OPINION

GRABER, Circuit Judge:

Plaintiff Animal Legal Defense Fund filed a Freedom of Information Act ("FOIA") request with the Food and Drug Administration ("FDA") regarding egg-production farms in Texas. The FDA released almost 400 pages of documents but redacted data regarding total hen population, number of hen houses, number of floors per house, number of cage rows per house, number of cage tiers per house, and number of birds per cage for each farm in question. Plaintiff filed this FOIA action seeking to compel the FDA to release the redacted data. The district court ordered the release of information regarding the number of birds per cage at each farm. But the court held on summary judgment that, under FOIA Exemption 4, the FDA properly withheld the other categories of information because its release was "likely to cause substantial competitive harm." See 5 U.S.C. § 552(b)(4). We affirm.

FACTUAL AND PROCEDURAL HISTORY

In late 2011, Plaintiff submitted a FOIA request to the FDA that sought the following:

The FDA released records related to inspections of eleven chicken egg-production facilities; one quail egg-production facility and food manufacturer; one food warehouse; and one food distribution center. But redactions appeared on 277 of the 398 pages that the FDA produced.

Plaintiff filed a complaint for injunctive and declaratory relief under FOIA, 5 U.S.C. § 552, seeking to compel the production of the following information regarding inspected egg-production facilities: total hen population; number of hen houses; number of floors per house; number of cage rows per house; number of cage tiers per house; and number of birds per cage. The FDA moved for summary judgment on the ground that FOIA Exemption 4 — which applies to "trade secrets and commercial or financial information obtained from a person and privileged or confidential," id. § 552(b)(4) — protected the redacted data. In support of its motion, the FDA submitted several declarations from experts who stated that releasing the requested information would enable competitors to learn a given egg producer's production rate, which in turn would allow the competitors to undercut the egg producer's prices and lure away customers. Plaintiff filed a cross-motion for summary judgment, supported by its own declarations from an economist and a food industry consultant. Those experts attested that releasing the withheld information would not facilitate competitive underbidding.

Plaintiff also asked to suspend briefing of FDA's summary judgment motion in order to permit discovery directed to whether the information sought was publicly available. The district court denied that request because, among other things, Plaintiff had not shown that the discovery it sought "is essential to litigating the motion for summary judgment."

After briefing and oral argument, the district court granted in part and denied in part both parties' summary judgment motions. The district court held that the FDA had fallen short of showing how releasing the number of birds per cage would "threaten any competitive harm" and ordered disclosure of that information. But the court concluded that the FDA had established that the release of the other five categories of redacted information — total hen population, number of hen houses, number of floors per house, number of cage rows per house, and number of cage tiers per house — was likely to result in substantial competitive harm due to underbidding.

Plaintiff timely appeals the court's grant of summary judgment in favor of the FDA on the redaction of those five categories of information, as well as the denial of third-party discovery.

STANDARDS OF REVIEW

"Our review of a grant of summary judgment in a FOIA case ... is slightly different than for other types of cases...." Yonemoto v. Dep't of Veterans Affairs, 686 F.3d 681, 688 (9th Cir.2012). We first determine, de novo, whether an adequate factual basis supports the district court's decision. Id. "Whether a particular set of documents gives the court an adequate factual basis for its decision is a question of law that the court reviews de novo." Lion Raisins, Inc. v. U.S. Dep't of Agric., 354 F.3d 1072, 1078 (9th Cir.2004). If no adequate factual basis exists, the case must be remanded for further development of the record. Yonemoto, 686 F.3d at 688.

If such a factual basis exists, we next treat the judgment as "if it were a bench trial," so that "the district court's conclusions of fact are reviewed for clear error." Id. (internal quotation marks omitted). On the other hand, "legal rulings, including [the district court's] decision that a particular exemption applies, are reviewed de novo." Id. As we noted in Lion Raisins, whether withheld information could be used by a food producer to undercut competitors is a determination that is "grounded in ... findings of fact." 354 F.3d at 1078. Therefore, if we determine that the district court had an adequate factual basis for reaching its decision, we must review for clear error the district court's conclusion that releasing the redacted information likely would cause substantial competitive harm. Id.

We review for abuse of discretion a district court's denial of discovery before ruling on summary judgment. U.S. Cellular Inv. Co. of L.A., Inc. v. GTE Mobilnet, Inc., 281 F.3d 929, 934 (9th Cir.2002).

