Petitioner, Consumers Union of U.S., Inc., challenges the Federal Trade Commission's "Used Car Rule," a Trade Regulation Rule governing the sale of used motor vehicles. See 16 C.F.R. Part 455 (1986). Petitioner objects to the Commission's decision to omit from the final rule the so-called "known-defects provision," which would have required used-car dealers to list on the window sticker affixed to any used car offered for sale the presence of certain mechanical defects of which they have knowledge. The principal issues addressed are the legality of the procedures by which the Commission withdrew the known-defects provision following remand of the initial rule by a reviewing court, and the substantive validity of the withdrawal.I

The tumultuous history of the Used Car Rule began in 1975, with the enactment of the Magnuson-Moss Warranty--Federal Trade Commission Improvement Act ("Magnuson-Moss Act"), Pub.L. No. 93-637, 88 Stat. 2183 (1975) (codified at 15 U.S.C. Secs. 2301-2312 (1982) and other scattered sections of 15 U.S.C.), which, in Sec. 109(b), directed the Commission to promulgate a rule regulating "warranties and warranty practices in connection with the sale of used motor vehicles," 15 U.S.C. Sec. 2309(b). In 1981, relying upon Sec. 109(b) and upon Sec. 18(a)(1) of the Federal Trade Commission Act ("FTC Act"), ch. 311, 38 Stat. 717 (1914), as added by the Magnuson-Moss Act, Sec. 202, 88 Stat. 2193 (codified as amended at 15 U.S.C. Sec. 57a(a)(1) (1982)), which grants the Commission rulemaking authority to regulate "unfair or deceptive" trade practices, the Commission promulgated a rule that required used-car1 dealers to post on a standard window sticker several consumer warnings, the terms of any warranty, and a list of certain specific mechanical defects known to the dealer. 46 Fed.Reg. 41,328 (1981) ("Initial Rule").

Timely petitions for review of the Initial Rule were filed in the United States Court of Appeals for the Second Circuit, pursuant to Sec. 18(e) of the FTC Act, 15 U.S.C. Sec. 57a(e), attacking primarily the known-defects provision. Miller Motor Car Corp. v. FTC, Nos. 81-4144, 81-4172, 81-4216. A lengthy delay interrupted the case, during which Congress vetoed the Initial Rule, see 128 CONG.REC. 10,343-44 (1982); 128 CONG.REC. H2856-58 (daily ed. May 26, 1982), and the Supreme Court, in turn, declared the legislative veto unconstitutional, see United States Senate v. FTC, 463 U.S. 1216, 103 S. Ct. 3556, 77 L. Ed. 2d 1402, 1403, 1413 (1983); United States House of Representatives v. FTC, 463 U.S. 1216, 103 S. Ct. 3556, 77 L. Ed. 2d 1402, 1403, 1413 (1983). While Miller was pending, the Commission announced its intention to consider modifying the Initial Rule. 48 Fed.Reg. 36,096 (1983). Shortly thereafter, alleging that the Commission had never afforded them the requisite opportunity to comment on the known-defects provision, the Miller petitioners moved the court to remand the case to the Commission pursuant to Sec. 18(e)(2) of the FTC Act, 15 U.S.C. Sec. 57a(e)(2), for additional oral and written presentations on the Initial Rule. By stipulation of the parties, in which the Commission disputed the alleged procedural irregularity, the court remanded the case pursuant to Sec. 18(e)(2), retaining jurisdiction.

After soliciting further evidence and comments on the Initial Rule, and particularly comments on the know-defects provision, 48 Fed.Reg. 55,874 (1983), the Commission (in addition to making minor changes not relevant here) revised the Initial Rule to delete the known-defects provision. 49 Fed.Reg. 45,692 (1984) ("Revised Rule").

Petitioner sought review of the Revised Rule in this court, challenging both the Commission's decision to omit the known-defects provision and the procedures it followed in arriving at that decision. The Second Circuit transferred Miller to this court and we designated the National Automobile Dealers Association ("NADA") and the National Independent Automobile Dealers Association, which were petitioners in Miller, as intervenors in this case.

We address first petitioner's contention that the Commission, in promulgating the Revised Rule, neglected to provide statutorily required procedures. The essence of the dispute is this: Section 18(e)(2) of the FTC Act, pursuant to which the Second Circuit remanded the Initial Rule to the Commission, provides that the reviewing court may, at the request of either the petitioner or the Commission, order the Commission to provide opportunity for "additional oral submissions or written presentations" if "there were reasonable grounds for ... failure to make such submissions and presentations in the proceeding before the Commission." 15 U.S.C. Sec. 57a(e)(2). It continues:

The Commission may modify or set aside its rule or make a new rule by reason of the additional submissions and presentations and shall file such modified or new rule, and the rule's statement of basis of [sic ] purpose, with the return of such submissions and presentations. The court shall thereafter review such new or modified rule.

