How Do Hub-and-Spoke Cartels Operate? Lessons from Nine Case

How Do Hub-and-Spoke Cartels Operate?

Lessons from Nine Case Studies



Joseph E. Harrington, Jr.

Patrick T. Harker Professor

Department of Business Economics & Public Policy

The Wharton School

University of Pennsylvania

3620 Locust Walk

Philadelphia, PA 19104

harrij@wharton.upenn.edu

24 August 2018

Abstract

Hub-and-spoke collusion is when …rms in a market coordinate their conduct by

communicating through an upstream supplier or downstream customer. This study

examines nine hub-and-spoke cartels towards understanding how they operate: What

is the collusive scheme? How do …rms achieve mutual understanding regarding that

scheme? What is the role played by the hub? How e¤ective is hub-and-spoke collusion?

The paper also discusses legal approaches to hub-and-spoke collusion.



The research assistance of Hsiang-Yen (Sam) Huang, Asad Hussain, and Sin Chit (Martin) Lai is grate-

fully acknowledged. This research has been supported by a grant from the Carol and Lawrence Zicklin

Center for Business Ethics Research, The Wharton School, University of Pennsylvania.

1 Introduction

A cartel is a collection of competitors who seek to coordinate their behavior for the purpose of

producing a supracompetitive outcome. Coordination is typically achieved by having cartel

memb ers directly and expressly communicate with each other. As an example, consider the

citric acid cartel. Its …ve members initially met to agree on a market allocation in terms

of market shares, and then regularly communicated to determine prices and exchange sales

data. Over a period of four years, about 25 meetings of the cartel were conducted along

with a dozen or so bilateral meetings between some of the cartel members.

Through these

meetings - whether in-person or over the phone - …rms coordinated on a collusive outcome,

monitored for compliance with that outcome, implemented punishments when there was

evidence of noncompliance, and more broadly dealt with the challenges of running a cartel.

While direct communication between cartel members is typical, and would seem to be the

most e¤ective means for delivering a supracompetitive outcome, it is not the only manner

in which colluding …rms share intentions and observations. In a number of cartels, …rms

instead communicated through a third party which had a vertical relationship with them.

Referred to as a hub-and-spoke cartel, the spokes are the colluding competitors and the hub

is an upstream supplier or downstream customer that facilitates collusion by the spokes.

The horizontal agreement among the spokes is referred to as the rim, for it connects the

spokes. Coordination occurs by each spoke communicating with the hub, and the hub

sharing information it has learned from one spoke with the other spokes. While this indirect

communication may be supplemented with some direct communication between the spokes,

the primary avenue for communication is through the hub.

Figure 1 shows two hub-and-spoke cartels in the market for toys. In the U.S. toys case,

downstream toy retailer Toys "R" Us was the hub and upstream toy manufacturers like

Hasbro, Mattel, and Fisher Price were the spokes.

Toys "R" Us orchestrated a group boycott

among toy manufacturers against warehouse clubs. So as to achieve mutual understanding

regarding the collusive scheme, each manufacturer learned about the intentions of the other

manufacturers by communicating through Toys "R" Us. In the UK toys case, the roles were

reversed. Upstream toy manufacturer Hasbro was the hub which facilitated collusion among

downstream toy retailers Argos and Littlewoods, who were dominant in the London market.

Argos and Littlewoods never directly communicated about enacting a common price increase

and instead exclusively conveyed their pricing intentions through Hasbro.

The objective of this paper is to investigate hub-and-spoke collusion towards addressing

the following questions: What are the types of collusive schemes implemented by hub-and-

spoke cartels? Given the lack of direct communication, how di¢ cult was it for …rms to achieve

some common understanding? Why did …rms choose to collude through an upstream supplier

or downstream customer? How essential was the hub’s role? How e¤ective was collusion?

To address these questions, the study’s approach is to engage in a series of case studies of

hub-and-spoke cartels. These questions will be asked of each of the hub-and-spoke cartels

These details are from Harrington (2006). See that study and Marshall and Marx (2012) for many

examples.

The Administrative Law Judge of the Federal Trade Commission concluded that there were 14 toy

manufacturers (spokes) involved. On appeal, the Commission reduced the number to seven, which included

the three companies cited in Figure 1.

examined, from which we’ll seek to identify both variation and commonality across these

cartels.

For a case to be included in our study, it must satisfy three criteria. First, there is

compelling evidence that there was collusion (e.g., a conviction). Second, communication

between competitors was primarily through an upstream supplier or downstream customer.

And, third, there is adequate do cumentation for identifying the collusive scheme and the

process by which they coordinated on that collusive scheme. After an extensive search, nine

cases have been found that satisfy these criteria: movie exhibition (U.S.), toys (U.S., UK),

paints and varnishes (Poland), pharmaceutical products (U.S.), replica football kits (UK),

cheese (UK), drugstore, perfumery, and hygiene products (Belgium), and e-books (U.S.). In

spite of the small number of cases (which means any conclusions are highly tentative) and the

usual sample selection bias associated with cartel studies,

I believe the study substantively

adds to our understanding of hub-and-spoke collusion.

Hub-and-spoke cartels have long been recognized as a phenomenon in economic and

legal circles. Consequently, there have been studies of individual cases which have provided

useful insight. What there has not been is a systematic investigation that seeks to identify

common elements of hub-and-spoke collusion. Odudu (2011), Sahuguet and Walckiers (2013,

2014), Zampa and Buccirossi (2013), Van Cayseele (2014), and Orbach (2016) raise some

of the economic and legal issues regarding hub-and-spoke collusion but do not provide a

comparative analysis of hub-and-spoke cartels towards addressing the questions of this paper.

There is also a small body of theoretical work on hub-and-spoke collusion which includes

Van Cayseele and Miegelsen (2013) and Sahuguet and Walckiers (2017), as well as some

papers motivated by the e-books case (which are referenced later). In contrast, the goal of

this paper is empirical as we assess how hub-and-spoke collusion works in practice.

Let me summarize some of the study’s …ndings. All nine hub-and-spoke cartels were in-

tended to reduce competition in a retail market. To do so, one of two collusive schemes were

deployed: 1) coordinated exclusion by upstream suppliers (spokes) against a rival or class

of rivals to a downstream retailer (hub); and 2) collusive price setting among downstream

retailers (spokes), often through the imposition of a recommended price by the upstream

supplier (hub). When the exclusion strategy was pursued, the downstream hub also threat-

ened exclusion against the upstream suppliers if they did not participate. It is an open

question whether collusion bene…tted or harmed those upstream …rms. When the scheme

instead had downstream …rms jointly raise their prices, the upstream hub purportedly ben-

e…tted in some cases by being able to raise its wholesale prices. In order to achieve mutual

understanding with regards to the collusive scheme, the hub typically engaged in extensive

back-and-forth bilateral communications with the spokes. These communications went well

beyond informing spokes of the scheme, as the hub had to convince each spoke that the other

spokes were planning to participate. The critical and essential role of the hub is very clear.

Our sample encompasses discovered and convicted cartels and thus excludes cartels that were not dis-

covered and those which were discovered but the evidence was inconclusive regarding collusion. As a result,

ine¤ective cartels could be undersampled (as they quickly collapsed and avoided discovery) or oversampled (as

their ine¤ectiveness extended to revealing evidence), while e¤ective cartels could be undersampled (as they

were skilled at avoiding detection) or oversampled (as they remained active long enough to be discovered).

If the reader is aware of other hub-and-spoke cartels that may satisfy these criteria, I welcome learning

about them.

In most cases, it was the hub that initiated collusion. The hub was the key player in both

coordinating on a collusive scheme and monitoring for compliance by the spokes with that

scheme. The hub performed its own monitoring, collected reports from spokes, and then

acted on this information by contacting and working with noncompliant spokes. However,

in spite of all of these e¤orts, hub-and-spoke cartels had relatively short duration compared

to average cartel duration. It appears that this short duration is due to them being more

easily detected than standard cartels.

Section 2 o¤ers an initial assessment of how hub-and-spoke collusion might di¤er from

when …rms collude using direct communication. Section 3 provides the case studies of nine

hub-and-spoke cartels, from which some regularities are distilled in Section 4. Section 5

examines the enforcement of competition law with a focus on how to establish that …rms

have engaged in unlawful hub-and-spoke collusion.

2 An Initial Assessment of Hub-and-Spoke Collusion

E¤ective collusion requires coordinating on a collusive scheme, monitoring for compliance

with that collusive scheme, and, in order to provide incentives to comply, punishment when

there is evidence of noncompliance. The participation in a cartel of an upstream supplier

or downstream buyer that acts as a conduit for messages between …rms can a¤ect all three

dimensions. First, communication may be less e¤ective because it goes through a third

party. Some information may be lost when a spoke’s message is conveyed to the hub who

then passes it along to other spokes. Second, communication may be less e¤ective and the

collusive scheme could be impacted because the third party is not neutral. As the hub has

vertical relations with the spokes, it will be a¤ected by collusion among the spokes and that

could in‡uence the hub’s conduct. Third, the hub may have information and instruments at

its disposal to make monitoring and punishing more e¤ective. Let me brie‡y elaborate on

each of these e¤ects. The case studies will ‡esh them out.

Collusion requires that competitors achieve some level of mutual understanding that they

are constraining competition and how they plan to constrain competition (i.e., the collusive

scheme). In a hub-and-spoke cartel, this mutual understanding is achieved by communicating

indirectly through the hub. Compared to when competitors directly communicate, indirect

communication may be less e¤ective for two reasons. First, messages between spokes may lose

information in the process of going through the hub. Collusion requires each …rm to achieve

some level of trust with the other cartel members so that it is con…dent that they will abide by

the collusive outcome. When involving face-to-face direct communication, it is not just words

that may create that trust; it may be in‡uenced by how the words are expressed (emphasis,

in‡ection, hesitation), facial expressions, eye contact, and body language. Experimental

research has shown that this extraneous information from direct interaction provides useful

cues for cooperation.

However, when …rms communicate through a third party, all but the

words are stripped away. In addition to the loss of those facets of messages, errors may be

introduced as the hub may forget to convey something said by a spoke or mistakenly insert

something that a spoke did not say. In short, messages communicated through a hub are apt

For example, see Centorrino et al (2015) and Sparks, Burleigh, and Barclay (2016). Manzini, Sadrieh,

and Vriend (2009) o¤ers a novel approach to measuring the signaling value of non-verbal gestures.

to b e less informative than when conveyed directly, and that could make it more di¢ cult for

…rms to coordinate on a collusive arrangement.

Second, the hub may intentionally distort messages or otherwise control the conversation

between spokes because it is not a neutral third party.

The hub will be impacted by

collusion and has a di¤erent objective than the spokes due to being an upstream supplier

or downstream customer. Hence, even if the hub and spokes all prefer collusion, the desired

form of collusion could di¤er between them. Of course, even with direct communication,

…rms may have an incentive not to tell the truth and try to mislead rival …rms.

However, as

competitors’preferences are likely to be more in line with each other than with an upstream

supplier or downstream customer, communication is likely to be less e¤ective with hub-and-

spoke collusion than with direct communications between colluding …rms.

Furthermore,

theoretical research has shown that the presence of an intermediary between communicating

parties can only reduce the informativeness of those parties’messages.

To elaborate on this second source of message distortion, consider a hub-and-spoke cartel

in which a manufacturer is the hub (e.g., Hasbro), retailers are the spokes (e.g., Argos

and Littlewoods), and the collusive outcome has retailers raise their prices which will then

support a rise in the wholesale price. As is well documented in our case studies, a retailer

is very concerned with avoiding a situation in which it raises its price only to …nd that

other retailers did not did not raise their prices; for such an outcome would cause the

retailer to su¤er a decline in demand and pro…t. Thus, a compliant retailer incurs a cost

for participating in a failed cartel and, critical for the current discussion, this cost is likely

to be more severe for a retailer than it is for the manufacturer. With the manufacturer

undervaluing the risk that the retailers face, it may try to mislead them in order to enhance

the chances of collusion. For example, the manufacturer might tell each retailer that the

other retailers have agreed to raise their prices when they have said no such thing. The

manufacturer bene…ts if this tactic succeeds in inducing all retailers to raise their prices, and

is only modestly harmed (relative to any complying retailers) if it fails. Such an incentive

to mislead could make communication problematic. If the spokes believe the hub is always

inclined to say that all other spokes have agreed to collude (whether or not it is true) then

the hub’s announcements would be uninformative to the spokes; hence, the hub would b e

unlikely to succeed in achieving mutual understanding among the spokes. Or suppose a

retailer communicated to the manufacturer that it would like all retailers to raise prices by

15% but the manufacturer prefers a 10% increase. Given it is the information gatekeeper,

the manufacturer could convey to the other retailers that this retailer’s pricing intention is

to raise price by 10%. More generally, given its strategic position in the communication

Examples in which a neutral third party was hired to assist a cartel include a taxi driver for a bidding

ring at stamp auctions (Asker, 2014) and a consulting company, AC Treuhand, for the organic peroxides

cartel (Marshall and Marx, 2012). In those cartels, …rm directly communicated. The role of the third party

was to assist in implementing and monitoring the collusive outcome and organizing meetings among cartel

members.

Members of regular cartels often have di¤erent preferences regarding the extent of the price increase,

the type of market allocation scheme, and, of course, the particular market allocation.

Crawford and Sobel (1982) prove that the informativeness of messages is reduced when the preferences

of the sender of a message are less aligned with the preferences of the receiver of the message.

This result is proven in Ambrus, Azevedo, and Kamada (2013), and holds as long as parties do not

randomize their actions.

network, the hub may choose to distort communications between the spokes in order to push

the collusive outcome in a direction that it favors. As we’ll see, there are cases in which the

hub used its position to implement an outcome that bene…tted it and quite possibly harmed

the spokes. In sum, mutual understanding among …rms is expected to be more problematic

when they indirectly communicate because each …rm is not sure what other …rms are saying

and hearing, and the upstream or downstream …rm that is intermediating may have an

incentive to engage in acts of omission or deception.

Thus far, our discussion has focused on how hub-and-spoke collusion is less e¤ective than

the standard cartel. However, there are ways in which the presence of a hub can make

collusion more e¤ective. First, the credibility of a hub’s messages may be enhanced due to

its relations with a spoke which go beyond participating in a collusive arrangement, and

that could act as a countervailing factor to the forces for incredulity just described. From

its previous business dealings with a spoke, a hub may have established some credibility

which it can draw upon. With regards to future business dealings with a spoke, a hub will

want to maintain a reputation for veracity and that will make its messages more credible. In

contrast, direct communications between …rms in a market is not common in a competitive

environment. This means that the members of a standard cartel cannot draw on an existing

reputation and incentives for maintaining an reputation when they communicate with each

other, while an upstream supplier or downstream buyer can do so in the context of a hub-

and-spoke cartel.

Second, as an upstream supplier or downstream customer, the hub may be in a better

position to monitor for compliance. In conducting normal business practices, the hub is regu-

larly communicating with the spokes, whether it is to take orders (in the case of an upstream

supplier) or to place orders (in the case of a downstream buyer). Those interactions provide

opportunities to collect information useful for monitoring. For example, when visiting stores,

it would be easy for an upstream supplier to observe the prices that a retailer is charging

and thus whether it is setting collusive prices. In addition, if the spokes are engaging in their

own monitoring, the hub can collect that information when interacting with the spokes. If

a downstream buyer learns from several upstream suppliers that a particular upstream sup-

plier is deviating, that provides a compelling basis for implementing a punishment against

that upstream supplier. In sum, given its vertical relations with the spokes, the hub is well

situated to observe and collect information relevant to monitoring and, with more e¤ective

monitoring, comes more e¤ective collusion.

Finally, the hub may possess instruments to induce spokes to comply with the collusive

outcome.

One instrument is the refusal to sell or buy from a spoke. If the hub is an

upstream supplier, it could threaten to refuse to supply downstream retailers who are un-

dercutting the collusive price. This would help stabilize collusion in two ways. First, the

prospect of foregone pro…t due to not having the supply to sell will help induce retailers to

comply. Second, in response to a retailer undercutting the collusive price, compliant …rms

are inclined to lower their prices to stave o¤ the loss of demand. However, if the deviating

retailer doesn’t have the supply to fully meet the additional demand then the demand loss

to compliant …rms is lessened which could induce them to continue to charge the collusive

price. Analogously, if the hub is a dominant downstream retailer then it could threaten to

For a broader discussion on how vertical restraints can aid collusion, see Levenstein and Suslow (2014).

refuse to purchase from noncompliant upstream suppliers.

A second instrument is the use

of side payments to stabilize collusion. The hub could o¤er monetary transfers to those …rms

that comply - for example, in the form of rebates - which would provide additional incentives

for the spokes to abide by the collusive arrangement. While monetary transfers could occur

between …rms in a regular cartel, they are more likely to create suspicions than if they were

from a supplier or customer.

This section has discussed several reasons for expecting hub-and-spoke collusion to op-

erate di¤erently from when …rms directly communicate and do not involve an upstream

supplier or downstream buyer. Let us now turn to examining our nine hub-and-spoke cartels

and assess to what extent they are di¤erent.

3 Cases of Hub-and-Spoke Collusion

Our coverage encompasses nine hub-and-spoke cartels in the markets for movie exhibition

(U.S.), toys (U.S., UK), paints and varnishes (Poland), pharmaceutical products (U.S.),

replica football kits (UK), cheese (UK), drugstore, perfumery, and hygiene products (Bel-

gium), and e-books (U.S.). For each case, I will describe the market, the collusive scheme, the

communication process through which …rms coordinated and monitored, and the disposition

of the legal case.

3.1 Movie Exhibition

The case concerns the distribution and exhibition of motion pictures.

The market for the

exhibition of movies included …rst-run theaters where newly released …lms were shown, and

subsequent-run (or second-run) theaters where those same …lms were shown for a lower ad-

mission price at a later date. Interstate Circuit operated 43 theaters and Texas Consolidated

operated 66 theaters, which were a mix of …rst-run and subsequent-run movie houses. The

two chains dominated the cities in which they had theaters, including having a monopoly

position in many of them.

As expressed by Interstate to the eight motion picture distributors, the subsequent-run

theaters (not owned by Interstate or Texas Consolidated) were competing too aggressively.

While the …rst-run theaters were showing a single …lm for an admission price of 40 cents,

many subsequent-run theaters were charging a vastly lower price and often showing two

…lms.

Let me note that a standard cartel could elicit the assistance of an upstream supplier to enforce a

collusive agreement. For example, in JTC Petroleum Co. v. Piasa Motor Fuels, Inc., 190 F.3d 775 (7th Cir.

1999), six road contractors formed a cartel to rig bids at local government tenders and, in the usual manner,

directly communicated. One of the contractors, JTC, did not abide by the collusive arrangement, so the

cartel enlisted three emulsi…ed asphalt producers to refuse to supply this essential input to JTC. Nevertheless,

most standard cartels do not involve upstream or downstream …rms and thus lack their assistance in making

collusion more e¤ective.

An exception is when competing …rms engage in inter-…rm sales as part of normal business practice. Such

sales allowed …rms to consummate transfers in the citric acid, lysine, and vitamins cartels; see Harrington

(2006) and Harrington and Skrzypacz (2011).

Interstate Circuit, Inc., et al. v. United States 306 U.S. 208 (1939)

Id. at 215.

In seventeen of the eighteen independent [subsequent-run] theatres ...the ad-

mission price was less than 25 cents. ... In most of them the admission was 15

cents or less. It was also the general practice in those theatres to provide double

bills either on certain days of the week or with any feature picture which was

weak in drawing power.

As this sti¤ competition was eating into admission sales at its …rst-run theaters, Interstate

sought to coordinate the motion picture distributors on a plan to control these independent

subsequent-run theaters. Hence, the movie exhibitor was acting as the hub to the movie

distributors who were the spokes.

A manager of Interstate initially engaged in bilateral communications with each of the

branch managers of the eight distributors. His plan had distributors require subsequent-

run theaters to charge no less than 25 cents for admission and to show only one …lm. If a

distributor did not follow that policy then Interstate threatened not to show the distributor’s

…lms in its …rst-run theaters. Clearly, this plan would bene…t Interstate, as it would raise the

price and lower the quality of rival exhibitors. It is less clear that it would bene…t distributors.

By raising the prices for subsequent-run theaters, more demand would be generated for …rst-

run theaters which had a higher pro…t margin. While that is likely to raise total industry

pro…ts, it is not immediate that the distributors would end up with more pro…ts. It depends

on the relative bargaining positions of the movie exhibitors and the movie distributors. It

is important to note here that the distributors’ participation does not require that they

are collectively made better only. It is su¢ cient that it is in the best interests of a movie

distributor to participate if it believed that the other distributors would participate, and it

believed Interstate’s threat of punishment for noncompliance. Let us consider a distributor’s

incentives to go along with the exclusionary plan.

