OPINION

McKEOWN, Circuit Judge:

This appeal, which centers on tin mill products used to make the tin cans commonly used to package food, teaches that there's no substitute for concrete evidence. Stanislaus Food Products Company claims that it pays artificially high prices as the result of an illegal market allocation agreement among the nation's leading tin manufacturers who agreed to cede the tin mill products market in the western United States to a single company, USS-POSCO Industries ("UPI").

Although echoes of price-fixing permeate the appeal, only a market allocation theory is at issue. But the market-allocation claim relies on a shaky economic theory, as the purported arrangement would not be rational in light of the circumstances. Even after extensive discovery, the evidence does not tend to exclude the possibility that the alleged conspirators were acting independently. We conclude that Stanislaus has failed to establish specific facts supporting a market allocation conspiracy and affirm summary judgment for the tin manufacturers.

BACKGROUND

Stanislaus, a tomato cannery located in Modesto, California, purchases its tin cans exclusively from Silgan, one of three major American tin can manufacturers. These purchases occur under long-term contract.

Silgan, in turn, purchases tin mill products from multiple suppliers, including United States Steel Corporation ("U.S. Steel") and UPI. U.S. Steel manufactures and sells tin mill products as well as hot band steel, which is a component of tin mill products. U.S. Steel has nationwide supply contracts with Silgan and the other major tin can manufacturers. UPI is a joint venture equally owned by U.S. Steel and POSCO America Steel Corporation ("POSCO America").1 As part of the joint venture, U.S. Steel and POSCO America supply UPI with hot band steel. UPI's tin mill product prices are set by a six-person management team that includes three appointees each from U.S. Steel and POSCO America.

Stanislaus initiated suit in California state court, where it first alleged price fixing claims against UPI, Silgan, and other unnamed conspirators in violation of state and federal antitrust law. The case features a complicated procedural history that we need not recount in full here. Stanislaus's third amended federal complaint, which excludes Silgan as a defendant,2 asserts that U.S. Steel agreed to exit the market and eliminate certain discounts to Silgan to allow UPI to charge monopolistic prices for tin mill products. Stanislaus described three 2006 UPI management committee meetings as the setting for the agreement. The district court granted in part and denied in part a motion to dismiss the third amended complaint. Dismissing a separate conspiracy to monopolize claim, the court left only the market allocation conspiracy claim to go forward. The "linchpin" of the market allocation agreement, according to the district court, would be proof of U.S. Steel's exit of the market.

After extensive discovery, the defendants moved for summary judgment. The briefing prompted the district court to observe that "the parties ha[d] sharply different views on what this case is about." Stanislaus pursued two distinct antitrust theories. The first was an "exit theory" of the conspiracy, which turned on U.S. Steel's exit from the market for tin mill products in the western United States — an approach that mapped to the surviving complaint allegations. Stanislaus's second theory was a new "partial allocation theory," not raised in the complaint, under which Stanislaus claimed that U.S. Steel did not aggressively compete on price. Under this theory, "[b]ecause UPI was left unchallenged, U.S. Steel effectively `allocated' to UPI a dominant, but not complete, position in the market."

Evaluating both theories, the district court granted the defendants' motion for summary judgment. The court began by considering whether there was a plausible economic motive for either alleged agreement. The court concluded that either type of conspiracy would be rational for UPI and POSCO, which both stood to gain from any supra-competitive pricing charged by UPI, but irrational for U.S. Steel, which competes with UPI by selling tin mill products through national supply contracts under which its customers elect the delivery destination. U.S. Steel thus "lacks the ability to unilaterally price discriminate against the western United States market"; this market structure did not make the prospect of forfeiting competitiveness on a national level rational for U.S. Steel, even taking into account that U.S. Steel's 50% ownership in UPI meant that it would benefit to some extent from UPI's profits. Against this backdrop, and "proceed[ing] with caution," the court concluded that Stanislaus's circumstantial evidence was insufficient to support an inference of conspiracy.

