GEORGE MASON UNIVERSITY SCHOOL OF LAW
SLOTTING CONTRACTS AND CONSUMER WELFARE
Joshua D. Wright
06-14
Forthcoming in Antitrust Law Journal
GEORGE MASON UNIVERSITY LAW AND ECONOMICS RESEARCH PAPER SERIES
This paper can be downloaded without charge from the Social Science Research Network at http://ssrn.com/abstract_id= 897394
Preliminary draft, do not cite without permission of the author. 08/25/06.
SLOTTING CONTRACTS AND CONSUMER WELFARE
Joshua D. Wright♦♦♦♦
Abstract
Slotting contracts involve manufacturer payments for retail shelf space. Slotting
is an increasingly important part of the competitive process in many product
markets, and has been the subject of congressional hearings, agency
investigations, antitrust litigation, and scholarly debate. However, very little is
known about the competitive consequences of slotting. This paper uses a unique
data set consisting of slotting contracts at military commissaries prior to an
exogenously imposed slotting ban to identify the impact of slotting on consumer
welfare. This natural experiment provides a rare opportunity to directly observe
the crucial policy counterfactual in the real world: would banning slotting
contracts increase consumer welfare? The analysis measures the impact of
slotting, at both the product and category levels, on prices, output, and product
variety. I find no evidence that slotting is anticompetitive. To the contrary,
slotting is associated with “brand-shifting” without category level effects.
Because slotting payments are passed on to consumers in competitive retail
markets, the results imply that competition for shelf space with slotting contracts
increase consumer welfare.
♦ Assistant Professor, George Mason University School of Law. I thank Bernard Black, Kenneth Elzinga, Bruce Johnsen, Benjamin Klein, Jonathan Klick, Latika Hartmann, Tom Hazlett, Keith Hylton, Kate Litvak, J.J. Prescott, Adam Rennhoff, Alison Sexton, Paul Stancil, and especially Wesley Hartmann for helpful comments and suggestions. Earlier versions of this paper were presented at the George Mason University Levy Workshop, the University of Texas Law School Center for Law, Business, and Economics, the International Industrial Organization Conference, and the Southeastern Association of Law Schools Conference. Brandy Wagstaff provided excellent research assistance. The Department of Defense Commissary Agency generously provided the dataset. Jay Manning and Robert Vitikacs of the Commissary Agency provided insight into commissary operations and the data. The George Mason Law and Economics Center provided financial support. All errors are my own.
1
1 Introduction
Slotting arrangements govern shelf space relationships between
manufacturers and retailers. These contract arrangements typically take the form
of a payment to the retailer in exchange for some form of promotional
consideration ranging from merely stocking the product to special displays or
preferred shelf location, such as an end-aisle display or “eye-level” shelf space.
Slotting arrangements have been the focus of government investigations,1
litigation,2 proposed legislation,3 and scholarly debate.4 Analogous payments
such as “payola” in the music industry have also resulted high profile
investigations, including Sony’s recent settlement prohibiting all future
payments for radio music spins resulting from New York Attorney General Eliot
1 The Federal Trade Commission has also expressed a considerable amount of interest in slotting allowances, holding a workshop culminating in the issuance of a report and a subsequent study on the competitive concerns of the practice. See FEDERAL TRADE COMM’N, REPORT ON THE FEDERAL TRADE
COMMISSION WORKSHOP ON SLOTTING ALLOWANCES AND OTHER MARKETING PRACTICES IN THE
GROCERY INDUSTRY (2001) [hereinafter FTC Report]; FEDERAL TRADE COMM’N STAFF STUDY, SLOTTING
ALLOWANCES IN THE RETAIL GROCERY INDUSTRY: SELECTED CASE STUDIES IN FIVE PRODUCT CATEGORIES (Nov. 2003) [hereinafter FTC Study] (examining the use of grocery slotting allowances in five product categories: fresh bread, hot dogs, ice cream, shelf-stable pasta, and shelf-stable salad dressing).
2 A variety of slotting allowance related issues have been raised in antitrust litigation. See Conwood Co. v. United States Tobacco Co., 290 F.3d 768 (6th Cir. 2002) (holding firm liable for combination of exclusionary conduct including product destruction, misleading retailers, and payments for exclusive product display space); R.J. Reynolds Tobacco Co. v. Philip Morris, Inc. (RJR II), 199 F. Supp. 2d 363 (M.D.N.C. 2002) (rejecting a claim that Philip Morris’s shelf space payment program violated the Sherman Act), aff’d per curiam, 67 F. App’x 810 (4th Cir. 2003); El Aquila Food Prods. v. Gruma Corp., 301 F. Supp. 2d 612 (S.D. Tex. 2003), aff’d, 131 F. App’x 450 (5th Cir. 2005); American Booksellers Ass’n v. Barnes & Noble, Inc., 135 F. Supp. 2d 1031 (N.D. Cal. 2001); FTC v. H.J. Heinz Co., 116 F. Supp. 2d 190 (D.D.C. 2000) (analyzing whether a reduction in slotting fees might be an efficiency in the context of horizontal merger analysis), rev’d, 246 F.3d 708 (D.C. Cir. 2001); Intimate Bookshop, Inc. v. Barnes & Noble, Inc., 88 F. Supp. 2d 133 (S.D.N.Y. 2000); FTC v. McCormick, FTC Docket No. C-3939 (2000) (settling charges that McCormick’s shelf space payments in the spice market violated the Robinson-Patman Act); Coca-Cola Co. v. Harmar Bottling Co., 111 S.W.3d 287 (Tex. App. 2003) (shelf space payments in violation of state antitrust laws). Slotting fees and other forms of shelf space payments were also central to Coca-Cola’s recent settlement with the European Commission that limits the amount of shelf space that Coca-Cola can purchase from retailers, as well as Coca-Cola’s ability to offer rebates conditioned on exclusivity or specified levels of sales. See Undertaking, Case Comp/39.116/B-2-Coca-Cola. 3 See Sen. B. 582 (Cal. 2005) (Figueroa), available at http://www.leginfo.ca.gov/pub/bill/sen/sb_0551-0600/sb_582_bill_20050411_amended_sen.pdf (as amended). 4 Section 2 summarizes this literature.
2
Spitzer’s efforts.5 Despite the substantial resources invested into studying,
litigating, and investigating slotting arrangements, we know surprisingly little
about the real world competitive consequences of the practice.6 The need for
empirical evidence is heightened because the antitrust doctrine governing these
practices is notoriously unsettled, inconsistent, and controversial.7 At least one
cause of the doctrinal uncertainty surrounding vertical distribution
arrangements more generally is the lack of empirical evidence capable of guiding
sensible antitrust policy.8 This paper seeks to fill that void by presenting
evidence that slotting contracts are pro-competitive in practice.
