The courts in ProMedica Health System, Inc. v. FTC and FTC v. Staples, Inc. articulated the rationale for clustering multiple relevant product markets for “analytical convenience.” An important takeaway from those cases is that clustering involves aggregating multiple product markets that share “similar competitive conditions.” Competitive conditions include barriers to entry, as well as the count and significance of competitors in the marketplace. In addition, the exercise of clustering multiple products/services is a permissive one; there is no requirement to add products/services to the cluster or to even cluster at all.
A key question in ProMedica was whether obstetrics (OB) services were properly analyzed separately from the other services typically included in the cluster of general acute care (GAC) services. Using a similar competitive conditions analysis, the ProMedica court noted that OB services were only offered by three of the four hospitals that competed in the broader GAC services market. Thus, OB services were not included in the GAC services cluster (as is done with most hospital cases), but were analyzed as a separate market where the merger resulted in even higher concentration than the remaining GAC services. The court also noted that tertiary services were properly excluded from the cluster as the geographic market would likely be broader, thus bringing in additional competitors.
In Staples II, there was significant debate over whether ink/toner was properly excluded from the cluster of consumable office supplies sold to large business customers. The court in Staples II found that the competitive conditions for ink/toner differed significantly from other office supplies in large part because customers made substantial purchases of ink/toner from printer and copier manufacturers. But those printer and copier manufacturers did not sell other consumable office products. Thus, unlike OB services in ProMedica, which was found to be a less competitive market than the cluster of GAC services, ink/toner was not included because it was found to be a more competitive market than the cluster of consumable office supplies.
Recent health care enforcement actions in which individual service lines were defined separately instead of clustered together include Reading/SIR and Sanford. Reading/SIR involved a specialty hospital that offered a host of services, including inpatient and outpatient orthopedic surgery as well as outpatient ENT surgery and outpatient surgery. The individual physician services involved in Sanford were adult primary care, pediatric, OB/GYN, and general surgery physician services. The competitive conditions in those two cases were not sufficiently similar across the various services to cluster these services together in one cluster market. For instance, in Reading/SIR, the FTC alleged that only hospitals can provide inpatient services. Thus, a significant third-party competitor participated in the market for outpatient orthopedic surgery but was absent in the market for inpatient orthopedic surgery. Further, as explained in prior hospital cases, entry conditions differ between inpatient and outpatient services.
Unlike cluster markets, which aggregate a number of individual relevant markets, a bundle market is the collection of products or services that comprise the relevant market where customers value suppliers offering a package of goods and benefit from the “one-stop shopping” experience. A bundle can also include value-added services that are offered in conjunction with the multiplicity of products in the bundle. Examples of bundle product markets in recent enforcement actions include supermarkets (Albertsons/Safeway, Ahold/Delhaize), discount general merchandise discount stores (Dollar Tree/Family Dollar), broadline food distribution (Sysco/US Foods, US Foods/SGA), casino services (Pinnacle/Ameristar, Penn/Pinnacle, Eldorado/Caesars), and managed care provider organization (MCPO) services (United/DaVita).
One-stop shopping is a key concept in the market definition analysis of supermarkets. Supermarket customers tend to shop for multiple products rather than a single item and shop based on the price of that basket of goods. The FTC defines supermarkets as offering “one-stop shopping for food and grocery products” that are typically at least 10,000 square feet in size and carry more than 10,000 different items. Walmart Supercenters are included as market participants because they offer a similar one-stop shopping experience to traditional supermarkets.
In contrast, the relevant market does not include “hard discounters, limited assortment stores, natural and organic markets, ethnic specialty stores, and club stores,” which also sell grocery items. These are excluded because “they offer a more limited range of products and services than supermarkets and because they appeal to a distinct customer type.”