ANALYSIS

Section 1 of the Sherman Antitrust Act renders illegal "[e]very contract, combination..., or conspiracy, in restraint of trade or commerce." 15 U.S.C. § 1. This case involves an alleged illegal agreement to allocate territory to UPI in order to reduce competition in the market for tin mill products. See Palmer v. BRG of Ga., Inc., 498 U.S. 46, 49, 111 S.Ct. 401, 112 L.Ed.2d 349 (1990) (per curiam) (citing United States v. Topco Assocs., Inc., 405 U.S. 596, 92 S.Ct. 1126, 31 L.Ed.2d 515 (1972)).

Before considering the evidence, we review the basic principles of summary judgment that guide this appeal. Summary judgment is appropriate "if the movant shows that there is no genuine dispute as to any material fact and the movant is entitled to judgment as a matter of law." Fed.R.Civ.P. 56(a). We view the facts and draw factual inferences in favor of Stanislaus, the non-moving party. T.W. Elec. Serv., Inc. v. Pac. Elec. Contractors Ass'n, 809 F.2d 626, 631 (9th Cir.1987). Still, to survive summary judgment, Stanislaus must establish a "genuine" factual dispute, which involves "more than ... some metaphysical doubt as to the material facts." Matsushita Elec. Indus. Co. v. Zenith Radio Corp., 475 U.S. 574, 586, 106 S.Ct. 1348, 89 L.Ed.2d 538 (1986); Fed.R.Civ.P. 56(a).

An agreement to restrain trade may be established by direct or by circumstantial evidence. See 7-Up Bottling Co. v. Archer Daniels Midland Co., Inc. (In re Citric Acid Litig.), 191 F.3d 1090, 1093 (9th Cir. 1999). Stanislaus relies upon circumstantial evidence to establish such an agreement here.

In Matsushita, the seminal case applying these principles to antitrust claims, the Supreme Court explained that to survive summary judgment on the basis of circumstantial evidence, "a plaintiff seeking damages for a violation of § 1 must present evidence `that tends to exclude the possibility' that the alleged conspirators acted independently." 475 U.S. at 588, 106 S.Ct. 1348 (quoting Monsanto Co. v. Spray-Rite Serv. Corp., 465 U.S. 752, 764, 104 S.Ct. 1464, 79 L.Ed.2d 775 (1984)). In other words, to establish the requisite factual dispute, Stanislaus "must show that the inference of conspiracy is reasonable in light of the competing inferences of independent action or collusive action that could not have harmed [Stanislaus]." Id.

Matsushita underscores that in making this determination, context is key. As the Supreme Court put it, "if the factual context renders [Stanislaus's] claim implausible — if the claim is one that simply makes no economic sense — [Stanislaus] must come forward with more persuasive evidence to support [its] claim than would otherwise be necessary." Id. at 587, 106 S.Ct. 1348.

As we proceed, we are mindful that "courts should not permit factfinders to infer conspiracies when such inferences are implausible." Id. at 593, 106 S.Ct. 1348 (citing Monsanto, 465 U.S. at 762-64, 104 S.Ct. 1464). The Supreme Court has observed that "mistaken inferences" are "especially costly" in cases that bear the risk of "chill[ing] the very conduct the antitrust laws are designed to protect." Id. at 594, 106 S.Ct. 1348 (citing Monsanto, 465 U.S. at 763-64, 104 S.Ct. 1464).

Because this case hinges on circumstantial evidence, our inquiry into whether Stanislaus's showing is sufficient to establish an agreement proceeds in two steps:

In re Citric Acid Litig., 191 F.3d at 1094 (citations omitted).

We begin by assessing the plausibility of Stanislaus's claims in light of their factual context. See Matsushita, 475 U.S. at 588, 106 S.Ct. 1348 (considering "the nature of the alleged conspiracy and the practical obstacles to its implementation"). In doing so, we cannot ignore the Supreme Court's prohibition on "permit[ting] fact-finders to infer conspiracies when such inferences are implausible," id. at 593, 106 S.Ct. 1348, lest we risk the "pernicious danger" of "deter[ring] pro-competitive conduct," In re Citric Acid Litig., 191 F.3d at 1094. Implausible claims require a "more persuasive" showing "that tends to exclude the possibility" of independent action. Matsushita, 475 U.S. at 587-88, 106 S.Ct. 1348. With these principles in mind, we turn to whether Stanislaus can establish its claim.

