Opinion for the Court filed by Chief Judge GINSBURG.
Opinion concurring in part and dissenting in part filed by Circuit Judge GARLAND.
GINSBURG, Chief Judge:McDonnell Douglas, a wholly owned subsidiary of Boeing, appeals from a judgment in favor of the Air Force in this “reverse” Freedom of Information Act case. McDonnell Douglas challenges as arbitrary and capricious and otherwise contrary to law the decision of the Air Force to release to Lockheed Martin Aircraft Center pricing information contained in the contract the Air Force awarded to McDonnell Douglas for the maintenance and repair of KC-10 and KDC-10 aircraft. We affirm the judgment of the district court insofar as it upheld the decision of the Air Force to release the Over- and Above Work Contractor Line-items (CLINs); we reverse that judgment insofar as it approved release of option year prices and Vendor Pricing CLINs.
I. ■ Background
In 1997 the Air Force issued a request for proposals (RFP) to perform maintenance and repair work on its fleet of KC-10 and KDC-10 aircraft. McDonnell Douglas submitted a bid which, in compliance with the requirements of the RFP, contained detailed pricing information both for the base year of the proposed contract and for subsequent years in which the Air Force would have the option to renew the contract. In June 1998 the Air Force awarded the contract to McDonnell Douglas. ; The contract, which provided for a base year and eight option years, incorporated the pricing information McDonnell Douglas had submitted in its bid.
. In July 1998 Lockheed asked the Air Force, pursuant to the FOIA, 5 U.S.C. §§ 551 et seq., for a copy- of the contract. The-Air Force duly notified McDonnell Douglas of Lockheed’s request, and McDonnell Douglas promptly objected.
Although ' McDonnell Douglas agreed some of the requested information must be disclosed, including the bottom-line price for the base year of the contract, the. Company maintained the option year prices and the prices listed Jn certain of the CLINs came within Exemption 4 of the FOIA, which exempts from disclosure “trade secrets and commercial or financial information obtained from a .person and privileged or confidential.” 5 U.S.C. § 552(b)(4). Exemption 4 does not itself prohibit an agency from disclosing commercial or financial information; it provides only that an agency is not compelled to disclose such information. The Trade Secrets Act, 18 U.S.C. § 1905, however, the scope of which is “at least coextensive with 'Exemption 4,” CNA Financial Corp. *1186v. Donovan, 830 F.2d 1132, 1151 (D.C.Cir.1987), effectively prohibits an agency from releasing information subject to the exemption. See McDonnell Douglas Corp. v. Widnall, 57 F.3d 1162, 1164 (D.C.Cir.1995) ("whenever a party succeeds in demonstrating that its materials fall within Exemption 4, the government is precluded from releasing the information by virtue of the Trade Secrets Act").1
Over a period of two years McDonnell Douglas sent the Air Force 11 submissions advancing its argument that option year prices, Vendor Pricing CLINs, and Over and Above Work CLINs are exempt from disclosure. Unpersuacled, the Air Force issued a Final Administrative Decision Letter in June 2000 concluding Exemption 4 was inapplicable and stating it would release the contract pricing information to Lockheed.
McDonnell Douglas then filed in district court a two-count complaint alleging the Final Decision (1) was arbitrary and capricious and contrary to law, in violation of the Administrative Procedure Act, 5 U.S.C. § 706(2)(A), and (2) violated the Trade Secrets Act. Upon cross-motions for summary judgment the district court held the decision of the Air Force to release the option year prices and the disputed CLINs "was not arbitrary or capricious," and hence did not violate the APA. McDonnell Douglas Corp. v. U.S. Dep't of the Air Force, 215 F.Supp.2d 200, 203 (D.D.C.2002). The court also granted summary judgment to the Air Force on McDonnell Douglas's claim under the Trade Secrets Act on the ground the Act "does not afford `a private right of action to enjoin disclosure in violation of the statute.'" Id. at 204 n. 2 (quoting Chrysler, 441 U.S. at 316-17, 99 S.Ct. at 1724). McDonnell Douglas now appeals, advancing only its APA claim.
