Manufacturer collusion: Strategic implications of the channel structure

• DOI: http://doi.org/10.1111/jems.12209
• Author: Markus Reisinger,Tim Paul Thomes
• Date: 01 December 2017
• Published date: 01 December 2017

Received: 30 March 2016 Revised: 17 February 2017 Accepted: 2 March 2017

DOI: 10.1111/jems.12209

ORIGINAL ARTICLE

Manufacturer collusion: Strategic implications of the channel

structure

Markus Reisinger1Tim Paul Thomes2

1Department of Economics, Frankfurt School

of Finance & Management, Frankfurt am

Main, Germany (Email: m.reisinger@fs.de)

2Düsseldorf Institute for Competi-

tion Economics (DICE), Universität

Düsseldorf, Düsseldorf, Germany

(Email: thomes@dice.hhu.de)

Wethank a Co-Editor and two anonymous

refereesas well as Stéphane Caprice, Jaques

Crémer,Tommy Staahl Gabrielsen, Germain

Gaudin,Teis Lunde Lømo, Volker Nocke,

DanielO’Br ien, Patrick Rey,Olga Rozanova,

Marius Schwartz,Shiva Shekhar, Yaron

Yehezkel,participants at the 2015 MaCCI

conferencein Mannheim, the 2015 Bergen

CompetitionPolicy Conference, the 2015 EEA

annualcong ress in Mannheim, the 2016 IOSE

conferencein Saint Petersburg, and seminar

participants at the Universitiesof Toulouse

andHamburg for very helpful comments and

suggestions.

Abstract

We investigate how the structure of the distribution channel affects tacit collusion

between manufacturers. When selling through a common retailer, we find—in con-

trast to the conventional understanding of tacit collusion that firms act to maximize

industry profits—that colluding manufacturers strategically induce double marginal-

ization so that retail prices are above the monopoly level. This lowersindustr y profits

but increases the profit share that manufacturers appropriate from the retailer. Com-

paring common distribution with independent (exclusive) distribution, we show that

the latter facilitates collusion. Despite this result, common retailing leads to lower

welfare because a common retailer monopolizes the downstream market. Fort he case

of independent retailing, we also demonstrate that contract offers that are observable

to the rival retailer are not necessarily beneficial for collusive purposes.

1INTRODUCTION

Large manufacturers that sell their products through retailers are often long-term competitors. An example is the automobile

industry, where in many countries, few big producers control a large share of the market over a long time horizon. For example,

General Motors, Ford, Chrysler, and Toyotahave supplied around 60% of the U.S. market during the last 10 years with relatively

stable shares (Plunkett, 2012).1Almost without exception, car producers sell to final consumers via authorized nonintegrated

dealers.2

A similar pattern applies to the market for beauty products, cosmetics, and toiletries. This market is dominated by few estab-

lished players, such as Estée Lauder, L’Oreal, Procter & Gamble, and Unilever.3In contrast to the automobile industry, these

companies typically follow a strategy of running a strong wholesale business with retailers carrying the products of multiple

competing manufacturers.4

The existing literature has pointed out that the channel structure has important effects on manufacturers’ profits (see, e.g.,

O’Brien & Shaffer, 1997 or Cachon & Kök, 2010 on common (multibrand) retailing, and Bonanno & Vickers, 1988 or Rey &

Stiglitz, 1995 on independent (exclusive) retailing).5This literature takes a static perspective, that is, it examines the strategic

implications of different forms of the structure in a one-shot game. It therefore neglects the effects that the distribution chan-

nel may have on repeated interaction between manufacturers. Likewise, the literature on long-term cooperation and collusion

between manufacturers has ignored channel considerations. Most papers assume that firms sell directly to final consumers,

whereas papers explicitly considering the vertical relationship between manufacturers and retailers (e.g., Nocke& White, 2007,

or Jullien & Rey, 2007) focus on vertical integration or the effects of contracts within the channel. However, in most indus-

tries, (i) dynamic considerations play an important role, implying that manufacturers likely base their pricing decisions not only

J Econ Manage Strat. 2017;26:923–954. © 2017 WileyPeriodicals, Inc. 923wileyonlinelibrary.com/journal/jems

924 JOURNAL OF ECONOMICS & MANAGEMENTSTRATEGY

on current but also on past prices, and (ii) manufacturers sell their products via retailers to final consumers, implying that the

channel structure affects pricing decisions.

