Stop & Shop Supermarket Co. v. Blue Cross & Blue Shield of Rhode Island

*59 BOUDIN, Chief Judge. Stop & Shop Supermarket Company (“Stop & Shop”) and Walgreen Eastern Co., Inc., (“Walgreen”) brought an antitrust suit against a number of defendants primarily based on section 1 of the Sherman Act, 15 U.S.C. § 1 (2000), lost certain of their claims on summary judgment, and then suffered a directed verdict at the jury trial held on the balance of their claims. They now appeal on several grounds as to certain defendants (the other defendants settled). We begin with a description of background events and proceedings in the district court. Blue Cross and Blue Shield of Rhode Island (“Blue Cross”) is the major health insurer in that state, offering various plans that cover, among other medical expenses, the cost of prescription drugs. Until 1997, Blue Cross managed drug benefits itself and provided a substantially “open” pharmacy system — that is to say, most subscribers could buy drugs at any pharmacy. Blue Cross determined what drugs it would reimburse, set (by negotiation) what would be paid to the pharmacies, and processed subscriber claims. Beginning in 1997, Blue Cross decided to use a pharmacy benefits manager to administer its prescription drug benefits. Such managers often set up a, “closed” network of pharmacies, providing greater insurance coverage to those subscribers who use network pharmacies. In exchange for inclusion in the network, and therefore increased volume of drug sales, the network pharmacies typically agree to provide drugs at lower prices, resulting in lower costs to the insurer. In this case, Blue Cross invited bids from managers and received three, one of which Blue Cross disqualified. The second manager was PharmaCare, a subsidiary of CVS, a well known major drug store chain (52 pharmacies in Rhode Island). The third manager, Wellpoint, proposed a closed network limited to pharmacies operated by Stop & Shop (18 pharmacies) and Walgreen (15 pharmacies). After obtaining further bids from Wellpoint and Phar-maCare, in December 1997 Blue Cross selected PharmaCare to manage a closed network that initially included the CVS pharmacies and most independent pharmacies in Rhode Island. During this period, PharmaCare was itself negotiating with yet another benefit manager — Provider Health Services, Inc. (“Provider”). Provider managed a closed network, comprised mainly of Brooks Pharmacies (42 pharmacies in the state), serving another insurer — United Healthcare of New England, Inc. (“United”)— doing business in Rhode Island. In February 1998, Provider agreed to allow CVS stores to join the United/Provider network; and in May 1998, PharmaCare allowed Brooks and other Provider pharmacies to join the Blue Cross/PharmaCare closed network. Ancillary to these arrangements, Brooks and Provider’s other pharmacies agreed— obviously for PharmaCare’s benefit — not to join other networks competing for Blue Cross’ business. PharmaCare in turn agreed not to admit into the Blue Cross/PharmaCare network new drug stores (beyond CVS, the independents, and the pharmacies in the United/Provider network). Blue Cross consented to the enlargement of its closed network and in November 1998 signed a formal three-year contract with PharmaCare. Not all Blue Cross customers are covered by plans that effectively restrict them to closed network pharmacies. Blue Cross offers multiple plans, and one set allows customers to fill prescriptions at any pharmacy without economic penalty. Blue Cross’ counsel estimated in oral argument *60 that perhaps two-thirds of Blue Cross’ customers are restricted to its closed network; our own review of the record suggests that the number — which obviously varies over time — may be closer to three-quarters. Unhappy with losing the opportunity to serve many Blue Cross customers on competitive terms, Stop & Shop brought the present action on June 9, 1999, against Blue Cross, PharmaCare and CVS, charging violations of federal and state antitrust laws; in March 2000, Walgreen joined as co-plaintiff and an amended complaint was filed on May 2, 2000. United Health, Provider, and Brooks were also initially defendants but they were dropped after agreeing to admit plaintiffs to the United/Provider network. On motions for summary judgment, the district court wrote a detailed opinion rejecting plaintiffs’ claims that any per se violation of the antitrust laws had been plausibly shown. Stop & Shop Supermarket Co. v. Blue Cross & Blue Shield of R.I., 239 F.Supp.2d 180, 195 (D.R.I.2003). It rejected claims that any of the arrangements comprised naked horizontal restraints such as a group boycott. Id. at 189-91. However, the court ruled that factual issues precluding summary judgment were raised as to whether any of defendants’ conduct was actionable under the rule of reason. Id. at 193. At this time, the