II. Analysis
As an initial matter McDonnell Douglas contends the Air Force misapplied the governing legal standard in determining that the option year prices, the Vendor Pricing CLINs, and the Over and Above Work CLINs are not exempt from disclosure pursuant to Exemption 4. McDonnell Douglas also argues the decision of the Air Force to disclose those data was arbitrary and capricious and contrary to law.
A. Standard of Review
We review de novo the district court's grant of summary judgment. See LaCedra v. Executive Office for United States Attorneys, 317 F.3d 345, 347 (D.C.Cir.2003). In determining whether the decision of the Air Force was arbitrary and capricious, we do not "substitute [our] judgment for that of the agency." Motor Vehicle Mfrs. Ass'n of United States, Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43, 103 S.Ct. 2856, 2866, 77 L.Ed.2d 443 (1983). On the other hand, *1187we do not defer to the agency's conclusory or unsupported suppositions. Id.
B. The National Parks Standard
That McDonnell Douglas was required to provide to the Air Force the option year prices and the information in the CLINs in order to compete for the contract is undisputed. That is no doubt why the parties agree the standard set out in National Parks & Conservation Association v. Morton, 498 F.2d 765 (D.C.Cir.1974) (National Parks 1), governs whether the contested information falls within the scope of Exemption 4. In National Parks I, we held financial information is "confidential" and therefore within the scope of Exemption 4 if it is required to be submitted to the Government and if its disclosure is "likely to cause substantial harm to the competitive position of the person from whom the information was obtained." Id. at 770.
McDonnell Douglas argues the Air Force misapplied the test of National Parks I because it required McDonnell Douglas to demonstrate "with certainty" release of the contested information would cause the Company substantial competitive harm. Indeed, according to McDonnell Douglas the Final Administrative Decision Letter is "riddled with demands for certainty." National Parks I, of course, does not require the party invoking Exemption 4 to prove disclosure certainly would cause it substantial competitive harm, but only that disclosure would "likely" do so. See id.; Gulf & W. Indus. v. United States, 615 F.2d 527, 530 (D.C.Cir.1979).
As we read the Final Administrative Decision Letter, the Air Force did not misinterpret or misapply the test of National Parks I. On the contrary, it expressly concluded "release of the requested information will not likely cause substantial competitive harm to [McDonnell Douglas]." (Emphasis in original). Although the Air Force did not use the word "likely" at every opportunity in the course of its analysis of whether disclosure would cause substantial competitive harm, it is quite clear the agency knew what was required to meet the National Parks I standard and sought to apply that standard accordingly.
McDonnell Douglas seizes upon the Air Force's statement that a competitor could not reverse-engineer the Company's commercially sensitive information "with certainty," but it does not follow that the Air Force required McDonnell Douglas to demonstrate substantial competitive harm would certainly occur. The issues are distinct: even if a competitor could determine the pricing strategy and markups McDonnell Douglas used in bidding for the present contract, the question would remain whether as a result McDonnell Douglas would likely suffer substantial competitive harm, with respect either to the option years in this contract or to future contracts. Therefore, we ~an not agree with McDonnell Douglas that the Air Force committed a wholesale error in applying the standard laid down in National Parks I, although we disagree with certain of the agency's more particularized conclusions, to which we now turn.
C. Option Year Prices
McDonnell Douglas argues release of the option year prices in the contract would likely cause it substantial competitive harm for two reasons. First, McDonnell Douglas anticipates its competitors would use their knowledge of those prices in an effort to convince the Air Force to rebid the contract rather than exercise its option annually to renew it. As the Air Force points out, however, lVlcDonnell Douglas failed to make this argument before the agency. Whether the Final Decision of the Air Force was arbitrary and capricious must be determined solely upon *1188the basis of the arguments and information before the agency at the time. See Walter O. Boswell Mem'l Hasp. v. Heckler, 749 F.2d 788, 792 (D.C.Cir.1984) ("If a court is to review an agency's action fairly, it should have before it neither more nor less information than did the agency when it made its decision"). McDonnell Douglas's argument that only the agency - and not the party challenging the agency's decision - is prohibited from advancing a post hoc argument is simply incorrect. See Military Toxics Project v. EPA, 146 F.3d 948, 957 (D.C.Cir.1998) (by failing to raise argument to agency appellant forfeited the argument and "may not raise it for the first time upon appeal").
