Search engine competition with network externalities

Journal of Competition Law and Economics

Volumn 8, Issue 1, 2012, Pages 73-105

  Argenton, Cédric ^a   Prüfer, Jens ^a  

^a TILBURG UNIVERSITY   (Netherlands)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 84857933606     PISSN: 17446414     EISSN: 17446422     Source Type: Journal    
DOI: 10.1093/joclec/nhr018     Document Type: Article

References (96)

1. See E.B. Boyd, Google on Facebook-Bing Social Search: What, Us Worry?, FAST COMPANY, Oct. 14, 2010, http://www.fastcompany.com/1695187/google-on-social-search-who-us-worried.
2. PwC, IAB Internet Advertising Revenue Report, at 13 (2010), available at http://www.iab.net/media/file/IAB_report_1H_2010_Final.pdf.
3. Press Release, comScore, comScore Releases November 2010 U.S. Search Engine Rankings (Dec. 15, 2010).
4. See Kris, What Is Search Marketing?, Type Media (posted Jan. 16, 2011), http://www.type-media.co.uk/search-marketing/what-is-search-marketing.
5. See Jared Newman, Will Bin's Back-End Make Yahoo Better?, PCWORLD, Aug. 25, 2010, http://www.pcworld.com/article/204134/will_bings_backend_make_yahoo_better.html.
6. There are other reasons for concern about the market for search engines, which we do not take as the primary objects of this study: media pluralism and privacy rights come to the forefront.
7. Press Release, Eur. Comm'n, Antitrust: Commission Probes Allegations of Antitrust Violations by Google (Nov. 30, 2010), http://europa.eu/rapid/pressReleasesAction.do?reference=IP/10/1624.
8. See, e.g., Italian Newspapers Ask Competition Watchdog to Investigate Google News, OUTLAW.COM, Aug. 31, 2009, http://www.out-law.com/page-10336.
9. Press Release, U.S. Dep't of Justice, Justice Department Requires Google Inc. to Develop and License Travel Software in Order to Proceed with Its Acquisition of ITA Software Inc. (Apr. 8, 2011), http://www.justice.gov/opa/pr/2011/April/11-at-445.html.
10. Thomas Catan & Amir Efrati, Feds to Launch Probe of Google, WALL ST. J., June 24, 2011, http://online.wsj.com/article/SB10001424052702303339904576403603764717680.html.
11. Google Should Be Watched Carefully, FINANCIAL TIMES, July 15, 2010, http://www.ft.com/cms/s/0/a84e8438-9049-11df-ad26-00144feab49a.html#axzz1izSgsNzm.
12. For a classical exposition, see WILLIAM J. BAUMOL, JOHN C. PANZAR & ROBERT D. WILLIG, CONTESTABLE MARKETS AND THE THEORY OF INDUSTRY STRUCTURE (Harcourt Brace Jovanovich 1982).
13. Whether the behavior of a market participant is susceptible to give rise to antitrust liability remains, of course, of prime legal interest even in that case, but there is no reason to expect the enforcement of competition law to lead to a significantly different market outcome in the medium run.
14. Amanda Spink, Dietmar Wolfram, Major B. J. Jansen & Tefko Saracevic, Searching the Web: The Public and Their Queries, 52 J. AM. SOC'Y FOR INFO. SCI. & TECH. 226 (2001).
15. Bernard J. Jansen, Mimi Zhang & Ying Zhang, Brand Awareness and the Evaluation of Search Results, World Wide Web Conference Poster Paper (2007).
16. Carrie Grimes, Diane Tang & Daniel M. Russell, Query Logs Alone Are Not Enough, World Wide Web Conference Paper (2007), available at http://dmrussell.googlepages.com/query_logs_are_not_enough_2007_02_12.pdf.
17. Ziv Bar-Yossef & Maxim Gurevich, Mining Search Engine Query Logs via Suggestion Sampling, Very Large Data Bases Coference Paper (2008).
