Opinion for the Court filed by Circuit Judge STARR.
Opinion concurring in part and concurring in the judgment filed by Circuit Judge SPOTTSWOOD W. ROBINSON, III.
STARR, Circuit Judge:These consolidated cases challenge an order of the Interstate Commerce Commission altering the rules governing cost recovery rate increases for railroads. Responding to certain perceived inequities under its original rules, the ICC rolled back cost recovery rail rates and conditioned protection of future cost-based rate increases on an agreement by the railroads to roll back rates when costs decline. Consolidated Rail Corporation (Conrail) contends that, in taking these steps, the Commission exceeded its statutory authority. Various shippers, in contrast, fully support the ICC’s authority to implement the new rules, but argue that the Commission’s actions fail fully to redress the problems under the previous regulations and are thus arbitrary and capricious.
As a preliminary matter, we must decide whether Conrail is properly in this case. Conrail did not file a petition for review of the Commission’s order; instead, Conrail seeks to substitute itself for the original petitioner, Association of American Railroads (Association), in No. 86-1574. Alternatively, Conrail seeks to intervene and continue to litigate that review proceeding. Conrail’s Motion to Substitute was occasioned by the decision of the Association and the American Short Line Railroad Association (ASLRA) to withdraw their joint petition for review. For the reasons to be set forth in Part II of this opinion, we deny Conrail’s Motion to Substitute or Intervene. In addition, our review of the shippers’ challenges persuades us that the Commission acted lawfully within its discretion; we therefore deny the shippers’ petitions for review.
I
Under the scheme of railroad rate regulation in effect since 1976, railroads may file rate tariffs as they see fit. 49 U.S.C. § 10701a(a) (1982). Shippers may, however, challenge as unreasonably high rates charged by railroads that have “market dominance.”- § 1071a(b), § 10707(a).1 During the late 1970s, such market-dominance challenges, coupled with the time-consuming “general rate increase” procedures relied upon by railroads, often delayed implementation of new tariffs, creating a “regulatory lag” between cost increases and recovery of those costs. See Western Coal Traffic League v. United States, 677 F.2d 915, 924-25 (D.C.Cir.), cert. denied 459 U.S. 1086, 103 S.Ct. 568, 74 L.Ed.2d 931 (1982).
In passing the Staggers Rail Act of 1980, Pub.L. No. 96-448, 94 Stat. 1895 (1980), Congress attempted to solve the problem of rate regulatory lag by establishing a mechanism that permits rail rates to be changed expeditiously without being subject to challenge, to reflect changes in rail costs. These cost recovery provisions require the Commission to develop, maintain, and publish the rail cost adjustment factor (RCAF), a quarterly index of railroad costs. § 10707a(a)(2)(B). Once determined, the quarterly RCAF is multiplied by a base rate2 to determine the adjusted base rate. § 10707a(a)(2)(A). The adjusted base rate is then used as a benchmark; railroad rates *397are conclusively presumed lawful — and thus insulated from challenge — “so long as the increased rate is not greater than the adjusted base rate.” § 10707a(b)(l). Those railroad rates that exceed the inflation adjustment may be challenged, but the Commission’s authority to suspend and investigate such rate increases is limited according to the size of the rate increase as a percent of the adjusted base rate.3
Responding to the Staggers Act, the ICC established an index of railroad costs, based on forecast data prepared by the Association, and began publishing the quarterly RCAF as required by section 10707a(a)(2)(B). Railroad Cost Recovery Procedures, 364 I.C.C. 841 (1981), aff'd, Western Coal, 677 F.2d 915. The implementing regulations permit railroads to file rate increases reflecting changes in the RCAF upon providing ten days’ notice. Ex Parte No. 290 (Sub-No. 2) Railroad Cost Recovery Procedures, decided Nov. 21, 1984, Joint Appendix (J.A.) at 172. Railroads can file increases in a special form of tariff, known as a “master tariff,”4 or set the increased rates in individual tariffs without reference to the master tariff. Rates falling within the zone of rate flexibility are protected regardless of the manner in which they are implemented.
It was but a short time before it became apparent that the Staggers Act’s cost recovery scheme (suffered from two significant shortcomings. The first stemmed from the nature of the RCAF. That mechanism is simply a forecast of future costs predicated upon data provided by the railroads. It has proved to be inaccurate. But the RCAF is also self-correcting, in that once actual data become available they are pressed into service in crafting the next quarterly RCAF. Initially, the Commission relied on this self-correction feature (along with the happy assumption that over the long haul any forecast errors would offset each other) to reject requests to establish procedures for correction of forecast errors. Brief for ICC at 4; see also Railroad Cost Recovery Procedures, 364 I.C.C. 841, 850 (1981) (discussion of frequency of adjustment of RCAF).