Discount General Merchandise Retail Stores
In Dollar Tree/Family Dollar, the FTC alleged a product market of “discount general merchandise retail stores.” The merging parties owned thousands of retail stores commonly referred to as “dollar stores.” According to the FTC, dollar stores are “small-format, deep-discount retailers that sell an assortment of consumables and non-consumables, including food, home products, apparel and accessories, and seasonal items, at prices typically under $10.” “Dollar stores differentiate themselves from other retailers on the basis of both convenience and value by offering a broad assortment” of general merchandise items at discounted prices in stores with small footprints located close to consumers’ homes or places of work. Customers often shop at dollar stores as a “fill-in” shopping trip.
Although Walmart is differentiated from dollar stores, in Dollar Tree/Family Dollar, Walmart was included in the relevant product market. According to the FTC, Walmart does not provide the same level of convenience as dollar stores given its locations, larger footprints, and greater assortment of products, but Walmart competes closely with dollar stores by offering a comparable or better value to consumers in terms of pricing. A nuanced point in the Dollar Tree matter is the treatment of other retailers, such as supermarkets, pharmacies, and mass merchandisers, based upon geography. The FTC explained that, “in certain geographic markets, typically characterized by high population density, where the number of and proximity of these other retailers is substantial . . . the collective presence of these other retailers acts as a more significant price constraint.” Thus, some retailers’ status as a market participant depended upon geography.
Broadline Food Distribution
Broadline food distribution shares a similar concept of one-stop shopping as supermarkets. The court in Sysco/US Foods found that broadline food distribution was a relevant market because “distributors market themselves to customers as a ‘one-stop shop,’ by virtue of their ability to supply most—if not all—food and related products needed by their customers.” The court further noted that “what is relevant for consideration here is not any particular food item sold or delivered by Defendants, but the full panoply of products and services offered by them that customers recognize as ‘broadline distribution.’” Product breadth and diversity, including branded and private label products, distinguished broadliners from other channels of distribution. For instance, specialty distributors carry a limited number of SKUs of niche products (e.g., fresh produce, meat, seafood) and cash-and-carry stores carry fewer items with less uniformity.
Broadline food distributors offer services that are not available from other food distributors. Frequent and flexible deliveries (including next-day delivery) was an important attribute that customers valued in broadliners. Moreover, broadline food distributors offered menu and “nutritional-meal planning services” while other modes of delivery generally did not offer comparable value-added services. This combination of product breadth and diversity with value-add services was described in the definition of broadline food distribution in the recent USF/SGA matter as well.
Based on FTC enforcement actions, casino services are also properly analyzed as a bundle of services. Casinos have been defined as “a combination of slot machine, video poker machine, and table gaming (i.e., gambling) services, and associated amenities that are used to drive gaming revenue, which typically include some combination of hotel accommodations, food and beverages, entertainment, and other amenities.” However, not all casinos offer all of the gaming options or associated amenities such as hotel and entertainment. For instance, in Pinnacle/Ameristar, Delta Downs Racetrack Casino in the Lake Charles, LA market is a “racino” (a combined horserace track and casino) that only offers slots, but not table games. Similarly, the Argosy Alton Casino was the only competitor in the St. Louis market that lacked a hotel.
Even if casinos offer most or all parts of the bundle, casino services are highly differentiated. In addition to lacking a hotel, the Argosy Alton Casino was described in the FTC’s complaint as “significantly smaller, lower-end, first-generation riverboat.” The Casino Queen, another competitor in the St. Louis market, was depicted as “a small, lower-end casino with only a small hotel and a recreational vehicle park.” In the Lake Charles market, the Pinnacle/Ameristar complaint alleged that the Isle of Capri Lake Charles “does not have the high-end amenities of [the Respondents’ facilities] and has 350 fewer slot machines, a third fewer tables, and half the number of hotel rooms.”
Managed Care Physician Organizations
Unlike Sanford, where individual physician services were analyzed separately, in United/DaVita the FTC alleged a bundle market of physician services sold to Medicare Advantage Organizations (MAOs). The bundle of services alleged to be the relevant market was managed care provider organizations, where MAO customers valued the combination of medical and non-medical services offered as an integrated product. MCPOs orchestrate networks of owned, employed, and affiliated providers (including hospitals, outpatient clinics, physician groups, and individual physicians) for the purpose of managing care of MAO plans’ patient population. Furthermore, MCPOs often employ a variety of clinical and non-clinical support personnel and have developed information technology systems dedicated to managing care utilization and monitoring patient care. Thus, MCPOs offer a broad set of providers that can independently manage an MAO’s network while effectively coordinating care, managing utilization, and containing costs for a covered patient population.