DISCUSSION

A. Disclosure Under FOIA

"Disclosure, not secrecy, is the dominant objective of FOIA." Shannahan v. IRS, 672 F.3d 1142, 1148 (9th Cir.2012) (internal quotation marks and brackets omitted). "We construe narrowly FOIA's nine exemptions." Id. The FDA relies on Exemption 4, 5 U.S.C. § 552(b)(4), "which is available to prevent disclosure of (1) commercial and financial information, (2) obtained from a person or by the government, (3) that is privileged or confidential."1 GC Micro Corp. v. Def. Logistics Agency, 33 F.3d 1109, 1112 (9th Cir.1994). Commercial information qualifies as "confidential" when disclosure is "likely ... to cause substantial harm to the competitive position of the person from whom the information was obtained." Id. 1112-13 (citing Nat'l Parks & Conservation Ass'n v. Morton, 498 F.2d 765, 770 (D.C.Cir.1974)).

B. Adequate Factual Basis

As noted above, we first must determine whether the district court had an adequate factual basis to reach its decision. Lion Raisins, 354 F.3d at 1079. "In making this determination, we may rely solely on government affidavits so long as the affiants are knowledgeable about the information sought and the affidavits are detailed enough to allow the court to make an independent assessment of the government's claim." Kowack v. U.S. Forest Serv., 766 F.3d 1130, 1132 (9th Cir.2014) (internal quotation marks omitted). That threshold is met here.

Several of the declarations by the FDA's experts stated that the egg-production industry was "highly" or "extremely competitive." One emphasized that "anything that changes costs by even a penny can make a huge difference." According to the experts, the redacted information was likely to cause substantial competitive harm because the competitors of the egg producers in question could use the information to form accurate estimates of each farm's or producer's rate of production and use those estimates to underbid. For example, one declarant stated that, once a competitor knows the production rate at an egg farm, the competitor is able to "enter the farm's regional market and offer to produce the same number of eggs per day for a lower price or a greater number of eggs per day for the same price and thereby lure away the farm's customers." As in Lion Raisins, 354 F.3d at 1079-80, the declarations in this case established an adequate factual basis. The declarations provided the district court with the identity of the information sought and the claimed exemption, and provided the necessary detail about the specific competitive harm that could arise from the release of the redacted information. See also Bowen v. FDA, 925 F.2d 1225, 1227-28 (9th Cir.1991) (holding that government affidavits that described the documents withheld, the statutory exemptions claimed, and the specific reasons for the agency's withholding provided an adequate factual basis for application of Exemption 4).

C. Review of District Court's Analysis for Clear Error

We next must decide whether the district court clearly erred in determining that the redacted information fell within Exemption 4's protection. "[The clear error] standard is significantly deferential, and we will accept the lower court's findings of fact unless we are left with the definite and firm conviction that a mistake has been committed." Lentini v. Cal. Ctr. for the Arts, 370 F.3d 837, 848-49 (9th Cir.2004) (internal quotation marks omitted).

"An agency seeking to withhold information under an exemption to FOIA has the burden of proving that the information falls under the claimed exemption." GC Micro Corp., 33 F.3d at 1113. "While conclusory and generalized allegations of competitive harm are insufficient to show that requested information is `confidential,'" the government need not show that releasing the documents would cause "actual competitive harm." Id. "Rather, the government need only show that there is (1) actual competition in the relevant market, and (2) a likelihood of substantial competitive injury if the information were released." Lion Raisins, 354 F.3d at 1079.

Plaintiff does not contest that there is actual competition in the egg-production market, and it also concedes that the redacted information could be used to estimate an egg farm's production capacity. The parties disagree, however, as to whether releasing the redacted information would likely cause "substantial competitive harm" to the affected egg producers and farmers.

Whether or not releasing the requested data would create a likelihood of substantial competitive harm was subject to dispute. But, on this record, the district court did not clearly err in finding that disclosure of the information was likely to cause commercial undercutting. The FDA provided declarations that explained how the information would facilitate accurate estimates of a farm's egg-production capacities and how those estimates could facilitate undercutting. For example, one declarant explained that the egg-production industry has a "tight profit margin"; industry experts estimate that an average profit is approximately 6.7 cents per dozen eggs sold. If a national egg producer were able to determine the production rates of its smaller competitors, it could direct its resources toward that market; and if the national producer were able to offer lower prices, "even a penny can make a huge difference" in the local company's ability to keep its customers.

Although the information sought may not provide a national egg producer with every piece of information that it would consider before entering a new market, knowing the production capacity of potential competitors could make the decision of whether or not to enter a competitor's market easier. By becoming aware of potential limitations in its competitors' production capabilities, a national producer could decide to focus all its resources on egg markets in which it could out-produce local competitors — whether in terms of efficiency, price, or total quantity. See Lion Raisins, 354 F.3d at 1081 (holding that releasing information that allows a raisin farmer to "infer the volume of its competitors' raisin sales" could facilitate undercutting and, therefore, create a likelihood of substantial competitive harm).

