Opinion for the court filed by Circuit Judge BUCKLEY.
BUCKLEY, Circuit Judge:Over the past fifteen years, the Federal Energy Regulatory Commission (“Commission” or “FERC”) has developed rules for adjudicating allegations of “price squeeze.” Price squeeze is a quasi-antitrust concept; it refers to an unjustified disparity between a public utility’s wholesale and retail electric power rates (“price discrimination”) that harms the ability of wholesale customers to compete with the utility in the retail market. These cases require us to review important aspects of the Commission’s price squeeze doctrine.
Petitioner Southern California Edison Company (“Edison” or “the utility”) is a public utility that sells electricity to customers in central and southern California. Petitioners the Cities of Anaheim, Riverside, Banning, Colton, and Azusa, California (collectively, “the Cities”), are municipalities that purchase electricity from Edison for resale through their own municipal electric systems. In these companion cases, Edison and the Cities seek review of two rulings in which the Commission found that Edison’s wholesale rates, as previously approved, contributed to price discrimination in 1976-1977 and 1979, and that the discrimination adversely affected the municipalities’ ability to compete with Edison for retail customers, thus creating price squeeze. The Commission ordered Edison to refund the amounts of the price discrimination by adjusting future billings. Southern Calif. Edison Co., 40 F.E.R.C. ¶ 61,371 (1987) (“Opinion 284”), reh’g granted in part, 50 F.E.R.C. H 61,275 (1990) (“Opinion 284-A ”); Southern Calif. Edison Co., 51 F.E.R.C. 1161,284 (“Opinion 347”), reh’g denied, 53 F.E.R.C. II 61,101 (1990) (“Opinion 347-A ”).
In Nos. 90-1236 and 90-1515, the Cities challenge the Commission's analysis of price discrimination. They claim the Commission acted improperly in rejecting alternative measurements of the discrimination and in accepting certain rate adjustments suggested by Edison. They also challenge the remedy ordered in the first proceeding and the Commission’s treatment of evidence submitted on rehearing in the second proceeding.
In Nos. 90-1369 and 90-1566, Edison also takes issue with the Commission’s approach to price discrimination. Edison claims there can be no price squeeze unless there is a rate disparity for the period the wholesale rate is in effect (the “locked-in” period) considered as a whole. In addition, the utility challenges a rate adjustment accepted by the Commission, the Commission’s use of “test year” data to calculate the wholesale rate of return, and the Commission’s analysis of discrimination against another wholesale customer, the City of Vernon, California (“Vernon”). Furthermore, Edison argues that the Commission improperly applied a presumption of anti-competitive effect, and that even if the presumption was properly invoked, the utility rebutted it. Lastly, Edison objects to the scope of the refund ordered by the Commission.
Because the issues substantially overlap, we address both Commission orders in a single opinion. We deny the Cities’ petitions for review in their entirety. We also deny Edison’s challenges to the price discrimination analyses and findings. But we grant Edison’s petitions for review on the critical issue of competitive harm. As to both periods of proven price discrimination, we conclude that Edison successfully rebutted the presumption of anticompetitive effects.
I. Background
A. Legal Framework
Under the Federal Power Act (“FPA”), 16 U.S.C. §§ 791a-828c (1988), FERC has jurisdiction over wholesale electric rates. Id. § 824(b)(1). Section 205(a) of the FPA requires that such rates be “just and rea*321sonable,” and section 205(b) forbids public utilities from maintaining any “unreasonable difference” in rates with respect to any sale within FERC’s jurisdiction. Id. § 824d(a), (b). Section 206 authorizes FERC to order a refund of any wholesale rate that is determined upon a hearing to be “unduly discriminatory.” Id. § 824e(a), (b).
In FPC v. Conway Corp., 426 U.S. 271, 96 S.Ct. 1999, 48 L.Ed.2d 626 (1976), the Supreme Court ruled that under section 205, FERC (formerly the Federal Power Commission) has jurisdiction to determine whether a utility’s wholesale rates are unreasonable or anticompetitive in relation to the utility’s state-regulated retail rates, even though the agency is without authority to fix retail rates. Id. at 277, 96 S.Ct. at 2003-04. Under section 206, FERC may remedy undue price discrimination traceable to the wholesale rate, id., and if a rate, though just and reasonable in its own right, will have anticompetitive effects visa-vis retail rates, the agency may adjust it downward to the lower range of the zone of reasonableness, see id. at 278-79, 96 S.Ct. at 2004-05.