The second difficulty, which occasioned the challenges here, arose when the unanticipated phenomenon of declining costs entered the regulatory picture. In the inflationary times to which Congress had responded in fashioning the Staggers Act, each adjusted base rate (and, correspondingly, the levels to which rates can be increased without being subject to challenge) rises when the RCAF goes up. When costs (and thus the RCAF) go down, each adjusted base rate likewise declines. However, in the latter situation, nothing in the ICC’s original regulations required railroads to reduce their RCAF-predicated rate accordingly. Thus, on those happy occasions in the early 1980s when the. RCAF *398declined, the Commission permitted the railroads to keep their cost-recovery rates in effect, even though those rates at times exceeded the relevant adjusted base rates. Ex Parte No. 290 (Sub-No. 2) Railroad Cost Recovery Procedures, decided Nov. 21, 1984 (refusal to order rate reduction in response to decline in RCAF), J.A. at 173-74; Ex Parte No. 290 (Sub-No. 2) Railroad Cost Recovery Procedures, (not printed) decided Apr. 19, 1982 (refusal to order hold-downs to adjust for decline in RCAF), J.A. at 122, 123 n. 1. Confronted with Congress’ failure to make specific provision for declining costs and a corresponding decline in the RCAF, the Commission elected not to order the reduction of cost-recovery rates.
Things changed in 1986, however. Events surrounding the 1986 first and second quarter RCAFs persuaded the Commission that the time had arrived for remedial action. Adjustments were now deemed to be in order. The background of this shift is as follows. After reviewing the forecast data filed by the Association in December 1985, the ICC approved a RCAF of 1.069 for the first quarter, up from the previous high of 1.057. This approval authorized a cost recovery rate increase of up to 1.1 percent.5 Pursuant to a broad grant of special permission, see 49 C.F.R. § 1312.17(k), the railroads took the increase for most of their traffic in a master tariff.
After the Association filed its forecast data, oil prices began to fall dramatically, an event unforeseen by the railroads in compiling their forecast. As a result of this erroneous prediction, the first quarter RCAF was overstated.6 Oil prices continued to drop, leading to a reduction in the second quarter RCAF to 1.023, 4.3 percent below the first quarter figure. Wanting procedures for requiring rate reductions when the RCAF declined, however, the Commission’s approval of the second-quarter RCAF had no effect on the master tariff implementing the 1.1 percent first quarter increase. As a result, the authorized cost recovery rate levels were once again above the levels of the adjusted base rates. See supra at 1365.
The ICC responded to this unhappy (for shippers) state of affairs by issuing a notice of proposed rulemaking. Railroad Cost Recovery Procedures, 51 Fed.Reg. 16,363 (1986) (codified at 49 C.F.R. pt. 1135) (proposed Apr. 25, 1986) (hereinafter Cost Recovery NPR). On October 17, 1986, the Commission served the primary decision under review, which modified the cost recovery scheme in four major respects. Railroad Cost Recovery Procedures, 3 I.C.C.2d 60 (1986) (hereinafter Cost Recovery). The ICC ordered, first, that the master tariff be rolled back to the December 1985 RCAF level of 1.057. Id. at 71. Recognizing that this action did not bring authorized cost recovery rates down to the levels that were actually cost justified, the Commission reconciled the disparity by adopting its second modification: implementation of a “banking” system. Under this system, railroads are not permitted to file cost recovery master tariffs until the “bank” — the disparity, as calculated by the Commission 7 — is exhausted. Id. at 72. The third *399change ordained by the ICC involved conditioning regulatory approval (and immunity from challenge) of cost recovery rate increases achieved through master tariffs or individual tariffs on the railroads’ agreement to roll back such rates when the RCAF declines. Id. at 72-73. The ICC established the 1.057 level as a floor for such rollbacks, with any decrease in the RCAF below that level to be accounted for by banking. Id. at 73. Finally, the Commission implemented a method for correcting forecast errors to be applied prospectively only. Id. at 73-74.8
In the wake of the Commission’s decision, shippers and railroads petitioned for review of the Commission’s order. In addition, several shippers filed a petition for reconsideration asking the ICC (1) to recompute and republish the actual RCAF values for the first and second quarters of 1986, and (2) to implement a one-time adjustment of the RCAF to offset the cumulative pre-1986 forecast errors. Ex Parte No. 290 (Sub-No. 2) Petition of Southern Electric System (SES) for Reconsideration or Clarification, filed Nov. 6, 1986, J.A. at 1001. The Commission denied the petition. Ex Parte No. 290 (Sub-No. 2) Railroad Cost Recovery Procedures, (not printed) decided Dec. 12, 1986, J.A. at 1030. The shippers sought review of this denial as well.
II
We turn first to the question of Conrail’s status in this proceeding. As we mentioned at the outset, a joint petition was filed by two railroad trade associations. Conrail did not file a petition for review, and indeed had not participated in the agency proceedings. To Conrail’s chagrin, the two trade associations dropped their suit, dismissing their petition for review. The train thus having stalled on the tracks, Conrail thereupon sought to pick up where the two trade-associations and its would-be champions had left off.
