abl-02
Dieses Dokument ist Teil der Anfrage „Amtsblätter bis 2018“
Amtsblatt der Bundesnetzagentur
für Elektrizität, Gas, Telekommunikation, Post und Eisenbahnen
484 – Mitteilungen, Telekommunikation, Teil A, Mitteilungen der Bundesnetzagentur – 2 2015
Economics Letters 120 (2013) 379–383
Contents lists available at SciVerse ScienceDirect
Economics Letters
journal homepage: www.elsevier.com/locate/ecolet
Price discrimination or uniform pricing: Which colludes more?
Niklas Horstmann ∗ , Jan Krämer
Karlsruhe Institute of Technology (KIT), Institute of Information Systems and Marketing (IISM), Englerstr. 14, 76131 Karlsruhe, Germany
highlights
• We study the impact of uniform pricing and price discrimination on tacit collusion.
• We conduct a laboratory experiment with two symmetric firms and markets.
• We find that price discrimination leads to higher prices than uniform pricing.
• These differences cannot be explained by existing theory.
article info abstract
Article history: Conventional wisdom attributes different economic outcomes of uniform pricing and price discrimination
Received 9 November 2012 to the heterogeneity in market conditions or market participants, such as differences in demand elasticity
Received in revised form or production costs. We offer a new explanation for the observed differences that relates to behavioral
24 April 2013
aspects rather than demand- or supply-side effects. In particular, in a symmetric Bertrand duopoly
Accepted 10 May 2013
Available online 18 May 2013
laboratory experiment, for which theory predicts no differences between the two pricing regimes, we find
that tacit price collusion is systematically higher under price discrimination than under uniform pricing.
JEL classification:
© 2013 Elsevier B.V. All rights reserved.
C92
L13
Keywords:
Price discrimination
Uniform pricing
Multimarket contact
Experimental economics
Collusion
1. Introduction In this note, we demonstrate in a laboratory experiment that
price discrimination leads to higher average prices than uniform
When firms sell their products in more than one (geographic) pricing even when firms and markets are symmetric. Thus, we
market, they may either charge the same price across markets identify a new explanation for differences in economic outcomes
(uniform pricing) or they may charge differentiated prices between the two pricing regimes that relates to their impacts
according to the specific market conditions (price discrimination). on tacit collusion, rather than cost or demand asymmetries. Pre-
According to conventional wisdom, firms should price discriminate vious experimental studies on tacit collusion have not consid-
whenever possible, due to asymmetric costs or differences in ered the possibility to price discriminate as a treatment variable
demand elasticity across markets. Although some exceptions to (Engel, 2007).
this conventional wisdom were identified (Dobson and Waterson, In this context, our findings also relate to the literature on mu-
2008), the existing literature agrees that price discrimination and tual forbearance (Edwards, 1955), which discussed the collusive ef-
uniform pricing generally yield different market outcomes when fects of multimarket contact. Whereas under price discrimination
there are differences in the market conditions. On the contrary, the underlying markets remain, in principle, independent, uniform
there is currently no theory that predicts differences in market pricing creates a bond between the markets that effectively makes
outcomes due to the two pricing regimes when there are no them one market. Porter (1980) argued that firms meeting in sev-
differences across markets. eral markets (price discrimination) may find it easier to tacitly col-
lude than firms meeting only in one market (uniform pricing). This
is because every colluding firm anticipates that a price deviation
∗ Corresponding author. Tel.: +49 721 608 48387; fax: +49 721 608 48399. in any one market will be punished by price cuts in all markets
E-mail addresses: horstmann@kit.edu (N. Horstmann), kraemer@kit.edu by the other firms. However, Bernheim and Whinston (1990) crit-
(J. Krämer). icized this view and argued that a rational price deviation should
0165-1765/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.econlet.2013.05.011
Bonn, 28. Januar 2015
Amtsblatt der Bundesnetzagentur
für Elektrizität, Gas, Telekommunikation, Post und Eisenbahnen
2 2015 – Mitteilungen, Telekommunikation, Teil A, Mitteilungen der Bundesnetzagentur – 485
380 N. Horstmann, J. Krämer / Economics Letters 120 (2013) 379–383
never occur only in one, but in all markets simultaneously, thus 3. Experimental procedure
rendering the multimarket retaliation as no more effective than the
retaliation in a single market environment. Moreover, the authors For each treatment condition, there were twelve sessions with
formally established an irrelevance result, which states that multi- four subjects each, i.e., 96 participants in total. The experiment
market contact cannot facilitate tacit collusion between symmetric was designed between subject, i.e., participants were exclusively
firms meeting in symmetric markets.1 Hence, our findings can also assigned to one treatment condition. In total, each subject par-
not be explained by the mutual forbearance theory. ticipated in three rounds. Each round consisted of ten consecu-
