We must decide whether purchasers were able to establish that a citric acid manufacturer conspired with competitors to fix prices and to allocate market share in violation of the federal antitrust laws.

I

Citric acid is a corn derivative with a wide variety of uses in the manufacture of food, soft drinks, detergents, and pharmaceuticals. Varni Brothers Corp. and 7-Up Bottling Company of Philadelphia, Inc. (collectively “Varni”) filed civil antitrust class actions against Cargill, Incorporated (“Cargill”), Archer Daniels Midland (“ADM”), Haarman & Reimer (“H&R”), Hoffman LaRoche (“HLR”), and Jungbun-zlauer (“JBL”), alleging that these firms conspired to fix citric acid prices and to allocate market share. These cases were consolidated and transferred to the Northern District of California. The district court certified a class consisting of all peo- *1093 pie and entities (other than government purchasers and the defendants themselves and affiliated companies) who had purchased citric acid directly from the defendants.

Conceding the existence of a conspiracy in the citric acid market but denying its participation therein, Cargill moved for summary judgment. ADM, H&R, HLR, and JBL (collectively the “admitted conspirators”) each admitted to conspiring to fix citric acid prices and reached settlements with Varni. Thus, the only question before the district court was whether Car-gill was a member of this conspiracy or, more precisely, whether Varni had produced sufficient evidence such that a reasonable factfinder could so infer. The district court concluded that Varni failed to satisfy this test because its evidence did not “tend[ ] to exclude the possibility that [Cargill] acted independently.” Matsushita Elec. Indus. Co. v. Zenith Radio Corp., 475 U.S. 574, 588 , 106 S.Ct. 1348 , 89 L.Ed.2d 538 (1986) (internal quotation marks and citation omitted). Accordingly, the court granted summary judgment in favor of Cargill and commented that “[a]p-parently hoping that quantity will substitute for quality, plaintiffs have submitted voluminous but weak circumstantial evidence that they argue indicates that Cargill was a member of the conspiracy.”

On appeal, Varni argues that Cargill’s participation in the conspiracy can be reasonably inferred from the circumstantial evidence in the record. Varni also appeals the district court’s denial of a discovery motion, an issue we consider in Part IV.

II

To avoid extensive repetition of the facts, we will first consider the legal standard by which we decide whether Varni has produced sufficient evidence to survive summary judgment. By understanding the proper legal framework applicable at the summary judgment stage, we can better analyze the evidence.

A

Price fixing is illegal per se under section 1 of the Sherman Act. See 15 U.S.C. § 1 ; United States v. Socony-Vacuum Oil Co., 310 U.S. 150, 223 , 60 S.Ct. 811 , 84 L.Ed. 1129 (1940). “On a claim of concerted price fixing, the antitrust plaintiff must present evidence sufficient to carry its burden of proving that there was such an agreement.” Monsanto Co. v. Spray-Rite Serv. Corp., 465 U.S. 752 , 763, 104 S.Ct. 1464 , 79 L.Ed.2d 775 (1984). Summary judgment “shall be rendered forthwith if the pleadings, depositions, answers to interrogatories, and admissions on file, together with the affidavits, if any, show that there is no genuine issue as to any material fact and that the moving party is entitled to a judgment as a matter of law.” Fed.R.Civ.P. 56(c); see also Celotex Corp. v. Catrett, 477 U.S. 317, 322 , 106 S.Ct. 2548 , 91 L.Ed.2d 265 (1986) (“Rule 56(c) mandates the entry of summary judgment, after adequate time for discovery and upon motion, against a party who fails to make a showing sufficient to establish the existence of an element essential to that party’s case, and on which that party will bear the burden of proof at trial.”).

Varni can establish a genuine issue of material fact by producing either direct evidence that Cargill conspired to fix citric acid prices or circumstantial evidence from which a reasonable factfinder could conclude that Cargill so conspired. Although Varni argues halfheartedly that it has produced direct evidence that Cargill entered into illegal price-fixing agreements with the admitted conspirators — in which case summary judgment would, of course, be inappropriate, see McLaughlin v. Liu, 849 F.2d 1205, 1208 (9th Cir.1988); T.W. Elec. Serv., Inc. v. Pacific Elec. Contractors Ass’n, 809 F.2d 626, 631-32 (9th Cir.1987) — Varni concedes that such evidence is, at best, weak. Having reviewed the proffered evidence, we can find nothing in the record that establishes, without requiring any inferences, that Cargill participated in the citric acid price-fixing conspiracy. See In re Baby Food Antitrust Litig., 166 *1094 F.3d 112, 118 (3d Cir.1999) (“Baby Food ”) (“Direct evidence in a Section 1 conspiracy-must be evidence that is explicit and requires no inferences to establish the proposition or conclusion being asserted.”). We thus agree with both parties that the essential question in this case is whether Varni’s circumstantial evidence is sufficient to permit a reasonable factfinder to conclude that Cargill was a participant in the citric acid conspiracy. 1

Matsushita is the Supreme Court’s foremost explanation of how summary judgment works in the antitrust context. See Matsushita, 475 U.S. at 585-97 , 106 S.Ct. 1348 . The Matsushita Court began by reaffirming hornbook law that courts are obligated to construe the evidence in the light most favorable to the non-moving party, to give the non-moving party the benefit of all reasonable inferences, and to refrain from making credibility determinations. See id. The Court cautioned, however, that, although plaintiffs are to be given the benefit of the doubt, they “must do more than simply show that there is some metaphysical doubt as to the material facts.” Id. at 588 , 106 S.Ct. 1348 . Importantly, “conduct as consistent with permissible competition as with illegal conspiracy does not, standing aloné, support an inference of antitrust conspiracy.” Id. at 586 , 106 S.Ct. 1348 . Instead, an inference of conspiracy is sustainable only if “reasonable in light of the competing inferences of independent action,” and “[t]o survive a motion for summary judgment ..., a plaintiff seeking damages for a violation of § 1 must present evidence ‘that tends to exclude the possibility’ that the alleged conspirators acted independently.” Id. at 588, 106 S.Ct. 1348 (quoting Monsanto, 465 U.S. at 764 , 104 S.Ct. 1464 ); accord City of Long Beach v. Standard Oil Co. of California, 872 F.2d 1401, 1404 (9th Cir.1989). The Court warned that permitting the inference of conspiratorial behavior from evidence consistent with both lawful and unlawful conduct would deter pro-competitive conduct — an especially pernicious danger in light of the fact that the very purpose of the antitrust laws is to promote competition. See id. at 593 , 106 S.Ct. 1348 (“[Cjourts should not permit factfinders to infer conspiracies when such inferences are implausible, because the effect of such practices is often to deter procompetitive conduct.”).