18. Steven Levy, Secret of Googlenomics: Data-fueled Recipe Brews Profitability, WIRED MAG., June 17, 2009, available at http://www.wired.com/culture/culturereviews/magazine/17-06/nep_googlenomics.
19. F. Silvestri, Mining Query Logs: Turning Search Usage Data into Knowledge, 4 FOUND. & TRENDS IN INFO. RETRIEVAL 1 (2010).
20. See Press Release, U.S. Dep't of Justice, Statement of the Department of Justice Antitrust Division on Its Decision to Close Its Investigation of the Internet Search and Paid Search Advertising Agreement Between Microsoft Corporation and Yahoo! Inc. (Feb. 18, 2010) [hereinafter U.S. DOJ Statement on Microsoft Yahoo! 2010].
21. See uTest, Q3 2009 Bug Battle Report: Battle of the Search Engines (2009), available at http://www.utest.com/sites/default/files/uTest%20Q3_09%20Search%20Engine%20Bug%20Battle.pdf.
22. Search engines also seek to gather other types of data, such as information about a user's profile and preferences by monitoring click-through rates to contents or online purchasing behavior, a point stressed by Daneil F. Spulber, The Map of Commerce: Internet Search, Competition, and the Circular Flow of Information, 5 J. COMPETITION L. & ECON. 633 (2009).
23. To some extent, Google's business model has recently consisted of providing users with free ancillary applications (including e-mail, calendar, personal page management system, browser, and others) that generate traffic in the first place and can be used to collect such individual-level information. We do not think, however, that this information is subject to network externalities of the type that we stress in this article, as it is specific to a particular user.
24. Amanda Spink, Dietmar Wolfram, Major B. J. Jansen & Tefko Saracevic, Searching the Web: The Public and Their Queries, 52 J. AM. SOC'Y FOR INFO. SCI. & TECH. 226 (2001) noted that query logs "provide a snapshot for comparison of public behavior while searching, a behavior that can also serve as a clue for improvement of search engines."
25. It is technically possible to sample search engines' drop-down suggestion menus as a way to indirectly mine their query logs of a particular engine, but this is no substitute for directly processing those logs.
26. See, e.g., Ziv Bar-Yossef & Maxim Gurevich, Mining Search Engine Query Logs via Suggestion Sampling (presented at the Very Large Data Base (VLDB) Conference, 2008).
27. See Part III. For a comparison of several key predictions of the model with reality.
28. Good points of entry include Stanley M. Besen & Joseph Farrell, Choosing How to Compete: Strategies and Tactics in Standardization, 8 J. ECON. PERSP. 673 (1994).
29. Nicholas Economides, The Economics of Networks, 14 INT'L J. INDUS. ORG. 117 (1996).
30. CARL SHAPIRO & HAL R. VARIAN, INFORMATION RULES: A STRATEGIC GUIDE TO THE NETWORK ECONOMY (Harvard Bus. Rev. Press 1999).
31. For detailed analyses of the policy implications of this line of research, see Nicholas Economides, Antitrust Issues in Network Industries, in THE REFORM OF EC COMPETITION LAW: NEW CHALLENGES 343 (Ioannis Lianos & Ioannis Kokkoris eds., Kluwer Int'l Law 2010).
32. Neil Gandal, Compatibility, Standardization and Network Effects: Some Policy Implications, 18 OXFORD REV ECON. POL'Y 80 (2002).
33. Kenneth J. Arrow, The Economic Implications of Learning by Doing, 29 REV. ECON. STUD. 155 (1962).
34. BOSTON CONSULTING GROUP, PERSPECTIVES on EXPERIENCE (Boston Consulting Group 1972).
35. See Theodore Paul Wright, Factors Affecting the Cost of Airplanes, 3 J. AERONAUTICAL SCI. 122 (1936).
36. More recently, C. Lanier Benkard, Learning and Forgetting: The Dynamics of Aircraft Production, 90 AM. ECON. REV. 1034 (2000).