district court said that Blue Cross represented about 60 percent of the customers in Rhode Island whose retail drug purchases were reimbursed; and United provided such benefits to about 25 percent. Stop & Shop, 239 F.Supp.2d at 183. These figures were the lynchpin of the pretrial report by plaintiffs’ expert witness, Dr. Bruce Stangle, who stressed the 85 percent figure in concluding that “an out-of-network entrant would be precluded from competing in a substantial portion of the relevant market.” Thereafter the district court considered in limine motions filed by Blue Cross. Subject to reconsideration at trial, the court granted a motion to exclude certain evidence as to the bidding process leading to PharmaCare’s selection and the related decisions of PharmaCare and Provider to expand their respective closed networks to include each other’s present pharmacy members. The court denied, again subject to reconsideration at trial, a motion to exclude key testimony from plaintiffs’ expert witness, Dr. Stangle. Trial began in June 2003. On the sixth day of trial, the district court upheld a defense objection to Dr. Stangle’s proposed testimony concerning the proper definition of the relevant market. Plaintiffs tendered a summary of the proposed testimony as an offer of proof and rested. The district court then granted a defense motion for judgment as a matter of law, holding (in an oral ruling from the bench) that absent an adequate market definition, the plaintiffs could not make out a rule of reason claim under the antitrust laws. The plaintiffs now appeal, arguing that the district court erred in granting partial summary judgment on the per se claims, in excluding several items of evidence including Dr. Stangle’s testimony on market definition, and in directing a verdict. Rulings on summary judgment and directed verdicts are reviewed de novo, Wolinetz v. Berkshire Life Ins. Co., 361 F.3d 44, 47 (1st Cir.2004) (summary judgment); Ahern v. Scholz, 85 F.3d 774, 793 (1st Cir.1996) (directed verdict); the standard for review of rulings excluding evidence depends on the nature of the underlying issue (fact, law, judgment call), see Blake v. Pellegrino, 329 F.3d 43, 46 (1st Cir.2003). *61 Per se claims. Plaintiffs begin their brief by contesting the district court’s summary judgment ruling that no legitimate per se claims were presented. As our prior decisions have explained, antitrust claims under section 1 of the Sherman Act ordinarily require a burdensome multi-part showing: that the alleged agreement involved the exercise of power in a relevant economic market, that this exercise had anti-competitive consequences, and that those detriments outweighed efficiencies or other economic benefits. This is the so-called rule of reason calculus. See, e.g., Eastern Food Servs. v. Pontifical Catholic Univ. Servs. Ass’n, 357 F.3d 1, 5 (1st Cir.2004); Fraser v. Major League Soccer, L.L.C., 284 F.3d 47, 59 (1st Cir.2002), cert. denied, 537 U.S. 885, 123 S.Ct. 118, 154 L.Ed.2d 144 (2002). This calculus is bypassed if the collusive arrangement falls instead within one of several categories {e.g., naked horizontal price fixing) in which liability attaches without need for proof of power, intent or impact. Eastern Food Servs., 357 F.3d at 4 & n. 1. For that reason plaintiffs typically try to bring their claims within per se rubrics. Whether a plaintiffs alleged facts comprise a per se claim is normally a question of legal characterization that can often be resolved by the judge on a motion to dismiss or for summary judgment. See, e.g., Addamax Corp. v. Open Software Found., Inc., 152 F.3d 48, 50-51 (1st Cir.1998). The most important per ^se categories are naked horizontal price-fixing, market allocation, and output restrictions. Sometimes group boycotts are called per se violations, but the label here is only minimally useful since many arrangements that are literally concerted refusals to deal have potential efficiencies and are judged under the rule of reason. See U.S. Healthcare, Inc. v. Healthsource, Inc., 986 F.2d 589, 593 (1st Cir.1993). Minimum retail price fixing is a rare vertical arrangement still comprising a per se violation — that is why car makers only “suggest” a retail price to dealers — but this offense is not charged by plaintiffs in this case. 1 Because the defendants moved for summary judgment, the complaint allegations did not have to be taken as true, see R.W. Intern. Corp. v. Welch Food, Inc., 13 F.3d 478, 487 (1st Cir.1994), but the plaintiffs were entitled to the benefit of the doubt: specifically, reasonable inferences were to be drawn in their favor and genuine factual disputes were properly reserved for trial so far as plaintiffs’ sworn version of the facts conflicted with the defendants’ sworn version. See Fed.R.Civ.P. 56(c); Douglas v. York County, 360 F.3d 286, 290 (1st Cir.2004). However, broadly