Based on Matsushita , this circuit has outlined a two-part test to be applied whenever a plaintiff rests its case entirely on circumstantial evidence. First, the defendant can “rebut an allegation of conspiracy by showing a plausible and justifiable reason for its conduct that is consistent with proper business practice.” Richards v. Neilsen Freight Lines, 810 F.2d 898, 902 (9th Cir.1987); accord City of Long Beach, 872 F.2d at 1406 ; Wilcox v. First Interstate Bank of Oregon, 815 F.2d 522, 526-28 (9th Cir.1987); T.W. Elec. Serv., 809 F.2d at 632, 635 . The burden then shifts back to the plaintiff to provide specific evidence tending to show that the defendant was not engaging in permissible competitive behavior. See Long Beach, 872 F.2d at 1406 ; Jeanery, Inc. v. James Jeans, Inc., 849 F.2d 1148, 1159 (9th Cir.1988); Wilcox, 815 F.2d at 525 ; Richards, 810 F.2d at 902, 904 ; T.W. Elec. Serv., 809 F.2d at 632 . 2

*1095 B

We have repeatedly applied this summary judgment framework, which, to repeat, we must apply whenever the plaintiff cannot establish every element of its case without asking the court to draw an inference in his favor. In Richards , a local trucking carrier, Foothills Express, accused major trucking companies of engaging in a concerted group boycott against Foothills Express. The defendant corporations used Foothills Express as an inter-liner for shipments to the Sacramento area: the defendants carried freight from areas nationwide to Sacramento using their own trucks and Foothills Express made the deliveries from Sacramento to surrounding areas. Foothills Express, however, began soliciting the business of defendants’ customers using information gained from carrier freight bills, a practice known as “back solicitation.” In response to Foothill Express’s back solicitation, each of the defendants cut off or substantially reduced its business with Foothills. See Richards, 810 F.2d at 900-01 .

Foothills Express alleged that the defendants had violated the antitrust laws by agreeing not to employ the services of any local carrier who engaged in back solicitation. In support of its claims, Foothills Express cited the fact that each defendant stopped using its services around the same time. The defendants, however, responded that it was in each defendant’s independent self-interest to refrain from doing business with local carriers who competed with them. Considering the defendants’ explanation, we held that the defendants had “met their burden of proffering a plausible justification for their conduct consistent with proper business practice.” Id. at 902 . Thus, the burden shifted to Foothills to demonstrate that the trucking companies acted pursuant to some sort of conspiracy, since the legality of their conduct depended crucially, as is often the case, on whether they had acted independently or pursuant to an agreement.

Foothills Express asserted that the existence of a conspiracy could be inferred from the concession by one of the defendant corporations’ presidents that there was “a gentlemen’s agreement” with the other major trucking companies against back solicitation. Richards, 810 F.2d at 903 . We acknowledged that it was possible to interpret this deposition testimony as evidence of an agreement amongst competitors, but noted that this evidence was not necessarily inconsistent with lawful behavior. The evidence could also plausibly be interpreted either as a “referfence] to industrywide acceptance of the competitive reality that back solicitation was an abuse of the interlining relation” or as “meanfing] there were discrete, bilateral understandings between each defendant on one hand, and its interlining company on the other” (rather than a horizontal agreement amongst the defendants). Id. Under both of these latter interpretations, defendants would have engaged only in lawful conduct. Because the circumstantial evidence was as consistent with permissible competitive behavior as with illegal conspiracy — and thus did not tend to exclude the possibility of legitimate behavior' — we concluded that the supposed smoking gun was not sufficient for Foothills to overcome defendants’ summary judgment motion. See id. at 903-04 .

We reached a similar conclusion in T.W. Electrical Service, in which electrical contractors accused the Pacific Electrical Contractors Association (“PECA”) of conspiring with unions to require all contractors to make certain contributions. Because PECA offered a legitimate, non-conspiratorial interpretation of its conduct, we emphasized that, to withstand PECA’s summary judgment motion, “the contractors must not only come forward with ‘specific facts’ showing the existence of a genuine issue of material fact, but also must set forth ‘specific facts’ that tend to exclude the possibility that PECA acted independently.” T.W. Elec. Serv., 809 F.2d at 635 . *1096 In light of the fact that the contractors failed to produce evidence showing that the contribution requirement was the product of an agreement between PECA and others, rather than a requirement unilaterally imposed by PECA, we affirmed the grant of summary judgment in favor of PECA. See id.

In Wilcox , commercial borrowers of the First Interstate Bank of Oregon (“FIOR”) sued FIOR alleging that it had conspired with other banks to fix interest rates. After the borrowers successfully convinced a jury to infer conspiracy from evidence of price parallelism, exchanges of pricing information, and opportunities to conspire, the district court granted judgment notwithstanding the verdict in favor of FIOR. Agreeing that there was no direct evidence of any agreement to set interest rates, that the circumstantial evidence of price fixing did not tend to exclude the possibility that FIOR set rates independently, and that FIOR did not act against its own independent self-interest, we affirmed. See Wilcox, 815 F.2d at 524-28 .

In another case, the City of Long Beach, which sold oil to various major oil companies at a price based on posted prices for Wilmington crude oil, alleged that the oil companies conspired to maintain uniformly low, noncompetitive posted prices. Although it lacked direct evidence of conspiracy, the City produced evidence that the defendants exchanged crude oil amongst themselves using a complicated “three-cut exchange system” rather than simply valuing oil based upon the posted prices. The City asserted that the companies used the three-cut exchange system because the posted prices were rigged: the oil corporations were making an “unusual effort” to maintain the elaborate three-cut exchange system, and the defendants did not demonstrate any advantage of this “cumbersome and inconvenient” system over valuing oil using posted prices. Long Beach, 872 F.2d at 1403, 1406 . Applying the legal framework outlined above, we concluded that evidence of such behavior, inexplicable absent a conspiracy, did tend to exclude the possibility that defendants acted independently and thus reversed the grant of summary judgment in the defendants’ favor. See id. at 1406-07 .

