Freeman v. San Diego Ass'n of Realtors

322 F.3d 1133, 2003 Cal. Daily Op. Serv. 2148, 2003 U.S. App. LEXIS 4091, 2003 WL 931834

• Court: Court of Appeals for the Ninth Circuit
• Decided: March 10, 2003
• Docket: Nos. 01-56199, 01-56231
• Status: Published

Opinion

OPINION

KOZINSKI, Circuit Judge:

Competition is the mainspring of a capitalist economy. Sometimes, however, cooperation can make markets more efficient; setting industry standards and pooling market data are two examples of arrangements that often benefit consumers. Antitrust laws acknowledge these benefits, but still treat the arrangements with skepticism, for seemingly benign agreements may conceal highly anticom-petitive schemes. We apply these principles to a case involving a real estate Multiple Listing Service.

Background¹

Real estate agents make a living matching buyers and sellers. Up-to-date information about properties on the market is a must. Long gone are the days when agents trawled the neighborhood on horseback in search of telltale “For Sale” signs. We’re now in the era of the Multiple Listing Service, or “MLS,” which lets agents share information about properties on the market with the help of a computerized database. Agents who subscribe to the MLS can peruse the listings of other subscribers and post their own.²

Care and feeding of an MLS involves more than just maintaining a database. Someone must enroll new MLS subscribers, bill and collect payments, ensure that postings comply with guidelines and provide support staff to answer subscribers’ questions. These “support services” are part of the necessary overhead of delivering an MLS.

Before 1992, twelve MLSs served San Diego County, each operated by a different real estate trade association serving subscribers in a particular region of the county. Some of these MLSs shared common databases of listings. The direct cost of maintaining each database was allocated among the associations using it. The associations each provided their own support services, which varied from one to the next, and they all set the prices they charged agents independently. There were four different MLS databases in total, so a real estate agent who wanted access to all properties in San Diego County had to subscribe to more than one MLS.

Eleven of the twelve MLS operators were local Associations of Realtors, professional groups with ties to the California Association of Realtors (CAR) and the National Association of Realtors (NAR).³ The Associations of Realtors provide many services in addition to MLS database access and support services to their real estate agent members. Under state law, however, they are required to offer MLS subscriptions to members and nonmembers alike. See Marin County Bd. of Realtors v. Palsson, 16 Cal.3d 920, 937-38, 130 Cal.Rptr. 1, 549 P.2d 833 (1976).

In 1990-91, these eleven associations decided to combine their databases. Not only would this give their agents access to all San Diego County properties through a single MLS, it would reduce operating costs because there would only be one database to maintain. Representatives met and decided to create a new entity called Sandicor with a single MLS database covering all of San Diego County. They considered it vital, though, that individual associations continue to provide local support services for the MLS. Comparing finan-cials, they discovered that the associations’ costs of providing services varied widely. The largest, San Diego Association of Realtors (SDAR), spent only $10 per subscriber per month, while two of the smallest, Fallbrook and Valley Center, spent close to $50.

The representatives considered two different business models on which to operate the new common database. The first was a “decentralized” arrangement in which each association would pay its share of the database costs but determine its own support service and pricing policies. The smaller associations balked at this proposal, fearing that SDAR would undercut their prices and threaten their viability. The representatives responded by adopting the current “centralized” business model instead.

In this model, Sandicor is a corporation, and the associations own its shares and appoint its directors. The associations operate under service agreements with Sandicor that outline in general terms the support services they must perform for subscribers. Associations sign up new MLS subscribers and collect MLS fees from subscribers on Sandicor’s behalf, but Sandicor determines the fee agents must pay to subscribe. Associations are prohibited from discounting Sandicor’s MLS fee or crediting any portion of it against other purchases. The service agreements between Sandicor and the associations specify a support fee that Sandicor pays each association in return for the support services the association provides to subscribers. Thus, subscribers don’t pay the associations for support services directly; they pay only Sandicor’s MLS fee, and Sandicor then returns part of that fee as a support fee to the associations. The support fee, like the MLS fee, is assessed on a per-subscriber basis.⁴

The associations originally set the support fee at $25 per subscriber per month.⁵ This was much higher than SDAR’s $10 costs but also considerably lower than Fallbrook’s and Valley Center’s $50 costs. This meant that Fallbrook and Valley Center would provide services to their subscribers at a loss. To make up the shortfall, the other associations agreed to pay them fixed monthly cash subsidies.

Defendants explain that this centralized model was “a fundamental matter for survival” because the smaller associations needed assurance that they could “continue to operate service centers as they had in the past and that the revenues they had received from MLS services would continue to be available.” If Sandicor had adopted a decentralized format, “the far larger San Diego Association of Realtors ... would undoubtedly have been able to offer different prices to MLS users than would Fallbrook,” which would have “threatened the future viability of local service centers.” Without the revenue guaranteed by the fixed support fee, “the smaller Associations would not have joined,” and Sandicor would not have had a “truly regional MLS.”

Plaintiffs’ Suit and Proceedings Below

Arleen Freeman and James Alexander are San Diego County real estate agents who subscribe to Sandicor’s MLS. They have no quarrel with Sandicor’s monopoly of the MLS database itself, recognizing that the single countywide database offers substantial overall benefits to consumers. They claim, however, that the price of Sandicor’s MLS is inflated because the support fees Sandicor pays the associations are fixed at a supracompetitive level. For example, although SDAR may spend only $10 per subscriber per month to provide support services,⁶ it receives more than twice that from Sandicor under its service agreement. As a result, plaintiffs claim, SDAR has made millions of dollars from inflated support fees. Real estate agents allegedly suffer because Sandicor passes on these higher support fees in the form of higher MLS fees. Plaintiffs allege that, by fixing support fees in the service agreements, defendants violated section 1 of the Sherman Act, which bars “[e]very contract ... or conspiracy[ ] in restraint of trade or commerce among the several States.” 15 U.S.C. § 1.

Plaintiff Freeman also sent a letter to Sandicor in which she proposed to operate a service center that would provide Sandi-cor’s MLS to subscribers. She claimed she could provide support services at a much lower price than the associations. Sandicor declined. Freeman believes Sandicor refuses to authorize new MLS support service providers for anticompeti-tive reasons, and that restrictive provisions in Sandicor’s shareholders’ agreement allow the associations to veto new sources of competition. Defendants respond that Sandicor refused the request because it was an obvious litigation ploy rather than a bona fide proposal. Plaintiffs nonetheless claim that the associations violated section 2 of the Sherman Act by conspiring to monopolize the market for support services. 15 U.S.C. § 2.⁷

Plaintiffs filed a class action in federal court seeking an injunction and damages in excess of $10 million, naming as defendants Sandicor, the associations, the California Association of Realtors and assorted officers and directors of these organizations.⁸ The district court refused to certify the class, dismissed the claims against essentially all the directors and officers, and dismissed certain other claims as abandoned. None of those decisions is before us.⁹

Following discovery, both sides moved for summary judgment. Defendants argued that their conduct had no substantial effect on interstate commerce and that they were immune from section 1 because they were a single entity under Copperweld Corp. v. Independence Tube Corp., 467 U.S. 752, 104 S.Ct. 2731, 81 L.Ed.2d 628 (1984). The district court rejected the first defense but sustained the second, holding that the associations were “not separate entities pursuing separate interests that conspired to create Sandicor.” It also found that Sandicor didn’t “buy” anything from the associations, holding that “the [associations] do not sell service center functions to Sandicor, but rather provide Sandicor’s MLS to users.”