Conrail styles its attempt to continue the now-abandoned suit as a Motion to Substitute under Fed.R.App.P. 43(b). But that reliance is misplaced. Rule 43(b) permits substitution “[i]f substitution of a party ... is necessary for any reason other than death.” “Necessary” means that a party to the suit is unable to continue to litigate, not (as Conrail would have it) that an original party has voluntarily chosen to stop litigating. This common-sense interpretation is bolstered both by the other clauses of Rule 43 and caselaw interpreting that Rule. Subsections (a) and (e) provide for substitution in situations where a party cannot continue an appeal due to a party’s death or removal from office. The most natural reading of subsection (b) is to permit substitution in similar situations where a party is incapable of continuing the suit, such as where a party becomes incompetent or a transfer of interest in the company or property involved in the suit has occurred. A few examples serve to illustrate the point. Rule 43(b) was successfully employed to effect the substitution of a landlord for a tenant when the suit involved the premises and rights in the property had reverted to the owner. Fred Harvey, Inc. v. Mooney, 526 F.2d 608 (7th Cir.1975); cf. Beghin-Say, Int’l v. OleBendt Rasmussen, 733 F.2d 1568 (Fed.Cir. 1984) (substitution of subsidiary for parent permitted when patents at issue had been assigned to subsidiary). Our research has yielded no instance of a court permitting a non-party to substitute in an appeal (or petition for review) from an agency action.
There exists a more fundamental problem with Conrail’s attempt to substitute itself for the Association, one which dooms its alternative request to be treated as a late-filing intervenor under Fed.R.App.P. 15(d). To permit either substitution or intervention in these circumstances would be to condone the impermissible — an evasion of clear jurisdictional requirements or*400dained by Congress for obtaining judicial review.
Direct review of ICC orders is governed by section 2344 of the Hobbs Act. 28 U.S.C. § 2344 (1982). Under that provision, a petition for review of a final ICC order must be filed within sixty days of its entry. In addition to the temporal requirement, the statute requires one seeking review to be a “party aggrieved,” a term which has uniformly been interpreted to require that petitioners be parties to any proceedings before the agency preliminary to issuance of its order. See, e.g., Simmons v. ICC, 716 F.2d 40, 42 (D.C.Cir. 1983). The requisites for intervention in a judicial review proceeding are less stringent; any party in interest to the proceeding before the Commission may intervene as of right in an action challenging an ICC order. 28 U.S.C. § 2323 (1982). Such motions, however, must be filed within 30 days of the filing of the petition for review. Fed.R.App.P. 15(d).
Conrail has failed to travel down any of these established tracks. Conrail not only chose not to file a petition for review during the sixty-day period prescribed by the Hobbs Act, it also failed to intervene within the thirty days specified in Rule 15(d). In addition, Conrail’s earlier failure to participate in the ICC proceeding casts grave doubt on its status as a “party aggrieved” within the meaning of section 2344.9
Permitting Conrail either to substitute itself for the now-departed trade association or to intervene at this late juncture not only sanctions an undisputed failure to comply with applicable statutes and rules, but it would squarely conflict with this court’s interpretive approach to the Hobbs Act. See Simmons, 716 F.2d 40. In Simmons, we rejected an intervenor’s attempt to continue a suit after the petition for review was dismissed on jurisdictional grounds. Although the intervenor in Simmons would have had standing to petition for review, its failure to file (whether by a motion to intervene or a petition for review) within the period prescribed by statute was viewed as requiring dismissal. According to the Simmons court, failure to dismiss the suit “would enable a late-filing party ... to perfect an appeal beyond the 60-day period which the statute prescribes.” Id. at 46. This the court would not condone since “ ‘[t]he sixty day period for seeking judicial review ... is jurisdictional.’ ” Id. (quoting NRDC v. NRC, 666 F.2d 595, 602 (D.C.Cir.1981)); cf. ICC v. Brotherhood of Locomotive Engineers, 482 U.S. 270, 107 S.Ct. 2360, 96 L.Ed.2d 222 (1987) (strict construction of Hobbs Act jurisdictional requirements).
Recognizing the restrictive nature of our precedent, Conrail seeks refuge in the Simmons court’s recognition of a possible exception to its holding: the Third Circuit’s decision in United States Steel v. EPA, 614 F.2d 843 (3d Cir.1979). Relying on U.S. Steel, Conrail argues that we are at liberty to ignore its various failures to comply with relevant jurisdictional requirements because an intervenor may continue a suit dropped by a party whose presence originally satisfied those requirements. The argument runs as follows: since the petition for review filed by the two trade associations established subject matter jurisdiction, and since the court “for good cause shown” may enlarge the time in which a party may move to intervene, Fed.R.App.P. 26(b), the court may treat Conrail’s motion for substitution as a motion to intervene and thereby allow it to continue the suit under the “rule” set out in U.S. Steel.
