D(p o ) = ns(p o w f ). Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 7 / 15

22 Negotiation between the supplier and dominant retailer "the deal-me-out bargaining outcome" (Binmore, Shaked and Sutton 89): The outside options do not a ect the sharing unless they are binding, i.e., the agents share their anticipated bilateral pro t not the gains from trade. The anticipated bilateral pro t is (p c) [D(p) ns(p w f )] + nw f s(p w f ) Chen ignored the supplier s pro ts from the fringe: nw f s(p w f ) Let γ 2 (0, 1) denote the dominant retailer s share over the anticipated bilateral pro ts. The supplier s outside option: Sell only to the competitive fringe, where The retail price is p o : D(p o ) = ns(p o w f ). The supplier earns n [F f + w f s(p o w f )]. Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 7 / 15

23 Retail price equilibrium Through F d, the supplier and dominant retailer share their trade surplus ensuring the supplier at least its outside option. Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 8 / 15

24 Retail price equilibrium Through F d, the supplier and dominant retailer share their trade surplus ensuring the supplier at least its outside option. Through setting w d, the retail price p is determined. Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 8 / 15

25 Retail price equilibrium Through F d, the supplier and dominant retailer share their trade surplus ensuring the supplier at least its outside option. Through setting w d, the retail price p is determined. Taking (F f, w f ) as given, the supplier and the dominant retailer set p : p arg max [(p c) [D(p) ns(p w p f )] + nw f s(p w f )]. Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 8 / 15

26 Retail price equilibrium Through F d, the supplier and dominant retailer share their trade surplus ensuring the supplier at least its outside option. Through setting w d, the retail price p is determined. Taking (F f, w f ) as given, the supplier and the dominant retailer set p : p arg max [(p c) [D(p) ns(p w p f )] + nw f s(p w f )]. Remark: p > p m since the distortion term nw f s 0 (p w f ) > 0. Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 8 / 15

27 Retail price equilibrium Through F d, the supplier and dominant retailer share their trade surplus ensuring the supplier at least its outside option. Through setting w d, the retail price p is determined. Taking (F f, w f ) as given, the supplier and the dominant retailer set p : p arg max [(p c) [D(p) ns(p w p f )] + nw f s(p w f )]. Remark: p > p m since the distortion term nw f s 0 (p w f ) > 0. To implement p > p m, the dominant retailer and the supplier negotiate w d > 0. Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 8 / 15

28 Retail price equilibrium Through F d, the supplier and dominant retailer share their trade surplus ensuring the supplier at least its outside option. Through setting w d, the retail price p is determined. Taking (F f, w f ) as given, the supplier and the dominant retailer set p : p arg max [(p c) [D(p) ns(p w p f )] + nw f s(p w f )]. Remark: p > p m since the distortion term nw f s 0 (p w f ) > 0. To implement p > p m, the dominant retailer and the supplier negotiate w d > 0. Sign p w w f = Sign d w f = How does the distortion term, nw f s 0 (p w f ), change in w f? Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 8 / 15

29 How does the retail price react to the fringe s wholesale price? Lemma 1. When the fringe s supply curve is not too concave, i.e., if C2: s (p-w f ) + (p w f c)s 00 (p w f ) > 0 holds, p is increasing in w f. The opposite is true if the fringe s supply is su ciently concave. Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 9 / 15

30 How does the retail price react to the fringe s wholesale price? Lemma 1. When the fringe s supply curve is not too concave, i.e., if C2: s (p-w f ) + (p w f c)s 00 (p w f ) > 0 holds, p is increasing in w f. The opposite is true if the fringe s supply is su ciently concave. Lemma 2. If the fringe s supply curve is su ciently convex, i.e., if w f s 00 (p w f ) = s 0 (p w f ) holds, w d decreases in w f. The opposite is true otherwise. Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 9 / 15

31 Result I Proposition 1. If the pass-through rate of the fringe s cost is higher than one, p w f > 1, we have w f γ > 0. The opposite is true if the pass-through rate is below one. Intuition: When setting w f, the supplier has a trade-o between the pro ts generated by the fringe: Z p nf f = n 0 w f s(x)dx its share over the bilateral pro ts with the dominant retailer: (1 γ) [(p c) [D(p ) ns(p w f )] + nw f s(p w f )] When γ increases, the latter has less value, so the supplier raises w f only if the margin of the fringe, p, improves by this change. w f if and Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 10 / 15

32 Result II Proposition 2. How the dominant retailer s buyer power a ects consumers and the fringe retailers depend on the pass-through rate of the fringe s wholesale price on the retail price, dp dw f : As γ " wf p p w f 2 (0, 1) & & wd & if s(.) is not too convex, % otherwise. p w f > 1 % % & if s(.) is not too convex, % otherwise. p w f < 0 & % & as opposed to: CHEN: wf p wd As γ " & & no e ect Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 11 / 15

33 E ects of buyer power on pro ts When the dominant retailer s bargaining power increases, the pro t of the supplier, π s, decreases (similar to Chen), the pro t of the dominant retailer, πd, always increases (di erent from Chen), the industry pro ts, πs + πd, increases (as opposed to Chen s result). Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 12 / 15

34 E ects of buyer power on the social welfare The social welfare is de ned as W = Z p D(x)dx + π s + π d When the dominant retailer s bargaining power increases, the welfare decreases if and only if 1) p w f > 1 or 2) 1 < p w f < 0 and γ is su ciently low. Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 13 / 15

35 Discussion Which demand functions lead to above one pass-through rate of input prices? Homogeneous goods: Su ciently convex demands: 2D 0 (p) + (p c)d 00 (p) > 0 Example: Cost amplifying demand functions: (a + p) α D(p) = b where a + p > 0, b > 0, α > 1 (Fabinger and Weyl 08). Di erentiated goods: For highly di entiated product categories, the pass-through of input prices are found to be above one (Besanko, Dubé and Gupta 05, Berck, Leibtag, Villas-Boas and Solis 09). Examples: Beer, dish detergent, and oat cereals. Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 14 / 15

36 Conclusion We analyze how buyer power of a dominant retailer a ects consumers and its rivals when supply contracts are allowed to be non-linear and the dominant retailer faces a competitive fringe. The pass-through rate of the fringe s wholesale price on the retail price is critical for the results. When the pass-through is above one, the buyer power increases the retail price, so harms consumers, increases the wholesale price to the fringe, decreases the social welfare. When the pass-through is between zero and one,the buyer power decreases the retail price, so bene ts to consumers, decreases the wholesale price to the fringe, increases the social welfare. Buyer power increases the pro ts of the dominant retailer, but reduces the supplier s pro ts. In total, it increases the industry pro t. Bedre-Defolie & Sha er (ESMT & UROC) Countervailing Power Hypothesis and Waterbed E ects 16/05/10 15 / 15