The paper proceeds as follows. Section 2 reviews the theoretical and
empirical literature on slotting arrangements, focusing on Klein and Wright’s
5 See, e.g., Press Release, Office of N.Y. Attorney Gen. Eliot Spitzer, Sony Settles Payola Investigation (July 25, 2005), available at http://www.oag.state.ny.us/press/2005/jul/jul25a_05.html (Sony BMG agrees to cease all payments for radio exposure, disclosed or otherwise, and to pay $10 million to the Rockefeller Philanthropy Advisors as a result of allegations that its payments “were designed to manipulate record charts, generate consumer interest in records and increase sales”). Similar payments occur in other retail environments, including record stores, drug stores, convenience stores, and book stores. See, e.g., Iris Rosenthal, Slotting Fees Continue to Spark Controversy in Retailing, 135 DRUG TOPICS 81 (January 21, 1991); James Surowiecki, Paying to Play, THE NEW YORKER, July 12, 2004; and Jeffrey A. Trachtenberg, Borders Sets Out to Make the Book Business Bussinesslike, WALL STREET JOURNAL, May 20, 2002, at B1. 6 Attempts to empirically document the competitive consequences of slotting contracts have not been fruitful. See, e.g., FTC Study, supra note 1, at 62 (“Our ability to comment about the relevance of the various theories of slotting based on our study is limited.”); U.S. GEN. ACCOUNTING OFFICE, SLOTTING
FEES: EFFORTS TO STUDY THESE PAYMENTS IN THE GROCERY INDUSTRY (2000) (testimony of Lawrence J. Dyckman before the Senate Committee on Small Business), available at http://www.gao.gov/archive/2000/ rc00295t.pdf) (“In short, despite repeated attempts over the last 8 months, we have been unsuccessful in gaining the cooperation needed from the industry to conduct this study.”). 7 See Joshua D. Wright, Antitrust Law and Competition for Distribution, 23 YALE J. ON REG. 169 (2006); William A. Kovacic, The Modern Evolution of U.S. Competition Policy Enforcement Norms, 71 ANTITRUST L.J. 377 (2003) (noting that federal vertical restraints and monopolization policy have exhibited the greatest disparities in enforcement activity over the last three decades). 8 Two recent surveys echo this concern and conclude that while further research is critical to sound policy in this area, the balance of evidence tips in favor of a presumption of legality. See James C. Cooper, et al., Vertical Antitrust Policy as a Problem of Inference, 23 INT’L J. INDUS. ORG. 639 (2005); Francine Lafontaine & Margaret Slade, Exclusive Contracts and Vertical Restraints: Empirical Evidence and Public
Policy, HANDBOOK OF ANTITRUST ECONOMICS (Paolo Buccirossi ed., Cambridge: MIT Press, forthcoming).
3
promotional services theory.9 Section 3 discusses the institutional foundations of
the empirical analysis: military commissaries, their product allocation decisions,
regulatory environment, the shelf space contracts, and the ban on those contracts
imposed at the end of 2001. Section 4 presents the data set. Section 5 presents
the empirical analysis and results and Section 6 concludes by examining the
implications of the evidence for antitrust policy.
2 The Economics of Slotting Contracts10
Slotting contracts involve manufacturer payments for shelf space and are
prevalent in markets with retail distribution. The contracts occasionally include
exclusive terms limiting the space available to a rival, but generally require the
retailer to commit a particular quantity or quality of shelf space to the supplier’s
product without any exclusive commitments.11 For example, Coca-Cola might
pay a retail chain $1000 for prominent display using “eye-level” shelf space for
six months. Alternatively, Coca-Cola might pay for the same shelf space by
allowing the retailer a discount from the wholesale price or other quantity-based
rebate, though the term “slotting fee” sometimes is reserved exclusively for the
use of upfront, lump-sum payments. The use of upfront, lump-sum payments
increased dramatically in the mid-1980s, both in terms of the number of products
9 The promotional services theory of slotting contracts is originally presented in Benjamin Klein & Joshua D. Wright, The Economics of Slotting Contracts, J.L. & ECON. (forthcoming 2007) [hereinafter Klein & Wright]. 10 This section draws substantially upon the discussion of competing slotting theories and evidence in Klein and Wright, supra note 9. 11 The FTC Study, supra note 1, at 57, reports that exclusivity was not prevalent in the five product categories studied (fresh bread, hot dogs, ice cream, shelf-stable pasta, and shelf-stable salad dressing). Benjamin Klein, Kevin M. Murphy & Joshua D. Wright, Exclusive Dealing and Category Management in Retail Distribution (unpublished working paper, 2006), provides a pro-competitive explanation for why retailers, effectively acting as bargaining agents for their consumers by internalizing each consumer’s independent buying decision, may offer exclusive shelf space in a product category as a way to lower prices.
4
covered and the magnitude of payment,12 though all forms of shelf space
contracts have attracted antitrust scrutiny.
While there have been a number of theoretical attempts to explain slotting
contracts as either efficiency-enhancing or anticompetitive, these explanations
have been fundamentally inconsistent with slotting data. As a result, very little
is known about the competitive consequences of shelf space contracts. This
Section summarizes the theoretical and empirical attempts to explain the use of
slotting contracts.
2.1 Theoretical Literature
Some economic models consider the possibility that manufacturers or
retailers strategically utilize slotting contracts to exclude rivals. The conventional
anticompetitive explanation contemplates manufacturers using shelf space
payments to disadvantage rivals by increasing their costs and increase barriers to
entry, ultimately increasing prices and reducing product variety.13 This
exclusionary theory of slotting has been the focus of the bulk of antitrust
litigation involving promotional payments,14 and in its “deep pockets” variant,
12 FTC Report, supra note 1, at 4, 11 & nn.18-19. 13 See FTC Report, supra note 1, at 34-41; FTC Study, supra note 1, at 3-4 (citing Greg Shaffer, Slotting
Allowances and Optimal Product Variety, ADVANCES IN ECONOMIC ANALYSIS & POLICY, Vol. 5, No. 1, Art. 3 (2005)). A second anticompetitive explanation suggests that slotting payments increase the cost of retail distribution and therefore favor incumbent firms who will be willing to pay more for the shelf space than competitors because a monopolist incumbent earning the monopoly rate of return will systematically outbid a potential entrant seeking to earn a competitive rate of return. See Steven C. Salop, Strategic Entry
Deterrence, 69 AM. ECON. REV. 335 (1979); FTC Study, supra note 1, at 3-4. These models depend on the presence of very large scale economies in manufacturing. An alternative possible concern is that a manufacturer may engage in predatory overpayment for shelf space. This is a condition that is difficult to identify and distinguish from the normal competitive process. See Benjamin Klein, Exclusive Dealing as
Competition for Distribution “On the Merits,” 12 GEO. MASON L. REV. 119 (2004).
14 See, e.g., Conwood Co. v. United States Tobacco Co., 290 F.3d 768 (6th Cir. 2002); Bayou Bottling, Inc. v. Dr. Pepper Co., 725 F. 2d 300, 303 (5th Cir. 1984); El Aquila Food Prods. v. Gruma Corp., 301 F. Supp. 2d 612, 621 (S.D. Tex. 2003), aff’d, 131 F. App’x 450 (5th Cir. 2005); RJR II, 199 F. Supp. 2d 363 (M.D.N.C. 2002), aff’d per curiam, 67 F. App’x 810 (4th Cir. 2003); Louisa Coca-Cola Bottling Co. v. Pepsi-Cola Metro. Bottling Co., 94 F. Supp. 2d 804, 816 (E.D. Ky. 1999); Beverage Mgmt., Inc. v. Coca-Cola Metro. Bottling Corp., 653 F. Supp. 1144, 1153-54 (S.D. Ohio 1986); FTC v. McCormick, FTC Docket No. C-3939 (2000).
5
depends on the assumption that capital markets are imperfect.15 Other highly
stylized theoretical models argue that slotting allowances are a mechanism by
which retailers with bargaining power are assumed to be able to use the
payments to exclude rivals by preventing them from selling a manufacturer’s
product.16
These anticompetitive theories of slotting are fundamentally inconsistent
with the evidence. The exclusionary theories of rival foreclosure involve at least
two necessary conditions for competitive harm not present for many slotting
contracts: they are typically not of sufficient duration to create risk of competitive
harm and frequently cover products that do not exhibit significant economies of
scale. The exclusionary theories also fail to explain the dramatic increase in the
prevalence of slotting beginning in the mid 1980s. The retailer bargaining power
explanations of slotting are equally inconsistent with the fact that small retailers
without any bargaining power use slotting and a large retailer, like Wal-Mart,
does not accept lump-sum, upfront fees.17 Perhaps more damning of the retailer
15 The Federal Trade Commission proffered a “portfolio effects” version of this argument, which was ultimately endorsed by the Supreme Court, to successfully challenge a conglomerate merger between Procter & Gamble and Clorox. The FTC argued that the post-merger firm would be able to utilize Procter & Gamble’s substantial advertising budget to more efficiently purchase preferred shelf space for Clorox products. Procter & Gamble Co., 63 F.T.C. 1465 (1963). Economists have recently revived these “deep-pockets” arguments, which depend on imperfect capital markets, in the context of slotting allowances. See Paul N. Bloom, et al., Slotting Allowances and Fees: Schools of Thought and the Views of Practicing
Managers, 64 J. MKTG. 92 (2000); Greg Shaffer, Slotting Allowances and Optimal Product Variety, ADVANCES IN ECONOMIC ANALYSIS & POLICY, Vol. 5, No. 1, Art. 3 (2005). 16 See, e.g., Leslie M. Marx & Greg Shaffer, Upfront Payments and Exclusion in Downstream Markets (unpublished paper, Aug. 2005); Patrick Rey, Jeanine Thal, & Thibaud Vergé, Slotting Allowances and Conditional Payments (unpublished paper, March 2005). Both papers present models where retailers with bargaining power demand slotting fees to increase their profits because slotting has the anticompetitive effect of excluding other retailers. These models are discussed in greater detail in Klein & Wright, supra note 9. 17 Klein & Wright, supra note 9, explain why Wal-Mart accepts promotional shelf space payments primarily in the form of lower wholesale prices rather than per unit time payments.