Id. By contrast, Sec. 18(d)(2)(B) of the FTC Act, 15 U.S.C. Sec. 57a(d)(2)(B), requires that "[a] substantive amendment to, or repeal of, a rule promulgated under subsection (a)(1)(B) of this section [as was the Initial Rule in the present case] shall be prescribed ... in the same manner as a rule prescribed under such subsection." That manner of prescription includes not merely the opportunity for oral presentation and written submission, 15 U.S.C. Sec. 57a(c)(2)(A), but also opportunity "to present such rebuttal submissions and to conduct (or have conducted ...) such cross-examination of persons as the Commission determines (i) to be appropriate, and (ii) to be required for a full and true disclosure" with respect to disputed issues of material fact, 15 U.S.C. Sec. 57a(c)(2)(B). Rebuttal submissions were permitted in the present case. The Commission did not, however, either permit cross-examination or determine that it was not "appropriate" or not "required for a full and true disclosure" with respect to disputed issues of material fact. It is this failure petitioner complains of--and whether the complaint is justified depends upon (1) whether Sec. 18(e)(2) requires such procedures, and (2) if not, whether Sec. 18(e)(2) or rather Sec. 18(d)(2)(B) governed the proceedings on remand.

The first point is not contested. As odd as it might seem to allow supplementation of the rulemaking record in a fashion different from that by which the original record itself was compiled, that is precisely what Sec. 18(e)(2) prescribes, in the limited circumstances in which it is applicable. And there is some reason for the oddity. The Sec. 18(e)(2) procedure, while not as demanding as that of Sec. 18(d)(2)(B) (or Sec. 18(c)) is fully as demanding as that of normal rulemaking under the Administrative Procedure Act ("APA"). See 5 U.S.C. Sec. 553(b), (c) (1982). In fact, it is somewhat more demanding, since it permits the remanding court to require receipt of not merely written comments but oral presentations. Congress evidently thought that this compromise procedure was a reasonable accommodation to the need for expedition where a lengthy completed Sec. 18(c) proceeding needs to be supplemented.

The issue reduces itself, therefore, to whether Sec. 18(e)(2) governed the remand. Petitioner's principal argument that it does not is based upon the simple (and simply rejected) premise that Sec. 18(d)(2)(B) categorically requires Sec. 18(c) procedures for "substantive amendment to ... a rule promulgated under subsection (a)(1)(B)." It does indeed contain that requirement, but the requirement cannot possibly be categorical, since it must be reconciled with Sec. 18(e)(2), which explicitly authorizes the Commission to "modify or set aside its rule or make a new rule by reason of the additional submissions and presentations." The manner of reconciliation is obvious: the traditional principle of construction that the more specific prevails over the more general. It is Sec. 18(e)(2), and not Sec. 18(d)(2)(B), that governs the narrow category of case in which a reviewing court authorizes reopening of the record under the former section. There is no question that is what occurred here. Though petitioner at one point puzzlingly asserts "the impropriety of treating the proceedings used by the Commission here as if they were made pursuant to a section [18(e)(2) ] remand," Reply Brief for Petitioner at 14 (emphasis added), it is entirely beyond doubt that that is precisely how they were "made." The Second Circuit remanded the Initial Rule to the Commission expressly "[p]ursuant to [Sec. 18(e)(2),] 15 U.S.C. Sec. 57a(e)(2)," Miller Motor Car Corp. v. FTC, Nos. 81-4144, etc., slip op. at 1 (2d Cir. Sept. 14, 1983) (Order), and directed the Commission to "reopen the record with respect to" the known-defects provision and "provide petitioners and all other interested persons with notice and an opportunity to submit comments and rebuttal thereto," leaving it to the Commission's discretion whether to allow oral presentations or any "other proceedings." Slip op. at 4 (Stipulation p 2, incorporated by reference into remand order).

Petitioner suggests, however, that a Sec. 18(e)(2) remand was improper because it is only available where evidence has been improperly excluded in the original proceeding. That limitation has absolutely no foundation in the text of the statute, and we decline to limit the plain meaning of the text by resort to bits of legislative history that describe some, but not necessarily all, of the purposes that the provision serves. See Block v. Meese, 793 F.2d 1303, 1310 (D.C.Cir.1986). Cf. Vargas-Gonzalez v. INS, 647 F.2d 457 (5th Cir.1981) (interpreting 28 U.S.C. Sec. 2347(c), a remand provision similar to Sec. 18(e)(2)). We do not fear that reading the text as it is written will, as petitioner threatens, enable the Commission to write the revocation provision of Sec. 18(d)(2)(B) out of existence whenever it wishes to "revok[e] a rule while a protracted judicial review of the rule is still pending." Reply Brief for Petitioner at 13. The Commission can only make use of the Sec. 18(e)(2) procedures by order of the reviewing court. Cf. Exxon Corp. v. Train, 554 F.2d 1310, 1316 n. 12 (5th Cir.1977) (interpreting 33 U.S.C. Sec. 1369(c), similar remand provision of Federal Water Pollution Control Act Amendments of 1972); Greater Boston Television Corp. v. FCC, 463 F.2d 268, 283 (D.C.Cir.1971) (interpreting 28 U.S.C. Sec. 2347(c)), cert. denied, 406 U.S. 950, 92 S. Ct. 2042, 32 L. Ed. 2d 338 (1972).