Conrail’s argument is doubly flawed: it ignores the distinguishing features of U.S. Steel and overstates the scope of that decision. For starters, the intervenor in U.S. Steel, unlike Conrail, intervened in a timely manner. In its timely motion, U.S. Steel represented that it could not rely on the original petitioner in the case as their respective interests were not identical. Second, the U.S. Steel court expressly grounded its holding on the record before it, emphasizing the intervenor’s presentation of its arguments to the agency itself and its vigorous efforts thereafter to protect its *401interest in court. Id. at 846 & n. 4; see also Horn v. Eltra Corp., 686 F.2d 439, 442 n. 2 (6th Cir.1982) (“U.S. Steel ... is obviously intended to be confined to its facts.”). In sum, informing U.S. Steel is the notion that where a party has taken steps to protect its interest, and made clear in timely fashion to both agency and court the existence of its interest, then the party should not be denied its day in court when the original petitioner withdraws. That is scarcely the situation here. Conrail’s failure to participate directly in the ICC proceeding and its later failure to seek to intervene in this court in a timely manner place it well beyond the limited bounds of U.S. Steel.
Attempting to rebut the charge that it was asleep at the switch, Conrail suggests that it was in fact a party in the ICC rulemaking and thus qualifies as a “party aggrieved” under section 2344. The point is, at best, tenuous. To be sure, “[t]he degree of- - participation necessary to achieve party status varies according to the formality with which the proceeding was conducted,” Water Transport Ass'n v. ICC, 819 F.2d 1189, 1192 (D.C.Cir.1987). But Conrail’s “participation” falls considerably far short of that required in the type of ICC proceeding at issue here. In contrast to the highly informal proceedings involved in Water Transport (where the ICC, having provisionally granted an exemption, simply requested interested parties to file protests within 15 days), here the ICC conducted an informal rulemaking in which parties were invited to (and did) make full presentations of their views. The petitioner upon which the court looked approvingly in Water Transport complied fully with the Commission’s request and filed its protest within the specified time. Here, in contrast, Conrail merely submitted affidavits to a trade association, which in turn presented them to the ICC as part of the trade association’s submission. Conrail was thus “present” only in the exhibits; it was scarcely a commenter in the proceedings, having chosen to rely on its trade association to represent its interests.
Conrail’s argument, in sum boils down to this: the court can, under Rule 43(b) or 15(d), exercise its discretion to permit Conrail to continue this suit. Although this argument may appeal to one’s sense of natural justice, it is not, upon analysis, grounded in law. In essence, Conrail would have us allow a party that has failed to comply with both the jurisdictional statute and applicable rules to litigate a suit that it likely could not have brought even had it filed a petition for review in a timely manner. Especially in light of Simmons, this we simply cannot do.
III
The shipper-petitioners’ challenges remain. The shippers argue that the ICC acted arbitrarily and capriciously in violation of the Administrative Procedure Act, 5 U.S.C. § 706(2)(A) (1982), in two respects: first, by adopting the 1.057 floor on rate rollbacks; and, second, by refusing to correct for cumulative past forecast error by either (1) recomputing and republishing an accurate RCAF for each preceding year or (2) adopting a one-time adjustment for pre1986 forecast errors. Another shipper, The Fertilizer Institute (TFI), argues in addition that the Commission erred by failing to order refunds to compensate for past discrepancies between the RCAF and actual costs. We address each of these contentions in turn.10
A
The shippers attack the 1.057 floor on two fronts.11 First, they assert that the *402ICC adopted an inferior remedy, namely banking, when a superior remedy in the form of rate rollbacks to RCAF levels below the 1.057 level was available and feasible. Banking, the shippers argue, not only fails to compensate them for the opportunity cost of money during periods of rate hold-downs, but may also lead to a mismatch between shippers injured by over-recoveries 12 and those benefitted by rate hold-downs based on credits in the bank.
The shippers’ attacks are unavailing. In the first instance, it is emphatically not the business of the judiciary to assess the relative merits of alternative agency actions and determine which is “better” or “best.” The existence of a “superior” remedy is not, standing alone, determinative of judicial review of the reasonableness of the device chosen by the agency. Second, and more fundamentally, the attributes of banking that, according to the shippers, mark it as “inferior” do not, in our view, render the ICC’s chosen remedy unreasonable. The ICC resolved the “time value of money” issue by applying an opportunity cost factor to compensate shippers for the delay in compensation which inheres in the banking approach. Cost Recovery, 3 I.C.C.2d at 70.13 As for the potential mismatch between injured and compensated shippers, the ICC observed that most shippers are repeat customers that will benefit fully from future rate hold-downs. Id. The shippers, tellingly, do not dispute the point. Instead, the shippers rely on an unsupported assertion that banking may not compensate certain shippers in particular circumstances.14 Speculation, however, will not suffice to render unreasonable a remedy that, the shippers concede, will redound to the benefit of the majority of their number.
The shippers’ second challenge focuses on the basis for the ICC’s decision. The Commission justified its decision to adopt the floor by pointing to the inability of railroads to determine which rates below the 1.057 level were effected through the cost-recovery process. Cost Recovery, 3 I.C.C.2d at 71.15 The shippers charge that *403the agency “does not identify the evidence which convinced it of the impossibility of this task.” This, the shippers say, is not surprising inasmuch as no credible evidence supports the railroads’ infeasibility claims. Brief for Shippers at 30.