As the Commission has come to describe it, “price squeeze” is
a form of price discrimination which results in an anticompetitive effect. Price discrimination occurs when a utility’s wholesale rate (in relation to costs) is higher than its retail rate (in relation to costs). Because of such price discrimination, it becomes difficult for the wholesale customer to pay the wholesale rate and resell the power at rates competitive with its supplier’s retail rates. Price discrimination becomes “price squeeze” when the disparity between wholesale and retail rates causes an anticompetitive effect, i.e., the wholesale customer loses or is likely to lose retail load, thereby “squeezing” the wholesale customer out of the retail market. Price squeeze may be caused by intentional actions of a wholesale supplier (“predatory” price squeeze) or by differences between the ratemaking policies and procedures of this Commission and state commissions (“regulatory” price squeeze).
Opinion 284, 40 F.E.R.C. at 62,151 (footnote omitted).
In the wake of Conway, FERC promulgated guidelines for adjudicating price squeeze allegations. To gain a hearing on the issue, a disgruntled wholesale customer must make a “prima facie” showing (a) that a disparity exists between relevant wholesale and retail rates and (b) that the customer competes in the same retail market as the utility. See 18 C.F.R. § 2.17(a) (1991). As will be seen, the hearing process is dominated by a progression of evidentiary presumptions. The burden throughout is on the utility to rebut the allegation of price squeeze and to justify the proposed wholesale rate(s). Id. § 2.17(e).
The proceedings are usually divided into two phases. In Phase I, the Commission resolves the traditional ratemaking issues, including cost of service, capitalization, and rate of return, to determine the “just and reasonable but for price squeeze” rate. The Commission, in its discretion, will proceed to Phase II and address price squeeze if the wholesale customer continues to allege it after the Phase I rate adjustment. See Opinion 284, 40 F.E.R.C. at 62,152; Arkansas Power & Light Co., 8 F.E.R.C. ¶ 61,131, at 61,510 (1979).
In its Phase I decision underlying Opinion 284, the Commission outlined a three-step structure for Phase II hearings. First, it determines whether there is price discrimination. Next, it determines whether the discrimination has exerted any anti-competitive effects. If the answer to both questions is yes, a price squeeze exists and a presumption arises that the wholesale rate is unduly discriminatory. Finally, unless some countervailing public policy or factual circumstance indicates that the discrimination is not undue, the Commission will fashion an appropriate remedy. Southern Calif. Edison Co., 8 F.E.R.C. ¶ 61,198, at 61,686-87 (1979) (“Opinion 62”), reh’g granted in part, 10 F.E.R.C. If 61,260 (1980) (“Opinion 62-A ”).
In Connecticut Light & Power Co., 8 F.E.R.C. 11 61,187, reh’g denied, 9 F.E.R.C. *322¶ 61,313 (1979), decided just two days before Opinion 62, the Commission announced the use of two additional evidentiary presumptions bearing specifically on the issue of anticompetitive effect. First, the existence of competition would be presumed upon a showing that the wholesale customer and the utility could potentially supply the same retail customers. Id. at 61.654. Second, drawing upon federal case law construing the antitrust statutes, the Commission determined that upon a prima facie showing of price discrimination and competition, “it is reasonable from an administrative point of view to presume that there is a reasonable probability of an anti-competitive effect.” 8 F.E.R.C. at 61,654-55. These presumptions served to focus the hearing process almost exclusively on the issue of price discrimination. See id.
Another, perhaps paramount, objective of the Connecticut Light decision was “administrative expediency”: “Because we believe that given the existence of price discrimination, it will almost invariably follow that there is at least a reasonable possibility of an anticompetitive effect, it would, in most cases, be a waste of time and resources to require intervenors to present extensive evidence on this issue.” Id. at 61.655. As set out below, the Commission in Opinion 284 declared the presumption of anticompetitive effect announced in Connecticut Light to be “no longer valid” for future cases, 40 F.E.R.C. at 62,166, but continued to apply it to Edison, id. at 62,-167.
Finally, the intent of the utility has no bearing on the price squeeze inquiry. Id. at 62,182 n. 18. See Boroughs of Ellwood City v. FERC, 731 F.2d 959, 977-78 (D.C.Cir.1984). In response to our opinion in Mid-Tex Electric Cooperative v. FERC, 773 F.2d 327 (D.C.Cir.1985), the Commission made clear that it considers itself obligated under the FPA to address not only predatory price squeeze, but also “regulatory” price squeeze — unintentional price squeeze resulting solely from differences between state and federal practices and policies. Order No. 474, 52 Fed.Reg. 23,-948, 23,956 (1987).
In deciding whether to remedy a particular instance of regulatory price squeeze, the Commission weighs the magnitude of the anticompetitive effect and balances the public’s interest in remedying it against the public interest served by the Commission practice or policy at issue. Id. The Commission believes this approach is consistent both with Conway’s emphasis on anticompetitive effects and with our statement in Boroughs of Ellwood City that “[i]t is primarily the effects of the price squeeze and its prospective remedy that should guide the Commission’s exercise of discretion,” 731 F.2d at 979 (emphasis in original). See 52 Fed.Reg. at 23,956.