6
market power theories is the fact that the growth in slotting fees has not been
associated with an increase in retailer profitability.18
The literature includes three pro-competitive theories of slotting contracts,
each claiming that the growth in slotting can be explained by the increase in new
supermarket products. Each of these theories has limited ability to explain
slotting contracts on established products, and, as I will demonstrate, also do not
explain slotting for new products.
The first of these theories asserts that the increase in new products
proportionally increases the transaction costs imposed on retailers, such as
entering new product information into a computer, warehousing the new
products, and physically placing the new products on the shelf.19 This
explanation fails because slotting payments are significantly greater in
magnitude than these costs, and it incorrectly predicts that slotting payments
will be consistent in magnitude across markets and do not cover established
products where additional transaction costs are likely to be minimal.
The second pro-competitive theory asserts that slotting compensates
retailers for the risks associated with dedicating shelf space to a new and
unproven product.20 These signaling models generally test the proposition that
slotting payments assist retailers in identifying successful products because
manufacturers are assumed to have superior information. Again, these models
are inconsistent with the fact that slotting contracts are observed with established
products and products with stable demand. Further, the models do not consider
18 See Klein & Wright, supra note 9. 19 See, e.g., Laurie Freeman, Paying for Retail Shelf Space, ADVERTISING AGE, Feb. 13, 1986, at 31. 20 See, e.g., Wujin Chu, Demand Signaling and Screening in Channels of Distribution, 11 MKTG. SCI. 324 (1997); Preyas S. Desai, Multiple Messages to Retain Retailers: Signaling New Product Demand, 19 MKTG. SCI. 381 (2000); Martin A. Lariviere & V. Padmanabhan, Slotting Allowances and New Product
Introductions, 16 MKTG. SCI. 112 (1997); FTC Study, supra note 1, at 1-2; K. Sudhir & Vithala R. Rao, Do
Slotting Allowances Enhance Efficiency or Hinder Competition?, 43 J. MKTG. RESEARCH 137 (May 2006).
7
why manufacturers adopt this particular form of contract to solve this
informational problem, rather than an alternative arrangement such as an
introductory price allowance or product failure fees. The most significant failure
as a theoretical matter for the first two explanations is that they fail to address
the critical economic question about these increased costs: why do manufacturers
compensate retailers for the increased costs in the form of a fixed fee rather than
consumers paying for these costs in the form of higher prices?
Mary Sullivan presents a third pro-competitive theory of slotting related
to the increase in new products that does address this fundamental question.21
Specifically, Sullivan contends that manufacturer payments to retailers for the
growth of new products have increased because the resulting increase in
supermarket shelf space costs per dollar of sales does not benefit consumers.
Sullivan’s argument hinges on the unrealistic assumptions that consumers do not
value brand extensions that do not reduce consumer search costs and do not
demand product variety. Because consumers are assumed not to be willing to
compensate supermarkets through increased margins or greater sales when
supermarkets increase product variety, slotting fees are necessary, according to
Sullivan, to allow supermarkets to recover their higher costs of providing
increased shelf space for stocking new products.
Contrary to Sullivan’s model, however, consumers are generally willing to
pay for increased variety that raises supermarket operating costs. Brand
extensions can generate significant value for consumers.22 Inter-retailer
competition should result in supermarket compensation for this consumer
benefit in the form of an increased margin and/or increased sales. This means
21 Mary Sullivan, Slotting Allowances and the Market for New Products, 40 J.L. & ECON. 461 (1997). 22 See, e.g., Jerry Hausman, Valuation of New Goods Under Perfect and Imperfect Competition, in THE ECONOMICS OF NEW GOODS 209-67 (Bresnahan & Gordon eds., 1997).
8
consumers “pay” for the increased costs of increased shelf space per dollar of
sales in an increased supermarket margin, even if there is no decrease in search
costs. Therefore, the growth in the number of new products would not require a
separate slotting contract to compensate supermarkets for their higher selling
costs.
2.2 Empirical Literature
There is also very little evidence regarding the competitive effects of
slotting contracts. The lack of evidence is explained primarily by firms’ rational
disinclination to share information in the present antitrust climate, and also
because slotting fees involve payments that are difficult to disentangle from
other trade promotion spending. For these reasons, most empirical
investigations of slotting have relied on aggregate level data and surveys rather
than actual slotting contracts.23 While these studies have provided useful
information regarding slotting trends and their implications, there have been
very few detailed studies of the use of slotting contracts at the firm level. These
exceptional studies uniformly reject the notion that slotting is anticompetitive.
For example, Rennhoff estimates a structural model of manufacturer
competition for retailer shelf space with slotting allowances.24 A structural
model imposes assumptions about the behavior of economic agents
(manufacturers, retailers, and/or consumers) to derive a relationship between
endogenous and exogenous variables that may be observed by an empirical
researcher as opposed to “reduced-form” estimation, which takes what is an
23 See, e.g., Sullivan, supra note 21 (using aggregate level data on the supply and demand for new products); Klein & Wright, supra note 9 (showing that the growth in slotting contracts over the past twenty years is consistent with the aggregate increase in manufacturer margins and prevalence of branded products across product categories). 24 Adam Rennhoff, Paying for Shelf Space: An Investigation of Merchandising Allowances in the Grocery Industry (working paper, on file with author, July 2004).
9
essentially “theory-free” approach. While Rennhoff does not directly observe
actual slotting contracts, he observes the magnitude of the manufacturer’s
merchandising allowances. Rennhoff estimates consumer demand in a single
product category, ketchup, and demonstrates that products enjoying placement
in premium shelf space experience increased demand.25 By imposing
assumptions about the strategic interactions of manufacturers and retailers,
Rennhoff’s model predicts that average retail prices are lower with slotting
allowances than they would be under a slotting ban.
Sudhir and Rao observe some data on allowances offered for new
products to a single, large supermarket chain in the northeastern United States
over a six-month period from June 1986 to February 1987.26 While they do not
observe the magnitude of the slotting payment, they observe whether a slotting
allowance was paid, and rely on survey results from a questionnaire completed
by the retail buyer assessing various manufacturer and product attributes.
Sudhir and Rao conclude that slotting is, on balance, pro-competitive because it
shifts new product risk from retailers to manufacturers, but that it also softens
retail competition. However, as Klein and Wright demonstrate, the growth in
slotting has not been correlated with an increase in retailer profitability.27
The study most closely related to ours is Bronsteen, Elzinga, and Mills’
examination of the competitive impact of Philip Morris’s “Retail Leaders”
program, which involved shelf space contracts specifying payments in exchange
for the supply of varying levels of display space and store signage in the cigarette
industry. Bronsteen, Elzinga, and Mills’ case study examines, and rejects, the
25 This is consistent with findings in the marketing literature. See, e.g., Xavier Drezè, et al., Shelf
Management and Space Elasticity, 70 J. RETAILING 301 (1994); Charles Areni, et al., Point-of-Purchase
Displays, Product Organization, and Brand Purchase Likelihoods, 27 J. ACAD. MKTG. SCI. 428 (1999). 26 Sudhir &. Rao, supra note 20. 27 Klein & Wright, supra note 9.