To begin, a distributor would be unlikely to comply unless it thought that su¢ ciently

many of the other distributors were planning to comply. For suppose instead that most

distributors did allow subsequent-run theaters to show their …lms at an admission price less

than 25 cents. If a distributor went along with Interstate’s scheme and only licensed its …lms

to subsequent-run theaters that charged at least 25 cents, subsequent-run theaters would

choose to exhibit the …lms of other distributors in order to be able to charge a price that was

competitive in the market. Hence, a compliant distributor could lose signi…cant demand if

the other distributors were noncompliant. Furthermore, the credibility of Interstate’s threat

of punishment for noncompliance - not showing the distributor’s …lms in its theaters - would

be put into question if many distributors did not go along. For Interstate may be willing to

go through with that threat against one or a few distributors but it would be very costly for it

to do so with many distributors, as then there would be many …lms it would not be showing

in its theaters. Thus, the credibility of Interstate’s punishment would be put into question

unless a distributor believed many other distributors were to comply. In sum, a distributor

would probably not comply if most other distributors did not comply because it would lose

signi…cant demand, and Interstate’s retaliatory threat would not be credible. However, if

all other distributors complied then a distributor would be at no competitive disadvantage

from complying, and Interstate’s threatened punishment for noncompliance could have been

credible.

Id. at 217-8.

For Interstate’s scheme to work, it must inform each distributor of its plan and have each

distributor believe that the other distributors were planning to go along with it. For that

purpose, Mr. O’Donnell, the manager of Interstate and Consolidated, sent out a letter to

each of the eight distributors with all of them named as addressees. That letter described the

plan to impose restrictions on Interstate’s competing exhibitors and the threatened response

of Interstate if a distributor did not participate. Here are some relevant excerpts from that

letter:

Interstate Circuit, Inc. will not agree to purchase produce to be exhibited in

its ‘A’ theatres at a price of 40c/ or more for night admission, unless distributors

agree that in selling their product to subsequent runs, that this ‘A’ product will

never be exhibited at any time or in any theatre at a smaller admission price than

25c/ for adults in the evening. In addition to this price restriction, we also request

that on ‘A’pictures which are exhibited at a night admission price of 40c/ or more

- they shall never be exhibited in conjunction with another feature picture under

the so-called policy of double-features. ... In the event that a distributor sees …t

to sell his product to subsequent runs in violation of this request, it de…nitely

means that we cannot negotiate for his product to b e exhibited in our ‘A’theatres

at top admission prices.

The letter was followed with bilateral meetings between Mr. O’Donnell and each distributor.

The U.S. Supreme Court viewed the letter as instrumental in achieving mutual under-

standing among distributors.

The O’Donnell letter named on its face as addressees the eight local repre-

sentatives of the distributors, and so from the beginning each of the distributors

knew that the proposals were under consideration by the others. Each was aware

that all were in active competition and that without substantially unanimous

action with respect to the restrictions for any given territory there was risk of

a substantial loss of the business and good will of the subsequent-run and inde-

pendent exhibitors, but that with it there was the prosp ect of increased pro…ts.

There was, therefore, strong motive for concerted action, full advantage of which

was taken by Interstate and Consolidated in presenting their demands to all in

a single document.

The Court concluded there was an unlawful agreement in spite of the lack of direct commu-

nication among the movie distributors.

It was enough that, knowing that concerted action was contemplated and

invited, the distributors gave their adherence to the scheme and participated in

it. Each distributor was advised that the others were asked to participate; each

knew that coop eration was essential to successful operation of the plan. They

knew that the plan, if carried out, would result in a restraint of commerce, ...

and knowing it, all participated in the plan.

See Interstate Circuit, Inc., et al. v. United States 306 U.S. 208, 216 n.3 (1939)

Id. at 222.

Id. at 227.

In sum, Interstate was a case in which collusion was initiated by the downstream (ex-

hibitor) …rm in response to aggressive competition from rival downstream …rms. The col-

lusive plan was exclusionary in that it proposed the upstream (distributor) …rms require

Interstate’s rival …rms to raise their prices and lower their quality as a condition of having

the upstream …rms’products. An upstream …rm would …nd it optimal to go along only if

it believed many other upstream …rms were to do so, which then required the downstream

…rm to achieve mutual understanding of compliance among the upstream …rms. In the ab-

sence of direct communication between the upstream …rms, the downstream …rm sought to

obtain mutual understanding among them by sending a letter with the collusive plan to all

upstream …rms while noting in the letter that it was sent to all upstream …rms. The letter

was supplemented by bilateral verbal communications in order to achieve enough con…dence

among upstream …rms that other upstream …rms were likely to comply. The Court found

the conduct illegal as it went beyond a vertical contract between one upstream …rm and one

downstream …rm. Rather, the downstream …rm sought to have coordinated adoption of the

collusive plan by the upstream …rms. Informing each upstream …rm that the other upstream

…rms were being asked to comply created a horizontal dimension to this conduct, and was

essential for Interstate’s plan to succeed because it would not have been in the interest of an

upstream …rm to comply unless it thought enough other upstream …rms were to do so.

3.2 Toys (U.S.)

The dominant U.S. toy retailer in the early-mid 1990s was Toys "R" Us (hereafter TRU)

with a 20% national market share.

In 18 metropolitan areas, its market share ranged

between 35 and 49%, and was over 50% in eight metropolitan areas and Puerto Rico. In the

upstream market, the leading toy manufacturers were Mattel (with a U.S. market share of

18%), Hasbro (17%), Tyco (3.2%), and Little Tikes (2.8%). For most of those manufacturers,

TRU was its most important customer, as it made up 28% of Mattel’s sales, 28% of Hasbro’s,

31% of Little Tikes’, 35% of Fisher-Price’s, and 48% of Tyco’s.

At the time, the retail landscape was disrupted by warehouse clubs which sold at lower

prices than TRU though in a more sparse, less attractive selling environment. Clubs included

Sam’s Club (a division of Wal-Mart with 256 stores in 1992), Pace (a division of Kmart with

115 stores), Costco (100 stores), Price Club (94 stores), and BJ’s Wholesale (39 stores). The

clubs sold fewer toys than TRU though could stock any product of a toy manufacturer. In

addition to selling the same products that were found in TRU stores, the clubs also worked

with toy manufacturers to create specially-packaged pro ducts that sold at a higher total

price but a lower unit price, so they provided better value.

With a gross margin of around 25%, TRU had entered the toy market as the low-price

retailer when compared to department stores (with gross margins of 40-50%) and small toy

stores. It was forced to defend that status against discount department stores like Wal-Mart

Facts are from In re Toys "R" Us, 126 F.T.C. 415 (F.T.C. September 25, 1997) (initial Decision, James

P. Timony). (hereafter, Toys "R" Us 1997); Toys "R" Us, Inc. v. FTC, 126 F.T.C. 415 (F.T.C. October

14, 1998) (opinion of the Federal Trade Commission, Chairman Robert Pitofsky) (hereafter, Toys "R" Us

1998); and Toys "R" Us, INC. v. F.T.C., 1999 U.S. 7th Cir. Briefs (on Petition for Review of a Final Order

of the Federal Trade Commission. Opinion of the Commission: Chairman Robert Pitofsky.) (hereafter, Toys

"R" Us 1999).

and Kmart, but the clubs were a more serious threat with gross margins of 9-12%.

By 1989, TRU senior executives were concerned that the clubs presented a

threat to TRU’s low-price image and its pro…ts. TRU knew that consumers form

opinions about a store’s relative prices based on a few visible items. TRU referred

to these products as “price image”or “price sensitive”items. ... TRU had already

lowered the prices of these popular items to meet Wal-Mart’s challenge, but the

clubs’marketing strategies threatened to bring prices even lower.

As of mid-1992, TRU reported that a warehouse club was within a …ve-mile radius of 238 of

its 497 U.S. stores.

TRU’s concern was that a customer, upon comparing the prices of TRU and a club for

the same item, would learn that TRU’s price was higher; no longer would TRU be seen as

the low-price outlet for toys. One strategic response would have been to lower prices, but

that would have cut into pro…t margins and could have fueled a price war with the clubs.

An alternative path was for TRU to avoid such price comparisons by preventing clubs from

carrying any toy that TRU carried. Of course, implementation of that strategy required the

participation of the toy manufacturers.

In a document drafted around Toy Fair 1993, Greg Staley from TRU’s inter-

national division summarized TRU’s policy as follows: "Our buying is simple -

we will not carry an identical item which is sold to a Warehouse Club. If we …nd

an item in both our assortments and those of a Club, we will discontinue carrying

that item immediately; and we reserve the right to take clearance markdowns to

dramatically accelerate the rate of sale on that item. In summary, the vendor

has to make a choice as to whom he sells an item - either us or them."

Thus, the strategy was one of exclusion - harm rival …rms by limiting the products they are

o¤ered by manufacturers. TRU was going to have manufacturers sell to the clubs only those

products that TRU did not stock, and it was going to induce the manufacturers to go along

by threatening not to carry their products.

While TRU would clearly be better o¤ with this collusive plan, what about the toy

manufacturers? If clubs were paying a lower wholesale price than TRU and the increase

in sales from the lower retail price at the clubs was insu¢ cient to o¤set the lower margins

being earned by manufacturers then collusion could be more pro…table for the manufacturers.

However, if the clubs paid the same, or almost the same, wholesale price as paid by TRU

then the manufacturers would be worse o¤ with TRU’s plan due to lower total sales from

excluding the clubs.

While we do not have the data to address the question of whether TRU’s exclusionary

plan made the toy manufacturers better or worse o¤, the documentary evidence reveals that

the manufacturers expressed an appeal to selling to the clubs; it was a source of growing

sales and made them less dependent on TRU. Playskool’s president noted "that his company

could not stop doing business with the clubs, and that in view of the consolidation in the

See Toys "R" Us 1998 at 427.

See Toys "R" Us, 1997 at 436-7.

retail trade it was important for Playskool to have other customers than TRU."

If indeed

the toy manufacturers liked having the clubs as an outlet for their products, that would pose

a challenge to TRU getting those manufactures to exclude the clubs.

A second challenge for TRU is that a toy manufacturer was concerned that it would be

at a disadvantage if it restricted its sales to the clubs and many other manufacturers did

not.

The toy companies were afraid of yielding a potentially important new chan-

nel of distribution to their competitors. Small changes in sales volumes have a

signi…cant e¤ect on toy manufacturers’overall pro…ts, and no retail channel other

than the clubs o¤ered similar opportunities for rapid growth. For example, Mat-

tel’s sales volume to the clubs increased by 87% between 1989 and 1991. Much

of this growth was a result of Sam’s emergence as a toy buyer, but sales to BJ’s,

Costco and Pace also increased at a rapid rate. By comparison, Mattel’s overall

sales grew by approximately 10% during this period.

Unilaterally restricting sales to clubs was not in the best interests of a toy manufacturer.

As argued above, a co ordinated restriction of sales by toy manufacturers may not have been

in their interests either. What made it sensible for a toy manufacturer to comply with the

collusive plan was TRU’s threat of exclusion if it did not.

While most – if not all – of the toy companies disliked having to choose

between what they saw as two bad options – (1) sell to TRU and restrict club

sales, or (2) sell to the clubs and risk retaliation from TRU – the decision was

made easier by the horizontal agreement which took the sting out of reducing

sales to the clubs. From the manufacturers’point of view, the boycott was the

second-best alternative.

An individual toy manufacturer found it better to comply than not, but only as long as other

toy manufacturers complied.

Finally, TRU faced a third obstacle to getting the toy manufacturers on board. It was

unclear how credible was TRU’s threat to cut o¤ purchases from a toy manufacturer if it did

not comply. Doing so would certainly harm TRU in the short run. The credibility of the

threat also hinged on most or all toy manufacturers complying. If all other toy manufacturers

complied, TRU might …nd it optimal to cut o¤ one toy manufacturer in order to establish

the credibility of its threat and keep other toy manufacturers from breaking ranks. However,

if several toy manufacturers did not comply, it would be very costly for TRU not to stock

their toys. Thus, a toy manufacturer could well believe that TRU would punish it for not

complying only if most or all of the other toy manufacturers were complying.

For all of these reasons, TRU’s collusive plan would be implemented by the toy man-

ufacturers only if each believed their rivals had bought into it. Only then would a toy

manufacturer not be at a competitive disadvantage from restricting supply to the clubs and

only then would it …nd TRU’s exclusionary threat to be credible. The toy manufacturers

clearly conveyed the need for assurances that all would be participating.

Id. at 448.

See Toys "R" Us 1998 at 552.

Toys "R" Us, Inc. v. FTC, 126 F.T.C. 415, 585 n.49 (F.T.C. October 14, 1998)

Mattel, Hasbro, Tyco, Little Tikes, Fisher-Price and others all wanted to know

how competitors were reacting to TRU. The manufacturers wanted assurances

from TRU that their competitors were subject to the same rule. They informed

TRU that they wanted a level playing …eld to avoid being placed at a competitive

disadvantage.

The challenge to TRU was clear: Convince each toy manufacturer that the other man-

ufacturers were going to restrict sales to the clubs. It met this challenge using bilateral

communications with each toy manufacturer and making each aware of its bilateral commu-

nications with rival toy manufacturers. There is no evidence that TRU ever communicated

to them as a group.

[TRU] tried to obtain a coordinated response from manufacturers by assuring

them that they would not be placed at a competitive disadvantage because TRU

was applying its policy to their competitors. ... The manufacturers all were aware

that TRU was communicating its policy to everyone and that uniformity was

contemplated. And everyone knew that without unanimity regular line product

sales to the clubs would recommence.

A key early contact was Hasbro, for whom TRU was its biggest U.S. customer.

In the fall of 1990, TRU’s CEO, Charles Lazarus, met with Hasbro’s execu-

tives and told them that the clubs were a threat to TRU because of their low

prices. He said that if Hasbro continued to aggressively supply the clubs ... that

this could a¤ect their business at TRU.

TRU went to each major toy manufacturer with a similar pitch, and each was informed that

their rivals were similarly being pressured to restrict sales to the clubs.

During conversations with manufacturers, TRU did not merely announce that

it would refuse to deal with manufacturers selling to the clubs, or inform man-

ufacturers that all manufacturers would be treated equally. Instead, TRU com-

municated the quid pro quo (i.e., I’ll stop if they stop) from manufacturer to

manufacturer.

[TRU vice president Roger] Goddu clari…ed that TRU engaged in these con-

versations with all the key toy manufacturing …rms. “We communicated to our

vendors that we were communicating with all our key suppliers, and we did that

I believe at Toy Fair 1992. We made a point to tell each of the vendors that we

spoke to that we would be talking to our other key suppliers.”

Furthermore, TRU would use one toy manufacturer’s acceptance to induce others to come

on board with the policy.

See Toys "R" Us 1997 at 436-7.

Id. at 430.

Id. at 432.

Id. at 447.

See Toys "R" Us 1998 at 555.

After Mattel agreed not to sell to the clubs the same products "based on the

fact that competition does the same", TRU told Hasbro that Mattel had agreed.

... Before committing not to sell certain products to the clubs, Little Tikes asked

TRU what its main competitor in the clubs (Today’s Kids) was going to do.

Goddu informed Little Tikes that Today’s Kids "was going to start doing less

business with the warehouse clubs" whereupon Little Tikes committed to restrict

its sales. ... Lazarus and Goddu told Sega that TRU had convinced Nintendo to

stop selling product to the clubs as part of TRU’s e¤ort to convince Sega to do

the same. TRU argued that Sega should stop selling b ecause TRU had convinced

Nintendo to stop.

Just before or at Toy Fair 1992, Hasbro’s then western regional sales man-

ager, James Inane, met with [Hasbro CEO] Verrecchia [who] said that he had just

come from a meeting with TRU, that TRU had met with Hasbro’s competitors,

including Mattel and Fisher-Price, and that they had agreed not to sell promoted

products to the clubs. Verrecchia said that because Hasbro’s competitors had

agreed not to sell promoted products, Hasbro would go along with the agree-

ment. Verrecchia told his sta¤ that Hasbro would not sell promoted products to

the clubs and that Hasbro would watch other manufacturers’sales to the clubs.

Hasbro would refrain from selling to the clubs until another manufacturer broke

the agreement.

As the FTC referred to it, TRU engaged in "shuttle diplomacy" by

reassuring each toy manufacturer that rivals would fall into line. It was only

after assurances were exchanged that the toy manufacturers, overcoming their

natural inclination to sell through all potential outlets, became willing to discrim-

inate against the clubs. At that point, a "conscious commitment to a common

scheme" was perfected, and a uniform, clearly interdependent, course of conduct

came into being.

TRU worked for over a year and surmounted many obstacles to convince the

large toy manufacturers to discriminate against the clubs by selling to them on

less favorable terms and conditions. The biggest hindrance TRU had to overcome

was the major toy companies’ reluctance to give up a new, fast-growing, and

pro…table channel of distribution, and their concern that any of their rivals who

sold to the clubs might gain sales at their expense. TRU’s solution was to build

a horizontal understanding – essentially an agreement to boycott the clubs –

among its key suppliers.

While there was some direct communications between the toy manufacturers,

it was

rare. The almost-exclusive channel for communication between toy manufacturers was

See Toys "R" Us 1997 at 432-4.

Id. at 449.

See supra note 19, at 586.

Id. at 552.

"In May of 1992, at a toy manufacturers’conference, Hasbro’s CEO Allan Hassenfeld discussed with

Tyco’s CEO Richard Grey what each company was doing or not doing with respect to the clubs." See supra

note 20, at 450.

through TRU, and it achieved the needed mutual understanding among the spokes. The

FTC concluded that many of the toy manufacturers

required assurances that rivals would sell on discriminatory terms to the clubs,

and ... were satis…ed with TRU’s assurances that such uniform policies would

be adopted. Evidence of that exchange of commitments –not necessarily direct

communications among the toy manufacturers but clearly through the interme-

diation of TRU –is present with respect to Mattel, Hasbro, Fisher Price, Tyco,

Little Tykes, Today’s Kids, and Tiger Electronics.

However, the job was not done. Even though many toy manufacturers had agreed to

TRU’s plan, they were continuously concerned with possible deviations by rival …rms. In

response, TRU actively monitored the toy manufacturers for compliance, and the toy man-

ufacturers themselves were instrumental in reporting a rival that supplied clubs outside of

the agreement. Figure 2 provides a list of infractions that was part of a memo to the CEO

of TRU. It states the noncompliant toy manufacturer and the product that was sold to the

clubs, and often o¤ers an explanation for the apparent infraction and a plan to rectify the

conduct.

The toy manufacturers had strong incentives to monitor and share any infractions with

TRU so that rival …rms would be brought into compliance.

[W]hen Mattel heard rumors that Hasbro and Tyco might be selling regular

line to the clubs, the president of Mattel’s Boy Division instructed that the clubs

be shopped and the information sent to TRU.

TRU promised to "take care of it" after Fisher-Price representatives com-

plained about Playskool product they found in Price Club.

In September 1991, Fisher-Price’s regional manager sent [Fisher-Price sales-

man John] Chase a copy of a TRU shopping report showing products of Hasbro,

Fisher-Price and Playskool found in Price Club. He told Chase that a TRU

executive had sent the report to Byron Davis, Fisher-Price’s vice-president for

sales. The words "Byron, you promised this wouldn’t happen" were written on

the report. After this event, Fisher-Price limited its club sales to special and

combination packs.

TRU admitted that it acted as a conduit between …rms regarding complaints of noncompli-

ance.

TRU’s President testi…ed: “I would get phone calls all the time from Mattel

saying Hasbro has this in the clubs or Fisher Price has that in the clubs.... So that

occurred all the time.”Goddu explained that, on the many occasions he received

these calls, he would “always thank them and tell them we would follow up.”...

See supra note 19, at 575.

See supra note 20, at 431.

Id. at 434.

Id. at 454.

TRU would speak to the o¤ending …rm and even assure the complainant that the

o¤ending …rm would be brought into line. Violations of TRU’s club policy were

thus detected and punished, serving to enforce the horizontal agreement. The toy

companies participated in this exchange of complaints, which was frequent and

continued over lengthy periods, e¤ectively making their competitors’compliance

a part of their agreements with TRU.

In sum, TRU had two sources of information to aid in monitoring for compliance with the

collusive scheme. One source was TRU shopping the clubs, and the second was reports from

toy manufacturers.