The shippers, we believe, are off the mark in both respects. Although the Commission did not specifically identify the testimony that persuaded it to adopt the 1.057 floor, a central element of the railroads’ “equitable” argument against rollbacks was the infeasibility of researching the basis for individual rate levels. As a consequence of this inability, the railroads argued, the practical effect of a rollback order below the 1.057 mark would be to decrease all rates, including those already below the relevant adjusted base rates. Extensive evidence was proffered to support this argument, evidence to which the shippers refer. See Brief for Shippers at 32-34; see also Brief for AAR as Intervening Respondent at 7-8. In its decision, the ICC explicitly weighed the railroads’ arguments against the interests of shippers and determined that a rollback combined with banking represented a reasonable accommodation of both sets of competing interests. In our view, the basis for the Commission’s conclusion in this respect is sufficiently clear to pass muster.
Similarly lacking in merit is the shippers’ challenge to the credibility of the railroads’ claims. Although the ICC referred to the practical difficulties attendant to rate reductions in its Notice of Proposed Rule-making and requested comments on the issue, Cost Recovery NPR, supra p. 1365, not a single shipper in reply comments addressed the substance of any the railroads’ arguments or their testimony with respect to feasibility. Brief for ICC at 39; Brief for AAR as Intervening Respondent at 8-9. Instead, the shippers assert here for the first time that the railroads have the technical capability to track the extent to which each individual rate has incorporated RCAF increases. Brief for Shippers at 32. Alternatively, the shippers contend that the administrative difficulties bemoaned by the railroads “are almost certainly overstated.” Reply Brief for Shippers at 20. No support, however, is offered for these broadsides;16 nor do the shippers attempt to rebut the railroads’ evidence that such “capabilities” were nonexistent at the time of the Notice of Proposed Rulemaking. See Brief for Shippers at 32-33; Reply Brief for Shippers at 17, 20-22. In the absence of evidence to the contrary, we are satisfied that the ICC’s determination (that rollbacks in conjunction with banking adequately accommodate the conflicting interests of shippers and railroads) is both reasonable and supported by the record. See American Pub. Gas Ass’n v. FPC, 567 F.2d 1016, 1029-30 (D.C.Cir. 1977), cert. denied, 435 U.S. 907, 98 S.Ct. 1456, 1457, 55 L.Ed.2d 499 (1978). (reviewing court must respect agency's wide latitude for difficult policy choices in ratemaking context).
B
The shippers’ next argument — that the ICC acted arbitrarily and capriciously in refusing to adjust the RCAF for prior cumulative forecast error — runs as follows: (1) the Commission’s decision leaves uneorrected a cumulative forecast error allegedly equal to .004 RCAF points for the period from the establishment of the RCAF in 1980 through the end of 1985;17 (2) this *404error will, as matters stand, never be corrected because the new adjustment mechanism will ensure that future forecasting errors will not offset the existing cumulative forecasting error as originally anticipated, see supra p. 1364; (3) correction of this cumulative error is consistent with the purely cost-recovery nature of the RCAF mechanism; and (4) the Commission failed to explain the basis for its rejection of such an adjustment in violation of the APA, 5 U.S.C. § 553(c) (1982). Brief for Shippers at 37-41. Although not without some force, the argument, we find, is ultimately unpersuasive.
The shippers’ approach is predicated not on a perceived violation of any statutory command, but on a balancing of equities. In essence, the shippers complain that it is unfair to leave untouched prior cumulative error. It would be more equitable, the shippers argue, for the Commission to apply its rule change retroactively and reduce a future RCAF to below cost justified levels to make up for all prior forecast errors. Although this would certainly eliminate the alleged cumulative error, it would not, as the shippers contend, necessarily rectify any past “overrecovery,” as it is unclear to what extent the railroads took their cost-justified increases. See supra p. 1370B n. 16. What is clear is that such an action would penalize the railroads who relied on the cost-recovery scheme previously approved by the ICC and this court. Moreover, far from furthering the cost recovery goal of the RCAF mechanism, a one-time correction would ensure that railroads would not recover fully their increased costs for the period affected by the adjustment. The equities would thus seem, at best, mixed.
Moreover, in determining the applicable time period for the new rules, the Commission did not reject out-of-hand the concept of retroactive application. Rather, the agency, explicitly considered the fairness of such an approach. The Commission determined that the sharp drop in oil prices in late 1985, which resulted in a large discrepancy between the authorized cost recovery rate levels and the actual cost justified rates, and the consequent flurry of shipper petitions, placed the railroads on notice as of January 1, 1986, that the existing cost-recovery scheme was subject to change. Cost Recovery, 3 I.C.C.2d at 70-72 & n. 13; see also supra note 7. By virtue of this notice, the ICC judged that application of the changes after this point would not penalize the railroads that had relied on the ICC-approved scheme; application before January 1, 1986, in contrast, would, the ICC found, do exactly that. Id. This reasoning not only appears reasonable, but this i& precisely the sort of judgment the Commission, and not the court, is designated by Congress to make. Although we recognize the possible inequity inuring to the shippers, we cannot conclude that the Commission’s resolution of a highly difficult situation rises to the level of impermissibility.
