Full opinion text
Opinion for the Court filed by Circuit Judge SPOTTSWOOD W. ROBINSON, III. SPOTTSWOOD W. ROBINSON, III, Circuit Judge: This reverse-Freedom of Information Act (FOIA) case features two important issues: the exact scope of 18 U.S.C. § 1905, commonly referred to as the Trade Secrets Act, and its relationship to FOIA Exemptions 3 and 4. In past cases, we have repeatedly touched upon these difficult questions, but never squarely decided them. We resolve them definitively today. I. Background We begin by reducing the background of this appeal to its essence. Appellant CNA is an insurance company doing business with the Federal Government. As a condition of receiving federal contracts, and by virtue of the mandate of Executive Order 11,246, it must submit to the appropriate governmental agency various materials demonstrating its performance in hiring, promoting, and otherwise utilizing women and minorities, as well as its affirmative action goals for the future. These materials take the form of EEO-1 reports and affirmative action programs. In 1977, when the FOIA requests instigating the present dispute were made, the agency charged with enforcing the insurance industry’s compliance with Executive Order 11,246 was the Department of Health, Education and Welfare (HEW), which in turn had delegated this responsibility to the Insurance Compliance Staff (ICS) of the Social Security Administration. In April of 1977, a group called Women Employed requested from ICS copies of the 1976-77 affirmative action programs and EEO-1 reports for CNA’s midwest regional office, the 1974-75 affirmative action programs and EEO-1 reports for CNA’s home office, and several documents prepared by ICS concerning CNA’s compliance with the executive order. On being notified by ICS both of the FOIA request and of the agency’s intention to honor it, CNA sought and obtained from the District Court an order restraining release pending completion of the administrative appeal process. Then began the long and tortuous journey of this case, in the course of which it traveled back and forth between the District Court and the agency, with one fairly brief stop in this court. The specifics of this five-year odyssey are reserved for a later section of this opinion, wherein CNA’s procedural challenges are discussed. It is sufficient for the present to note that in determinations rendered in November, 1980, and July, 1981, the Office of Federal Contract Compliance Programs (OFCCP) of the Department of Labor — which by then had assumed the enforcement responsibilities previously exercised by ICS — decided to release, pursuant to FOIA and regulations of OFCCP, all the material requested. In thus discarding CNA’s objections to disclosure, OFCCP decided that the Trade Secrets Act is not a withholding statute within the meaning of Exemption 3. It then proceeded to consider whether the requested material fell within Exemption 4. Applying the test established by this court in National Parks & Conservation Association v. Morton, OFCCP discounted CNA’s claim that revelation of its affirmative action programs and EEO-1 reports would precipitate substantial competitive harm. CNA had asserted, through various affidavits and depositions, that any such disclosure would, inter alia, facilitate raiding of its employees, with a concomitant increase in recruiting and training costs; deleteriously affect employee morale; generate adverse publicity by virtue of public “misconstruction” of the data; and reveal to competitors CNA’s plans to enter, expand, or contract different markets and product lines. OFCCP dismissed these concerns as overstated and speculative; it concluded that the information in the affirmative action programs and EEO-1 reports was neither so detailed nor so specific to CNA that it would have any particular competitive value to a rival. The agency’s estimates in this regard were buttressed by its view that the data at issue, which were by then four to seven years old, were stale. Having found no threat of competitive harm that would bring the requested material within Exemption 4, OFCCP concluded that FOIA mandated its disclosure. The District Court sustained OFCCP’s final determination. It agreed that the Trade Secrets Act is not an Exemption 3 withholding statute, and that substantial competitive harm was the appropriate standard for resolving the Exemption 4 question. After denying CNA’s request for a de novo evidentiary hearing, the court held that OFCCP’s determination was not “arbitrary, capricious, an abuse of discretion or otherwise not in accordance with law.” Accordingly, it granted summary judgment for the agency and directed that the documents be released within fifteen days. We granted a stay of this direction pending appeal. CNA now complains vigorously that it has not been afforded at any point — either before the agency or in the District Court — an evidentiary hearing on its claims of competitive harm. CNA’s procedural challenges were largely resolved by our holding in National Organization for Women v. Social Security Administration (NOW), and are discussed in Part IV of this opinion. Parts II and III address CNA’s substantive claims respecting the scope and interrelationship of the Trade Secrets Act and Exemptions 3 and 4 of FOIA, and review CNA’s contention that OFCCP improperly applied Exemption 4. II. The Trade Secrets Act and Exemption 3 The third exemption of FOIA, as amended in 1976, excepts from mandatory disclosure material specifically exempted from disclosure by statute (other than [the open meeting provisions of 5 U.S.C. § 552b]), provided that such statute (A) requires that the matters be withheld from the public in such a manner as to leave no discretion on the issue, or (B) establishes particular criteria for withholding or refers to particular types of matters to be withheld. We have had occasion in the past to recount the events leading up to the 1976 amendment, and to explicate the legislative goals that must inform our resolution of Exemption 3 questions. Without reiterating that now-familiar history, we note simply that Subsections (A) and (B) were designed to narrow the scope of the prior-existing exemption by excluding those broadranging statutes that give an agency “cart blanche [sic ] to withhold any information [it] pleases.” The “unmistakable thrust” of Exemption 3 as amended is “to assure that basic policy decisions on govemmental secrecy be made by the Legislative rather than the Executive branch.” CNA contends that the Trade Secrets Act qualifies under both Subsections (A) and (B) as an Exemption 3 withholding statute. That Act reads as follows: Whoever, being an officer or employee of the United States or of any department or agency thereof, publishes, divulges, discloses, or makes known in any manner or to any extent not authorized by law any information coming to him in the course of his employment or official duties or by reason of any examination or investigation made by, or return, report or record made to or filed with, such department or agency or officer or employee thereof, 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; or permits any income return or copy thereof or any book containing any abstract or particulars thereof to be seen or examined by any person except as provided by law; shall be fined not more than $1,000, or imprisoned not more than one year, or both; and shall be removed from office or employment. After careful consideration, we believe the Trade Secrets Act does not satisfy the strictures of either Subsection (A) or (B). We have heretofore explained that Subsection (A) “is too rigorous to tolerate any decisionmaking on the admininstrative level. It embraces only those statutes incorporating a congressional mandate of confidentiality that, however general, is ‘absolute and without exception.’ ” CNA’s insistence that the Trade Secrets Act is such a ban on disclosure of trade secrets and other business data ignores the crucial phrase confining the prohibition to only those revelations that are “not authorized by law.” If the sole possible source of disclosure-authorization were congressional enactment, we might agree that the Act contemplates no decisionmaking on the administrative level, and thereby satisfies Subsection (A) of Exemption 3. The Supreme Court has made clear, however, that duly promulgated agency regulations reasonably related to the purpose and scope of a statutory grant of substantive rulemaking authority can provide the requisite legal sanction for disclosure. We think this potential for administrative authorization precludes the application of Subsection (A). Whenever an agency has power to adopt substantive regulations governing public access to its files, the decision whether the Trade Secrets Act is to bar divulgence of commercial and financial material contained therein ordinarily will be made by the agency, not by Congress. More specifically, disclosure will be “not authorized by law” just to the extent that the agency elects not to endorse release. Of course, the fact that an agency possesses discretion to control the applicability of the Trade Secrets Act is fatal only to a Subsection (A) claim. Even statutes conferring considerable amounts of administrative discretion can fall within Subsection (B) of Exemption 3 if they either establish “particular criteria for withholding” or refer to “particular types of matters to be withheld.” To meet the first prong of Subsection (B), the statute must “incorporate[ ] a formula whereby the administrator may determine precisely whether disclosure in any instance would pose the hazards that Congress foresaw.” The Trade Secrets Act is patently deficient in this regard. As we have already explained, the Act, by prohibiting only unauthorized disclosures, leaves at least some agencies in a position to opt out of its strictures. Yet nothing in the Act directs or guides an agency in deciding whether it ought to exercise its power to authorize revelation of officially collected commercial and financial data. Moreover, going a step beyond the agency’s initial decision to permit some disclosure, the Trade Secrets Act in no way channels the discretion of agency decision-makers in supplying authorization for specific cases. Suppose, for example, an agency with substantive power to promulgate public access regulations duly adopts a rule stating: Disclosure of trade secrets and other business information will be deemed ‘authorized’ for purposes of the Trade Secrets Act only when the administrator determines that release is consistent with the public interest. This hypothetical regulation embodies the kind of administrative carte blanche that Congress intended to curb by its amendment of Exemption 3, yet nothing in the Trade Secrets Act suggests that such a wideranging regulation would be ineffective as a source of disclosure authority. Devoid of any normative content to steer the agency in determining when revelation of data would be deleterious, the Act is susceptible of invocation to bar disclosure at the whim of an administrator acting under such a regulation. Indeed, one of the fundamental difficulties with classifying the Trade Secrets Act as an Exemption 3 statute is that agencies conceivably could control the frequency and scope of its application through regulations adopted on the strength of statutory withholding authorizations which do not themselves survive the rigors of Exemption 3. The Trade Secrets Act is no more successful in meeting the second prong of Subsection (B). The language of the Act is exceedingly broad. Extending by its terms to information pertaining to “trade secrets,” “processes,” “style of work,” “apparatus,” “confidential statistical data,” the amount or source of “income, profits, losses, or expenditures” and even the “identity” of “any person, firm, partnership, corporation, or association,” “encompasses virtually every category of business information likely to be in the files” of an agency. The “oceanic” quality of the Act is enhanced by the fact that it is addressed to every “officer or employee” of “any [federal] department or agency;” moreover, it applies to any information discovered “in the course of [the employee’s] employment or official duties or by reason of any examination or investigation made by, or return, report, or record made to or filed with” the department, agency, or employee. In sum, the Trade Secrets Act appears to cover practically any commercial or financial data collected by any federal employee from any source in performance of the duties of his or her employment. In determining in a prior case that Section 7(c) of the Export Administration Act does not satisfy the second prong of Subsection (B), we explained that a nondisclosure statute’s general applicability to anything that might happen to be encompassed within [an] array of information-gathering functions undermines any notion that [the statute] represents a congressional determination of the advisability of secrecy for any “particular type[ ] of matter[ ],” if for no other reason than that the agency has the power radically to expand the quantity and diversity of information in its files to intercept matter of a sort that Congress well might not have contemplated when considering the need for confidentiality. We think that reasoning is equally applicable here. Admittedly, the Trade Secrets Act lists several specific categories of data, but cumulation of the parts may at some point become so extensive that what have been described as parts are in fact the whole; and that is true here. The comprehensive catalogue of items in the Trade Secrets Act accomplishes essentially the same thing as if it had simply referred to “all officially collected commercial information” or “all business and financial data received.” Given this encyclopedic character, and absent any limitation as to the agencies covered or the sources of data involved, we unhesitatingly conclude that the laundry list of information that the Act sets out does not specify “particular types of matters” for purposes of satisfying Subsection (B) of Exemption 3. For these reasons, we hold that the Trade Secrets Act does not, by virtue of Exemption 3, erect a disclosure bar that is impervious to the mandate of FOIA. Our view that this legislation is one of those prior-existing nondisclosure edicts that have been at least partially superseded by FOIA’s newer public access mandate is reinforced by several general observations about the two statutes. This court has had occasion to remark that the Trade Secrets Act is “merely a general prohibition against unauthorized disclosure.” This was perhaps a rather abbreviated way of noting that the Act seems to embody a congressional judgment that private commercial and financial information should not be revealed by agencies that gather it, absent a