Qualitative and Quantitative Evidence Supporting Bundle Markets
Several merger cases demonstrate that both qualitative and quantitative evidence can be important in supporting bundle markets, especially in cases in which a critical inquiry is whether a bundle market is defined too narrowly. In Sysco, for example, the FTC’s expert conducted an aggregate diversion analysis to demonstrate that a hypothetical monopolist of broadline food distribution could profitably increase prices even though other modes of distribution offer some products in the bundle. Further, the Sysco court took guidance from two previous merger cases litigated in the D.C. District Court and D.C. Circuit Court––Staples I and Whole Foods.
In Staples I, office supplies superstores (OSS) was found to constitute a relevant product market apart from other retail outlets where office supplies were sold: “the unique combination of size, selection, depth, and breadth of inventory offered by the superstores distinguishes them from other retailers.” It is notable that Staples I involved OSS retail stores as a bundle market whereas Staples II involved the cluster of office supplies sold on a business-to-business basis. The court in Staples I cited evidence that supported a finding of an OSS-only market including:
price zones based on the presence (or absence) of other OSS;
pricing data demonstrating that prices in areas where Staples faced no competition from other OSS were 13 percent higher than areas where all three OSS were present; and
Staples changed its pricing behavior based on entry of other OSS (but not other retailers).
Moreover, Judge Hogan, who presided over the Staples I case, made the following observation after visiting various retailers of office supplies (including Staples, Office Depot, CompUSA, Best Buy, CVS, Kmart, Giant Food, and Walmart): “The unique combination of size, selection, depth and breadth of inventory offered by the superstores distinguishes them from other retailers.” He went on to conclude that “[n]o one entering Staples or Office Deport would mistakenly think he or she was in Best Buy or CompUSA. You certainly know an office superstore when you see one.”
In Whole Foods, the relevant product market was found to be “premium and natural organic supermarkets” (PNOS), which excluded traditional supermarkets. The D.C. Circuit pointed to quantitative analyses that supported a finding of a PNOS market, including a comparison of margins based on the presence (or absence) of another PNOS competitor as well as an event study comparing the effect on prices of entry by PNOS competitors versus conventional supermarkets. The court also cited a Whole Foods study called “Project Goldmine,” which analyzed the likely reaction of customers in cities where Whole Foods planned to close Wild Oats stores. According to that study, Whole Foods stores would capture most of the revenue from closed Wild Oats stores despite the presence of numerous conventional supermarkets that were more proximate to the Wild Oats stores.
The concept of cross-shopping is often brought up by defendants as evidence that the market should include certain competitors. The district court in Whole Foods spent five pages discussing the cross-shopping habits of customers who shopped at both PNOS stores and traditional supermarkets. However, in finding that PNOS was a separate market from traditional supermarkets, Judge Tatel stated “[t]hat Whole Foods and Wild Oats have attracted many customers away from conventional grocery stores . . . tells us nothing about whether [they] should be treated as operating in the same market as conventional grocery stores.” In Sysco, the court similarly acknowledged that customers, especially independent restaurants, purchase across different distribution channels, including from specialty distributors. Nevertheless, the court cited the Areeda & Hovenkamp treatise, which admonished the relevance of cross-shopping on market definition: “It would be improper to group complementary goods into the same relevant market just because they occasionally substitute for one another.”
Because cluster and bundle markets often involve a customer purchasing multiple products and services from a common supplier, this can cause confusion over which of the two analyses is appropriate. The defendants in Staples II and ProMedica argued that the court should adopt a bundle market to expand the FTC’s proposed market definition. Staples argued that ink and toner should be included in the market because customers valued the bundle, and, in fact, the court acknowledged that customers preferred to one-stop shop with office supply vendors. However, the court also noted that customers’ preference for the option of purchasing additional items did not necessitate that those items be included in the market definition. Similarly, the ProMedica court found that a cluster analysis was more appropriate than a bundle analysis. Although insurers “typically bargain for all of a hospital’s services in a single negotiation,” and therefore appreciate the bundle, they “do not demand from each hospital a package of services.”