C

Notwithstanding our many cases applying this legal framework for analyzing circumstantial evidence cases, Varni asks us to apply a very different summary judgment standard whereby it need not produce evidence tending to exclude the possibility that defendants acted independently even though it relies entirely on circumstantial evidence. Varni cites only a single case, In re Coordinated Pretrial Proceedings in Petroleum Products Antitrust Litigation, 906 F.2d 432 (9th Cir.1990), in support of its proposed legal standard. The plaintiffs in that case, however, presented direct evidence of conspiracy, see id. at 456-57 , 459-60 & n. 22, thus making dicta any discussion therein of the standard applicable when plaintiffs rely exclusively on circumstantial evidence. See United States Nat’l Bank of Oregon v. Independent Ins. Agents of Am., Inc., 508 U.S. 439 , 463 n. 11, 113 S.Ct. 2173 , 124 L.Ed.2d 402 (1993) (recognizing the “need to distinguish an opinion’s holding from its dicta”); Delta & Pine Land Co. v. Sinkers Corp., 177 F.3d 1343, 1349-50 (Fed.Cir.1999) (declining to follow dicta).

The requirement that a plaintiff who relies solely on circumstantial evidence of conspiracy — as Varni does — must produce evidence tending to exclude the possibility that defendants acted independently follows directly from the Supreme Court’s opinion in Matsushita and is, as we have explained, well-established in both this circuit, see Long Beach, 872 F.2d at 1406 ; Jeanery, 849 F.2d at 1159 ; Wilcox, 815 F.2d at 525 ; Richards, 810 F.2d at 902, 904 ; T.W. Elec. Serv., 809 F.2d at 632 , as well as in our sister circuits. See Mitchael v. Intracorp, 179 F.3d 847, 858 (10th Cir.1999); Baby Food, 166 F.3d at 123 ; City of Tuscaloosa v. Harcros Chems., Inc., 158 F.3d 548, 569-70 (11th Cir.1998), petition for cert. filed, 67 U.S.L.W. 3773 *1097 (U.S. June 9, 1999) (No. 98-1997); Comer Pocket of Sioux Falls, Inc. v. Video Lottery Techs., Inc., 128 F.Sd 1107, 1112 (8th Cir.1997), cert. denied, — U.S. -, 118 S.Ct. 1054 , 140 L.Ed.2d 116 (1998); Clorox Co. v. Sterling Winthrop, Inc., 117 F.3d 50, 55 (2d Cir.1997); Thompson Everett, Inc. v. National Cable Adver., L.P., 57 F.3d 1317, 1323 (4th Cir.1995); Wallace v. Bank of Bartlett, 55 F.3d 1166, 1170 (6th Cir.1995); Reserve Supply Corp. v. Owens-Corning Fiberglas Corp., 971 F.2d 37, 49 (7th Cir.1992); In re Beef Indus. Antitrust Litig., 907 F.2d 510, 514 (5th Cir.1990).

With the appropriate legal standard in mind, we now analyze the evidence in the record.

Ill

Varni offers a plethora of evidence which it claims shows that Cargill was a member of the citric acid conspiracy. As Varni emphasizes, the crucial question is whether all the evidence considered as a whole can reasonably support the inference that Cargill conspired with the admitted conspirators to fix prices. See Continental Ore Co. v. Union Carbide & Carbon Corp., 370 U.S. 690, 699 , 82 S.Ct. 1404 , 8 L.Ed.2d 777 (1962). Thus, we consider the evidence offered by Varni, and then ask whether, in light of all the evidence considered together, a reasonable factfinder could have concluded that Cargill conspired to fix prices.

A

Varni contends that the admitted conspirators and Cargill used the industry’s trade association, the European Citric Acid Manufacturers Association (“ECAMA”), as a vehicle for conducting illegal conspiratorial activities. The claim, in a nutshell, is that ECAMA functioned as a front organization for conspiratorial activities.

ECAMA held semi-annual meetings that were attended, on a regular basis, by “masters” (the term used by the conspirators to describe high-level executives involved in the conspiracy who negotiated the broad terms of the conspiracy) and “sherpas” (lower-level corporate executives responsible for implementing the details of the conspiracy), among others. Varni argues that these meetings provided an opportunity for competitors to reach illegal agreements and that it is unlikely that these meetings were legitimate in fight of the fact that many of those attending ECAMA meetings were active players in the conspiracy.

Having reviewed the record independently, however, we agree with the district court that there is no evidence to support Varni’s theory that these meetings were used to pursue illegal ends. Although the evidence makes clear that representatives of the admitted conspirators gathered in “unofficial” meetings to fix prices and to allocate market shares — often in the same city as, and within a few.days of, officially scheduled ECAMA meetings — there is no evidence that illegal activities took place during ECAMA meetings attended by Cargill representatives. Nor is there evidence that Cargill participated in any of the “unofficial” meetings. Instead, the apparently uncontroverted testimony of Hans Hartmann, president of H&R and a master in the conspiracy, was to the contrary: that official ECAMA meetings were legitimate and that the illegal activities took place in separate, non-ECAMA meetings. According to Hartmann, the ECA-MA gatherings merely provided a convenient excuse for the conspirators to be in the same city at the same time.

Lacking evidence that the admitted conspirators and Cargill negotiated or reached price-fixing agreements during ECAMA meetings, Varni alleges that the citric acid manufacturers used ECAMA as a mechanism for negotiating illegal agreements with Chinese citric acid producers, whose expanding market share had caused increasing concern among ECAMA members. Varni cites the following evidence: (1) that ECAMA gathered information about Chinese citric acid producers; (2) that ECAMA commissioned a study of *1098 “the Chinese situation as to location, production, distribution, prices”; (3) that ECAMA held a meeting to discuss a report on Chinese citric acid producers; (4) that an executive of Gadot Petrochemical Industries, Ltd. (“Gadot”) circulated a memorandum advising ECAMA to correspond with Chinese citric acid producers to “stabilize and regulate Chinese production and exports”; and (5) that handwritten notes of an ADM employee at a March 1994 ECAMA working group meeting contained the following statement — “Undertaking is a confidential agreement to maintain price. Producers must police.” 3

We have little trouble concluding that the first three pieces of evidence do not tend to exclude the possibility of legitimate behavior. Gathering information about pricing and competition in the industry is standard fare for trade associations. If we allowed conspiracy to be inferred from such activities alone, we would have to allow an inference of conspiracy whenever a trade association took almost any action. As the Supreme Court has recognized, however, trade associations often serve legitimate functions, such as providing information to industry members, conducting research to further the goals of the industry, and promoting demand for products and services. See, e.g., Maple Flooring Mfrs. Ass’n v. United States, 268 U.S. 563, 567 , 45 S.Ct. 578 , 69 L.Ed. 1093 (1925).