37. The literature on the learning curve is voluminous. Recent noticeable contributions include (but are not limited to) Pauls S. Adler & Kim B. Clark, Behind the Learning Curve: A Sketch of the Learning Process, 37 MGMT. SCI. 267 (1991).
38. Willard I. Zangwill & Paul B. Kantor, Towards a Theory of Continuous Improvements and the Learning curve, 44 MGMT. SCI. 910 (1998).
39. Eelke Wiersma, Conditions that Shape the Learning Curve: Factors that Increase the Ability and Opportunity to Learn, 53 MGMT. SCI. 1903 (2007).
40. Frederick J. Riggins, Charles H. Kriebel & Tridas Mukhopadhyay, The Growth of Interorganizational Systems in the Presence of Network Externalities, 40 MGMT. SCI. 984 (1994).
41. Eric T. G. Wang & Abraham Seidmann, Electronic Data Interchange: Competitive Externalities and Strategic Implementation Policies, 41 MGMT. SCI. 401 (1995).
42. For accessible introductions, see Bernard J. Jansen, Search Log Analysis: What It Is, What's Been Done, How to Do It, 28 LIB. & INFO. SCI. RES. 407 (2006).
43. Fabrizio Silvestri, Mining Query Logs: Turning Search Usage Data into Knowledge, 4 FOUND. & TRENDS IN INFO. RETRIEVAL 1 (2010).
44. See Hal R. Varian, The Economics of Internet Search, 96 RIVISTA DI POLITICA ECONOMICA 9 (2006).
45. Daniel F. Spulber, The Map of Commerce: Internet Search, Competition, and the Circular Flow of Information, 5 J. COMPETITION L. & ECON. 633 (2009).
46. David S. Evans, The Economics of the Online Advertising Industry, 7 REV. NETWORK ECON. 359 (2008).
47. James D. Ratliff & Daniel L. Rubinfeld, Online Advertising: Defining Relevant Markets, 6 J. COMPETITION L. & ECON. 653 (2010).
48. See Anindya Ghose & Sha Yang, An Empirical Analysis of Search Engine Advertising: Sponsored Search in Electronic Markets, 55 MGMT. SCI. 1605 (2009).
49. Greg Taylor, Search Quality and Revenue Cannibalization by Competing Search Engines (Mimeo, Oxford Univ., 2010).
50. Song Yao & Carl F. Mela, A Dynamic Model of Sponsored Search Advertising, 30 MARKETING SCI. 447 (2011).
51. See Yongmin Chen & Chuan He, Paid Placement: Advertising and Search on the Internet (Networks, Electronic Commerce, & Telecommunications (NET) Institute, Working Paper No.06-02, 2006).
52. Benjamin Edelmann, Michael Ostrovsky & Michael Schwarz, Internet Advertising and the Generalized Second-price Auction: Selling Billions of Dollars Worth of Keywords, 97 AM. ECON. REV. 242 (2007).
53. Susan Athey & Glenn Ellison, Position Auctions with Consumer Search, Q. J. ECON. (forthcoming).
54. Avi Goldfarb & Catherine Tucker, Search Engine Advertising: Channel Substitution When Pricing Ads to Context, 57 MGMT. SCI. 458 (2011).
55. Avi Goldfarb & Catherine Tucker, Substitution between Online and Offline Advertising Markets, 7 J. COMPETITION L. & ECON. 37 (2011).
56. Neil Gandal, The Dynamics of Competition in the Internet Search Engine Market, 19 INT'L J. INDUS. ORG. 1103 (2001).
57. Rahul Telang, Uday Rajan & Tridas Mukhopadhyay, Market Structure For Internet Search Engines, 21 J. MGMT. INFO. SYS. 137 (2004).
58. Rufus Pollock, Is Google the Next Microsoft? Competition, Welfare and Regulation in Online Search, 9 REV. NETWORK ECON. 1 (2010).