B. Procedural History
1. The First Proceeding
In October 1975, Edison filed for a $16.9 million increase in its wholesale rates for R-l (small resale) and R-2 (large resale) customers. FERC accepted the rates for filing effective February 1, 1976, subject to a hearing. Earlier, in June 1974, Edison had applied to the California Public Utility Commission (“CPUC”) for an increase in its A-8 rate for “very large power” retail customers. Although Edison’s costs of service for R-2 and A-8 customers were similar, the rate filings differed; the proposed R-2 rate was somewhat higher, primarily because of differences in state and federal rules on the timing of cost pass-throughs. This rate disparity was compounded when the CPUC delayed full approval of the new retail rate until January 1977.
The Cities intervened before FERC, claiming, among other things, that the disparity between the R-2 and A-8 rates caused a regulatory price squeeze. An administrative law judge (“AU”) agreed with the Cities on “traditional” cost-of-service and rate-of-return issues but rejected the price squeeze claim. 3 F.E.R.C. II 63,033 (1978).
In Opinion 62, the Commission substantially affirmed the AU’s cost-of-service and rate-of-return conclusions. 8 F.E.R.C. at 61,677-84. As to price squeeze, it found that Edison and the Cities were competitors, and that the Cities had made a prima *323facie showing under 18 C.F.R. § 2.17(a), which Edison had failed to rebut. Id. at 61,687-94. The Commission rejected any focus on filed rates and ruled that the proper rates for a price discrimination comparison are the “effective” retail rates (the rates authorized and in effect during the relevant period) and the “just and reasonable but for price squeeze” wholesale rate. Id. at 61,689-90. The Commission invoked the Connecticut Light presumption of anti-competitive effects and remanded to the AU to determine whether a price discrimination existed using the appropriate rates. Id. at 61,693-94. On rehearing, the Commission expanded the scope of the Phase II proceedings, in part so that Edison could offer additional evidence to rebut the newly adopted presumption of anticompetitive effect. Opinion 62-A, 10 F.E.R.C. at 61,-510-11. We affirmed the “traditional” Phase I rate issues decided in Opinion 62 in Anaheim v. FERC, 669 F.2d 799 (D.C.Cir.1981).
The AU on remand found that a regulatory price squeeze had existed for the first eleven to eleven-and-a-half months that the R-2 rate was in effect, and he ordered Edison to refund approximately $3.8 million to its wholesale customers, principally the Cities. See Phase II Initial Decision, 34 F.E.R.C. 11 63,086, at 65,303-04 (1986). For the remaining two-and-a-half years of the locked-in period (which ended when Edison’s next R-2 rate filing became effective on August 16, 1979), the AU found that “the R-2/A-8 rate relationship was highly favorable to the R-2 customers.” Id. at 65,294. In the Phase II hearing, the Cities introduced a new analysis that quantified price discrimination between the R-2 rate and a composite of all of Edison’s retail rates. The AU, however, excluded the evidence as outside the scope of the remand. See 23 F.E.R.C. 11 63,109 (1983), reconsid. denied, 34 F.E.R.C. at 65,279-80.
On review, the Commission agreed that the R-2 rate as modified in Phase I was unduly discriminatory from February 1, 1976, until January 12, 1977, when the retail rate increased and the disparity was reversed. See Opinion 284, 40 F.E.R.C. at 62,160-63. In confirming the existence of price discrimination, the Commission affirmed the use of an analysis comparing Edison’s rate of return (the margin of revenues above costs) on the “just and reasonable but for price squeeze” R-2 rate with its rate of return on the “effective” A-8 rate. The Commission summarily affirmed the exclusion of the Cities’ broader, composite rate-of-return analysis. To derive the wholesale rate of return, the Commission used the estimated costs for “test year” 1976 that Edison had filed with its original rate application rather than the actual R-2 costs of service experienced by Edison. Id. at 62,152-55, 62,182 n. 22.
The Commission summarily affirmed the AU’s treatment of two adjustments Edison included in its rate-of-return analysis. Id. at 62,157-60. First, Edison wanted to increase the retail rate of return to compensate for tax accounting differences: FERC permitted Edison to pass the cost of tax obligations on to its R-2 customers before the taxes were actually paid (tax “normalization”), while the CPUC allowed only flow-through tax accounting. Because FERC required consistent treatment of wholesale and retail costs in the comparative rate-of-return analysis, the retail rate of return was distorted, according to Edison. The AU rejected the tax normalization adjustment. See 34 F.E.R.C. at 65,-274-75.