Second, McDonnell Douglas argues disclosure of the option prices in the contract likely will cause it substantial competitive harm because, in the event the Air Force does decide to rebid the contract, its competitors will be able to use that information to underbid it. The Air Force responds with two reasons for which rebidding is unlikely. First, "option years of contracts are usually exercised," and a contract to service military aircraft is "even less susceptible to a new competition on the basis of price than most [contracts]." Second, "under the standards mandated by the [Federal Acquisition Regulations], the option years of the contract will be exercised unless the market changes considerably," in which event McDonnell Douglas would also "necessari-iy change [its] pricing strategy" and thereby "greatly diminish any potential competitive value" of knowing the option year prices. Now it is McDonnell Douglas's turn to point out that neither of these considerations played any role in the decision here under review: the Air Force never suggested McDonnell Douglas would not likely be harmed because the contract would not likely be rebid. We do not rely upon counsel's post hoc rationale for upholding an agency's action. See Bowen v. American Hosp. Ass'n, 476 U.S. 610, 626-27, 106 S.Ct. 2101, 2111-12, 90 L.Ed.2d 584 (1986); Yukon-Kuskokwim Health Corp. v. NLRB, 234 F.3d 714, 718 (D.C.Cir.2000).2
The Air Force also suggests three reasons even a rival bidder that did know McDonnell Douglas's option year prices would not know what it would take to underbid McDonnell Douglas. First, the Air Force claims a rival would never know the exact "price to beat" because McDonnell Douglas's nominal option prices are subject to revision pursuant to the Economic Price Adjustment Clause in the contract. As McDonnell Douglas points out, however, the index used to determine the final option price in each option year is identified on the face of the Adjustment Clause and is publicly available. If the option year prices in the contract were released, then it would be a matter of simple arithmetic to calculate the adjusted option year prices. See Greenberg v. FDA, 803 F.2d 1213, 1218 (D.C.Cir.1986) (combination of allegedly confidential information and publicly available information sufficient evidence of competitive harm to defeat summary judgment motion).
Second, the Air Force contends a rival's price, to be attractive, would have to be lower than McDonnell Douglas's bid by enough to offset the "costs of disrupting *1189operations.” This transaction cost argument, however, did not figure in the agency’s decision that McDonnell Douglas would not likely be harmed by release of its option year prices, and hence can play no part in our review.
Finally, the Air Force argues that, even if it does rebid the contract, McDonnell Douglas is unlikely to suffer competitive harm because the new bid price “would be only one of several evaluation factors for award” of the new contract. The RFP for the existing contract listed the six criteria for evaluating bids — logistics, maintenance/repair/modifications, management, safety/fire protection, quality, and cost/ price — which were “to be weighted equally” in selecting the winning bid. The Air Force contends, therefore, any rebid contract would be awarded to the bidder that offered the “best value” based upon all six factors and not necessarily to the bidder with the lowest price.
As Boeing correctly points out, we considered and rejected this argument in McDonnell Douglas Corp. v. NASA, 180 F.3d 303 (1999). In that case the agency argued disclosure of the contested information would not cause McDonnell Douglas to be underbid “because price is only one of many factors used by the. government in awarding contracts.” Id. at 306. We thought the argument “too silly to do other than to state it, and pass on,” id., but the argument’s resurrection here suggests we should take greater pains to explicate the problem with the Air Force’s position.
Simply put, release of the option year prices in the present contract would likely cause McDonnell Douglas substantial competitive harm because it would significantly increase the probability McDonnell Douglas’s competitors would underbid it in the event the Air Force rebids the contract. . See Gulf & W. Indus., 615 F.2d at 530 (substantial competitive harm likely where disclosure “would allow competitors to estimate, and undercut, its bids”). Because price is the only objective, or at least readily quantified, criterion among the six criteria for awarding government contracts, submitting the lowest price is surely the most straightforward way for a competitor to show its bid is superior. Indeed, price is by statute the only factor that “must be considered in the evaluation of a proposal.” 10 U.S.C. § 2305(a)(3)(A)(ii) (emphasis added). Whether price will be but one of several factors to be weighted equally in any future RFP, therefore, is necessarily somewhat speculative.