tive repetitions of the Bertrand stage game, which we refer to as
2. Experimental design periods. Within each round, there was a fixed partner matching.
However, after each round, participants were matched with a new
We consider an industry with two distinct markets, A and B, partner that they did not previously encounter. Thus, each subject
in which two symmetric, price competing firms, i ∈ {1, 2}, offer played with all other participants of the same session for exactly
a homogeneous product for T periods, respectively. The supply of one round (i.e., for ten periods). Since firms were designed to be
one unit of the product to either market implies marginal cost of symmetric, we avoided labelling subjects in any order. Instead, a
c to each firm. The number of consumers per market is N. Denote firm’s current partner was referred to as ‘the other firm’.
i’s price for market X ∈ {A, B} by pXi . Then, according to Bertrand Every effort was made to ensure salience in the experiment.
competition, the demand of firm i in market X in each period is Participants were equipped with a calculator and the experimental
given by software provided a forecast tool for demand and profit in the
if pXi < pX−i and pXi ≤ v next round, given a subject’s expectation of both firms’ prices.
N
X X X Moreover, a history of previous prices within the same round and
Di (pi , p−i ) = N/2 if pXi = pX−i and pXi ≤ v
the same group was provided. However, there was no exchange
if pXi > min{pX−i , v},
0 of information or interaction between subjects in different groups,
i.e., no population feedback (Bruttel, 2009). To avoid budget effects,
where −i is the index of the other firm and v is the consumers’
homogeneous willingness to pay. Consequently, i’s total profit in the earnings of only one round were paid out. Participants threw
each period is a dice to determine which of the last two rounds was paid out to
them. The first round, which was declared a practice round, was
πi (pAi , pBi , pA−i , pB−i ) = DAi · (pAi − c ) + DBi · (pBi − c ) not relevant for the final payoff and thus it is not considered in
in case firms are allowed to price discriminate across markets. the subsequent statistical analysis. The experimental instructions
Similarly, if firms commit to uniform pricing, pi = pAi = pBi . provided to the subjects covered all stated design features of the
It is well known that the unique strict Nash equilibrium of the experiment, including the number of periods and rounds as well
above Bertrand stage game is as how the profits and their final payment would be determined.2
∗ ∗ ∗
pAi = pA−i = ⌈c ⌉ and pBi = pB−i = ⌈c ⌉
∗
The experiment was computerized using z-Tree (Fischbacher,
2007). All sessions were run at the Karlsruhe Institute of
under price discrimination, where ⌈·⌉ returns the smallest feasible Technology in Karlsruhe, Germany, in May and June 2012, and April
price level that is larger than its argument. Likewise, under uniform
2013. Participants were recruited via the ORSEE platform (Greiner,
pricing
2004). Subjects were exclusively students of economic fields. None
p∗i = p∗−i = ⌈c ⌉. of the 24 sessions lasted more than one hour. No initial budget was
Further, under reasonable assumptions about the equilibrium given to the participants. A subject’s average monetary earning was
concept of the finitely repeated Bertrand game, the above unique 10.86 EUR.