Second, the AU accepted Edison’s proposed synchronization of fuel costs and revenues. The A-8 rate tariff approved by the CPUC included an “energy cost adjustment clause” (“ECAC”) and a related ECAC “balancing account.” The ECAC allowed Edison to recover estimated rises in fuel and purchased power costs without filing for a rate increase. The R-2 rate included a comparable “fuel cost adjustment” (“FCA”). The ECAC balancing account reflected amounts that Edison under- or overcollected in its ECAC recoveries due to errors in cost estimation. These amounts were amortized and included as charges or credits in subsequent retail billings. The AU agreed with Edison that in deriving the rate of return, the retail revenues should be adjusted to match the corre*324sponding fuel costs that were subject to recovery under the ECAC or that were held in the balancing account, and he approved a similar adjustment in wholesale revenues to account for the FCA. He rejected analyses submitted by the Cities and FERC staff that failed to include the ECAC adjustment. See id. at 65,271-74.
On the question of anticompetitive effects, the Commission decided to retract the Connecticut Light presumption, but only prospectively. 40 F.E.R.C. at 62,166-67. It concluded that the presumption was “no longer valid” because “[pjrice differences of limited magnitude and duration may not result in an injury to competition, and the Supreme Court’s decision in Conway makes clear that the anticompetitive effect of utility rates should be our concern.” Id. at 62,166. Although such a presumption might have made sense in the context of predatory price squeeze, the Commission reasoned, it was inconsistent with the regulatory price squeeze policy announced in Order 474, which contemplates a critical evaluation of effects evidence. See id.
Henceforth, those alleging price squeeze would have the burden in Phase II of demonstrating, via wholesale and retail market analyses, that the utility possesses monopoly power and that the alleged price discrimination will likely exacerbate market concentration at the expense of the utility’s competitors. Id. at 62,167-68. The Commission stated its intention that such proceedings would focus on objective criteria and would be guided by antitrust principles. Id. at 62,168. Although the Commission recognized that it could have applied the new rule to the present case, it decided not to do so because “[ejvidence has been submitted on this issue in reliance on our prior policy, and it would be inappropriate in this proceeding to depart from the standards enunciated in Opinion Nos. 62 and 62-A.” Id. at 62,167.
The Commission then found that Edison failed to rebut the Connecticut Light presumption. Id. at 62,170-72. The Commission ruled that it was not enough that the Cities had lost no actual retail load; Edison also had to show that there was no adverse effect on “potential” competition — that is, no “reasonable probability” of competitive harm. See id. at 62,165, 62,168-69. Edison introduced expert testimony indicating that the decisions of large power customers are made on the basis of long-term cost relationships and would not have been affected by a rate disparity of the amount and duration at issue. But the Commission rejected this testimony as “conjecture”; it relied instead on the AU’s reasoning that “witnesses have testified persuasively in other proceedings that the only valid basis for analyzing price squeeze is to evaluate what was confronting the wholesale customer from day-to-day.” Id. at 62,172.
Edison argued that any potential anti-competitive effect resulting from the eleven-and-a-half months of price discrimination was offset by the subsequent two-and-a-half years of favorable wholesale rates. The Commission disagreed, reasoning that a “mere price preference” in a later period will not erase the earlier harm to competition because “the antitrust laws were enacted for the protection of competition— not competitors.” Id. at 62,169-70 (emphasis in original). Although it did riot reject the notion that a subsequent reversal in a rate disparity might “dissipate[ ]” an anti-competitive effect, the Commission ruled that a mere showing of “reverse price discrimination” is not sufficient; the utility must show that the reverse rate disparity “is or is likely to be of sufficient magnitude and duration to have a countervailing effect on competition so that, overall, during the entire course of the wholesale/retail rate relationship, competition in the relevant market has not been harmed.” Id. at 62,170.
The Commission quantified the price discrimination suffered by the Cities as approximately $2.6 million. See id. at 62,164-65. It found the discrimination “undue” and ordered a refund. It required Edison to reduce its R-2 rate of return for the relevant period and, if necessary, eliminate the effects of tax normalization. See id. at 62,178-79. The refund was ordered for all members of the R-2 class, including those *325customers, like Vernon, that had not intervened. Id. at 62,180.
2. The Second Proceeding
In January 1979, Edison filed for another increase in the R-2 rate. FERC accepted the new rate for filing effective August 16, 1979, and scheduled a hearing in which the Cities and Vernon intervened. The challenged rate was locked in for a period of twenty-three months, until a superseding rate became effective July 16, 1981. See Phase II Initial Decision, 37 F.E.R.C. 11 63,-005, at 65,010-11 (1986).
In Phase II, the parties introduced rate-of-return analyses that compared Edison’s “just and reasonable but for price squeeze” wholesale rate of return with a composite rate of return derived from a time-weighted average of the various retail rates effective during the locked-in period. Edison and FERC staff identified two different effective retail rate levels, and they maintained that no price discrimination occurred with respect to either of the two subperiods or with respect to the locked-in period as a whole. See id. at 65,013-15. The Cities and Vernon identified seven subperiods of differing retail rates; they claimed price squeeze only during the first two of these subperiods, which constituted the initial five-and-a-half months of the locked-in period. See id. at 65,021, app. A; Opinion 347, 51 F.E.R.C. at 61,878.