Petitioner further suggests that the Sec. 18(e)(2) remand was unlawful because the Second Circuit "held no hearing to determine the propriety of the remand," Reply Brief for Petitioner at 15, and made no specific finding of the requisite "reasonable grounds for ... failure to make" the relevant submissions and presentations in the original proceeding, 15 U.S.C. Sec. 57a(e)(2). But the statute does not require those formalities. Finally, petitioner's assertion that "there is no indication that the Second Circuit was aware of [Sec. 18(d)(2)(B) ], or that the court analyzed the propriety of the statutory remand in light of that section," Reply Brief for Petitioner at 15, is patently incredible. The Miller petitioners had specifically directed the court's attention to the cumbersome FTC Act rulemaking procedures: it was the Commission's alleged failure to follow them that was the very basis of their procedural attack. The court could not have missed the statutory directive, codified a mere subsection away from Sec. 18(e)(2), that applied the same procedures to the amendment or repeal of an FTC rule. And even if it were unaware of that provision it would have assumed the same result in any event, since procedural equation of revocation and promulgation is the general (indeed, as far as we are aware, the invariable) rule. See 5 U.S.C. Sec. 551(5) (1982); Motor Vehicle Manufacturers Association v. State Farm Mutual Automobile Insurance Co., 463 U.S. 29, 41, 103 S. Ct. 2856, 2865, 77 L. Ed. 2d 443 (1983) ("State Farm "); see also 15 U.S.C. Sec. 2058(h) (1982) (Consumer Product Safety Act); 21 U.S.C. Sec. 371(e)(1) (1982) (Food, Drug, and Cosmetic Act).

We are unable, on the record before us, to identify the particular reasons why the Second Circuit found "reasonable grounds for the ... failure to make" the submissions that were the subject of its Sec. 18(e)(2) remand in the original proceeding. That it made such a remand, however, is unquestionable, and that decision is, at the very least, the law of this case if not an entirely separate determination of another court immune from collateral attack here. Even if it is merely the former, we cannot reconsider it absent a finding of clear error causing "manifest injustice." See Laffey v. Northwest Airlines, Inc., 740 F.2d 1071, 1082-83 & n. 18 (D.C.Cir.1984) (per curiam), cert. denied, 469 U.S. 1181, 105 S. Ct. 939, 83 L. Ed. 2d 951 (1985); City of Cleveland v. FPC, 561 F.2d 344, 346 (D.C.Cir.1977) (law of case binding on administrative agency). We can find no "manifest injustice" produced either by the protraction of these rulemaking proceedings or by the deprivation of cross-examination (which is standard in the vast majority of rulemakings).

Before turning to the merits, we pause to consider our standard of review. We may not set aside the Revised Rule unless it is "arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law," see 15 U.S.C. Sec. 57a(e)(3) (1982) (incorporating by reference 5 U.S.C. Sec. 706(2)(A) (1982)), or the factual determinations on which it is based are "not supported by substantial evidence in the rulemaking record ... taken as a whole," 15 U.S.C. Sec. 57a(e)(3)(A). See American Financial Services Association v. FTC, 767 F.2d 957, 985 (D.C.Cir.1985), cert. denied, --- U.S. ----, 106 S. Ct. 1185, 89 L. Ed. 2d 301 (1986). Although this court flirted briefly with the notion that the "substantial evidence" standard of the FTC Act called for a more intensive review than the "arbitrary and capricious" standard applicable to rulemaking under the APA, see American Optometric Association v. FTC, 626 F.2d 896, 904-05 (D.C.Cir.1980), our more recent pronouncements are to the contrary. We have held that in the context of the APA, the substantial evidence test (which is applied only to formal adjudication and formal rulemaking, see 5 U.S.C. Sec. 706(2)(E)) and the arbitrary and capricious test (which governs review of all proceedings, see 5 U.S.C. Sec. 706(2)(A)) "are one and the same" insofar as the requisite degree of evidentiary support is concerned. Association of Data Processing Service Organizations, Inc. v. Board of Governors of the Federal Reserve System, 745 F.2d 677, 683 (D.C.Cir.1984). We have noted, moreover, that the APA standard of review must be deemed unaltered unless "Congress's intent to make a substantive change [is] clear" on the statute's face. Id. at 686. As application of these principles would require, our most recent review of a trade rule described the substantial evidence test of Sec. 57(e)(3)(A) in the same fashion as the substantial evidence test of the APA is described, i.e., as demanding " 'such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.' " See American Financial Services Association v. FTC, 767 F.2d at 985 (citations omitted). Compare Consolo v. FMC, 383 U.S. 607, 619-20, 86 S. Ct. 1018, 1026-27, 16 L. Ed. 2d 131 (1966). This is, as we have said, the same degree of evidentiary support needed to satisfy the arbitrary and capricious standard.