equilibrium of the Bertrand stage game is also the unique price
equilibrium of the repeated Bertrand game. For example, Farrell 4. Results
and Maskin (1989) showed that the price equilibrium of the
Bertrand stage game is the unique weakly renegotiation proof We aggregate our data by computing the average market price
price equilibrium of the repeated Bertrand game. It is also the over all ten periods of a round. Note that under price discrimination
unique subgame perfect equilibrium. In conclusion, the theoretical the average is taken also across markets. Thus, at the group
prediction of both pricing scenarios is equivalent in terms of level an observation is uniquely identified by treatment (UP or
equilibrium prices and hence in terms of profits and consumer PD), session (1–12), group (1–2), and round (1 or 2). Thus, there
surplus. are 48 observations for each of the treatments. However, note
In the experiment, participants played T = 10 repeated inter- that due to our matching scheme, observations from a single
actions (periods) of the Bertrand stage game. Profits were accumu- session are not statistically independent. We control for this by
lated over the periods. For a more direct relation between reward means of a hierarchical mixed-effects regression model and by
signals and participants’ decisions, the model was parameterized considering only the session-averaged market prices, respectively.
using EUR instead of an experimental currency unit. Marginal costs First, however, in Table 1 we report the descriptive statistics with
were set to c = 30 cent. Each market had N = 10 consumers with respect to a subject’s average price and profit, and a group’s average
a willingness to pay of v = 50 cent each. The minimum price in- market price as a measure for tacit collusion. Moreover, Fig. 1
crement was chosen to be 1 cent. Treatments differed only with re- shows the average market price for both treatments over the ten
spect to whether participants could price discriminate (PD) or were periods and contrasts it to the equilibrium price. Table 1 and
restricted to uniform pricing (UP) between the two markets. As Fig. 1 already indicate two notable deviations from the theoretical
noted above, the unique∗
strict
∗
Nash equilibrium entails that both prediction. First, prices have a positive offset from marginal costs,
firms choose prices pA = pB = 31 cent for both markets (treat- i.e., from the theoretical equilibrium. This is in line with previous
ment PD) or p∗ = 31 cent as the uniform price (treatment UP) experimental results on Bertrand competition (cf. Engel, 2007).
during all periods. Second, there seem to be differences in market prices and hence
in tacit price collusion between the treatments. On average, the
market price is 4.15 cent (10.71%) higher for the PD treatment.
1 In their model, Bernheim and Whinston consider an infinite time horizon,
whereas we consider a finite time horizon. However, note that collusion is harder
to sustain with a finite time horizon (Harrington, 1987) and thus, the irrelevance
result remains to hold in the present context. 2 The instructions as well as screenshots are provided in the Appendix.
Bonn, 28. Januar 2015
Amtsblatt der Bundesnetzagentur
für Elektrizität, Gas, Telekommunikation, Post und Eisenbahnen
486 – Mitteilungen, Telekommunikation, Teil A, Mitteilungen der Bundesnetzagentur – 2 2015
N. Horstmann, J. Krämer / Economics Letters 120 (2013) 379–383 381
Table 1
Summary statistics.
Treatment Variable Obs. Mean Std. dev. Min Max
UP Avg. price 96 40.45 4.75 31.80 50.00
Avg. market price 48 38.74 4.92 31.10 49.70
Subject’s profit 96 873.96 532.04 10.00 2140.00
PD Avg. price 96 44.50 5.06 34.45 50.00
Avg. market price 48 42.89 5.84 33.10 49.60
Subject’s profit 96 1289.17 625.27 255.00 2240.00
pricing. This result offers the insight that even under symmetric
market conditions the mere possibility to be able to engage in
differential pricing may facilitate collusion and thus lead to higher
prices than price uniformity.
This result bears important policy implications. For example,
competition policy may investigate more closely the impact on
competition when firms switch from uniform pricing to discrim-
inatory pricing. Furthermore, whether price discrimination should
be allowed for different geographic markets is currently under con-
sideration by many national regulatory authorities in the telecom-
munications domain. Currently, telecommunications operators are
bound by a universal service obligation, which usually includes a
uniform pricing constraint. In order to stimulate investments in so-
called next generation networks, regulators are considering mov-
ing towards a geographically segmented regulation, which would
imply the possibility for price discrimination.3 As our results show,
such a relaxation of the pricing constraints may also have unex-
pected consequences on consumers’ surplus and should therefore
Fig. 1. Average market price over time across treatments as a measure for tacit
price collusion. The boundaries of the gray corridor depict the average of minimum
be closely scrutinized by regulators.
and maximum market prices across markets for the price discrimination treatment. Of course, our results are subject to some limitations. Although
our competition model is believed to be fairly robust to alternative
Table 2 theoretical explanations (e.g., other-regarding preferences, hetero-
Results of the mixed-effects regression model on the average geneous products), we only considered price competition. Thus, it
market price. might be worthwhile to investigate whether our empirical results
Covariate Value z p would also hold in the context of quantity (Cournot) competition.