The Air Force nonetheless presses the view, which the district court accepted, that its present argument “differs markedly” from the argument we rejected in NASA. McDonnell Douglas, 215 F.Supp.2d at 208. The district court restated the Air Force’s present argument as follows: “even if underbidding were to occur, the fact that price is just one of many factors means that the effect of the underbidding would be diluted by other factors.” Id. The district court reasoned that unlike the NASA, which had argued competitors would not underbid McDonnell Douglas, the Air Force acknowledged that competitors might underbid McDonnell Douglas but argued the Company would not be harmed because the Air Force would consider nonprice factors in awarding the contract. Id. The district court clearly used the term “underbid” to mean “bid a lower price,” but that is not how we used the term in NASA; there “underbidding” refers to making a bid more attractive overall, not with respect only to price, so it will be chosen by the Government. See NASA, 180 F.3d- at 303 (“[T]he agency ‘reasoned’ that underbidding due to the disclosure would not occur because price is only one of the many factors used by the government in awarding contracts”). Obviously, any competitor would try to bid a lower price than *1190McDonnell Douglas; the NASA was not so foolish as to deny that. Instead, the agency there made the same argument as does the Air Force here, viz., that the NASA would not accept a bid merely because it offered a lower price but instead would consider nonprice factors in selecting the winning bid. We rejected this argument summarily in NASA, and we reject it again here for the reason given in the preceding paragraph.
We conclude disclosure of McDonnell Douglas’s option year prices would likely cause McDonnell Douglas substantial competitive harm by informing the bids of its rivals in the event the contract is rebid.3 Consequently, the option year prices fall within the scope of Exemption 4, and the decision of the Air Force to release them was contrary to law.
D. Vendor Pricing CLINs
McDonnell Douglas argues the release of prices for certain CLINs composed predominantly of the costs of materials and services it procures from other vendors would enable its competitors to derive the percentage (called the “Vendor Pricing Factor”) by which McDonnell Douglas marks up the bids it receives from subcontractors. McDonnell Douglas argues release of the disputed CLINs would likely harm its competitive position because Lockheed “most likely obtained quotations from the same vendors,” with “the same or nearly the same pric[es].”
In explaining its decision to release those CLINs the Air Force stated it is “entirely possible,” indeed “not uncommon,” for a subcontractor to “quote different prices ... to- different prime contractors.” Therefore, the Air Force reasoned, McDonnell Douglas’s competitors could not “with any degree of certainty” derive its Vendor Pricing Factor from disclosure of the Vendor Pricing CLINs and hence McDonnell Douglas is unlikely to suffer substantial competitive harm.
The problem with this line of reasoning is its premise, namely, the mere supposition that McDonnell Douglas and Lockheed received — or may well have received — significantly different prices from the same vendors bidding for the same subcontract. The Air Force provided no actual evidence, nor did it claim special knowledge based upon its experience, to support this proposition, apart from the rather casual observations that it was “entirely possible” and “not uncommon.” This is tantamount to the Air Force saying it would “not be surprised” if the rival bidders had received different prices from subcontractors, but it is far short of asserting (let alone substantiating) that it was “likely” they did so.4 Nor does it seem probable as a matter of economic *1191theory. In a competitive market for subcontracted work, a rational subcontractor would quote each prime contractor the lowest price consistent with covering its costs. Any difference in the prices it quotes different prime contractors should, in theory, reflect differences in the costs of supplying them. Therefore, it is reasonable to presume, as McDonnell Douglas did in its submissions to the Air Force, that its “competitors obtained similar pricing from various vendors to support those tasks.”5
Perhaps the Air Force, which we recognize has “knowledge of the government contracting industry,” McDonnell Douglas, 215 F.Supp.2d at 208, has reason to believe the markets in which its prime contractors purchase goods and services are not effectively competitive.6 Having failed to explain how its knowledge or experience supports that understanding, however, the decision of the Air Force is neither “at least as compelling as McDonnell Douglas’s,” McDonnell Douglas, 215 F.Supp.2d at 209, nor “well-reasoned, logical[,] and consistent,” CNA Financial, 830 F.2d at 1155; it is, in short, arbitrary and capricious.