Both sides appeal. The district court had jurisdiction under 28 U.S.C. §§ 1331 and 1337. We have jurisdiction under 28 U.S.C. § 1291, and, except as noted below, we review de novo, Morrison v. Hall, 261 F.3d 896, 900 (9th Cir.2001).

Interstate Commerce

The Sherman Act applies to restraints of trade or commerce “among the several States.” 15 U.S.C. §§ 1-2. Monopolizing the local lemonade stand doesn’t get you into federal court. To make a federal case, a plaintiff must show that the activities in question, although conducted within a state, have a “substantial effect on interstate commerce.” McLain v. Real Estate Bd. of New Orleans, Inc., 444 U.S. 232, 242, 100 S.Ct. 502, 62 L.Ed.2d 441 (1980). Uncontradicted evidence shows that Sandicor’s MLS mediated more than $23 billion of home sales over a four-year period, affecting approximately $10 billion in interstate home-mortgage financing. The effects of home sales on the interstate mortgage market are a sufficient connection with interstate commerce, see id. at 246, 100 S.Ct. 502, and $10 billion is “substantial” by any standard.

Defendants urge that we consider only the interstate effects caused by the illegal conduct itself. According to their economics expert, the cost of an MLS subscription is trivial compared to the typical real estate sales commission. Thus, even if the cost of the MLS was inflated, it would have only a de minimis effect on the commissions real estate agents charge, and thus no effect on the number of houses sold. Defendants misunderstand the legal standard. The Supreme Court has explained:

To establish the jurisdictional element of a Sherman Act violation it would be sufficient for petitioners to demonstrate a substantial effect on interstate commerce generated by respondents’ [“infected”] activity. Petitioners need not make the more particularized showing of an effect on interstate commerce caused by the alleged conspiracy to fix [prices], or by. those other aspects of respondents’ activity that are alleged to be unlawful.

Id. at 242-43, 100 S.Ct. 502; see also Summit Health, Ltd. v. Pinhas, 500 U.S. 322, 331, 111 S.Ct. 1842, 114 L.Ed.2d 366 (1991) (reaffirming this standard). McLain tells us to look at the infected activities, not the infection. Thus, we examine the MLS, not the alleged price fixing. Even if the price of the MLS has no substantial effect on interstate commerce, the MLS itself does. Defendants’ economist may be right, but he answered the wrong question.

Defendants perceive a split in Ninth Circuit authority between cases applying the infected activities test and those considering a defendant’s “ ‘general business’ activities.” See Mitchell v. Frank R. Howard Mem’l Hosp., 853 F.2d 762, 764 n. 1 (9th Cir.1988) (citing W. Waste Serv. Sys. v. Universal Waste Control, 616 F.2d 1094, 1097 (9th Cir.1980)). Neither flavor helps defendants; what they want is the “illegal conduct” test that McLain rejected. And that, we are not at liberty to provide.

Price Fixing

1. No antitrust violation is more abominated than the agreement to fix prices.¹⁰ With few exceptions, “ ‘price-fixing agreements are unlawful per se under the Sherman Act and ... no showing of so-called competitive abuses or evils which those agreements were designed to eliminate or alleviate may be interposed as a defense.’” Arizona v. Maricopa County Med. Soc’y, 457 U.S. 332, 345, 102 S.Ct. 2466, 73 L.Ed.2d 48 (1982) (quoting United States v. Socony-Vacuum Oil Co., 310 U.S. 150, 218, 60 S.Ct. 811, 84 L.Ed. 1129 (1940)). The dispositive question generally is not whether any price fixing was justified, but simply whether it occurred.

None of the relevant facts is disputed. Prior to 1992, the associations made independent decisions about how to price their support services, even though many of them shared common databases. When they decided to form a countywide database, they debated whether to continue that system of independent pricing or to set a fixed, uniform fee that they would receive for providing support services. They opted for the latter arrangement and selected a fee that was more than double the cost of the most efficient association. They admit that they fixed the fee in order to ensure that financially weaker associations would make more money than under a competitive regime, and thus concede by implication that they intentionally fixed the fee at a supracompetitive level.

The district court nonetheless rejected plaintiffs’ theory of the case because it was not convinced that the associations sold anything to Sandicor. The court explained that “[a] review of the Service Center Agreements ... reveals the Associations do not sell service center functions to Sandicor, but rather provide Sandicor’s MLS to users.” The terms of the service agreements, however, are unambiguous: “The Association will provide the following [support] services to SANDICOR MLS participants,” Service Agreement ¶ 1; “[i]n consideration of the above services, SANDICOR will pay to the Association the fees itemized on Schedule A,” id. ¶ 2. Sandicor is paying for something. What it’s buying is the contractual right to have the associations provide support services to its MLS subscribers. That the subscribers receive the support services as third-party beneficiaries doesn’t change the fact that Sandicor is the buyer and the associations are the sellers. Cf. FTC v. Superior Court Trial Lawyers Ass’n, 493 U.S. 411, 422-23, 110 S.Ct. 768, 107 L.Ed.2d 851 (1990) (treating the government as the “customer” when it pays for representation of indigent criminal defendants). If Elle Woods pays someone to walk her dog, she’s the buyer and the dog-walker is the seller, even though Bruiser gets the exercise. And if the Cambridge dog-walking cartel starts fixing prices, it’s hardly a defense for them to say “We just walk dogs!”

Sandicor charges subscribers for their use of the MLS; its MLS fee includes the support services provided by the associations. The support fee Sandicor in turn pays the associations for support services was fixed at a level more than twice what it cost the most efficient association to provide them. These inflated support fees harm subscribers if Sandicor passes them on in the form of higher MLS fees. The district court found that “there is no evidence ... [that Sandicor] passes on an inflated price to consumers.” This finding is contrary to the record.

There is unmistakable evidence that Sandicor not only considered its costs in setting MLS fees but, in fact, priced near cost and thus may have passed on the inflated support fees almost dollar for dollar. For example, an April 1997 association bulletin explains that “[Sandicor’s] pricing takes the total amount required to run Sandicor and divides by the total number of agents on the system. The amount to run Sandicor includes ... the amount Sandicor sends back to each Association, by contractual arrangement, to provide you with local service.” North San Diego County Ass’n of Realtors, Sandicor Board of Directors Approves Per Capita Fee Structure, Service Center Supplement, Apr. 1997, at 1, E.R. at 441 (emphasis omitted). Other materials in the record are to the same effect. See, e.g., Letter from William C. Stegall, President/CEO of Sandicor (June 20,1997), E.R. at 454 (stating that Sandicor would study ways in which “costs can be reduced” so that it could sell MLS at the “lowest possible cost in the future for all of our consumers”); San Diego Ass’n of Realtors, M.L.S.—Why $27 Per Month? (Jan. 3, 1992), E.R. at 256 (“[Sandicor’s MLS] price was determined by the lowest possible cost to users....”). This evidence is sufficient to show that Sandicor’s prices reflected its costs to some extent and thus that an inflated support fee injured plaintiffs. In contrast, the record does not support the counterintuitive claim that Sandicor is some sort of accounting pyrrhonist that sets its prices in utter disregard of its costs of doing business.¹¹