At first glance, the last two pieces of evidence appear more ominous. The Ga-dot memorandum' suggests an illegal course of conduct — that ECAMA contact the Chinese producers and agree to stabilize prices. However, minutes of an ECA-MA meeting indicate that the suggestion “was quickly overviewed and it was decided’not to pay attention to that message for most of the information contained in it [was] against the spirit of the anti-trust law.” It would not be reasonable to infer that Cargill engaged in illegal activities merely from evidence that an illegal course of action was suggested but immediately rejected. 4 Finally, the handwritten notes that constitute the fifth piece of evidence were taken at a meeting at which Cargill was not present. Thus, even if we interpret the notes as Varni asks us to, it would nevertheless not be reasonable to infer from such notes that Cargill conspired to fix prices.

B

Varni argues that, even if ECAMA was not directly used as a mechanism for fixing prices, it served as a front organization for the illegal distribution of detailed, firm-specific information on citric acid production and sales, information which facilitated the policing of agreed-upon market share allocations.

ECAMA collected figures on production and sales from each of its members, audited this information on an annual basis, and produced statistics aggregated by country on citric acid production and sales. ECA-MA members reported production and sales information to an independent agent of ECAMA, who distributed information aggregated by country to members. The reason that only aggregate statistics were distributed was to preserve the confidentiality of individual members’ sales and production figures. It is uncontested that these activities served the legitimate purpose of informing members of worldwide citric acid conditions and were, in themselves, wholly legal. Cargill vigorously *1099 maintains that ECAMA distributed only aggregate information.

Varni contends, however, that ECAMA engaged in the additional illegal step of distributing firm-specific data to each member firm so that they could verify that all conspirators were adhering to agreed-upon market shares. The only evidence produced by Varni in support of its allegation that members received firm-specific information is a table with detailed, firm-specific market share information found in Cargill’s business records. Although Var-ni asserts that this information could have come only from ECAMA, Cargill maintains that these figures were generated from its own internal market research, and Varni has offered no evidence to the contrary. Moreover, the only other evidence in the record relating to this issue is Hart-mann’s testimony that the admitted conspirators policed their conspiracy by exchanging sales and production figures on a monthly basis by telephone, which clearly supports Cargill’s position.

Lacking solid evidence that ECAMA ever disclosed firm-specific information to members, Varni falls back upon its argument that such an inference is reasonable in light of the fact that ECAMA hired auditors to verify the information submitted by members. According to Varni, the only conceivable reason ECAMA would have had to hire an independent firm to audit production and sales data is to allow conspirators to police their price-fixing conspiracy.

Cargill, however, offers a legitimate explanation — that the auditing process served the perfectly legal purpose of guaranteeing the reliability of the aggregate statistics on worldwide market conditions that ECAMA distributed to members. Cargill joined ECAMA to get “a more accurate picture of world supply/demand conditions.” Cargill, like the other firms, valued having accurate information on market conditions throughout the world, because such information was being used to make strategic business decisions. If any ECAMA member submitted inaccurate data (whether intentionally or not), the resulting aggregate statistics would be inaccurate. Decisions made on the basis of inaccurate data are, of course, likely to be sub-optimal. If firms had to worry about whether their competitors were submitting accurate information, they would likely have been less willing to rely on the information distributed by ECAMA in making strategic business decisions. Auditing eliminated this worry.

Applying the legal framework outlined in Part II, we conclude that (1) Cargill’s participation in an organization that collected and audited members’ production and sales figures to calculate and distribute aggregate market statistics is as consistent with legitimate behavior as with conspiratorial behavior, and (2) Varni has failed to produce evidence tending to exclude the possibility that Cargill acted independently.

C

Lacking evidence that ECAMA itself served illegal ends, Varni falls back upon the argument that Cargill’s decision to join ECAMA in the first place constitutes indirect evidence that Cargill was a participant in the citric acid conspiracy. Its theory is the same here as with much of the other evidence — that Cargill’s behavior was inconsistent with its own self-interest and that, since no non-conspiring rational actor would have decided to do what Cargill did, Cargill must have been conspiring.

Cargill asserts that it joined ECAMA to obtain accurate information on worldwide market conditions in the citric acid industry, which it valued because it was considering expanding into foreign markets. In joining, Cargill believed that information on worldwide supply and demand in the citric acid industry would assist it in deciding which markets to enter, when to enter them, and how best to do so.

Varni, however, maintains that the costs of joining ECAMA far outweighed any *1100 benefits, such that Cargill must have received some other (ie., conspiracy-related) benefit from joining. The costs relate to Cargill’s obligation, as an ECAMA member, to disclose sales and production information to an independent agent of ECA-MA who would calculate the aggregate statistics distributed to members. Cargill itself concedes that it considered the danger that its production and sales figures would fall into the hands of competitors to be a drawback of joining ECAMA. Varni argues that, on the benefits side, Cargill gained little from joining ECAMA because ECAMA’s focus was on the European citric acid market and Cargill did not compete in that market.

In response, Cargill points to the uncon-troverted evidence in the record that it did explicitly weigh the costs and benefits of joining ECAMA and did take steps to assure that its data would be kept as confidential as possible. When ECAMA and Cargill representatives met in February 1993 to negotiate and discuss terms of membership, Robert Simpson, a Cargill executive, conveyed Cargill’s concerns about joining ECAMA. Simpson stated that he recognized that, as an ECAMA member, Cargill would have to disclose production and sales information to ECA-MA so that ECAMA could compute aggregate statistics, but wanted to be absolutely sure that Cargill-specific figures would not fall into the hands of competitors. In a letter of March 12, 1999, ECAMA’s Secretary responded, explaining that Cargill’s data would be kept confidential and that only aggregate market data would be provided to other firms. The Secretary did, however, acknowledge that when Cargill joined ECAMA, ECAMA members would be able to derive a one-time estimate of Cargill’s sales and productions figures by comparing the pre-Cargill figures with the new figures after Cargill’s joinder, with the difference roughly equaling Cargill’s sales and production figures.

Simpson evaluated the costs and benefits of joining ECAMA in an internal memorandum. By joining ECAMA, Cargill would get a better picture of world supply and demand conditions, which would be of increasing importance as Cargill expanded worldwide. In addition, Cargill stood to benefit from ECAMA’s work on product development and other industry-related matters. Simpson noted that “[t]he only drawback to ECAMA membership [was] disclosure of Cargill’s sales volume and production capacity,” but concluded that this drawback was minimal because, “[a]f-ter several detailed discussions with ECA-MA, I feel comfortable with the confidentiality of Cargill trade data and ECAMA’s position on national competition laws (antitrust).” Weighing the costs against the benefits, Simpson recommended that Car-gill join ECAMA, and Cargill subsequently did so in May 1998.