The Commission asserts that the gravamen of the present appeal is simply its "decision not to engage in rulemaking," see Brief for Respondent at 23--so that presumably our review would be governed by the deferential standard applicable to an agency refusal to adopt a rule not required by law. See Professional Drivers Council v. Bureau of Motor Carrier Safety, 706 F.2d 1216, 1223 n. 24 (D.C.Cir.1983) ("[W]hile an agency may possess the authority to regulate an area, substantial discretion will be accorded an agency's decision not to promulgate regulations."); id. at 1220-21, 1223; WWHT, Inc. v. FCC, 656 F.2d 807, 818 (D.C.Cir.1981) ("It is only in the rarest and most compelling of circumstances that this court has acted to overturn an agency judgment not to institute rulemaking."); id. at 816-17. We must reject this contention. When an agency has declined entirely to enter a particular arena of regulation its discretionary judgment as to where its administrative and enforcement resources should best be expended is prominently involved. That is not the case where the agency merely fails to include in a final rule a proposed provision that is closely related to the provisions that are adopted. The latter case, which is what we have here, has never been thought to present an issue of agency refusal to adopt a rule but rather an issue of the rationality of the rule it did adopt, given the alternatives that were suggested and rejected. See Vermont Yankee Nuclear Power Corp. v. NRDC, Inc., 435 U.S. 519, 551, 98 S. Ct. 1197, 1215, 55 L. Ed. 2d 460 (1978); American Financial Services Association v. FTC, 767 F.2d at 988; cf. Public Citizen v. Steed, 733 F.2d 93, 103-04 (D.C.Cir.1984) (failure to consider alternatives to rescission).

At the other end of the spectrum, we reject petitioner's argument that our review must be particularly strict and probing in the present case because the agency has revoked a preexisting rule. Reply Brief for Petitioner at 17-22. We have already disapproved that characterization of what has occurred. Section 18(e)(2) furnishes the Commission an opportunity to promulgate the ultimate rule with the benefit of the new evidence that Sec. 18(e)(2) accommodates, not merely an opportunity to repeal the old rule if repeal can be substantiated on the basis of that new evidence. The proceedings conducted after remand in the present case were part of the same rulemaking proceeding as that which produced the Initial Rule; and any revisions made during those proceedings are to be treated like any other agency decision to alter its course during an ongoing rulemaking proceeding--that is, the Commission "is clearly free to reject an early conclusion after further study and reflection. All that is important is ... that the final [action is] supported by substantial evidence." Marathon Oil Co. v. EPA, 564 F.2d 1253, 1272 (9th Cir.1977) (citing Consolo v. FMC, 383 U.S. at 620, 86 S.Ct. at 1026).

The Commission rejected the known-defects disclosure alternative because in its view "the record does not support ... a conclusion that [its] benefits ... outweigh its costs." 49 Fed.Reg. at 45,712. We address each side of the Commission's balance, in turn.

The Commission correctly reasoned that the benefits of the provision would depend upon (1) the extent to which dealers operating under it would know of specific mechanical defects in the vehicles they sell, and (2) the extent to which the stickers would convey this knowledge in an accurate and unconfusing way. 49 Fed.Reg. at 45,712-13. It found inadequate support to establish that either would be considerable.

As to the former factor, it must be borne in mind that the rule does not require dealers to inspect their cars before sale, nor does it include the "optional inspection" provision that had been suggested, which would have established a standard inspection and would have required the sticker to show, along with the known defects, whether that inspection had been conducted. See 49 Fed.Reg. at 45,718-20 (Revised Rule rejecting mandatory and optional inspection); 46 Fed.Reg. at 41,375-76 (Initial Rule rejecting mandatory and optional inspection). Without one or the other of these provisions, whose omission petitioner does not contest, there would obviously be no incentive for dealers to inform themselves of specific mechanical defects (as opposed to the general operability of the vehicle which would be apparent to a purchaser). As the Commission said:

[H]onest dealers who learn of defects must reveal their knowledge on the disclosure portion of the window sticker, whereas dealers who avoid gaining this knowledge may honestly leave the sticker blank. ... Thus, a dealer who regularly inspects and honestly discloses all "known defects" may be put at a competitive disadvantage relative to dealers who do not inspect. This factor may then have the unintended and perverse effect of discouraging, rather than encouraging, inspections and disclosure of defects.