In its rate-of-return analysis, Edison once again proposed a tax normalization adjustment to the retail rate of return and an ECAC adjustment to retail revenues. The ALJ relied on the initial decision in the earlier proceeding to reject tax normalization and accept the ECAC adjustment, which in this instance also included similar adjustments for interim tax changes and conservation expenses. 37 F.E.R.C. at 65,-017-19; see 51 F.E.R.C. at 61,879 & n. 14.
The AU refused the Cities’ proposal to decrease retail revenues to account for a “fuel collection balance adjustment” (“FCBA”) credit. The CPUC had required Edison to pay the FCBA credit to its retail customers during the locked-in period to make up for overcollections in 1972-1976 resulting from unusually wet weather, which had caused greater-than-expected production at Edison’s hydroelectric plants. See 37 F.E.R.C. at 65,017.
The AU ultimately rejected the Cities’ and Vernon’s Phase II claims. He concluded that “[cjommon sense and logic ... compel a finding that the entire locked-in period must be considered for price squeeze and price discrimination purposes.” Id. at 65,019. He accepted Edison’s argument that because Vernon’s retail load was ninety percent industrial, the only justifiable comparison for Vernon was between the R-2 rate of return and Edison’s “large power/very large power” (“LP/VLP”) retail rate of return; this comparison revealed no price discrimination. Id.
On review, the Commission found that the Cities had established price discrimination for the first five-and-a-half months of the wholesale rate. The important consideration for the Commission was whether the period of price discrimination could have created potential anticompetitive effects, not whether it comprised the entire locked-in period. The Commission stated that “it is conceivable that a severe price discrimination lasting for fk months could have an anticompetitive effect, while a minimal price discrimination lasting over an entire 23-month locked-in period might not.” Opinion 347, 51 F.E.R.C. at 61,884. It found the amount of the discrimination to be approximately $1 million, which was sufficient to support a finding of price squeeze. Id. at 61,884-86 (relying on City of Groton v. Connecticut Light & Power Co., 662 F.2d 921, 934 (2d Cir.1981)).
The Commission summarily affirmed the AU’s use of test year data and his treatment of the ECAC and tax normalization adjustments. 51 F.E.R.C. at 61,880. But it overruled the AU’s decision on the FCBA credit. The Commission included the credit in the effective retail rates because it was substantial and of adequate duration to affect the decisions of retail customers. The Commission also reasoned that it would be unfair to recognize the associated overcollections as revenue in the prior period (which had benefited Edison in an earli*326er price squeeze case brought by the Cities) but not to include the later refunds as part of retail costs. Id. at 61,882-83 (discussing Southern Calif. Edison Co., 16 F.E.R.C. 11 61,185, at 61,420 (1981) (“Opinion 128 ”)).
Following its decision in Opinion 284, the Commission then invoked the presumption of anticompetitive effects. It found that Edison had failed to rebut the presumption of “potential” effects because Edison’s evidence did not show how the rate disparity affected retail customers “from day to day.” 51 F.E.R.C. at 61,890. That the Cities had absorbed the price discrimination while still offering a competitive rate was irrelevant because “the diminution of a municipal customer’s competitive advantage caused by the discriminatory rate itself is an anticompetitive effect.” Id. The Commission again rejected Edison’s offset argument. Id.
After determining that no significant policy considerations or factual circumstances mitigated the discrimination, the Commission ordered Edison to submit a compliance filing to establish precisely the amount of the price squeeze for the five-and-a-half month period. Contrary to the AU, the Commission found that Vernon was entitled to a refund on the same basis as the Cities, as were all of Edison’s wholesale customers. Id. at 61,891-93.
On rehearing, the Cities submitted an affidavit from their expert purporting to show that if the Commission adopted the ECAC and other adjustments approved in Opinion 347, there would be three subperiods of price discrimination: the first two- and-a-half months of the locked-in period (but not the remainder of the five-and-a-half months previously alleged) and two later subperiods. The affidavit calculated the amount of discrimination that purportedly occurred during each subperiod. On the basis of this submission, the Cities urged the Commission to amend its findings and increase the amount of the refund to approximately $2.6 million.
The Commission refused to accept the affidavit as proof of price discrimination during additional subperiods. It held that the Cities had had ample opportunity to raise the new claims at an earlier stage and that it was now too late to expand the scope of the proceeding. The Commission did, however, accept the affidavit as a concession by the Cities that price squeeze had occurred only during the first two-and-a-half months and only in the amount of $518,000. It stated that even though the period and amount of this discrimination were small, the discrimination could still have harmed competition, and Edison had not rebutted the presumption of anticompetitive effects created by the shortened period. See Opinion 347-A, 53 F.E.R.C. at 61,318-20.