I. RATIONALITY AND JUSTIFICATIONS FOR THE CHALLENGED ACTIONS

In light of the shifting allegations, our first step is to identify the alleged conspirators and the exact subject matter of the alleged conspiracy. Stanislaus alleges that in 2006, UPI, POSCO, and U.S. Steel agreed to allocate the western United States market for tin mill products to UPI, enabling UPI to dominate the market and raise its prices.

As the district court noted, Stanislaus's exact theory of the alleged agreement has been in flux. The operative complaint alleges that "[a]fter 2006, pursuant to the Market Allocation Agreement, U.S. Steel exited the Relevant Market, and UPI faced no meaningful competition whatsoever." It also posits that U.S. Steel fully exited the market for the western United States and refused to enter into new long-term contracts with its largest customers in the region. By the time the parties briefed summary judgment, Stanislaus had shifted its theory to focus on U.S. Steel's failure to price competitively against UPI. This reframing came about because the evidence showed that U.S. Steel had never exited the market. In fact, U.S. Steel continued selling to Silgan (including for west coast delivery). Absent evidence of U.S. Steel's exit from the market, we focus our analysis on whether Stanislaus can show this second type of agreement.

We first analyze whether the alleged conspirators would have had a rational motivation to conspire.3 For UPI, the alleged agreement would certainly have made sense: it stood to gain by being able to charge higher prices in the absence of meaningful competition in the region. POSCO, a joint owner of UPI that does not sell tin mill products in the United States, similarly stood to gain by reaping the benefit of those increased prices. So far, so good for Stanislaus. But these incentives represent only one side of the alleged conspiracy.

The key to the alleged agreement — and to Stanislaus's case — is U.S. Steel's agreement to stop competing in that region. U.S. Steel occupies quite a different position than do the other alleged conspirators because it also manufactures and sells tin mill products throughout the United States. Critical to the context of this case, U.S. Steel has nationwide supply contracts with all of the major tin can manufacturers. These contracts and their timing are independent of any of the conspiracy claims. Under these contracts, U.S. Steel sells tin mill products F.O.B. U.S. Steel's mill, which means that the customer selects where U.S. Steel is to ship the products and pays for shipping costs to the chosen destination.

The structure of U.S. Steel's contracts means that the price and other terms are negotiated without U.S. Steel knowing whether a customer will request items be sent, say, to California or to New York. The geographic neutrality is a significant practical obstacle to the viability of the alleged allocation agreement: in order not to compete on price in the western United States, U.S. Steel would need to stop competing on price nationwide or refuse customers. Both options risk losses to U.S. Steel's bottom line and make little economic sense.

A scheme like Stanislaus alleges would not be rational unless U.S. Steel had little competition outside of the western United States or the potential payoff through ownership of UPI was likely to be significant. But other manufacturers compete in this market, including Arcelor-Mittal-Dofasco, Tata Steel (formerly, Corus), Rasselstein, Dongbu, Bao, JFE, and R.G. Steel (formerly, Severstal). In fact, most of these manufacturers have at some time after 2006 shipped products to Silgan's facilities in the western United States.

Stanislaus argues that it is "self-evident" that an alleged conspiracy was in U.S. Steel's economic interest because U.S. Steel did not in fact take market share from UPI, despite U.S. Steel's prices being lower. Even assuming Stanislaus's market analysis is correct, this theory boils down to retrospective guesswork; it says nothing of the economic rationality of the agreement in the first place. Additionally, the fact that U.S. Steel undercut UPI on price is consistent with competitive behavior. See Matsushita, 475 U.S. at 594, 106 S.Ct. 1348 ("[C]utting prices in order to increase business often is the very essence of competition.").

Silgan's position is a key aspect of Stanislaus's claim. UPI offers testimony that it was able to secure Silgan's agreement in 2008 to a long-term contract that raised Silgan's price for tin mill products because Silgan, a knowledgeable player in the market, wanted to lock in lower long-term prices. The price agreed to, while higher than previous prices, was still lower than those of other suppliers, with the exception only of U.S. Steel. Silgan wanted to avoid what its executives termed the "hockey stick" impact4 of a dramatic price increase that it expected would ensue if Silgan waited to renegotiate in 2010, when the prior contract expired.