conscious choice in favor of disclosure by someone with power to impart the force of law to that decision. The Act attempts to forestall casual or thoughtless divulgence — disclosure made without first going through a deliberative process — with an opportunity for input from concerned parties. If we are correct in this assessment of the motivating force behind the Trade Secrets Act, then our holding, which effectively recognizes that FOIA can provide the requisite legal authorization to disclose agency-gathered commercial and financial data, does no violence to the legislative aim. FOIA is itself the product of a most thoughtful and deliberate weighing process, in which Congress considered the views of those who submit information, those who collect it, and those who desire it. Congress has twice since adjusted the balance in order more finely to tune the operation of FOIA to real-world concerns and constraints. To the extent that commercial and financial data within the ambit of the Trade Secrets Act do not fall within one of the FOIA exemptions, particularly Exemption 4 or 6, the decision has been made that on balance it ought to be unmasked in the interest of informing the public of the bases for governmental action. To the extent that such data as trade secrets and confidential financial information are excepted from mandatory disclosure by one or more of the exemptions that Congress has incorporated into FOIA, the Trade Secrets Act will bar a discretionary release unless, after notice and comment, an agency, possessing delegated power to do so, promulgates, a contrary rule having the force of law. Clearly, this resolution does not undermine the foundation of the Trade Secrets Act by throwing open the door to wholesale, haphazard revelation of private financial and business data in the possession of governmental agencies. ÍII. The Trade Secrets Act and Exemption 4 Having determined that the Trade Secrets Act is not a withholding statute of sufficient rigor or particularity to satisfy Exemption 3, we next address the interrelationship of the Act and Exemption 4. Specifically, we consider the position taken by some commentators, and urged here by agency counsel, that, despite its apparent sweep, the Trade Secrets Act is no broader than its three predecessor statutes were combined. This is of much more than academic interest to federal agencies that currently do not have public access regulations qualifying as legal authorizations for disclosures otherwise prohibited by the Trade Secrets Act. If the range of the Act is narrower than the scope of Exemption 4, there will be some commercial and financial data that these agencies will be free to release in their discretion, though they are not required to do so by FOIA. If, on the other hand, the reach of the Act is at least coextensive with that of Exemption 4, a finding that requested material falls within that exemption will be tantamount to a determination that these agencies cannot reveal it. A. The Trade Secrets Act The Trade Secrets Act came as part of the 1948 revision and codification of the Federal Criminal Code. In this process, a committee of legislators, judges, lawyers, and law book publishers gathered together criminal provisions scattered throughout the United States Code and organized, edited, and consolidated them; and the product was then enacted into positive law as Title 18. According to the Reviser’s Notes, the Trade Secrets Act was “[bjased on” three preexisting provisions. Section 216 of Title 18, which had made it a misdemeanor for any federal officer or employee to “divulge or make known in any manner whatever not provided by law” the “operations, style of work, or apparatus of any manufacturer or producer” visited by the employee in the course of his or her official duties, or “the amount or source of income, profits, losses, [or] expenditures” disclosed in income tax returns; Section 1335 of Title 19, also a misdemeanor statute, which had prohibited federal officers and employees from disclosing, “in any matter whatever not provided for by law ... trade secrets or processes ... embraced in any examination or investigation” of the Tariff Commission; and Section 176a of Title 15, which had directed that “[a]ny statistical information furnished in confidence” to the Bureau of Foreign and Domestic Commerce was to be held confidential and used “only for the statistical purposes for which it [was] supplied.” It is apparent from a comparison of the provisions of the Trade Secrets Act with those of the three earlier statutes from which the 1948 recodification drew its language, that the resulting whole was considerably greater than the sum of its parts. The remarkably scant legislative history of the 1948 Criminal Code is virtually silent on the Trade Secrets Act. Except for brief remarks on some purely editorial matters, the Reviser’s Note to the Act merely lists the antecedent statutes and declares that “[m]inor changes were made in translations and phraseology.” Neither the floor debates, the committee hearings nor the committee reports contain any reference to the new section embodying the Act. It is contended that, in these circumstances, we should apply the canon of construction that absent a clear showing of legislative intent to work a change in the prior law, a codified provision is to be given the same substantive content as its source statute, despite any changes in phraseology. Resort to this principle, the argument continues, will demonstrate that the Trade Secrets Act prohibits disclosure only of the fairly limited kinds of material that were encompassed by the three antecedent statutes. We think the question of statutory interpretation presented here is not nearly so simple. Indubitably, when Congress enacted the 1948 revision and codification of Title 18, its understanding was that “[t]he original intent of Congress is preserved.” But testimony by members of the drafting commission during two sets of House Committee hearings, as well as statements in the House Report, make clear that preservation of the “original intent” was not synonymous with absence of substantive change. The point was often made that the enactment represented a “revision” as well as a “codification,” and that the former involved reconciling conflicts, harmonizing incongruities and inconsistencies traceable to piecemeal enactment, and omitting some superseded sections while updating others. As Representative Keogh, sponsor of the legislation in the House and chairman of the Revision Committee, testified, “[t]he policy that [the Revision Committee] adopted ... was to avoid wherever possible and whenever possible the adoption in our revision of what might be described as controversial substantive change of law.” This emphasis on avoidance of “controversial” change was reiterated during the hearings. Thus, while we must not attribute to the 1948 enactment any substantial disruption in prior congressional purpose or policy without first discerning clear evidence that such a departure was intended, the mere fact that we are dealing with a revision-codification does not dictate slavish adherence to the predecessor statutes without regard to whether the new version is essentially compatible with what went before. We are mindful that several times in the last quarter-century the Supreme Court has had occasion to address variances between codified laws and their statutory antecedents, and has