49 Fed.Reg. at 45,713.

It was reasonable for the Commission to conclude that the degree of information dealers would have concerning specific defects under the disclosure provision would be no more, though it might well be less, than what they currently possess. And there was substantial evidence to support the related finding that it is uncertain whether, currently, dealers "ordinarily know about specific defects," 49 Fed.Reg. at 45,712, since the walk-around inspections and test drives that they customarily perform before purchasing cars convey knowledge only of the car's general condition, not of specific defects. See, e.g., Record Transcript ("Tr.") 1557-58, 1576 (even experienced mechanic may not notice certain defects); Tr. 4077 (independent dealers lack repair facilities and trained personnel to inspect certain systems such as brakes and transmission); Tr. 4107 (defects not always apparent from test drive); Record at D-10, p. 3 (same); id. at D-19, p. 3 (dealers have no way of knowing exact defect information even after inspection); id. at D-56, p. 2 (dealers generally unaware of specific defects after car renovated). There is no reason for dealers to engage in a more probing inspection because most buyers "are likely to perform no more than a similarly superficial examination." 49 Fed.Reg. at 45,713 (footnote omitted); see, e.g., BUREAU OF CONSUMER PROTECTION, FTC, SALE OF USED MOTOR VEHICLES, FINAL STAFF REPORT 84-87 (1978) ("Final Staff Report"); Tr. 3982-83 (test drive around block uninformative); Tr. 5213-14 (state official testifies that consumers often perform ritualistic inspection under hood without knowing what to look for); Tr. 5422 (consumers' test drives are generally less than a mile). Instead, dealers spend more time and money "detailing" the car, i.e., reconditioning the car's appearance, catering to customers' perceptions that a " 'good looking' car is also mechanically sound." 49 Fed.Reg. at 45,713 (footnote omitted); see, e.g., Final Staff Report at 98 & n. 94; Tr. 265-66, 1585-86, 1785, 4103, 4431-32; Record at D-7, p. 2.

In addition to being unable to conclude that dealers would generally have knowledge of specific mechanical defects, the Commission judged that the sticker system would fail to convey that knowledge accurately and might even mislead potential purchasers. See 49 Fed.Reg. at 45,713-16. It based this judgment on two studies, and on a common-sense prediction of a window sticker's effect, informed by record evidence of dealer and purchaser behavior.

The first of the studies, prepared for the FTC by the Center for Public Representation, Madison, Wisconsin, entitled An Investigation of the Retail Used Motor Vehicle Market: An Evaluation of Disclosure and Regulation (1977) ("Wisconsin Study"), had been cited by the Commission in promulgating the Initial Rule as support for the proposition that the known-defects provision would be effective. 46 Fed.Reg. at 41,342 n. 255. "[U]pon close review," however, the Commission discerned certain contradictory indications within the study that rendered it, at best, "inconclusive." 49 Fed.Reg. at 45,714. The study compared Wisconsin purchasers' awareness of the mechanical condition of used cars after enactment of a Wisconsin known-defects law with Wisconsin purchasers' awareness before that enactment, and with the awareness of purchasers in Minnesota, which had no such law.2 It showed that fewer Wisconsin buyers knew about the defects of the cars they bought than Minnesota buyers. See Final Staff Report at 121 (38.7% in Wisconsin, compared to 40.7% in Minnesota). Moreover, the Wisconsin known-defects law did not appear to alter significantly purchasers' perception that they lacked information on the car's condition, see id. at 116 (31.7% pre-law, compared to 28.5% post-law), or that their dealer had accurately apprised them of the car's condition, id. at 117 (62.6% pre-law, compared to 62.8% post-law).

The second study on which the Commission relied, prepared by the Bureau of Social Science Research, Washington, D.C., was entitled A Report on a National Survey of Private Buyers and Sellers of Used Automobiles (1982). This study, the so-called "Baseline Survey," submitted to the Commission after it promulgated the Initial Rule, was a nationwide survey of car buyers and sellers designed as a pre-Rule yardstick against which to gauge the effectiveness of the Used Car Rule. It revealed that before purchasing a used car, Wisconsin residents were no more likely to know about defects, and were no more likely to learn about defects from dealers, than were their out-of-state counterparts. Nor were they any more likely to be "very satisfied" with their cars; indeed, they were significantly more likely to be "very dissatisfied." See 49 Fed.Reg. at 45,715 (62% of Wisconsin residents "very satisfied" versus 63% elsewhere; 12% "very dissatisfied" versus 5% elsewhere).

The generalizability of field studies is a matter that "rests within the expertise of [the agency], and upon which a reviewing court must be most hesitant to intrude." State Farm, 463 U.S. at 53, 103 S.Ct. at 2872. We see no basis for disturbing the Commission's judgments here.3

Finally, the Commission was concerned that the window sticker would affirmatively mislead potential purchasers, leading them to believe that a car with no known defects was actually defect-free, or that a car with no known defects was necessarily better than a car in which several defects were so obvious as to surface upon the dealer's cursory inspection. See 49 Fed.Reg. at 45,715-16. "[T]he extensive evidence of dealer misrepresentations concerning the mechanical condition of their cars," id. at 45,716, gave it little cause to hope that dealers would correct this misimpression. This is the type of prediction of commercial behavior and consumer perception that is uniquely within the Commission's expertise and for which statistical support is neither readily available nor necessary. See FCC v. National Citizens Committee for Broadcasting, 436 U.S. 775, 813-14, 98 S. Ct. 2096, 2121-22, 56 L. Ed. 2d 697 (1978); American Home Products Corp. v. FTC, 695 F.2d 681, 686 (3d Cir.1982).