speaking what happened in this case is largely undisputed, although some of the details are obscure. In a nutshell, the arrangements that concern plaintiffs were two. First, Blue Cross contracted for a closed network for its subscribers that excluded the plaintiffs after a bidding contest that plaintiffs say was flawed and that they should have won. Second, Blue Cross and its pharmacy benefits manager agreed with United and its manager that their respective network pharmacies would be admitted to each other’s closed networks; these arrangements included exclusionary restrictions and (plaintiffs suggest) other more sinister collaboration as to price. We start with the creation of Blue Cross’ closed network. The alleged unfair *62 bidding aside, this is nothing other than an exclusive dealing arrangement, slightly complicated by the involvement of three or four sets of parties rather than the usual two. Blue Cross, either directly or indirectly, is buying prescription drugs for its subscribers. See Kartell v. Blue Shield of Mass., Inc., 749 F.2d 922, 924-25 (1st Cir.1984), ce rt. denied, 471 U.S. 1029, 105 S.Ct. 2040, 2049, 85 L.Ed.2d 322 (1985). Blue Cross’ closed network effectively gives certain drug stores the exclusive right to supply such drugs to most of its customers. This certainly inconveniences Blue Cross subscribers for whom more outlets are a benefit. But, if Blue Cross is a competent negotiator, the closed network should lower the cost to Blue Cross of supplying drugs to customers (because most suppliers will cut prices in exchange for increased volume, cf. U.S. Healthcare, 986 F.2d at 591). Blue Cross, might pass the savings on to customers (lower premiums, smaller co-payments, broader coverage) or keep the savings itself and pay its executives more (if competition among health insurers is inadequate and state regulation absent). Either way, the closed network is simply an exclusive dealing arrangement which is not a per se violation of the antitrust laws. See Tampa Elec. Co. v. Nashville Coal Co., 365 U.S. 320, 327-29, 81 S.Ct. 623, 5 L.Ed.2d 580 (1961); Eastern Food Servs., 357 F.3d at 8. The arrangement might still be unlawful under the rule of reason depending upon the particular circumstances — that is, depending upon the balance between efficiencies gained and any harm to competition that could be shown, id. at 5, but we are concerned for the moment only with whether per se treatment was warranted. It was not. Nothing is changed by plaintiffs’ claim that the bidding was unfair. “Bid rigging” of a certain kind is a per se violation of the antitrust laws, e.g., JTC Petroleum Co. v. Piasa Motor Fuels, Inc., 190 F.3d 775, 777 (7th Cir.1999) (Posner, C.J.); but this refers to horizontal price fixing whereby two or more suppliers (or occasionally purchasers) secretly fix the price at which they will bid. See id. When Blue Cross, through its benefits manager, gave exclusive rights to CVS and certain other pharmacies, it was not bidding at all; it was inviting bids and making its own decision as to which bid to accept. Ordinarily, it would be in Blue Cross’ interest to accept the lowest cost bid, assuming services were equivalent. Plaintiffs say that Blue Cross manipulated the bidding {e.g., by giving CVS information about Wellpoint’s bid), which conceivably could happen if Blue Cross were corrupt or incompetent (an alternative, benign reason would be to press PharmaCare to improve its offer). But the antitrust laws are not meant to police bad management; the market (or the insurance regulator) is expected to do that. This brings us to the first part of the second transaction, namely, the agreements that let the Blue Cross and United pharmacies serve in each other’s closed networks. These are undoubtedly horizontal agreements, Blue Cross and United being competitors (as are PharmaCare and Provider), and so draw closer scrutiny. But on their face, they are not exclusionary or otherwise anti-competitive: they allow more pharmacies to compete for the same consumer’s business (e.g., Brooks can supply Blue Cross customers) and give customers more options. The main anti-competitive threat, to the extent it exists, lies not in admission of new pharmacies but in ancillary provisions that might exclude others by agreement. It appears that the United/Provider network remained free to' admit other phar *63 macies: it did in fact admit the plaintiffs as part of an agreement to settle this lawsuit. But the district court says that the contracts precluded the Blue Cross/PharmaCare network from admitting any other new pharmacies (beyond the United/Provider pharmacies), Stop & Shop, 239 F.Supp.2d at 184, so the contracts did for some period restrict Blue Cross/PharmaCare from admitting plaintiffs. Like the original exclusive dealing contract, this is a possible antitrust violation, but it is not a per