59. Not all search engines are profitable. As of March 2010, "reports indicate that Redmond [Microsoft] has spent as much as $5bn on its nine-month-old search revamp, and the company continues to pour cash into the would-be Google killer."
60. Cade Metz, One Day, Bing Will Actually Make Money, THE REGISTER, Mar. 2, 2010, http://www.theregister.co.uk/2010/03/02/ballmer_on_bing/. Compare this with the performance of search engines 2 and 3 in our model displayed in Figure 2.
61. See Jean-Charles Rochet & Jean Tirole, Two-Sided Markets: A Progress Report, 37 RAND J. ECON. 645 (2006) for a technical overview of the theory.
62. Jean-Charles Rochet & Jean Tirole, Introduction to the Symposium on Two-sided Markets, 3 COMPETITION POL'Y INT'L 146 (2007) introduce several less-technical papers.
63. Avi Goldfarb & Catherine Tucker, Online Display Advertising: Targeting and Obtrusiveness, 30 MARKETING SCI. 389 (2011) show that targeted advertising leads to negative externalities only when obtrusive.
64. The French Competition Authority's online advertising sector inquiry reached the same conclusion. Avis No. 10-A-29 du 14 Décembre 2010 sur le Fonctionnement Concurrentiel de la Publicité en Ligne [Opinion No. 10-A-29 of 14 Dec. 2010 on the Competitive Operation of Online Advertising], Autorité de la Concurrence [Competition Authority] (Fr.).
65. Positive indirect network externalities on the users' side would only reinforce market tipping.
66. Since 2010, Yahoo! has used Bing's search technology and does not invest in its own search technology anymore
67. For an exposition of those models, see SIMON P. ANDERSON, ANDRE DE PALMA & JACQUES-FRANCOIS THISSE, DISCRETE CHOICE THEORY OF PRODUCT DIFFERENTIATION (MIT Press 1992).
68. In a first approximation, the marginal cost of running an additional query can be taken to be zero. In any case, variable costs can always be subsumed into the p variable, which would then stand for net revenue per user. This formulation implicitly assumes that quality affects the fixed cost of production rather than the variable cost. This is likely: variable costs mostly come from the huge energy requirements needed to run server farms, whereas quality is directly related to the work of engineers and software developers.
69. We are aware of the fact that a static model is not the best tool to study market dynamics. The dynamic extension of our model can be worked out, especially if there is no quality persistence over time. We use the static model to illustrate our main ideas.
70. The search engine industry arguably meets the conditions identified by Joshua S. Gans, When Is Static Analysis a Sufficient Proxy for Dynamic Considerations? Reconsidering Antitrust and Innovation, in 11 INNOVATION POLICY AND THE ECONOMY 55 (Josh Lerner & Scott Stern eds., Univ. of Chicago Press 2010) for static analysis to be a sufficient proxy for dynamic considerations in innovative industries.
71. 1 = ε) because then firm 2 could set a higher quality and (almost) capture the entire market demand, thereby making a profit.
72. The development of market shares on a global level followed a similar pattern in the last decade. The main difference was that Google's dominance was even more pronounced on a global scale, reaching about 85 percent market share at the end of 2010.
73. See After Record Profits: Google Stock Shoots Up, FinancesJournal.com, http://financesjournal.com/taxes/record-profits-google-stock-shoots-7114.html.
74. The results are qualitatively identical whether we consider two, three, or more firms.
75. As our analysis has shown, quality and welfare under an uncontested monopoly are even lower.
76. See equations (A.29) and (A.30) in the Appendix.
77. All or a fraction of the old query data may stop being relevant after some time. Thus, it is possible that exponential decay or truncation impacts the evolution of N over time. That would not change the qualitative predictions of the model: a firm that has been dominant in the recent past would presently benefit from a competitive advantage.
78. See equations (A.10) and (A.11) in the Appendix or Figure 2.
79. Eytan Adar, User 4XXXXX9: Anonymizing Query Logs (presented at the World Wide Web Conference, 2007), available at http://www.cond.org/anonlogs.pdf.