That a purchaser suffers higher prices from passed-on costs doesn’t necessarily mean he can sue. Illinois Brick denies standing to indirect purchasers in many circumstances. Ill. Brick Co. v. Illinois, 431 U.S. 720, 97 S.Ct. 2061, 52 L.Ed.2d 707 (1977). This rule has exceptions, however — two of which apply here. First, Illinois Brick doesn’t apply to equitable relief, see Lucas Auto. Eng’g, Inc. v. Bridgestone/Firestone, Inc., 140 F.3d 1228, 1235 (9th Cir.1998), which is one of the remedies plaintiffs seek. Second, indirect purchasers can sue for damages if there is no realistic possibility that the direct purchaser will sue its supplier over the antitrust violation. See Royal Printing Co. v. Kimberly-Clark Corp., 621 F.2d 323, 326 (9th Cir.1980). The associations own Sandicor. They appoint its board of directors, and they are accused of conspiring with it. There’s no realistic possibility Sandicor will sue them.¹²

Were we to grant immunity from section 1 merely because defendants nominally sell services through another entity rather than to consumers directly, we would risk opening a major loophole for resale price maintenance and retailer collusion. Consider the following: Ford can’t sell cars to its dealers wholesale for $20,000 and require them to mark up the price exactly $4000 before reselling them to the public — that would be resale price maintenance, a per se violation of section 1. See Bus. Elecs. Corp. v. Sharp Elecs. Corp., 485 U.S. 717, 724, 108 S.Ct. 1515, 99 L.Ed.2d 808 (1988) (citing Dr. Miles Med. Co. v. John D. Park & Sons Co., 220 U.S. 373, 31 S.Ct. 376, 55 L.Ed. 502 (1911)). Nor can the dealers collude among themselves to peg the retail markup at $4000. See Plymouth Dealers’ Ass’n v. United States, 279 F.2d 128, 134 (9th Cir.1960). Can Ford and its dealers circumvent these prohibitions by agreeing that the dealers, while continuing to bear the same economic risks, will nominally sell the cars on Ford’s behalf for $24,000 and receive a $4000 “reimbursement” from Ford for each car sold? We think not. Yet this is in material respects analogous to what Sandicor and the associations did here. Sandicor charges MLS subscribers $44 per month; an association collects this fee from each subscriber and hands it over to Sandicor, which then returns $22.50 to the association as the support fee.¹³ But the association might as well simply collect $44, send $21.50 to Sandicor and keep the other $22.50 for itself; defendants’ counsel forthrightly conceded that “economically, it could be described either way.” For that matter, the associations could purchase MLS database access from Sandicor for $21.50 on an as-needed basis and then agree among themselves to resell it with support services for exactly $22.50 more— again, there’s no practical economic difference. Defendants can’t turn a horizontal agreement to fix prices into something innocuous just by changing the way they keep their books.¹⁴

The discount ban, while not necessary to our finding of a section 1 violation, certainly supports our conclusion. Agreements not to offer discounts are per se violations of section 1. Catalano, Inc. v. Target Sales, 446 U.S. 643, 648, 100 S.Ct. 1925, 64 L.Ed.2d 580 (1980) (per curiam). If associations were permitted to refund part of Sandicor’s MLS fee to subscribers, they could compete on price notwithstanding the contractually fixed support fee.¹⁵ Without the ban, the price-fixing structure might well have collapsed. We can conceive of no legitimate justification for it, and defendants have offered none.

No one doubts that Sandicor and the associations may set policies necessary to maintain the MLS database. See United States v. Realty Multi-List, Inc., 629 F.2d 1351, 1368 (5th Cir.1980) (allowing an MLS operator to enforce “reasonably ancillary restraints”). Obviously, for example, they can dictate standards for data entry. Sandicor also retains considerable freedom to specify services that the associations must provide. Cf. Cont’l T.V., Inc. v. GTE Sylvania Inc., 433 U.S. 36, 59, 97 S.Ct. 2549, 53 L.Ed.2d 568 (1977). But this flexibility does not extend to what the associations did here.

The associations engaged in price fixing, and plaintiffs have standing to sue them. The associations purposely fixed the support fee they charged Sandicor at a supracompetitive level.¹⁶ Sandicor passed on some portion of that inflated support fee to agents, who paid higher prices for the MLS as a result. This is precisely the type of injury the antitrust laws are designed to prevent.

2. Defendants argue that they are immune from section 1 because they constitute a “single entity” and are thus incapable of conspiring with one another. Section' 1, like the tango, requires multiplicity: A company cannot conspire with itself. Copperweld Corp. v. Independence Tube Corp., 467 U.S. 752, 769, 104 S.Ct. 2731, 81 L.Ed.2d 628 (1984). If two erstwhile competitors combine to become a single economic entity — by merger or acquisition, for example — the act of combination may violate the antitrust laws, but their subsequent relations are generally immune from section 1.

The single-entity rule is relevant in a variety of contexts. It applies to a company and its officers, employees and wholly owned subsidiaries. Id. at 769, 771, 104 S.Ct. 2731. It also applies to subsidiaries controlled by a common parent, Thomsen v. W. Elec. Co., 680 F.2d 1263, 1265-66 (9th Cir.1982), firms owned by the same person, Las Vegas Sun, Inc. v. Summa Corp., 610 F.2d 614, 616, 618 (9th Cir.1979), and a firm owned by a subset of the owners of another, Copperweld, 467 U.S. at 773 n. 21, 104 S.Ct. 2731 (citing Sunkist Growers, Inc. v. Winckler & Smith Citrus Prods. Co., 370 U.S. 19, 29, 82 S.Ct. 1130, 8 L.Ed.2d 305 (1962)). It applies to principal-agent relationships, Calculators Haw., Inc. v. Brandt, Inc., 724 F.2d 1332, 1336 (9th Cir.1983), and to “partnerships or other joint arrangements in which persons who would otherwise be competitors pool their capital and share the risks of loss as well as the opportunities for profit.” Maricopa County Med. Soc’y, 457 U.S. at 356, 102 S.Ct. 2466; see also Hahn v. Or. Physicians’ Serv., 868 F.2d 1022, 1029 n. 5 (9th Cir.1988). The theme in these cases is economic unity. Where there is substantial common ownership, a fiduciary obligation to act for another entity’s economic benefit or an agreement to divide profits and losses, individual firms function as an economic unit and are generally treated as a single entity.

Some decisions have found a single entity even in the absence of economic unity. In City of Mt. Pleasant v. Associated Electric Cooperative, Inc., 838 F.2d 268 (8th Cir.1988), the Eighth Circuit held that a joint venture of independently owned regional electric cooperatives was a single entity. Id. at 271. It opined that “legally distinct entities cannot conspire among themselves if they ‘pursue! ] the common interests of the whole rather than interests separate from those of the [group] itself.’ ” Id. at 274 (quoting Copperweld, 467 U.S. at 770, 104 S.Ct. 2731). Crucially, though, there was no evidence that “defendants are, or have been, actual or potential competitors.” Id. at 276. Similarly, Williams v. I.B. Fischer Nevada, 794 F.Supp. 1026 (D.Nev.1992), aff'd, 999 F.2d 445 (9th Cir.1993) (per curiam), found that a fast food franchisor and its franchisees were a single entity; among other things, the franchisees were located too far apart to be effective competitors, even though they could vary their prices. See id. at 1031.