se violation. The reason is that the closed pharmacy arrangement is valuable to participating pharmacies in part because it directs volume to them; thus, the United/Provider pharmacies had a direct interest, in exchange for allowing CVS to compete for their captive subscribers, in not only being allowed to compete for Blue Cross’ customers but in making sure that yet additional new member pharmacies did not unreasonably dilute this benefit. This does not mean that the ancillary restriction is lawful but only that per se condemnation is not appropriate. Joint ventures involving direct competitors not infrequently exclude other competitors. Cf. N.W. Wholesale Stationers, Inc. v. Pac. Stationery & Printing Co., 472 U.S. 284, 296-97, 105 S.Ct. 2613, 86 L.Ed.2d 202 (1985). Imagine a research and development consortium between a dozen small manufacturers that by agreement excluded any new entrants; the arrangement might enhance competition with larger manufacturers, and yet the original members might be unwilling to commit resources to the venture if other competitors — even small ones — -were able to enter at will and share in the inventions. See XIII Hovenkamp, Antitrust Law ¶ 2115a (1999). A further explicit provision of the reciprocal expansions barred the United/Provider pharmacies from participating in other networks competing for Blue Cross’ business. Stop & Shop, 239 F.Supp.2d at 184. This, too, is not a per se violation. By admitting the Brooks pharmacies into its own closed network, Blue Cross and Phar-maCare were in effect including them in a joint venture. There are sometimes legitimate reasons why one party to a joint venture can insist as the price of entry that a new member limit its existing competitive freedom. XIII Hovenkamp ¶ 2213. Here, PharmaCare was creating and administering a network for Blue Cross and, in the course of doing so, it would be providing favored access to network pharmacies, bolstering their connection with subscribers, and conceivably giving them information valuable in their servicing of the customers. PharmaCare might legitimately be unwilling to expand the network to include pharmacies who were at the same time preparing to join a new consortium to replace PharmaCare as Blue Cross’ manager. Despite plaintiffs’ looser description, the restraint does not prevent Brooks pharmacies from joining other networks but only from involving themselves in attempts to supplant PharmaCare with Blue Cross. The restraint might still be unjustified— with efficiency gains outweighed by competitive harm — and so perish after a rule of reason analysis. But restraints that are truly ancillary to a larger efficiency-gaining enterprise — here, the expanded closed network — are not normally condemned per se without looking at likely consequences. Addamax Corp., 152 F.3d at 52; see XIII Hovenkamp ¶ 2213c(l). If the rhetoric of older group boycott cases were taken at face value, such agreements might appear to fall within a per se ban. Cf. Associated Press v. United States, 326 U.S. 1, 13-14, 18-19, 65 S.Ct. *64 1416, 89 L.Ed. 2013 (1945). After all, in two different aspects, the ancillary agreements are promises by one competitor or group of competitors to another not to deal with a third competitor (CVS, through PharmaCare, not to include Stop & Shop and Walgreen in the PharmaCare network; Brooks not to consort with Stop & Shop and Walgreen to compete for the Blue Cross contract). But the rhetoric cannot be taken literally. After all, every joint venture among competitors that limits membership fits the lay definition of “an agreement not to deal,” and at least in recent years the Supreme Court has flatly rejected the per se label for those that have some efficiency achieving benefits. N.W. Wholesale Stationers, 472 U.S. at 295-98, 105 S.Ct. 2613. To the extent the group boycott label is useful at all to describe a per se violation, it is principally a warning against anticom-petitive secondary boycotts — e.g., manufacturers who agree not to supply a store that buys from a discounting manufacturer. Cf. Fashion Originators’ Guild of Am., Inc. v. FTC, 312 U.S. 457, 465-67, 61 S.Ct. 703, 85 L.Ed. 949 (1941); U.S. Healthcare, 986 F.2d at 593. This brings us finally to the intimations in plaintiffs’ brief of more sinister collaboration. In attacking the dismissal of per se claims, the plaintiffs accuse the district court of “ignor[ing] compelling evidence of per se illegal, horizontal agreements and their anticompetitive motives and effects,” a statement followed by bullet points that cross-reference back to the briefs’ statement of facts. Putting aside the bid rigging and ancillary-agreement charges already dealt with above, the pertinent factual charge is as follows: After Blue Cross’s selection of Phar-maCare’s bid in late 1997, representatives of CVS and representatives of PharmaCare, engaged in a series of meetings, discussions, and written correspondence with representatives of competing pharmacies Brooks and the independent pharmacy members of Provider. (J.A. 210-217, Stmt.