80. Syam Menon & Sumit Sarkar, Minimizing Information Loss and Preserving Privacy, 53MGMT. SCI. 101 (2007) explore ways to anonymize large data sets such as query logs.
81. We are aware of the potential public choice difficulty that handing over enforcement and monitoring to a public agency could create. This is why we suggest to have at least two technical and institutional infrastructures do this job, which should be ideally placed in at least two jurisdictions. Competition between (or at least duplication of) agencies may preempt any attempt to abuse any one agency's power.
82. Here, we follow EINER ELHAUGE & DAMIEN GERADIN, GLOBAL COMPETITION LAW AND ECONOMICS (Hart Publishing 2007).
83. See Otter Tail Power Co. v. U.S., 410 U.S. 366 (1973); Aspen Skiing Co. v. Aspen Highlands Skiing Corp., 472 U.S. 585 (1985).
84. See, most recently, Verizon Commc'ns v. Law Offices of Curtis V. Trinko, 540 U.S.398 (2004).
85. See Commercial Solvents: Cases 6 and 7/73, Commercial Solvents and Others v. Comm'n, 1974 E.C.R. 223; Case 7/97, Oscar Bronner GmbH & Co KG v. Mediaprint Zeitung und Zeitschriftenverlag GmbH & Co., 1998 E.C.R. I-7791; Case C-148/01, IMS Health GmbH & Co. OHG v. Health GmbH & Co.KG, 2004 E.C.R. I-5039; Case T-201/14, Microsoft Corp. v. Comm'n, 2007 E.C.R. II-3601.
86. See Aspen, 472 U.S. 585; Eastman Kodak v. Image Technical Services, 504 U.S. 451 (1992). In Aspen, the case revolved around the possibility for one ski domain manager to issue bundled tickets giving access to all mountains in one ski resort, including some managed by another company. In Kodak, eighteen independent service providers were complaining that Kodak had restricted access to replacement parts needed to service Kodak equipment.
87. This power was recognized by case law before the enactment of the regulation. See Case C-62/86 AKZO Chemie BV v. Commission, 1991 E.C.R. 3359, ¶155.
88. Frank P. Maier-Rigaud, Per Hellström & Friedrich Wenzel Bulst, Remedies in European Antitrust Law, 76 ANTITRUST L.J. 43 (2009).
89. For introductory surveys, see Mark Armstrong, The Theory of Access Pricing and Interconnection, in 1 HANDBOOK OF TELECOMMUNICATIONS ECONOMICS 295 (Martin E. Cave, Sumit K. Majumdar & Ingo Vogelsang eds., North-Holland 2002).
90. Ingo Vogelsang, Price Regulation of Access to Telecommunications Networks, 41 J ECON. LIT. 830 (2003).
91. Council Directive 95/46, 1995 O.J. (L 281) (EC).
92. See Opinion 1/2008 on Data Protection Issues Related to Search Engines, Advisory Working Party (adopted Apr. 4, 2008) (EC), Data Protection available at http://ec.europa.eu/justice/policies/privacy/docs/wpdocs/2008/wp148_en.pdf. The Working Party was created by Counceil Directive 95/46, art. 29 (EC).
93. See the source mentioned in footnote 3 for this assessment.
94. See Google Anonymizing Search Records To Protect Privacy, Search Engine Land, http://searchengineland.com/google-anonymizing-search-records-to-protect-privacy-10736 for details on how anonymization can work in practice.
95. For a reading of the European Microsoft case along similar lines, see Pierre Larouche, The European Microsoft Case at the Crossroads of Competition Policy and Innovation: Comment on Ahlborn and Evans, 75 ANTITRUST L.J. 933 (2009).
96. We are aware that classical Stackelberg competition uses quantity levels, not quality levels, as a strategic variable. Still, we feel that our model is sufficiently close to the original Stackelberg model to warrant the use of its name.