Intercept 38.07 31.36 <0.001 Future work may also address the role of elastic demand, i.e., het-
PD 4.15 2.58 0.010 erogeneous willingness to pay among consumers, which may al-
Round 1.34 1.61 0.107 ter the collusive strategy. Likewise, it would be interesting to see
whether our results carry over to settings in which there are more
In order to test for differences in the average market price than two firms or markets.
between treatments, we consider the following two-level linear
random-intercept model, which controls for the potential depen- Acknowledgments
dence of observations within one session:
Financial support from the German Science Foundation (DFG) is
pij = (β0 + ζj ) + βPD · PD + βRound · Round + ϵij ,
gratefully acknowledged. We particularly thank the editor and an
where pij is the average market price of group i in session j, PD anonymous referee for valuable comments.
is the treatment dummy, Round is a dummy for first or second
payout relevant round, and ζj is the error component shared be- Appendix. Experimental instructions
tween observations of the same session. Table 2 reports the results,
which show that the average market prices are significantly higher The following experimental instructions are for the price discrimi-
for the PD treatment, whilst the round has no significant impact, nation treatment and were translated from German. The instructions
i.e., there is no learning effect. Also by a one-tailed non-parametric for the uniform pricing treatment are identical except with respect to
Mann–Whitney U test on session averages, the market price is sig- the specifics of the treatment.
nificantly higher under price discrimination (p = 0.0471). These
findings suggest that the possibility to differentiate prices between
geographic markets facilitates tacit price collusion more than uni- A.1. Preliminary remarks
form pricing. Hence, consumers’ surplus decreases in the transition
from uniform pricing to price discrimination. Welcome to the experiment and thank you very much for your
participation. If you read through these instructions carefully and
5. Conclusions
Contrary to existing theory, we find tacit price collusion to be 3 For example, recently the German legislator has explicitly enacted that a
significantly higher under price discrimination than under uniform differentiation of retail prices in next generation networks is not abusive per se.
Bonn, 28. Januar 2015
Amtsblatt der Bundesnetzagentur
für Elektrizität, Gas, Telekommunikation, Post und Eisenbahnen
2 2015 – Mitteilungen, Telekommunikation, Teil A, Mitteilungen der Bundesnetzagentur – 487
382 N. Horstmann, J. Krämer / Economics Letters 120 (2013) 379–383
consider them during the experiment, you can earn an amount The profits of firm 2 in both markets are calculated as follows:
of money that depends on your decisions and the decisions of • Profit in market A: 0 · (40 cent − 30 cent) = 0.00 euro.
the other participants. Please address the person in charge of the • Profit in market B: 10 · (45 cent − 30 cent) = 1.50 euro.
experiment in case of questions. Please do not talk to the other • Profit of retail period: 0.00 euro + 1.50 euro = 1.50 euro.
participants during the entire experiment.
Throughout the experiment we will use the currency euro and
A.2.3. Software display
its subunit cent. The euro that you will have earned by the end of
Fig. 2 shows the display of the experiment software during
the experiment will be paid to you in cash.
a retail period. The individual parts will be explained in detail
The experiment is divided into several rounds. The decisions below.
and the results of each round are not interdependent.
In each round, you will simulate the decision of a firm that sells Markets
a good to consumers. There is exactly one other firm apart from The upper part of the display represents the two markets A and
you (which will be called ‘‘the other firm’’ in the following). You B exemplarily and repeats the key decision factors.
will be competing against this other firm.