To sum up, let me utilize the FTC’s succinct synopsis of the case.

The record demonstrates that TRU organized and enforced a horizontal agree-

ment among its various suppliers. Despite TRU’s considerable market power, key

toy manufacturers were unwilling to refuse to sell to or discriminate against the

clubs unless they were assured that their competitors would do the same. To over-

come that resistance, TRU gave initial assurances that rival toy manufacturers

would commit to comparable sales programs; TRU representatives then acted

as the central player in the middle of what might be called a hub-and-spoke

conspiracy, shuttling commitments back and forth between toy manufacturers

and helping to hammer out points of shared understanding; toy manufacturers’

commitments were carefully conditioned on comparable behavior by rivals; and,

after the discriminatory program was in place, TRU and the toy manufacturers

worked out a program to detect, bring back into line, and sometimes discipline,

manufacturers that sold to the clubs.

The evidence is compelling that TRU’s collusive strategy of exclusion was e¤ective in

reducing toy manufacturers’ sales to the clubs. Mattel and Hasbro’s sales to the clubs

dropped from $32.5 million in 1991 to $10.7 in 1993.

The clubs’share of U.S. toy sales,

which had risen from 1.5% to 1.9% during the pre-boycott years of 1991-92, fell to 1.4% by

1995.

TRU’s internal documents revealed that, while the clubs were a competitive threat

as of late 1992, they were not considered as such by 1993:

In December of 1992, TRU included clubs located near TRU stores when it

calculated its [competition] index. TRU explained this decision by noting that

“[w]arehouse clubs have been a strong competitive force this season.”Clubs were

withdrawn from later competition indices in 1993 –after TRU’s club policy was

put into e¤ect – because clubs were then thought to have “no signi…cant . . .

impact on TRU stores.”

See supra note 19, at 558.

Id. at 575.

Toys "R" Us 1999 at 18.

See supra note 19, at 600.

Toys "R" Us, Inc. v. FTC, 126 F.T.C. 415, 539 n.15 (F.T.C. October 14, 1998)

The reduced supply of the clubs translated into higher prices for consumers. Products

sold by TRU but not by discounters had margins as high as almost 40%.

By expanding

the set of products exclusive to TRU, it would have expanded the set of products for which

it could set a high margin. For 1993, the FTC "found that the elimination of competitive

pressure from the clubs cost consumers as much as an extra $55 million per year on top-

selling products purchased at TRU alone."

As another measure of e¤ect, TRU and three

toy manufacturers settled private litigation for $56 million.

The Administrative Law Judge (ALJ) ruled that TRU and 14 toy manufacturers had

engaged in a per se violation of Section 5 of the FTC Act. In response to the increased

competition from warehouse clubs, the ALJ noted that "TRU could have announced a uni-

lateral policy by TRU and a refusal to deal with suppliers that did not comply. [T]he

issue is whether TRU went further, entering agreements with each manufacturer."

The

ALJ concluded that "TRU orchestrated a horizontal conspiracy among its suppliers [and

t]he major manufacturers knew that TRU was contacting the other manufacturers with the

same prop osal and that concerted action was invited."

TRU appealed the ALJ’s decision

to the Commission, which was denied. TRU then appealed that decision to the Seventh

Circuit Court which a¢ rmed the FTC’s decision and, in doing so, viewed the case as more

compelling than Interstate.

The Commission’s theory, stripped to its essentials, is that this case is a

modern equivalent of the old Interstate Circuit decision. ... [T]he TRU case if

anything presents a more compelling case for inferring horizontal agreement than

did Interstate Circuit, because not only was the manufacturers’decision to stop

dealing with the warehouse clubs an abrupt shift from the past, and not only is

it suspicious for a manufacturer to deprive itself of a pro…table sales outlet, but

the record here included the direct evidence of communications that was missing

in Interstate Circuit.

In Interstate, the key piece of evidence was the letter sent by the hub (Interstate Circuit) to

the spokes (movie distributors) which indicated that all spokes were receiving it. A spoke’s

inference that other spokes were intending to comply would have only been based on knowing

that the other spokes had received the letter. With Toys "R" Us, there was instead bilateral

verbal communications b etween the hub (TRU) and the spokes (toy manufacturers) which

not only encouraged a spoke to participate but made clear that other spokes were similarly

being encouraged and that some had already agreed to participate.

As in Interstate Circuit, there was an invitation clearly addressed to all of the

participants in the proposed conspiracy. Like the listing of all the …lm distributors

as addressees in the letter sent by Interstate Circuit, TRU, in Goddu’s phrase,

“made a point of telling”its suppliers that its club “policy”was to be extended

Id. at 530.

See supra note 40, at 30. Scherer (2004) provides some of the economic analysis.

Scherer (2004)

See supra note 20, at 147.

Id. at 162.

Toys "R" Us v. FTC, 221 F.3d 928, 935 (7th Cir. 2000)

to each and every one of them. Each therefore knew that the others were asked

to make a similar decision.

And there was more in Toys "R" Us, for the spokes made clear in their communications

with the hub that their participation depended on the other spokes participating. We then

have express communication by a spoke to the hub that it needed a joint action among

manufacturers. That is something which only could only b e inferred by the hub in Interstate.

3.3 Toys (UK)

The previous two cases had a downstream …rm responding to competition in its market by

recruiting upstream …rms to coordinate on an exclusionary action against rival downstream

…rms. Coordination was achieved through bilateral communications between the downstream

hub and each of the upstream spokes. In Interstate, upstream movie distributors imposed

higher prices and lower quality on the downstream hub’s competitors in the movie exhibition

market, and in Toys "R" Us upstream toy manufacturers limited their product o¤erings to

warehouse clubs that were competing against downstream retailer Toys "R" Us.

We now turn to a more common form of hub-and-spoke collusion: The hub is an upstream

manufacturer or wholesaler, the spokes are retailers, and collusion has the downstream …rms

raising retail prices. Interestingly, the …rst case we’ll consider is the reverse of Toys "R" Us

in that a toy manufacturer is the hub and toy retailers are the spokes.

The setting is the London toy retailing market where the top chains are Argos and

Littlewoods (which is also referred to as Index).

A key upstream …rm was Hasbro which

was one of the largest toy and game manufacturers in the United Kingdom. Hasbro learned

in 1998 that retailers were "unhappy with the margins they were receiving on Hasbro’s

branded products."

A meeting was held among key Hasbro employees in October 1998 to

develop initiatives to raise margins. As noted by Hasbro Sales Director David Bottomley:

"The listing and pricing initiatives came about as a result of low margins that were a concern

across the entire industry and shared by Argos and Littlewoods."

The "pricing initiative" was for retailers to charge the recommended retail price (RRP).

The "listing initiative" was for Hasbro to o¤er rebates to retailers to induce them to con-

tinue stocking certain Hasbro products that were in jeopardy of being removed from a toy

retailer’s o¤erings. The pricing initiative would clearly bene…t retailers. Presumably, the two

initiatives must have bene…tted Hasbro given that it was the …rm responsible for devising

and promoting them. These initiatives were developed by Hasbro employees Ian Thomson

(Account Manager for Littlewoods) and Neil Wilson (Account Manager for Argos). They

were also supported at a senior level by Mike McCulloch, the Head of Marketing and Sales,

and Sales Directors David Bottomley and Mike Brighty.

Our focus will be on the pricing initiative for it was the one that required Hasbro to

achieve mutual understanding between Argos and Littlewoods. While there were other toy

See supra note 19, at 586.

Ensuring facts are from CA/98/8/2003 Agreements between Hasbro UK Ltd, Argos Ltd & Littlewoods

Ltd …xing the price of Hasbro toys and games, [2004] 4 UKCLR 717.

See Argos Ltd & Anor v. O¢ ce of Fair Trading [2006] EWCA Civ 1318 [114].

Id. at { 98.

retailers in the market, Bottomley of Hasbro felt that "Argos and Littlewoods were key to

the success of the pricing initiative since they were the market leaders - if they could be

persuaded to maintain prices at RRP then other retailers would follow suit."

However,

getting both of them to charge the RRP required each being assured the other would do so.

Argos is generally accepted as the price setter and leader in the market.

However, Hasbro considered that Argos would have been very unlikely to make a

commitment to follow Hasbro’s RRPs unless it was reassured that doing so would

not result in its catalogue prices being undercut by those in the Index catalogue

[of Littlewoods]. Littlewoods is the main catalogue competitor to Argos. ...

Argos and Littlewoods monitor, in particular, each others’ prices very closely

and produce regular analyses showing how often each undercuts and is undercut

by the other. Since b oth companies o¤er a price-match guarantee, neither can

a¤ord to have prices that are seriously out of step with the other. It was therefore

necessary to reassure Argos that Littlewoods would also be committed to RRPs.

For its part Littlewoods required the same assurance of commitment by Argos.

This concern of mutual compliance was especially acute with their catalog prices. As catalogs

were issued only twice a year, a retailer could …nd itself at a price disadvantage for six months

if it complied with the RRP and the other retailer did not.

Both Argos and Littlewoods were concerned about undercutting by any re-

tailer, but each had a special concern about undercutting by the other. This was

because they were the largest catalogue retailers, directly competing with each

other, and because their retailing formats meant that they both had to commit

themselves to a price for a forthcoming season without knowledge of the other’s

intention except for the previous catalogue which was, by de…nition, out of date.

Further, unlike with ordinary retailers where an agreement to price at X could

be given public e¤ect on the next day or within a very short space of time, any

“agreement”or “understanding”that the other catalogue retailer would price at

an agreed price (say RRP) would not be seen to be implemented until much later

when it would be to o late to change one’s own catalogue.

Initially, Hasbro focused the pricing initiative on its core games and Action Man product

line. These were high-volume well-advertised products for which pricing had been particu-

larly aggressive. Starting in late 1998, members of Hasbro’s sales team were speaking with

Argos’ buyers about having retailers price at the RRPs. The initial reaction of the retailers

was not encouraging.

Littlewoods was concerned about the feasibility of Hasbro’s pricing initiative

and in particular expressed doubts about Hasbro’s ability to prevent undercutting

by Argos. Ian Thomson states in his witness statement: "It was at this point

Id. at {48.

Id. at { 47.

Id. at { 96.

that Mike McCulloch intimated ... that he had been having discussions with the

major opposition (Argos) and they were of the same opinion i.e. that they could

not agree to the new pricing structure for fear of being undercut. It did need

the agreement of both parties in order for the plan to work, but that if Index

would agree to go along with it then Mike McCulloch, using this knowledge,

was con…dent that he could persuade them to do the same. John McMahon

[of Littlewoods] said that he would play ball and go along with the plan but

if they (Argos) reneged on the deal and did not stick to the retail prices in

their 1999 Autumn Winter Catalogue and he (Index) did, he would be seriously

disadvantaged. If this happened as a result he would do some serious price cutting

in the next Index catalogue launch.

Intent on surmounting such skepticism, Hasbro engaged in extensive bilateral communi-

cations with Argos and Littlewoods. The goal was for Argos and Littlewoods to commonly

believe they would charge the RRPs. These communications also involved …nding values for

the RRPs that would be acceptable to both retailers.

Hasbro set the RRPs after separate discussions with Argos, Littlewoods, and

other retailers. ... Argos and Littlewoods then selected, independently from each

other, the Hasbro products they would include in their catalogues. Neil Wilson,

Hasbro’s account manager for Argos, describes how the pricing initiative then

worked in practice: "When I was given the products selected for the catalogue, I

established which were the common products carried by the majority of retailers

(not speci…cally Index) and asked Argos what its price intentions were in relation

to each of these products. I did not do this for products that were not common. I

informed Argos what the Hasbro RRPs for the common products were and asked

them whether any of our RRPs were a problem for them to match. ... Having

determined Argos’pricing intentions and passed these on to the other account

managers within Hasbro, I received information from those account managers

regarding the intentions of other retailers to go with RRPs. I then reverted to

Argos and said, without being speci…c, that it was my belief that the future retail

price of a product would or would not be at the RRP. I told Argos which products

this related to. I never mentioned the name of the retailer who was involved or

quanti…ed exactly the price that retailer would go out at. I simply said to Argos

that it was my belief from what retailers told us that this or that product would

or would not be at the RRP."

Neil Wilson of Hasbro was communicating with Argos and Ian Thomson of Hasbro was

communicating with Littlewoods. The two of them shared information with the common

objective of each retailer becoming con…dent that the other retailer would charge the RPPs

on the Hasbro products that they commonly carried.

[David Bottomley states:] "It is incorrect to suggest that Neil and Ian were

acting unilaterally in putting together this proposal: it was based on detailed

Id. at { 49.

Id. at { 53.

discussions and conversations that they had with Argos and Littlewoods about

pricing at RRPs. Each was aware that similar discussions were taking place with

the other and that a big e¤ort was being made to get all retailers to price at RRP."

Neil Wilson states: "Argos were fully aware that the pricing initiative involved

Hasbro talking to other retailers." ... Ian Thomson states: "There was no doubt

that Alan Burgess [Littlewoods’s buyer of boys’toys] knew that I was passing on

to the Argos account handler (Neil Wilson) the contents of our discussion and

that I would con…rm the Argos intentions back to him after Neil had concluded

his discussions with Argos."

The O¢ ce of Fair Trading found "no evidence that Argos and Littlewoods spoke directly"

and that "con…dential information was exchanged between them with Hasbro acting as the

…xer or middleman".

In spite of the initial concerns of Argos and Littlewoods that the other

would undercut the RRPs and the lack of direct communications between them, Hasbro felt

that they achieved the exchange of assurances required for the pricing initiative to work.

[David Bottomley, Neil Wilson and Ian Thomson made] it clear that there

was an informal agreement, understanding or tacit arrangement whereby Argos

and Littlewoods co-operated with Hasbro by indicating that they would or might

price the particular products in question at or near RRP on the understanding

that the other retailer would also do so, at the same time making it clear again

and again that if the other reneged, the former would immediately respond.

Bottomley ... states that "what existed between Hasbro and Argos and Hasbro

and Littlewoods was an understanding that, because of the obvious bene…t to

everyone in the industry, prices would be at or near RRP."

While the pricing initiative started on a restricted set of products, it was soon extended

to other categories because, as conveyed in a meeting at Hasbro, "it was crucial that we

maintained retail price stability as far as possible across our key brands so that the initia-

tives could succeed."

With the …rst pricing initiative having succeeded, there was cautious

optimism: "Littlewoods’reaction to Hasbro’s proposal was similar to Argo’s reaction: it was

positive, but also concerned about undercutting."

Argos’senior buyer Sue Porritt "felt it

was great that Hasbro could help maintain retail price stability, but said that Argos would

react if it was undercut in order to remain competitive."

Obtaining a state of common understanding among retailers to implement the proposed

collusive strategy was just the starting point to success. A retailer continued to be concerned

that its rival would undercut the RRPs, in spite of its announced intention to abide by them

(as conveyed by Hasbro). To ensure compliance as well as maintain retailers’con…dence in

the scheme, Hasbro monitored the retailers’prices. This issue arose at the very …rst Hasbro

meeting on the pricing initiative.

Id. at { 97.

Id. at { 103.

Id. at { 140.

Id. at { 65.

23 October 1998 meeting - Discussion took place about the listing and the

pricing initiatives (under which Hasbro would try to get retailers to list at RRPs).

Account managers were briefed to undertake audits of toy retailers and if they

found that prices were not at RRPs they were to have conversations with them

to try and persuade them to adopt RRPs.

The importance of monitoring was very much recognized by Hasbro.

Hasbro conducted its own monitoring to detect undercutting by retailers. Ian

Thomson states in his witness statement: "The emphasis on price monitoring

now was to ensure that our other customers would fall in line so that Argos

and Index would be con…dent that our plan was working throughout the UK.

This would reduce the risk of them going back to price cutting in the following

catalogues."

Given their concerns about having higher prices than rivals, Argos and Littlewoo ds were

also incentivized to monitor. However, rather than directly contact the deviator, they would

inform Hasbro.

Neil Wilson states: "Argos monitored other retailers’ prices. If they found

out that a retailer was not at the Hasbro RRP, they contacted me to …nd out

why there was a di¤erence. When Argos called me about the apparently lower

price of another retailer, they contacted me to see if Hasbro could do something

about it, i.e. get the other retailer to go back to RRP. The understanding was

that if Hasbro could give Argos an assurance that the other retailer would put

the price back up to the RRP, Argos would also remain at the RRP. If not, Argos

would have to make a decision about how it would price the product –usually

by matching the competitor’s price."

In response to receiving this information, Hasbro would seek to bring the recalcitrant retailer

into compliance.

Neil Wilson describes the process: "[O]nce I had spoken to Argos, I contacted

the account manager in Hasbro who dealt with the retailer in question. He or

she in turn called the buyer of the retailer who had the lower price. The account

manager sought to …nd out why the price was lower and to persuade the retailer

to go back to the RRP. Often the lower price turned out to be a temporary

promotion, for instance to clear out stock, or a simple mistake, as most retailers

were eager to charge RRPs. I then informed Argos whether we were able to do

anything and either provided the reassurance they sought or said that we could

do nothing. Argos knew that this was the process that was going on."

Id. at { 45.

Id. at { 85.

Id. at { 86.

Id. at { 90.

Hasbro’s coordinating and monitoring practices proved successful.

The Argos and Littlewoods Autumn/Winter 1999 catalogues were the …rst

catalogues for which the Hasbro account managers for Argos and Littlewoods

had applied the [pricing initiative]. When the catalogues were published in July

1999, it became clear to Hasbro that Argos and Littlewoods had priced nearly

all the Action Man products and core games at the levels they had indicated to

Hasbro, normally at Hasbro’s RRPs. This had been very di¤erent in the three

previous catalogues.

At a year-end Hasbro meeting in 1999, it was noted that the "retail pricing initiative has

worked."

And, as predicted by Hasbro, it was su¢ cient to get Argos and Littlewoods on

board in order for all major London toy retailers to comply. Argos was the price leader and,

with the exception of Littlewoods, the "rest of [the] retailers were price followers."

Other retailers would have been able to see easily from the catalogues that

RRPs were being followed. From the statements made by Hasbro employees,

it would seem that other retailers broadly followed Argos/Littlewoods pricing

practices and that as a result there was little deviation from Hasbro RRPs. Mike

McCulloch states: "As far as other retailers [are] concerned, [there was] no need

to communicate; they had bought into [the] initiative, and were happy to follow

Argos price lead."

In an email on May 18, 2000, Ian Thomson and Neil Wilson (who were the Hasbro

Business Account Managers for Littlewoods and Argos respectively) shared their pricing

initiative, along with a report of its success, more broadly within Hasbro:

Neil and I have spoken to our respective contacts at Argos and Index and

put together a proposal regarding the maintenance of certain retails within our

portfolio. This is a step in the right direction and it is fair to say that both

Accounts are keen to improve margins but at the same time are taking a cautious

approach in case either party reneges on a price agreement.

In response, they received an e¤usive email from Hasbro Sales Director Mike Brighty:

Ian . . . This is a great initiative that you and Neil have instigated!!!!!!!!!

However, a word to the wise, never ever put anything in writing, its highly illegal

and it could bite you right in the arse!!!! suggest you phone Lesley and tell her

to trash? Talk to Dave. Mike

Id. at { 57.

Id. at { 63.

Id. at { 55.

Id. at { 67.

Id. at { 73.

The O¢ ce of Fair Trading did "bite them in the arse" when they found all three …rms

to have infringed the Competition Act 1998.

Hasbro, Argos and Littlewoods have entered into an overall agreement and/or

concerted practice to …x the price of certain Hasbro toys and games. This over-

all agreement included two bilateral price-…xing agreements and/or concerted

practices which in themselves constitute infringements: one b etween Hasbro and

Argos and the other between Hasbro and Littlewoo ds. The agreements were en-

tered into in 1999 and ... came to an end no earlier than 15 May 2001 and no later

than 14 September 2001. The OFT takes the view that these agreements... had,

as their object and e¤ect, the prevention, restriction or distortion of competition

in the supply of certain Hasbro toys and games in the UK.

Penalties of £ 17.28 million and £ 5.37 million were levied on Argos and Littlewoods respec-

tively. The penalty for Hasbro was calculated to be £ 15.59 million, but which it avoided by

having received full leniency for cooperating with the OFT. Argos and Littlewoods appealed

the OFT’s decision to the UK Competition Appeal Tribunal (CAT) which then dismissed

the appeal. The CAT’s judgment was taken to the Court of Appeal by Argos and Little-

woods on the grounds that there was no communications between the two retailers and such

communications were necessary to conclude there was an agreement or concerted practice.