On the other hand, in Los Angeles Memorial Coliseum Commission v. National Football League, 726 F.2d 1381 (9th Cir.1984), we held that the teams of the NFL are not a single entity.¹⁷ We noted that “[t]he member clubs are all independently owned,” id. at 1389-90, and that “profits and losses are not shared,” id. at 1390. We further observed that “the NFL clubs do compete with one another off the field as well as on to acquire players, coaches, and management personnel.” Id. This competitive element distinguishes Los Angeles Memorial Coliseum from cases like Mt. Pleasant and Williams.

Although the single-entity inquiry is fact-specific, Williams, 999 F.2d at 447, a few general guidelines emerge. First, in the absence of economic unity, the fact that joint venturers pursue the common interests of the whole is generally not enough, by itself, to render them a single entity. “[A] commonality of interest exists in every cartel.” L.A. Mem’l Coliseum, 726 F.2d at 1389. Retailers have a common interest in distributing a supplier’s product, but they certainly are not exempt from section 1. See Plymouth Dealers, 279 F.2d at 134. And, the Court routinely scrutinizes joint ventures that involve aspects of common interest. See, e.g., NCAA v. Bd. of Regents, 468 U.S. 85, 104 S.Ct. 2948, 82 L.Ed.2d 70 (1984); Broad. Music, Inc. v. CBS, 441 U.S. 1, 99 S.Ct. 1551, 60 L.Ed.2d 1 (1979); United States v. Topco Assocs., 405 U.S. 596, 92 S.Ct. 1126, 31 L.Ed.2d 515 (1972); United States v. Sealy, Inc., 388 U.S. 350, 87 S.Ct. 1847, 18 L.Ed.2d 1238 (1967); see also 1 Julian O. von Kalinowski et al., Antitrust Laws and Trade Regulation § 11.02[2], at 11-17 (2d ed.2000) (noting that cases finding joint venturers to be incapable of conspiracy are the “exception”).

Second, in the absence of economic unity, the fact that firms are not actual competitors is also usually not enough, by itself, to render them a single entity. Absence of actual competition may simply be a manifestation of the anticompetitive agreement itself, as where firms conspire to divide the market. See Maricopa County Med. Soc’y, 457 U.S. at 344 n. 15, 102 S.Ct. 2466 (division of markets is per se illegal). Cases have required instead that the constituent entities be neither actual nor potential competitors, City of Mt. Pleasant, 838 F.2d at 276, cf. Williams, 794 F.Supp. at 1031.

Finally, where firms are not an economic unit and are at least potential competitors, they are usually not a single entity for antitrust purposes. This rough guideline fairly captures the holdings of the cases above.

These principles resolve the issue here.¹⁸ The associations do not function as an economic unit in providing support services. Sandicor is a corporation, so agreements wholly internal to the company and its officers and employees are presumably immune.¹⁹ But the agreements at issue extend beyond Sandicor to its shareholder associations contracting on their own accounts. See Fraser v. Major League Soccer, L.L.C., 284 F.3d 47, 57 (1st Cir.2002) (“That a stockholder may be insulated by Copperweld when making ordinary governance decisions does not mean automatic protection when the stockholder is also an entrepreneur separately contracting with the company.”). The associations have no substantial common ownership; they are mutual benefit corporations independently owned by their respective members. Their profits thus don’t all wind up under the same corporate mattress. Even assuming that the associations act as Sandicor’s “agents” for some purposes, their decision to fix the price of the support services they sell to Sandicor at a supracompetitive level — services on which the associations bear the economic risk — was not an exercise of agency authority on Sandicor’s behalf. Cf. Oltz v. St. Peter’s Cmty. Hosp., 861 F.2d 1440, 1450 (9th Cir.1988). Nor do the associations share profits among themselves; Sandicor pays support fees on a per-subscriber basis, so an association’s revenues are tied to the number of agents it signs up. An association that attracts lots of agents doesn’t share the fruits of that industry with one that attracts few.

The associations are also at least potential competitors. Nothing innate in the economics of a countywide MLS requires an agent to subscribe through one association rather than another. Because Sandi-cor pays support fees on a per-subscriber basis, competition among associations to sign up new MLS subscribers is also competition in the market to sell support services to Sandicor. The more subscribers an association signs up, the more support services it sells to Sandicor. If one association offered better support services — • longer hours, a nicer building, more patient help staff — it could attract more subscribers.

The associations are also now actual competitors. Before 1994, rules of the National Association of Realtors required an agent to join the association in the geographical area where the agent’s office was located. But that year NAR adopted the “board of choice” system, which allows agents to choose their association. Under board of choice, the associations compete for members.²⁰ SDAR, for example, sent letters to other associations’ members offering free trials. Another , association considered using flyers and a presentation to retain or attract members in response to board of choice. Agents could consider an association’s MLS support services in deciding whether to become a member; at a minimum, those services are part of the total package that an association offers.²¹

Defendants sabotage their theory by their own admissions. They concede they fixed support fees in part because SDAR “would undoubtedly have been able to offer different prices to MLS users than would Fallbrook” under the board of choice regime, whose arrival was imminent when they formed Sandicor. They explain: “The board-of-choice prospect buttressed the attractiveness of a centralized MLS structure, working hand-in-glove with the underlying notion of fairness of offering the same services at the same prices to all participants.” In other words: “The prospect of having to compete with one another buttressed the attractiveness of a cartel, where we could fix prices and services in ways we thought were fair.” Rarely do antitrust defendants serve up their own heads on so shiny a silver platter.

The associations are not a single entity, and so their joint venture is not immune from scrutiny under section l.²²

3. Defendants argue that, if we don’t find them categorically immune from section 1, we should evaluate their conduct under the more forgiving “rule of reason” rather than the per se rule. Under the rule of reason, we look to the particular facts of the case to determine whether a challenged restraint is likely to enhance or harm competition. See Nat’l Soc’y of Prof'l Eng’rs v. United States, 435 U.S. 679, 690, 98 S.Ct. 1355, 55 L.Ed.2d 637 (1978). Although price fixing is almost always a per se violation of section 1, there are “very narrow” exceptions. United States v. A. Lanoy Alston, D.M.D., P.C., 974 F.2d 1206, 1209 (9th Cir.1992). In Broadcast Music, Inc. v. CBS, 441 U.S. 1, 99 S.Ct. 1551, 60 L.Ed.2d 1 (1979), and NCAA v. Board of Regents, 468 U.S. 85, 104 S.Ct. 2948, 82 L.Ed.2d 70 (1984), the Court applied the rule of reason to conduct that arguably constituted price fixing. BMI held that a blanket license for recordings wasn’t per se illegal even though it “fixed” a “price” in some literal sense. 441 U.S. at 8-9, 99 S.Ct. 1551. The Court stated that per se rules are inappropriate in novel contexts, id. at 9-10, 99 S.Ct. 1551, and observed that “a necessary consequence of an aggregate license is that its price must be established.” Id. at 21, 99 S.Ct. 1551. NCAA applied the rule of reason to broadcast restrictions on college football games that involved some elements of price fixing. 468 U.S. at 99-100, 104 S.Ct. 2948. It justified this approach on the theory that, in college football, “horizontal restraints on competition are essential if the product is to be available at all.” Id. at 101, 104 S.Ct. 2948.