Decision and testing environment
A.2. Setup of a round In the lower part of the display you can enter your prices for the
two markets. In addition, you can test the consequences of your
In each round, you and the other firm offer a good in two pricing decision. In order to do so, you have to indicate your prices
markets. In each market there are 10 consumers. as well as what you expect the prices of the other firm to be, by
The provision of the good will cost you 30 cent per consumer using the slider.
in market A and in market B. You and the other firm produce equal As soon as you release the slider, the values in the table beneath
goods. Therefore, the consumers will buy the good from the firm are being updated. The table shows the effects of the respective
that offers the good at a lower price. The consumers are willing to price combination on the demand and on the profit in one retail
pay a maximum of 50 cent for the good. period.
One round consists of 10 retail periods. In each retail period you For each market the demand and the profit of your firm and the
offer your good to all consumers. Thereby, you choose an individual other firm are being shown. The table beneath shows the demand
price for market A and market B. and the profit accumulated over both markets for your firm and the
other firm.
Please use the slider for your firm to set your prices. Note that
A.2.1. Demand
the actual prices of the other firm are set by the corresponding firm
The demands in market A and market B will be determined and not by you. The slider for the other firm only serves as a means
separately and depend on your prices and the prices of the other of decision support.
firm. All consumers will always demand the good that is offered Please note: You cannot only use the testing environment in
at the lower price. However, they are not willing to pay more than order to test the effects of your own pricing decision, but also to
50 cent per retail period. estimate the possible reactions to your current pricing decision of
Thus, if you offer your good in a market at a higher price than the the other firm.
other firm, there will be no demand for your good in this market.
The same applies if one of the two firms offers the good at a price Retail history
higher than 50 cent. If both firms offer the same price in a market, See Fig. 3.
each firm will receive half of the demand in this market. If both Clicking the button ‘‘History’’ opens the retail history. The main
prices of the same market are above 50 cent, there will be no parameters of the past retail periods are shown in this section:
demand for both firms in this market.
• Your price for market A.
• The price of the other firm for market A.
A.2.2. Profit • Your price for market B.
Your profit in a retail period depends on your prices and • The price of the other firm for market B.
your demand. Therefore, the profit in one market is calculated as • Your profit in both markets.
follows: • The profit of the other firm in both markets.
Profit = Demand · (Price − Costs of Provision). Another click on the button ‘‘History’’ closes the retail history
again.
Your profit in a retail period is calculated by summing up your
profit in market A and your profit in market B. The profits in both
A.3. Course of the experiment
markets are being accumulated over all 10 retail periods and make
your overall profit.
Overall, 3 rounds (0–2) are being played. Round 0 is a test run.
Each round consists of 10 retail periods. The firm pairings are being
Example
randomly determined all over again in each round. However, it is
Firms 1 and 2 offer the following prices: excluded that you will ever play again with the same firm.
The information regarding the round and the course of the
Market A Market B
experiment are always being shown at the top level of the screen.
Firm 1 35 cent 50 cent
At the end of each round, your accumulated profits of all
Firm 2 40 cent 45 cent 10 retail periods that represent your payoff are being shown in
euro. You do not have to memorize this value. At the end of the
The profits of firm 1 in both markets are calculated as follows:
experiment, the payment of exactly one round will be paid out
• Profit in market A: 10 · (35 cent − 30 cent) = 0.50 euro to you. It will be chosen at random which of the rounds is being
• Profit in market B: 0 · (50 cent − 30 cent) = 0.00 euro cashed out. To this end, you will have to roll a dice. The test run,
• Profit of retail period: 0.50 euro + 0.00 euro = 0.50 euro. round 0, is not included.
Bonn, 28. Januar 2015
Amtsblatt der Bundesnetzagentur
für Elektrizität, Gas, Telekommunikation, Post und Eisenbahnen
488 – Mitteilungen, Telekommunikation, Teil A, Mitteilungen der Bundesnetzagentur – 2 2015
N. Horstmann, J. Krämer / Economics Letters 120 (2013) 379–383 383
Fig. 2.
Fig. 3.
A.4. Concluding remarks References
Do not hesitate to ask questions. As long as they refer to these Bernheim, B.D., Whinston, M.D., 1990. Multimarket contact and collusive behavior.