II. Discussion
A. The Cities’ Petitions
1. Exclusion of Evidence
The Cities first complain of the Commission's decision to strike their composite rate-of-return study in the earlier proceeding. They claim this decision denied them due process. We disagree.
To obtain a Phase II hearing, the Cities had the initial burden of establishing the existence of competition and a disparity between the R-2 rate and relevant retail rate(s). In Opinion 62, the Commission found that the Cities competed with Edison in three ways. They competed directly for industrial customers located in, or thinking of locating in, their service areas; they faced “yardstick” competition, which refers to pressure brought by retail customers on municipal power systems to keep their rates competitive with Edison’s; and they competed for exclusive service areas, called “franchises,” granted by the CPUC. See 8 F.E.R.C. at 61,684-85, 61,687, 61,693. We accept the Cities’ contention that the yardstick and franchise concepts potentially involve all classes of retail customers, not just large industrial firms eligible for the A-8 rate.
The Commission, however, further found that the Cities had established only an R-2/A-8 rate disparity. See id. at 61,687. The Cities “clearly stated ... that the retail rate that is the focus of our inquiry herein is the A-8 industrial rate.” Id. *327And “[virtually all of the cities’ testimony ... centered on their allegation that they are comparable to those customers receiving the A-8 rate and that this is the rate the cities should be receiving.” Id. The alleged R-2/A-8 disparity was therefore the sole focus of Edison’s rebuttal evidence. In remanding to the AU, the Commission instructed that the Phase II hearing would “not provide an opportunity to relitigate the cities’ prima facie showing [and] Edison’s rebuttal.” Id. at 61,694. Thus, the AU could reasonably conclude that evidence purporting to show price discrimination for all retail rates was inadmissible as beyond the scope of the proceedings. See 23 F.E.R.C. at 65,282-83.
The Cities cry foul because their prima facie case was based on a “bill comparison” study, a flat comparison of billed rates that is meaningful only if the two rate classes have comparable costs of service. They assert that when the Commission subsequently changed its view and preferred the more flexible rate-of-return approach, which in theory can identify price discrimination with respect to any retail class, due process required an opportunity to present such evidence. Cf. Hatch v. FERC, 654 F.2d 825, 835 (D.C.Cir.1981).
The Cities received that opportunity. On rehearing of Opinion 62, the Commission “recognize[d] that price squeeze proceedings must still be regarded as a developing area of the law,” and that as important new rulings are made, “[e]quity may require that the parties be given a further opportunity to make their case.” Opinion 62-A, 10 F.E.R.C. at 61,510. Accordingly, in Phase II both Edison and the Cities were ultimately allowed to introduce competing rate-of-return analyses comparing the R-2 and the A-8 rates.
The Commission was not obligated to entertain an entirely new prima facie case, however, because the Cities were aware at the outset of their burden to establish the rate disparity that would become the focus of the price squeeze inquiry. Cf. Cities of Batavia v. FERC, 672 F.2d 64, 87 (D.C.Cir. 1982) (inconsistent with due process to find in favor of one party simply because “an adversary has failed to make a showing it was not aware would be required of it”) (quoting Commonwealth Edison Co., 9 F.E.R.C. ¶ 61,184, 61,345 (1979) (emphasis added)).
The Cities cannot maintain that the Commission rigidly refused to consider any methodology but bill comparison. The Cities themselves had also introduced a transfer price analysis, comparing the average costs of transmission and generation for all wholesale and retail customers, which the Commission termed “an acceptable approach to employ in' establishing a price squeeze claim.” 8 F.E.R.C. at 61,694. The Commission rejected this particular transfer price analysis because it was flawed and because the Cities’ own choice to zero in on the A-8 class rendered the study “not particularly helpful.” See id. at 61,695, 61,698 n. 97.
Thus, the Commission was open to new methodologies. (In fact, the Commission embraced the rate-of-return test no later than a month after Opinion 62 and well before Opinion 62-A. See Commonwealth Edison Co., 8 F.E.R.C. II 61,277, reh’g denied, 9 F.E.R.C. 1161,184 (1979), aff'd in relevant part, Cities of Batavia, 672 F.2d at 90.) Any loss of opportunity to expand the case to cover all retail rates was caused not by a lapse of adequate process, but by the Cities’ failure to anticipate or initiate a new way to demonstrate price discrimination.