held that the meaning of the original statute survived changes in phraseology. We think, however, that close examination of those cases reveals a far more careful and thoughtful approach to interpretation than wooden citation of a canon of construction would suggest. In United States v. Cook, a defendant argued that consolidation of two statutes into a single anti-embezzlement provision effectively excluded a class of employees from the reach of the law. The Court began its analysis by pointing out that this class had been covered by one of the predecessor statutes; indeed, it observed, Congress had emphasized that coverage just two years prior to the codification. Noting that no “plausible reason” appeared for distinguishing between the class of employees at issue and all other groups of employees plainly subject to the statute, the Court concluded that to accept the defendant’s reading was to presume the occurrence in 1948 of a significant and seemingly illogical change in congressional thinking. Nothing in the legislative history hinted at such a deviation; moreover, the language used in the codified version was, in the Court’s view, capable of intercepting the disputed class. In these circumstances, the Court held that the antecedent statutes furnished the appropriate measure of the codified provision. City of Greenwood v. Peacock, a case decided contemporaneously with Cook, also saw the Court scrupulously assessing preexisting legislative intent and policy before determining that a change in phraseology did not work a difference in substance. At issue was a provision which, as codified, governed removal to federal courts of certain state criminal actions implicating civil rights. The Court demonstrated at some length that the argued-for construction would divorce the removal provision from the specific historical and statutory context in which it had developed, and give it a tenor not responsive to the purpose for which it was enacted. The proposed reading also would have expanded the scope of a statute which the Court was inclined to construe cautiously because of the obvious implications of federalism. It therefore looked carefully for some sign of congressional intent to disturb the balance previously reached in this delicate area, and finding none, held that the codified language did not accomplish the modification suggested. Two other cases reflect a similar pattern of analysis. Had the Court in Fourco Glass Company v. Transmirra Products Corporation accepted a proposed interpretation of the codified venue provision there at issue, the result would have been an effective overruling of one of its decisions, squarely on point, rendered just six years prior to codification. As the legislative history gave no hint of congressional intent to reverse that ruling, the Court quickly concluded that the newer enactment had not modified the prior law. Contributing to this disposition was the Court’s recognition that any other reading would have permitted a generic venue provision to supersede another provision specifically designed to set venue for the class of cases there involved. The Court’s reluctance to regard changed language in a codified statute of general applicability as evidence of a congressional policy shift in a specialized area of the law was also evident in Tidewater Oil Company v. United States. According to the petitioner’s reading of the interlocutory appeal provisions there involved, the codification of Title 28 had effected a major transformation of appellate jurisdiction over govemmentally-initiated civil antitrust cases. The Court reviewed the course of past legislative activity on that point and concluded that it was a subject which consistently had been treated as “a peculiarly distinct matter.” The petitioner’s position presupposed not only that Congress had intended to alter the law in this hitherto discrete area simply by modifying an across-the-board appeal provision, but also that in the process Congress had abandoned policies which for many years had guided legislative action on the subject. The Court also perceived the potential for a host of practical problems were the statute to be read as the petitioner suggested. Accordingly, instead of attributing to the codification so drastic a result, the Court searched the legislative history for some evidence of legislative intent, and found not only an absence of affirmative expression of congressional purpose to change existing law on the point, but also some material suggesting the opposite intention. In two other cases, the Court declined to find in the language of codified statutes a legislative determination to alter time-honored legal principles. The fact that a phrase had been deleted during codification of provisions governing awards of costs was held insufficient in Alyeska Pipeline Service Company v. Wilderness Society to demonstrate that Congress meant to disturb the “longstanding rule” restricting recovery by victorious litigants of their attorneys’ fees. Similarly, in Muniz v. Hoffman, the Court was reluctant to adopt an expansive reading of a section that “created an exception to the historic rule that there was no right to a jury trial in contempt proceedings.” To have applied the codification provision literally in that case would have disturbed a preexisting interaction between various labor statutes and upset a previously established legislative compromise that had survived challenge just the year prior to codification. In those circumstances, the Court insisted on some clear evidence of congressional intent to work “a substantial change in accepted practice” through the 1948 revision. Looking at these cases in their entirety, it seems to us that the Supreme Court has clearly expressed its distrust of phraseology bom of the revision-codification process in situations where the new language, if taken literally, would demonstrably conflict with settled precedent or policy, or significantly impede the operation of other, preexisting statutes. In other words, the Court’s refusal to accept the purported facial meaning of recodified provisions has been the product, not of an automatic reaction to the mere fact that the language was different, but rather of a wariness to regard particular word-changes as indicative of intended changes of a substantive nature absent some evidence other than the language difference per se. By contrast, a case in which the codified version presents a rule worded differently from but not fundamentally inconsistent with what went before would seem to present far less reason for judicial hesitance to conclude that Congress meant exactly what it said. At the same time, we recall that the Court has sometimes expressed its reasoning in codification cases in strong and unqualified terms. In Muniz, for example, it warned that “[i]n view of the express disavowals in the House and Senate Reports on the revisions of ... the Criminal Code ..., it would seem difficult at best to argue that a change in the substantive law could nevertheless be effected by change in the language of a statute without any indication in the Reviser’s Note of that change.” Of course, as earlier stated, the Reviser’s Note to the Trade Secrets Act contains nothing suggesting a conscious decision to criminalize disclosures not within the purview of any of the three predecessor provisions. Nevertheless, and realizing that the question is exceedingly