¶¶ 89-113). As a result, these competitors agreed to set identical prescription reimbursement rates, both directly and by an equalizing mechanism that entailed payments by CVS to Brooks and Provider’s independent pharmacies for each United prescription that CVS filled. (¿.A. 216, 218-19, Stmt. ¶¶ 108-10, 120; J.A.1987-89; J.A. 3053-54; J.A. 3081-84; J.A. 3129-32.) The parties likewise discussed identical co-payment levels. (J.A. 2923.) This description insinuates wrongdoing but without the precision needed to advance the argument. Obviously any arrangement that reciprocally admitted United’s pharmacies into the Blue Cross network and Blue Cross’ into United’s would involve some arrangements as to various payment matters. This would not permit CVS and Brooks to agree in general on prices at which prescription drugs could be sold to the public. But the respective insurers and their benefit managers would be entitled to discuss with newly entering pharmacies reimbursement and co-payment rates; and there would be nothing startling about new-comers expecting the same reimbursement as earlier network members. 2 *65 Partial integration of the two networks — each operating at three levels, insurer, benefit manager, and retailer — involves settling on component payments which — at least within each network — may well involve identical payments to all participating pharmacies. It was plaintiffs’ job to explain in detail to us just what the arrangements were and why they plausibly constituted antitrust violations. In the context of partial integration, simply to refer to “identical principal reimbursement rates,” “an equalization mechanism,” and “identical co-payment levels” is to substitute innuendo for analysis. This fatal obscurity has one exception. Immediately after the paragraph quoted above, there is a further paragraph describing a February 1998 meeting between PharmaCare and two Stop & Shop representatives. PharmaCare had recently won the Blue Cross contract, and it was then negotiating with United pharmacies for reciprocal inclusion. Stop & Shop was asking to be included as well. According to a memorandum describing the meeting prepared by a Stop & Shop representative, the PharmaCare president said no. The memorandum says that the reasons PharmaCare gave for this refusal were that the negotiations .to set up the new PharmaCare network had been difficult, that it had required “horse trading” (apparently a reference to the United/Provider negotiations), and that adding a “third” chain (apparently referring to Stop & Shop on top of CVS and Brooks) would make the situation even more complex. The memorandum concluded its recitation by saying: [The PharmaCare president] also lectured John and I about our industry not being farsighted [sic] to stick together on pricing issues and that we had only ourselves to blame for the extremely low reimbursement rates in the market. In their opening brief, plaintiffs describe this last statement as “another reason” for plaintiffs’ exclusion from the expanded Blue Cross network. The intimation (by a benefits manager owned by CVS) that the retail pharmacies in general ought to stick together to raise reimbursement rates paid to them by insurers might well interest the Antitrust Division, but plaintiffs have a grievance only if their refusal to adhere was “another reason” for their exclusion. This claim is unsupported by the language of the memorandum or the follow-up depositions of the participants which we have ourselves read. The trial. The absence of plausible per se claims in no way dooms the plaintiffs’ case. The initial core arrangement— the creation of a closed network by Blue Cross and PharmaCare comprising CVS and various independent pharmacies — is a classic exclusive dealing arrangement. To simplify slightly (and without repeating details) Blue Cross, in exchange for better prices, gave its business exclusively to one group of pharmacies, agreeing for three years not to deal with others including Walgreen and Stop & Shop. Such agreements are not universally forbidden by the Sherman Act — indeed, they are quite common — but may, depending on the circumstances, unreasonably restrain trade. XI Hovenkamp, Antitrust Law ¶¶ 1802-07, 1810-1814, 1821 (1998). Because such agreements can achieve legitimate economic benefits (reduced cost, stable long-term supply, predictable prices), no presumption against such agreements exists today. Compare Standard Oil Co. of Cal. v. United States, 337 U.S. 293, 306-307, 313-14, 69 S.Ct. 1051, 93 L.Ed. 1371 *66 (1949), with Tampa Elec. Co., 365 U.S. at 334, 81 S.Ct. 623. Indeed, courts tend to be skeptical of such claims because it is not in the long-term