Defendants offer three reasons why BMI and NCAA apply. They first claim that the inherent cooperative aspects of the MLS as a joint venture warrant more deferential review. They next argue that fixing support fees is necessary to realize the “underlying notion of fairness of offering the same services at the same prices to all participants.” Finally, they contend that Fallbrook and Valley Center’s refusal to join the MLS on a decentralized basis made it necessary to fix support fees if the countywide MLS was to exist at all.

We reject the first argument because any elements of novelty and cooperation in the MLS are irrelevant to whether support fees are fixed or set competitively. Plaintiffs challenge the associations’ support fees, not Sandicor’s MLS fee, and fixed support fees are not a “necessary consequence” of the countywide MLS. BMI, 441 U.S. at 21, 99 S.Ct. 1551. There is no obvious technical reason that the fees must be fixed; support fees before 1992 were set independently even when two associations shared the same database. And defendants do not claim that the efficiencies of the MLS would be impaired if support fees were set competitively. Cf. id. at 20, 99 S.Ct. 1551 (noting that blanket licenses allow consumers to avoid the substantial transaction costs of negotiating licenses on a song-by-song basis); Nat’l Bancard Corp. v. VISA U.S.A., Inc., 779 F.2d 592, 601-02 (11th Cir.1986). The only apparent effect of the fixed support fee is to raise the price that consumers pay.

NCAA is likewise inapplicable. To fall within that case’s protection, the price fixing need not itself be essential, see NCAA, 468 U.S. at 101, 104 S.Ct. 2948, but it must still be reasonably ancillary to the legitimate cooperative aspects of the venture. See Regents of the Univ. of Cal. v. ABC, 747 F.2d 511, 517 (9th Cir.1984) (holding per se treatment likely applicable where “[b]y any account, the purpose and effect of the horizontal restraints imposed ... have little, if any, bearing on the operative rules”); cf. Gen. Leaseways, Inc. v. Nat’l Truck Leasing Ass’n, 744 F.2d 588, 595 (7th Cir.1984) (characterizing NCAA as a case where there was a “plausible connection between the specific restriction and the essential character of the product” and holding that per se treatment was appropriate where “the organic connection between the restraint and the cooperative needs of the enterprise that would allow us to call the restraint a merely ancillary one is missing”). The fixed support fee is not reasonably ancillary to the combined MLS database. Defendants offer no explanation for how it improves the efficiency of the MLS or has any effect at all beyond raising prices. Moreover, the support fee is significant relative to the price of the service as a whole; it represents more than half of what consumers pay.

We likewise reject the argument relying on the “underlying notion of fairness of offering the same services at the same prices to all participants.” The “fairness” of uniform pricing is not a relevant consideration in an antitrust case; consumers are presumed to prefer lower prices to the satisfaction of knowing they paid the same inflated price as everyone else. Nor can the price fixing be justified as necessary to ensure uniform support services. Even assuming arguendo that defendants have a legitimate interest in standardizing support services, fixing the price of those services is not a reasonably tailored means of achieving the goal. Sandicor’s service agreements already specify the services that the associations must provide. Fixing the price of those services has only the most speculative tendency to promote further uniformity.

We turn, finally, to the claim that price fixing was justified to convince the smaller associations to join the countywide MLS. This theory at least attempts to explain why the restraint itself was necessary to the joint venture. Nonetheless, it fails to state a valid defense.

We reject some justifications as a matter of antitrust policy, even though they might show that a particular restraint benefits consumers. Among these are theories that “depend! ] on power over price for their efficacy.” 7 Phillip E. Areeda & Herbert E. Hovenkamp, Antitrust Law ¶ 1504c, at 361 (2d ed.2003). In Professional Engineers, for example, the Court held that competitive bidding could not be banned on the theory that it would tempt engineers to do shoddy work. 435 U.S. at 692-94, 98 S.Ct. 1355. The Court categorically rejected the argument that the “restraint on price competition ultimately inures to the public benefit by preventing the production of inferior work and by insuring ethical behavior.” Id. at 693-94, 98 S.Ct. 1355. It noted that “[t]he logic of this argument rests on the assumption that the agreement will tend to maintain the price level; if it had no such effect, it would not serve its intended purpose.” Id. at 693, 98 S.Ct. 1355. This logic doomed the defense because the Sherman Act presumes that competition, not cartel piicing, best ensures quality products for consum-éis, regardless of any empirical evidence to the contrary.²³

Like the defendant in Professional Engineers, the associations here seek to justify not only fixing prices, but intentionally fixing prices at a supracompetitive level. This defense is very different from the ones raised in BMI and NCAA. Defendants in BMI justified blanket licenses on the ground that negotiating song licenses on an individual basis is impractical. 441 U.S. at 20, 99 S.Ct. 1551. And the NCAA justified its broadcast restrictions at least in part on the need to maintain competitive balance among teams. NCAA, 468 U.S. at 117-20, 104 S.Ct. 2948; see also Gen. Leaseways, 744 F.2d at 595. Neither argument depends on the fact that the fixed price is supra-competitive. Sandicor and the associations, by contrast, fixed prices precisely in order to keep them above competitive levels. Their ultimate purpose may have been to entice the smaller associations to join the countywide MLS, but the fee had this effect only because it was supracompetitive, not because it was fixed. Defendants seek to justify not only the forbidden practice (price fixing) but its forbidden effect (supracompetitive prices), and this they may not do. See Nat’l Soc’y of Prof'l Eng’rs, 435 U.S. at 693, 98 S.Ct. 1355. Finns cannot fix prices as a mere quid pro quo for providing consumers with better products. Antitrust law presumes that competitive markets offer sufficient incentives and resources for innovation, and that cartel pricing leads not to a dedication of newfound wealth to the public good but to complacency and stagnation.²⁴

The presumption that higher prices are bad for consumers is, in any case, amply borne out by the record. The associations’ scheme bore no reasonable relation to its asserted goal. Fallbrook and Valley Center served about 400 subscribers, less than 3% of the San Diego County total. By fixing support fees, Fall-brook and Valley Center together may have made an additional $6000 per month (on top of their cash subsidy of $3000).²⁵ Here’s the catch: The support fee was fixed at $25 not just for those 400 subscribers, but for all 14,000 subscribers in San Diego County. And, other than the trivial $3000 per month cash subsidy, only 3% of the countywide overcharge actually went to Fallbrook and Valley Center. The other 97% went into the coffers of the other associations. In the case of SDAR alone, this could have amounted to more than $1.2 million per year. Defendants are thus trying to justify overcharging consumers millions by the need to provide an implicit subsidy worth less than one-twentieth that amount. If the other associations believed it was essential to involve Fallbrook and Valley Center in the MLS, they could have allowed price competition for support services and increased the subsidies to these two associations. Instead, they adopted an anticompetitive scheme that was completely out of proportion to its asserted purpose.²⁶

Moreover, defendants assume that, without Fallbrook and Valley Center, there couldn’t be a true countywide MLS. But Sandicor could have formed without them and then competed for their 400 subscribers. An MLS’s listings come from its subscribers, so its coverage is simply a function of who subscribes. Other than NAR’s now-defunct geographical restrictions, nothing prevented subscribers in Fallbrook or Valley Center from choosing Sandicor’s MLS over their local service.²⁷ The competitive option at a minimum was not so plainly unworkable that price fixing was necessary as a matter of law.²⁸

Stripped to its essentials, defendants’ argument is that some of the firms they wanted to include in the joint venture were so inefficient that they could survive only under cartel pricing. Defendants’ concern for the weakest among them has a quaint Rawlsian charm to it, but we find it hard to square with the competitive philosophy of our antitrust laws. Inefficiency is precisely what the market aims to weed out. The Sherman Act, to put it bluntly, contemplates some roadkill on the turnpike to Efficiencyville.