difficult, we think the Trade Secrets Act, as it emerged from the revision, did exactly that. We base these conclusions partly on the view that a literal reading of the Act does not reflect a true conflict with the apparent original intent of the antecedent statutes. For example, prohibiting federal employees from disclosing trade secrets obtained in the course of any of their official duties is obviously not the same as only prohibiting them from disclosing those trade secrets revealed in Tariff Commission proceedings, but surely this difference does not amount to such a conflict as would raise our suspicion that Congress did not really mean what it ultimately said. If anything, it might be argued that, as no reasonable basis appears for predicating the confidential status of trade secrets when they come into the possession of the Government, the Act expresses more perfectly the fundamental intent of Congress that such data be protected from unauthorized disclosure. Our conclusion also rests in significant part on a perception of how much violence the sought-after construction would do to the language of the Trade Secrets Act. To arrive at the “true” meaning of the Act, one would have to parse every significant phrase and qualify each segment with conditions not even faintly suggested by the words themselves. Thus, disclosure of information that “concerns or relates to” “trade secrets [or] processes” would be prohibited only if it originated in Tariff Commission proceedings; secrecy of “operations, style of work, or apparatus” would be mandated only if these facts were known to the employee because of a personal visit to the site of manufacture. Similarly, “confidential statistical data” would be shielded if it was furnished in confidence to the Bureau of Foreign and Domestic Commerce, while nondisclosure of the “amount or source of income, profits, losses or expenditures” of any person or business would be required only if the employee discovered that information from examination of a tax return; and the proscription on revealing information about the “identity” of a person or corporation would apply only if its release were made in the course of publishing statistics on domestic commerce. We recognize that “ ‘words are inexact tools at best,’ ” but there is a vast difference between dissecting a statute to arrive at its essential meaning and butchering it. In the codification cases we have discussed, the Supreme Court was confronted with questions of meaning that turned on the addition or deletion of single words, or at most a phrase. In dealing with legislative projects the size of the 1948 revision and codification, we do well to bear in mind the danger of attaching too much significance to such small changes, for these may easily be the product of inadvertence or a failure to appreciate the full consequences of minute alterations. In the case at bar, however, the variation is on a much grander scale. To say that the drafters constructed the Trade Secrets Act as they did and yet thought and intended that it mean exactly what its statutory forerunners had meant is to charge them with a degree of carelessness or sheer blindness that is utterly inconsistent with everything the legislative history reveals about the high quality of the Revision Committee, and that we are unwilling to do. Quite importantly, it is a criminal statute that we are construing here. To be sure, the case against a construction that substantially rewrites the statutory language is arguably weaker when the result narrows rather than expands the reach of the provision in question. Still, we are considerably more reluctant to engraft numerous and significant qualifications onto an apparently clear, unambiguous text where the provision is criminal rather than civil in nature. We realize, however, that the drafters of the 1948 revision and codification were relying on the normal disinclination of courts lightly to infer intentional changes in meaning from mere changes in phrasing. We recognize, too, that one of the primary goals toward which the revisers labored in their years of study and reformulation was that of rendering the law more readily accessible and comprehensible. As Representative Keogh remarked when he introduced the bill that eventually was enacted, [i]t has been our aim in preparing this bill to furnish a modern, simple and understandable code of the Federal criminal law, in keeping with our motto, “making the laws understandable is as important as making the laws.” For all of these reasons, then, we conclude that the scope of the Trade Secrets Act is not delimited by that of its three predecessor statutes. To accept the facially unambiguous language of the Act as accurately expressing the congressional will does not require us to attribute to the 1948 enactment an abandonment of settled legal principles or an abrogation of preexisting precedent. It does not force us to assume that Congress sub silentio adopted a position it had earlier considered and rejected. It does not present us with a result that appears illogical or arbitrary in light of the general policies that have informed prior legislative action in the area, nor does it compel us to attribute to Congress an intent to alter the law in a hitherto discrete and specialized area by modifying a provision of general applicability. On the contrary, the changes wrought by the Trade Secrets Act represent a uniform, comprehensive, and reasonable though perhaps stringent approach to discouraging unauthorized disclosures of private commercial and financial data entrusted to the Government. It is our considered view, therefore, that the scope of the Act is at least co-extensive with that of Exemption 4 of FOIA, and that, in the absence of a regulation effective to authorize disclosure, the Act prohibits OFCCP from releasing any information in CNA’s affirmative action programs and EEO-1 reports that falls within Exemption 4. B. Application of Exemption 4 We next consider CNA’s contention that OFCCP erred in its interpretation and application of the legal standard summoned by FOIA Exemption 4. The controlling test in this circuit was articulated more than a decade ago in National Parks & Conservation Association v. Morton, and has since been consistently followed. In pertinent part, it states that commercial or financial information is “confidential” under Exemption 4 if disclosure of the information “is likely to ... cause substantial harm to the competitive position of the person from whom the information was obtained.” This criterion has been interpreted to require both a showing of actual competition and a likelihood of substantial competitive injury. Because OFCCP explicitly stated the proper standard to be applied, our review of the agency decision is confined to the question whether OFCCP correctly applied that standard to the facts of this case. As an initial matter, we note that the scope of review of OFCCP’s decision is governed by the Administrative Procedure Act. Review of the informal agency action contested here does not involve the “substantial evidence” test, for no hearing on the record was required. Rather, OFCCP’s action may be set aside only on one or more of a limited number of bases, of which the relevant one is that it is “arbitrary, capricious [or] an abuse of discretion.” Our inquiry must be “searching and careful,” but “[a] court is not empowered to substitute its judgment for