Karen LeCraft Henderson, Circuit Judge:

At issue in this Privacy Act case is the decision of the National Science Foundation (NSF) not to identify who among twelve peer reviewers prepared written evaluations of an ultimately unsuccessful research grant proposal. NSF withheld their names pursuant to Privacy Act exemption (k)(5) which protects the identity of a source who, under an express promise of confidentiality, provides an agency with information regarding, inter alia, the suitability or qualifications of an applicant for a federal contract. The district court granted NSF summary judgment and we affirm.

I.

NSF is an independent agency of the federal government charged with promoting scientific and engineering progress in the United States. 42 U.S.C. Section(s) 1861 et seq. NSF does not itself conduct research. Instead the agency operates a number of programs designed to initiate, support and promote basic research in the physical, natural and social sciences. NSF has traditionally used grant agreements to support scientific research projects. The agency receives grant proposals primarily from educational institutions but also from commercial firms (typically small businesses) and, less frequently, from individuals. Joint Appendix (JA) 17. Each proposal must name a principal investigator or "PI," the scientist responsible for conducting the research and publishing the results. JA 176. The PI is the key figure behind an NSF grant. NSF is heavily dependent on an advisory "peer review" process to evaluate grant proposals. JA 152. That is, an NSF grant proposal is reviewed by outside experts knowledgeable in the particular scientific or engineering field represented by the proposal. Reviewers are asked to evaluate a proposal according to four general criteria, one of which assesses the applicant's research performance competence, including the capability of the investigator. JA 86. NSF either solicits peer reviews by mail or assembles a panel of peer reviewers to review grant proposals. JA 88.

Panel review typically works like this. A peer review panel convenes to consider a large number of proposals. In addition to full panel consideration, each proposal receives several written evaluations drafted by individual panel members. The full panel rates each proposal either "highly recommended," "recommended" or "not recommended." JA 145. An NSF Program Officer then considers the written reviews, the panel recommendation and other factors and makes a funding recommendation. Id. NSF supervisory personnel then review the Program Officer's recommendation and, after weighing various considerations, decide whether to fund the proposal. JA 143.

When NSF rejects a grant proposal the agency notifies the PI and furnishes a summary of the review panel's discussion (if any) as well as verbatim copies of written reviews but redacts the authors' names. JA 312. NSF's policy is to protect the confidentiality of all of the peer reviewers who draft written comments. See infra Part III. As the General Accounting Office (GAO) has observed, "Confidentiality . . . helps ensure that peer reviewers will give candid comments on grant applications by protecting reviewers from possible reprisal by applicants." GAO Report, Peer Review: Compliance With the Privacy Act and Federal Advisory Committee Act (1991) ("GAO Report") at 4 (JA 226).

II.

Dynamic In Situ Geotechnical Testing, Incorporated (Dynamic) submitted a grant proposal to NSF in April 1993. Dynamic's application listed appellants Wanda and Robert Henke (Dynamic's co-owners) as the project's PI and co-PI, respectively, meaning that they were the scientists responsible for carrying out the research. A peer review panel consisting of twelve experts from universities and other government agencies convened in August 1993 to evaluate Dynamic's proposal and thirty-six others. Before the meeting four of the twelve panelists had each prepared written comments regarding Dynamic's proposal. The panel recommended against funding the proposal and NSF ultimately denied funding. Thereafter the agency, in accordance with its standard practice, notified Wanda Henke and provided her with a summary of the panel discussion and verbatim copies of the four written reviews with authors' names redacted.

Wanda Henke then filed with NSF a Privacy Act request seeking the names of the four peer reviewers who prepared the written comments as well as the names of the other eight panel members who evaluated Dynamic's proposal. In response NSF disclosed the names of the twelve panel members but refused to indicate which four authored the written comments. NSF relied on 45 C.F.R. Section(s) 613.6(a) which implements Privacy Act exemption (k)(5). Exemption (k)(5) provides that an agency may promulgate rules to exempt from the Privacy Act's access provisions a system of records that is investigatory material compiled solely for the purpose of determining suitability, eligibility, or qualifications for Federal civilian employment, military service, Federal contracts, or access to classified information, but only to the extent that the disclosure of such material would reveal the identity of a source who furnished information to the Government under an express promise that the identity of the source would be held in confidence.

5 U.S.C. Section(s) 552a(k)(5) (emphasis added). NSF has construed "Federal contracts" in exemption (k)(5) to encompass NSF grant agreements. 45 C.F.R. Section(s) 613.6(a).