Although the customers in Staples II and ProMedica tended to purchase multiple products from a single vendor, this dynamic alone was not sufficient to apply a bundle market analysis. Putting aside whether the aggregation of individual products and services is properly analyzed as a cluster or bundle, the court in Staples II conducted a separate product market analysis focused on specific types of customers. In following the Merger Guidelines Section 4.1.4, the court undertook a targeted customer analysis to find a product market of “Large B-to-B Customers.” Indeed, even though the court characterized the product market as a cluster market, the discussion about the distinct needs and preferences of these targeted customers seems to rely on a bundle-type analysis. For instance, large business-to-business customers in Staples II valued: (1) sophisticated IT capabilities including customizable product catalogs and electronic procurement systems; (2) personalized, high-quality customer services; and (3) national next-day and desktop delivery. This analysis closely mirrors the analysis the Sysco court undertook in finding a market for targeted “National Customers.”
The relationship among products differs across cluster and bundle markets, affecting not only how they are assessed but also the importance of defining them correctly. The exercise of clustering is a permissive one. Each product or service that is included in a cluster market constitutes a relevant market, and aggregating those individual markets to a cluster market for evaluation is allowable only if each product or service included in the cluster meets a similar competitive conditions requirement. If the similar competitive conditions requirement is in question (such as ink/toner in Staples II), there is no risk in “not including” a certain product/service in the cluster and evaluating it separately.
In contrast, bundles are considered a relevant product market for an entirely different reason––the pricing and other strategic decisions for the individual products in a bundle are so intertwined that the bundle itself is a distinct product. Importantly, unlike in cluster markets, the individual products in a bundle need not face similar competitive conditions, and thus the risks involved in “excluding” products/services from the bundle can be profound. For example, had broadline food distribution not included breadth and diversity as a feature of the relevant product market described in Sysco, then specialty distributors and cash-and-carry outlets may well have been included as market participants. Accordingly, market shares for Sysco and US Foods in this broader market might not have met the Merger Guidelines threshold for presumptively unlawful transactions.
Historical Use of “Cluster”
In some cases, the Supreme Court has improperly used the term “cluster” when the relevant product market at issue was a bundle market. In United States v. Philadelphia National Bank (PNB), the Court stated the “cluster of products (various kinds of credit) and services (such as checking accounts and trust administration) denoted by the term ‘commercial banking’ . . . composes a distinct line of commerce.” But the Court noted that competitive conditions differed between checking account services and the personal loan services, suggesting that such services should not be clustered together based on the principle subsequently provided by ProMedica and Staples II. Seven years later, in United States v. Phillipsburg National Bank, the Supreme Court again used the term “cluster” to refer to commercial banking services but expanded on the one-stop shopping dynamic: “A customer who uses one service usually looks to his bank for others as well . . . customers are likely to maintain checking and savings accounts in the same local bank even when higher savings interest is available elsewhere.” Thus, the Court’s analysis suggests that commercial banking services should properly be analyzed as a bundle market.
In United States v. Grinnell Corp., the Court found central service stations to be a relevant product market that included fire sprinkler, fire protection, burglary, and other protection systems. The Grinnell court took guidance from PNB, stating there was “a comparable cluster of services.” Although the Court stated that it would be “unrealistic . . . to break down the market into the various kinds of central station protective services,” there was never any discussion about grouping together the different services for analytical convenience.
The Court noted that there is a “single basic service––the protection of property through use of a central service station.” Furthermore, “central station companies recognize that to compete effectively, they must offer all or nearly all types of service[s].” Hence, both supply- and demand-side dynamics suggested that the bundle of alarm services constituted a relevant market. Unfortunately, the courts in both Grinnell and Sysco use the term “cluster” when in fact the relevant product markets were bundle markets.