To the extent that Cargill’s decision is consistent with conspiratorial behavior, the evidence is equally well, if not better, interpreted as a decision in Cargill’s own independent self-interest. Cargill offered a sound reason for joining ECAMA, and, importantly, Varni’s evidence does not tend to exclude the possibility that Cargill, acting independently, reasonably concluded that the benefits of membership outweighed the costs. Varni’s theory might be validated if information on worldwide market conditions were truly useless to Cargill, but this information was valuable because Cargill was considering expanding abroad. Finally, we note that Varni’s argument that information on worldwide market conditions was worthless to Cargill is inconsistent with its attacks on Cargill for not expanding operations rapidly enough outside of the United States. See infra Part III.D.

D

Varni argues that Cargill’s failure to expand its production capacity more rapidly than it did constitutes circumstantial evidence that it was conspiring to fix prices. When it entered the citric acid market, Cargill’s production facility had an annual production capacity of 56 million pounds. In 1992, Cargill expanded capaci *1101 ty to 80 million pounds. In March 1992, Cargill issued a press release announcing that it would double capacity to 160 million pounds per year. Sometime in or before October 1992, however, Cargill decided to increase capacity from 80 million to only 120 million pounds. 5 Varni contends that Cargill's decision to expand capacity by 50 percent rather than by 100 percent is circumstantial evidence that Cargill was conspiring to fix prices and allocate market share.

The timing of this decision, however, does not fit with Varni’s theory of the conspiracy. Varni contends that Cargill was a vigorous competitor until it turned conspirator in 1993, but the decision to expand capacity to 120 million rather than 160 million pounds was made by late 1992. Because “the factual context renders [appellants’] claim implausible,” no reasonable factfinder could conclude from this evidence that Cargill was part of the conspiracy. Matsushita, 475 U.S. at 587 , 106 S.Ct. 1348 .

Moreover, Cargill did expand rapidly. It increased its annual production capacity from 56 million to 80 million and then to 120 million — in only a matter of years. Faulting a company for expanding too slowly where it doubled capacity in a few years would subject countless strategic business decisions to second-guessing by courts. Courts have recognized that firms must have broad discretion to make decisions based on their judgments of what is best for them and that business judgments should not be second-guessed even where the evidence concerning the rationality of the challenged activities might be subject to reasonable dispute. See Reserve Supply, 971 F.2d at 49-53 ; Valley Liquors, Inc. v. Renfield Importers, Ltd., 822 F.2d 656, 662 (7th Cir.1987). 6 Cargill has explained why it did not expand even more rapidly: it was concerned that an excessively rapid expansion in citric acid supply would undermine prices and wanted to avoid precipitating a costly price war as it had done when it first entered the citric acid market (incurring substantial losses as a result).

E

Varni contends that Cargill “decided” to stop increasing its market share in 1993, which only a conspirator would have done. We note, as an initial matter, that it is not reasonable to infer that a firm is engaging in illegal activities merely from the fact that it failed to continue to increase market share. If a conspiracy could be inferred so easily, then no firm against whom an antitrust claim is levied would ever be entitled to summary judgment unless it continually increased its market share. In light of the fact that firms enjoy discretion in exercising business judgment, see supra Part III.D, such a rule of law would be untenable.

In any case, there is no evidence in the record supporting Varni’s claim that Cargill’s market share flattened after 1993. Varni attempts to rely on market share statistics contained in an internal Cargill document cited in the report of its economic expert. See infra n. 9. That document, however, was not included as part of the summary judgment record. To be considered in a motion for summary judgment, “documents must be authenticated by and attached to an affidavit that meets the requirements of [Federal Rule of Civil Procedure] 56(e) and the affiant must be a person through whom the exhibits could be admitted into evidence.” Hal Roach Studios, Inc. v. Richard Feiner & Co., 896 F.2d 1542, 1550-51 (9th Cir.1989) (internal quotation marks omitted). The document on which Varni relies for its market share figures was not submitted as part of the *1102 summary judgment record, and thus we do not consider it.

Varni attempts to get this document in through the back door. First, it argues that this data is admissible because it is discussed in an expert report which was part of the record. The law is clear, however, that an expert report cannot be used to prove the existence of facts set forth therein. See Paddack v. Dave Christensen, Inc., 745 F.2d 1254, 1261-62 (9th Cir.1984); see also Brooke Group Ltd. v. Brown & Williamson Tobacco Corp., 509 U.S. 209, 242 , 113 S.Ct. 2578 , 125 L.Ed.2d 168 (1993) (“Expert testimony is useful as a guide to interpreting market facts, but it is not a substitute for them.”); Rebel Oil Co. v. Atlantic Richfield Co., 51 F.3d 1421, 1436 (9th Cir.1995) (same). Varni makes the alternative argument that the expert report can be used to show the truth of facts stated therein because it is a summary of the data admissible under Federal Rule of Evidence 1006. Varni, however, has failed to show that the underlying document that it neglected to attach as evidence to its opposition motion qualifies as a voluminous writing that cannot be conveniently examined. See Fed.R.Evid. 1006 (“The contents of voluminous writings, recordings, or photographs which cannot conveniently be examined in court may be presented in the form of a chart, summary, or calculation.”).

Because there is no evidence in the record establishing that Cargill’s market share was constant between 1993 and 1995, 7 any inference founded upon that factual assertion — even one drawn by an economic expert — is necessarily unreasonable. See Brooke Group, 509 U.S. at 242 , 113 S.Ct. 2578 (“When an expert opinion is not supported by sufficient facts to validate it in the eyes of the law, or when indisputable record facts contradict or otherwise render the opinion unreasonable, it cannot support a jury’s verdict.”).

F

Varni argues that the fact that Cargill’s list prices mirrored those of its competitors is evidence of conspiracy. A section 1 violation cannot, however, be inferred from parallel pricing alone, see Wallace, 55 F.3d at 1168 ; Wilcox, 815 F.2d at 526 ; Reserve Supply, 971 F.2d at 50 , nor from an industry’s follow-the-leader pricing strategy, see United States v. International Harvester Co., 274 U.S. 693, 708-09 , 47 S.Ct. 748 , 71 L.Ed. 1302 (1927); Wilcox, 815 F.2d at 526 . Parallel pricing is a relévant factor to be considered along with the evidence as a whole; if there are sufficient other “plus” factors, an inference of conspiracy can be reasonable. See Petroleum Prods., 906 F.2d at 441-60 .