In February 1994 the Henkes filed this Privacy Act suit seeking the names of the four peer reviewers. See 5 U.S.C. Section(s) 552a(g)(1)(B), (g)(3)(A). On cross-motions for summary judgment the plaintiffs argued that exemption (k)(5) was inapplicable on the grounds that (1) the reviewers did not act under an express promise of confidentiality and (2) the reviewers provided information in connection with an application for a federal grant, not a federal contract. The district court granted NSF summary judgment on both issues and the plaintiffs appeal. Our review is de novo. Sellers v. Bureau of Prisons, 959 F.2d 307, 310 (D.C. Cir. 1992). We address each issue in turn.

III.

Exemption (k)(5) protects the identity of a source only if he "furnished information to the Government under an express promise that the identity of the source would be held in confidence." The four peer reviewers who drafted comments in connection with Dynamic's proposal used an NSF evaluation form which informed them that "[i]t is the policy of the Foundation that reviews will not be disclosed to persons outside the Government, except that verbatim copies without the name and affiliation of the reviewer will be sent to the principal investigator." JA 98.

The form further states that "the identity of reviewers will be kept confidential to the maximum extent possible." Id. This of course constitutes an express promise of confidentiality. NSF has presumed that a peer reviewer asked to draft a review of an NSF grant proposal and assess the PI's capabilities not only desires an assurance of confidentiality but would not provide a candid review (i.e., a critical, thorough review) without it. Cf. Department of Justice v. Landano, 113 S.Ct. 2014, 2023-24 (1993) (although presumption is not all encompassing, government entitled to presume FBI source is confidential where generic circumstances, such as use of paid informants, support inference of confidentiality in FOIA exemption 7(D) context). The record contains scientists' affidavits and refers to studies showing that absent reviewer anonymity the quality of reviews likely would suffer and, indeed, many (if not most) peer reviewers would decline to assist NSF or would do so less often. Reviewers fear harassment and lawsuits from disgruntled PIs, Aff. of Dr. Aubrey M. Bush, NSF Program Manager at 2 (JA 153), and "quite naturally do not want to offend a colleague whom they may like as a person, or hope to work with in the future," Aff. of Dr. Constance A. Sancetta, NSF Associate Program Manager at 2 (JA 155), especially "senior researchers in positions of authority," Aff. of Daniel J. Madden, NSF Program Director at 2 (JA 166). Furthermore, because peer reviewers and PIs often switch hats — a peer reviewer may be the PI on a future grant application and vice versa — an identified peer reviewer might fear reprisal at the hands of a disgruntled PI. Sancetta Aff. at 2 (JA 155). It bears emphasizing that many specialized scientific communities are relatively small. Aff. of Dr. Stephen J. Mackwell, NSF Program Director at 1 (JA 163). In sum we hold that there is sufficient evidence on this record to show that each of the four peer reviewers asked to draft written comments on Dynamic's proposal furnished information to NSF under an express promise that NSF would hold in confidence his or her identity.

The plaintiffs contend that the requirement in subsection (k)(5) for an express promise of confidentiality is not satisfied unless the agency can demonstrate that the reviewers requested or desired non-disclosure. The district court disagreed, stating that "whether the reviewers expressed a desire to keep their identities confidential is wholly irrelevant to the Court's determination." JA 392. We would not go quite that far. Agencies must use subsection (k)(5) sparingly. Indeed, it is incumbent upon an agency to determine that promises of confidentiality are necessary. That determination, however, may be made categorically. Nothing in either the statute or the case law requires that NSF apply subsection (k)(5) only to those particular reviewers who have expressly asked for an exemption and would otherwise have declined to participate in the peer review process.

IV.

Privacy Act exemption (k)(5) protects the identity of confidential sources who provide agencies with information regarding the suitability or qualifications of applicants for "Federal contracts." The plaintiffs contend that peer reviewers who assist NSF are not covered by the exemption because they provide information in connection with federal grants, not federal contracts. The Privacy Act does not define the term "Federal contracts" and exemption (k)(5)'s sparse legislative history is silent on whether Congress intended the term to exclude grant agreements. But the plaintiffs urge that a grant is not a contract, that Congress understood the distinction and that by not including the term "grants" in exemption (k)(5) Congress categorically declined to protect confidential sources providing information in connection with federal grants. We disagree.