The Court of Appeal sided with the CAT:

Argos must have known or could have reasonably foreseen that its discussion

with Hasbro re‡ected Hasbro’s discussions with other retailers. In our view

such conduct was a “form of practical coordination”which knowingly substituted

practical cooperation for the risks of competition. In particular, those reciprocal

contacts reduced uncertainty on Argos’part as to what other retailers’pricing

intentions were, and reduced uncertainty on Hasbro’s part on what Argos’prices

would be. That, in turn, facilitated Hasbro’s conversations with other retailers

especially Littlewoods, with a view to ensuring that they to o priced at RRPs.

In other words, downstream …rms can have an exchange of assurances by communicating

through an upstream …rm.

3.4 Pharmaceutical Products

Parke, Davis & Company (hereafter, Parke Davis) manufactured and sold pharmaceutical

products to retailers and drug wholesalers (who would then sell them to retailers).

The

particular retail market of interest was the Washington, D.C. metropolitan area. It had been

Parke Davis’s p olicy to have a suggested minimum retail price (SMRP). In early 1956, some

Id. at p. 1.

See supra note 51, at { 703.

Ensuing facts are from United States v. Parke, Davis & Company, 164 F. Supp. 827 (D.D.C. 1958)

(hereafter referred to as Parke, Davis (1958)) and United States v. Parke, Davis & Co., 362 U.S. 29 (1960)

(hereafter referred to as Parke, Davis (1960)).

drug store chains in the DC area started pricing below the SMRP and one retailer - Dart

Drug Company - engaged in deep and well-advertised price cuts.

Frustrated with the retail pricing situation, the manager in charge of the DC area devised

a strategy after learning he could legally refuse to sell to retailers who priced below the SMRP.

He informed retailers of this policy in July 1956, and also told wholesalers that they would

be denied supply should they supply retailers who did not respect the SMRP. Nevertheless,

…ve retailers continued with their low prices. Parke Davis cut o¤ their supplies though some

retailers were still pricing below the SMRP and selling out of their inventories. Parke Davis

then modi…ed its policy in July for some retailers: It would only cut o¤ sales to the retailer

(and to the wholesaler who sold to it) if the retailer advertised a price below the SMRP.

Dart Drug initially respected the advertising policy, in response to which Parke resumed

supplies. However, Dart Drug along with other retailers were back to advertising prices

below the SMRP by September. At that point, Parke Davis stopped refusing to sell to them

because an investigation by the Antitrust Division of the U.S. Department of Justice had

been launched in response to a complaint from Dart Drug.

As thus far described, the actions of Parke Davis could well be nothing more then the

exercise of its legal right to refuse to supply wholesalers and retailers. Furthermore, the

communications it had with them were bilateral; either with a single wholesaler or a single

retailer. However, the DOJ argued there was a violation of Section 1 of the Sherman Act.

As summarized by the Supreme Court:

[T]he Government introduced evidence showing that appellee had (1) an-

nounced a policy of refusing to deal with retailers who failed to observe appellee’s

suggested minimum resale prices or who advertised discount prices on appellee’s

products, (2) discontinued direct sales to those retailers who failed to abide by

the announced policy, (3) induced wholesale distributors to stop selling appellee’s

products to the o¤ending retailers, (4) secured unanimous adherence by inform-

ing a number of the retailers that, if each of them would adhere to the announced

policy, one of their principal comp etitors would also do so, and (5) permitted the

retailers to resume purchasing its products after they had indicated willingness

to observe the policy.

It is item (4) that shifted Parke Davis’actions into unlawful territory.

The credibility of Parke Davis’s threat to cut o¤ a retailer’s supply for having priced

below the SMRP relied on enough retailers complying. If many were charging below the

SMRP, implementation of the threat would mean shutting o¤ the supplies of many retailers,

and that would signi…cantly harm Parke Davis’market sales. However, if a retailer expected

other retailers to be pricing at the SMRP then, should it price below the SMRP, it became far

more believable that Parke Davis would go through with the threat and refuse to supply it.

To achieve those expectations, Parke Davis not only communicated its plan with wholesalers

and retailers but also that it was communicating with other wholesalers and retailers.

In order to insure that retailers who did not comply would be cut o¤ from

sources of supply, representatives of Parke Davis visited the wholesalers and told

Parke, Davis (1960), syllabus.

them, in e¤ect, that not only would Parke Davis refuse to sell to wholesalers

who did not adhere to the policy announced in its catalogue, but also that it

would refuse to sell to wholesalers who sold Parke Davis products to retailers

who did not observe the suggested minimum retail prices. Each wholesaler was

interviewed individually, but each was informed that his competitors were also

being apprised of this. The wholesalers, without exception, indicated a willing-

ness to go along. Representatives called contemporaneously upon the retailers

involved, individually, and told each that, if he did not observe the suggested

minimum retail prices, Parke Davis would refuse to deal with him, and that,

furthermore, he would be unable to purchase any Parke Davis products from the

wholesalers. Each of the retailers was also told that his competitors were b eing

similarly informed.

Of particular concern to retailers was that Dart Drug comply, for it had been pricing

well below the SMRP. That it advertised its prices would have heightened this concern

among other retailers, for a well-advertised discount to the SMRP price could attract many

customers from retailers who were charging the SMRP.

When interviewed, the president of Dart Drug Company indicated that he

might be willing to stop advertising, although continuing to sell at discount prices,

if shipments to him were resumed. Each of the other retailers was then told

individually by Parke Davis representatives that Dart was ready to discontinue

advertising. Each thereupon said that, if Dart stopped advertising, he would

also. On August 28, Parke Davis reported this reaction to Dart. Thereafter,

all of the retailers discontinued advertising of Parke Davis vitamins at less than

suggested minimum retail prices.

The District Court dismissed the DOJ’s complaint on the grounds that

[t]here is no evidence that ... Parke, Davis ever conferred or discussed its sales

policies with more than one wholesaler or more than one retailer at a time, nor

that it made the enforcement of its policies as to any one wholesaler or retailer

dependent upon the action of any other wholesaler or retailer.

It is true that Parke Davis’s communications were always one-on-one with a retailer or

wholesaler. It never convened a meeting of retailers and wholesalers at which it encouraged

collective adoption of prices that respected the SMRP or an agreement not to advertise

prices below the SMRP. However, coordinated adoption of a plan can be achieved through a

hub-and-spoke arrangement whereby the hub (Parke Davis) communicates with each of the

spokes (retailers, like Dart Drug, and wholesalers) and conveys a plan to price at the SMRP

and that the hub is meeting with other spokes to achieve a joint adoption of this plan. That

is how the Supreme Court viewed the matter when it reversed the District Court’s judgment.

It accepted the DOJ’s argument that

Parke, Davis (1960), 362 U.S. 33.

Parke, Davis (1960), 362 U.S. 35.

Parke, Davis (1958), at 55.

what Parke Davis did here by entwining the wholesalers and retailers in a

program to promote general compliance with its price maintenance policy went

beyond mere customer selection, and created combinations or conspiracies to

enforce resale price maintenance in violation of §§ 1 and 3 of the Sherman Act.

Furthermore, the Supreme Court noted that Parke Davis acted in this manner because: "It

was only by actively bringing about substantial unanimity among the competitors was Parke

Davis able to gain adherence to its policy."

3.5 Replica Football Kits

The product market comprises authentic reproductions of shirts, shorts, and socks for Eng-

lish football clubs. The supply of these replica football kits required licensing rights from

the football club (in this case, Celtic, Chelsea, Manchester United, and Nottingham Forest),

production by a manufacturer (Umbro), and marketing and distribution by retailers (All-

sports, Blacks, JJB, JD, and Sports Soccer).

As a matter of standard practice, Umbro

sought to control both the distribution and retail pricing of the replica football kits. MU

(which was the parent company of Manchester United) had concerns about maintaining the

brand and did not want its kits in supermarkets where it might be sold as a "loss leader."

Re‡ecting MU’s preferences, Umbro refused to supply the major supermarkets. Umbro also

set a recommended retail price (RRP).

Starting in 1998, Sports Soccer became more aggressive in its pricing in order to establish

itself as a major sports retailer. Its business model was to price below Umbro’s RRP and

sell large volumes.

Mr. Att…eld [Sports Soccer’s account manager] states: "Retailers have always

complained about Sports Soccer. The complaints, however, became particular[ly]

intense from around 1999, when Sports Soccer started to develop as a credible and

important competitor to established major retailers, such as JJB. On a number

of occasions, JJB made threats to cancel orders because of Sports Soccer’s pricing

practices."

In mid-1999, JJB announced that it was planning to sell adult replica shirts at or below

£ 40.00, which was about £ 3 less than Umbros’RRP. At outlets located near those of Sports

Soccer, JJB was o¤ering a 20 percent discount on all products including replica kits. As

JJB was Umbro’s largest customer, other retailers felt a need to respond with lower retail

prices and, consequently, they put pressure on Umbro to lower the wholesale price in order

for them to maintain their pro…t margins.

The state of the retail market led JJB to lobby Umbro to quell Sports Soccer’s aggressive

pricing.

Parke, Davis (1960), 362 U.S. 37, 38.

Parke, Davis (1960), 362 U.S. 46.

All facts are from CA/98/06/2003 Price-…xing of Replica Football Kit, 1 August 2003 [2004] CAT 24.

Id. at { 157.

[T]he starting point is that JJB had been badgering Umbro for some time

to do something about the fact that Sports Soccer was already selling England

replica kit at a discount, and that a crucial selling period was approaching during

which it would be particularly important for JJB that it should not have to face

or engage in a price war.

This "crucial selling period" was associated with the Euro 2000 which would substantively

increase demand for kits. There was potentially large pro…t to be earned for retailers like

JJB if margins could b e raised.

In response to retail prices below RRPs, Umbro lowered its RRPs and, starting in April

2000, sought to have all retailers’price at RRPs.

Umbro made it clear that it wished retailers to adhere to its RRPs or "high

street" prices for the resale of its Replica Kit and that discounting of its key

products was detrimental to its brand. For example, JD and First Sport have

con…rmed that Umbro pursued policies designed to persuade retailers to adhere

to its RRPs or "high street" prices and that their own respective policies of

generally pricing at Umbro’s RRPs were well known to Umbro.

Umbro conducted bilateral communications with each retailer. The message was not just

that it should price at or above RRPs, but that other retailers were planning to do so. Such

communications could deliver the assurances needed for a retailer to price at RRPs without

the risk of another retailer undercutting it.

Mr Ashley of Sports Soccer has said that during May and/or June 2000 Mr

Ronnie of Umbro contacted Mr Ashley, Mr Hughes of Allsports, Mr Knight of

Blacks, Mr Sharpe of JJB and possibly Mr Makin of JD. The contact was by

telephone to seek agreement that each retailer would price the England home

Replica Shirt for the duration of England’s participation in Euro 2000 at £ 39.99.

Sports Soccer had agreed to this and had understood from Umbro that Allsports,

Blacks, JJB and possibly JD had made similar agreements with Umbro, (Umbro

states that Mr Ashley’s agreement was conditional upon this).

Representative of these communications and the understanding surrounding them, con-

sider the following triadic relationship between JJB, Umbro, and Sports Soccer.

Mr. Ronnie [of Umbro] spoke to Mr Sharpe [of JJB] about JJB’s pricing

intentions, in resp onse to which Mr. Sharpe told him that JJB would sell at

High Street prices unless others discounted. ... [I]t must have been apparent to

Mr. Sharpe, even if there was no express reference to Sports Soccer, that Mr.

Ronnie wanted the information in the context of taking steps to guard against

discounting. Mr Sharpe was among those who had spoken to Umbro sta¤ about

the Sports Soccer discounting policy. It would be extraordinary to suppose that,

See supra note 51, at { 92.

Id. at { 120.

Id. at { 166.

when Mr. Ronnie asked him about JJB’s own attitude to pricing England replica

kit during Euro 2000, it did not occur to Mr. Sharpe that this was connected

with the question of discounting by Sports Soccer and attempts to prevent it

occurring, which Mr. Sharpe and others at JJB had b een asking for over some

time past.

With JJB’s pricing intention, Umbro then went to sp eak with Sports Soccer.

The second stage in the process was Mr. Ronnie’s call to Mr. Ashley [of

Sports Soccer], to persuade him that he should raise Sports Soccer’s prices for

the England shirt. It is evident that Mr. Ashley was very reluctant to do so.

It required pressure from Umbro by way of veiled or not so veiled threats as

regards supplies. It also required an assurance that other retailers would not be

discounting the shirts. Mr. Ronnie gave him that assurance. He did not mention

any retailer by name, but because JJB was so dominant in the …eld, any assurance

as to retailers must have been taken, and intended to be taken, as including JJB.

Mr. Ashley agreed to raise Sports Soccer’s prices on that basis, conditionally

on the others also raising or maintaining their prices to or at the same level. It

seems to us that, in turn, he must have recognised that others concerned would

be told of his agreement. He knew, from what Mr. Ronnie told him about other

retailers, that Umbro had b een in touch with the other retailers about their

pricing intentions, and that these had been passed on to him. He must have

realised that what he told Mr. Ronnie about Sports Soccer’s intentions would,

correspondingly, be passed back to the others, including, necessarily, JJB.

Finally, Umbro returns to speaking with JJB in order to assure them about Sp orts Soccer’s

pricing.

Then at the third stage, Mr. Ronnie telephoned Mr. Sharpe again and told

him that Sports Soccer had agreed to raise their prices and to sell at High Street

prices. He did so in order to make it known to JJB that Umbro had, as asked,

"done something" about Sports Soccer’s discounting, by securing an agreement

that it would come to an end as regards this product. He also needed to make

sure that JJB knew of this because of Mr Sharpe having mentioned that JJB

might discount if others did, so that JJB should be aware that, at any rate if

Sports Soccer kept to their agreement, JJB would not need to discount.

However, this was not a pure hub-and-spoke cartel, however, for there were also direct

communications between retailers.

On 8 June 2000 at about 13.00 hours, Mr Hughes of Allsp orts, Mr Whelan

and Mr Sharpe of JJB and Mr Ashley of Sp orts Soccer met in Mr Hughes’house.

... [T]he purpose of the meeting was to discuss "the state of the market for

See supra note 51, at { 94.

Id. at { 97.

Id. at { 98.

replica kit including the crippling price war between" Allsports, JJB and Sports

Soccer. Allsports denies that any agreement was reached at the meeting and

states that the JJB representatives merely restated JJB’s public pricing policy

on Replica Kit and that Sports Soccer refused to give an indication as to its

future pricing policy. However, Sports Soccer has stated that, at this meeting, it

agreed with Allsports and JJB that it would price the MU home adult Replica

Shirts at £ 39.99 at launch and for an unspeci…ed period thereafter.

That Umbro promoted direct communication among the spokes suggests that it did not

believe hub-spoke bilateral communications would prove su¢ cient to deliver the mutual

understanding among retailers that they would price at Umbro’s RRPs.

A letter dated 13 July 2000 from Mr Prothero of Umbro to Mr Richards of

MU says: "As you know Umbro have worked very hard in agreeing a consensus

to the price of the new Manchester United jersey. At one stage we even managed

to get Messrs Hughes [of Allsports], Ashley [of Sp orts Soccer] and Whelan [of

JJB] in the same room to agree this issue."

Mr Ashley of Sports Soccer has also con…rmed that Umbro had requested it

to attend a meeting with Allsports and JJB to discuss retail pricing on the MU

home Replica Kit as Sports Soccer’s assurances to Umbro regarding its pricing

intentions were not su¢ cient for the other retailers.

Sports Soccer con…rmed that "at meetings of this sort it requested and received assurances

over the pricing intentions of other retailers,"

and the OFT opined that "Sports Soccer

would require such information in order to ensure that agreements reached with Umbro on

retail pricing did not put it at a disadvantage."

According to Umbro’s internal documents, these e¤orts were successful. From its May

2000 monthly management report:

There has been a major step forward in the retail price of England [and] the

launch of Manchester United. JJB, Sports Soccer, First Sports, JD Sports and

all:sports have all agreed to retail their adults shirts at £ 39.99. This is following

England being sold at various retail prices through April and May ranging from

£ 24.99 to £ 29.99, £ 32.99 or £ 32.99 with a free £ 9.99 cap at JD Sports.

Crucially, Sports Soccer, which had been one of the most aggressive retailers, was in com-

pliance:

Sports Soccer complied with the agreement in relation to the launches of the

Chelsea and Celtic launches in May 2000, when all the relevant retailers sold the

Replica Shirts at High Street Prices.

Id. at { 187.

Id. at { 197.

Id. at { 187.

Id. at { 165.

Id. at { 358.

Id. at { 346.

The OFT found the parties had engaged in unlawful agreements from March 2000 to

November 2001, and concluded "that at least Allsp orts, Blacks, and JJB (together with

Umbro and Sports Soccer) all agreed, in or around late May to early June 2000, to co-

ordinate their pricing of the new MU home Replica Shirt that was launched on 1 August

2000."

100

The OFT recognized that

RRPs are not unlawful when they simply operate as recommended prices. In

this case, however, the OFT is satis…ed that RRPs and "high street" prices during

the period of the infringement operated as focal points for concerted behaviour.

Umbro applied pressure to certain retailers for them to adhere to RRPs or "high

street" prices.

101

The UK Competition Appeal Tribunal (CAT) supported the OFT’s decision, which was

then a¢ rmed by the Court of Appeal. In its decision, the Court of Appeal succinctly

expressed the hub-and-spoke communications that were the basis for JJB and Sports Soccer

having a "meeting of minds" that they would both raise their prices:

[T]he Tribunal was entitled to …nd that (1) JJB provided con…dential price

information to Umbro in circumstances in which it was obvious that it would or

might be passed on to Sports Soccer in support of Umbro’s attempt to persuade

Sports Soccer to raise its prices (thereby adopting the pricing policy which JJB

explicitly wanted adopted by all signi…cant retailers), (2) Umbro did use the

information in relation to Sports Soccer in that way, (3) Sports Soccer did agree

to raise its prices in reliance on this information, and foreseeing that others

including JJB would be told of its agreement, and later did raise its prices as it

had agreed to do, and (4) Umbro did tell JJB of this, thereby making it clear

to JJB that it would be able to maintain its prices at their current level, as it

did.

102

The replica football kits case has a number of similar features to the UK toys case. In

response to intense retail price competition and complaints from at least one retailer, an

upstream manufacturer sought to raise retail prices by engaging in bilateral communications

with retailers in order to achieve mutual understanding among them to set prices at the

manufacturer recommended level. The replica football kits case di¤ers in that the hub

also orchestrated a meeting of the spokes for them to engage in direct communication.

Apparently, the hub was not su¢ ciently convinced that indirect communication would deliver

an exchange of assurances among the spokes.

3.6 Cheese

In response to low raw milk prices in the United Kingdom,a group representing farmers

- Farmers for Action (FFA) - put pressure on processors - who purchase the raw milk to

produce milk and cheese - to support a higher farmgate price.

103

In July 2002, the FFA

100

Id. at { 451.

101

Id. at { 125.

102

See supra note 51, at { 102.

103

Facts are from Decision of the O¢ ce of Fair Trading, CA98/03/2011, Dairy retail price initiatives, 26

July 2011, (Case CE/3094-03).

organized farmers to blockade dairy processing facilities. The FFA also lobbied supermarket

chains to raise the retail prices for milk and cheese, on the grounds that it could support a

higher wholesale price from the processors which, they hoped, would lead to higher prices

for raw milk.

While the initial impetus for raising retail prices came from the farmers, the case is about

collusion between the processors and supermarkets. The processors - Dairy Crest, Glanbia,

and McLelland - manufactured cheese and other dairy products which they sold to retailers.

The involved supermarket chains were Arla Foods, Asda Stores, Sainsbury’s Supermarkets,

Safeway Stores, and Tesco Stores. The vertical relations were that Dairy Crest supplied

Asda, Safeway, Sainsbury’s, and Tesco; Glanbia supplied Asda, Safeway, and Sainsbury’s;

and McLelland supplied Asda, Safeway, Sainsbury’s, and Tesco.

The case involved several episodes but our attention will generally be limited to a collusive

plan to raise cheese prices. The aim of the 2002 Cheese Initiative "was to …nancially assist

UK dairy farmers by subsidising an increase in the farmgate price through co-ordinated

wholesale and retail price increases."

104

This was to be achieved by raising the retail price

of cheese by £ 200 per metric ton "through staggered increases over three weeks starting on

20 October 2002".

105

As is clear from the documentary evidence provided below, critical

to its implementation was that retailers jointly consummate this price increase. As both

the upstream processors and downstream supermarkets had been approached by the FFA to

increase wholesale and retail prices, the thought of raising prices was on their minds even

prior to the communications between processors and supermarkets.