It is true that Cargill’s list prices have generally been nearly identical to those of ADM and H&R. Between 1990 and 1997, ADM, H&R, and Cargill always changed list prices within a month of one another and generally did so in the same month. We must, however, consider the evidence in the record as a whole. See Continental Ore, 370 U.S. at 699 , 82 S.Ct. 1404 . Although there appears to have been little competition in citric acid list prices, Cargill did price aggressively in actual contracts. An internal Cargill document stated its strategy of generally “[s]upport[ing] industry price increase^]” but “[selectively competing] at strategic accounts.” As part of this strategy, Car-gill reached agreements to supply most of the citric acid needs of Coca-Cola, Pepsi, Kraft General Foods, Procter & Gamble Beverage, and Quaker Oats, and these five accounts constituted the majority of Car-gill’s sales.

Also relevant is Cargill’s success in gaining 22 percent of the United States market share by 1995, which came largely at the expense of ADM and H&R. A May 1993 H&R internal monthly report expressed exasperation with Cargill’s success, report *1103 ing as follows: “We as a company can no longer take a position of ‘maintaining our share of the business’, [sic] Why? Because we are losing! Cargill has new capacity coming on stream and they will continue to go after anyone they can get. Each time we play it conservative in our negotiations and price offerings Cargill beats us by half to one cent. Most of the time its [sic] a price issue.” A second H&R internal report from 1994 on the domestic citric acid market concluded that Cargill had gained market share from ADM and H&R by consistently underpricing them. A third H&R document, dated 1995, similarly reported that “Cargill has been quite aggressive with citric price offers.”

Considering the foregoing evidence together, we conclude that the evidence does not tend to exclude the possibility that Cargill acted legally in its pricing decisions.

G

Varni asks the court to infer Car-gill’s role in the conspiracy from the fact that representatives of Cargill attended meetings and had telephone conversations with individuals who have been identified as masters and sherpas. Varni does not offer any specific details with regard to illegal discussions, but instead merely asks us to infer participation in the conspiracy from the opportunity to do so. Such meetings, at least in and of themselves, do not tend to exclude the possibility of legitimate activity. See Baby Food, 166 F.3d at 126 (“[C]ommunications between competitors do not permit an inference of an agreement to fix prices unless ‘those communications rise to the level of an agreement, tacit or otherwise.’ ”) (quoting Alvord-Polk, Inc. v. F. Schumacher & Co., 37 F.3d 996, 1013 (3d Cir.1994)).

H

Varni argues that the fact that Cargill's files contained copies of competitors’ price lists is circumstantial evidence that Cargill was conspiring with competitors to fix prices. Varni relies upon United States v. Container Corp. of America, 393 U.S. 333 , 89 S.Ct. 510 , 21 L.Ed.2d 526 (1969), in which the Supreme Court held that the reciprocal exchange of price information pursuant to an agreement by the defendants was concerted action sufficient to establish a price-fixing conspiracy. There is, however, a crucial difference between that case and the case before us. Unlike the plaintiffs in Container who presented evidence of an agreement to exchange pricing information, see Container, 393 U.S. at 340 , 89 S.Ct. 510 ; accord Baby Food, 166 F.3d at 125-26 , Varni has made no showing — or even any claim — that there was an agreement to exchange pricing information.

Although the possession of competitor price lists is consistent with conspiracy, it does not, at least in itself, tend to exclude legitimate competitive behavior. See Baby Food, 166 F.3d at 125 ; Wallace, 55 F.3d at 1169 ; Stephen Jay Photography, Ltd. v. Olan Mills, Inc., 903 F.2d 988, 996 (4th Cir.1990). There are many legal ways in which Cargill could have obtained pricing information on competitors. See United States v. United States Gypsum Co., 438 U.S. 422 , 441 n. 16, 98 S.Ct. 2864 , 57 L.Ed.2d 854 (1978); Petroleum Prods., 906 F.2d at 455 (“To allow an inference of agreement from such evidence, standing alone, would have the effect of deterring competitors from obtaining information about other companies’ actual retail prices, even through legitimate means. We can think of few inferences that would have a more undesirable distorting effect on market conduct.”). Varni has not, for example, produced evidence tending to exclude the possibility that Cargill received these price lists legitimately from customers after they were distributed by competitors as Cargill claims.

I

Varni contends that a 1994 H&R document recommending that H&R “gain share at accounts where no significant effect on ADM, Cargill, JBL share” is cir *1104 cumstantial evidence of Cargill’s role in the conspiracy because H&R would have been concerned about taking market share from Cargill only if they were conspiring with one another. Cargill responds with an innocent explanation' — -that H&R avoided competing directly with Cargill because H&R had consistently lost when it tried to do so due to Cargill’s lower prices, and that H&R apparently found it easier simply to compete with other firms instead. 8

In a similar case, plaintiffs produced what they characterized as direct evidence of a horizontal conspiracy. Upon close analysis, however, this court concluded that such evidence could be interpreted as evidence of a horizontal conspiracy but also could be construed in a benign light. We held that, in such a situation, the purported “smoking gun” was not sufficient in itself to preclude summary judgment because evidence as consistent with legitimate behavior as illegal behavior cannot independently support an inference of conspiracy. See Richards, 810 F.2d at 903-04 .

The Third Circuit recently reached a similar conclusion when it was confronted with purported smoking gun evidence in an antitrust price-fixing case involving the baby food industry. In Baby Food, an internal Heinz memorandum contained the following statement:

Baby Food, 166 F.3d at 120 . The Baby Food plaintiffs contended that the “truce” was evidence of a price-fixing conspiracy— namely, that Heinz’s hands were “tied” because they had agreed not to lower prices. The Third Circuit rejected this argument and concluded instead that “the single use of the term [‘truce’] in a highly competitive business environment and in the face of continuing fierce competition is as consistent with independent behavior as it is with price-fixing.” Id. at 127 .

Considering Varni’s interpretation of this evidence in light of Cargill’s explanation, we conclude that the district court did not err in ruling that this evidence did not tend to exclude the possibility that Cargill was acting lawfully.

J

Varni also cites scattered evidence from 1991 and 1992 which purportedly supports an inference that Cargill was a conspirator. In its briefs, however, Varni conceded that “[u]ntil mid-1993, Cargill competed with the Admitted Conspirators.” At oral argument, Varni argued that Car-gill joined the conspiracy in October 1992 — when Cargill indicated its receptiveness to joining EC AMA — rather than in May 1993 — when Cargill actually joined ECAMA. Even under this revised theory, all circumstantial evidence pre-dating October 1992 is irrelevant because it does not comport with Varni’s theory that Cargill joined the conspiracy in October 1992.