We also find, on reflection, no fault with the Commission’s response to the shippers’ request for a recomputation. The ICC was required only to provide a clear and concise statement responding to relevant comments. Baltimore Gas and Elec. Co. v. United States, 817 F.2d 108, 117 (D.C. 1987). The Commission's response to the shippers' petition, which stated that the agency would not “calculate the pre-1986 cumulative forecast error ... because of [its] determination not to apply [its] new forecast error rules retroactively,” Ex Parte No. 290 (Sub-No. 2) Railroad Cost Recovery Procedures (not printed), decided Dec. 12, 1986, J.A. at 1031, evinced the Commission’s judgment that there was nothing unique to the forecast error methodology so as to justify retroactive application of the new forecast adjustment procedures. The ICC thus adhered to its determination, explained above, that application of the amended cost recovery rules prior to January 1, 1986, would unfairly penalize the railroads. In the context of the entire proceeding, the basis for the ICC’s rejection of the request was clear.
The shippers champion an alternative remedy that, as they see it, would have *405adequately corrected for the cumulative past forecast error: recomputation and republication of corrected RCAF values for each year since 1980.18 Although in requesting the ICC to take this action the shippers acknowledged that the ICC might not be “absolutely obligated” to publish such values, the shippers now attack the agency’s failure to do so as a violation of the “the requirement of 49 U.S.C. § 10707a(a)(l)(B) that the RCAF values published by the ICC be accurate.” Brief for Shippers at 42. We disagree.
The shippers’ contention is predicated on what is, at best, a somewhat strained construction of the statute. True enough, the Staggers Act contemplates frequent adjustments in the RCAF to track cost changes as currently and accurately as feasible. See 49 U.S.C. § 10707a(a)(2)(B) (“the Commission shall, as often as practicable but in no event less often than quarterly, publish a [RCAF]”). By implementing a new forecast error adjustment methodology, the ICC sought to further this undisputed aim of the Act. Nothing in the Staggers Act or its legislative history, however, specifies how the ICC is to correct the inevitable errors in a current cost index, which is of course based on forecast data. Certainly. nothing in the statute even hints at, much less articulates, a requirement that the ICC is obliged to revisit prior years, formally change old RCAFs and then republish values that have no effect on current cost-recovery rates. Absent a statutory directive to the contrary, the ICC’s choice of a correction mechanism based on adjustments in future RCAFs was, we believe, a permissible construction of the statute and a reasonable exercise of the agency’s discretion.
C
The final attack on the ICC’s order is waged in solitary fashion by The Fertilizer Institute. TFI contends that the Commission acted arbitrarily in refusing to order “refunds” to remedy past cost-recovery “overcharges.” 19 This is so, TFI argues, because both the Staggers Act and principles of equity require refunds if a railroads fails to rescind prior RCAF-based rate increases during periods of declining costs. Brief for TFI at 6-15. TFI also argues that the Commission failed adequately to address the issue of refunds, thereby violating the APA, 5 U.S.C. § 553(c).
We discern no error in the ICC’s treatment of the “refunds” issue. Notwithstanding TFI’s contentions, the statute does not in fact require refunds. Section 10707a, cited as authority by TFI, simply establishes a safe harbor for cost-based rate increases by the railroads, with no provision for declines in the RCAF. The legislative history is similarly silent as to the proper treatment of the RCAF in periods of deflation. See, e.g., Cost Recovery, 3 I.C.C.2d at 65 (“Congress never even thought about whether or how the RCAF procedures would operate in a period of deflation.”). To read into this section a mandatory remedy for a situation never *406contemplated by Congress stretches the statute beyond recognition. In addition, such a construction is incompatible with the careful statutory scheme erected by Congress, which delineates the circumstances in which the ICC may award damages. See 49 U.S.C. §§ 10701a(a), 10704(a)(1), 11705(b)(2) (the ICC may award damages only if it finds after a full hearing that a railroad rate violates a specific substantive provision of the Interstate Commerce Act).
Second, equitable principles, far from requiring the ICC to order refunds in fact tend to counsel against such an award. The railroads established their cost-recovery rates pursuant to judiciary approved Commission rules. Requiring them to pay refunds, based on a determination that the earlier Commission-approved rates were impermissible, runs counter to the well-established prohibition against retroactive ratemaking. See, e.g., Arizona Grocery Co. v. Atchison, T. & S.F. Ry., 284 U.S. 370, 390, 52 S.Ct. 183, 186, 76 L.Ed.2d 348 (1932) (where" the Commission has approved the maximum reasonable rate a carrier may charge, it may not later require a carrier that adhered to the established rules to pay reparations measured by what the Commission now determines it should have decided earlier to be a reasonable rate); Sea Robin Pipeline Co. v. FERC, 795 F.2d 182, 189 n. 7 (D.C.Cir.1986) (“FERC may not order a retroactive refund based on a post hoc determination of the illegality of a filed rate’s prescription.”). Absent a showing that the rates actually charged by the railroads ran afoul of the specific requirements of the Interstate Commerce Act, a showing no one even argues was made here, the Commission’s decision not to order refunds constituted a reasonable exercise of discretion.