"A contract is a promise or set of promises for the breach of which the law gives a remedy, or the performance of which the law in some way recognizes as a duty." Restatement (Second) of Contracts (1979) Section(s) 1; see 1 Williston, Contracts Section(s) 1, 18 (3d ed. 1957). A contract has certain essential elements, to wit, competent parties, lawful subject matter, legal consideration, mutuality of assent and mutuality of obligation. 1 Williston, Contracts Section(s) 1:1 (4th ed. 1990) (discussing "classic concept of contract"); see Black's Law Dictionary 322 (6th ed. 1990); see also Farrington v. Tennessee, 95 U.S. 679, 685 (1878) (discussing Trustees of Dartmouth College v. Woodward, 17 U.S. (4 Wheat.) 518 (1819)). An NSF grant agreement includes the essential elements of a contract and establishes what would commonly be regarded as a contractual relationship between the government and the grantee. NSF "agrees to support a specified level of effort for a specified period of time" and to cover "the costs of the project to be performed under the provisions of the award instrument" while "[t]he grantee agrees to the performance of the project, to the prudent management of the funds provided by the grant, and to the provisions of the award instrument." NSF Grant Policy Manual (July 1989) at I-3 (JA 308). The award instrument in turn imposes numerous conditions to which the grantee must agree to receive the grant funds. Acceptance of the grant occurs when the grantee takes action to obtain the grant funds from the government, unless NSF requires written acceptance. Id. at I-4 (JA 309). NSF can suspend or terminate a grant when "the grantee has materially failed to comply with the terms and conditions of the grant." JA 277-78. Finally, the government can, at least in some circumstances, sue to enforce the conditions of an NSF grant agreement. See United States v. Marion County Sch. Dist., 625 F.2d 607, 609 (5th Cir. 1980) ("As the Supreme Court has long recognized, the United States may attach conditions to a grant of federal assistance, the recipient of the grant is obligated to perform the conditions, and the United States has an inherent right to sue for enforcement of the recipient's obligation in court."), cert. denied, 451 U.S. 910 (1981); accord Brown v. Califano, 627 F.2d 1221, 1232 n. 67 (D.C. Cir. 1980). There is no indication that Congress intended the term "contracts" to have a different or specialized meaning within the context of exemption (k)(5).

The plaintiffs point out that in the Grants Act of 1958 (Grants Act), 42 U.S.C. Section(s) 1891-93 (repealed), and its successor, the Federal Grant and Cooperative Agreement Act of 1977 (FGCAA), 31 U.S.C. Section(s) 6301 et seq., Congress drew a distinction between grants and procurement contracts. But neither statute supports the categorical proposition from which the plaintiffs seek to benefit — that a federal grant agreement can never be deemed a federal contract, that the two are somehow inherently mutually exclusive. And unlike the Grants Act and FGCAA, which were enacted to address whether and when agencies should use grant agreements instead of procurement contracts (and thus distinguished between grants and contracts), exemption (k)(5) is not a statute designed to address funding by federal agencies and does not distinguish, either expressly or by implication, between grant agreements and contracts. It just says "Federal contracts."

More to the point, the plaintiffs' reliance on the Grants Act and the FGCAA proves too much. A quick review of why Congress enacted those two statutes belies the plaintiffs' theory that Congress intended a formalistic interpretation of "contracts" in exemption (k)(5). Before Congress passed the Grants Act all federal agencies authorized to fund basic scientific research were authorized to use procurement contracts but only three agencies were authorized to use grant agreements as well. Most of the agencies authorized to fund basic scientific research, then, were forced to comply with burdensome procurement regulations. In an effort to give agencies more flexibility, Congress enacted the Grants Act. See generally H.R. Rep. No. 2640, 85th Cong., 2d Sess. (1958). The statute gave all affected agencies the discretion to use either a grant agreement or a procurement contract when funding basic scientific research. 42 U.S.C. Section(s) 1891-93, repealed by FGCAA.

A decade later Congress created the Commission on Government Procurement (Commission) to study procurement by federal agencies and report to Congress. Pub.L. No. 91-129, 83 Stat. 269 (1969). In studying the relationship between federal grants and procurement, the Commission found a good deal of chaos. For one thing, agencies (and sometimes the same agency) were using grant agreements and procurement contracts interchangeably for the same types of projects. See Report of the Commission on Government Procurement (Dec. 1972) at 157 (JA 292). The Commission discovered, moreover, that Congress's enabling legislation was inconsistent regarding the circumstances under which agencies were to use grant agreements rather than procurement contracts. Id. at 159 (JA 294). The Commission recommended legislation to "distinguish assistance relationships as a class from procurement relationships by restricting the term `contract' to procurement relationships and the terms `grant,' `grant-in-aid,' and `cooperative agreement' to assistance relationships." Id. at 162-63 (JA 297-98); see also id. at 164 (JA 299) ("The term `contract' should be restricted to procurement relationships.") (emphasis added).