10
plaintiffs’ theory in the antitrust dispute against Philip Morris, which asserted
that the Retail Leaders program would result in higher Philip Morris prices
relative to rival brands.28 One important characteristic of this important case
study is that it measures the impact of actual slotting contracts on retail prices.
The present study improves upon the case study approach by broadening the
analysis to include hundreds of product markets, and most importantly, by
exploiting the exogenously imposed “natural experiment” resulting from the
slotting ban to estimate the impact of slotting on retail prices, as well as category
level output and product variety.
While the existing empirical studies unequivocally contradict the
predictions of the anticompetitive theories, they leave significant room for
improving our understanding of the impact of slotting contracts on consumer
welfare across product markets.
2.3 The Promotional Services Theory of Slotting Contracts29
The promotional services theory of slotting contracts recognizes that the
supply of retail shelf space is a form of promotion that increases some
consumers’ reservation values for a product without generating significant inter-
retailer effects. In contrast to abstract economic models of retailing that assume
that retailers exist only to reduce search costs, the promotional services theory
begins with the empirical reality that retailers have some discretion over product
mix and shelf space allocation, and that these decisions influence consumer
purchasing behavior.
Retailer supply of premium shelf space creates “promotional sales” that
would not occur without the promotion. These promotional sales are
28 Peter Bronsteen, et al., Price-Competition and Slotting Allowances, 50 ANTITRUST BULL. 267 (2005); RJR II, 199 F. Supp. 2d 363 (M.D.N.C. 2002), aff’d per curiam, 67 F. App’x 810 (4th Cir. 2003). This case is discussed in Wright, supra note 7, and Klein & Wright, supra note 9. 29 This section summarizes the theory originally presented in Klein & Wright, supra note 9.
11
particularly profitable for manufacturers of products with relatively large
margins of wholesale price over marginal cost. But the fact that manufacturer
profits are increasing in the supply of promotional shelf space is not a sufficient
condition to generate slotting contracts. It merely sets the foundation for the
fundamental economic question regarding slotting contracts: why do
manufacturers and retailers find it necessary to enter into slotting contracts?
That is, why do manufacturers not simply sell products to retailers at wholesale
prices and allow competitive retailers to determine independently which
products to feature?
The promotional services theory addresses this question by identifying a
general incentive incompatibility between manufacturers and retailers with
respect to the supply of the desired quantity and quality of promotional shelf
space. There are two primary economic reasons for this incentive
incompatibility: (1) manufacturer margins over marginal cost are large relative to
retailer margins on the same product; and (2) increasing the supply of
promotional shelf space allocated to a particular brand does not generate
significant inter-retailer effects. These conditions are very realistically satisfied in
many settings involving differentiated products distributed through retailers to
consumers. Klein and Wright demonstrate that when the supply of premium
shelf space generates profitable incremental manufacturer sales without shifting
sales between retailers relative to a reduction in price, slotting contracts are a
consequence of the normal competitive process.
The intuition driving the theory is straightforward. Because the retailer’s
decision to allocate eye-level shelf space to Coke or Pepsi creates significant
profits for the favored manufacturer, but does little to shift inter-retailer sales or
increase retailer traffic, the manufacturer must compensate the retailer to ensure
the joint profit-maximizing allocation of shelf space.
12
Klein and Wright also address a second important economic question that
has received a good deal of attention from antitrust regulators: why do
manufacturers sometimes rely on per unit time payments rather than wholesale
price discounts or rebates? Klein and Wright show that a per unit time payment
offers a significant advantage over wholesale price discounts because the latter
create the incentive for retailers to lower the price of the manufacturer’s product,
undermining the manufacturer’s shelf space payments. The promotional
services theory, therefore, predicts that wholesale price discounts will be an
efficient method to compensate retailers for the provision of promotional shelf
space when there is not significant inter-retailer competition on the particular
product.
The promotional services theory suggests that slotting contracts increase
consumer welfare because payments are passed on to consumers in the highly
competitive retail environment.30 The key to understanding this prediction is the
fact that shelf space payments are increasing a supermarket’s traffic or sales.
One can expect competitive supermarkets to pass on slotting payments in the
form of lower prices and higher quality, thereby increasing store traffic and the
“price” the retailer can collect for its shelf space. In a competitive retail
environment, supermarkets that do not pass on rents earned from shelf space
payments in the form of lower prices or increased non-price amenities that
consumers value will sacrifice significant sales and also will collect lower shelf
space payments in the future. Because shelf space payments are ultimately to be
30 See, e.g., Joshua Wright, Vons Grocery and the Concentration-Price Relationship in Grocery Retail, 48 UCLA L. Rev. 743 (2001) (demonstrating that competition in the grocery retail market has remained vigorous despite substantial increases in concentration over the past two decades).
13
completely dissipated regardless of their particular form, shelf space payments
can be expected to increase consumer welfare.31
3 DeCA Commissaries
The Department of Defense operates grocery retail stores, called
commissaries, for active and retired military personnel and their families. The
Defense Commissary Agency (“DeCA”), formed in 1991, manages these
commissaries and is responsible for product assortment decisions, negotiations
with suppliers, pricing, and day-to-day operations. DeCA operates 275
commissaries, 171 within the 48 contiguous United States, and 104 located
elsewhere throughout the world. 32 Annually, DeCA commissaries generate
approximate $5 billion in revenue. While commissaries are somewhat similar to
traditional supermarket chains in terms of store content, management, and
competitive dynamics,33 there are some important and obvious differences
between commissaries and conventional private sector supermarkets. For
example, commissaries typically offer fewer items, compete less rigorously on
non-price dimensions, and are open for fewer hours. Table 2 summarizes some
key differences between private supermarkets and commissaries.
The most prominent difference between supermarkets and commissaries
is that the latter are not-for-profit operations. DeCA’s operating expenses are
31 Klein & Wright, supra note 9 (demonstrating that retailer profitability remained constant between 1980-2003, the time period during which shelf space payments dramatically increased in both frequency and magnitude).
32 A list of the geographic locations of the commissaries is available at http://www.commissaries.com/ realign_maps/alpha_list.cfm. 33 Commissary stores, according to 10 U.S.C. § 2484(a), are “similar to commercial grocery stores and may sell merchandise similar to that sold in commercial grocery stores.” DeCA’s 1995 total sales of $5.4 billion made it the tenth largest supermarket chain in the United States that year. Demographic Section, EXCHANGE AND COMMISSARY NEWS, Oct. 15, 1996, at 86. A 1995 study estimates that in Hampton Roads, Virginia, local commissaries accounted for approximately 8.4% of all grocery retail sales, making DeCA the third leading retail chain in the region. Stephanie Stoughton, Supermarkets Fight Commissary
Proposal, THE VIRGINIAN-PILOT, Sept. 8, 1995, at D1 (citing Food World study).
14
paid primarily through its annual appropriation of approximately $1 billion. The
DeCA mission is “to provide a significant non-pay compensation benefit to
military families.”34 Specifically, DeCA seeks to generate 30% in savings for
military families relative to conventional supermarkets.35
It is also important to note that commissaries are generally not isolated
from inter-retailer competition. A 1975 General Accounting Office examination
of 27 urban commissaries found that each one had at least four grocery stores
within a five mile radius.36 The significant growth in the number of
supermarkets over the past 30 years suggests that the number of grocery stores
in these areas has increased.37 For instance, a 1975 study found that each of 27
urban commissaries examined had at least four grocery stores within five miles.
Significantly, however, private manufacturers compete, and contract, for shelf
space at both commissaries and supermarkets.