Plaintiff submitted its own declarations, which asserted that the production information it seeks is insufficient to affect the market. Nevertheless, under our special standard of review for FOIA cases, and in view of the extensive FDA affidavits, we see no clear error. The incomplete data could allow egg producers to make more accurate — if imperfect — estimates of their competitors' production capabilities and sales than they could without the redacted information. Due to the competitiveness of the egg-production industry, where "even a penny can make a huge difference," even a slight upgrade in the accuracy of projections might have a large effect on competition. Although the information may not afford egg producers their competitors' exact profit-per-egg statistics, the FDA need only establish, as the district court correctly noted, "a likelihood of substantial competitive harm, not a certainty."2

D. Third-Party Discovery

The district court did not abuse its discretion by denying third-party discovery. In response to a summary judgment motion, a non-moving party may obtain relief pursuant to Federal Rule of Civil Procedure 56(d) if it "shows by affidavit or declaration that, for specified reasons, it cannot present facts essential to justify its opposition." A party seeking further discovery must show that there is "some basis for believing that the information sought actually exists." Blough v. Holland Realty, Inc., 574 F.3d 1084, 1091 n. 5 (9th Cir.2009) (internal quotation marks omitted). Further, a party seeking discovery must show that it lacks the "essential facts" to resist the summary judgment motion. Cal. Union Ins. Co. v. Am. Diversified Sav. Bank, 914 F.2d 1271, 1278 (9th Cir.1990).

Plaintiff here sought additional discovery to show that the sought-after information was already publicly available. The district court ruled that the evidence Plaintiff sought was not sufficiently similar to the information requested through discovery; Plaintiff's request was grounded in speculation; and allowing discovery of "an individual farm's egg production could improperly give Plaintiff information that it could not obtain through its FOIA request." That ruling fell within the district court's range of discretion.

AFFIRMED.

PER CURIAM, concurring:

We write separately to explain why we think that our circuit should reconsider the standard of review that we apply to summary judgments in FOIA cases.

We generally review de novo a district court's grant of summary judgment. "Summary judgment is appropriate when, viewing the evidence in the light most favorable to the nonmoving party, there are no genuine questions of material fact and the district court correctly applied the underlying substantive law." Campbell v. PricewaterhouseCoopers, LLP, 642 F.3d 820, 824-25 (9th Cir.2011). Typically, of course, the district court does not make factual findings at summary judgment. Rand v. Rowland, 154 F.3d 952, 957 n. 4 (9th Cir.1998) (en banc).

In FOIA cases, by contrast, we allow the district court to make factual findings, and we review those findings for clear error. Schiffer v. FBI, 78 F.3d 1405, 1409 (9th Cir.1996). That peculiar standard means that a dispute of material fact does not necessarily defeat summary judgment. See Yonemoto v. Dep't of Veterans Affairs, 686 F.3d 681, 688 n. 5 (9th Cir.2012) ("Our cases do not explain why [we review for clear error], and one can question whether it should be. By definition, summary judgment may be granted only when there are no disputed issues of material fact, and thus no factfinding by the district court."). But we see no good reason to depart from our traditional standard of review in FOIA cases. See generally Rebecca Silver, Comment, Standard of Review in FOIA Appeals and the Misuse of Summary Judgment, 73 U. Chi. L.Rev. 731 (2006) (arguing that de novo review should apply in FOIA appeals).

As a threshold matter, "[s]ummary judgment is the procedural vehicle by which nearly all FOIA cases are resolved." Office of Information Policy, U.S. Dep't of Justice, Guide to Freedom of Information Act: Litigation Considerations 104 (2013). In this case, though, the parties presented contradictory declarations as to the likelihood of substantial competitive harm, making summary judgment an inappropriate vehicle for resolving that issue. See In Def. of Animals v. U.S. Dep't of Agric., 501 F.Supp.2d 1, 8 (D.D.C.2007) (stating that summary judgment in a FOIA case is "improper" when a "dispute is genuine and factual," even though the contention on which it is based may be "doubtful on the basis of the evidence before the court"); Pub. Citizen Health Research Grp. v. FDA, 953 F.Supp. 400, 403 (D.D.C.1996) (concluding that "contradictory" claims by the parties made summary judgment "an inappropriate vehicle" for resolution of a FOIA case and scheduling a bench trial).