E. Over and Above Work CLINs
Finally, McDonnell Douglas challenges as arbitrary and capricious the decision of the Air Force to disclose CLINs containing the hourly labor rates McDonnell Douglas charges the Air Force for “Over and Above Work,” that is, maintenance and repair work not required under the contract, which the Air Force is not required to direct to McDonnell Douglas. The Company first complains a competitor with knowledge of these rates would be able to underbid it for Over and Above Work. Because .McDonnell Douglas did not make this specific argument to the Air Force prior to its issuance of the Final Decision, we shall not consider it. McDonnell Douglas also complains, however, that knowledge of these CLINs “would clearly place such a competitor at a distinct advantage over [McDonnell Douglas], in any contract (commercial or military) awarded *1192on the basis of a price comparison.” Specifically, McDonnell Douglas claims it was “no secret” — based upon local newspaper articles published in 1996 — the Company paid the “going wage” at its new Boeing Aerospace Support Center (BASC), located at the former Kelly Air Force Base in San Antonio; releasing the labor rates it charges for Over and Above Work, therefore, would enable its competitors, to calculate the “overall markup (or labor pricing factor)” McDonnell Douglas uses, to its detriment in bidding on future contracts.
In an effort both to make sense of and to refute this claim, the Air Force responds that a competitor could not “safely assume” McDonnell Douglas either pays its blue collar employees the “prevailing wage” or meets the “standard for fringe benefits” in the area, both as determined by the Department of Labor. • Moreover, the Air Force found McDonnell Douglas submitted “significantly different” rates for Over and Above Work in the present contract and in a contract to service KC-135 aircraft at the BASC; a competitor therefore could not reasonably infer the Over and Above Work CLINs in the present contract accurately reflect the Labor Pricing Factor McDonnell Douglas will use in bidding on some future contract.
McDonnell Douglas presents neither a viable theory nor any evidence to support its claim that release of the Over and Above Work CLINs would enable a competitor'to derive its Labor Pricing Factor. The 1996 newspaper articles McDonnell Douglas submitted to the Air Force to prove its competitors know how much the Company pays its mechanics are not persuasive. For example, a newspaper report that the “average blue collar worker” at Kelly AFB earned $28,352 per year two years before the BASC was established there,see http://www.boeing.com/defenses-pace/aerospace/maintenance/basc.html, does not reveal anything about the prevailing wage for McDonnell Douglas employees performing work under the contract at the BASC — then or now. Nor does McDonnell Douglas present any evidence that the cost of the benefits it currently provides to mechanics is publicly known; the average salary at Kelly AFB reported in 1996 specifically “exclud[ed] benefits,” and incoming contractors (such as McDonnell Douglas) had announced that “retirement and vacation [benefits] would differ from what civil servants [were then receiving].”
Based upon the publicly available information the Company submitted to the Air Force, the agency reasonably concluded McDonnell Douglas failed to carry its burden of showing release of the Over and Above CLINs was likely to cause it substantial competitive harm. Therefore, the decision of the Air Force to release the CLINs was not arbitrary and capricious. See CNA Fin. Corp., 830 F.2d at 1155-56.
F. Business Judgment and Risk Assessment
In a coda to its brief, the Air Force argues disclosure of “the type of pricing information that is at issue” in this case is unlikely to cause substantial competitive harm to McDonnell Douglas because a rival company can never understand fully or model precisely the “business judgment and risk assessment” that go into another firm’s pricing decisions.