K

Varni points to the trial testimony of ADM employee Barrie Cox in a criminal proceeding involving a price-fixing conspiracy in the lysine industry, in which Cox stated that he had discussions regarding the “bidding price for certain [citric acid] accounts” with someone from Cargill, and maintains that Cox’s testimony constitutes direct evidence of price-fixing by Cargill. Cox’s testimony at the lysine trial does not constitute direct evidence, however, because it still requires an inference that the price discussions were conspiratorial in nature. Furthermore, Cox’s trial testimony, *1105 not subjected to cross-examination, does not possess the requisite specificity to withstand summary judgment. See Baby Food, 166 F.3d at 125 (“Evidence of sporadic exchanges of shop talk among field representatives who lack pricing authority is insufficient to survive summary judgment.”) (citing Krehl v. Baskin-Robbins Ice Cream Co., 664 F.2d 1348, 1357 (9th Cir.1982)).

Subsequent to oral argument in this case, Varni filed a request for judicial notice of and motion to supplement the record with the government’s sentencing memorandum in the lysine conspiracy case. Varni relies in particular on an FBI investigative report attached to the sentencing memorandum, which summarizes statements allegedly made by ADM employee Cox in an FBI interview. According to the report, Cox claimed that he had discussions with Cargill employee William Gruber regarding the bidding price for specific citric acid accounts. Varni argues that Cox’s statements, as described in the FBI report, constitute specific, direct evidence of price-fixing by Cargill in the citric acid industry.

.Cargill has vigorously opposed Varni’s request for judicial notice, offering several arguments as to why the substance of Cox’s FBI interview statements is not a proper subject for judicial notice. Cargill makes the following arguments: (1) Cox’s statements in the FBI report are inadmissible “triple hearsay”; (2) the statements are inadmissible under Federal Rule of Evidence 403 because they lack probative value; (3) as a general rule, the summary judgment record cannot be supplemented on appeal and is limited to material before the district court; and (4) under Federal Rule of Evidence 201, only facts “not subject to reasonable dispute” can be proper subjects of judicial notice, and Cox’s statements fail to satisfy this requirement. The arguments offered by Cargill against Varni’s request for judicial notice possess considerable force.

We need not resolve the dispute over whether judicial notice of this evidence would be proper, however, because the documents Varni seeks to add to the record would not change the result in this case, for several reasons. First, the documents support Cargill’s theory that Cargill was not part of the “Group of Four” or the “Group of Five” (the terms used by the citric acid conspirators in referring to themselves). Second, Cox’s statements in the FBI report show, at best, that Cox engaged in sporadic price discussions with one individual at Cargill, which evidence is neither sufficient to survive summary judgment under Baby Food nor probative of the industry-wide conspiracy alleged by Varni. Third, the documents upon which Varni seeks to rely do not comport with Varni’s theory that Cargill joined the conspiracy when it joined ECAMA in May 1993, since the FBI report on which Varni relies makes clear that the conspirators considered asking Cargill to join the conspiracy in 1995 — some two years after Cargill joined ECAMA. In sum, even if we were to grant Varni’s request for judicial notice, the new evidence would not alter the outcome because it appears to be inconsistent with Varni’s theory of the case.

L

Applying the legal framework set forth in Matsushita and developed through several subsequent cases in this circuit, we have concluded that none of the various pieces of evidence offered by Varni supports a reasonable inference that Cargill conspired to fix prices because none of the evidence, when considered individually, tends to exclude the possibility that Cargill acted independently. 9 The crucial ques *1106 tion, however, is whether the evidence considered as a whole can support an inference of conspiracy on Cargill’s part. As we stated at the outset of this analysis, such an inference may still be plausible when looking, as we must, at the big picture. Summing the iotas up, of course, makes it more difficult to say whether a reasonable factfinder could conclude that Cargill was more likely than not a conspirator, especially in light of the fact that this case is unique in that several citric acid manufacturers have already admitted to conspiring about prices.

Considered as a whole, the evidence in the record, though it clearly shows that several citric acid producers conspired to fix prices and to allocate market shares, does not tend to exclude the possibility that Cargill acted independently — and thus does not support a reasonable inference that Cargill was involved in the citric acid price-fixing conspiracy. Cargill has offered reasonable legitimate explanations for joining ECAMA, for its strategic decisions relating to expansion, for its citric acid pricing, and for the various other pieces of evidence offered by Varni purportedly showing that Cargill engaged in price fixing. We note that all four major citric acid manufacturers admitted to conspiring to fix prices but none identified Cargill as a co-conspirator. All in all, there is no more than a scintilla of evidence that Cargill was a participant in the citric acid conspiracy, and the existence of “a scintilla of evidence of concerted, collusive conduct” is not sufficient for Varni to overcome Cargill’s summary judgment motion. Baby Food, 166 F,3d at 137. 10 Thus, the district court did not err in granting summary judgment in favor of Cargill.

IV

Finally, we address whether the district court abused its discretion in denying Var-ni’s motion to compel Coopers & Lybrand L.L.P. (“C&L-US”) to produce various documents in its possession and in the possession of Societe Fiduciare Suisse Coopers & Lybrand (“C&L-Switzerland”), which had performed certain accounting and auditing work for ECAMA.

A

C&L-Switzerland and C&L-US are both members of Coopers & Lybrand International (“C&L International”), an association (organized under the laws of Switzerland) consisting of member accounting firms around the world (including C&L-US and C&L-Switzerland). By virtue of membership in C&L-International, member firms are allowed to utilize the Coopers & Lybrand name. Although members use the “Coopers & Lybrand” name, each firm is autonomous. Firms do not share profits or losses, nor do they have any management, authority, or control over other member firms. In addition, C&L-International does not exercise management, authority, or control over member firms. Of particular relevance to the case at hand, C&L-US does not have any economic or legal interest in C&L-Switzerland, and C&L-Switzerland has no such interest in C&L-US.

*1107 Varni served C&L-US with a subpoena to produce various documents in its possession and in the possession of C&L-Switzerland. C&L-US produced all the relevant documents in its possession and wrote a letter to C&L-Switzerland asking if it would voluntarily turn over the documents sought by Varni. C&L-Switzerland declined. Varni then filed a motion to compel C&L-US to produce C&L-Switzerland’s documents. The magistrate judge ruled that, under Federal Rule of Civil Procedure 45(a), Varni could compel only C&L-US to produce those records in its legal control. Finding that C&L-US lacked legal control over documents held by C&L-Switzerland, the magistrate denied the motion to compel in an order subsequently adopted by the district court. Varni argues that the district court abused its discretion by denying its motion to compel.

B

The Federal Rules of Civil Procedure require a party served with a subpoena for records to produce those records that are in its “possession, custody or control.” Fed.R.Civ.P. 45(a). The documents in question are neither in the possession nor custody of C&L-US. Thus, the only question before us is whether C&L-US has “control” over documents in the possession of C&L-Switzerland.

