1. Naked Exclusion: An Experimental Study of Contracts with Externalities Claudia M. Landeo Kathryn E. Spier University of Alberta Harvard Law School Economics Department and NBER North American Meetings of the Econometric Society Carnegie Mellon University June 20, 2008

2. Rasmusen, Ramseyer, and Wiley (AER, 1991) argue that exclusion may arise when the incumbent can deter entry through exclusive contracts with just a subset of the buyers. –Economies of scale. –Coordination failures among the buyers. Segal and Whinston (AER, 2000) refine RRW’s argument. –Divide-and-conquer strategies. Theoretical Framework

3. Theoretical Framework (cont.) Suppose there are three players: an incumbent monopolist, and two buyers. First Stage: The Contracting Stage. –The seller simultaneously offers transfer payments x 1 and x 2 to the buyers in exchange for exclusivity. –After observing both offers, the buyers simultaneously decide whether to accept or reject the offers. Second Stage: The Entry Stage. –Scale economies imply that entry will be deterred if one (or both) buyers accepted the offers in Stage 1. Third Stage: The Market Pricing Stage.

4. Numerical Examination Our experiment focuses on the Contracting Stage. The buyers’ payoffs at the acceptance subgame reflects the Second and Third Stages: The incumbent seller’s payoffs: –If both buyers reject the offers (no-exclusion), the seller receives zero profits. –If both buyers accept the offers (exclusion), the seller receives 1950 – x 1 - x 2. –If only buyer i accepts the offer (exclusion), the seller receives 1950 – x i. To reduce the subjects' computational costs, we restrict the incumbent seller's offers to a small set: x i ϵ {100, 650, 800, 1100}, i = 1, 2. AcceptReject Accept(x 1, x 2 )(x 1, 0) Reject(0, x 2 )(1000, 1000) Buyer 1 Buyer 2

5. Theoretical Predictions: No Discrimination Suppose the incumbent cannot discriminate (x 1 = x 2 = x). The acceptance subgame becomes a symmetric coordination (stag hunt) game. -Proposition 1. There are multiple subgame perfect Nash equilibria when the incumbent cannot discriminate: - Exclusion equilibria: The incumbent offers x ϵ {100, 650, 800} and both buyers accept. - Equilibria with entry: the incumbent offers x ϵ {100, 650, 800} and both buyers reject.

6. Theoretical Predictions: Discrimination Now suppose the incumbent can discriminate. - Proposition 2. There are multiple subgame perfect Nash equilibria, all of which involve exclusion. In these equilibria, x 1 + x 2 ≤ 1200 and both buyers accept. - (100,1100) and (1100,100), the divide-and-conquer offers. - Other SPNE include (100,100), (100,650), (100,800).

7. Experimental Design Two offer treatments: – No-discrimination and discrimination  theoretical model Two communication treatments: – No-communication and two-way buyer-buyer communication  coordination games Two buyers’ payoff treatments: – Exogenous and endogenous payoffs  coordination games with endogenous payoffs (fairness and reciprocity) Eight-condition, between-subject design. Sixteen 70-minute to 90-minute sessions. 215 subjects in total.

8. Findings: Offers and Exclusion Rates - No Discrimination Condition (100, 100) (650, 650) (800, 800) Total Offers EN/ND/NC 5 112 3 120 [.00,.00] [.96,.84] [1.00, 1.00] EN/ND/C 9 74 37 120 [.00,.00] [.39,.07] [.59,.24] Endogenous (exclusion rate) Exogenous (exclusion rate) [From Table 6 on page 18]

9. Findings: Offers and Exclusion - Discrimination Condition (100, 800)/ (100, 1100)/ (650, 650) Total (800, 100) (1100,100) Offers EN/D/NC 20 83 31 144 [.25,.50][1.00,.99] [.84,.71] EN/D/C 6 113 7 132 [.00,.17] [.88,.69][.43,.00] Endogenous (exclusion rate) Exogenous (exclusion rate) (From Table 6 on page 18)

10. Findings: Effects of Discrimination on Exclusion Rate (last six rounds) *** Note: *** denotes significance at the 1%.

