COMMUNICATION AND MARKET SHARING: AN EXPERIMENT ON THE EXCHANGE OF SOFT AND HARD INFORMATION

• Author: Andreas Freitag,Christian Thöni,Catherine Roux
• DOI: http://doi.org/10.1111/iere.12480
• Date: 01 February 2021
• Published date: 01 February 2021

INTERNATIONALECONOMIC REVIEW

Vol. 62, No. 1, February 2021 DOI: 10.1111/iere.12480

COMMUNICATION AND MARKET SHARING: AN EXPERIMENT ON THE

EXCHANGE OF SOFT AND HARD INFORMATION∗

By Andreas Freitag, Catherine Roux, and Christian Thöni1

University of Basel, Switzerland; University of Basel, Switzerland; University of Lausanne,

Switzerland

We study the role of communication in collusive market sharing. In a series of Cournot oligopoly experi-

ments with multiple markets, we vary the information that f‌irms can exchange: hard information—verif‌iable

information about past conduct—and soft information—unbinding information about future conduct. We f‌ind

that the effect of communication on the f‌irms’ ability to collude depends on the type of information available:

Whereas market prices increase only slightly with hard information, the price raise due to soft information is

substantial. Our results point to the types and contents of communication that should be of particular concern

to antitrust authorities.

1. introduction

What role does communication play in collusive market sharing? We study this question

in a laboratory experiment. Our f‌indings suggest that the effect of communication on mar-

ket sharing depends on the type of communication involved: Although nonbinding one-time

communication on future planned conduct can establish market sharing almost perfectly, the

exchange of disaggregated information on recent past behavior—which allows the identif‌ica-

tion and selective punishment of deviators—increases market prices only slightly. Antitrust

rules condition enforcement on information that is observable and verif‌iable in court. There-

fore, evidence on whether f‌irms communicated serves as a smoking gun in antitrust proce-

dures against collusion. To understand which type of communication carries the highest col-

lusive potential is thus central for the design of effective antitrust rules.

Market sharing is a collusive practice prohibited by the European Community competi-

tion rules.2A market sharing agreement typically determines exclusive territories along ge-

ographical borders, the so-called home markets. The colluding f‌irms refrain from entering

each other’s home market with the result that each market remains insulated from the others.

∗Manuscript received August 2018; revised April 2020.

1We thank Masaki Aoyagi, three anonymous referees, Miguel Brendl, Aaron Edlin, Miguel Fonseca, John Mayo,

João Montez, Georg Nöldeke, Hans-Theo Normann, Luìs Santos-Pinto, and Alois Stutzer, as well as audiences at

various seminars and conferences for valuable comments and discussions. The usual disclaimer applies. This re-

search project is part of the Swiss Competence Center for Energy Research SCCER CREST of the Swiss Innovation

Agency Innosuisse. Financial support from the SNSF (grant no. 100018_182185) is gratefully acknowledged. Please

address correspondence to: Catherine Roux, Faculty of Business and Economics, University of Basel, Peter Merian-

Weg 6, 4002 Basel, Switzerland. Phone: +41 61 207 23 98 E-mail: catherine.roux@unibas.ch.

2We follow the terminology of Bellef‌lamme and Bloch (2004, 2008): by “market sharing,” we designate the recip-

rocal collusive agreement in which each f‌irm is assigned monopoly rights over a market to distinguish it from “pro-

duction quotas” where the f‌irms jointly restrict output within a market. Other names for these arrangements are to be

found in the literature, for example, spheres of inf‌luence and multimarket contact (Bernheim and Whinston, 1990),

collusion at the extensive and intensive margin (Byford and Gans, 2014), collusion with and without trade (Lom-

merud and Sørgard, 2001; Colonescu and Schmitt, 2003; Bond and Syropoulos, 2008), exclusive territories, and sales

quotas (Harrington, 2006).