We recoil, as does McDonnell Douglas, from the implication of this argument, namely, a per se rule (or at least a strong presumption) that all constituent pricing information — as opposed to the bid price itself — is to be disclosed; such a rule would be squarely at odds with the protection we have always understood Exemption 4 to provide for such pricing information. See, e.g., NASA, 180 F.3d at 306; Widnall, 57 F.3d at 1164; Gulf & W. In*1193dus., 615 F.2d at 530. To be sure, there may be, as the Air Force suggests, too many “unascertainable,” Acumenics Research & Technology v. Department of Justice, 843 F.2d 800, 808 (4th Cir.1988), or “fluctuating,” Martin Marietta Corp. v. Dalton, 974 F.Supp. 37, 40 (D.D.C.1997), variables for a firm to model exactly or to pinpoint precisely a rival’s pricing strategy, but pinpoint precision is not required to inflict substantial competitive harm.
Before concluding, we respond to two claims made by our dissenting colleague. First, our analysis in Parts II.C. and II.D does not come “close to a per se rule” that contract line-item prices “may never be revealed to the public through the [FOIA].” Dissent at 1194, 1202. McDonnell Douglas has not urged such a rule, and we have not considered one. Our decision that disclosure of such pricing information is not required in this case turns upon the particular facts that make disclosure here “likely ... to cause substantial harm to the competitive position of the person from whom the information was obtained.” National Parks I, 498 F.2d at 770.
In any event, Judge Garland’s suggestion that disclosure of the vendor prices implicates “the core purpose of FOIA,” Dissent at 1194, is doubtful. The decision of the Air Force “to expend a specified amount of public funds,” id. at 1204, has already been disclosed to the public; indeed, the total contract price paid by the Government “is routinely made public,” James T. Reilly, 1 FedeRal Infoemation DISCLOSURE § 14.84 (3d ed.2004), because that disclosure informs citizens about “what their government is up to.” Dep’t of Justice v. Reporters Comm. For Freedom of Press, 489 U.S. 749, 773, 109 S.Ct. 1468, 1481, 103 L.Ed.2d 774 (1989). The pricing information Lockheed seeks, however, has little to do with the core purpose of the FOIA, namely, “ ‘contributing significantly to public understanding of the operations or activities of the government.’ ” Dep’t of Defense v. FLRA, 510 U.S. 487, 495, 114 S.Ct. 1006, 1012, 127 L.Ed.2d 325 (1994) (quoting Reporters Comm., 489 U.S. at 775, 109 S.Ct. at 1482). On the contrary, the information now in suit reveals the internal workings of the contractor, not those of the Government, and would seem to shed little if any light upon the “agency’s performance of its statutory duties.” Bibles v. Or. Natural Desert Ass’n, 519 U.S. 355, 356, 117 S.Ct. 795, 795, 136 L.Ed.2d 825 (1997) (per curiam). Cf. Reporters Comm., 489 U.S. at 773, 109 S.Ct. at 1481 (purpose of FOIA “is not fostered by disclosure of information about private citizens ... that reveals little or nothing about an agency’s own conduct”). For all the Government tells us, the contractor’s margin on particular line items should be a matter of indifference to the purpose of the FOIA. The decision made by the Air Force was whether to accept McDonnell Douglas’s or Lockheed’s overall bid; whether the lower bidder marked up one cost element by a large margin and another by a small margin, in the course of making its bid competitive overall, is not self-evidently relevant to the question what the “government is up to.”
That the vendor prices at issue here do not seem to concern the core purpose of the FOIA does not mean, however, that line-item pricing information is per se protected from disclosure. As noted above, McDonnell Douglas did not argue for a per se rule, and we do not endorse such a rule in deciding certain prices are protected from disclosure in this case.
III. Conclusion
For the foregoing reasons, we reverse the judgment of the district court insofar as it upholds the decision of the Air Force *1194to release the option year prices and the Vendor Pricing CLINs in McDonnell Douglas’s KC-10 and KDC-10 contract. We affirm the judgment of the district court insofar as it authorizes the Air Force to release the Over and Above Work CLINs.
So ordered.
. The Trade Secrets Act provides a criminal penalty for anyone who
publishes, divulges, discloses, or makes known in any manner or extent not authorized by law any information coming to him in the course of his employment or official duties .. which information concerns or relates to the trade secrets, processes, operations, style of work, or apparatus, or to the identity, confidential statistical data, amount, or source of any income, profits, losses, or expenditures of any person, firm, partnership, corporation or association.