Congress responded by enacting the FGCAA in 1978, several years after Congress passed exemption (k)(5). (Although the FGCAA did not pass both houses until 1978, Congress considered similar legislation from 1973 to 1977.) The FGCAA drew in large measure from the Commission's report and attempted to tighten the distinction between federal assistance relationships and federal procurement relationships. In the statement of findings Congress recognized that "uncertainty as to the meaning of such terms as `contract', `grant', and `cooperative agreement' and the relationships they reflect causes operational inconsistencies, confusion, inefficiency, and waste for recipients of awards as well as for executive agencies." Pub.L. No. 95-224 Section(s) 2(a)(2), 92 Stat. 3 (1978); see S. Rep. No. 449, 95th Cong., 2d Sess. 6-7 (1977) (noting "inappropriate use of grants to avoid the requirements of the procurement system" and "often inconsistent" ways agencies used grants "even for similar programs and projects"). In the statement of purpose Congress explained that it hoped to establish uniform criteria and promote increased discipline regarding the selection and use of contracts, grants and cooperative agreements. Pub.L. No. 95-224 2(b)(2), (3), 92 Stat. 3 (1978); see S. Rep. No. 449, 95th Cong., 2d Sess. 6 (1977) ("No uniform statutory guideline exists to express the sense of Congress on when executive agencies should use either grants, cooperative agreements or procurement contracts.").

This is the backdrop for Congress's 1974 enactment of exemption (k)(5). In light of the uncertainty over the meanings of "contract" and "grant," the lack of uniform statutory or regulatory standards dictating when either instrument should be used and the recognition that agencies were using both instruments to fund similar projects, we doubt Congress intended exemption (k)(5) to employ a rigid definition of "contracts," one contemplating a neat division between procurement contracts and grant agreements. Put another way, it is doubtful that Congress intended the protection of a confidential source to hinge on what Congress knew would, in practice, be an arbitrary variable — the agency's choice of grant agreement versus contract. See Report of the Commission on Government Procurement at 157 (JA 292) (observing that for basic scientific research some agencies used grants while others used contracts). Rather, it is more likely that Congress intended the protection of confidential sources to turn on the government's need for candid information and the reduced likelihood of obtaining such candid information absent an assurance of confidentiality.

Moreover, why would Congress protect the identity of a source providing information on an applicant for a procurement contract but not a source providing information on an applicant for a grant agreement? We find nothing in the record to substantiate the plaintiffs' suggestion that Congress was more concerned with the suitability and qualifications of procurement contract applicants than with those of grant applicants.

Indeed, a good argument can be made in support of the reverse proposition: inasmuch as an agency has less day-to-day control over a grantee than it does over a contractor who performs under a procurement contract, the agency has a stronger interest in determining up front that the grant recipient is qualified and competent. The plaintiffs suggest instead that Congress may have seen a more pressing need for source protection in the procurement context where a source frequently is asked to reveal personal details about an applicant (the suggestion being that such a source, fearing reprisal, likely will not talk, or talk candidly, absent an assurance of confidentiality). Brief of Appellants at 40. But the same could be said of the peer reviewer asked to evaluate the merits of a proposal and the capabilities of its PI. As Wanda Henke explained in an affidavit submitted below, NSF peer reviewers "focus on the research proposed by, and research capabilities of, the specifically named PIs. In fact, the reviews are frequently written in a manner that is quite personal to the particular PIs who have submitted the proposal." JA 169-70.

Finally, the plaintiffs rely on the interpretation of the Office of Management and Budget (OMB), the agency charged with implementing the Privacy Act. See 5 U.S.C. Section(s) 552a(v). In 1975, shortly after Congress passed the Privacy Act, OMB issued interpretive guidelines for the entire Act. Regarding exemption (k)(5), OMB stated, "The term `Federal contracts' covers investigatory material on individuals being considered for employment on an existing Federal contract as well as investigatory material compiled to evaluate the capabilities of firms being considered in a competitive procurement." OMB Privacy Act Guidelines, 40 Fed.Reg. 28,949, 28,974 (1975). The plaintiffs emphasize that OMB did not suggest that the term "contracts" encompasses grants and urge that we defer to OMB's contemporaneous interpretation of the exemption. It is far from clear, however, that OMB intended the statement quoted above to be exhaustive or that OMB squarely addressed the issue presented here: whether Congress intended the general term "Federal contracts" to include a federal grant agreement that contains the essential elements of a contract. Indeed, in a 1991 letter to GAO, prepared in response to the GAO Report, supra Part I, OMB suggested that an agency may protect the identities of peer reviewers via exemption (k)(5) so long as the agency first publishes a regulation stating its intention to do so as required by 5 U.S.C. Section(s) 552a(k). GAO Report at 16, 20 (JA 238, 242) (January 17, 1991 Letter from James B. MacRae, Jr., Acting Administrator of OMB's Office of Information and Regulatory Affairs, to James F. Bouck, GAO Evaluator).

We have considered the plaintiffs' other arguments and find them without merit. We therefore hold that the term "Federal contracts" in Privacy Act exemption (k)(5) encompasses a federal grant agreement if the grant agreement includes the essential elements of a contract and establishes a contractual relationship between the government and the grantee. We conclude that the NSF grant agreement is within the scope of exemption (k)(5) and that NSF properly invoked the exemption and therefore is entitled to summary judgment.

* * *

For the foregoing reasons the judgment of the district court is

Affirmed.