Commissaries also face a unique regulatory environment. For instance,
commissaries do not sell private-label products, which have earned an increasing
share of conventional supermarket sales in recent years.38 The most important
34 U.S. GEN. ACCOUNTING OFFICE, DEFENSE COMMISSARIES, REPORT TO THE SENATE COMMITTEE ON
SMALL BUSINESS AND ENTREPRENEURSHIP 5 (Dec. 2002) [hereinafter Defense Commissaries]. 35 This estimate appears to be overstated. Because private retailers sell some items at higher margins than commissaries, which are limited to a 5% fixed margin, such savings would only be possible if commissaries purchased at significantly lower costs than large private supermarket chains. While the Robinson-Patman Act does not apply to commissaries, which would allow manufacturers the opportunity to offer different prices to commissaries, a Congressional Budget Office study suggests that manufacturers do not do so. CONGRESSIONAL BUDGET OFFICE, THE COSTS AND BENEFITS OF RETAIL ACTIVITIES AT
MILITARY BASES (Oct. 1997). 36 U.S. GEN. ACCOUNTING OFFICE, THE MILITARY COMMISSARY STORE (1975). 37 The number of supermarkets has increased about 20% during this time period, from 26,997 in 1972 to 34,252 in 2004. Annual Report of the Grocery Industry, PROGRESSIVE GROCER (various years). 38 Private label sales had captured up to 20.7% of supermarket unit sales and 16.9% of dollar sales according to the Private Label Manufacturers Association (citing Information Resources, Inc.). The absence of private label products does not appear to have compromised DeCA’s bargaining power vis-à-vis manufacturers as DeCA generally secures “most-favored nation” pricing provisions with all suppliers. Defense Commissaries, supra note 34.
15
regulatory constraint involves commissary pricing: commissaries sell food and
household items tax free at cost plus a 5% surcharge over the wholesale price.39
3.1 Commissary Shelf Space Decisions
Commissary shelves are not unlike those in private supermarkets. Both
offer a similar range of products,40 though commissaries are smaller than today’s
largest supermarkets by approximately 25% in terms of square footage, and carry
about 40% fewer Stock Keeping Units (“SKUs”) than the average supermarket.41
This is explained, in part, by the fact that commissaries do not sell beer, wine, or
general merchandise items. Commissaries typically offer a comparable number
of products per category. Commissaries also keep more limited hours than
supermarkets, averaging 48 hours per week in 1996 compared to 131 hours for
supermarkets.42
DeCA purchases typically involve a competitive bidding process, though
federal law exempts the purchase of brand name products from such bidding.43
39 10 U.S.C. § 2484(d) reads in pertinent part:
(d) Uniform sales price surcharge.--The Secretary of Defense shall apply a uniform surcharge equal to five percent on the sales prices established under subsection (e) for each item of merchandise sold in, at, or by commissary stores.
As discussed infra, the fixed retail margin allows direct observation of the impact of the slotting ban because manufacturers cannot substitute towards alternative methods of paying for shelf space, such as reducing the wholesale price. 40 Federal law enumerates the categories of products that commissaries may sell, including: health and beauty aids, meat and poultry, fish and seafood, produce, food and non-food grocery items, bakery goods, dairy products, tobacco products, delicatessen items, frozen foods, and magazines and other periodicals. 10 U.S.C. § 2484(b). Commissaries may sell non-enumerated products as the “Secretary of Defense may authorize” upon notification of Congress, but are not authorized to carry beer & wine, greeting cards, and most general merchandise items. Id. at § 2484(c). 41 Defense Commissaries, supra note 34. 42 Annual Report of the Grocery Industry, PROGRESSIVE GROCER (April 1996). 43 See 10 U.S.C. § 2304(c)(5). This is to be compared to bidding procedures that are imposed for non-brand products set forth in Section 2304 mandating a sealed bid auction where feasible. DeCA may also
16
The “brand name exception” is triggered only where the commercial item “is
regularly sold outside of commissary stores under the same brand name as the
name by which the commercial item will be sold in, at, or by commissary
stores.”44
DeCA’s purchasing practices for brand name products are similar to the
negotiations of a large chain of private supermarkets. These practices include,
for example, entry into category management relationships with suppliers.45
DeCA adopted category management in the mid 1990s, and accordingly, engages
in intensive data-based analysis of nearly all of the 170 product categories.46
DeCA assigns category managers and buyers to each category who meet with
suppliers during a category review. Suppliers present DeCA with sales data,
product selection recommendations, and display recommendations.47 DeCA
escape the competitive procedures for product purchase under a number of other scenarios, including “unusual and compelling urgency,” or national emergency. Id. at § 2304(c)(2). 44 10 U.S.C. § 2484(f). The term “regularly sold outside of commissary stores” is to be interpreted by evaluating “sales of the item on a regional or national basis by commercial grocery or other retail operations consisting of multiple stores.” Id. It is worth noting that DeCA sales do not include new product introductions, therefore negating the possibility that these contracts are caused by new product risk. 45 Category management typically involves designating a category captain in a product category and conferring upon the captain some power to provide input and otherwise participate in the retailer’s shelf space allocation decisions. Klein, Murphy, and Wright analyze this arrangement, showing that category management is a mechanism for controlling dealer free-riding incentives while allowing for greater product variety than exclusive dealing. Klein, Murphy & Wright, supra note 11. 46 DeCA utilizes data from Information Resources, Inc. (“IRI”), a data source commonly employed by private supermarkets. 47 DeCA notes that the suppliers making such presentations include both “leading” manufacturers such as Kraft, General Mills, and Pepsi, as well as other companies and distributors. DeCA describes its category management review process in eight steps, as follows: (1) DeCA announces a category review and invites companies supplying products to provide information and analysis; (2) DeCA conducts meetings with interested manufacturers, brokers, and distributors at DeCA’s headquarters in Fort Lee, Virginia; (3) DeCA’s category manager obtains IRI and scanner data on product performance; (4) DeCA analyzes IRI data as well as data presented by participating suppliers and prepares a schedule of category decisions indicating assortment decisions; (5) DeCA officials develop and sign-off on shelving plans called “plan-o-grams” illustrating how products will be arranged on the shelves. These plans are carried out with the assistance of the participating suppliers and distributors; (6) DeCA releases category review decisions and solicits comments from participating companies; (7) DeCA responds in writing to each company that
17
states that it “relies on market data” that it obtains from independent sources
and that “large suppliers do not select the products that commissaries sell,” but
notes that suppliers such as Kraft Foods, Inc., and General Mills, Inc., and other
companies in the category often have input on display considerations.48
One obvious limitation of the study is that it involves non-profit
commissaries rather than private supermarkets. This fact raises some questions
regarding the external validity of the results. However, this concern is somewhat
mitigated by three observations. First, the manufacturers offering slotting
contracts to both private supermarkets and commissaries are both profit
maximizing enterprises. Second, the unique regulatory framework imposed on
DeCA, such as fixed retail margins, actually enhances the ability to isolate
changes in manufacturer behavior by holding retailer behavior constant. In
other words, because commissary retailers cannot re-optimize by shifting sales to
higher margin products after the slotting ban, since all products are sold at the
fixed 5% margin, the estimates herein measure the “true” effect of the ban.
Third, as discussed, one might expect the consumer benefits from shelf space
payments to be even larger in the private supermarket context, since the
payments are a significantly greater percentage of total revenue in private
supermarkets than in commissaries. This suggests that these estimates might
underestimate the pro-competitive effects of slotting in private supermarkets.
3.2 DeCA’s Slotting Contract Ban
DeCA’s report to Senator Christopher Bond, the Ranking Member of
Senate Committee on Small Business and Entrepreneurship, states that:
objects to DeCA’s decisions; and (8) DeCA removes all deleted items from the stock system after 90 days. Defense Commissaries, supra note 34, at app. III. 48 Id.
18
DeCA does not require its suppliers to pay for shelf space in
commissaries, the agency accepts and stocks products for sale without
charging slotting fees. However, in recent years the agency participated
in a small number of promotional arrangements, called “performance
based agreements,” under which suppliers paid a negotiated fee for
preferred product display space . . .. The revenue from these performance
based agreements . . . has been used to fund commissary construction.