18 U.S.C. § 1905. Although the proprietor of commercial information does not have a private right of action to enforce § 1905, it may seek review of an agency action that violates the Trade Secrets Act on the ground it is "contrary to law," per § 10 of the Administrative Procedure Act, 5 U.S.C. § 702. Chrysler Corp. v. Brown, 441 U.S. 281, 317, 99 S.Ct. 1705, 1725, 60 L.Ed.2d 208 (1979); Widnall, 57 F.3d at 1164.
. Although we will remand a matter to an agency where the agency's initial explanation of its decision was inadequate, e.g., Northeast Md. Waste & Disposal Auth. v. EPA, 358 F.3d 936, 949-50 (D.C.Cir.2004), we do not typically remand to permit the agency an opportunity to adopt an entirely new explanation first suggested on appeal. The Air Force had numerous opportunities to argue that "rebidding is not likely," Dissent at 1199, and to offer the detailed analysis Judge Garland presents in his dissent; it never did so. We see no reason to give the Air Force yet another chance.
. The dissent faults the court for not resolving McDonnell Douglas’s additional claim that disclosure of the option year prices in the contract was also likely to cause it competitive harm because its competitors could reverse-engineer certain sensitive pricing factors from the option prices. See Dissent at 1200-01. Because we conclude that disclosure of the option year prices would likely cause McDonnell Douglas competitive harm by enabling competitors to undercut its prices, the court has no occasion to continue on in dicta to decide whether McDonnell Douglas would also suffer competitive harm from reverse-engineering of its sensitive pricing factors. •
. The Air Force’s conclusory statement that "it is not uncommon” for subcontractors to quote different prices to different primes is not a "prediction” or a "forecast” regarding “the repercussions of disclosure,” to which we ordinarily defer; it is a "declaration of empirical fact.” CNA Fin. Corp., 830 F.2d at 1155. Although we may defer to the "predictive judgment” of an agency absent record evidence, id., we will not defer to a declaration of fact that is "capable of exact proof” but is unsupported by any evidence. Cf. id.
. The presumption is reasonable because McDonnell Douglas can not be faulted for failing to produce evidence of the prices its subcontractors charged to other primes, suqh as Lockheed. See Maj. Francis Dymond, DoD Contractor Collaborations, 172 Military Law Review 96, 139 (noting "the secretive nature of collaborations and their complexity”). Because only the Air Force received information from both primes, the burden of producing such evidence should lie, if anywhere, with it. Cf. Occidental Petroleum v. SEC, 873 F.2d 325, 342 (D.C.Cir.1989) ("It is far more efficient, and obviously fairer, to place the burden of production on the party who claims that the information is publicly available”).
. Relying upon analyses contending that the market for weapons systems is not competitive, Dissent at 1197 n.5, 1198 n.6, Judge Garland says it is "plausible” for the Air Force to speculate that the market for servicing aircraft is not competitive either, because "[o]ngoing relationships between the vendor and different primes ... could affect the vendor's decision of the price to charge.” Setting aside whether the teaming arrangements and joint ventures characteristic of weapons manufacturing are pro- or anticompetitive, we have no basis for assuming the market for servicing the DC/KC/KDC-10 family of aircraft — which aircraft are used in both civil and military aviation and, hence, have multiple civilian and military customers-and suppliers — is characterized by the same alleged failures as the oligopolistic market for weapons systems. In any event, if the Air Force had offered some explanation' or evidence of the sub/prime contractor market at issue here along- the lines of that now supplied by Judge Garland, then perhaps its argument would be persuasive. To be sure, the Air Force need not provide "the kinds of evidence more usually associated with elaborate antitrust proceedings,” National Parks & Conservation Assn v. Kleppe, 547 F.2d 673, 681 (D.C.Cir.1976), but its assertion regarding-subcontractor pricing is not a prediction; it is an assertion capable of, but entirely lacking in, substantiation. Cf. CNA Fin., 830 F.2d at 1155.