Our past cases reasoned that we owe substantial deference to the district court in FOIA cases because of their unique nature. See, e.g., Assembly of Cal. v. U.S. Dep't of Commerce, 968 F.2d 916, 919 (9th Cir.1992). "Because there will rarely be any genuine issues of material fact — the document says whatever it says — the case may usually be decided on summary judgment." Id. To make its decision, the district court often reviews sensitive documents in camera, a process that we have described as "a trial on a hidden record." Id. The district court's characterization of the document in this context "more closely resembles a finding of fact than a conclusion of law." Id. Therefore, we grant substantial deference to the district court. Id.; see also Schiffer, 78 F.3d at 1409 ("[W]e endorsed the [clear error] standard because in FOIA cases the district court's findings of fact effectively determine our legal conclusions." (internal quotation marks omitted)).

Although the FOIA statute requires that district courts "determine the matter de novo," it is silent as to the appropriate standard of review for appellate courts. 5 U.S.C. § 552(a)(4)(B).1 We originally adopted our deferential standard of review in reliance on a D.C. Circuit Court's footnote, without explanation. See Church of Scientology of Cal. v. U.S. Dep't of Army, 611 F.2d 738, 743 (9th Cir.1980) (citing Mead Data Cent., Inc. v. U.S. Dep't of Air Force, 566 F.2d 242, 251 n. 13 (D.C.Cir. 1977)). The D.C. Circuit has since abandoned the FOIA-specific standard of review, and it now applies ordinary summary judgment principles in FOIA cases. See Petroleum Info. Corp. v. U.S. Dep't of Interior, 976 F.2d 1429, 1433 & n. 3 (D.C.Cir.1992) (noting that the D.C. Circuit "applies in FOIA cases the same standard of appellate review applicable generally to summary judgments" but that, in contrast, the Ninth Circuit applies "a clearly erroneous standard"). Likewise, the Second, Sixth,2 Eighth, and Tenth Circuits also apply de novo review when evaluating FOIA summary judgment decisions. See TPS, Inc. v. U.S. Dep't of Def., 330 F.3d 1191, 1194 n. 5 (9th Cir.2003) (collecting cases).

We acknowledge that some other circuits appear to use a deferential standard of review similar to our own. See Silver, 73 U. Chi. L.Rev. at 740-43. But those circuits all appear to have adopted the standard without explanation or analysis, and at least one has questioned whether a deferential standard of review is appropriate. See Stephenson v. IRS, 629 F.2d 1140, 1144 (5th Cir.1980); Chilivis v. SEC, 673 F.2d 1205, 1210 (11th Cir.1982); Antonelli v. DEA, 739 F.2d 302, 303 (7th Cir. 1984) (per curiam); Lame v. U.S. Dep't of Justice, 767 F.2d 66, 70 (3d Cir.1985); Willard v. IRS, 776 F.2d 100, 104 (4th Cir. 1985). But see Flightsafety Servs. Corp v. Dep't of Labor, 326 F.3d 607, 611 n. 2 (5th Cir.2003) (per curiam) (noting the circuit split and choosing not to take a firm stand because the case's outcome remained "the same whether the district court's judgment [was] reviewed de novo or for clear error").

De novo review would be consistent with our usual summary judgment standards. As the Second Circuit has explained, de novo review also is consistent with FOIA's history and purpose:

Halpern v. FBI, 181 F.3d 279, 288 (2d Cir.1999) (citations, internal quotation marks, and paragraph break omitted).

Even if we assume that the sensitive nature of documents withheld under a FOIA exemption calls for deference in some contexts, why we defer to the district court in cases such as this one — where the factual inquiry on which the summary judgment turns is one that does not depend on a review of withheld information — remains unclear. Here, the district court found that the release of the egg-production data was likely to cause substantial competitive harm by reviewing declarations and testimony that went well beyond, and depended little on, the redacted information. That review process did not concern what the documents said; rather, it centered on what effect their release would have because of the kind of data involved. The district court ultimately decided that the FDA's declarations were more persuasive than those submitted by Plaintiff. But the district court was in no better position to make that determination at summary judgment than we are on appeal. See Grand Cent. P'ship, Inc. v. Cuomo, 166 F.3d 473, 478 n. 2 (2d Cir. 1999) (holding that, in a FOIA case where "no witnesses were heard and no credibility findings were made," "the district court was in no better position to evaluate the record than" the circuit court).

In sum, if ordinary principles applied, summary judgment would not be appropriate because the record contains a disputed issue of material fact, and we would reverse and remand for further proceedings. Under our current FOIA standard, however, we must affirm. We urge our court to take up, en banc, the appropriate standard of review in FOIA cases.