In Brown Shoe, the Supreme Court properly found the relevant product market to be the cluster of men’s, women’s, and children’s shoes. The Court stated that “whether considered separately or together, the picture of this merger is the same.” Additionally, when considering whether to analyze submarkets within these categories, the Court stated that “[f]urther division does not aid us in analyzing this merger” and that further subdivision of the general shoe market based on “age/sex” or “price/quality” distinctions would be “impractical” and “unwarranted.”
There are several open questions with respect to cluster and bundle market analyses. In ProMedica, the Sixth Circuit noted competitive conditions between OB and GAC services markets differed in two respects. As discussed above, there were a different set of competitors in the two markets––one fewer competitor participated in the OB market. In addition, the court noted that ProMedica’s market share in OB services (71 percent) was significantly greater than its share in GAC services (47 percent). But would the market share difference alone warrant keeping OB out of the cluster (where the same number of competitors participated in both markets)? Assume that the same number of competitors participated in the marketplace as GAC services, but other facts were different:
What if ProMedica’s premerger share of OB services was 90 percent?
What if ProMedica offered OB services at two hospitals in the area versus its competitors offering OB services at just one facility each?
What if ProMedica was closely affiliated with the largest independent OB group that tended to perform their deliveries at a ProMedica hospital?
Instead, what if ProMedica employed that large OB group?
We posit that, in some instances, the similar competitive conditions test justifies that a service be analyzed separately from the cluster even if the same market participants compete in all markets. Returning to the principle that clustering is permissive and not required, there is no danger in doing so, because the proper approach absent clustering is to separately analyze each individual service.
Given that there is less guidance from case law on bundle markets, there are myriad issues that are unresolved. As previously discussed, there appears to be some subjectivity as to which components of the bundle are necessary for competitors to be counted as a market participant. To highlight this ambiguity, consider the following questions related to market participants included in the markets for broadline distribution, casino services, and supermarkets:
Which products/services must a food distributor need to offer to be considered a broadliner? A minimum number of total SKUs? A minimum number of food categories (e.g., meat, seafood, produce, etc.)? Is a sufficient number of private label SKU offerings a requirement? What about next-day delivery?
Which companies are properly considered market participants of casino services? In Pinnacle/Ameristar, competitors that lacked table games or a hotel were still viewed as market participants. If a company does not offer table games, is there a minimum number of slot machines or video gaming required? What if a property offered no slots or video gaming and only a limited number of poker tables?
What criteria must be met for a food retailer to be considered a supermarket? A minimum number of products? A minimum number of categories? A minimum number of branded products? A minimum square footage area? Is a brick-and-mortar offering necessary?
Market definition is often the dispositive issue in antitrust merger cases. The Merger Guidelines explain that market definition focuses on substitutability of demand for products in a candidate market. However, multi-product firms selling several products and services that are not reasonable substitutes for each other can raise a separate question: should products be grouped together and considered as a cluster or bundle market? As explained above, clustering involves grouping products that face independent but similar demand conditions for analytical convenience. Clustering should be of little consequence to inferences of the likely competitive effects of a merger. Hence, a reasonable approach is “when in doubt, don’t cluster.”
In contrast, grouping products together and considering them as a bundled product can have profound impacts on the analysis of the competitive effects of a merger. Incorrectly grouping products as a bundle may result in structural measures that overstate or understate the likely competitive effects of a merger. A product market properly defined as a bundle of products may exclude firms that only offer part of the bundle as market participants. For example, specialty food distribution is not included in broadline food distribution and club stores are not included as supermarkets. Bundle analysis may apply to other industries, such as technology platforms or fintech, which offer multiple services, including some free services. While ride-sharing and online food delivery services are clearly separate markets, could the bundle of such services constitute a relevant market?
Regardless of whether products are clustered or bundled, relevant markets must satisfy the hypothetical monopolist test as prescribed by the Merger Guidelines. Moreover, any mistakes in improperly clustering or bundling will likely be corrected by the antitrust agencies or courts in a rigorous analysis of competitive effects.