Given the multiple channels at work, it is useful to summarize the various communi-

cations that were intended to raise cheese prices by the supermarkets. First, there were

bilateral communications between a processor and a supermarket, with a processor trying to

convince a supermarket that other supermarkets were going to increase prices. Second, there

were meetings between a supermarket and several of the processors supplying it. In those

meetings, a supermarket expressed a desire to raise its cheese prices but that it was critical

that other retailers do so as well. Third, there were bilateral communications between the

FFA and a supermarket, whereby the latter expressed the need for a collective price increase

and the former shared information about other retailers’price intentions.

Consistent with a standard hub-and-spoke cartel, a processor (hub) was disseminating

pricing intentions of supermarkets (spokes) in order to provide assurances to a supermarket

that it could raise price knowing that its competitors would do likewise. On September 17,

2002, Dairy Crest’s Senior Account Manager for supermarket Asda met with Asda’s Category

Manager for Dairy. At that meeting, Dairy Crest’s manager provided a slide presentation

titled "Asda Brie…ng Document Raw Milk Pricing" which demonstrated "that Dairy Crest

prop osed to Asda that it should participate in an initiative to subsidise a farmgate price

increase through retail price increases on dairy products in addition to milk."

106

One slide

stated: "Move Cheese Prices at retail up by £ 200 per tonne.", and another stated: "Action

- Move the whole market forward."

107

which the OFT appropriately interpreted as Dairy

Crest informing Asda "that it was not being expected to increase its cheese retail prices

104

Id. at { 5.14.

105

Id. at { 5.158.

106

Id. at { 5.122.

107

Id. at { 5.122.

unilaterally, but as part of a wider market move which also involved the implementation of

retail price increases by its competitors."

108

On the same day, Dairy Crest made a "virtually

identical presentation" to another retailer.

109

These communications are consistent with an

upstream processor promoting an increase in retail cheese prices with each of its customers,

while conveying to those customers that it was part of a plan for all supermarkets to raise

prices. Such was clearly expressed in a note by Dairy Crest’s Commercial Director for its

Tesco account which stated that Dairy Crest "[p]roposed to Asda that by early November

all accounts would have followed the market moves... Asda - Tesco - Sainsbury’s - Safeway

- [a retailer] - [a retailer]."

110

If Asda were to raise its retail prices, it would not be alone.

There were also documented communications in which a processor conveyed the explicit

pricing intentions of one supermarket to another supermarket, which would have led to

mutual understanding among the supermarkets regarding price increases.

On 4 November 2002, [Dairy Crest’s Category Manager (Cheese and Spreads)

for its Tesco account] sent an e-mail and attached spreadsheet entitled "Suggested

RSP" to [Tesco’s Senior Cheese Buyer]. The e-mail reads as follows: "I have

attached a spreadsheet which shows the suggested rsp’s [retail selling prices] of

cheese lines that we supply Asda following the price increase. My understanding

is that Asda will be applying £ 200 per tonne."

111

In an interview with the OFT, that Dairy Crest employee explained:

By sending this email, I was showing [Tesco’s Senior Cheese Buyer] that the

suggested price of cheese was the same across the board and not speci…c to Tesco.

The prices mentioned were only suggested prices and not guaranteed prices. In

my view it was not a de…nitive statement that Asda’s prices would go up by the

amount listed; I stated that it was "My understanding" that they would do so

rather than stating that they de…nitely would.

112

Thus, Dairy Crest was not sharing the actual prices of Asda but rather its intended prices,

and presumably doing so with the goal of a¤ecting Tesco’s prices. Similarly, processor

McLelland sent an email on October 21, 2002 to Tesco which conveyed the pricing intentions

of rival retailer Sainsbury: "I have provided the recommended Retail going forward plus the

position to protect your own margin. ... Sainsbury are con…rming that the new retails on

Branded pre-pack will be in place Tuesday this week."

113

In addition to the bilateral communications that McLelland conducted with its customers,

it also utilized a public forum to promote mutual understanding among retailers. An article

in the Dairy Industry Newsletter stated:

108

Id. at { 5.171.

109

Id. at { 5.127.

110

Id. at { 5.171.

111

Id. at { 5.332.

112

Id. at { 5.335.

113

Id. at { 5.251.

SUPERMARKETS AGREE £ 200/tonne PRICE INCREASE Apparently, as

of this week, all the major UK supermarkets are to implement an across-the-

board £ 200/tonne increase in cheese prices as from next month. Asda were the

last of the major chains to ‘come on board’. Sainsbury led the initiative, followed

by [a retailer], [a retailer] and [a retailer]. Tesco, initially reluctant, have agreed

to a price increase. It is understood that Sainsbury are putting retail prices up

this week, the others will follow.

114

McLelland shared this article with Sainsbury, which would have helped shore up Sainsbury’s

belief that other supermarkets would be raising prices. In addition, McLelland may have

provided the information for the article.

On 24 September 2003, [McLelland’s National Account Manager for its Sains-

bury’s account] e-mailed a copy of an article taken from the Dairy Industry

Newsletter website of the same date to [Sainsbury’s Senior Cheese Buyer]. ...[The]

covering e-mail suggests that McLelland may have played a role in providing in-

formation for the article: "Please read the below feature on the retail move. As

you can see, we have positioned you favourably in terms of moving the price

forward, and as the Dairy Industry News is the ‘Bible’ of the dairy farming

community, this has been viewed as a very favourable move by Sainsbury’s."

115

A public announcement of a market-wide increase in retail prices could have gone a long way

to providing the assurances that the supermarkets needed.

Though the FFA was not prosecuted by the OFT, it also engaged in communications to

promote common expectations among retailers about higher cheese prices. When the FFA

was lobbying supermarkets for a price hike, some supermarkets clearly expressed that there

needed to be a market-wide movement in prices:

On 4 October 2002, [Sainsbury’s General Manager for Dairy and Cheese]

wrote to [an o¢ cial at FFA] and informed him of the status of discussions between

Sainsbury’s and its processors regarding a cheese price increase. The letter reads

as follows: "With regards to cheese we are still discussing the implementation of

cost price increases with all our processors. It is intended that we will pass on

an increase in our buying prices by £ 200/tonne in approx 3 weeks, for all our

standard cheese range, provided other retailers also accept this. I must stress

that if others do not generally support this initiative, I will have to withdraw my

support for cheese, if I …nd I am uncompetitive in the wider market place."

116

Though not associated with the 2002 Cheese Initiative, the FFA had shared pricing intentions

between supermarkets for fresh liquid milk (FLM), around the same time. Tesco had received

letters from the FFA which had been sent

114

Id. at { 5.510.

115

Id. at { 5.510.

116

Id. at { 5.231.

to [an FFA o¢ cial] by Safeway and [a retailer’s] representatives, following an

approach by FFA. In the letters, these retailers expressed support for an increase

in retail prices for FLM in order to pass back monies up the supply chain to

farmers (to achieve an increase in the farmgate price for raw milk), provided that

competing retailers would also increase their retail prices. FFA was circulating

these letters between retailers.

117

It is quite possible that the FFA also engaged in such practices for cheese.

The …nal documented channel was between a supermarket and several of its suppliers.

A meeting on September 13, 2002 involved Tesco and its two suppliers, Dairy Crest and

McLelland. Based on a processor’s employee’s notes from the meeting,

[i]t is also evident that [Tesco’s Category Manager for Dairy] gave a clear

and direct indication that Tesco was contemplating increasing its cheese retail

prices as part of any price initiative with the note recording him as stating that:

"Cheese and spread values have crashed over the last 3 years and Tesco has been

selling them at a loss. Cautiously optimistic that Tesco can now start to move

retail prices forward in this area but ... very di¢ cult to move out of line with

other competitors. ... [Tesco’s manager] senses there is a mood to move some of

these prices forward."

118

The OFT commented in its decision that

[i]n disclosing that it was prepared to increase its cheese retail prices ... to its

supplying processors in the context of discussions which had as their purpose the

co-ordination of a market-wide cheese retail price increase, Tesco may be taken

to have intended and did, in fact, foresee that its supplying processors would

make use of that information to in‡uence conditions on the cheese retail market

by passing it to other retailers in order to facilitate further and wider retail price

increases.

119

The message from these communications are clear: A supermarket is willing to raise

its cheese prices as long as competing supermarkets are planning to do so. In order to

achieve that common course of actions, there were various forms of communications utilized

to deliver the required mutual understanding among supermarkets. By conveying its pricing

intentions to a processor like Dairy Crest and to the FFA, a processor and FFA could share

those pricing intentions with other supermarkets. If a supermarket anticipated such sharing,

which is reasonable given that other supermarkets’ pricing intentions were being shared

with them, then these retailers could have achieved the exchange of assurances necessary to

consummate a rise in retail cheese prices. As summarized by the OFT:

The type of information disclosed by retailers chie‡y consisted of expressions

of willingness to increase retail prices on condition that their competitors either

117

Id. at { 2.70.

118

Id. at { 5.80.

119

Id. at { 5.119.

led or followed the retail price increase and information concerning future retail

price increases (such as the levels of the retail price increases and the dates on

which those increases would be implemented).

120

With regards to the legal case, the OFT found Asda, Dairy Crest, Glanbia, McLelland,

Safeway, Sainsbury’s, and Tesco to have participated "in a single overall concerted practice

which had as its object the prevention, restriction or distortion of competition in respect of

cheddar and British territorial cheese retail prices in 2002."

121

All but Tesco admitted their

guilt. As a result of these practices, the OFT concluded that retail prices rose for cheddar and

British territorial cheeses. Tesco appealed the OFT’s decision to the Competition Appeal

Tribunal which overturned the judgment regarding conduct in 2002 but a¢ rmed Tesco’s

guilt for behavior in 2003.

3.7 Paints and Varnishes

In the early 2000s, the Do-It-Yourself (DIY) market grew in Poland and with it came in-

creased competition among DIY retail chains.

122

Of particular note was a decline in the retail

prices of paint and varnishes. The reduced retail price margins were not just a concern for

the retail chains but also the upstream manufacturers. With rising prices for raw materials,

the upstream manufacturers wanted to raise the wholesale prices of paints and varnishes but

were …nding resistance from the downstream retailers, who were concerned that it might

mean yet smaller margins.

In response to this situation, paint and varnish manufacturer Polifarb Cieszyn Wroc÷aw

(PCW) created a program to induce retailers to set higher retail prices. Referred to as a

"price stabilizing system," PCW put forth recommended retail prices for its ten best-selling

products and encouraged retailers to price no lower than the recommended level. To induce

compliance, PCW created a rebate system:

Stores that maintain PCW suggested prices as their retail prices will be re-

ceiving an additional stabilizing rebate in their invoice as a bonus for price com-

pliance. If a given store does not comply with the prices suggested by PCW, all

supplies of all products will be stopped and the stabilizing rebate will be put on

hold until prices are brought back to the suggested level.

123

A chain would be considered out of compliance if just one store priced below the recom-

mended price.

If PCW had independently negotiated such a policy with each chain, there would not

have been collusion. However, as with Hasbro, the retail chains were concerned that other

chains may not comply by pricing below PCW’s recommended prices. If that were to occur,

a retailer in compliance would lose market share. In light of the recent history of aggressive

pricing, it was natural for the retailers to be skeptical about compliance.

120

Id. at { 5.45.

121

Id. at { 7.2.

122

The ensuing facts are from Bolecki (2011).

123

Id. at 32.

The compliance by all major DIY chains ... with the prices set by the suppliers

was the key to the success of the price stabilization program. A deviation by one

of the retailers generally caused an immediate reaction from other DIY store

chains –the latter would either lower or threaten to lower their retail price so as

to remain competitive with each other.

124

The Polish competition authority - O¢ ce of Competition and Consumer Protection (or

Urz ¾ad Ochrony Konkurencji i Konsumentów or UOKIK) - found no evidence of any com-

munications between the downstream retail chains. What they did …nd was a very engaged

upstream manufacturer intent on assuring that all retailers adopt the policy. The UOKIK

determined that

PCW had played the role of a "mediator" who would "appease disputes",

explain price di¤erences, and inform retailers about how quickly would the prices

of their competitors return to the agreed level. The manufacturer would notify

retailers of all price changes of their competitors (most often increases), no matter

how small, so as to prevent price changes by other trading partners.

125

It was PCW’s e¤orts in support of a coordinated adoption of its price-stabilizing program

among retailers that turned a resale price maintenance program into hub-and-spoke collusion.

In addition to monitoring current prices for compliance, there were at least two other

avenues through which a upstream manufacturer could learn about a retail chain’s future

prices:

1) Price information associated with ... joint marketing campaigns organised

by the supplier and its retailers. ... The campaign would be commissioned by the

given DIY chain but the manufacturer would usually participate in its costs. ...

[W]hen sending the draft for a supplier’s approval, retailers would often disclose

at the same time that product’s intended retail price. If the price shown on the

draft was lower than the price recommended by the supplier, the latter would try

to persuade the DIY chain to raise it, for example by threatening to withdraw

the co-…nancing of the campaign. 2) Suppliers would obtain information from

retailers on their intended retail prices in the course of standard conversations or

e-mail contacts.

126

PCW’s success was not lost to other paint and varnish manufacturers, who were also

…nding resistance to wholesale price increases due to shrinking retail margins. Soon manu-

facturers Akzo Nobel and Tikkurila formed their own hub-and-spoke cartels in order to raise

the retail prices for their paints and varnishes. As the UOKIK noted:

Akzo Nobel was transmitting to its retailers information on price changes in-

tended by their competitors. As shown by the collected evidence, the company

was not doing that at the request of the chains but rather to convince them that

124

Id. at 33.

125

Id. at 32.

126

Id. at 36.

the given participant in the agreement would indeed comply with the arrange-

ments. Such assurances would give the supplier more certainty as far as the

compliance with the arrangements by those participating in the agreement.

127

The UOKIK found no evidence of any communication among the three upstream manu-

facturers. Though these were then three separate hub-and-spoke conspiracies, they were not

independent occurrences for Akzo Nobel and Tikkurila were imitating the success of PCW’s

hub-and-spoke cartel.

3.8 Drugstore, Perfumery and Hygiene Products

From 2002 to 2007, a hub-and-spoke cartel raised prices for drugstore, perfumery, and hy-

giene (DPH) in supermarkets throughout Belgium. Seven retailers were involved - Carrefour,

Colruyt, Cora, Delhaize, Intermarché, Makro, and Mestdagh - and 11 suppliers - Beiersdorf,

Belgium Retail Trading, Bolton, Colgate, DE HBC, GSK, Henkel, L’Oreal, Procter & Gam-

ble, Reckitt, and Unilever. The cartel was discovered due to Colgate applying for leniency.

All 18 companies settled and the Belgian Comp etition Authority imposed an aggregate …ne

of 174 million euros.

128

The objective of the cartel was "to increase and stabilize the selling prices of domesti-

cally produced DPH products to consumers at similar or near similar levels at the [retailer]

level."

129

Each upstream supplier, such as Unilever, ran its own hub-and-spoke cartel with its

downstream suppliers. While there was no evidence of any communication among suppliers,

the simultaneity of these hub-and-spoke cartels and the similarity of their practices suggest

otherwise.

130

The documented communications between a supplier’s account manager and a

retailer’s purchasing agent or department manager took the form of emails, telephone calls,

and conversations in stores. There was no evidence of communications between retailers.

Presumably with the cooperation of its leniency awardee, the Belgian Competition Au-

thority provided a detailed description of how a round of collusive price setting was con-

ducted.

131

It had four stages that are labelled initial, negotiation and consultation, imple-

mentation, and control. The initial phase had either the supplier (hub) or a retailer (spoke)

127

Id. at 39.

128

Facts are based on a Google translation of Autorite Belge de la Concurrence, Décision n



ABC-2015-

I/O-19-AUD du 22 juin 2015, A¤aire CONC-I/O-06/0038 – Hausses coordonnées des prix de vente de

produits de parfumerie, d’hygiène et de droguerie (which translates as: Belgian Authority of Competition,

Decision No. ABC-2015-I / O-19-AUD of June 22, 2015, Case CONC-I / O-06/0038 - Coordinated in-

creases in the selling prices of perfumery, hygiene and drugstore products). Other references are "Belgian

Competition Authority settles its …rst cartel case," www.elexica.com/en/legal-topics/antitrust-and-merger-

control/29-belgian-competition-authority-settles-its-…rst-cartel-case (downloaded 05/22/2018); and Mattioli

(2016).

129

Google translation of Belgian Authority of Competition, Decision No. ABC-2015-I / O-19-AUD of June

22, 2015, Case CONC-I / O-06/0038 - Coordinated increases in the selling prices of perfumery, hygiene and

drugstore products, para. 20.

130

Like with the Paints and Varnishes case in Poland, it is p os sible that one of the suppliers implemented a

hub-and-spoke cartel and, upon learning about its success, other suppliers imitated it by creating their own.

131

The description in Mattioli (2016, p. 262) con…rms the translation of Autorite Belge de la Concurrence,

Décision n



ABC-2015-I/O-19-AUD du 22 juin 2015, A¤aire CONC-I/O-06/0038 –Hausses coordonnées des

prix de vente de produits de parfumerie, d’hygiène et de droguerie.

contact the other to propose a price increase for certain products. The hub then informed

all of its spokes that a coordinated price increase was in the works. During the negotiation

and consultation phase, the supplier would discuss with its retailers what should be the

magnitude of the price increase, the date at which it should be implemented, and which

retailers were to participate. The information that a supplier received from a retailer was

then conveyed to the other retailers. Once an agreement had been reached, the supplier

informed each retailer of the products, the new retail prices, the implementation dates, and

the participating retailers.

The communications conducted by a supplier during the negotiation and consultation

phase served to achieve mutual understanding among the retailers regarding the price in-

creases for certain products. The implementation phase was also crucial for dispelling re-

tailers’concerns that other retailers would not comply. It was during the implementation

phase that retailers were to adopt the new prices, which might take a week for all retailers

to do. Some retailers would not raise their prices until the supplier showed them transaction

receipts from rival retailers documenting that they were indeed charging the new prices.

Also critical to ensuring continued compliance was the control phase which had the supplier

monitoring the stores to ensure that the new prices were in place. The information acquired

through store inspection would be shared by the supplier with the participating stores. Re-

tailers would also monitor, and then share with the supplier any evidence of noncompliance,

and ask them to intervene with the deviant retailer.

The Belgian Competition Authority concluded that the hub-and-spoke cartel had mixed

success in that some agreed-upon price increases were implemented, while others were not.

3.9 e-books

In the book publishing market, the traditional pricing strategy was to initially release a b ook

in hardcover at a high price and then, at a later time, release a paperback version at a lower

price. Referred to as "windowing," it allowed a publisher to generate higher pro…ts through

intertemporal price discrimination. Compared to charging a constant price over time, higher

margins were earned on those consumers who most valued the product - and thus bought it

early at the high price - while still selling to the broader market (who valued it less) through

its lower-priced paperback edition.

e-books (also known as digital or electronic books) presented a new format for publishers.

Introduced by Amazon in 2007 with the Kindle, Amazon chose a pricing strategy that was a

disruptive departure from the windowing strategy. For New Releases and New York Times

bestsellers, Amazon charged a retail price of $9.99, which was signi…cantly below the typical

$30 price of a hardcover b ook.

132

$9.99 was at or slightly below the wholesale price Amazon

paid to the publishers. Amazon’s pricing strategy was focused more on promoting adoption

of the Kindle platform than making money on e-book sales. Amazon dominated the e-book

132

Facts are from Complaint, United States v. Apple, et al., No. 12 CV 2826, 2012 WL 1193205 (S.D.N.Y.

Apr. 11, 2012); United States v. Apple Inc., 952 F. Supp. 2d 638 (Opinion and Order, Southern District

Court of New York, Judge Denise Cote, July 10, 2013) (referred to as Apple 2013) and United States v.

Apple, Inc., 791 F.3d 290 (2d Cir. 2015) (referred to as Apple 2015). For some analyses of the case, see

Gilbert (2015) and Klein (2017).

market with a market share of almost 90%.

The publishing industry for "trade" books (i.e., general interest …ction and non-…ction

books) was dominated by the "Big Six" - Hachette, HarperCollins, Macmillan, Penguin,

Random House, and Simon & Schuster. Their market share of New York Times Bestsellers

exceeded 90%, and they sold almost 50% of e-books. For several reasons, the Big Six "saw

Amazon’s $9.99 pricing strategy as a threat to their established way of doing business."