We also reject the Cities’ contention that the AU’s exclusion of the study was arbitrary, capricious, and an abuse of discretion. We have previously held that the Commission may designate a particular target of competition, such as industrial customers, and that it may limit its inquiry to a rate-of-return comparison between the wholesale rate and the large industrial rate. See Boroughs of Ellwood City, 731 F.2d at 973-74. Contrary to the Cities’ assertion, such a limited comparison is not “nonsensical” or “illogical[ ]” simply because it fails to quantify all of the price discrimination that may affect competition between the utility and the wholesale customer. Brief for Petitioning Cities, No. 90-*3281236, at 18, 30. The Cities rely on Commonwealth Edison, in which the Commission stated that the relevant question for the rate-of-return test “is which retail customers comprise the group for which there is competition ..., not which group of retail customers is of like size to the wholesale customers.” 8 F.E.R.C. at 61,850. But there the Commission was simply ruling that a utility may not disprove price discrimination by using a rate-of-return analysis that focuses on an irrelevant subset of the target group. See id. at 61,849-50. As the AU pointed out in granting Edison’s motion to strike, the Commission in Commonwealth, as here, accepted a relevant target group that comprised only industrial customers despite evidence of broader competition. See 23 F.E.R.C. at 65,283.
2. ECAC Adjustments
Next, the Cities challenge in both cases the Commission’s acceptance of Edison’s ECAC adjustments. They argue that because the fuel costs affected by the ECAC and ECAC balancing accounts were not actually billed to retail customers during the months at issue, they should not be considered part of the retail rates in the comparative rate-of-return analysis. The Cities maintain that these adjustments are impossible to reconcile with the Commission’s treatment of tax normalization and the FCBA credit. We uphold the adjustments as within the Commission’s discretion and not arbitrary or capricious.
In refining the comparative rate-of-return analysis, the Commission has decided to focus on the “effective” retail rate against which the wholesale customers are actually competing — that is, the rate authorized by the state and in effect during the relevant period. See Opinion 62, 8 F.E.R.C. at 61,690. The Commission could reasonably conclude that the fuel cost increases subject to recovery under the ECAC were part of the effective rate from the point of view of the retail customers. The AU found that the ECAC and ECAC balancing' accounts were integral parts of the rate tariffs authorized by the CPUC; therefore, rational purchasers, especially the A-8 customers, would presumably have known that their rates were highly sensitive to fluctuations in Edison’s costs. See 34 F.E.R.C. at 65,272. We refuse to impose on the Commission a rigid definition of “effective” that would include only the amounts of electricity bills actually mailed to retail customers in a particular period.
Furthermore, the Commission acted within its discretion in treating the retail and wholesale rates alike with respect to the fuel cost/revenue relationship. Both rates included similar adjustment clauses that allowed the utility to recoup fuel and purchased power cost increases in the short term. In determining the just and reasonable wholesale rate of return in Phase I, the Commission included the FCA amounts as revenue; in Phase II, the parties similarly “synchronized” R-2 fuel costs and revenues in their various rate-of-return studies. See id. at 65,271. Because both the FCA and the ECAC operated to make Edison whole for actual costs incurred and both were designed to insulate Edison’s profit margins from uncontrollable changes in fuel costs, the AU concluded that the methods for deriving the wholesale and retail rates of return should be consistent, even though the lag in recovery of fuel expenses on the wholesale side was shorter. Id. at 65,271-72.
We recognize that the price squeeze doctrine is meant to remedy discrimination caused by “the imperfections of regulation,” Cities of Batavia, 672 F.2d at 91, and “the paradox” of dual ratesetting, Boroughs of Ellwood City, 731 F.2d at 974. We also recognize that matching fuel costs with revenues eliminated any short-term wholesale/retail disparity caused by the different cost recovery schedules for the FCA and ECAC. But the rate-of-return analysis also suffers from imperfections and discrepancies, and it is largely within the Commission’s discretion to refine that analysis from case to case. Thus, the Commission could conclude that erasing the disparity by disregarding both the FCA and the ECAC lags was preferable to aggravating it by eliminating the FCA lag alone, as *329the Cities proposed in their study. See 53 F.E.R.C. at 61,318.
The ECAC adjustment is also consistent with the Commission’s handling of the FCBA credit. In Opinion 128, the Commission found that the overcollections associated with the later FCBA credit should be included in calculating the retail revenues in the earlier period. The overcollections were part of the effective rates because the charges were not “interim” or “subject to refund” when collected; the CPUC had authorized them as “final” and “lawful.” 16 F.E.R.C. at 61,422. Because there was no ECAC balancing account in the authorized rates, the excess billings could only be credited prospectively, following an investigation by the CPUC. See id. at 61,420, 61,423; Opinion 347, 51 F.E.R.C. at 61,883. Thus, taking account of the credits later ordered by the CPUC would produce retail rates against which the Cities had never actually competed. 16 F.E.R.C. at 61,422. In contrast, the ECAC mechanism was part of the final rate actually in effect at the relevant times; retail charges were continually subject to automatic fine tuning without separate action by the CPUC.