“Control is defined as the legal right to obtain documents upon demand.” United States v. International Union of Petroleum & Indus. Workers, 870 F.2d 1450 , 1452 (9th Cir.1989) (“International Union”). In International Union, the Department of Labor had served a subpoena on the International Union of Petroleum and Industrial Workers (“IUPIW”), seeking records from local unions affiliated with IUPIW. The court concluded that IUPIW lacked legal control over documents in the possession of local unions because IUPIW and each local union were separate entities under the law and because the contract governing the union relationship did not expressly give IUPIW the right to obtain the records of local unions upon demand. See id. at 1452-53. Although there was a theoretical possibility that IUPIW could dissolve a local union and thereby gain access to the documents of the dissolved union, the court deemed this remote possibility insufficient to give IUPIW “control” for purposes of Federal Rule 45(a). The court emphasized that proof of theoretical control is insufficient; a showing of actual control is required. See id. at 1453-54.

Varni concedes — as it must — that the district court’s ruling was correct under an application of the legal control test, because there is no dispute that C&L-US lacks the legal ability to obtain documents from C&L-Switzerland. Just as IUPIW and the local unions were separate entities under the law, so are C&L-US and C&L-Switzerland. Also as in International Union, there is no contract giving C&L-US the right to compel C&L-Switzerland to furnish it with documents in C&L-Switzerland’s possession.

Recognizing that the district court’s ruling is entirely consistent with International Union, Varni urges this court to reject the legal control test used in International Union. It asks us instead to define “control” in a manner that focuses on the party’s practical ability to obtain the requested documents and argues that International Union does not foreclose the application of this praetical-ability-to-obtain-documents test, because International Union’s decision to apply the legal control test rather than the practical-ability-to-obtain-documents test is purportedly “an unexamined assumption,” which stare de-cisis does not compel the present court to follow.

Even if International Union does not conclusively settle the question, we conclude — consistently with all of our sister circuits who have addressed the issue — that the legal control test is the proper standard under Rule 45. See Cochran Consulting, Inc. v. Uwatec USA, Inc., 102 F.3d 1224, 1229-30 (Fed.Cir.1996); In re Bankers Trust Co., 61 F.3d 465, 469 *1108 (6th Cir.1995); Chaveriat v. Williams Pipe Line Co., 11 F.3d 1420, 1426 (7th Cir.1993); Gerling Int’l Ins. Co. v. Commissioner, 839 F.2d 131, 140-41 (3d Cir.1988); Searock v. Stripling, 736 F.2d 650, 653 (11th Cir.1984). Ordering a party to produce documents that it does not have the legal right to obtain will oftentimes be futile, precisely because the party has no certain way of getting those documents. Varni claims that C&L-US has the practical ability to obtain documents from C&L-Switzerland because C&L-Switzerland has voluntarily furnished C&L-US with documents and information in the past. With respect to the ECAMA documents, however, C&L-US asked C&L-Switzerland to produce those documents, but C&L-Switzerland refused. There is no mechanism for C&L-US to compel C&L-Switzerland to produce those documents, and it is not clear how Varni wants C&L-US to go about getting the ECAMA documents, since C&L-Switzerland could legally — and without breaching any contract — continue to refuse to turn over such documents. Because C&L-US does not have legal control over C&L-Switzerland’s documents, Varni could not compel C&L-US to produce those documents.

V

For the foregoing reasons, we conclude that the district court did not err either by granting summary judgment in favor of Cargill or by denying Varni’s motion to compel.

AFFIRMED.

. In its briefs, Varni argued that it need show only a "slight connection” between Cargill and the conspiracy to survive summary judgment. See, e.g., United States v. Dunn, 564 F.2d 348, 357 (9th Cir.1977). At oral argument, however, Varni expressly disclaimed any reliance upon this theory.

. This second requirement flows directly from Matsushita . As then-Judge Kennedy explained in Richards :

Richards, 810 F.2d at 903 (citing Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 252, 106 *1095 S.Ct. 2505 , 91 L.Ed.2d 202 (1986); 6 P. Areeda, Antitrust Law ¶ 1405, at 21 (1986)).

. Varni also attempts to rely on a search warrant affidavit by FBI Agent Brian Shepard, which reports what a witness had told the FBI agent that a master had told the witness. The district court did not abuse its discretion in deciding not to consider this evidence because it was hearsay. See Quevedo v. Trans-Pacific Shipping, Inc., 143 F.3d 1255, 1258 (9th Cir.1998).

. Importantly, we note that Varni’s theory of the case is that Cargill conspired with the admitted conspirators to fix prices and to allocate market shares amongst themselves. Varni has not articulated a conspiracy theory wherein Cargill or the admitted conspirators actually conspired with Chinese producers to fix prices or to allocate market shares.

. This expansion was completed by 1994. Cargill also implemented improvements such as installing a liquid extraction recovery process, which increased the efficiency of the manufacturing process.

. For the same reason, we reject Varni’s assertion that illegal conduct can be inferred from Cargill’s failure to expand more rapidly in Europe.

. In fact, the only market share data actually in the record — and thus the only data properly before this court — indicates that Cargill’s market share continued to increase even after 1993.

. This document is not direct evidence of conspiracy because there remains a gap between the document and the conclusion that Cargill was a conspirator, a gap that can be bridged only by making some sort of inference. See Baby Food, 166 F.3d at 120-21 .

. Varni argues that the reports of its expert economist, Robert G. Harris, mandate the denial of Cargill’s summary judgment motion. Harris concluded that a company acting independently would not have done many of the things that Cargill did. Purported to be irrational absent conspiratorial behavior on the part of Cargill were Cargill’s decision to join ECAMA (the strongest piece of evidence in Harris's view), the virtual identity in list prices among citric acid producers, Cargill's failure to increase its market share between *1106 1993 and 1995, its failure to expand its annual production capacity by more than it did, its decision not to enter the European citric acid market sooner, and the fact that, "given the number and type of known contacts — to say nothing of other possible communications— between Cargill and personnel of the admitted conspirators, it is inconceivable that Car-gill remained ignorant of the existence of the conspiracy.” We have already considered all the evidence discussed by Harris, see supra Part III.A-III.L, and our conclusion that no piece of evidence tends to exclude the possi-bilily that Cargill acted independently is not affected by the conclusoiy statements in the expert report to the contrary. See Brooke Group, 509 U.S. at 242 , 113 S.Ct. 2578 .

. Because the evidence must, on summary judgment, be construed to the greatest extent reasonably possible in favor of finding an antitrust violation, rare is the antitrust case in which there is absolutely no evidence which could support some inference of an antitrust violation.