Finally, the ICC adequately explained the basis for its decision not to satisfy TFI’s request for refunds. The Commission made the requisite statement in response to the issue raised, observing that damages are improper without a specific showing that particular rates are unreasonably high, and that the Commission “will not award reparations here in the absence of a showing that rates are unreasonable.” Cost Recovery, 3 I.C.C.2d at 64 n. 5. No more was required.
******
Our examination of the challenged aspects of the Commission’s order satisfies us that the agency acted reasonably and in conformity with applicable statutory provisions. Moreover, reviewing the Commission’s modifications as a whole (as this court has held is proper in complex cases such as this), we conclude that the “overall effect” of the agency’s action was reasonable. See American Public Gas Ass’n, 567 F.2d at 1029-30. Accordingly, the petitions for review are denied.
Judgment accordingly.
. "Market dominance” is defined in section 10709, 49 U.S.C. § 10709. The test for determining whether a carrier has market dominance is set forth in section 10709(d).
. The base rate for the transportation of particular commodities is established by section 10707a(a)(l). The current base is the rate that was in effect for the particular commodity on October 1, 1984. § 10707a(a)(l)(A)(iii).
. The amount of the rate increase over the adjusted base rate railroads are permitted to take without being subject to challenge varies over time. During the four years following passage of the Staggers Act, additional rate increases of six percent or less of the adjusted base rate fell within the railroads’ "zone of rate flexibility,” and were not subject to suspension or challenge. See § 10707a(e)(l); but see § 10707a(e)(2)(A) (ICC may investigate a rate increase that results in a revenue-variable cost percentage equal to or in excess of the amounts listed in section 10707a(e)(2)(A)). The statute sets the zone of rate flexibility for the following four years at four percent or less of the adjusted base rate. § 10707a(d)(1). Any rate increase falling outside the zone of rate flexibility is subject to challenge and suspension, with the burden of proof in a rate proceeding resting on the railroad.
. A "master tariff’ embodies rate increases (or other changes) potentially applicable to an unlimited number of individual movements. A master tariff is incorporated by reference in the individual rate tariffs applicable to each movement. Publication of broadly applicable tariff changes in master tariff form requires special permission from the ICC, but is nonetheless preferred by the railroads because it greatly increases the speed with which rate increases may be instituted on a large number of individual tariffs, while eliminating much of the burden and expense associated with preparing and publishing amendments to each individual rate tariff. See 49 C.F.R. § 1312.17(k); Regulations Governing the Transfer of General Increases from Master Tariffs into the Individual Tariffs of Railroads or Rail Ratemaking Organizations, 358 I.C.C. 158 (1977). Master tariffs used to implement cost recovery rate increases are referred to as "rail carrier cost recovery” (RCCR) tariffs.
. The RCAF of 1.057 means that adjusted base rates, and thus the maximum rate increases railroads could have taken before January 1, 1986, without fear of challenge, were 5.7 percent higher than the base rates: the RCAF of 1.069 means that they were 6.9 percent higher. The 1.1 percent rate increase authorization was derived by subtracting 1.057 from 1.069 and dividing the result by 1.057.
. Adjustment for forecast error would have reduced the first quarter RCAF from 1.069 to 1.062, which means that the forecasted RCAF was overstated by approximately .7 percent and that, had costs been accurately projected, the railroads would have been authorized to take a cost recovery increase of approximately .5 percent instead of the 1.1 percent authorized. See Railroad Cost Recovery Procedures, 3 I.C.C.2d 60, 72 (1986) (hereinafter Cost Recovery) (applying forecast error adjustment mechanism to 1986 RCAFs).
. The ICC calculated the size of the "bank" by adding the difference between the actual/adjusted RCAF and the maximum RCAF rate levels for each of the four quarters of 1986. Cost Recovery, 3 I.C.C.2d at 72. Although a disparity existed prior to 1986, the agency declined to go further back in calculating the bank because January 1986 was when deflation first arrived, requiring the adjustments, and the railroads had full notice that their RCAF levels might not be sustained. Id. at 72 n. 13.
. Although it declined to apply the correction procedure retroactively, the Commission did apply it to the first and second quarter 1986 RCAFs, increasing the size of the bank by the amount of the cumulative forecast error for those two periods. See Ex Parte No. 290 (Sub-No. 2) Railroad Cost Recovery Procedures, (not printed) decided Dec. 12, 1986, J.A. at 1031.
. In consequence, it is by no means clear that Conrail could have brought this action even if it had filed a petition for review within the requisite time period.
. With the withdrawal of the two railroad trade associations and denial of Conrad's Motion to Substitute, the question of the Commission's authority to order rate rollbacks is not before us. That being the case, we express no opinion as to that issue.
. The shippers challenge both the original rollback to the 1.057 floor and the establishment of 1.057 as a floor on future RCAF rollbacks. By letter dated December 23, 1987, counsel for the shippers informed the court that railroad costs in the first quarter of 1988 will be at a sufficiently high level to eliminate all remaining credits in the bank, thereby permitting railroads to take cost-recovery tariff increases. Letter from Michael F. McBride, Attorney for Baltimore Gas and Electric Co., to George Fisher, *402Clerk, U.S. Court of Appeals for the D.C. Circuit (Dec. 23, 1987). Although exhaustion of the bank appears to moot the shippers' challenge to the floor on rate reductions as it applies to 1986-87, the propriety of retaining the 1.057 floor on rate reductions for the future remains for decision.