The agency discontinued offering its suppliers performance-based
agreements for 2002 to assure that revenue obtained through these
agreements was not limiting vendors’ ability to reduce product prices and
because the Congress had provided for funding commissary
construction.49
As one can verify with a quick review of the sample contract attached as
Appendix A, the there are no meaningful differences between “performance
based agreements” (“PBAs”) and the slotting contracts observed in private
supermarkets.50
DeCA officials decided not to enter into any PBAs for fiscal year 2002 after
using shelf space contracts in both 2000 and 2001.51 This decision was made after
a number of Senate Small Business Committee hearings, which included a
request that DeCA self-report on its own shelf space allocation practices, and in
particular, its use of slotting. DeCA publicly stated that the shelf space contract
program was terminated because Congress granted separate funding for
commissary construction, obviating the need for additional revenues from the
49 Defense Commissaries, supra note 34, at 2.
50 Senator Bond nonetheless issued a press release announcing that he was “heartened by findings that the Defense Commissary Agency does not accept so-called slotting fees or cash payments from big manufacturers to control prime shelf space and limit the display of competing products produced by small manufacturers.” Press Release, Senate Comm. on Small Bus. 51 DeCA justified the decision to no longer accept trade spending dollars because the revenue needed for commissary construction had been secured through the National Defense Authorization Act of 2002. Defense Commissaries, supra note 34, at 8.
19
contracts. A reasonable interpretation of the sequence of events might be that
DeCA ceased practices as a result of the Senate and Department of Defense
inquiries. Regardless, the ban is exogenous to the bargaining process between
manufacturers and DeCA, which allows empirical examination of the
relationship between shelf space contracts and consumer welfare.
4 The Data
The primary source of data are the 32 actual shelf space contracts in
operation at DeCA commissaries from 2000-01, the two years prior to the slotting
ban. While there are 32 contracts, each individual contract may govern several
different SKUs offered by a given manufacturer across several product
categories. For instance, Procter & Gamble might sell shampoo, dog food, and
soap, and sell several SKUs within each product category. However, the shelf
space obligations governing the promotion of each SKU would appear in a single
contract.
These contracts exhibit significant variation across several dimensions,
including form of payment (fixed, variable, or both), magnitude of payment,
duration, and the performance sought from DeCA. The performance sought
might include supplying an end-cap, a specified amount of shelf space, a display
rack, signage, or other forms of promotional effort. Appendix A contains a
sample slotting contract from the data set. Table 1 summarizes the product
categories where slotting contracts were observed in commissaries, as well as the
total value of the payments accepted by the commissaries in each year.
The data also include a set of “plan-o-grams,” shelf space schematics
which detail the precise shelf layout of SKUs in a product category at DeCA
commissaries during the relevant time period. These plan-o-grams indicate the
number of square inches of shelf space allocated to each product, exactly where
the product is allocated on the shelf, and next to what products. Appendix B
20
contains a sample plan-o-gram setting forth the shelf space allocation for juices in
commissaries in 2005.
I also observe the prices and quantities of each SKU sold by DeCA
commissaries in the United States in product categories where a slotting contract
exists. The dataset covers the sales of 11,688 SKUs in 73 product categories from
2000-03, generating approximately $2.3 billion in revenues. Commissary prices
are set nationally and negotiated annually. Therefore, an observation is a price
and quantity combination during a given year. There are also several unique
institutional details of the commissary system, discussed in Section 4, which
inform the empirics.
5 Do Slotting Contracts Harm Consumers?
Despite the substantial resources that have been invested into
Congressional hearings, a Federal Trade Commission Report and Study, federal
prosecutions in both the United States and the European Union resulting in
settlements, and a growing body of antitrust cases involving slotting contracts, it
remains the case that very little is known about their competitive consequences.
A complete analysis of the consumer welfare benefits associated with slotting
contracts would include the following:
(1) Shelf space payments may reduce the price of the slotted good
directly;
(2) Shelf space contracts may stimulate demand for the slotted
product, creating “promotional sales” and increasing output;
(3) Retailers may use shelf space payments to engage in non-price
competition, i.e., advertising, provision of free parking, wider
aisles, a deli, more cash registers, more employees, etc., in order to
increase store traffic;
21
(4) Retailers may use shelf space payments to subsidize price
competition on “staple” products likely to increase store traffic, i.e.,
milk, bananas, diapers, and other staple items.
The data allow measurement of (1) and (2), but not (3) and (4), which the
promotional services theory suggests are quite important since they increase
store traffic, increase shelf space value, and lead to greater future slotting
payments. On the other hand, the analysis should capture any net
anticompetitive effects of slotting on price, output, or product variety. This
suggests that the estimated effects should be interpreted as an upper bound on
anticompetitive effects.
I take a two-step approach to the consumer welfare analysis. In Section
5.1, I consider the product-level effects to determine whether the slotting ban
affected shelf space composition. Specifically, I examine whether the slotting ban
shifted the unit sales and prices of slotted products relative to non-slotted
products. One can think of this step of the analysis as testing for whether the
slotting ban resulted in compositional changes in shelf space allocation from
slotted to previously non-slotted products. This test alone gives us information
on whether slotting is having an impact on shelf space allocations, but alone is
not sufficient to draw conclusions about consumer welfare. The second step of
the analysis addresses the question of whether these compositional effects
resulted in net injury to consumers. Since the pre-ban shelf space allocation
allows for payments to retailers that are ultimately passed on to consumers,
however, the post-ban allocation can only improve welfare if category output
increases, category level prices fall, or product variety increases. Section 5.2
considers the relationship between slotting and category output, prices, and
product variety.
5.1 Differences-in-Differences Estimation of Product Level Effects
22
The basic strategy for identifying the competitive effects of slotting
contracts is to exploit the natural experiment provided by the slotting ban at the
end of 2001. I compare the effects of the ban on a “control group” consisting of
the non-slotted products within a particular product category to the effects on
the “treatment group,” which consists of the slotted products in the same
product categories. This is the “differences-in-differences” or “double
differences” estimate made possible by the natural experiment.
The intuition behind the strategy is simple: compare the change in prices
on slotted products while controlling for changes that occur in the control group
of non-slotted products in the same stores. Technically, one need not estimate a
regression equation. One could calculate the difference in mean prices for
“slotted” and “non-slotted” products before and after the ban and calculate the
difference between those averages. The regression analog to this procedure is to
estimate the following two equations:
(1)
(2)
Pit (Qit ) is the price (quantity) of a specific SKU, where i indexes whether the
product was slotted or non-slotted, and t indexes whether the observation is
before or after the policy change. Si is a dummy variable that takes on the value
of 1 if the SKU is slotted and zero if it is non-slotted. Therefore, βS measures the
treatment group specific effect to account for average differences between slotted
ln ( ) ( ) ( )
ln ( ) ( ) ( )
it DD i t s i A t it
it DD i t s i A t it
p S A S A
q S A S A
α β β β ε
α β β β ε
= + × + + +
= + × + + +
23
and non-slotted products. At is a dummy variable that takes the value of 1 if the
observation is after the slotting ban and zero if the observation is before the ban.
ΒA therefore measures the time trend common to both groups. α is a constant
which allows for a product-specific fixed effect. Finally, the double differences
estimator is βDD , which measures the true effect of the slotting ban. This
estimate answers the crucial counterfactual: what would happen to the prices
and unit sales of slotted products relative to non-slotted products in a world
without slotting contracts?
5.1.1 Quantity Effects
These regressions capture the possibility that slotting contracts generate
“promotional sales,” thereby increasing unit sales that would not otherwise
occur. If slotting contracts induce promotional sales, one would expect the ban
to reduce the output of the slotted products relative to non-slotted products
which are unaffected by the ban. Significant effects, therefore, represent a shift in
the composition of shelf space before and after the ban towards previously non-
slotted products. Fifteen product categories generate statistically significant
quantity effects at the 10% level, 9 at the 5% level, and 6 at the 1% level.
Consistent with the notion that slotting contracts generate promotional sales, the
output of slotted products relative to non-slotted products decreased in 13 out of
15 of these categories. That is, sales shifted towards the previously non-slotted
products after the ban. Of these categories, on average, relative unit sales of
slotted products declined by 13.36% relative to non-slotted products. Table 3
summarizes the results for the 19 categories with the strongest quantity effects.