133

First, e-books at a retail price of $9.99 were taking sales away from their more pro…table

hardcover sales. Second, the low e-book price might create increasing resistance from con-

sumers to the hardcover price, which would force publishers to reduce the wholesale price

for hardcover books. Third, while Amazon was currently selling e-books at or below the

wholesale price it was paying publishers, it might eventually use its dominant position in

the e-book market to demand a lower wholesale price in order for Amazon to have positive

margins at a retail price of $9.99. "As Hachette’s [Chairman and CEO David] Young put

it, the idea of the ‘wretched $9.99 price point becoming a de facto standard’ for e-books

‘sickened’him."

134

Starting as early as September 2008, the CEOs of the Big Six publishers were privately

meeting about once a quarter to discuss industry matters, including Amazon’s e-book retail-

ing practices. Since publishers competed over authors and agents rather than prices, they

“felt no hesitation in freely discussing Amazon’s prices with each other and their joint strate-

gies for raising those prices.”

135

Some of the policies debated including raising the wholesale

price for e-books and creating their own e-book platform as an alternative to Amazon’s.

Though the publishers did not settle on any policy, their meetings yielded a common

understanding that Amazon’s pricing model had to be changed and that could only happen

if the publishers adopted a uni…ed front. As noted by President and CEO Carolyn Reidy of

Simon & Schuster:

We’ve always known that unless other publishers follow us, there’s no chance

of success in getting Amazon to change its pricing practices [and] without a crit-

ical mass behind us Amazon won’t "negotiate," so we need to be more con…dent

of how our fellow publishers will react if we make a move.

136

This point resonated with a Penguin executive in a report made to the Penguin Group Board

of Directors:

[It] will not be possible for any individual publisher to mount an e¤ective

response [to Amazon] because of both the resources necessary and the risk of

retribution, so the industry needs to develop a common strategy.

137

At the same time that the publishers were in the midst of searching for an alternative to

the current arrangement with Amazon, Apple was preparing to launch the iPad on January

27, 2010. Apple’s goal was to have a well-stocked iBookstore from which iPad users could

133

Apple 2015 at 299.

134

Id. at 300.

135

Id. at 300.

136

Apple 2013 at 651.

137

Id. at 650.

purchase and download e-books by the time of the launch. Hence, Apple needed to work

out an arrangement with the Big Six publishers, and it knew that they wanted to raise the

e-book price above $9.99. As a result, Apple and the publishers were natural allies; Apple

wanted to e¤ectively compete with Amazon in the e-book market, and the publishers wanted

to have an alternative e-book platform to reduce the market power of Amazon.

138

As director of Apple’s digital content stores, Eddy Cue was in charge of negotiations with

the Big Six. The pricing arrangement with Amazon (which was referred to as the "wholesale

model") had the publisher charge a …xed wholesale price for each e-book sold by Amazon,

with Amazon setting the retail price (which was $9.99). Cue prop osed a wholesale model but

with higher price p oints. Hachette (and later HarperCollins) suggested instead an "agency

model" in which the publishers set the retail prices and the publisher and Apple shared the

revenues. While initially resistant, Cue soon embraced it though with Apple retaining some

control over retail prices.

Apple settled on an agency model with a 30% commission, the same commis-

sion it was using in its App Store. Agency would give the Publishers the control

over e-book pricing that they desired, and ensured that Apple would make a

pro…t from every e-book sale in its iBookstore without having to compete on

price. Apple realized, however, that in handing over pricing decisions to the

Publishers, it needed to restrain their desire to raise e-book prices sky high. It

decided to require retail prices to be restrained by pricing tiers with caps.

139

Apple proposed retail price caps of $14.99, $12.99, and $9.99, depending on the book’s

hardcover price.

However, this strategy would not solve the publishers’problem if Amazon continued to

sell e-books at $9.99, for then the publishers would be forced to set the iBookstore retail price

close to $9.99, in order to remain competitive with Amazon. In that situation, publishers

would be making 70% of $9.99, while it was receiving the wholesale price of $9.99 (or larger)

from Amazon. Recognizing that the agency model needed to be augmented with a plan to

control price competition between Apple and Amazon, Apple devised a two-prong strategy.

First, the publishers would force Amazon to adopt the agency model. Second, Apple would

add a Most Favored Nation (MFN) clause to their contracts with publishers.

By moving Amazon to the agency model, the publishers would regain control over the

retail price and thus could set a higher price. The MFN clause mandated that:

If, for any particular New Release in hardcover format, the ... Customer Price

[in the iBookstore] at any time is or becomes higher than a customer price o¤ered

by any other reseller . . . , then [the] Publisher shall designate a new, lower

Customer Price [in the iBookstore] to meet such lower [customer price].

140

138

“As [Apple Senior Vice President] Cue saw it, Apple’s most valuable bargaining chip came from the fact

that the publishers were desperate ‘for an alternative to Amazon’s pricing policies and excited about . . .

the prospect that [Apple’s] entry [into the ebook market] would give them leverage in their negotiations with

Amazon.’”(Apple 2015 at 659)

139

Apple 2013 at 659.

140

Apple 2015 at 304.

Thus, if a publisher o¤ered a lower price for its e-book at Amazon (perhaps due to pressure

from Amazon), it would b e contractually bound to lower its price at the iBookstore to that

same level. An MFN clause would weaken the incentive for Amazon to bargain for a lower

retail price - since it would not result in a price advantage over Apple - and strengthen the

incentive of a publisher to resist a lower retail price - as it would be forced to lower the retail

price at the iBookstore, too.

Apple’s scheme would not necessarily make e-books sales more pro…table for publishers.

For example, consider a hardcover book priced at $25. Apple would cap the retail price

at $12.99 and, if that is the price at which a publisher sold the e-book, it would earn 70%

of $12.99 or $8.75, which is less than the wholesale price Amazon was paying. For some

publishers, e-book gross revenue was expected to decline by 17%.

141

Nevertheless, Apple’s

pricing scheme appealed to the publishers for several reasons. First, higher e-book prices

would raise hardcover sales, which had a higher margin. Second, regaining control over

retail prices could prove more pro…table in the long run if Apple raised the price caps, which

would be quite likely once the e-book market became better established. Third, it would

allow Apple to e¤ectively enter the market and that would weaken the position of Amazon

when bargaining with the publishers.

Summarizing, and as argued by the DOJ in its complaint,

142

the adoption of the Apple

Agency Agreements would raise e-book prices by taking away retail pricing authority from the

retailers, having e-books prices set according to the Apple price caps, and resulting in higher

retail prices for other retailers (speci…cally, Amazon) through their adoption of the agency

model and an MFN clause which would ensure the price caps would not be less than those

set by Apple. In addition to this anti-competitive e¤ect, the plan was exclusionary in that it

sought to use the combined market power of the publishers to force Amazon to switch from

wholesale pricing to the agency model. An interesting wrinkle is that Amazon was dominant

in the e-bo ok retail market in which case the coordinated actions of the publishers could be

viewed as a countervailing force to Amazon. Indeed, the mutual interests of the publishers

and Apple resided in reducing Amazon’s market dominance.

Having devised this plan, Apple now had to get publishers to accept it. Apple quickly

learned that it would not succeed if each negotiation with a publisher was conducted in

isolation of the negotiations with other publishers. Publishers were willing to go along with

Apple’s plan - which involved higher retail prices and pressuring Amazon to accept the

agency model - only if all or almost all publishers were on board.

Each Publisher Defendant required assurances that it would not be the only

publisher to sign an agreement with Apple that would compel it either to take

pricing authority from Amazon or to pull its e-books from Amazon. The Pub-

lisher Defendants continued to fear that Amazon would act to protect its ability

to price e-books at $9.99 or less if any one of them acted alone. Individual Pub-

lisher Defendants also feared punishment in the marketplace if only its e-books

suddenly became more expensive at retail while other publishers continued to

allow retailers to compete on price. As Mr. Cue noted, "all of them were very

concerned about being the only ones to sign a deal with us." Penguin explicitly

141

Apple 2013 at 667.

142

Apple 2012 at 91.

communicated to Apple that it would sign an e-book distribution agreement with

Apple only if at least three of the other "major" publishers did as well.

143

Beginning in Decemb er 2009, Cue’s strategy was to negotiate with a publisher while

keeping it apprised of what the other publishers were intending to do. Cue would be the

hub communicating between the spokes, who were the CEOs of the Big Six. He began by

meeting with each of the Big Six in New York City over December 15-16, 2009. In the

following week, he met again with executives from three publishers and

explained that he had met with all of the Big Six the preceding week, and

had come to the conclusion that the way forward would involve four components.

First, the e-book “industry”needed to move to the agency model, which would

allow the Publishers to set the prices and introduce what Cue euphemistically

termed “some level of reasonable pricing.” Second, Apple would need a 30%

margin on e-bo oks sold through Apple. Third, he proposed setting prices for

New Release e-bo oks at $12.99, that is, $3 over Amazon’s $9.99 price. Finally,

to remove all retail price competition, the Publishers would have to adopt the

agency model for all of their e-tailers.

144

Critical is that Apple was telling each of those publishers that all publishers had to get

Amazon to change its pricing policy. Over January 4-5, 2010, Cue wrote essentially identical

emails to each of the publishers and, for the three that he did not meet in late December,

he began with: “After talking to all the other publishers and seeing the overall bo ok envi-

ronment, here is what I think is the b est approach for e-books.”

145

On January 11th, each

publisher received from Apple its proposed e-book Agency Distribution Agreement and each

was assured that all publishers were getting the same terms.

The goal was to get all publishers under contract prior to the iPad launch on January

27th. Though, as of January 16th, Apple did not have a single contract signed, they soon

started falling like dominoes. Simon & Schuster was the …rst to sign, which led Cue to send

"substantively identical e-mails to Macmillan and Penguin stating that Apple had completed

its …rst agency agreement and was ‘very close’ on two more."

146

On January 22nd, Cue

informed Apple CEO Steve Jobs

that he had commitments from Hachette, S&S, Macmillan, and Penguin that

they would sign. At this point, Penguin required assurance that three other

Publishers were also signing Agreements. As Cue admits, in these …nal days the

Publishers needed reassurance that they would not be alone in signing an agency

agreement with Apple because they feared Amazon’s reaction, reassurance that

Cue readily provided.

147

Come January 26th, …ve of the Big Six had signed, with Random House deciding it would

stick with the wholesale model. As the District Court noted:

143

Id. at { 69.

144

Apple 22013 at 660.

145

Id. at 661.

146

Id. at 674.

147

Id. at 674.

[I]n less than two months, Apple had signed agency contracts with …ve of

the six Publishers, and those Publisher Defendants had agreed with each other

and Apple to solve the “Amazon issue” and eliminate retail price competition

for e-books. ... This would not have happened without Apple’s ingenuity and

persistence. Apple’s task had not been easy, but it had succeeded. As Reidy

acknowledged in an email to Cue on January 21, working with the Publishers

had been like “herding ... cats.”

148

There were also direct communications among the publishers during this period as

"[d]uring their negotiations with Amazon, the Publisher Defendants shared their progress

with one another."

149

Of particular note is how the publishers joined forces when Amazon re-

taliated against Macmillan. When Macmillan CEO John Sargent met with Amazon, he told

Amazon that they could either adopt the agency model or not receive the Kindle versions

of its new releases for seven months (which was the length of time that these books were

classi…ed as "new releases" under the contract with Apple). Amazon responded by removing

the buy buttons at its web sites for Macmillan print and Kindle versions; the books were

listed at Amazon but a customer could not buy them. Macmillan expected the support of

the other publishers, for Sargent wrote:

The deal that 5 of us did with Apple meant someone was gonna have to do

it […rst]. ... The optics make it lo ok like I stood alone, but in the end I had no

doubt that the others would eventually follow.

150

And indeed they did.

Hachette’s Nourry [wrote] to Sargent: “I can ensure you that you are not

going to …nd your company alone in the battle” with Amazon. The next day,

Penguin’s Makinson similarly wrote, “[j]ust to say that I’m full of admiration for

your articulation of Macmillan’s position on this. Bravo.”Internally, Hachette’s

Nourry told Young that he wanted to “enter in the battle as soon as possible,”

and in an allusion to Macmillan’s small size, that he was “thrilled to know how

A[mazon] will react against 3 or 4 of the big guys.”Over the weekend, it b ecame

obvious to Amazon that its strategy had failed. The feedback was mixed, but

included intense criticism of Amazon by customers and publishers.

151

On January 31st, Amazon announced that it would adopt the agency model with Macmillan,

and had also done so with Hachette, HarperCollins, and Simon & Schuster by March 2010.

Though the key retailer to convert was Amazon, the …ve publishers also switched other

retailers - such as Barnes & Noble (which had its Nook e-book reader) and Google’s e-

bookstore - to the agency model.

In sum, Apple proved to be an e¤ective hub in coordinating the adoption of the agency

model between …ve of the Big Six publishers and e-book retailers Apple and Amazon. The

District Court summarized well the coordinating role played by Apple.

148

Id. at 678.

149

Id. at 681.

150

Id. at 680.

151

Id. at 680.

A chief stumbling block to raising e-book prices was the Publishers’fear that

Amazon would retaliate against any Publisher who pressured it to raise prices.

Each of them could also expect to lose substantial sales if they unilaterally raised

the prices of their own e-books and none of their competitors followed suit. This

is where Apple’s participation in the conspiracy proved essential. It assured each

Publisher Defendant that it would only move forward if a critical mass of the

major publishing houses agreed to its agency terms. It promised each Publisher

Defendant that it was getting identical terms in its Agreement in every material

way. It kept each Publisher Defendant apprised of how many others had agreed

to execute Apple’s Agreements. As Cue acknowledged at trial, “I just wanted to

assure them that they weren’t going to be alone, so that I would take the fear

awa[y] of the Amazon retribution that they were all afraid of.”

152

The evidence shows that the adoption of the agency model by Apple and Amazon caused

e-book prices to rise.

153

To begin, e-book prices were generally set at Apple’s price caps.

In the …ve months that followed [the opening of the iBookstore], the Publisher

Defendants collectively priced 85.7% of their New Release titles sold through

Amazon and 92.1% of their New Release titles sold through Apple within 1%

of the price caps. This was also true for 99.4% of the NYT Bestseller titles on

Apple’s iBookstore, and 96.8% of NYT Bestsellers sold through Amazon.

154

Figure 3 was o¤ered by one of Apple’s economic experts at trial. After the iBookstore

opened in April 2010, Amazon’s e-book prices went up for all publishers (except Random

House which did not adopt the agency model). A statistical analysis by an expert witness

for the DOJ found that the average price of e-books rose by 16.8% in the six months after

Amazon adopted the agency model compared to six months before (see Table 1). Finally, a

study has tracked e-bo ok prices after publishers switched back to the wholesale model over

September 2012 to September 2013 as part of their settlement with the DOJ (De los Santos

and Wildenbeest, 2017). With retailers gaining control of prices at that point, Amazon’s

e-book prices fell by 18%. In sum, e-book prices rose when publishers adopted the agency

model in 2010 and fell when they moved away from the agency model over 2012-13. The

evidence is compelling that Apple’s hub-and-spoke cartel had an anti-competitive e¤ect on

prices.

In response to the civil suit …led against the …ve publishers and Apple by the DOJ and 33

states and territories, all …ve Publisher Defendants eventually settled. Apple chose to go to

trial where it was found guilty of violating Section 1 of the Sherman Act.

155

It lost on appeal

to the Second Circuit Court,

156

which resoundingly supported the judgment of the District

152

Id. at 692-3.

153

For recent theories exploring the rationale for and implications of the agency model in the context of a

market like ebooks, see Gaudin and White (2014), Foros, Kind, and Sha¤er (2017), and Johnson (2017). De

los Santos and Wildenbeest (2017) o¤ers a test of these various theories.

154

Apple 2013 at 682.

155

United States v. Apple Inc., 952 F. Supp. 2d 638 (Opinion and Order, Southern District Court of New

York, Judge Denise Cote, July 10, 2013)

156

United States v. Apple, Inc., 791 F.3d 290 (2d Cir. 2015)

Court, and the Supreme Court denied Apple’s petition for certiorari. Both the District and

Circuit courts found Apple to have committed a per se violation and even commented that

it would have been found guilty under the rule of reason.

Because this conspiracy consisted of a group of competitors - the Publisher

Defendants - assembled by Apple to increase prices, it constituted a “horizontal

price-…xing conspiracy”and was a per se violation of the Sherman Act. It con-

cluded, moreover, that even if the agreement to raise prices and eliminate retail

price competition were analyzed under the rule of reason, it would still constitute

an unreasonable restraint of trade in violation of § 1. In the district court’s view,

Plainti¤s’ experts persuasively demonstrated that the agreement facilitated an

“across-the board price increase in e-books sold by the Publisher Defendants”

and a corresponding drop in sales. Apple, on the other hand, failed to show that

“the execution of the Agreements,” as opposed to the launch of the iPad and

“evolution of digital publishing more generally”(which were independent of the

Agreements), “had any pro-competitive e¤ects.”

157

As vertical agreements are a feature of hub-and-spoke cartels, it should be emphasized

that the those agreements are typically legal. Exempli…ed in the current case, the Apple

Agency Agreement was lawful. What was unlawful was Apple coordinating publishers so

that they would collectively adopt it with Apple and Amazon. Nevertheless, a remedy for

the horizontal agreement could mean prohibiting the associated vertical agreement, which is

what happ ened in the e-books case.

4 An Analysis of Hub-and-Spoke Collusion

Though most collusion occurs in intermediate goods markets, all of the cases of hub-and-

spoke collusion had the object of reducing competition in a retail market. Collusion took one

of two forms. In six of the nine cases, it was standard price …xing: Retailers coordinated on

a higher common price, with an upstream manufacturer as the hub and downstream retailers

as the spokes. In two cases - movie exhibition and U.S. toys - the collusive scheme was ex-

clusionary. A dominant downstream retailer (hub) coordinated upstream suppliers (spokes)

in harming a class of rivals to the hub. The e-books case is a hybrid of the two schemes. It

had an exclusionary component - in that it harmed Amazon which was a downstream rival

Apple (hub) - and it was designed to raise downstream retail prices, which it did.

4.1 Properties of Collusive Schemes

4.1.1 Exclusion of Downstream Firms

The exclusionary scheme implemented in the markets for movie exhibition and toys (U.S.)

was motivated by the aggressiveness of a class of competitors to a dominant retailer. With

low prices and high quality (in the form of showing two …lms), second-run theaters were

157

Id. at 312.

taking demand from dominant movie theater chain Interstate. Warehouse clubs were taking

demand from toy retailer Toys "R" Us through lower prices, and threatening Toys "R" Us’

reputation as the lowest-price retailer. Rather than respond with lower retail prices, these

two dominant retailers chose to harm their rivals with the assistance of upstream suppliers.

Interstate had upstream movie distributors make as a condition for them to supply second-

run theaters that those theaters raise their prices and only show one …lm per admission.

Toys "R" Us had upstream toy manufacturers not supply any products to warehouse clubs

that they supplied to Toys "R" Us.

The participation of the downstream retailer is easy to understand. It was the …rm

that devised the collusion plan and it would clearly bene…t from excluding a class of rival

retailers. More challenging is explaining the participation of the upstream suppliers. To

shed light on this issue, we ask two related questions. First, does collusion raise upstream

suppliers’ pro…ts compared to competition? If it does then upstream suppliers have an

incentive to make collusion work. There is still the issue of accomplishing that goal because

it requires coordination; each upstream supplier’s participation relies on it believing that

the other suppliers will also participate. Second, is it an equilibrium for upstream …rms to

participate? That is, if the other upstream …rms complied with the collusive plan, would

it be in the best interests of an upstream …rm to comply? While collusion may be easier

to implement when it bene…ts all involved parties, that condition need not be satis…ed for

collusion to occur. It is enough that, given all other …rms collude, each …rm’s pro…ts are

higher by also colluding.

In evaluating whether exclusion against a class of downstream retailer raised upstream

suppliers’pro…ts, there are some opposing forces at play. On the one hand, more intense

downstream competition can bene…t upstream suppliers by raising downstream, and thereby

upstream, demand. On the other hand, this class of downstream competitors caused demand

to shift from high-margin sales (at …rst-run theaters and Toys "R" Us) to low-margin sales

(at subsequent-run theaters and warehouse clubs).

158

As a result, total pro…ts for the vertical

chain may have fallen. What the implication of such a decline is for upstream …rms’pro…ts

depends on how pro…t sharing between the upstream and downstream …rms is impacted by

exclusion. If more intense downstream comp etition increases the bargaining power of the

upstream manufacturers then they might receive a larger share of pro…ts without exclusion

which could yield more upstream pro…ts in spite of lower total pro…ts for the vertical chain.