The RAND Journal of Economics 21 (1), 1–26.
instructions and not to possible strategies, we will answer your Bruttel, L.V., 2009. Group dynamics in experimental studies—The Bertrand paradox
questions as far as possible. Please remember: The better you have revisited. Journal of Economic Behavior & Organization 69 (1), 51–63.
understood these instructions, the more money you can make. Dobson, P.W., Waterson, M., 2008. Chain-store competition: customized vs. uniform
Before the experiment starts, you will be asked some questions pricing. University of Warwick, Department of Economics. Mimeo.
Edwards, C.D., 1955. Conglomerate bigness as a source of power. In: Business
on the screen with regard to the understanding of the rules Concentration and Price Policy. Princeton University Press, pp. 331–352.
and the course of the experiment. Please enter the respective Engel, C., 2007. How much collusion? A meta-analysis on oligopoly experiments.
answer into your computer. Afterwards, the experiment will start Journal of Competition Law & Economics 3 (4), 491–549.
Farrell, J., Maskin, E., 1989. Renegotiation in repeated games. Games and Economic
automatically. Behavior 1 (4), 327–360.
In case of any questions, please remain seated and give the Fischbacher, U., 2007. z-tree: zurich toolbox for ready-made economic experi-
person in charge of the experiment a hand signal. Please wait until ments. Experimental Economics 10 (2), 171–178.
the person in charge of the experiment has arrived at your seat. Greiner, B., 2004. An online recruitment system for economic experiments.
In: Kremer, K., Macho, V. (Eds.), Forschung und Wissenschaftliches Rechnen
Talk as quietly as possible when asking the question. Please remain 2003, GWDG Bericht 63. Gesellschaft für Wissenschaftliche Datenverarbeitung,
seated after the end of the experiment as well and wait for further Göttingen, pp. 79–93.
instructions from the person in charge of the experiment. You can Harrington Jr., J.E., 1987. Collusion in multiproduct oligopoly games under a finite
horizon. International Economic Review 28 (1), 1–14.
make notes on the pad that is laid out for you on the table during Porter, M.E., 1980. Strategic Interaction: Some Lessons from Industry Histories for
the experiment. Please leave the experiment instructions, the Theory and Antitrust Policy. Division of Research, Graduate School of Business
calculator as well as the note pad at the table after the experiment. Administration, Harvard University.
Bonn, 28. Januar 2015
Amtsblatt der Bundesnetzagentur
für Elektrizität, Gas, Telekommunikation, Post und Eisenbahnen
2 2015 – Mitteilungen, Telekommunikation, Teil A, Mitteilungen der Bundesnetzagentur – 489
Tacit collusion under multimarket contact with
identical firms and markets
Niklas Horstmann∗ Jan Krämer†
September 10, 2014
Abstract
According to Bernheim and Whinston’s (The RAND Journal of Economics 21
(1), 1990) irrelevance result, multimarket contact may not facilitate tacit collusion
if identical firms meet in identical markets. In contrast, we offer a novel behavioral
explanation why multimarket contact may facilitate tacit collusion for this case. By
means of an economic laboratory experiment without explicit communication, we
show that a firm can implicitly communicate the collusive intention through its price
setting behavior, i.e., by price signaling. We find that multimarket contact facilitates
tacit collusion because, in contrast to single market contact, firms can send differen-
tiated price signals.
Keywords: multimarket contact; tacit collusion; price signaling; uniform pricing con-
straint
JEL classification: C92; L13; L41
∗
Corresponding author. Karlsruhe Institute of Technology; horstmann@kit.edu.
†
University of Passau; jan.kraemer@uni-passau.de.
We thank participants at the 2014 Conference of the European Association for Research in Industrial
Economics (EARIE) in Milan for valuable comments. Financial support from Deutsche Forschungsgemein-
schaft (DFG) is gratefully acknowledged.
