A model of biased intermediation

• Published date: 01 December 2019
• Author: Alexandre Cornière,Greg Taylor
• Date: 01 December 2019
• DOI: http://doi.org/10.1111/1756-2171.12298

RAND Journal of Economics

Vol.50, No. 4, Winter 2019

pp. 854–882

A model of biased intermediation

Alexandre de Corni`

ere∗

and

Greg Taylor∗∗

We study situations in which consumers rely on a biased intermediary’s advice when choosing

among sellers. We introduce the notion that sellers’ and consumers’ payoffs can be congruent

or conflicting, and show that this has important implications for the effects of bias. Under

congruence, the firm benefiting from bias has an incentive to offer a better deal than its rival

and consumers can be better-off than under no bias. Under conflict, the favored firm offers lower

utility, and bias harms consumers. We study various policies for dealing with bias and show that

their efficacy also depends on whether the payoffs exhibit congruence or conflict.

1. Introduction

Consumers often turn to an intermediary for advice when choosing among firms. The inter-

mediary’sjob is to provide information about which are the best products and which products best

fit each consumer’s needs. However, the intermediary does not necessarily have the consumer’s

best interests in mind. Particular concerns have repeatedly been raised about the fact that a ver-

tically integrated intermediary has a clear incentive to bias its advice in favor of its own product

offerings rather than those of rivals. These concerns have given rise to high-profile investigations

in many markets by regulators and competition authorities, who worry about potentially harmful

effects of bias.

For example, in the current debate about online platform regulation, a major issue is that

of “own-content bias,” that is when a multisided platform that is also present on one side of

the market behaves in such a way as to steer consumers toward its own products or services.

Google, for instance, has been investigated in many jurisdictions for promoting its own products

in search results.1In 2017, the European Commission imposed a €2.42bn fine after Google was

∗Toulouse School of Economics, Universityof Toulouse Capitole; alexandre.de-corniere@tse-fr.eu.

∗∗University of Oxford; greg.taylor@oii.ox.ac.uk.

This article was previously circulated under the title “Quality Provision in the Presence of a Biased Intermediary.” We

are grateful to Diane Coyle, Jacques Cr´

emer, Serena Drufuca, Mikhail Drugov, Anthony Dukes, P´

eter Es¨

o, Renaud

Foucart, Doh-Shin Jeon, Ben Hermalin, Paul Klemperer,Laurent Lamy, Jon Levin, Meg Meyer,Ines Moreno de Bar reda,

David Myatt, Martin Obradovits, Sander Onderstal, R´

egis Renault, Andrew Rhodes, Michael Riordan, Andre Veiga,

Chengsi Wang,Chris Wilson, Peyton Young,Jidong Zhou, two anonymous referees, and participants at various seminars

and conferences for useful comments and discussions. The authors gratefully acknowledge the financial support of

this research by the NET Institute, www.NETinst.org. A. de Corni`

ere acknowledges funding from ANR under grant

ANR-17-EUR-0010 (Investissements d’Avenir program).

1Edelman and Lai (2013) provide evidence that Google’s own-content bias drives more traffic to its affiliate

flight-search service than it would get otherwise.

854 C2019, The RAND Corporation.

DE CORNI `

ERE AND TAYLOR / 855

found guilty of favoring its own comparison shopping website. Before that, the United States

(US) Federal Trade Commission (FTC) and the High Court of England and Wales had ruled

against fining Google for related practices.2Other companies may soon face similar charges: the

European Commission recently launched an investigation to determine whether Amazon uses

its dominant position in the e-commerce market to favor its own, private-label sellers.3Some of

Apple’s practices also tend to favor its own services at the expense of competitors (see Kr¨

amer

and Schnurr, 2018, for a thorough discussion). The European Commission, particularly active

when it comes to the issue of platform regulation, recently put forward a proposal to require

platforms to, among other things, be transparent regarding “how they treat their own goods or

services compared to those offered by their professional users.”4This issue may become more

prevalentin the not-too-distant future with the g rowthof the market for virtual assistants, cur rently

dominated by vertically integrated firms like Amazon, Google, and Apple. Forbesalready repor ts

that “Eighty-five percent of Amazon customers select the recommended Amazon product when

voice shopping.”5

The mechanisms of own-content bias mayalso apply to situations where a leading application

or hardware provider (the “intermediary”) chooses one of its own complementary applications as

the default option for consumers: Safari as the default browser for iPhones, Google as the default

search engine for its browser Chrome, etc. Being the default option is a powerful way to foster

usage of a product or service, as the default bias is well documented (Sunstein and Thaler, 2008).

Companies sometimes choose long-term contracts as an alternative to integration: for example,

the technology press reports that heavy bidding forced Google to pay $300 million for the right

to be the default search engine in Mozilla Firefox and $1 billion for similar rights across Apple’s

suite of products.6

The idea that an intermediary might provide biased recommendations because of financial

incentives is, of course, not restricted to the digital economy. Similar concerns have been raised

in the pharmaceutical and financial industries, where doctors and financial advisers sometimes

receive commissions potentially leading them to recommend one treatment/investment over an-

other which would be better suited to their patient/client (Engelberg, Parsons, and Tefft, 2013;

Cookson et al., 2019; Egan, Forthcoming). Default option bias may also apply there, as when a

bank offers its own insurance alongside a loan.

Biased intermediation raises two main concerns: first, are consumers directed towardinferior

or higher priced products or services? Second, does bias distort firms’ incentives in a way that

harms consumers? Even though these concerns are common to the situations described above,

a certain degree of heterogeneity remains as to the institutional or strategic environments. A

key concern in the price comparison website industry, for example, has been the effect of paid

endorsements on final consumer prices. Search engines and websites, by contrast, are mostly free

to access, so that attention in the search bias debate has focused on innovation and investment

decisions rather than on pricing. It is therefore doubtful that the analysis of a particular market

could be transposed directly to another one. We show that, in order to understand the effects

of bias, it is often possible to reduce complex strategic heterogeneity to a simpler problem of

identifying the relationship between firms’ and consumers’ marginal payoffs.This insight allows

2See ec.europa.eu/competition/elojade/isef/case_details.cfm?proc_code=1_39740 for the EC’sdecision, www.ftc.

gov/sites/default/files/documents/public_statements/statement-commission-regarding-googles-search-practices/130103

brillgooglesearchstmt.pdf for the FTC’s one, and www.bailii.org/ew/cases/EWHC/Ch/2016/253.htmlfor the High Court’s

decision in Streetmap.EU Ltd v Google Inc. & Ors (2016); all pages accessed October 12, 2018.

3See www.ft.com/content/a8c78888-bc0f-11e8-8274-55b72926558f, accessed October 12, 2018.

4A summary of the proposal is available at europa.eu/rapid/press-release_IP-18-3372_en.htm, accessed October

12, 2018.

5See www.forbes.com/sites/kirimasters/2018/06/30/amazon-voice-commerce-a-huge-opportunity-for-brands-or-t

oo-early-to-tell/#2080f0a83d43, accessed October 16, 2018.

6See allthingsd.com/20111222/google-will-pay-mozilla-almost-300m-per-year-in-search-deal-besting-microsoft-

and-yahoo/ and www.recode.net/2014/4/16/11625704/marissa-mayers-secret-plan-to-get-apple-to-dump-google-and-def

ault-to, accessed October 12, 2018.

C

The RAND Corporation 2019.