Balanced against this finding of (at best) indeterminate benefits were the costs of the proposed provision. Petitioner attempts to trivialize these by focusing only on the dealer's minimal expense of listing known defects on a window sticker. Brief for Petitioner at 23. This ignores entirely the provision's massive enforcement costs, and its likely frustration of a central objective of the Used Car Rule. 49 Fed.Reg. at 45,716-18.

To enforce the known-defects provision, the Commission would have to prove, among other things, that a defect was present at the time of purchase (not just that it manifested itself shortly after purchase), and that the dealer knew of the defect. The Commission noted how difficult such proof had been in cases alleging known defects in new cars:

In most of these cases there have been thousands of vehicles with the relevant manufacturing defect, manufacturers have had regular reporting systems to identify problems, and engineering analyses have been able to establish the cause of the problem. Even so, establishing a manufacturer's knowledge of a defect has been one of the most difficult aspects of these cases. Establishing that an individual used car dealer has knowledge of a particular defect in a series of sales of individual cars of different makes, models and ages would be a very difficult, resource-intensive undertaking.

49 Fed.Reg. at 45,717. The Commission concluded that the only way it could be done would be to field two battalions of enforcement personnel, the first of which would pose as potential shoppers, have the car inspected, and report to the dealer the defects found, and the second of which would return to the scene, ask to see the same car, and catch the dealer in a failure to disclose any of those defects. Id. at 45,716. Given the number of used-car dealers, even an effective "spot check" enforcement through a method such as this would be enormously time-consuming and expensive.

Perhaps an even greater cost, in the Commission's view, would be the provision's effect of frustrating an overarching purpose of the Used Car Rule. 49 Fed.Reg. at 45,718. The Revised Rule--requiring dealers to post a window sticker that discloses the terms of any warranty, warns purchasers not to rely on dealers' oral promises, lists a car's mechanical and safety systems and their potential major defects, and encourages consumers to obtain a "pre-purchase inspection"--is aimed at inducing consumers not to rely upon dealer-controlled information. The Commission believed that objective, and in particular the objective of encouraging third-party inspections, would be compromised by including on the sticker itself a dealer representation concerning mechanical condition, as though that were reliable, and by the dealer's ability to assure the buyer that "If we knew of any problems, we'd have to tell you about them." 49 Fed.Reg. at 45,716. This predicted effect seems to us plausible, and we have no basis for disregarding it, even if it lacks supporting statistical data, see NRDC, Inc. v. SEC, 606 F.2d 1031, 1052 (D.C.Cir.1979); American Home Products Corp. v. FTC, 695 F.2d at 686.

In sum, we find the Commission's assessment of the costs and benefits of the proposal to be a reasonable one, supported by the requisite substantial evidence; and we likewise find reasonable the conclusion that there is inadequate reason to believe that the benefits predominate. This is not to say, of course, that a reasonable person might not reach the opposite conclusion on both many of the subsidiary points and on the ultimate balancing judgment--as the Commission itself originally did, when it issued the Initial Rule. "[T]he possibility of drawing two inconsistent conclusions from the evidence does not prevent an administrative agency's finding from being supported by substantial evidence." Consolo v. FMC, 383 U.S. at 620, 86 S.Ct. at 1026 (citations omitted). Nor does the fact that the Commission at one point reached one reasonable conclusion bar it from reaching a contrary reasonable conclusion upon further reflection. See Marathon Oil Co. v. EPA, 564 F.2d at 1272.

Petitioner raises for the first time on appeal a challenge to the impartiality of Commission Chairman Miller, based upon two of his statements. The first, allegedly made at a press conference shortly after Congress vetoed the Initial Rule, contained the assertion that "[t]his rule [without the known-defects provision] I should stress is the rule that I think best." Reply Brief for Petitioner at A-1 (reprinting one page of uncited transcript). The second, allegedly made to a reporter near the end of the period for rebuttal comments,4 asserted that the Revised Rule would not contain a known-defects provision, id. at A-2 (reproducing uncited newspaper article), and explained that the provision was unnecessary because "most dealers are not mechanics, anyway," id. at A-3 (affidavit of Bruce Ramsey paraphrasing Chairman Miller). At least as to the first of these statements, we have great doubts whether petitioner's objection is timely. See United States v. L.A. Tucker Truck Lines, Inc., 344 U.S. 33, 37, 73 S. Ct. 67, 69, 97 L. Ed. 54 (1952) ("[Clourts should not topple over administrative decisions unless the administrative body has not only erred but has erred against objection made at the time appropriate under its practice."); Sanderlin v. United States, 794 F.2d 727, 734 (D.C.Cir.1986). Both objections are so insubstantial, however, that they are more readily disposed of on their merits.