Related to this point, an upstream toy manufacturer commented that the warehouse clubs

made it less dependent on TRU, and a natural consequence of that is more bargaining power.

It is then possible that competition from this class of downstream …rms could have led to

a rise in upstream suppliers’demand and an improvement in their bargaining position with

the dominant downstream retailer, and that might have them better o¤. In that case, the

upstream suppliers would have been worse o¤ under a coordinated strategy of exclusion.

In light of these countervailing forces and the absence of appropriate data and empirical

analysis, it remains an open question whether the upstream spokes were made better or worse

o¤ with the exclusionary scheme promoted by the downstream hub. However, some indirect

evidence can be brought to bear. An argument in favor of them having been made better

o¤ is that an upstream …rm could have reported the scheme to the authorities and thereby

158

This e¤ect is emphasized in Klein (2017).

prevented its implementation. However, that tactic does come with a potential cost in the

form of retaliation by the downstream dominant retailer. Souring a key vertical relation

might have made the upstream supplies compliant even if they were harmed by the collusive

scheme.

Indirect evidence in favor of the upstream suppliers having been worse o¤ under collusion

is that the downstream hub found it necessary to use the threat of exclusion against any

upstream …rm that did not comply. For suppose collusion did make an upstream …rm worse

o¤. In that case, an upstream …rm would prefer to destabilize any collusion and, as a result,

would do what is in its own best short-run interests. In the absence of an exclusionary threat

from the downstream retailer, an upstream …rm’s short-run best interests are not to comply

with the collusive plan, and that is irresp ective of what other upstream …rms are expected

to do. For if other upstream …rms were expected not to comply then an upstream …rm

would obviously not want to comply as evidenced by its conduct under competition. If they

were expected to comply then the a noncompliant upstream …rm’s demand would be even

higher, in which case it would have a stronger incentive to supply those downstream …rms in

violation of the collusive plan. For example, if all other toy manufacturers did not supply the

warehouse clubs with their best-selling products then a toy manufacturer would have even

stronger demand from warehouse clubs. Similarly, if the other movie distributors restricted

theaters to set higher admission prices and show only one …lm then a movie distributor

would face stronger demand from those theaters if it did not place those restrictions on

them. Thus, if collusion did harm the upstream suppliers and given the unilateral incentives

not to comply, it would seem the downstream retailer would fail in implementing its plan

to exclude rival retailers. This would then provide a motivation for the downstream retailer

to alter the incentives faced by upstream suppliers by, for example, threatening exclusion

against them: If an upstream …rm did not comply then the downstream …rm would not buy

the upstream …rm’s products (either not leasing their …lms or purchasing their toys). While

a threat of exclusion against an upstream …rm for not abiding by the collusive plan is not

inconsistent with collusion bene…tting upstream …rms, a threat of exclusion (or something

like it) is required if collusion does harm upstream …rms.

Let us now turn to the second question: Is it an equilibrium for upstream …rms to

participate in the collusive plan? If collusion resulted in higher pro…ts for upstream …rms

than were earned under competition, an upstream …rm would …nd it optimal to comply if it

expected all (or enough) other upstream …rms to comply and the increase in long-run pro…ts

from collusion is su¢ cient to deter a …rm from any short-gain it might receive from deviating

(and most likely causing a subsequent return to competition). In that situation, an upstream

…rm would comply if it believed other upstream …rms would comply and collusion was

su¢ ciently more pro…table than comp etition. If instead collusion resulted in lower pro…ts for

upstream …rms than were earned under competition, an upstream …rm would …nd it optimal

to comply only if exclusion by the downstream retailer was su¢ ciently harmful and credible.

For the Toys "R" Us case, "su¢ ciently harmful" means that an upstream …rm preferred

making fewer sales to warehouse clubs (as part of the collusive plan) to fewer sales to Toys

"R" Us (as part of the exclusionary punishment for deviating from the collusive plan). Given

that it would be very costly for the downstream retailer to cut o¤ purchases from multiple

upstream suppliers, the credibility of this exclusionary threat relied on each upstream …rm

believing that the other upstream …rms would collude. In that case, noncompliance by an

upstream supplier would only require the downstream retailer to exclude that one upstream

supplier. As long as collusion is su¢ ciently pro…table for the upstream …rms or the threat of

exclusion was su¢ ciently harmful for the upstream …rms, their participation in the collusive

plan would be optimal if each upstream …rm believed that the other upstream …rms were

complying with the collusive plan. It is then critical for the success of collusion that the

downstream hub convince the upstream spokes that they are all planning to comply with

the collusive plan.

There are several points made for when collusion involved exclusion against a class of

downstream retailers. First, the dominant downstream retailer (hub) formulated this plan

and was made better o¤ with its implementation. Second, it is unclear whether the upstream

suppliers (spokes) were better o¤. Third, it is unnecessary for upstream suppliers to be better

o¤ with collusion; all that is needed is that it is in each upstream supplier’s interests to

collude given other upstream suppliers are expected to collude. Fourth, critical to providing

upstream suppliers with the incentive to collude is a threat of exclusion against them if they

don’t, and that threat is credible only if an upstream supplier believes others will comply.

4.1.2 Price Fixing by Downstream Firms

The more common collusive scheme was for an upstream manufacturer (hub) to coordinate

downstream retailers (spokes) in raising their prices. With the exception of the UK cheese

case, the upstream manufacturer established a recommended (or minimum suggested) retail

price to which all retailers were to adhere. Generally, the motivation for collusion was low

retail margins. In some cases, it was the retailers who complained to the manufacturer, as

when London toy retailers lamented to Hasbro about the low margins on Hasbro’s toys in

their stores. With the replica football kits case, retailer JJB complained to manufacturer

Umbro about Sports Soccer’s low prices.

159

In contrast, the instigation for collusion appar-

ently came from the upstream manufacturers in the Polish paints and varnishes and U.S.

pharmaceutical products cases.

In general, retailers can be expected to bene…t from a collusive scheme which results

in a common price increase (though exceptions are discussed below). Less obvious is that

upstream manufacturers would bene…t. Higher retail prices mean lower retail demand and,

therefore, lower demand for the manufacturers’products. As manufacturers do not receive

those higher prices charged in the retail market, manufacturers’pro…ts would seem to decline

from this brand of hub-and-spoke collusion.

A common rationale for a manufacturer desiring higher retail prices such as through

resale price maintenance (whereby a manufacturer requires retailers to set some minimum

price) is that it promotes competition among retailers in non-price dimensions (such as

sales personnel and advertising) which can raise demand for the manufacturers’products.

160

However, there was no documentary evidence in the six cases that this consideration was

relevant, and it was de…nitely not an explicit part of the collusive scheme that retailers

159

JJB was the leading retailer and Sports Soccer was a smaller …rm intent on growing its market share

through low prices. Thus, the situation is similar to the movie exhibition and U.S. toys cases. That JJB did

not act as a hub and pursue an exclusionary response to Sports Soccer is likely due to it being les s dominant

than Interstate and Toys "R" Us.

160

See, for example, chapter 6.2 in Motta (2004).

engage in more demand-promoting activities. In two cases, documents state that higher

retail prices bene…tted upstream …rms by allowing them to raise their prices. In the paints

and varnishes cases, the upstream manufacturer was …nding retailers resistant to accepting a

higher wholesale price because competition among retailers had compressed margins. Thus,

the manufacturer saw higher retail prices as a necessary …rst step to raising its own prices.

The UK cheese case began with the objective of raising the price of raw milk. If supermarkets

raised retail prices then processors could raise the prices they charged to supermarkets which

would then allow processors to raise the prices they paid to farmers for raw milk. In the UK

toys case, Hasbro could have bene…tted as part of a quid pro quo arrangement. In exchange

for supporting the pricing initiative which raised retail prices, retailers would support the

listing initiative that resulted in Argos and Littlewoods continuing to stock certain Hasbro

toys that were in jeopardy of being dropped by the retailers.

In two cases, there was some evidence that not all retailers bene…tted b ecause they

were discontent with the current market allocation. At the time of collusion, Dart Drug

in the pharmaceutical products case and Sports Soccer in the replica football kits case had

been pricing below their competitors with the apparent goal of increasing market share.

Sports Soccer had that as an explicit goal, while Dart Drug’s policy of advertising products

at deep discounts is supportive. The most compelling evidence that those two companies

preferred not to collude is that investigations were begun because they …led complaints with

the competition authority. It is also interesting that upstream manufacturer Parke, Davis

threatened to exclude supplies to any retailer that did not comply with the collusive scheme.

Such a policy is consistent with a need to alter the incentives of some retailers in order to

get them to raise their prices.

161

With UK toys, replica football kits, and pharmaceutical products, there was one hub-and-

spoke cartel. The exclusive manufacturer of a product - Hasbro, Umbro, and Parke, Davis

respectively - coordinated their retailers. In the Polish paints and varnishes and the Belgian

drugstore, perfumery, and hygiene (DPH) products cases, several hub-and-spoke cartels were

operating simultaneously. Each of the three leading manufacturers of paints and varnishes

had its own hub-and-spoke cartel with downstream retailers. The various manufacturers of

DPH products - such as Colgate, Proctor & Gamble, and Unilever - ran hub-and-spoke cartels

with an overlapping set of retailers, such as Carrefour, Delhaize, and Intermarché.

162

How

did it come about that several upstream …rms all formed hub-and-spoke cartels? In the paints

and varnishes cases, there was no evidence of communication between the manufacturers.

Could retailers have shared information with other manufacturers regarding its collusion

with another manufacturer? Could information had ‡owed so easily between manufacturers

and retailers that the presence of a hub-and-spoke cartel had become widely known?

There are several takeaways. First, downstream retailers (spokes) generally bene…t by

collusion that raises retail prices. Second, there are exceptions to that rule when a down-

stream retailer is interested in growing its market share (and, indeed, it is that desire for

growth which may have led it to set low prices and could have been the reason for collusion).

161

For an economic analysis showing the e¤ect of threatening exclusion on collusive prices, see Van Cayseele

and Miegielsen (2013).

162

The UK cheese case also involved multiple hub-and-spoke cartels with the hubs being the three proces-

sors which shared some retailers. However, multiple cartels was probably necessary because the processors

supplied a homogeneous product.

Third, it is not immediately obvious how the upstream supplier (hub) bene…ts from higher

retail prices though in some cases it was apparently intended to allow them to raise their

wholesale prices.

4.2 Communications to Achieve Mutual Understanding

There were some real challenges in getting all spokes on board with a collusive scheme. First,

a spoke would not comply unless it thought enough other spokes were going to comply. In

U.S. toys, UK toys, and e-books, it is particularly well documented that a spoke needed the

assurance that the other spokes were going to comply. As a result, a hub could not pursue

a strategy of sequentially getting spokes on board over an extended length of time; there

must be coordinated adoption. Second, in a few cases, it is arguable that some spokes did

not want to comply even if other spokes complied, and they went along only because of the

threat of exclusion. Presumably, that made the hub’s task harder, for it would seem more

challenging to shift …rms to an equilibrium that makes them all (or at least some) worse o¤.

Achieving mutual understanding among spokes that all would be abiding by the collusive

arrangement required extensive bilateral communication between the hub and each spoke.

These communications involved more than sharing the intentions of other spokes. The

hub had to convince each spoke that others would indeed go through with their intentions.

Though, by design, we have focused on cases in which the hub played the primary role in

coordination, it is still notable how little direct communication there was among the spokes.

There was some in the e-books case, where direct communication among publishers preceded

the arrival of Apple, and the hub in the replica football kits case organized a meeting among

the spokes. Still, by and large, direct communication was absent.

Crucial to the e¤ectiveness of collusion is monitoring for compliance. If a …rm does not

anticipate being monitored then it is likely to cheat. Here, monitoring also served to assure

a spoke that other were complying, and that may have been crucial for a spoke to continue

to comply. The hub actively monitored spokes for compliance, and this is well documented

in U.S. toys, UK toys, paints and varnishes, and DPH products. Sp okes also monitored each

other - as documented in U.S. toys and UK toys - and conveyed their …ndings to the hub,

upon which the hub would contact a deviant spoke and seek to get it back into compliance.

Whether it was related to coordination or monitoring, spokes generally did not directly

communicate with each other. They almost exclusively communicated through the hub;

whether it was to convey pricing intentions or report that another spoke was not complying.

This lack of direct communication could be due to the su¢ ciency of indirect communication

for e¤ective collusion. It could also re‡ect the recognition that direct communication was

more dangerous from a legal perspective. While communications between an upstream …rm

and a downstream …rm are to be expected in a competitive market, communications between

competitors are not.

163

As the competition authority had strong incentives to document direct communications between the

spokes, it is unlikely that spokes did directly communicate and there was a failure to …nd evidence of it.

4.3 Role of the Hub

One of the most compelling regularities for these cases is the crucial role played by the

hub. To begin, it was often the hub that had object. Collusion was initiated by the hub in

the cases involving movie exhibition, U.S. toys, replica football kits, paints and varnishes,

and pharmaceutical products. In the UK toys and e-books cases, the spokes might have

voiced complaints for which collusion could be a response, but it was the hub that was

entrepreneurial in developing a collusive plan. Only in cheese (for which object originated

with farmers) and DPH (for which there is a lack of evidence as to who initially had object)

can one not say that the hub was the primary source of object.

As described above, the hub was essential in achieving mutual understanding. It was the

hub that went back and forth between spokes to obtain their intentions and then share them

with the other spokes. The proposition that the hub could facilitate an agreement inad-

vertently - for example, by incidentally mentioning a retailer’s pricing intention to another

retailer - is contrary to the extraordinary e¤ort that went into convincing spokes. That was

the case whether the hub was Toys "R" Us, Umbro, or Apple.

164

Finally, the hub was pivotal in monitoring. It performed its own monitoring, collected

reports from spokes, and then acted on this information by contacting and working with non-

compliant spokes. As noted by the Polish competition authority in the paints and varnishes

case with regards to hub PCW,

PCW had played the role of a "mediator" who would "appease disputes",

explain price di¤erences, and inform retailers about how quickly would the prices

of their competitors return to the agreed level.

165

It is natural to ask with hub-and-spoke cartels, why the spokes did not just directly

communicate. While that is an option, the multiple and crucial roles played by the hub

suggests that collusion would not have occurred but for the hub’s intervention. The hub was

the entrepreneur when it came to forming a cartel, and its strategic lo cation at the nexus of

a communication network made it highly e¤ective in coordinating and monitoring.

4.4 E¤ectiveness of Collusion

Two key metrics of the e¤ect of collusion are the overcharge (how much higher is price) and

duration (the length of time during which prices are higher). As the case material generally

does not have overcharge estimates (with e-books being an exception), our attention will

focus on duration.

Considering all cartels, and not just hub-and-spoke cartels, most data sets report a mean

duration of (discovered) cartels between 5 and 8 years.

166

As reported in Table 3, what is

164

This property could be the product of sample selection bias. Perhaps collusion that occurred with only

minimal communications from the hub was less likely to be detected and prosecuted.

165

Bolecki (2011), p. 32.

166

A survey of several studies (Levenstein and Suslow, 2006) …nds mean duration of discovered cartels

between 4.5 and 7.5 years. Taking account of selection bias with having been discovered, Harrington and

Wei (2017) …nds, for a data set with a mean duration of discovered cartels of 5.8 years, a mean duration of

(discovered and undiscovered) cartels in the range of 5.3 to 6.8 years.

striking about hub-and-spoke cartels is their brevity. These cartels had mean and median

duration of around 2 years with a maximum duration of 4.5 years.

Table 3: Cartel Detection and Duration

Market Country Time until detection Duration (of e¤ect)

Movie exhibition U.S. ? 1 year

Toys U.S. 2 years 3 years

Toys U.K. 1.5 years 1.5 years

Paints and varnishes Poland ? 3 years

Pharmaceutical products U.S. 3 months 3 months

Replica football kits U.K. 6 months 1 year

Cheese U.K. 2 years 2 years

Drugstore, perfumery, Belgium 4 years 4.5 years

and hygiene products

e-books U.S. 2 years 3 years

The short duration re‡ects the rapidity with which the cartel was detected.

167

In the

movie exhibition and U.S. toys cases, it is possible that the exclusionary action in‡icted on a

class of competitors to the hub resulted in them reporting it, though there is no evidence of

such. In the replica football kits case, Sports Soccer complained to the O¢ ce of Fair Trading,

while Dart Drug from the pharmaceutical products case complained to the U.S. Department

of Justice. Recall that both were low-priced …rms seeking to expand their market share prior

to cartel formation, and apparently were recalcitrant spokes.

More speculative is that detection was due to "leaky" information ‡ows. With a standard

cartel among rival …rms, the cartel members strive to keep their communications con…dential

because they know that revealing any such communications creates a high likelihood of

prosecution and conviction. That is not the case with a hub-and-spoke cartel, for an upstream

…rm and downstream …rm regularly communicate under competition. Furthermore, it is

perfectly reasonable for a retail …rm to convey its pricing intentions to an upstream supplier.

Perhaps it is the normalcy of those communications that led members of a hub-and-spoke

cartel not to be as diligent in keeping it con…dential, which could then result in other parties

learning that an upstream supplier is sharing the pricing intentions of one retailer with

another retailer. The widespread ‡ow of such information is consistent with the presence of

simultaneous hub-and-spoke cartels as occurred in paints and varnishes, DPH, and cheese.

5 Hub-and-Spoke Collusion and the Enforcement of

Competition Law

Having examined how hub-and-spoke cartels operate, let us turn to the enforcement of com-

petition law. After brie‡y reviewing relevant U.S. jurisprudence, approaches are discussed

167

The time of detection is the earliest time for which there is evidence that a competition authority

suspected collusion, as re‡ected in receiving a complaint, requesting information from a cartel member,

opening an investigation, or having a leniency application.

for establishing when …rms have engaged in illegal hub-and-spoke collusion.

168

Hub-and-spoke collusion involves an information exchange in which a …rm (hub) collects

and disseminates information on price or supply intentions among its upstream suppliers or

downstream customers (spokes) for the purpose of restraining competition. An immediate

enforcement challenge is that many vertical information exchanges are lawful due to having

procompetitive justi…cations. It is standard business practice for an upstream supplier and

a downstream customer to communicate, and that includes a downstream retailer expressing

its pricing intentions to an upstream supplier. The law also gives considerable discretion

to upstream and downstream …rms in laying out the terms of their arrangement, which

could include a downstream …rm agreeing not to buy from another upstream supplier or an

upstream …rm agreeing not to sell to another downstream …rm. As noted in Toys "R" Us

(1998):

We …nd that TRU’s conduct violates Section 5 of the FTC Act. In doing

so, we do not intrude on the right of a trader unilaterally to announce terms on

which it will deal with suppliers, even if those terms disadvantage a rival. That

is a company’s long-recognized right under United States v. Colgate & Co., 250

U.S. 300 (1919), rea¢ rmed by the Supreme Court in 1984 in Monsanto Co. v.

Spray-Rite Servs. Co., 465 U.S. 752 (1984).

169

Clarifying when vertical information exchanges crossover into illegal territory, the FTC in

Toys "R" Us (1998) went on to say: "What a …rm cannot do is to coordinate with rival …rms

on the terms of that arrangement."

170

It is the presence of a rim - a horizontal agreement

among the spokes - that moves vertical agreements into illegal territory. Multiple vertical

agreements between a hub and spokes that do not connect the spokes through an agreement

(and thus are "rimless") is not actionable.

Though vertical agreements - such as resale price maintenance and tying - are considered

under the rule of reason in the U.S., hub-and-spoke cartels are per se illegal according

to the same standards as regular cartels; it is a per se o¤ense when there is evidence of

communication su¢ cient to establish a horizontal agreement among the spokes.

171

Per se

illegality clearly applies to the spokes, for they are the ones with the horizontal agreement. In

implicating the hub, "plainti¤s must demonstrate both that a horizontal conspiracy existed,

and that the vertical player was a knowing participant in that agreement and facilitated the

scheme."

172

Having done that, per se illegality applies to the hub as well; it is seen to be as

culpable as the spokes.

While p er se illegality pertains to the horizontal agreement, it is not applied to the indi-

vidual vertical agreements. Nevertheless, the court may choose to prohibit certain vertical

agreements as part of a remedy, as it did in the e-books case. While the Apple Agency

168

This discussion has been informed by the analysis in Orbach (2016) for the U.S., Odudu (2011) for the

UK, and Prewitt and Falls (2015) for the EU.

169

See supra note 19, at 1.

170

Id. at 1.