In fact, several empirical studies provide evidence consistent with the conjecture that pricing in many industries is based on

long-term competitive dynamics. Studying the automobile industry, Sudhir (2001a) demonstrates that prices in the compact

and midsize segment of the U.S. car market during the 1980s are indicative of manufacturer collusion and cannot be explained

by short-term competition.6Similar observations can be made in markets where the prevalent channel structure is common

retailing. For example, in the soft drink industry, Saltzman, Levy, and Hilke (1999) provide detailed evidence that bottlers

colluded on the prices charged to supermarkets and general merchandisers during the 1980s and the early 1990s. Likewise,

using data on food categories for two suburban retail stores, Sudhir (2001b) finds that manufacturer pricing is consistent with

cooperation.

Building on these considerations, the objective of this paper is to examine how the structure of the distribution channel affects

the strategic choices of manufacturers aiming to achieve cooperative outcomes. We seek to address the following questions on

the channel structure in a dynamic setting. How does the channel structure affect collusive behavior betweenmanufactures? Are

cooperation strategies under common retailingfundamentally different from those under independent retailing? How does con-

tract observability affect tacit collusion between competing manufacturers?Which channel structure leads to a higher welfare?

Are the findings robust to changes in the contractual form?

To answer these questions, we consider a simple infinitely repeated game with twosingle-product manufacturers contracting

either with a common retailer or with independent (exclusive) retailers. In the baseline model, the contractual form is a two-

part tariff.7We focus on tacit collusion between manufacturers and determine contracting decisions in the collusive agreement

and the critical discount factors above which collusion can be sustained. For expositional simplicity, we suppose that the retailer

(or the retailers) is short-lived. However,we demonstrate that all our results car ry overto t he case of long-livedretailers. Within

this framework, the channel structure has several distinct effects on manufacturers’ collusion.

Concerning the first research question, we show that when manufacturerssell t hrough a common retailer,tacit collusion works

in a fundamentally different way than in case they sell directlyto final consumers. In particular, colluding manufacturers propose

contracts with per unit wholesale prices above marginal cost, deliberately accepting double marginalization. Thus, the industry

profits under collusion are not maximized and even below those of the static game. However, in doing so, manufacturers obtain

a larger share of the channel profits.

The intuition underlying this finding lies in the common retailer’s opportunity to threaten a manufacturer with rejecting his

offer and only selling the rival’s product. This effect allows the retailer to pit manufacturers against each other and keep part of

the industry profits. We show that when tacitly colluding, manufacturers face a trade-off between maximizing channel profits

and mitigating the common retailer’s opportunity of rejecting one contract offer. Our analysis reveals that they achieve the

latter by raising the collusive wholesale price above the industry profit maximizing level. Setting the collusive wholesale price

sufficiently high lowerst he common retailer’sprofit from selling only one manufacturer’s product, thereby squeezing the share of

the industry profits that can be kept by the retailer. By contrast, when manufacturers compete against each other (as in the static

game), each one offers a contract that maximizes the bilateral profit of the manufacturer and the retailer, which avoids double

marginalization and maximizes channel profits. Thus, the manufacturers’ collusive strategy is to get a larger share of a smaller

“pie.”

Instead, when manufacturers sell through independent (exclusive) retailers, tacit collusion works in a similar way as in the

case in which manufacturers sell directly to final consumers. This holds independentlyof whether or not contracts are observable

to the rival retailer. An independent retailer obtains only an offer from the own manufacturer,t herebylacking the opportunity to

sell a different product. Therefore, colluding manufacturers set wholesale prices that maximize industry profits.

We then compare the channel structures with respect to their impact on the stability of manufacturer collusion. As explained

above, the common retailer can keep part of the industry profits. This leads to higher collusion profits for manufacturers under

independent retailing. However, for the same reason, manufacturers can also realize higher static profits and therefore higher

profits along the punishment phase with independent retailers.

We demonstrate that the result is nevertheless unambiguous in favor of independent retailing due to a novel effect: defection

from collusion implies that the deviant induces the common retailer to reject the offer of the nondeviating manufacturer because

the low wholesale price of the deviant makes the rival manufacturer’s contract unattractivefor the common retailer. This leads to

monopolization of the downstream market. Therefore, the ratio betweendeviation and collusion profits is higher under common

retailing than under independent retailing, implying that the incentive to deviate is larger in the formerregime. As a consequence,

manufacturer collusion can be sustained for a larger range of discount factorsif products are sold through independent retailers.

Despite this result, we show that welfare is larger with independent retailers. The reason is that a common retailer sets

monopoly prices in the downstream market, whatever the wholesale prices manufacturers have chosen. Even if the discount

REISINGER AND THOMES 925

factor is such that manufacturer collusion is possible under independent retailing but not under common retailing, the effect that

downstream competition is eliminated under the latter dominates.