As for U.S. Steel, the prices it charged Silgan were lower than the prices Silgan agreed to pay UPI. And the terms of the U.S. Steel-Silgan agreement were set before the alleged conspiracy began. A plausible justification for U.S. Steel's pricing strategy to Silgan is obvious: U.S. Steel was competing on price. The price increase by U.S. Steel to one of Silgan's competitors that Stanislaus relies on here not only did not involve Stanislaus's supplier, it also still left U.S. Steel's prices to that competitor lower than UPI's. The same competitive justification for pricing thus applies, with the increase in price explained by UPI's price leadership. See Areeda & Hovencamp ¶ 1429b (discussing how oligopolistic rationality can provide for price increases through price leadership).

Stanislaus argues that the fact that POSCO did not enter the western U.S. market is evidence of a conspiracy to allocate that market to UPI. But the fact that POSCO did not enter a new market just to compete with its own joint venture is perfectly justifiable, as "[f]irms do not expand without limit and none of them enters every market that an outside observer might regard as profitable, or even a small portion of such markets." Bell Atl. Corp. v. Twombly, 550 U.S. 544, 569, 127 S.Ct. 1955, 167 L.Ed.2d 929 (2007) (alteration in original) (quoting Areeda & Hovencamp ¶ 307d, at 155 (Supp.2006)).

The potential benefit of this scheme for U.S. Steel is just not apparent. To come out ahead, U.S. Steel would have had to gain more than it would lose from abandoning a competitive position in the western market for tin mill products. This prospect is speculative at best. Even factoring in the potential of charging UPI a higher price for hot band steel and the potential profits to be derived from U.S. Steel's 50% ownership in UPI, we note that U.S. Steel supplies just half of UPI's hot band steel. These factual circumstances do not alter our conclusion that U.S. Steel's participation in the alleged conspiracy is economically implausible.

The burden shifts to Stanislaus to establish its claim through "specific evidence tending to show that the [coconspirators were] not engaging in permissible competitive behavior." In re Citric Acid Litig., 191 F.3d at 1094.

II. EVIDENCE OF CONSPIRACY

Stanislaus identifies several categories of evidence that it argues "tend[] to show" the market allocation agreement. While some of the evidence may have a negative tinge, we conclude that the evidence does not, taken together, tend to exclude the possibility of permissible independent behavior. Matsushita, 475 U.S. at 588, 106 S.Ct. 1348. The notion, for example, that U.S. Steel did not compete hard enough or aggressively enough for Stanislaus's taste is difficult to assess or quantify. In any event, that characterization proffered by Stanislaus is not the applicable standard. The question we should instead ask is whether there is sufficient specific evidence of a market allocation agreement — the crux of Stanislaus's remaining antitrust claim. Considered against the backdrop of the market for tin mill products during this period, Stanislaus has failed to provide sufficient "specific evidence" to support a finding of conspiracy in light of the lack of economic rationality of the purported agreement. In re Citric Acid Litig., 191 F.3d at 1094.

A. Parallel pricing and pricing patterns

When UPI raised its prices, U.S. Steel also raised its prices to some customers. Stanislaus asserts that this allowed UPI to maintain artificially high prices in an agreed-upon absence of competition. Of course, "[p]arallel pricing is a relevant 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." Id. at 1102 (citation omitted).

Key to Stanislaus's theory is that this pricing behavior was contrary to U.S. Steel's self-interest. Specifically, Stanislaus argues that U.S. Steel's decision to raise its price to one of Silgan's competitors despite having lower hot band steel costs than in the previous year was not in U.S. Steel's economic self-interest. See In re Text Messaging Antitrust Litig., 630 F.3d 622, 628 (7th Cir.2010) (noting that falling costs increase a seller's profit margin and motivate the seller to reduce its price). This argument, which pinpoints a single year, fails to account for pricing patterns over multiple years. To get the full picture, we must juxtapose the price of tin against the cost involved in making that tin. In particular, the cost of hot band steel — a significant factor in tin manufacturing — only dropped in 2009 after having jumped, sharply, in 2008. The cost of making tin was therefore in flux, creating uncertainty for manufacturers. U.S. Steel may have raised tin prices in 2009 to make up for losses caused by skyrocketing material costs a year earlier — or as a hedge against future price increases in the market for component parts.