11. Findings: Determinants of Buyer’s Acceptance Marginal Effects Own Offer.33*** (.0001) Partner’s Offer.11** (.0002) Partner’s Reject Intention–.67*** (.0539) Divide-and-Conquer Offers.45*** (.0421) Endogeneity.29*** (.0522) Observations1008 Note:***and ** denote significance at the 1% and 5%, respectively. [Table 9 on page 25]

12. ExclusionNo Exclusion (A, A), (A, R), (R, A) (R, R) No-Communication Cooper et al. (1992) 1.00.00 EX/ND/NC 1.00.00 EN/ND/NC 1.00.00 Communication Cooper et al. (1992).09.91 EX/ND/C.24.76 EN/ND/C.59.41 Communication and Endogeneity: A Comparison with Previous Studies Offers: (800,800) [Table 10 on page 26]

13. Concluding Remarks Our paper makes several contributions. –Naked Exclusion literature: –Our findings provide support for Segal and Whinston’s predictions about divide-and-conquer strategies under communication. –We show that communication between the buyers influences not only the exclusion rate but also the offers made by the incumbent seller. –Our results suggest that fairness and reciprocity may lead to higher exclusion rates than previously anticipated. –Experimental Economics literature: –We provide the first empirical test of coordination games with endogenous payoffs.

14. Concluding Remarks (cont.) Implications for Public Policy: –Our findings suggest that Naked Exclusion may be surprisingly easy for incumbent firms to achieve. –Even in the absence of discrimination, when adequate communication channels were not available, our subjects failed to coordinate on their preferred equilibria and entry was deterred. –Coordination was particularly elusive when the incumbent seller had a human identity. The human face of a sales representative (an agent for the seller) might elicit fairness and reciprocity from the agents representing the buyers, and facilitate the exclusion of faceless rivals (in the event of contracts perceived as “kind.”) –Our results provide an efficiency rationale for permitting communication among non-competing buyers. Communication among buyers might serve the public interest by facilitating entry.

15. Concluding Remarks (cont.) Other Applications: -Contracting with externalities: corporate takeovers, licensing, mergers, debt bailouts (Segal, 1999). -Bargaining with externalities: joint-and-several liability (Kornhauser and Revesz, 1994), complex litigation (Spier, 2002; Che and Spier, forthcoming), plea bargaining (Bar- Gill and Ben-Shahar, 2007).

16. Extra Slides

21. Extension: Private Offers

23. Result 1: Under communication, discrimination significantly increases exclusion. *** Note: *** denotes significance at the 1%. **

24. Result 2: Under no-discrimination, or under discrimination and exogeneity, communication significantly reduces exclusion. Note:*** and ** denote significance at the 1% and 5%, respectively. *** *

25. Result 3: Under communication, endogeneity of the offers significantly increases exclusion Note:*** and * denote significance at the 1% and 10%, respectively. *****

26. Result 4: Communication significantly affects the choice of offers by sellers. It induces the choice of the divide-and- conquer offers in discrimination environments, and reduces the likelihood of choosing (650, 650) offers. Conditions(650, 650)(100, 1100) or (1100, 100) EN/ND/NC vs. -.317***n. a EN/ND/C (.052) Observations 240 EN/D/NC vs. -.164*.282* EN/D/C (.121)(.141) Observations 276276 Note:***and ** denote significance at the 1% and 5%, respectively. [Table 8 on page 23]

27. Experimental Design Two offer treatments: – no-discrimination and discrimination Two communication treatments: – no-communication and two-way buyer-buyer communication. Two buyers’ payoff treatments: – exogenous and endogenous payoffs. Eight-condition, between-subject design. Sixteen 70-minute to 90-minute sessions. – undergraduate and graduates students from Northwestern University. – between 9 and 21 human subjects per session. – 215 subjects in total.

28. Experimental Design, cont. Context-free environment. Networked personal computers and a software designed for this experiment. Eight versions of the game corresponding to the eight experimental conditions. Experimental currency: token (650 tokens = $1).

29. Experimental Design, cont. At the beginning of the session – written instructions provided to the subjects. – instructions verbally presented. Roles and groups assigned randomly. 3 practice rounds and 12 actual rounds. Subjects paid according to their performance.

30. Directions for Future Research? Additional trials on private offers vs. public offers. –Our experimental trials suggest that the exclusion rates will be lower when this is the case (73% versus 82% in the E/D/NC case.) –Explore belief formation. Testing models of downstream competition. –Although the Anheuser Busch example had distributors with exclusive territories, this is not always the case. –Simpson and Wickelgren (AER, 2007) and Fumagalli and Motta (AER, 2006). Other important topics in Antitrust. –Breach of contract, bundling, volume discounts, etc.