5.1.2 Price Effects
These regressions measure the impact of the slotting ban on the price of
the slotted product relative to non-slotted products. It is important to note that
24
these regressions do not measure category level price effects. Generally, the
slotting contract ban had little effect on prices. Price effects are significant (at the
10% level) in just 28 of 73 product categories, with prices increasing after the
slotting ban in 17 of these categories.52 On average, the prices of previously
slotted products increased after the ban by 3.7% relative to non-slotted products.
Table 4 presents the price results for the same 19 product categories appearing in
Table 3 where slotting appears to have had the greatest impact.
There are, however, two important limitations on the product level
estimates. It is possible that commissaries would re-optimize shelf space
allocations after the slotting ban. In other words, retailer behavior might change
in important ways in the absence of slotting contracts. This is important because
such re-optimization would result in changes in the manner in which products in
both the treatment and control groups were displayed after the ban. Because
products that were not slotted prior to the ban might be affected by this post-ban
change in retailer behavior, I am only able to estimate relative effects, i.e., how
the sales of previously slotted products changed relative to non-slotted products.
These qualifications aside, the results provide evidence that slotting contracts
have the intended effect, increasing sales of the slotted product. This result
confirms our intuition that shelf space allocations change when slotting contracts
are present.
A separate economic question is whether this change in shelf space
allocations and sales in a world with slotting contracts generates lower consumer
welfare than the equilibrium without slotting. Because slotting generates
payments that are passed on to consumers, the post-ban allocation can only
improve consumer welfare if it is associated with category level benefits such as 52 Price effects are significant at the 5% level in 21 product categories and significant at the 1% level in 13 categories.
25
lower prices, higher output, or higher product variety to offset this loss of
payments. Section 5.2 examines these category level effects.
5.2 Category Level Effects
To measure the intensity of slotting within a product category, I construct
the variable, “SLOTSHARE,” which is the total category sales attributable to
slotted products before the ban. Because there is substantial variance in slotting
intensity across product categories prior to the ban, it is possible to examine
whether increased slotting affects consumer welfare. Specifically, I estimate the
following equation:
(3)
The crucial policy variable of interest is βDD, which can be interpreted as the
effect of the slotting ban on category output weighted by the intensity of slotting
within the product category. I run similar regressions using category prices and
product variety as the dependent variables and find no evidence to support the
theory that slotting is anticompetitive, or that the ban improved consumer
outcomes by any measure. The category level regressions do not completely
measure the welfare effects of slotting because they cannot fully capture the pass-
through of payments to consumers, which might occur on any number of
margins. The category level regressions should therefore be interpreted as a
lower bound on consumer welfare benefits. Because I find no evidence of
anticompetitive effect at the category level, the results suggest that slotting
contracts increases consumer welfare when one accounts for pass-through.
5.2.1 Category Output
Table 5 summarizes the results from the category output regressions.
There is no evidence that product categories with greater intensity of slotting
ln ( ) ( ) ( )it DD t s A t it
q SLOTSHARE A SLOTSHARE Aα β β β ε= + × + + +
26
experienced any increase or decrease in output relative to those where slotting is
less prevalent. Interestingly, there is a significant negative correlation between
the intensity of slotting and sales. In other words, slotting appears to be more
intense in categories with less sales volume.53
5.2.2 Category Prices
Table 6 summarizes the results from the category level price regressions.
There is no evidence of differential price effects in product categories with
greater slotting shares, or that the slotting ban produced lower prices. In sum,
there is no evidence that slotting contracts impact category level prices.
5.2.3 Category Variety
The FTC Study raises the concern that slotting contracts might reduce
product variety.54 I measure whether product categories with greater slotting
shares experienced different changes in product variety, as measured by the total
number of SKUs in the category, after the slotting ban. Table 7 summarizes the
results, which find no evidence that slotting contracts reduce product variety.
6 Conclusions, Antitrust Implications, and Future Research
The analysis of the slotting contracts governing shelf space relationships
at military commissaries in fiscal years 2000 and 2001 increases our
understanding of slotting contracts and their consequences.55 DeCA’s slotting
contract ban provides a natural experiment which allows systematic analysis of
the effects of slotting on prices, quantities, and product variety at both the
product and category levels. The results suggest that slotting contracts primarily
53 This may be explained by a correlation between lower volume sales categories and high-margin product categories where the promotional services theory predicts slotting will be most profitable. 54 See FTC Study, supra note 1, at 3-4 (citing Shaffer, supra note 15).
55 Bronsteen et al., supra note 28, analyze the competitive effects of actual shelf space contracts associated with Philip Morris’s Retail Leaders’ program.
27
involve “brand-shifting” without any significant impact on category level prices,
output, or product variety. This result is consistent with the promotional
services theory, which suggests that manufacturers with large margins seeking
incremental “impulse sales” must compensate retailers for providing premium
shelf space because consumers are not willing to pay for the full cost of the
promotion.56 The results call into question the proposition, frequently asserted in
antitrust litigation and the economics literature, that slotting contracts are
anticompetitive.
These results should not be interpreted as evidence that slotting has
ambiguous consumer welfare effects despite the fact that the DeCA slotting ban
was not associated with significantly lower category output or higher category
prices. This study systematically underestimates the consumer welfare benefits
of slotting because it does not capture the benefits of slotting revenue passed-
through to consumers. For example, slotting revenues might result in generally
lower retail prices, increased amenities, or other measures to increase store
traffic. While these payments amount to less than 1% of commissary revenue,
these pass-through effects are very significant in private supermarkets, where
slotting revenues are a considerably larger fraction of total revenue.57 The results
therefore suggest that antitrust law should recognize that slotting and other
forms of competition for distribution are “competition on the merits,” and are
unlikely to involve manufacturer exclusion or retailer attempts to extract
monopoly rents.
There are other important economic questions regarding the use of
slotting contracts. For example, this data allows one to measure how slotting
56 This does not imply social inefficiency. See Gary Becker & Kevin M. Murphy, A Simple Theory of
Advertising as a Good or Bad, 108 Q.J. ECON. 941 (1993). 57 Slotting is estimated to contribute to up to 50-75% of net supermarket profits. John Stanton, Rethinking Retailers’ Fees, FOOD PROCESSING, Vol. 60, No. 8, at 32 (1999).
28
contracts influence shelf placement, the impact of slotting on private label
brands, and on smaller firms. The data also allow for a systematic analysis of
firm and industry level determinants of shelf space contracts. Future research
will address these questions.
29
References
Areni, Charles, Duhan, Dale & Kiecker, Pamela, “Point-of Purchase Displays,
Product Organization, and Brand Purchase Likelihoods,” 27(4) Journal of
the Academy of Marketing Science 428-41 (1999).
Becker, Gary and Murphy, Kevin M., “A Simple Theory of Advertising as a Good
or Bad,” 108 Quarterly Journal of Economics 941 (1993).
Bloom, Paul, Greg Gundlach and Joseph Cannon [2000]. “Slotting Allowances
and Fees: Schools of Thought and the Views of Practicing Managers,” 64
Journal of Marketing, 92-108.
Bronsteen, Peter, Elzinga, Kenneth G., & Mills, David, “Price-Competition and
Slotting Allowances,” 50 Antitrust Bulletin 267 (2005).
Chu, Wujin [1992]. “Demand Signaling and Screening in Channels of
Distribution,” 11 Marketing Science (Fall), 327-247.
Cooper, James C.; Froeb, Luke M.; O’Brien, Dan; and Vita, Michael G.
“Vertical Antitrust Policy as a Problem of Inference.” 23 International
Journal of Industrial Organization 639 (2005).
Congressional Budget Office, “The Costs and Benefits of Retail Activities at
Military Bases.” (October 1997).
Desai, Preyas S. [2000]. “Multiple Messages to Retain Retailers: Signaling New
Product Demand,” 19 Marketing Science 381-89.
Dreze, Xavier, Hoch, Stephen P., & Purk, Mary E., “Shelf Management and Space
Elasticity,” 70 Journal of Retailing 301 (1994).
Federal Trade Commission Staff Report [2001], “Report on the Commission
Workshop on Slotting Allowances and Other Marketing Practices in the
Grocery Industry.”