1
Bonn, 28. Januar 2015
Amtsblatt der Bundesnetzagentur
für Elektrizität, Gas, Telekommunikation, Post und Eisenbahnen
490 – Mitteilungen, Telekommunikation, Teil A, Mitteilungen der Bundesnetzagentur – 2 2015
1 Introduction
Detection and prevention of collusive behavior are among the primary concerns of compe-
tition authorities. This includes the exposure of explicit collusion (i.e., cartels) as well as
the hindrance of tacit collusion in an effort to create competitive markets. Consequently,
a substantial amount of economic research is devoted to characterize market conditions
that facilitate tacit collusion among competing firms.1 In a nutshell, this research indi-
cates that i) a low number of firms (Dufwenberg and Gneezy, 2000; Huck et al., 2004) and
ii) multimarket contact (Yu and Cannella, 2013) are, above and beyond the possibility to
communicate explicitly in the form of cheap talk (Fonseca and Normann, 2012), considered
to be the main drivers of tacit collusion. However, to date the relationship between multi-
market contact and tacit collusion is only supported by field studies (reviewed by Yu and
Cannella, 2013), but could not be clearly confirmed in controlled laboratory experiments
(Güth et al., 2010; Phillips and Mason, 1992; Feinberg and Sherman, 1988). Moreover, pre-
vious experimental research indicates that without explicit communication “firms virtually
never coordinated successfully, not even the duopolies” (Fonseca and Normann, 2012, p.
1770).
In this article we reconsider the role of communication and multimarket contact in the
context of collusive behavior. Thereby, we build on the insight that firms can coordinate on
a collusive state without the need for explicit communication by implicitly communicating
their intention for collusive play solely through their price setting behavior. This implicit
communication through prices is known as price signaling in the extant experimental lit-
erature (Hoggatt et al., 1976; Durham et al., 2004; Davis et al., 2010). Unfortunately,
the term signaling is ambiguous in the economic literature. In a similar context, it is
also used to refer to explicit communication in the form of cheap-talk about future prices
1
As tacit collusion lacks formal agreements, it is difficult to obtain empirical evidence for its existence.
The few field studies on the identification of tacit collusion are reviewed by Feuerstein (2005). Laboratory
experiments on tacit collusion are surveyed by Potters and Suetens (2013) as well as Engel (2007).
2
Bonn, 28. Januar 2015
Amtsblatt der Bundesnetzagentur
für Elektrizität, Gas, Telekommunikation, Post und Eisenbahnen
2 2015 – Mitteilungen, Telekommunikation, Teil A, Mitteilungen der Bundesnetzagentur – 491
(Holt and Davis, 1990; Cason, 1995; Cason and Davis, 1995). More generally signaling is
associated with signaling games in which players use signals to reveal their type (Spence,
1974). Note that in this article we will refer to (price) signaling exclusively as implicit
communication through price setting behavior. We suggest that such price signaling is
facilitated by multimarket contact, because firms meeting in several geographic markets
can set different prices (distinct price signals) on each market. This allows multimarket
firms to signal prices more efficiently in contrast to single market firms, which only have a
single price (one price signal) at their disposal.
The main message of this article is twofold. First, in contrast to previous experimental
studies (e.g., Güth et al., 2010), we are able confirm under controlled laboratory conditions
that multimarket contact facilitates tacit collusion. Second, we show that price signaling
can facilitate tacit collusion under certain conditions—a theory that was also dismissed in
previous experimental studies (e.g., Plott, 1982; Davis et al., 2010). Hence, our results bear
important insights for competition policy by highlighting that limiting firms’ possibilities
to engage in price signaling can effectively mitigate the emergence of tacit collusion. This
is particularly relevant if collusion is suspected among firms that meet in several geograph-
ically distinct, but otherwise relatively homogeneous markets (such as telecommunications
or airline markets), where a uniform pricing constraint could therefore be an effective tool
to render price signaling ineffective, and hence, may undermine the process that establishes
collusion.
The remainder of this article is organized as follows. Section 2 reviews the related
literature and Section 3 provides the theoretical background of our study. Subsequently,
the experimental design and procedure is described in Section 4. Sections 5 and 6 present
the results and finally, Section 7 offers a discussion and policy implications.