We then turn to the question if contract observability facilitates collusion under independent retailing. Previous studies on

static models (e.g., Coughlan, 1985; Rey & Stiglitz, 1995) find that contract observability gives rise to a strategic effect: man-

ufacturers set higher wholesale prices, resulting in a dampening of the competitive pressure in the retail market and in higher

profits. Thus, in a static framework, contract observability is always profitable.

In a dynamic setting, we demonstrate that contract observability imposes two opposing effects on the sustainability of man-

ufacturer collusion. On the one hand, retailers can immediately react to a deviation of the rival manufacturer, which makes

collusion more stable. On the other hand, due to the strategic effect, the punishment following a deviation is less severe, leading

to a destabilization of collusive agreements. We show that the second effect dominates if products are close substitutes. The

intuition is that in this case, the retailer of the nondeviating manufacturer is constrained by the high wholesale price and the

possibility to react immediately loses importance. As a consequence, we obtain that observable contracts facilitate collusion if

competition is low or moderate, whereas private contracts are more suitable for collusive purposes if competition is fierce. This

result provides a strategic rationale for secret contracting, which is not encompassed by static models.

Finally, we show that the main insights derived with two-part tariffs carry over to the case with linear wholesale price

contracts.8In particular, tacit collusion under common retailing leads to increased wholesale prices that lower channel profits

but increase the share manufacturers can keep. Turning to independent retailing, the only difference is that contract observ-

ability unambiguously favors collusion. The intuition is that, due to the strategic effect, manufacturers need a lower wholesale

price under public contracts to realize identical collusion profits as under private contracts. This makes deviation always more

profitable under secrecy of contracts.

Our paper has also interesting implications for antitrust authorities. It demonstrates that a distribution channel of exclusive

retailers is more prone to collusion compared to a channel with a common retailer. For example, the “Guidelines on Vertical

Restraints” issued by the European Commission state that ”(when) most or all of the suppliers apply exclusive distribution this

may (...) facilitate collusion.”9Our paper provides a rationale for this statement by demonstrating a clear mechanism why this

effect occurs. However,this does not imply that independent retailing is anticompetitive relative to common retailing. Therefore,

it does not justify to consider the former channel structure more suspiciously.

The rest of the paper is organized as follows:Section 2 presents the relation to previous literature. Section 3 sets up the model.

Section 4 characterizes the equilibrium under common retailing. Section 5 provides the equilibrium analysis in case manufactur-

ers sell through independent retailers for both secret and public contracts. Section 6 studies sustainability of cooperation under

the different channel structures and provides a welfareanalysis. Section 7 analyzes linear wholesale price contracts, and Section 8

concludes.

2RELATION TO THE LITERATURE

This paper contributes to the literature on competition in manufacturer–retailer relationships in several aspects. First, it extends

the recent and growing literature on collusion in vertical settings. This literature started with Nocke and White (2007) who

analyze whether vertical integration can facilitate tacit collusion between manufacturers under two-part tariffs. They show that

this is indeed the case because the downstream affiliate of the integrated firm is no longer a potential buyer for a deviating

manufacturer.10 Normann (2009) considers linear upstream contracts and shows that a similar result obtains even with double

marginalization. Jullien and Rey (2007) demonstrate in a model with stochastic demand that resale price maintenance reduces

profits in a static framework, but helps to facilitate collusion because it simplifies the detection of deviations. Piccolo and

Reisinger (2011) analyze exclusive territories and find that they help to sustain collusion if wholesale contracts are observable

because this enables retailers to react to deviations. Piccolo and Miklòs-Thal (2012) analyze collusion at the retail level and

find that retailers can collude through manufacturers via above-cost wholesale prices and negative fixed fees. Finally, Gilo and

Yehezkel(2016) consider collusion between retailers and a common supplier. They show that the presence of a common supplier

helps retailers to sustain collusion (that is, vertical collusion is easier to maintain than horizontal collusion) and demonstrate

how collusion profits are distributed between the firms as the discount factor varies.11

To the best of our knowledge, our paper is the first one analyzing the effects of different channel structures on manufacturer

collusion in a dynamic framework and how changes in the contractual form affect the respective outcomes.12

Our paper also relates to the work analyzing strategic effects of the channel structure in static environments. One strand of this

work focuses on a frameworkwith multiple manufacturers and a common retailer. For example, Choi (1991) analyzes wholesale

price contracts and distinguishes between the scenarios in which either manufacturers or retailers are the Stackelberg price