Assuming that both statements were made as alleged, neither approaches the "clear and convincing evidence" that must be produced to prove that Chairman Miller had "an unalterably closed mind on matters critical to the disposition of the proceeding." Association of National Advertisers, Inc. v. FTC, 627 F.2d 1151, 1170 (D.C.Cir.1979), cert. denied, 447 U.S. 921, 100 S. Ct. 3011, 65 L. Ed. 2d 1113 (1980). The first of them indicated that he then favored deletion of the known-defects provision, but in no way demonstrated that that view could not be changed by the rulemaking proceedings that were to follow. The second of them likewise represents not an indication of unalterable prejudgment, but a prediction of Commission action and (implicitly) an announcement of Chairman Miller's own considered position. It was inappropriate to provide either before all the relevant material was in--or indeed before the Commission's formal opinion issued. But that impropriety (still assuming it occurred) gives no indication of a mind that has been closed to the evidence in the past or that would disregard any significant new material subsequently introduced. This court has declined to find invalidating prejudice (in rulemaking cases) on substantially greater evidence of predisposition. See id. at 1172-74; United Steelworkers of America, AFL-CIO-CLC v. Marshall, 647 F.2d 1189, 1209-10 (D.C.Cir.1980), cert. denied, 453 U.S. 913, 101 S. Ct. 3148, 69 L. Ed. 2d 997 (1981).

Denied.

While I agree with the majority's treatment in Parts I, II, III, and V dealing with the standard of review, the applicability of Sec. 18(e)(2), and the bias challenge, I am writing separately to express concern about the majority's wholehearted embrace in Part IV of "the Commission's assessment of the costs and benefits of the [known defects] proposal [as] a reasonable one, supported by the requisite substantial evidence." Maj. op. at 426.

The critical finding in the Commission's explanation of its decision to reverse its earlier position and to eliminate the known defects disclosure provision from the Used Car Rule is that "dealers do not ordinarily possess knowledge about specific defects" in the used cars they sell. 49 Fed.Reg. at 45,712. This finding is central to the Commission's ultimate conclusion that the potential benefits of the provision are so minimal that they do not outweigh its costs. Yet the Commission's rationale for promulgating the weaker rule it finally did adopt rests on a seemingly contradictory premise that substantial numbers of dealers are presently misrepresenting the condition of used cars to customers. See id. at 45,702 ("many used car dealers have knowingly misrepresented the mechanical condition of the cars they sell and thereby cause substantial injury to consumers"). Indeed such a premise is a necessary predicate of a rule that requires stickers to warn potential buyers not to depend on oral assurances of dealers, to check systems for themselves, and to consider obtaining independent inspections. Were dealers content to sit back and let customers judge used cars for themselves, the rule would make no sense.

In order to adequately explain its rejection of the known defects disclosure requirement, the Commission had to distinguish between the pervasiveness of dealer misrepresentations based only on general knowledge of a car's condition and those based on knowledge that specific defects exist. There is admittedly conflicting evidence in the record concerning the thoroughness with which dealers inspect used cars and the extent to which those inspections reveal specific defects. In the end, however, I cannot say that the Commission could not properly conclude on the rulemaking record that dealers "ordinarily" lack knowledge about specific defects. I do feel compelled, however, to point out that there was considerable evidence that in a great many cases dealers do know about specific defects in the cars they sell. The Commission found in 1981 that "[d]ealers acquire a great deal of information regarding the mechanical condition of vehicles they sell," 46 Fed.Reg. at 41,354, and it listed a number of sources through which dealers acquire that information:

The record shows that dealers go to great lengths to learn the mechanical condition of the cars they sell. They obtain knowledge of defects from various sources. The record reflects that, at auctions, dealers may inspect after purchase and in fact, may rescind or renegotiate if problems found are severe. After purchase by dealers, additional defects are discovered during further inspections, appearance reconditioning and repairs. In addition, dealers also become aware of defects through third-party service contract company inspectors, ... through sellers, and through selling or servicing the car previously.

Id. at 41,342. It is, of course, next to impossible to quantify, on a record devoid of any statistics or sampling about the degree of dealer knowledge, whether more dealers do or do not know about specific defects.1 On the anecdotal evidence that did exist, I must give the Commission the benefit of the doubt, but not without reservations.

The majority and the Commission attempt to bolster sparse evidence on dealer knowledge with the theory that requiring disclosure of known defects would create a disincentive for dealers to inspect cars for specific defects, so that "the degree of information dealers would have concerning specific defects under the disclosure provision would be no more, though it might well be less, than what they currently possess." Maj. op. at 423. Of course, this disincentive, if it exists, cannot be counted as a cost of the known defects disclosure proposal, since without the requirement consumers will get no benefit from any information dealers may have, anyway. But once we accept the finding that "ordinarily" dealers do not know of specific defects, it becomes hard to label as arbitrary or capricious the Commission's failure to require disclosure from dealers that do know about defects unless it appears that the benefits to consumers would be substantial enough to make worthwhile a rule for the extraordinary situation. This record simply does not provide that kind of evidence.

