(dissenting).
while I agree with most of the majority opinion, I disagree with the result. I think it unlikely that the district court, on remand, will find a policy reason strong enough to prevent disclosure; I would af-f|rm its judgment now.
National Parks and Conservation Association v. Morton 498 F 2d 765 (D.C.Cir.1974); and ^ majority>s 0pinion make clear that Congress did not intend to apply the word “confidential” automatically whenever an agency official agrees to a business request to keep information confidential. Otherwise, given the temptation of government and business officials to follow the path of least resistance and say “confidential” whenever they seek to satisfy the government’s vast information needs, the exemption would expand beyond what Congress intended. Thus, National Parks (which I agree is illustrative and does not lay down a dispositive test) looked beyond the simple promise of confidentiality and examined whether either the business or government had a legitimate interest in ascribing confidential status to the information. And, the majority holds that the government can invoke the exemption only if it shows how disclosure will significantly harm some relevant private or governmental interest.
The record, in my view, shows that the legitimate business and governmental interests against disclosure are unusually weak in this case. First, the BSSG has entered into a consent decree with the Attorney General of Massachusetts promising to make all future anonymous salary compilations public. It is not obvious why there would be a significantly greater need to *13keep past compilations private. Moreover, as the Commonwealth’s suit against the BSSG suggests, employers who collect cost information, circulate it only among themselves, and deny others access to it, risk antitrust difficulties. Their practice may be viewed as an agreement among buyers of a factor (secretarial services), see Mandeville Island Farms, Inc. v. American Crystal Sugar Co., 334 U.S. 219, 68 S.Ct. 996, 92 L.Ed. 1328 (1948); Cordova v. Bache & Co., 321 F.Supp. 600 (S.D.N.Y.1970); 1 P. Areeda & D. Turner, Antitrust Law ¶ 229a (1978), to circulate price information, see United States v. Container Corp. of America, 393 U.S. 333, 89 S.Ct. 510, 21 L.Ed.2d 526 (1969); Sugar Institute, Inc. v. United States, 297 U.S. 553, 56 S.Ct. 629, 80 L.Ed. 859 (1936); American Column & Lumber Co. v. United States, 257 U.S. 377, 42 S.Ct. 114, 66 L.Ed. 284 (1921); Tag Manufacturers Institute v. FTC, 174 F.2d 452 (1st Cir. 1949); P. Areeda, Antitrust Analysis ch. 3C (3d ed. 1981), where the purpose or effect of the agreement is to suppress competition and where the agreement provides no offsetting benefit through the public dissemination of the information. I do not know whether the Commonwealth Attorney General would have proven such an antitrust violation in his suit, for the parties settled the case by consent decree. But, the legitimate business need for confidentiality is far from obvious in the case of a confidential salary information-sharing agreement among competitors.
Second, the government’s need for confidentiality seems slight. The government activity at issue — the setting of salaries — is peripheral to the main job of the Federal Reserve Board and its regional banks. Normally government salary setting is conducted in public. Of course, the Federal Reserve Banks set their salaries differently and they are authorized by the Board to find out what comparable businesses pay their workers. But, these facts do not show any special need to enter into what was evidently a legally questionable, private salary information-sharing agreement among various competing organizations. Rather, the only interest to which the government can point is its general interest in keeping a promise of confidentiality — an interest that is at stake whenever a promise of confidentiality is made, but an interest that is weakest when the information is related as peripherally to the agency’s main mission as it is here. To accept the threat to this interest as determinative in this case would come close to making a simple “request plus promise” necessarily sufficient to apply the “confidentiality” exemption.
In the majority’s view, these matters are better explored on remand. However, this litigation over a few documents has gone on for some time. It has been the subject of several district court opinions. No information of great moment is involved. And, the fact that a directly related state court consent decree exists prevents the case from having much precedential significance. It thus seems to me better to settle the matter now than to remand the case for yet another district court determination, another opinion, and possibly still another appeal. The government has had sufficient opportunity to demonstrate its interest in the confidentiality of the information. I do not think it has done so adequately. It is unlikely that the district court can reach a different result under the standards the majority instructs it to apply. Therefore, I would simply affirm the district court’s result.