3
Bonn, 28. Januar 2015
Amtsblatt der Bundesnetzagentur
für Elektrizität, Gas, Telekommunikation, Post und Eisenbahnen
492 – Mitteilungen, Telekommunikation, Teil A, Mitteilungen der Bundesnetzagentur – 2 2015
2 Related literature
Multimarket contact and collusion
Edwards (1955) was the first to link multimarket contact to incentives for cooperative be-
havior. According to his mutual forbearance hypothesis, multimarket contact reduces the
competitive intensity, because firms that meet in multiple markets fear to trigger a price
war across all markets if they undercut their rivals in any one market. The mutual forbear-
ance hypothesis has stimulated considerable empirical research on multimarket contact in
several industry contexts, including airlines (Evans and Kessides, 1994), cement (Jans and
Rosenbaum, 1997), telecommunications (Parker and Röller, 1997), hotels (Fernandez and
Marin, 1998), and media (Waldfogel and Wulf, 2006). See Yu and Cannella (2013) for a
recent and comprehensive overview. By and large, this research confirmed a relationship
between multimarket contact and tacit collusion. Scott (2008) reviews merger cases in
the US, especially over different geographic areas, and makes a strong case for considering
multimarket contact as a potential anti-competitive harm in conglomerate and horizontal
mergers. In their seminal article, Bernheim and Whinston (1990) explain theoretically un-
der which conditions multimarket contact may indeed facilitate tacit collusion. However,
the authors also establish an irrelevance result stating that multimarket contact may not
explain mutual forbearance in situations where identical firms experiencing identical and
constant returns to scale meet in identical markets. Therefore, to date “most researchers
assume that mutual forbearance requires asymmetric markets, rivals, and competitive po-
sitions” (Yu and Cannella, 2013, p. 77).
Price signaling and collusion
By contrast, we challenge the irrelevance result by Bernheim and Whinston (1990) from a
behavioral point of view and argue that multimarket contact may facilitate tacit collusion
4
Bonn, 28. Januar 2015
Amtsblatt der Bundesnetzagentur
für Elektrizität, Gas, Telekommunikation, Post und Eisenbahnen
2 2015 – Mitteilungen, Telekommunikation, Teil A, Mitteilungen der Bundesnetzagentur – 493
also when the irrelevance result holds, i.e., when identical rivals with identical constant
returns to scale technology meet in identical markets. More precisely, we conjecture that
price signaling may be used to coordinate on a collusive state and that multimarket contact
renders these price signals more efficient.
This conjecture on price signaling and collusion is not new per se. However, previous
research could not find any convincing evidence on the effectiveness of price signaling on
the emergence of tacit collusion (Plott, 1982; Davis et al., 2010; Potters and Suetens, 2013),
nor did it consider price signaling in the context of multimarket contact, i.e., what we will
refer to as signal efficiency.
Initial observations of signals in repeated price competition experiments have been
made by Hoggatt et al. (1976) and Friedman and Hoggatt (1980). Plott (1982) discusses
these early attempts to model the effect of signals and conjectures that price signaling
occurs, but has no clear effect on tacit collusion.2 Hoggatt et al. (1976) conduct oligopoly
experiments with repeated price decisions. They differentiate between pulses (“sequence of
two or three successive price changes which sum to zero”, Hoggatt et al. (1976, p. 263)) and
steps (“price change of unusually large magnitude”, Hoggatt et al. (1976, p. 263)). Only
the latter are found to have a temporary effect on the price development and to be more
probable in a positive (negative) direction, if a firm’s price is low (high). Comparably,
in an auction experiment, where information about losing bids is a treatment variable,
Dufwenberg and Gneezy (2002) find prices to be supra-competitive if bidders are informed
about the losing bids in previous periods. They hypothesize that this is due to signaling
behavior during repeated interaction.
Durham et al. (2004) observed signaling behavior in pricing decisions in an extensively
repeated posted offer market experiment. According to Durham et al. (2004, p. 155), “a
price signal is defined as any price submitted by any firm that is greater than or equal
2
Surprisingly, price signaling has not been addressed again until 20 years later.
5
Bonn, 28. Januar 2015