Nevertheless, other parts of the Commission's rationale, and particularly its discussion of the potential costs of the known defects disclosure provision, strike me as so far-fetched that I cannot acquiesce silently in the majority's acceptance of them.

The Commission concluded, without citing any record evidence, that "the defect disclosure provision included in the 1981 Rule may confuse consumers and cause them to make inaccurate assumptions about the condition of a car after reading the defect disclosure." 49 Fed.Reg. at 45,715. Although it acknowledged that the window sticker mandated by the 1981 Rule "contained a warning that the absence of a disclosed defect does not necessarily mean that the car is free from defects," the Commission stated that "there is no evidence that this warning would be effective." Id. at 45,716 n. 301. In fact, there was evidence of the effectiveness of the 1981 warning in preventing consumer confusion about the meaning of the defects disclosure statement in the record. Market research conducted by the Commission before the adoption of the 1981 Rule compared consumer reaction to two versions of the defects disclosure section of the window sticker. One of the versions included the sentence "If nothing is listed, the car is not necessarily free of defects," which was incorporated into the window sticker required by the 1981 Rule. The study found that this sentence, along with the reference to the list of defects that must be disclosed if known (which was also included in the window sticker) "left no doubt in the consumers' minds" that the listing of defects was not necessarily complete. Exploratory Research Into Consumer Attitudes Toward the Used Car Buyers Guide 17-18 (1981). Moreover, the Commission itself apparently relied on this evidence in 1981, when it noted that "the disclosure is drafted so that consumers will be aware of the fact that dealers have no obligation to inspect prior to purchase." 46 Fed.Reg. at 41,343. The Commission offered no explanation in 1984 of why it ignored this market research evidence in order to reach the opposite conclusion that consumers would be confused by the defects disclosure.

The Commission also argued that the defects disclosure requirement would reduce the effectiveness of other sections of the rule by "encouraging [consumers] to focus their attention on dealer-controlled information about a car's mechanical condition" rather than seeking independent inspections. 49 Fed.Reg. at 45,716. The Commission cited no evidence for this conclusion and, in fact, the record suggests that a purchaser who receives a defects disclosure statement may be more likely to seek an independent inspection. See Baseline Survey at 26, Table 10 (19% of Wisconsin used car purchasers had cars inspected before purchase, as opposed to 17% in other states). But even if the Commission is correct in assuming that the defects disclosure requirement will discourage consumers from seeking third-party inspections, this fact represents a significant cost only if consumers would be likely to have used cars inspected before purchase in the absence of the requirement. The Commission acknowledged that currently "few consumers actually seek independent inspections by a qualified mechanic," 49 Fed.Reg. at 45,706, and it presented no evidence that consumers would be more likely to obtain such inspections after adoption of the rule. In fact, the record shows that the principal reasons why few customers currently obtain pre-purchase inspections are that dealers generally refuse to allow independent inspections and that, even when inspections are allowed, consumers are not willing to accept the inconvenience and expense that they entail. See 49 Fed.Reg. at 45,701 n. 157 ("the record demonstrates that off-premises inspections are not allowed on a regular basis"); Final Staff Report at 7 ("pre-sale inspection by an independent mechanic is inconvenient and inefficient to the consumer and unpalatable to even the honest dealer, who must either risk the theft of his vehicle or incur the cost of an employee's time to safeguard it"); id. at 87-89 ("considerable information holds that off-premises inspections are not allowed"). None of these factors that currently make prepurchase inspections unlikely is affected by the 1984 Used Car Rule. Thus, even after adoption of the rule, most consumers will presumably continue not to obtain independent inspections. What, if anything, the new rule will do for consumers is obscure at best, and this part of the Commission's rationale for dropping the known defects proposal insubstantial.

I am well aware that the Commission's decision to eliminate the defects disclosure provision from the Used Car Rule is rooted in a fundamental shift in the agency's view of how the public interest may best be served. Consumer self-help and a self-correcting free market have replaced aggressive regulation as the order of the day. Nonetheless, the Supreme Court has reaffirmed only recently that " '[a]n agency's view of what is in the public interest may change, either with or without a change in circumstances. But an agency changing its course must supply a reasoned analysis....' " Motor Vehicle Manufacturers Association v. State Farm Mutual Automobile Insurance Co., 463 U.S. 29, 57, 103 S. Ct. 2856, 2874, 77 L. Ed. 2d 443 (1983) (quoting Greater Boston Television Corp. v. FCC, 444 F.2d 841, 852 (D.C.Cir.1970)). This one makes it, but just under the wire.