We find all the asserted defenses legally deficient or factually unsupported. The record compels the conclusion that defendants violated section 1 of the Sherman Act by fixing support fees. The district court should have denied summary judgment to defendants and granted it to plaintiffs instead.

Conspiracy To Monopolize

. Plaintiffs also allege a conspiracy to monopolize the market for support services. To prevail, they must show “specific intent to monopolize and anticompeti-tive acts designed to effect that intent.” Hunt-Wesson Foods, Inc. v. Ragu Foods, Inc., 627 F.2d 919, 926 (9th Cir.1980). “[N]o particular level of market power or ‘dangerous probability of success’ has to be alleged or proved ... where the specific intent to monopolize is otherwise apparent from the character of the actions taken.” Id.

Plaintiffs’ section 2 case centers around restrictive provisions in Sandicor’s shareholders’ agreement and its rejection of Freeman’s offer to run a competing service center. The shareholders’ agreement reads as follows:

¶ 4.2. To pass a Major Corporate Resolution, ... the holders of not less than two-thirds of the outstanding Shares cast by not less than two Shareholders must vote ... in favor of the resolution. ¶ 4.3. A Major Corporate Resolution shall be defined as follows:

(2) Any decision that substantially changes the corporate structure, including but not limited to, the admission of new Shareholders, or entering into an agreement with any other Board of Realtors or Shareholder.

Freeman proposed that Sandicor sell MLS database access to her directly at a rate that excluded the support fee, and that she resell it to agents with her own support services. Essentially, she sought to run a service center on a “decentralized” basis. Sandicor refused, explaining that her proposal was incompatible with Sandicor’s centralized structure. Plaintiffs attribute this rejection to the shareholders’ agreement. They draw a parallel to Associated Press v. United States, 326 U.S. 1, 65 S.Ct. 1416, 89 L.Ed. 2013 (1945), where the Supreme Court invalidated provisions in AP’s bylaws requiring a majority vote among existing newspapers to grant membership to a competitor. Id. at 12, 65 S.Ct. 1416.

The district court rejected the section 2 claim in part because Freeman had no bona fide intent to operate a competing service center. Even accepting this to be the case, we conclude that plaintiffs have standing to sue. That Freeman had no bona fide competitive intent might undermine her standing as a competitor, but she is also an indirect purchaser of support services. If the associations exclude competitors, she suffers inflated prices, and she has standing to sue as an aggrieved consumer. See In re Ins. Antitrust Litig., 938 F.2d 919, 926 (9th Cir.1991); Thompson v. Metro. Multi-List, Inc., 934 F.2d 1566, 1571 (11th Cir.1991).

Nonetheless, the shareholders’ agreement does not explain Sandicor’s rejection of Freeman’s offer. It applies. only to “admission of new Shareholders, or entering into an agreement with any other Board of Realtors or Shareholder.” Freeman was not asking to become a shareholder of Sandicor, only a service center operator, and we see no reason why the latter status compels the former. Moreover, although she proposed an agreement with Sandicor, she is not a Board of Realtors or a shareholder, so she is not covered by the second half of the provision either. The supermajority requirement didn’t apply, so it can’t have been the reason Sandicor rejected her offer.

This doesn’t end the section 2 inquiry, however. Even though there was no causal relationship between the shareholders’ agreement and the rejection, they each might still be independent “anticompetitive acts” sufficient to support a section 2 conspiracy claim. See Hunt-Wesson Foods, 627 F.2d at 926 (“[T]he [anticompetitive] act may be no more than the agreement itself.”).²⁹ We must determine whether the evidence supports a finding that either act was driven by a specific intent to monopolize.

We find no genuine issue as to the adoption of the shareholders’ agreement. Although there are some parallels between this case and Associated Press, the distinctions are significant. First, the agreement in Associated Press distinguished between new members who would compete with existing members and new members who would not. 326 U.S. at 10-11, 65 S.Ct. 1416. Sandicor’s shareholders’ agreement makes no comparable distinction. The absence of facial discrimination against competitors makes the inference of anticompetitive intent much weaker. Second, as noted above, the agreement applies only to new service agreements with other Boards of Realtors or shareholders. It doesn’t apply to agreements with people like Freeman. This limitation on the agreement’s scope further attenuates the inference of monopolistic intent.

Unlike the bylaws in Associated Press, see id. at 6, 65 S.Ct. 1416, the restrictive provisions here are not so anticompetitive on their face that they can be condemned without further evidence that they were adopted with specific intent to monopolize. Cf. Northwest Wholesale Stationers, Inc. v. Pac. Stationery & Printing Co., 472 U.S. 284, 298, 105 S.Ct. 2613, 86 L.Ed.2d 202 (1985) (“[N]ot all ... refusals to deal are predominantly anticompetitive.”). The associations may have adopted them for entirely benign reasons — for example, to ensure that new shareholders would provide support services effectively or that they would not compromise the integrity of the database. Absent additional evidence that the restrictive provisions were adopted for improper reasons, they don’t support a conspiracy to monopolize claim.³⁰ No such evidence exists in the record, so plaintiffs have failed to raise a genuine issue that the agreement violates section 2.

Finally, we consider whether the rejection of Freeman’s proposal violated section 2. Defendants may not “refus[e] to deal in order to create or maintain a monopoly absent a legitimate business justification.” Image Technical Servs., Inc. v. Eastman Kodak Co., 125 F.3d 1195, 1209 (9th Cir.1997); see also Aspen Skiing Co. v. Aspen Highlands Skiing Corp., 472 U.S. 585, 601, 105 S.Ct. 2847, 86 L.Ed.2d 467 (1985). We agree with the district court that there is insufficient evidence that “Sandicor’s purpose in precluding Plaintiff Freeman from operating a service center was to stifle competition.” Sandicor’s professed reason — that Freeman’s proposal was not consistent with its current business model — is plausible. Freeman wanted to operate a service center on a decentralized basis, but Sandicor doesn’t operate that way; it sells MLS subscriptions directly to agents and then reimburses its service centers for the support services they provide. It’s true that one prominent element of Sandicor’s business model is illegal price fixing, but that doesn’t mean Freeman can demand Sandicor change the way it does business to suit her wishes.³¹ Plaintiffs needed evidence that the stated reason for Sandicor’s refusal was not the actual reason, and they failed to produce it. The only evidence they point to is the shareholders’ agreement, but it couldn’t have affected Sandicor’s decision because it didn’t apply to what Freeman wanted to do.