interest of the company that grants the “exclusive deal” to drive out of business competitors of the grantee. Here, Blue Cross would be disserved by making CVS a monopolist, which could then exploit Blue Cross by demanding higher reimbursement. Still, an excluded supplier remains free to offer evidence that, in the individual instance, the anti-competitive consequences of an exclusive contract outweigh the benefits. This almost always requires a showing of injury to competition; some savings are likely or else the buyer would ordinarily not agree to forego dealing with other suppliers, and in any event an agreement that caused no harm would not be worth condemning. But harm does not mean a simple loss of business or even the demise of a competitor but an impairment of the competitive structure of the market. See Brown Shoe Co. v. United States, 370 U.S. 294, 344, 82 S.Ct. 1502, 8 L.Ed.2d 510 (1962). If an exclusive dealing contract cuts off stores like Walgreen from an unduly large portion of the available market for its goods, it and others like it might cease to provide prescription drugs. And if this led or was likely to lead to a shortage of competing drug stores (and new entry was difficult), the few remaining existing competitors might then be able to conspire or otherwise misbehave without being disciplined by competition. Where such foreclosure and negative effects are the result of an agreement, the Sherman Act may condemn the agreement as unreasonable. Eastern Food Servs., 357 F.3d at 8; see XI Hovenkamp ¶ 1802b. Accordingly, at trial, it was critical to any attack on the exclusive dealing arrangement — and almost any other non-per-se claim one could imagine — that plaintiffs establish a relevant market and harm within it. For the exclusive dealing contract, the first step would be to show the extent of foreclosure resulting from the Blue Cross contract with CVS and others in the PharmaCare network, taking account of other existing foreclosures (e.g., by United/Provider until its settlement with plaintiffs). Cf. Tampa Elec. Co., 365 U.S. at 327-29, 81 S.Ct. 623; Eastern Food Servs., 357 F.3d at 8. Plaintiffs sought to offer their market definition evidence primarily, as is typical, through an economist, Dr. Stangle. Dr. Stangle’s position in his pretrial report was that the relevant market was “the retail sale of health care financed or insurance reimbursed pharmaceutical products” (the product dimension of the market) in Rhode Island (the geographic dimension). Obviously, excluding retail sales that are not financed or reimbursed increases the percentage size of the foreclosed market. In explaining why he distinguished between reimbursed purchases and all others, Dr. Stangle pointed to the much smaller direct cost to the Blue Cross customer who purchased the same drug from a closed network pharmacy (e.g., CVS) as opposed to a non-network pharmacy (e.g., Walgreen). The defendants protested that this ignored the customer’s “true” cost of the CVS drugs which included a share of the insurance premium that the subscriber (or his employer) paid to Blue Cross and also ignored other elements affecting the comparison, such as partial reimbursement for out-of-network purchases. Pointing to these supposed flaws in Dr. Stangle’s market definition, the district court refused to allow the jury to consider it. Then, as already recounted, the plaintiffs made a proffer of the testimony for *67 the record and rested, the defense moved for a directed verdict, and the judge granted the motion and entered judgment for the defendants. The judge said that any-remaining evidence in the record supporting the plaintiffs’ market definition was too thin to permit the jury to find the proposed market in the absence of expert testimony. Plainly, for high-cost prescription drugs, whether insurance will cover purchases at a particular pharmacy tends to be crucial to consumer choice, and Dr. Stangle was correct in retorting to the defense that a customer who has paid his insurance premium (or had it paid for him) will — at least for high priced drugs — seek out closed network pharmacies if reimbursement is higher and shun those not within the insurer’s closed network. At the point of sale, the customer is interested in what he pays and gets reimbursed, not some, imputed (and now sunk) insurance premium cost. Unfortunately for Dr. Stangle’s market definition, the concern in an ordinary exclusive dealing claim by a shut-out supplier is with the available market for the supplier. Here, for Walgreen- and Stop & Shop, their potential customers are presumptively all retail customers for prescription drugs — not just that smaller sub-group who are insured or reimbursed. To say that some sub-group of customers is foreclosed proves nothing by itself about the impact on pharmacies. 