that of the agency.” CNA and OFCCP disagree on what material is within the purview of Exemption 4. The data that OFCCP proposes to release can be grouped into three principal types: statistics on the racial and sexual composition of the workforce within various CNA departments; goals developed for equal employment purposes; and “applicant flow information” showing the percentage, by race and sex, of applicants hired from without and employees promoted from within. CNA’s objections and the responses thereto by OFCCP may also be arranged in three occasionally overlapping categories. First, OFCCP says that much of the information sought by CNA to be confined is already publicly available, and this assertion has not been contested before this court. To the extent that any data requested under FOIA are in the public domain, the submitter is unable to make any claim to confidentiality — a sine qua non of Exemption 4. We do not further consider CNA’s arguments regarding these data. Second, several of CNA’s claims with respect to other information relate not to alleged competitive harm but rather to anticipated displeasure of its employees or to adverse public reaction. CNA has protested, for example, that release of information on the number of women and minorities hired might result in unfavorable publicity. It also fears that its employees may become “demoralized” following disclosure of data showing the percentage of individuals promoted. We have previously found such complaints unrelated to the policy behind Exemption 4 of protecting submitters from external injury. These proffered objections simply do not amount to “harm flowing from the affirmative use of proprietary information by competitors.” The remaining disagreements between CNA and OFCCP concern the long-range consequences of the release of the data at issue. CNA submitted affidavits predicting a number of harmful effects; OFCCP, while offering no independent evidence, has answered these contentions with its own predictions of the repercussions of disclosure. One noteworthy objection by CNA to revelation of applicant flow data is that it would enable competitors more easily to direct their recruiting efforts to the best sources of potential employees. OFCCP counters with the logical rejoinder that these data will not be of any particular help to competitors since the employee-source and employee-position categories are broad, and the applicant pool is a function of the labor market and beyond an individual competitor’s control. These and other similar contentions presented no more than two contradictory views of what likely would ensue upon release of information that CNA sought to protect. In each case, OFCCP retorted with reasonable and thorough prognoses of its own. We thus are confronted by the type of judgments and forecasts courts traditionally leave largely to agency expertise, with judicial review limited by the narrow standard sanctioned. After careful consideration of OFCCP’s decision, we are satisfied that it cannot in any way be characterized as arbitrary, capricious, or an abuse of discretion. CNA’s objections were answered fully, and OFCCP’s explanations of anticipated effects were certainly no less plausible than those advanced by CNA. Each of OFCCP’s explanations is well reasoned, logical and consistent, and predictive judgments are not capable of exact proof. We find OFCCP’s application of Exemption 4 entirely rational, and therefore legally permissible. We find support for these conclusions in the deference the Supreme Court has accorded agency forecasts in other contexts. In a typical example, FPC v. Transcontinental Gas Pipe Line Corp., the Court upheld action by the Federal Power Commission based in part on a prediction that widespread, unrestricted direct sales of natural gas would probably result in price increases. This prediction was grounded solely on the agency’s expertise, and not on any material in the administrative record. The Court specifically rejected the argument that the Commission “should have adduced testimonial and documentary evidence to the effect that this forecast would come true.” More recently, in FCC v. National Citizens Committee for Broadcasting, the Court held that an agency’s “judgmental or predictive” determinations need not be supported by record evidence. The Court upheld, in the context of formal rulemaking regulations, barring initial licensing and transfer of broadcast facilities when a station and a daily newspaper in the same community shared common ownership. The Court explained that “ ‘a forecast of the direction in which future public interest lies necessarily involves deductions based on the expert knowledge of the agency.’ ” We are aware that FOIA applies to all executive agencies subject to its requirements, and that no agency is charged singly with its enforcement. OFCCP, however, stands out in its endeavor to acquire additional expertise by hiring a consultant to evaluate the data submitted by CNA. The report submitted by OFCCP’s outside expert served as a fair substitute for agency experience and, when coupled with OFCCP’s own evaluation of CNA’s objections, produced a well-supported agency forecast. Given the Supreme Court’s recognition of agencies’ freedom to make predictive determinations in more formal settings without additional evidence, we conclude that the same discretion must be indulged to OFCCP’s evaluations of the effect of release of the documents in dispute. IV. Procedural Issues In addition to the substantive questions we have discussed, CNA presents procedural challenges to the manner in which OFCCP and the District Court undertook to adjudicate its claims of serious competitive harm. A review of the procedural course by which this case arrived in this court will assist explanation of these claims. A. Procedural History After obtaining from the District Court, in April, 1977, an order restraining release pending completion of the administrative appeal process, CNA filed with OFCCP a five-volume index specifying its objections to FOIA disclosure. CNA supported its factual allegations with affidavits or depositions of its director of employment, its vice-president for personnel, an outside personnel recruiter — or “headhunter” — and a consultant with a doctorate in business administration. No oral presentation was made. In October, 1978, OFCCP issued its first decision, which ordered release of the affirmative action programs and EEO-1 reports, with wage and certain identifying data deleted. CNA promptly returned to the District Court seeking an injunction against any disclosure pending completion of judicial review. On the District Court’s refusal to impose the requested restraint, CNA applied to this court and was granted an interim injunction. We also stayed all proceedings in the case in order to benefit from the Supreme Court’s impending decision in Chrysler Corporation v. Brown. When Chrysler issued, we remanded the case to the District Court with directions to “order the appropriate agenc[y] to make new administrative determinations in accordance with [its] normal rules of practice and in light of the Supreme Court’s opinion in Chrysler Corporation v. Brown.” Thus, in November, 1979, the matter was again before OFCCP. At this juncture, CNA asked the agency to afford it an evidentiary hearing, at which CNA presumably would have presented expert testimony and examined witnesses on whose