Of course, Stanislaus's critique of U.S. Steel's pricing behavior is also undercut by the fact that its prices were still lower than those of UPI. This "price point" had potential, even if unrealized, to attract customers away from UPI. This behavior is not credibly viewed as against U.S. Steel's self-interest or as tending to exclude independent behavior.5

B. Information exchanges and other communications

To support an inference of conspiracy, Stanislaus offers evidence of information exchanges between U.S. Steel and UPI. Interfirm communications, particularly among high-level executives, are a "plus factor" that can bolster the inference of conspiracy. See In re Publ'n Paper Antitrust Litig., 690 F.3d 51, 62 (2d Cir. 2012); In re Flat Glass Antitrust Litig., 385 F.3d 350, 368-69 (3d Cir.2004).

Executives at U.S. Steel were aware of the contract between UPI and Silgan. In 2008, the president of UPI emailed U.S. Steel executives an overview of the price and other terms associated with the new contract.

The alleged conspirators argue that these communications are irrelevant without evidence that they affected U.S. Steel's price to Silgan in 2009. See In re Flat Glass Antitrust Litig., 385 F.3d at 369; In re Baby Food Antitrust Litig., 166 F.3d 112, 125 (3d Cir.1999) (requiring "evidence that the exchanges of information had an impact on pricing decisions"). To show the lack of such an impact, the alleged conspirators point to a provision in the 2005 agreement between U.S. Steel and Silgan that limits subsequent price increases for tin mill products to 2% in any given year.

Exchanges of information like this call to mind the requirement that Stanislaus "must show more than a conspiracy in violation of the antitrust laws; [it] must show an injury to [it] resulting from the illegal conduct." Matsushita, 475 U.S. at 586, 106 S.Ct. 1348. Absent deviation from the contracted price between U.S. Steel and Silgan, we agree that any negative inference is more difficult for Stanislaus to show and that the interfirm communications are of limited use. U.S. Steel's knowledge of the new contract between its joint venture and Silgan has little legal significance if it did not prompt action on U.S. Steel's part.

Even if the U.S. Steel-Silgan contract was not affected, Stanislaus argues that this information prompted U.S. Steel to change the pricing for Silgan's competitors. Despite this relatively attenuated chain of cause-and-effect, we nonetheless take a deeper look at this allegation.

In particular, Stanislaus asks that we take note of internal communications that prove that UPI "made pricing decisions based on its understanding of [U.S. Steel's] agreement not to compete on price" in the region. In August 2008, the president of UPI wrote, "There is a lot going on regarding 2009 industry pricing for tin .... Right now our position on 2009 West Coast pricing for everyone other than Silgan is major increases consistent with Arcelor Mittal's announcement and our understanding of what U.S. Steel is doing, plus a major additive for freight to the West Coast." UPI internally predicted that "[h]igh [p]rices and [t]ransportation [c]osts [w]ill [k]eep [e]astern [s]teel in the [e]ast." These communications arose in the context of observations about the difficult state of the tin market in the previous year due to skyrocketing hot band steel prices, increased fuel and freight costs, inexpensive imports, and fluctuations in demand.

None of this evidence bolsters the inference that Stanislaus would have us draw from these communications. Considered against the backdrop of market conditions in the industry, it is ambiguous what the president of UPI meant when he wrote of "our understanding of what U.S. Steel is doing." Stanislaus would have a jury infer that "what U.S. Steel [was] doing" was deliberately failing to price competitively, as agreed. It is just as plausible, if not more so, that the phrase referred to what U.S. Steel was doing in the face of the same challenging conditions in the tin market at the time. Although we do not totally discount the evidence as irrelevant, the evidence is insufficient to support the existence of a market allocation conspiracy.