3 This is the same defect we recently addressed in Eastern Food Services. There, as here, a shut-out supplier , complained that the foreclosed customers (in Eastern, they were students and faculty seeking food services on a university campus) werp foreclosed by the university’s exclusive dealing contract with another vendor. 357 F.3d at 3-4, 6-7. But the impact of the foreclosure on the supplier depended not on the impact on the students and faculty but on how many unforeclosed vending machine customers remained elsewhere. Id. at 6-7. Walgréen and Stop & Shop sell prescription drugs to lots of customers including those whose purchases are not "reimbursed. Conceivably, the latter could be so small a group that foreclosure of a large percentage of reimbursed customers would still be fatal, or there might be some special circumstance that made separate consideration of the sub-group appropriate. But the former possibility would still have to be proved, normally by a proper market definition; and of the latter, there is no hint in this case. Conceivably, some adjustment to account for the omission of self-paying customers could be devised from existing evidence: plaintiffs say that Blue Cross and United insure 70 percent of Rhode Islanders. But even if a figure representing the entire market could be derived, the number foreclosed by Blue Cross, (and formerly by United) remains unknown because a significant portion of Blue Cross’ customers have policies that do not effectively restrict them to the closed network. Nor is it our job to build plaintiffs’ case for them. The plaintiffs refer to other record evidence that they say was available to the jury to establish the same reimbursed-drugs product market without testimony from Dr. Stangle. The evidence is described in some detail; for example, the Stop & Shop executive in charge of phar *68 macy products testified to the same large differential in out of pocket costs to reimbursed and unreimbursed customers. But this simply repeats the same mistake in focus without the Ph.D. It may be worth adding that even a high number would not necessarily establish an antitrust violation. XI Hovenkamp ¶¶ 1820b at 147, 1821d. How much of the market must remain open to support decent competition depends on scale economies (retail pharmacies are different than car makers), on ease of re-entry, and on other factors. Cf id. at ¶ 1821d. The still somewhat useful Learned Hand formula for monopoly power has no counterpart in exclusive dealing law. United States v. Aluminum Co. of America, 148 F.2d 416, 424 (2d Cir.1945). But reliable numbers are an essential starting point and were not supplied. For exclusive dealing, foreclosure levels are unlikely to be of concern where they are less than 30 or 40 percent. See Jefferson Parish Hosp. Dist. No. 2 v. Hyde, 466 U.S. 2, 45-46, 104 S.Ct. 1551, 80 L.Ed.2d 2 (1984) (O’Connor, J., concurring); Hovenkamp, Federal Antitrust Policy 436-37 & nn. 43-45 (2d ed.1999) (collecting cases). But while low numbers make dismissal easy, high numbers do not automatically condemn, but only encourage closer scrutiny based on factors just mentioned. There are a few cases to the contrary, Hovenkamp, Federal Antitrust Policy, at 437 n. 49, but they cannot be reconciled with Tampa, 365 U.S. at 329, 333, 81 S.Ct. 623; see also Roland Machinery Co. v. Dresser Indus., Inc., 749 F.2d 380, 393-94 (7th Cir.1984) (Posner, J.); XI Hovenkamp, ¶¶ 1820b at 147, 1821d. The plaintiffs also complain that the district court should have permitted them to offer at trial evidence on two other matters: “[first] that Blue Cross manipulated its bidding process and the selection of PharmaCare to manage its closed pharmacy network .... [and second] evidence showing the mutual, concerted expansion of the defendants’ closed pharmacy networks.” This evidence, say the plaintiffs, was relevant to their broad rule of reason case even if neither incident was a per se violation. Plaintiffs devote only three paragraphs (and one long quotation from an old warhorse) to explaining why or how this evidence might be relevant to their “broader challenge to the defendants’ course of conduct. ...” The warhorse is the reminder in Chicago Board of Trade v. United States, 246 U.S. 231, 238, 38 S.Ct. 242, 62 L.Ed. 683 (1918), since repeated in other cases, that in assessing a restraint, a broad array of prior history of the restraint, motive, surrounding conditions and the like may be pertinent. Whether evidence was wrongly excluded depends in part on what it was offered to prove. Plaintiffs’ brief says that the two excluded incidents were relevant to show “deliberate, concerted action by the defendants” and “that the concerted action unreasonably restrained or tended to restrain trade in the relevant market.” The former proposition is admitted (there are explicit agreements); and the latter is simply an abstract description of the Sherman Act’s