Federal Trade Commission Staff Study [2003], “Slotting Allowances in the Retail
Grocery Industry: Selected Case Studies in Five Product Categories.”
30
Hausman, Jerry A. [1997]. “Valuation of New Goods under Perfect and
Imperfect Competition in The Economics of New Goods” (Bresnahan and
Gordon, eds.), University of Chicago Press, pp. 209-37.
Klein, Benjamin and Wright, Joshua D., “The Economics of Slotting
Contracts,” forthcoming Journal of Law and Economics (2007).
Klein, Benjamin, “Exclusive Dealing as Competition for Distribution ‘On the
Merits’,” 12 George Mason Law Review 119 (2003).
Klein, Benjamin, Murphy, Kevin M., and Wright, Joshua D., “Exclusive Dealing
and Category Management in Retail Distribution,” unpublished working
paper (2006).
Kovacic, William A., “The Modern Evolution of U.S. Competition Policy
Enforcement Norms,” 71 Antitrust L.J. 377 (2003).
Lafontaine, Francine, & Slade, Margaret, “Exclusive Contracts and Vertical
Restraints: Empirical Evidence and Public Policy.” Forthcoming in
Handbook of Antitrust Economics, Paolo Buccirossi (ed.), Cambridge:
MIT Press.
Lariviere, Martin A. & Padmanabhan [1997]. “Slotting Allowances and New
Product Introductions” 16(2) Marketing Science 112-28.
Marx, Leslie M. & Greg Shaffer [August 2005]. “Upfront Payments and
Exclusion in Downstream Markets”, unpublished paper.
Rennhoff, Adam, “Paying for Shelf Space: An Investigation of Merchandising
Allowances in the Grocery Industry (July 2004), available at
http://www.pages.drexel.edu/~adr24/rennhoffshelf.pdf.
Rey, Patrick, Jeanine Thal & Thibaud Vergé [2005]. “Slotting Allowances and
Conditional Payments,” unpublished paper, March 2005.
Salop, Steven C., “Strategic Entry Deterrence,” 69 American Economic Review
335 (1979).
31
Shaffer, Greg [2005]. “Slotting Allowances and Optimal Product Variety,” 5(1)
Advances in Economic Analysis & Policy: Article 3.
Stanton, John, “Rethinking Retailers’ Fees,” 60(8) Food Processing 32 (1999).
Sudhir, K. & Rao, Vithala, “Do Slotting Allowances Enhance Efficiency or Hinder
Competition?”, 43(2) Journal of Marketing Research 137 (2006).
Sullivan, Mary W., “Slotting Allowances and the Market for New Products,” 40
Journal of Law & Economics 461 (1997).
Wright, Joshua, [2001]. “Vons Grocery and the Concentration-Price Relationship
in Grocery Retail,” 48 UCLA L. Rev. 743.
Wright, Joshua D., “Antitrust Law and Competition for Distribution,”
23 Yale Journal on Regulation 169 (2006).
32
TABLE 1. SUMMARY OF SLOTTING CONTRACTS
Year No. of Agreements Product Categories Estimated Revenue
2000 18 Candy, pet supplies, breakfast foods, butter, margarine, baked goods, cheese, drinks, snack foods and home care products
$1.7 million
2001 14 International products, snack foods, baked goods, drinks, frozen foods, health foods,
kitchen aids, insecticides
$1.2 million
33
TABLE 2. CHARACTERISTICS U.S. OF SUPERMARKETS AND COMMISSARIES (1995)
Average Commissary
Average Supermarket*
Hours Open Per Week
48
131
Stores open Sunday (%)
40
99
Number of Items Stocked
9,600
22,000
Average Customer Transaction**
64 19
Weekly Sales Per Store** 380,000 201,000
Weekly Sales Per Employee**
6,600 3,400
SOURCE: The Costs and Benefits of Retail Activities at Military Bases (Table 2) (citing Congressional
Budget Office based on data from the April 1995 annual report of Progressive Grocer magazine and the
Defense Commissary Agency).
* Supermarket data are for chain stores with annual sales of more than $2 million.
** Commissary sales are adjusted to reflect the commercial retail value of the goods sold.
34
TABLE 3. PRODUCT LEVEL QUANTITY EFFECTS BY CATEGORY
Log (Quantity)
Product Category DD Estimator
(Standard Error) % Change P-Value
Tomato Paste -3.956 (0.766)
-98.086 0.000
Health Food Bars -3.785 (1.033)
-97.729 0.000
Diced Tomatoes -1.695 (0.503)
-81.640 0.001
Pastries 1.567 (0.437)
379.225 0.001
Chili Beans -2.636 (0.860)
-92.835 0.003
Soda -0.978 (0.330)
-62.394 0.003
Tea Mix -2.21 (0.947)
-89.030 0.023
Donuts/Muffins -0.355 (.163)
-29.883 0.030
Pretzels -0.597 (0.284)
-44.954 0.036
Snack Cakes -0.365 (0.191)
-30.580 0.056
Cookies -0.423 (0.239)
-34.492 0.077
Tortilla Chips -0.481 (0.276)
-38.184 0.082
Specialty Pet Food -0.655 (0.366)
-48.056 0.085
Cheese -0.407 (0.236)
-33.436 0.086
Popcorn 1.472 (0.878)
335.794 0.103
Insecticides 0.522 (0.326)
68.540 0.111
Tea Bags -0.631 (0.391)
-46.794 0.112
Coffee Filters 1.598 (0.994)
394.314 0.116
Tea -0.799 (0.551)
-55.022 0.150
35
TABLE 4. PRODUCT LEVEL PRICE EFFECTS BY CATEGORY
Log (Price)
Product Category DD Estimator
(Standard Error) % Change P-Value
Tomato Paste 0.430 (0.016) 53.737 0.008
Health Food Bars -0.567 (0.038) -43.295 0.144
Diced Tomatoes 0.004 (0.019) 0.4 0.836
Pastries -0.064 (0.024) -6.171 0.010
Chili Beans -0.081 (0.047) -7.799 0.089
Soda 0.005 (0.030) 0.518 0.864
Tea Mix 0.047 (0.035) 4.861 0.174
Donuts/Muffins 0.024 (0.014) 2.433 0.080
Pretzels 0.089 (0.018) 9.293 0.000
Snack Cakes 0.004 (0.013) 0.437 0.728
Cookies 0.006 (0.019) 0.604 0.754
Tortilla Chips -0.011 (0.015) -1.101 0.449
Specialty Pet Food
0.003 0.322 0.789 Cheese 0.028
(0.022) 2.83 0.211 Popcorn 0.14
(1.07) 15.081 0.196 Insecticides -0.055
(0.018) -5.315 0.003 Tea Bags -0.004
(0.038) -0.419 0.911 Coffee Filters -0.014
(0.051) -1.422 0.780 Tea -0.045
(0.014) -4.419 0.001
36
TABLE 5. CATEGORY LEVEL QUANTITY EFFECTS
Log (Quantity)
Variable
Coefficient Estimate
(Std. Error) P-Value
Slotshare*After -0.188 (0.500)
0.707
Slotshare -0.636 (0.358)
0.077
After 0 .388 (0.191)
0.042
Constant
14.51614 (0.135)
0.000
Number of Observations 492
37
TABLE 6. CATEGORY LEVEL PRICE EFFECTS
Log (Average Price)
Variable
Coefficient Estimate
(Std. Error) P-Value
Slotshare*After 0.106 (0.164)
0.518
Slotshare -0.042 (0.118)
0.722
After 0.013 (0.063)
0.836
Constant 0.307 (0.045)
0.000
Number of Observations 492
38
TABLE 7. CATEGORY LEVEL VARIETY EFFECTS
(# of SKUs)
Variable
Coefficient Estimate
(Std. Error) P-Value
Slotshare*After 1.019
(41.965) 0.981
Slotshare -10.013
(29.692) 0.736
After -0.631
(16.151) 0.969
Constant 96.641 (11.442)
0.000
Number of Observations 492
39
Appendix A: Slotting Contract Sample
40
41
42
Appendix B: Plan-O-Gram Sample
43
44
45