C. Spot market sale

Stanislaus also points to evidence about a spot market sale, a circumstance in which buyers individually negotiate one-time sales. But the evidence offered here of a single spot market sale from U.S. Steel to one of Silgan's competitors — thus not involving Stanislaus's supplier at all — adds too little to qualify as "specific" evidence in support of a conspiracy. In 2008, U.S. Steel sold 1,000 tons of tin mill products to one of Silgan's competitors on the spot market. The customer directed the tin mill products to be shipped to the western United States. The sale price was higher than the national contract price between U.S. Steel and that same customer; U.S. Steel would not negotiate despite the customer's complaints about the high price. Although U.S. Steel's one-time price does not appear to have been as competitive as its prices set by long-term contract, what is missing is a link to competitive pricing vis-a-vis UPI. The record does nothing to place the sale in context — for example, it does not reflect how frequently these types of sales occurred or whether high spot market prices somehow displaced its other sales. We thus treat the spot sale as somewhat of an outlier.

Several U.S. Steel executives communicated internally about this sale. Upon learning of the high price, U.S. Steel's CEO inquired, "Is this because UPI is full?" Another U.S. Steel executive answered, "I am told this is NOT a supply issue — but a price issue." In another exchange, one U.S. Steel executive commented, "not sure this is high enough but OK now," to which another wrote, "Greed is a deadly sin."

The meaning of these emails is ambiguous at best, but it does not bear on the key question of whether U.S. Steel agreed to or did stop competing with UPI.

Perhaps anticipating that nothing concrete can be inferred from the emails, Stanislaus characterizes the exchanges as exhibiting an "incongruous[ly] high level of senior management attention" to what it calls a "de minimis order" when compared to U.S. Steel's total annual production of 1.2 million tons. This subjective gloss on a single aspect of the evidence does not implicate the claimed conspiracy or its impact on pricing to Stanislaus.

D. Expert report

Finally, Stanislaus argues that the district court improperly ignored its expert report. The district court reviewed "the two-page portion of Dr. Rausser's report that Plaintiff direct[ed] the Court's attention to," but went on to note that it "is nothing more than a broad overview of Plaintiff's argument." The district court declined to mine the 82-page report for "substantive opinions that are not cited in its briefing," citing Carmen v. San Francisco Sch. Dist., 237 F.3d 1026, 1031 (9th Cir.2001), for the proposition that courts "need not examine the entire file for evidence establishing a genuine issue of fact, where the evidence is not set forth ... with adequate references so that it could conveniently be found."

The explanation that Stanislaus now offers linking the summary conclusions to various components of the expert report is too little, too late. Summary judgment proceedings focus on whether there are genuine issues of fact. Specific citations, not bulk references, are essential to pinpoint key facts and factual disputes. The district court was not required to put the puzzle together from a boxful of facts, and, in line with Carmen, may permissibly decide the motion without mining the entire document for more substantiation.

Nonetheless, on de novo review, and despite the breezy reference, we give Stanislaus the benefit of a full review. In doing so, we are sympathetic to the district court's predicament. Not only is the report 82 pages, but the evidentiary references are found, without much detail, in 285 footnotes, a listing of more than 180 sets of documents, multiple websites, and an array of other evidence.

Although the expert report is well crafted and provides useful background on the optics of the industry, it also contains extensive discussion related to price-fixing claims that are not at issue in the appeal. Even Stanislaus's belated explanation that the expert summary relates to two broad sections of the report does little to identify factual issues or cite to specific evidence supporting Stanislaus's argument. This circumstance highlights why the appellate rules require specific citations to the record. Fed. R.App. P. 28(a)(8)(A), 28(e); 9th Cir. R. 28-2.8. However, consideration of the evidence (which we dug out ourselves) dealing with market allocation does not change our analysis. The broad conclusions and cited evidence essentially mirror Stanislaus's arguments elsewhere and nothing in the market analysis rescues Stanislaus's case.6

CONCLUSION

In sum, Stanislaus failed to meet its burden to show "specific evidence" of a market allocation agreement.

AFFIRMED.