175

© (2020) by the Economics Department of the University of Pennsylvania and the Osaka University Institute of So-

cial and Economic Research Association

176 freitag, roux, and thöni

Market sharing agreements jeopardize the unity of a common market and, thus, the European

Commission (EC) takes a particularly tough stance on these arrangements.3

Evidence on interf‌irm communication and explicit agreements is crucial for the antitrust en-

forcement against collusive practices because it makes anticompetitive contacts between f‌irms

visible and verif‌iable in court and thereby facilitates legal prosecution. Despite this fact, col-

luding f‌irms commonly engage in incriminating communication, and they do so even in the

face of large monetary f‌ines (see, e.g., Genesove and Mullin, 2001; Harrington, 2006; Lev-

enstein and Suslow, 2008; Hyytinen et al., 2019). Hence, communication must be a valuable

tool for establishing collusive agreements. Although interf‌irm communication has recently at-

tracted substantial attention from economic researchers, we still know little about what type

of communication facilitates collusion.4

According to European case law, communication between f‌irms is more likely to be consid-

ered harmful if the information exchanged (i) is f‌irm specif‌ic instead of aggregated, (ii) is con-

cealed from customers and potential entrants, and (iii) concerns either recent conduct or fu-

ture plans in relation to pricing and supply (Kühn, 2001). Information on recent behavior and

decisions is considered as “hard” because it is in principle available and verif‌iable for the par-

ties. In contrast, communication about planned future market conduct conveys information

that is “soft” in the sense that it is unbinding and about intentions, which are not verif‌iable

at the time the f‌irms take their decisions (Kühn, 2001, p. 170). Communication of soft infor-

mation is known as cheap talk in economic theory. Cheap talk can enhance collusion when it

helps f‌irms resolve the strategic uncertainty that typically occurs in games with multiple equi-

libria (Farrell, 1987).5Although the impact of cheap talk in pure coordination games (like the

battle of the sexes) is well established in the literature, its effect in oligopoly games is subject

to debate among economic theorists (Farrell and Rabin, 1996; Whinston, 2008).

Regarding the exchange of hard information, Stigler (1964) argued that it can facilitate col-

lusion very effectively. The heart of Stigler’s argument is that to detect and punish deviations

from a collusive agreement, f‌irms must be able to monitor their coconspirators’ actions. This

ability to monitor makes punishment a credible threat that disciplines the cartel members.6

Stigler’s work has inspired antitrust policy on information exchange between f‌irms. Accord-

ingly, the EC distinguishes between the exchange of aggregate industry and that of individual

f‌irm data7: Although it does not object the exchange of information on sales and production

as long as the data do not allow to identify individual f‌irms, the EC has taken the stance in

various antitrust decisions that the publication of individualized data would make markets ar-

tif‌icially transparent and, thereby, less competitive.8

3In a landmark case, the two major producers of soda ash, Solvay and ICI, were sanctioned with the largest an-

titrust f‌ine imposed by the EC at that time (1990) for having operated a market sharing agreement. The EC found

the f‌irms guilty of having conf‌ined their soda ash activities to their traditional home markets, with Solvay supplying

continental Western Europe and ICI serving the United Kingdom, during at least 17 years (EC IP/90/1057 and OJ L

152, 1991).

4A recent focus is on the precise content of communication that supports collusion. For example, Harrington et al.

(2016) compare the collusive effects of price announcements with those of unrestricted communication. Fonseca et al.

(2018) and Fischer and Normann (2019) use text mining to understand what kind of language is most useful to sup-

porting collusion.

5See, for example, Cason (1995) for a discussion of cheap talk in collusion in relation to the antitrust investigation

around the tariff publishing system in the U.S.airline industry where airlines were alleged to use nonbinding price sig-

nals to raise fares.

6Green and Porter (1984) show that imperfect observability of rivals’ past actions makes collusion more diff‌icult

in the sense that temporary price wars are needed to sustain collusion. Abreu et al. (1986) extend Green and Porter

(1984)’s model to include optimal punishment strategies. Kandori (1992) shows that the set of equilibria increases to-

ward more collusive outcomes when observability improves. Aoyagi and Fréchette (2009) and Aoyagi et al. (2019)

provide corresponding experimental evidence.

7See Kühn and Vives (1995) for a detailed analysis of the EC’s competition policy regarding information exchange.

8See, for example, the decisions on UK Agricultural TractorsExchange (OJ L 20, 1993), Fatty Acids (OJ L 3, 1987).

Market transparency is also an important element in the assessment of coordinated effects in merger reviews: The