171

The per se illegality of hub-and-spoke cartels was a¢ rmed in Toys "R" Us, Inc.,v. Federal Trade

Commission, Appellee. No. 98–4107, U.S. Court of Appeals for the Seventh Circuit 221 F.3d 928, August

1, 2000.

172

United States v. Apple, Inc., 952 F. Supp. 2d 638, 690–91 (pp. 112-3)

Agreement was lawful, the Second Circuit prohibited its adoption because publishers had

illegally coordinated a near-industry-wide adoption of it.

The next issue is how one proves that …rms have a hub-and-spoke conspiracy; that is,

…rms have a horizontal agreement, even though they did not directly communicate, and it

was facilitated by an upstream or downstream …rm. Interstate (1939) set a key precedent

when it stated:

It was enough that, knowing that concerted action was contemplated and

invited, the distributors gave their adherence to the scheme and participated in

it. Each distributor was advised that the others were asked to participate; each

knew that coop eration was essential to successful operation of the plan. They

knew that the plan, if carried out, would result in a restraint of commerce, ...

and knowing it, all participated in the plan.

173

It is argued that the current interpretation of Interstate (1939) is that vertical relations

comprise illegal hub-and-spoke collusion when:

(1) two or more competitors enter into vertical agreements with a single up-

stream or downstream …rm; (2) the vertical agreements could bene…t each com-

petitor only if its rivals enter into similar agreements; and (3) the …rm that facil-

itates all the vertical agreements persuades each competitor that its competitors

will take a similar action.

174

This legal perspective captures some forms of hub-and-spoke collusion, but not all. For

example, suppose retailer A is charging the manufacturer’s recommended price (MRP) and

complains to upstream supplier B about retailer C pricing below the MRP. Firm B relays

this complaint to retailer C and, subsequently, C raises its price to the MRP. As there is no

reason for …rm A to complain to …rm B unless it is to solicit B’s assistance in raising …rm

C’s price, the intent of …rm A’s communication is to facilitate coordination between retailers

A and C. That is presumably …rm B’s intent as well when it passes the complaint along to

…rm C. When …rm C receives the complaint, it could take it as an invitation to coordinate

on the MRP, and …rm C accepts that invitation when it raises its price to the MRP. All

of the essential elements of an unlawful horizontal agreement are present, but there are no

vertical agreements. It would seem to be an unnecessary …ction, and a bit of a stretch, to

view upstream supplier B passing along the complaint of …rm A to …rm C as an agreement

between …rms A and B or between B and C.

175

The UK’s Competition Appeal Tribunal developed an approach referred to as the "A to

B to C" test. Hub-and-spoke collusion is said to exist if

173

Interstate Circuit, Inc., et al. v. United States 306 U.S. 208, 226 (1939)

174

Orbach (2016), p. 6.

175

One should not attach any signi…cance to the collusive price being the MRP; the MRP only serves as a

focal point. If there is a "conventional" retail price (e.g., a price that all retailers have charged in the recent

past), that could take the place of the MRP and thereby remove any role of the upstream …rm with regards

to the price level.

(i) retailer A discloses to supplier B its future pricing intentions in circum-

stances where A may be taken to intend that B will make use of that information

to in‡uence market conditions by passing that information to other retailers (of

whom C is or may be one), (ii) B does, in fact, pass that information to C in

circumstances where C may be taken to know the circumstances in which the in-

formation was disclosed by A to B, and (iii) C does, in fact, use the information

in determining its own future pricing intentions, then A, B, and C are all to be

regarded as parties to a concerted practice having as its object the restriction or

distortion of competition.

176

Alternatively stated, the conditions are: (i) spoke A discloses intentions on future conduct

(such as prices) to hub B for which A intends that B will pass onto spoke C; (ii) B passes

information to C, where C knows it was passed on with A’s understanding; and (iii) C uses

the information in determining its own conduct (such as pricing). In its contribution to

an OECD report on information exchanges, the UK summarized this test by stating that

"the provision of, receipt of or passing on of information between competitors through an

intermediary in circumstances where it can be taken for one to have intended to in‡uence

the market conduct of the other is anticompetitive".

177

The relevance of that test for the U.S. is not immediate because the law in the UK is

more expansive than that in the U.S.. Like the EU, the UK prohibits "concerted practices"

which is a category of practices that falls short of an agreement. A concerted practice

could be an information exchange that, by reducing uncertainty between …rms, results in

supracompetitive prices, but does not go so far as to form an agreement. This means plainti¤s

could prove there is a violation of Chapter I of the UK Competition Act 1998 (or Article

101 of the Treaty on the Functioning of the European Union) without showing there is an

agreement, which must be done to establish a violation of Section 1 of the Sherman Act.

In light of the di¤erences in the law, let me reformulate the "A to B to C" test from the

perspective of U.S. jurisprudence. The issue is when do communications between vertically

related …rms form a per se illegal horizontal agreement. Consider a vertical information

exchange between …rm A (spoke) and …rm B (hub) and between …rm C (spoke) and …rm B

where …rms A and C are competitors and …rm B is either an upstream supplier or downstream

buyer to …rms A and C. A horizontal agreement has two elements to it: …rm A invites …rm

C to collude, and …rm C accepts that invitation. In the context of a hub-and-spoke cartel, I

prop ose that …rm A invites …rm C to collude when: 1) the information conveyed by …rm A

to …rm B would have facilitated a horizontal agreement had it been conveyed directly to …rm

C; and 2) when conveying the information to …rm B, …rm A could have reasonably foreseen

that …rm B would convey the information to …rm C.

In determining when these two conditions are satis…ed, there is considerable jurisprudence

regarding condition (1). As to condition (2), here are four methods for substantiating that

a spoke could have reasonably foreseen that the hub would share the information without

another spoke.

176

See supra note 51, at { 141.

177

OECD, “Information Exchange Between Competitors under Competition Law,”DAF/COMP(2010)37,

2010, pp. 286-7

First, the hub expresses to spoke A that it intends to share A’s expressed intention with

spoke C. While it is possible that spoke A may not b elieve the hub, the burden should be

on the spoke to convince the court why the hub should not have been believed.

Second, the hub provides information from sp oke C to spoke A that is of a similar type

to the information that spoke A provided to the hub. In concluding that condition (2) is

satis…ed, one is drawing the inference that spoke A understands the information exchange is

reciprocal. If hub B conveys the pricing intention of spoke C to spoke A then, when spoke

A conveys its pricing intentions to hub B, it is reasonable for spoke A to expect that hub B

will convey spoke A’s pricing intentions to spoke C. This makes sense because if it is optimal

for the hub to share such information from spoke C to spoke A, it is likely to be optimal for

the hub to share similar information from spoke A to spoke C.

178

However, one does want

to be careful, for spoke A might have provided its pricing intentions to B without any intent

of it being shared by B to C, and it was purely on B’s own initiative that B collected C’s

pricing intentions and conveyed them to A. In that instance, B might be culpable but A

is not. The following requirement should su¢ ce to deal with that situation: If A shares its

pricing intentions with B after having received C’s pricing intentions from B then A could

reasonably foresee that the pricing intentions it shared with B will be passed on to C.

Third, it is in the hub’s best interests to share a spoke’s information with other spokes,

and it is reasonable for a spoke to be aware that it is in the hub’s best interests to share

it. A spoke’s own conduct can serve to satisfy this requirement. Suppose the hub is better

o¤ if all spokes choose some common course of action (e.g., raising prices), and suppose a

spoke conveys to the hub that it is willing to adopt that action if and only if all other spokes

do so (e.g., it is willing to raise price if all other spokes raise their prices). It would then

be in the best interests of the hub to share the spoke’s intention with the other spokes and,

furthermore, the spoke would be cognizant of it because its message to B is implicitly a

request for it to do exactly that.

179

Fourth, it is in the sp oke’s best interests to convey this information to the hub only if the

information is shared with other spokes or, more to the point, it will in‡uence the conduct

of other spokes. An example is when A complains to B about C’s low prices. There is little

point in A sharing this information unless it hopes B would act on it by getting C to stop

charging low prices. Or suppose a retailer (spoke) expresses to the hub its intent to raise

price. If this information was not to be shared with other spokes (and thus was not to be

used to enact a coordinated rise in retail prices) then it might not be in the best interests

of the retailer to share this information for it could be the basis for the upstream …rm to

raise its wholesale price to the retailer (knowing that the retailer could maintain its current

margin given it is raising its retail price). If instead the pricing intention was shared with

178

"When A not only gives commercially sensitive information to B, but also receives commercially sensitive

information from B originating from C, i.e. there is a bidirectional ‡ow of information, A can reasonably

assume that information it provides to B will or is likely to be passed on to C." Odudu (2011), p. 234

179

"[P]roblematic vertical information exchanges often occur when suppliers are attempting to cajole retail-

ers into accepting the case for higher wholesale prices as a result of higher costs. A retailer may individually

accept the case for higher prices but might take the view that it will only move its prices if the "market"

moves too. By sharing information on individual retailers’willingness to increase prices, the supplier can

manipulate market sentiment and facilitate the wider acceptance of the case for a price increase." Prewitt

and Fails (2015), p. 69

other retailers then it could lead them to all raise their prices. Thus, by this argument, a

spoke would share its intention to raise price only if it expected the hub to share it with

other spokes.

If any of the four conditions are satis…ed, it should be su¢ cient to conclude that …rm A

invited …rm C to collude. That would be a violation of Section 5 of the FTC Act. For there

to be an agreement (in violation of Section 1 of the Sherman Act), U.S. law also requires

acceptance of A’s invitation by C. Evidence of such an acceptance could be market conduct

by C which is consistent with its acceptance. For example, A told B that A and C should

enact a common price increase (or A complained to B that C was pricing too low), B conveys

that information to C, and C subsequently raises its price. A second avenue to concluding

that A and C have an agreement is that the previously mentioned conditions to substantiate

an invitation to collude by A to C also apply for C to A. That is, …rm C also conveyed

information to …rm B which, had it been conveyed directly to …rm A, would have facilitated

a horizontal agreement, and …rm C could have reasonably foreseen that …rm B would convey

the information to …rm A. If that condition holds with regards to C’s communications with

B (as well as A’s communications with B) then we can conclude that …rms A and C have

the requisite "meeting of minds." When both …rms extend invitations to collude, there is an

agreement.

6 Concluding Remarks

Hub-and-spoke collusion is a common method that …rms have used to restrain competition

in retail markets. In most cases, the hub is the instigator of collusion. Communicating

indirectly through an upstream supplier or downstream buyer poses challenges in achieving

mutual understanding among …rms. Where one meeting among all …rms may be su¢ cient to

achieve a "meeting of minds" in a standard cartel, it can take many back-and-forth bilateral

communications between the hub and each sp oke for the spokes to reach an agreement.

The documentary evidence is rife with spokes expressing concerns that other spokes will

not comply with the collusive plan, and it is almost exclusively the hub’s responsibility to

alleviate those concerns. At the same time there are these communication challenges, the

involvement of an upstream supplier or a downstream buyer can facilitate collusion. Given

its interaction with the spokes as part of normal business practices, the hub has access to

information relevant to monitoring for compliance with the collusive scheme. It may also

have instruments for punishing noncompliant …rms, such as refusal to buy or sell, and for

rewarding compliant …rms, such as rebates. Turning to the bottom line, it appears that hub-

and-spoke cartels are less e¤ective than standard cartels as re‡ected in shorter duration.

There are a number of open questions. It is not always clear how upstream …rm(s)

bene…t by restraining competition in the downstream market. When the hub is a dominant

downstream retailer, do upstream suppliers bene…t when the collusive outcome excludes a

class of downstream retailers and, if so, how? When the hub is an upstream manufacturer,

does it bene…t when the collusive outcome is supracompetitive retail prices and, if so, how?

When do es a hub-and-spoke cartel form rather than a standard cartel among retailers?

Finally, why are hub-and-spoke cartels less e¤ective than standard cartels?

7 References

References

[1] Ambrus, Attila, Eduardo M. Azevedo, and Yuichiro Kamada, “Hierarchical Cheap

Talk,”Theoretical Economics, 8 (2013), 233-261.

[2] Asker, John, “A Study of the Internal Organization of a Bidding Cartel,” American

Economic Review, 100 (2010), 724-762.

[3] Bolecki, Antoni, “Polish Antitrust Experience with Hub-and-Spoke Conspiracies,”Year-

book of Antitrust and Regulatory Studies, 4 (2011), 25-46.

[4] Centorrino, Samuele, Elodie Djemai, Astrid Hopfensitz, Manfred Milinski, and Paul

Seabright, “Honest Signaling in Trust Interactions: Smiles Rated as Genuine Induce

Trust and Signal Higher Earning Opportunities,” Evolution and Human Behavior, 36

(2015), 8-15.

[5] Crawford, Vincent and Joel Sobel, “Strategic Information Transmission,”Econometrica,

50 (1982), 1431-1451.

[6] De los Santos, Babur and Matthijs R. Wildenbeest, “E-book Pricing and Vertical Re-

straints,”Quantitative Marketing and Economics, 15 (2017), 85-122.

[7] Foros, Øystein, Hans Jarle Kind, and Greg Sha¤er, “Apple’s Agency Model and the Role

of Most-Favored-Nation Clauses,”RAND Journal of Economics, 48 (2017), 673-703.

[8] Gaudin, Germain and Alexander White, “On the Antitrust Economics of the Electronic

Books Industry,”Tsinghua University, September 2014.

[9] Gilbert, Richard J., “E-books: A Tale of Digital Disruption,” Journal of Economic

Perspectives, 29 (2015), 165-184.

[10] Harrington, Joseph E. Jr., “How Do Cartels Operate?,” Foundations and Trends in

Microeconomics, 2 (2006), 1-108.

[11] Harrington, Joseph E., Jr. and Andrzej Skrzypacz, “Private Monitoring and Communi-

cation in Cartels: Explaining Recent Collusive Practices,”American Economic Review,

101 (2011), 2425-2449.

[12] Harrington, Joseph E., Jr. and Yanhao Wei, “What Can the Duration of Discovered

Cartels Tell Us About the Duration of All Cartels?,” Economic Journal, 127 (2017),

1977-2005.

[13] Johnson, Justin P., “The Agency Model and MFN Clauses,”Review of Economic Stud-

ies, 84 (2017), 1151-1185.

[14] Kaplow, Louis, Competition Policy and Price Fixing, Princeton: Princeton University

Press, 2013.

[15] Klein, Benjamin, “The Apple E-Books Case: When is a Vertical Contract a Hub in

a Hub-and-Spoke Conspiracy?,”Journal of Competition Law & Economics, 13 (2017),

423-474.

[16] Kühn, Kai-Uwe, “Fighting Collusion by Regulating Communication Between Firms,”

Economic Policy: A European Forum, 32 (April 2001), 167-197.

[17] Levenstein, Margaret C. and Valerie Y. Suslow, “What Determines Cartel Success?,”

Journal of Economic Literature, 44 (2006), 43-95.

[18] Levenstein, Margaret C. and Valerie Y. Suslow, “How Do Cartels Use Vertical Re-

straints? Re‡ections on Bork’s The Antitrust Paradox,”Journal of Law and Economics,

57 (2014), S33-S50.

[19] Manzini, Paola, Abdolkarim Sadrieh, and Nicolaas J. Vriend, “On Smiles, Winks and

Handshakes as Coordination Devices,”Economic Journal, 119 (2009), 826-854.

[20] Marshall, Robert C. and Leslie M. Marx, The Economics of Collusion - Cartels and

Bidding Rings, Cambridge, Mass.: The MIT Press, 2012.

[21] Mattioli, Eva, “Hub and Spoke: Towards a Belgian Precedent?,” Journal of European

Competition Law & Practice, 7 (2016), 261-266.

[22] Motta, Massimo, Competition Policy: Theory and Practice, Cambridge: Cambridge

University Press, 2004.

[23] Odudu, Okeoghene, “Indirect Information Exchange: The Constituent Elements of Hub

and Spoke Collusion,” Journal of European Competition Law and Practice, 7 (2011),

205-242.

[24] Orbach, Barak, “Hub-and-Spoke Conspiracies,”The Antitrust Source, April 2016, 1-15.

[25] Prewitt, Elizabeth and Greta Fails, “Indirect Information Exchanges to Hub-and-Spoke

Cartels: Enforcement and Litigation Trends in the United States and Europe,” Com-

petition Law & Policy Debate, 1 (2015), 63-72.

[26] Sahuguet, Nicholas and Alexis Walckiers, “The Economics of Hub-and-Spoke Collu-

sion,”HEC Montréal, February 2013.

[27] Sahuguet, Nicholas and Alexis Walckiers, “Hub-and-Spoke Conspiracies: The Vertical

Expression of a Horizontal Desire?,”Journal of European Competition Law and Practice,

5 (2014), 711-716.

[28] Sahuguet, Nicholas and Alexis Walckiers, “A Theory of Hub-and-Spoke Collusion,”

International Journal of Industrial Organization, 53 (2017), 353-370.

[29] Sparks, Adam, Tyler Burleigh, and Pat Barclay, “We Can See Inside: Accurate Pre-

diction of Prisoner’s Dilemma Decisions in Announced Games Following a Face-to-Face

Interaction,”Evolution and Human Behavior, 37 (2016), 210-216.

[30] Van Cayseele, Patrick and Simon Miegelsen, “Hub and Sp oke Collusion by Embargo,”

KU Leuven, Center for Economic Studies, Discussion Paper Series DPS 13.24, December

2013.

[31] Van Cayseele, Patrick J. G., “Hub-and-Spoke Collusion: Some Nagging Questions

Raised by Economists,”Journal of European Competition Law and Practice, 5 (2014),

164-168.

[32] Zampa, Gian Luca and Paolo Buccirossi, “Hub and Spoke Practices: Law and Eco-

nomics of the New Antitrust Frontier,” Competition Law International, 91 (2013), 91-

110.

Hasbro

Littlewoods

Argos

Figure 1b: Toys (UK)

Hasbro Mattel Fisher Price

Toys “R” Us

Figure 1a: Toys (U.S.)

Table 2: List of Noncompliance Episodes Reported to Toys “R” Us

Manufacturer

Description

COMMENTS

Hasbro

Puppy Surprise

Shipped early. No more will be shipped to warehouses.

Binney & Smith

[various]

Per Brent Blaine: Understood our concern. Going

forward they will offer special packs only for ‘93.

Commitments already made for ‘92.

Mattel

Barbie Dream House

Sold LY mdse. Will not sell again.

Huffy sports

Graphite Ultra Pak

Per Dave Allen, VP Sales: They admit their mistake.

Effective immediately only special Backboards will be

sold to clubs.

Playtime

(Div. of Tyco)

Super Saturator

Per Howard Abrams. SVP Sales: Pleaded ignorance.

He’s now aware and other than some prior

commitment, they will only sell club “special” items or

items we don’t carry.

Today’s Kids

Activity Rocker

Little Golfer

All Star Baseball

Per Jim Stephens: They needed the business but fully

understand our position. They will sell special items

going forward.

Tyco

123 Firehouse Blocks

Deluxe Set Magnadoodle

DB Nursery/Playground

Per Ken Shumaker: These are goods shipped last year –

prior to their new “no ship” policy on current goods we

carry.

Century

Elite Car Seat

Vendor will stop shipping BJ’s.

Fisher-Price

Nursery Monitor

They have agreed to stop selling this item to the clubs.

Safety 1st

Swivel Bath Seat

They have agreed to stop selling the clubs this item.

Playskool Baby

(a Hasbro Div.)

Nighttime Feeder

We have reached a corporate agreement on the sale of

this item to the club stores.

Kransco

Swim Seater

Will not be selling like items to them next year. Will

change graphics/packaging to differentiate item in

future

Morey Boogie

Sting Ray Board

Admitted they screwed up – will not happen again.

Will continue to sell them but in a “completely”

different packaging and graphics on the boards.

Nintendo

Asst.

“Not getting it from Nintendo” per Randy. They will

“look into.”

Sega

Asst.

Will continue to sell as long as Nintendo is in

Warehouse Clubs.

Source: In the Matter of Toys "R" Us, FTC Docket No. 9273, Initial Decision, James P. Timony,

Administrative Law Judge, September 25, 1997, pp. 21-22.

Figure 2: Impact of Collusion on ebook Prices

Source: United States v. Apple Inc., Opinion and Order, Southern District Court of New York,

Judge Denise Cote, July 10, 2013, pp. 95.

Table 2: Impact of Collusion on ebook Prices

Source: United States v. Apple Inc., Opinion and Order, Southern District Court of New York,

Judge Denise Cote, July 10, 2013, pp. 96.