rule of reason and is not a honed antitrust theory for which specific evidence might or might not be relevant. Chicago Board of Trade is not an endorsement of kitchen-sink antitrust in which anything that might alarm a jury is made admissible. Plaintiffs had, so far as appears, only one substantial and relevant antitrust theory — namely, that Blue Cross had adopted a competitively harmful exclusive dealing arrangement, made more fearsome by its reciprocal expansion and coupled with overbroad ancillary restraints. For this, collaboration was patent — no one *69 denied the existence of the contracts. The allegedly flawed bidding process added nothing to proof of collaboration and had no demonstrated antitrust significance. In their brief on appeal, plaintiffs suggest that the bid process evidence showed that Blue Cross was determined to exclude plaintiffs for the indefinite future and “would have permitted an inference that the three-year terms of the network agreements were illusory.” The only “restraint” within the meaning of the Sherman Act was the contractual limitation on Blue Cross’ ability to add new pharmacies; any unilateral decision thereafter to exclude plaintiffs would not extend the pertinent “restraint.” The reciprocal expansion incident was arguably more relevant. It did not affect the extent of total foreclosure — United had not previously included plaintiffs; but (merely as an example) some of the ancillary terms seemingly made it harder for the plaintiffs to join either network or start a competing network. Yet the district judge’s in limine ruling did not exclude proof of the terms — only proof of the details of negotiations, and at trial the district judge apparently reversed the tentative exclusion and allowed in some of the details. There is yet another problem. As already noted, neither the alleged sham bidding process nor the reciprocal expansion of closed networks was a per se violation although the latter, particularly with its ancillary restraints, had a bearing on any rule of reason attack on the core exclusive dealing arrangement. It is not easy to think of a rule of reason analysis that does not depend on showing adverse effects on competition in a properly defined relevant market. Cf. Augusta News Co., 269 F.3d at 47. This predicate failed with Dr. Stan-gle’s testimony. Some antitrust cases are intrinsically hopeless because (as in Eastern) they merely dress up in antitrust garb what is, at best, a business tort or contract violation. By contrast, Blue Cross’ adoption of a closed network whose' impact was arguably reinforced by its reciprocal expansion coupled with ancillary restraints, might be an unreasonable foreclosure of a properly defined market. However, as plaintiffs omitted the proof, one simply cannot tell. Whether or not there was an antitrust violation affecting the plaintiffs, some of Blue Cross’ customers will doubtless be inconvenienced by restricting their purchases to the closed network. If use of a closed network reduces costs for Blue Cross and also reduces or holds down the price of a closed market policy, this may be a legitimate outcome — especially if an open market policy is also an available option. There are few free lunches in the world of commerce. The possibility always remains that a dominant company may act inefficiently or may unfairly exploit its customers. The usual check for such abuses is competition (here, United is an obvious competitor for Blue Cross) but competition may sometimes be inadequate. In such cases antitrust may not always offer customers much protection, Aluminum Co. of America, 148 F.2d at 429; but state regulation — sometimes wisely and sometimes not — is usually free to fill the gap. Affirmed. Footnotes [1] The per se categories are discussed with relevant citations in Eastern Food Services, 357 F.3d at 4-5 & n. 1, and other of our decisions. On minimum vertical price fixing, see Augusta News Co. v. Hudson News Co., 269 F.3d 41, 47 (1st Cir.2001). [2] As for the "equalization” payments by CVS to Brooks and Provider's independent pharmacies, Blue Cross/Pharmacare apparently reimburses its pharmacies for certain prescription drugs at lower rates than those at which United/Provider reimburses its pharmacies for the same drugs. To compensate United's pharmacies for agreeing to the lower reimbursement rates, CVS and the other Blue Cross pharmacies agreed to pay United’s pharmacies $.25 for every United subscriber's prescription that they filled. This might or might not be a permissible arrangement, but it is not naked price fixing by competitors. [3] One of the defendants suggests that Rhode Island is too large a geographic market because customers shop locally. Reflection will reveal that — whatever the correct market— this argument is a different version of the same mistake of focusing on the customer rather than the supplier.