. The shippers use the term "overrecovery” or "overcharge" to refer to the collection of rates above the level of adjusted base rates. The railroads in contrast, use these terms to refer to the collection of unreasonable rates. In addressing the shippers’ arguments, we merely adopt their proposed usage solely for convenience, without authoritatively embracing it.
. The ICC calculates the opportunity cost factor by applying the three-month Treasury Bill rate in effect seven days before the AAR’s quarterly RCAF filing to the accumulated credits in the bank. This calculation is done quarterly. Cost Recovery, 3 I.C.C.2d at 70.
. As examples of those who may not benefit from delayed rate hold-downs, the shippers point to the categories of seasonal shippers and shippers that cut back on their rail transportation use after having shipped large volumes during overcharge periods. Brief for Shippers at 25. The shippers, however, offer no concrete examples of such alleged mismatches; we are thus loathe to condemn an otherwise sensible agency approach in the absence of indications that theoretical possibilities have ripened into concrete realities of significant moment. The shippers also observe that if the credit account in the bank were large during a period of low inflation, the degree of mismatch between shippers injured by railroad overrecoveries and shippers benefitting from the rate hold-downs likely would increase dramatically. Id. But the converse seems just as likely to be the case. As the time span of the hold-down increases, the likelihood that non-repeat shippers and seasonal shippers will again transport commodities via railroad obviously increases.
. The infeasibility of tracing cost recovery rate increases stems from the statutory requirement that cost-recovery master tariffs be cancelled within two years and the rate adjustments taken by those tariffs be incorporated into individual tariffs. 49 U.S.C. § 10762(d)(2). The 1.057 RCAF-based master tariff first became effective in July 1984, more than two years before the ICC's October 1986 decision. Thus, to rescind RCAF-based rate increases below the 1.057 level, railroads would have to research numerous individual tariffs to determine whether cost recovery increases had been taken and amend each individual tariff. Rolling back rates to the 1.057 level, in contrast, required cancellation of a single master tariff, the tariff approved by the ICC in early 1986 and not yet incorporated into the individual tariffs. See Cost Recovery, 3 I.C.C.2d at 71 n. 12.
. The shippers quote one railroad witness’ testimony to the effect that the railroad had taken most RCAF-justified tariff increases in full in an apparent attempt to show that an across-the-board rate reduction below the 1.057 RCAF level would not be over-exclusive. See Brief for Shippers at 31-32; see also Reply Brief for Shippers at 19 n. 25 (quoting two additional railroad witnesses for the same proposition). The shippers neglect to mention, however, that the witnesses also testified that they immediately offset these increases by selective rate decreases effected via a myriad of techniques. See Brief for ICC at 39 n. 23; Brief for AAR as Intervening Respondent at 8 & n. *. More importantly, the fact that some railroads may have taken the permitted cost-recovery increases says little if anything about their ability to track increases in individual tariffs.
. AAR disputes the shippers' figure, arguing that the error was at most .001. Brief for AAR as Intervening Respondent at 25. Even accepting the shippers’ figures, AAR contends, the total cumulative forecast error amounts to .003. *404Id. In light of our disposition of this issue, we have no occasion to resolve this dispute.
. In its petition requesting the ICC to correct for past forecast error, SES asked the ICC to set forth revised RCAF values only for the four quarters of 1986. Ex Parte No. 290 (Sub-No. 2) Petition of SES for Reconsideration or Clarification, filed Nov. 6, 1986, J.A. at 1001, 1003-04. This was necessary, SES argued, because many private rail transportation contracts between railroads and shippers contain escalation clauses based on the RCAF and the banking methodology adopted by the Commission would not allow parties to determine the effect of the ICC’s decision on these contract rates. Id. at 1003-04 & n. 2. "Thus, even if the Commission is not absolutely obligated to set forth recomputed index values for 1986, by doing so it can greatly facilitate the amicable resolution of potentially thousands of contractual rate adjustments.” Id. at 1004.
. "Refunds," as TFI uses the term, is somewhat of a misnomer. The Interstate Commerce Act authorizes "refunds" in two contexts. First, a shipper may obtain refunds of overcharges — i.e., charges exceeding the amount specified in the applicable tariff. 49 U.S.C. §§ 11705(b)(1), 11706(b). Nobody contends that such overcharges are involved here. Second, a shipper may obtain a refund of a tariff increase under section 10707(d)(1) if the ICC finds, in an investigation begun under section 10707(a) before the proposed increase took effect, the increase to be unreasonable. 49 U.S.C. § 10707(d)(1). No such investigation of the rate increases at issue was requested or conducted. Thus, what TFI appears to be requesting are damages under section 11705(b)(2) for rates charged pursuant to unlawful tariffs. See 49 U.S.C. § 11705(b)(2).