opinions OFCCP intended to rely. By letter from the director of OFCCP, such a hearing was denied; instead, the agency offered to accept any written submissions CNA cared to make. In November, 1980, OFCCP issued its second and most detailed decision. As we stated at the outset, the agency rejected CNA’s legal arguments on the scope of FOIA Exemptions 3 and 4 and the Trade Secrets Act, and, as well, its factual assertions of competitive harm that allegedly would be precipitated by disclosure. Returning once again to the District Court, CNA filed an opposition to the agency’s decision. It tendered four supplemental affidavits aimed at controverting OFCCP’s conclusions about staleness of the requested data, and attacking several other elements of the agency’s reasoning. At this point, some collateral procedural problems arose. It appeared that the agency, in forwarding the administrative record to the District Court, omitted three depositions CNA had submitted at the onset of the controversy. More significantly, CNA discovered that OFCCP had engaged its own outside expert, Professor Boyd Fjelsted, to review CNA’s submissions and render an opinion on the substantiality of the competitive threat they described. CNA demanded the opportunity to review and respond to Fjelsted's reports; OFCCP adamantly refused to reveal them; and the District Court ordered in camera inspection. After reviewing the reports, the court sustained OFCCP’s refusal to disclose them on grounds that they constituted privileged “deliberative materials” of an “agency decisionmaker.” The court agreed with CNA, however, that the three omitted depositions should have been made a part of the record upon which the agency rested its decision. It therefore remanded the matter to OFCCP with directions that the agency consider the depositions. On remand, CNA again requested an evidentiary hearing, which again was denied. At the same time, it tendered to OFCCP the four supplemental affidavits it had presented to the District Court. OFCCP’s third decision, issued in July, 1981, reaffirmed its earlier conclusions. It found the depositions either merely cumulative of other material it had already reviewed, or, in some instances, actually inconsistent with that material. The agency refused to take the four supplemental affidavits into account on grounds that they were beyond the scope of the District Court’s remand order. OFCCP represented nonetheless that it had looked at the affidavits and found them insufficient to change its views. Back in the District Court again, CNA requested alternatively a remand to OFCCP with directions to hold an evidentiary hearing, or resolution of the factual issues de novo in the District Court. The court turned down the plea for remand. It held that, contrary to CNA’s interpretation, OFCCP regulations did not contemplate an evidentiary hearing in order to evaluate a contractor’s opposition to FOIA-release of its affirmative action materials. The court also denied CNA’s demand for consideration of the issues de novo. It reasoned that the Administrative Procedure Act fixed the standard and scope of review at determination whether OFCCP’s decision was arbitrary or capricious on the basis of the administrative record. Accordingly, the court granted judgment for the agency without independently receiving any evidence. B. Adequacy of OFCCP Procedures CNA’s broadest challenge is to the procedure employed by OFCCP in evaluating the competitive effect of releasing CNA’s filings. CNA maintains that the agency’s refusal to afford an evidentiary hearing violated its rights under the Due Process Clause, the Administrative Procedure Act, and OFCCP’s own regulations. CNA apparently believes that such a hearing is essential at the agency level to assure an adequate resolution of factual issues. We cannot agree. We note initially, as did the District Court in its opinion, the regulation governing contractors’ objections to disclosure of affirmative action materials. It requires a contractor to identify “the reasons why such information is not disclosable” and, after an initial determination by OFCCP personnel, directs the agency to inform the contractor of its decision. The regulation also provides for appeal of that ruling to the director of OFCCP, who must then render a “final determination.” The regulation makes no mention of an evidentiary hearing, or indeed of any review procedures at all. We can hardly take issue with the District Court’s finding that OFCCP did not transgress its own skeletal constraints. Hardly more substantial is CNA’s claim that the agency factfinding procedures ran afoul of Section 10 of the Administrative Procedure Act and impinged on the full spectrum of review assured by the Due Process Clause. This court recently entertained a nearly identical complaint in NOW, in which we found the OFCCP review sufficient to allay concerns about fairness to the submitters of information. The holding in NOW is dispositive on this issue, and we dismiss without further discussion CNA’s claim in that regard. C. OFCCP’s Expert’s Report CNA registers an additional objection predicated upon OFCCP’s refusal to produce Dr. Fjelsted’s report for CNA’s examination and rebuttal. CNA believes that the information contained in this report weighed heavily in OFCCP’s assessment of competitive effect, and that CNA’s inability to respond specifically to the expert’s conclusion constituted reversible error. The agency, for its part, asserts that the report was a deliberative, predecisional document privileged against disclosure through discovery. We discuss this issue separately because it demands careful balancing of the competing interests of both sides. Our analysis begins, as it must, with a guarantee attending agency adjudication. A precept fundamental to the administrative process is that a party must have an opportunity to refute evidence utilized by the agency in decisionmaking affecting his or her rights. A decade ago, in Ralpho v. Bell, we declined to uphold a property valuation by the Micronesian Claims Commission that was based in part on evidence unavailable to Ralpho. A “value study,” conducted and used by the Commission in assessing property claims, was actually an assemblage of interviews, records, and other data relating the average price of goods and services in Micronesia. We held that, the Commission, by denying Ralpho an opportunity to inspect and counter the factual information compiled in the study, impermissibly truncated even the minimal procedures required of an agency. The general requirements of disclosure to a litigant of material to be considered by an agency in an adjudicative proceeding is modified where the agency asserts a privilege respecting material generated in the process of agency decision-making. When agency material is “deliberative” or “recommendatory” in character, and does not of itself inject new factual data into the calculus, the agency is privileged to withhold it. This privilege is designed to ensure the full measure of agency decisionmaking; by removing the chilling effect of possible future disclosure, inhibitions on candid expression are dissolved. The materials at issue here consist solely of Dr. Fjelsted’s analysis of the information already submitted by CNA to OFCCP and his recommendations as to the course he felt it should follow. Dr. Fjelsted did not provide any new data of his own. The District Court conducted an