Plaintiffs have failed to raise a genuine issue that either the execution of the shareholders’ agreement or the rejection of Freeman’s proposal was driven by a specific intent to monopolize. The district court correctly granted summary judgment in favor of defendants on the conspiracy to monopolize claim.

Claims Against the California Association of Realtors

The California Association of Realtors isn’t a party to any of the offending agreements, but plaintiffs allege that CAR lawyers encouraged the associations’ antitrust violations. Plaintiffs have failed to turn up any evidence to support this theory. There is some evidence suggesting that CAR encouraged a corporate form for Sandicor, but this doesn’t show that CAR encouraged the associations to fix support fees. A CAR attorney did opine that fixed support fees were legal. But nothing indicates that she recommended the arrangement, and dispassionate legal advice is not an antitrust violation. See Tillamook Cheese & Dairy Ass’n v. Tillamook County Creamery Ass’n, 358 F.2d 115, 118 (9th Cir.1966). Finally, plaintiffs argue that CAR is violating antitrust laws by financing the defense of this lawsuit. CAR’s financial support certainly explains plaintiffs’ desire to sue it, but it’s hardly illegal.

Discovery Sanctions

Shortly after discovery closed, Sandicor’s president Bill Stegall, who had been diagnosed with cancer, resigned and revealed to plaintiffs’ counsel that documents had been withheld. Plaintiffs moved to compel discovery of the documents and asked for sanctions. A magistrate judge found that the failure to produce the documents was willful, and awarded plaintiffs their costs of filing the motion under Fed.R.Civ.P. 37(a)(4). The district court approved the award, and we review it only for abuse of discretion. Rio Props., Inc. v. Rio Int’l Interlink, 284 F.3d 1007, 1022 (9th Cir.2002).³²

Relying on Badalamenti v. Dunham’s, Inc., 896 F.2d 1359 (Fed.Cir.1990), defendants contend that the court abused its discretion because they had objected to the discovery requests as ambiguous, and plaintiffs never “tested” those objections. But plaintiffs’ motion to compel discovery is the only testing that Rule 37(a)(4) requires. The magistrate didn’t find the objections substantially justified, so plaintiffs recover the costs of bringing the motion. Fed.R.Civ.P. 37(a)(4); Rockwell Int’l, Inc. v. Pos-A-Traction Indus., Inc., 712 F.2d 1324, 1325-26 (9th Cir.1983).

Conclusion

Antitrust law doesn’t frown on all joint ventures among competitors — far from it. If a joint venture benefits consumers and doesn’t violate any applicable per se rules, it will often be perfectly legal. The decision to combine MLS databases fits comfortably within this category. See Realty Multi-List, 629 F.2d at 1367-68 (extolling the many virtues of MLSs). The further decision to fix support fees does not.

We AFFIRM the district court’s holding that defendants’ activities substantially affected interstate commerce. We also AFFIRM summary judgment in favor of defendants on the section 2 claim and summary judgment in favor of CAR on all claims. Finally, we AFFIRM the court’s award of discovery sanctions.

We REVERSE, however, on the merits of the principal section 1 claim. Plaintiffs’ evidence is sufficient to prove a violation of section 1 by Sandicor and the associations. Defendants have presented no evidence that refutes the legal essentials of plaintiffs’ case. The district court should have denied summary judgment to defendants and awarded it to plaintiffs instead.

We remand to the district court for further proceedings consistent with this opinion.

AFFIRMED in part, REVERSED in part and REMANDED. Costs to appellants, except that appellee California Association of Realtors shall recover its costs from appellants.

1 . The facts we describe are uncontested unless otherwise noted.

2 . State law and industry practice distinguish between “brokers” and “salespersons.” We use the term "agents” throughout as a generic placeholder for real estate professionals who are typical MLS subscribers.

3 .The other association was unaffiliated and didn’t participate in any of the arrangements at issue in this case.

4 . Although Sandicor has experimented with different pricing policies, at most relevant times it charged fees based on the number of individuals who used the MLS. Thus, a real estate office with four agents using the MLS would pay four subscription fees.

5 . The fee has varied somewhat over the years but remains at approximately the same level.

6 . Ten dollars is the figure quoted at the 1991 meeting; plaintiffs claim that SDAR's cost of providing support services is actually much less. It is unclear from the briefing and record whether the $10 figure also included SDAR's share of the MLS database costs.

7 . Plaintiffs' briefs argue both monopolization and conspiracy to monopolize theories, but the Second Amended Complaint alleges only the conspiracy claim in the market for support services. Second Amended Complaint ¶¶ 158-160. The complaint also alleges monopolization and conspiracy to monopolize in the MLS market itself, id. ¶¶ 161-168, but plaintiffs abandoned arguments related to that market on appeal.

8 . Plaintiffs had earlier sued in California state court for violations of state antitrust law. The trial court dismissed the suit on the pleadings, and the court of appeal affirmed. Freeman v. San Diego Ass’n of Realtors, 77 Cal.App.4th 171, 91 Cal.Rptr.2d 534 (1999). Defendants rely on the state decision as persuasive authority but do not contend that it has preclu-sive effects on plaintiffs' federal claims. The state suit has no res judicata effect because federal courts have exclusive jurisdiction and California doesn’t apply res judicata to claims the first forum was incompetent to hear. See Marrese v. Am. Acad. of Orthopaedic Surgeons, 470 U.S. 373, 379, 382, 105 S.Ct. 1327, 84 L.Ed.2d 274 (1985); Pension Trust Fund for Operating Eng'rs v. Triple A Mach. Shop, Inc., 942 F.2d 1457, 1460 (9th Cir.1991). The suit also has no relevant issue preclusive effects here because the pertinent issues are not "identical.” Lucido v. Superior Court, 51 Cal.3d 335, 341, 272 Cal.Rptr. 767, 795 P.2d 1223 (1990). The state suit challenged different conduct' — namely, an alleged conspiracy to fix Sandicor's MLS fee. The state court's single-entity analysis is thus not preclusive, because it was based in part on the nature of the price-fixing allegation. See Freeman, 77 Cal.App.4th at 192-94 & n. 21, 91 Cal.Rptr.2d 534. Moreover, because the state suit was dismissed on the pleadings, we doubt that it has any issue preclusive effects in this case at all. The purely legal issue of the sufficiency of the claims under California law is not identical to the federal issues here. See id. at 183 n. 9, 91 Cal.Rptr.2d 534 (noting that the Cartwright and Sherman Acts are "not coextensive”). The state suit is, of course, res judicata as to plaintiffs' pendent state law claims. The district court dismissed those claims under 28 U.S.C. § 1367, and plaintiffs haven’t asked us to revive them.

9 . The only remaining individual defendant was Anita Alkire, who filed for bankruptcy during the proceedings. Plaintiffs have abandoned their claims against her.

10 . One manual captures the principle nicely in question and answer format: "[Q.] May competitors agree to fix prices? [A.] Duh. What do you think?” Eliot G. Disner, Antitrust Law for Business Lawyers § 4.06, at 82 (2001).