in camera inspection of Dr. Fjelsted’s reports and concluded: A review of the reports and the affidavit by Kenneth G. Patton, acting director of the OFCCP, demonstrates that ... [Fjelsted’s reports] contain document-by-document recommendations to the agency as to how it should act on CNA’s claims that disclosure would lead to competitive harm. These recommendations were part of the agency give-and-take by which its final determination was made. It is clear enough to us that OFCCP was entitled to shield from discovery reports consisting solely of analyses of data and recommendations of agency action predicated thereon. Unlike the appellant in Ralpho v. Bell, CNA was not confronted with an unascertainable store of knowledge which it was called upon haphazardly to rebut. The factual information relied on by Dr. Fjelsted was, of course, available to CNA; indeed, much of it was supplied by CNA itself. Furthermore, Dr. Fjelsted’s analysis, though perhaps influential in OFCCP’s appraisal of the commercial impact of the contested material, was in no sense binding, for OFCCP was free to modify or even reject these analyses and recommendations. We conclude that CNA has no legally cognizable basis for complaining of OFCCP’s decision to withhold Dr. Fjelsted’s report. It likewise is clear that the agency’s privilege to withhold the reports is unaffected by the fact that they were prepared by a consultant from outside the agency. In Ryan v. Department of Justice, we recognized that [i]n the course of its day-to-day activities, an agency often needs to rely on the opinions and recommendations of temporary consultants, as well as its own employees. Such consultations are an integral part of its deliberative process; to conduct this process in public view would inhibit frank discussion of policy matters and likely impair the quality of decisions. Thus, in that case we held exempt from production those nonfactual portions of responses by Senators to questionnaires propounded by the Attorney General. Any material that might reveal decisionmaking or policymaking activity was considered privileged. Similarly, courts have repeatedly found that a privilege attaches to reports of outsiders commissioned by an agency to perform agency work, when such reports would be protected if compiled within the agency itself. Whether the author is a regular agency employee or a temporary consultant is irrelevant; the pertinent element is the role, if any, that the document plays in the process of agency deliberations. If information communicated is deliberative in character it is privileged from disclosure, notwithstanding its creation by an outsider. A moment’s reflection will reveal the reason why. Professor Davis, in discussing the difficulty of the task confronting agency decisionmakers, has commented that “able administrators ... have almost uniformly concluded that deciding officers should have the assistance both of reviewing staffs and of agency specialists.” Moreover, it is clear that “deciding officers need the special strength that comes from ready access to staff specialists.” Then, too, federal agencies occasionally will encounter problems outside their ken, and it clearly is preferable that they enlist the help of outside experts skilled at unravelling their knotty complexities. This is not to say that any material derived from an outside expert is inviolable; factual data, for example, are still susceptible to discovery. But where, as here, a consultant is retained to evaluate information and submit recommendations as to decisions thereon, the advice or opinion transmitted to the agency is subject to privileged withholding. To force an exposure is to “stifle honest and frank communication” between agency and expert by inhibiting their free exchange of thought. D. Proceedings Before the District Court CNA’s final procedural challenge is to the level of scrutiny afforded in the District Court. CNA believed that it was entitled to de novo review, replete with testimony and cross-examination. The District Court disagreed, confining itself to an examination of the record compiled before OFCCP. Our reading of the Supreme Court’s teachings in Camp v. Pitts and Citizens to Preserve. Overton Park, Inc. v. Volpe, together with our own conclusion in NOW, convinces us that the District Court behaved entirely correctly. Both Camp and Overton Park authorize de novo judicial review under the Administrative Procedure Act only when the agency’s “factfinding procedures are inadequate.” Our decision in NOW upheld the procedures employed by the OFCCP in reverse-FOIA actions — procedures which have largely been replicated here. We thus sustain the District Court in its refusal to review the case de novo. V. Conclusion We agree with the District Court that “the agency’s decision thoroughly discusses CNA’s objections and presents ‘a reasoned and detailed basis for its decision.’ ” Because the agency’s decision to release the documents found to be outside Exemption 4 must be upheld, and because the procedures employed by OFCCP were legally sufficient, the stay instituted pending appeal is dissolved and the decision of the District Court is Affirmed. . "Reverse-FOIA” actions are now a common species of FOIA litigation. Jurisdiction over these cases is conferred by 28 U.S.C. § 1331(a) (1982), while § 10(a) of the Administrative Procedure Act (APA), 5 U.S.C. § 702 (1982), supplies the cause of action. Chrysler Corp. v. Brown, 441 U.S. 281, 317 & n. 47, 99 S.Ct. 1705, 1725 & n. 47, 60 L.Ed.2d 208, 234 & n. 47 (1979). Typically, a submitter of information — usually a corporation or other business entity required to report various and sundry data on its policies, operations, or products — seeks to prevent the agency that collected the information from revealing it to a third party in response to the latter’s FOIA request. The agency's decision to release the data normally will be grounded either in its view that none of the FOIA exemptions applies, and thus that disclosure is mandatory, or in its belief that release is justified in the exercise of its discretion, even though the data fall within one or more of the statutory exemptions. . The Trade Secrets Act is part of the 1948 revision and codification of Title 18, the Criminal Code. See Act of June 25, 1948, ch. 645, § 1905, 62 Stat. 683, 791, amended by Pub.L. No. 96-349, § 7(b), 94 Stat. 1158 (Sept. 12, 1980). The text of the Trade Secrets Act appears infra at text accompanying note 39. . 5 U.S.C. § 552(b)(3) (1982). The text of Exemption 3, the withholding statute exemption, appears infra at text accompanying note 35. . 5 U.S.C. § 552(b)(4) (1982). The text of Exemption 4, the trade secrets and confidential financial information exemption, appears infra at note 71. . We háve acknowledged, and even indicated our views on, these questions in prior cases. See, e.g., United States Int'l Trade Comm'n v. Tenneco West, 261 U.S.App.D.C. 341, 345-346 & nn. 4-5, 822 F.2d 73, 77-78 & nn. 4-5 (1987); Webb v. Department of Health & Human Serv., 225 U.S.App.D.C. 19, 26 n. 48, 696 F.2d 101, 108 n. 48 (1982); Goland v. CIA, 197 U.S.App.D.C. 25, 36 n. 61, 607 F.2d 339, 350 n. 61 (1978), cert. denied, 445 U.S. 927, 100 S.Ct. 1312, 63 L.Ed.2d 759 (1980); National Parks & Conservation Ass'n v. Kleppe, 178 U.S. App.D.C. 376, 389-390, 547 F.2d 673, 686-687 (1976); Charles River Park “A", Inc. v. HUD, 171 U.S.App.D.C. 286, 292 n. 7, 519 F.2d 935, 94