11 . A firm will normally pass on some portion of its cost-per-unit savings to consumers even if it is a profit-maximizing monopolist. This has nothing to do with altruism; a monopolist just makes more money by reducing its price in response to a cost decline and thus selling to more consumers. See Richard A. Posner, Antitrust Law: An Economic Perspective 248 (1976) ("[I]f the monopolist’s marginal cost declines (other things being the same), he will reduce his price....”). The theoretical economics of this case are more nuanced because the associations are shareholders of Sandicor. They could conceivably induce Sandicor to set MLS fees at a level that maximizes shareholder profits (dividends plus support fees) rather than its own. The level of support fees might then be less relevant (except perhaps as a device to make the MLS look less profitable than it really is). The record contains no evidence that Sandicor ever set MLS prices in this fashion, however. Instead, it indicates that Sandicor set prices based on costs.

12 . Defendants argue that an indirect purchaser who proves it can sue under Illinois Brick because of a unity of interest necessarily proves that it can't sue under Coppei-weld because of the very same unity of interest. This argument is clever but unpersuasive. Copper-weld looks at whether the defendants are commonly owned, not at whether they commonly own some other entity. Royal Printing, on the other hand, doesn't care who owns whom in the distribution chain. Thus, Cop-perweld doesn't apply here (as we will see) because the associations are independently owned, but Royal Printing applies because the associations own Sandicor.

13 . These figures reflect 1999 MLS and support fees.

14 .The district court rejected the section 1 claim in part because the associations’ support services and MLS database access were not "separate and distinct products or services” with "separate markets.” It relied heavily on tying concepts. Distinct product markets are crucial to a tying claim, but they are largely irrelevant to a price-fixing claim. Many forms of price fixing, for example, involve retail services that are obviously inseparable from the product itself. See, e.g., Cal. Retail Liquor Dealers Ass’n v. Midcal Aluminum, Inc., 445 U.S. 97, 102-03, 100 S.Ct. 937, 63 L.Ed.2d 233 (1980); Plymouth Dealers, 279 F.2d at 134. That the retail services can legally be tied to the product doesn’t mean that retailers can agree to fix prices.

15 . A refund of part of Sandicor's MLS fee from an association to an agent is equivalent to a reduction in the support fee that Sandi-cor pays the association coupled with a reduction in the MLS fee that the agent pays Sandicor.

16 . Plaintiffs claim several other instances of price fixing. Sandicor’s board of directors, which is made up of association representatives, fixed the associations' retail prices on books, "tour inputs,” computer assistance by association staff, and lockboxes and replacement keypads. Defendants question plaintiffs’ standing to challenge these practices, observing that they did not specifically allege having purchased these products and services. We leave the issue for the district court to resolve on remand. Plaintiffs further challenge Sandicor’s policy of mandating "uniform” support services and a standardized splash screen. These are nonprice restraints subject to rule-of-reason analysis. See Cont’l T.V., 433 U.S. at 59, 97 S.Ct. 2549. We leave them for the district court to consider in the first instance as well.

17 . Los Angeles Memorial Coliseum was decided before Copperweld, but nothing in the latter impugns our holding in the former.

18 . We may resolve the issue of capacity to conspire because the relevant facts are not disputed. See L.A. Mem’l Coliseum, 726 F.2d at 1387.

19 . We assume this to be the case, although we do note that each association’s ownership interest in Sandicor is automatically adjusted every year to reflect the relative number of subscribers it serves, so that the associations have eliminated many of the elements of risk-sharing traditionally associated with the corporate form.

20 . The district court held that "[c]ompetition for members is outside the scope of the Associations’ joint venture.” But MLS subscriptions are one of the services associations sell to their respective members. An association that attracts more members thereby attracts more subscribers, and thus earns more support fees from Sandicor.

21 . The associations may have been actual competitors even before board of choice, because under Palsson, 16 Cal.3d at 937-38, 130 Cal.Rptr. 1, 549 P.2d 833, they had to sell MLS subscriptions to nonmembers, who were not subject to NAR's geographical restrictions.

22 .We note that our colleagues on the First Circuit recently disdained the Copperweld defense in a monogrammatically similar case. See Fraser v. Major League Soccer, L.L.C., 284 F.3d 47, 59 (1st Cir.2002) ("The case for expanding Copperweld is debatable and, more so, the case for applying the single entity label to [an] MLS.”).

23 . Professional Engineers was a rule of reason case, but the principle applies with even more force in the per se context.

24 . It does not matter that Fallbrook and Valley Center would have operated at a loss in a competitive environment. Their precarious financial situation may have explained their intransigence, but it does not transform it into a viable defense. If there is any argument the Sherman Act indisputably forecloses, it is that price fixing is necessary to save companies from losses they would suffer in a competitive market. See United States v. Socony-Vacuum Oil Co., 310 U.S. 150, 221, 60 S.Ct. 811, 84 L.Ed. 1129 (1940). While a competitor who fixes prices to stem his losses may be a more sympathetic character than one who does so to fatten his purse, he enjoys no favored legal position. Defendants' argument is more complex than the traditional cry of "ruinous competition," but it is cut from the same cloth.

25 . The $6000 figure is based on an assumed competitive price of $10, SDAR's stated costs in 1991. This is concededly a ballpark estimate, but the exact number is irrelevant to the point made.

26 . We also note that Fallbrook and Valley Center merged into larger associations nearly a decade ago. Even if defendants’ argument excused price fixing at the outset, we are at a loss to see how it justifies the persistence of the regime, given that the purported necessity no longer exists.

27 . We don’t mean to suggest that NAR’s restrictions would have been a valid defense while they existed. As essentially self-imposed limitations quite likely anticompetitive in their own right, they are hardly a defense to per se liability.

28 . Plenty of market defect theories do try to explain why consumers might continue using a product even after a better one becomes available. Switching costs, information asymmetries, and the defect du jour, the dreaded network externality, see United States v. Microsoft Corp., 253 F.3d 34, 49-50 (D.C.Cir.2001); Michael L. Katz & Carl Shapiro, Network Externalities, Competition, and Compatibility, 75 Am. Econ. Rev. 424 (1985), all come to mind. But how are we to know, for example, whether Fallbrook and Valley Center's installed base would give them an insuperable advantage over a more efficient competitor? There are plenty of arrows in the competitive quiver that a firm might use to dislodge an incumbent: aggressive marketing, pen-

etration pricing, free trials, money-back guarantees — are these techniques so obviously ineffective that a firm has no choice but to violate the law? Network externalities generate substantial dissent even within the economics community, and not only over how they should shape antitrust analysis, Microsoft, 253 F.3d at 49-50, but even over whether they exist at all, see, e.g., S.J. Liebowitz & Stephen E. Margolis, Network Externality: An Uncommon Tragedy, J. Econ. Persp., Spring 1994, at 133. We need not decide the point, but they seem a weak basis for an exception to a per se rule.

29 . The signing of the original shareholders' agreement antedates the limitations period, but the agreement has been re-executed several times since.

30 . That the agreement itself is not anticom-petitive doesn’t mean that it could never be anticompetitively applied. Even if the voting mechanism is permissible on its face, voting against new shareholders for anticompetitive reasons may still be a violation of section 2.

31 . Although decentralizing Sandicor would be one way to remedy the price-fixing violation, it is not the only way, and Freeman had no right to insist on it. Whether the district court may now order Sandicor to decentralize as a matter of equitable relief is a different story, and an issue we need not decide.

32 . The magistrate also awarded plaintiffs the costs of further discovery. That sanction hasn’t yet been approved by the district court and isn’t before us.