Nor is the ECAC decision at odds with rejection of the tax normalization adjustment. Under the required flow-through accounting, the effective retail rate(s) did not include charges for tax costs until Edison actually paid them. Edison elected to take advantage of the tax normalization option permitted by FERC to raise R-2 revenues. This decision caused a rate disparity that could have affected the purchasing decisions of retail customers during the locked-in period because the tax costs might not have passed to the retail customers for another ten to twenty years, if at all. See 34 F.E.R.C. at 65,274-75. Moreover, we have said that price discrimination generally exists where a rate disparity results from the utility’s manipulation of its administrative options, as opposed to a difference in cost of service. See Boroughs of Ellwood City, 731 F.2d at 977-79. It was thus appropriate for the Commission to consider the tax normalization issue in step three of the price squeeze analysis, in determining whether the discrimination was “undue,” rather than in step one. See 40 F.E.R.C. at 62,174-75.
3. Corrigan Affidavit
We uphold the Commission’s handling of the new evidence submitted by the Cities on rehearing of Opinion 347. In response to the Commission’s decision affirming Edison’s ECAC adjustment, the Cities included with their petition for rehearing an affidavit from an expert witness, Timothy R. Corrigan, that reformulated the Cities’ rate-of-return analysis to include the adjustment. The affidavit indicated the existence of price discrimination in subperiods not previously alleged by the Cities. See 53 F.E.R.C. at 61,318-19. The Commission rejected the new study as proof of additional discrimination; it reasoned that
acceptance of Edison’s adjustment to the [ECAC] balancing account does not permit Cities to expand the scope of this already lengthy proceeding at this late date. Cities have long been aware of this adjustment to the balancing account since Edison proposed this adjustment not only in this proceeding but also in the proceeding which is the subject of Opinion No. 284. If Cities were of the opinion that the Commission’s acceptance of Edison’s adjustment would result in price discrimination during additional subperiods, they could have raised that contention long before this stage of the proceeding.
Id. at 61,319.
Nevertheless, the Commission found the submission relevant to the issue of remedy and denied Edison’s motion to strike. Because the Corrigan affidavit indicated, contrary to the Cities’ earlier contentions, that no price squeeze had occurred in the second subperiod, the Commission accepted the affidavit as a concession on this point. Id.
The Cities claim that the Commission’s refusal to recognize additional periods of price squeeze and its selective use of the evidence violated their due process rights and was arbitrary, capricious, and an abuse of discretion. We disagree, largely for the *330reasons we discussed above in rejecting their challenge to the exclusion of the composite rate-of-return study in the companion case. The Cities knew early on of Edison’s ECAC arguments and easily could have submitted the modified analysis of price discrimination at the outset of Phase II. There was no fundamental unfairness and nothing arbitrary in refusing to recognize such expansive new evidence long after the close of the hearing.
The Cities further argue that the Commission should have treated their submission as a motion to reopen the record. We have held that “[t]he Commission should address even improperly styled motions to supplement the record under relevant criteria.” Cooley v. FERC, 843 F.2d 1464, 1473 (D.C.Cir.), cert. denied, 488 U.S. 933, 109 S.Ct. 327, 102 L.Ed.2d 344 (1988). But in Cooley we also stated that “at some point an agency must be able to say, ‘Enough is enough,’ ” and thus “[w]e normally reverse an agency’s decision not to reopen the record only for an abuse of discretion.” Id. (internal quotes omitted). Under its own rules of procedure, the Commission has discretion to reopen a hearing “for good cause.” 18 C.F.R. § 385.716(a). The Commission acted within this broad discretion here because it addressed criteria relevant to reopening the record — the timing of the request and the opportunity afforded the Cities to explore the issue at any point before or during the Phase II hearing.
We also uphold the Commission’s deeming the affidavit a concession that no price squeeze had occurred in the second subperiod. See 53 F.E.R.C. at 61,319. The submission was a representation by the complaining parties that they no longer contested a portion of their original case. In the absence of any controversy over the second subperiod, the agency would have wasted time and resources by requiring Edison to prepare a compliance filing for the entire five-and-a-half months.
B. Edison’s Petitions
1. Price Discrimination
For its part, Edison attacks the Commission’s price discrimination findings on several grounds. None merits extended discussion.
First, Edison argues that as a matter of law, the entire locked-in period must be the basis for assessing price discrimination: If, as in both cases here, the Commission does not find price discrimination over the locked-in period considered as a whole, it cannot make a price squeeze finding. In support of this claim, Edison points to a single sentence in Conway in which the Court stated that the decision to remedy an alleged price squeeze would involve “an examination of the entire “factual context in which the proposed wholesale rate will function.’ ” 426 U.S. at 280, 96 S.Ct. at 2005 (quoting Conway Corp. v. FPC, 510 F.2d 1264, 1273 (D.C.Cir.1975)).
By “entire factual context,” however, the Court was signifying its approval of our conclusion that “the Federal Power Act does not contemplate that [FERC] is to operate in an airtight chamber, insulated from nonjurisdictional factors,” 510 F.2d at 1272. FERC may consider the effect on competition of the relationship between the wholesale rate and the nonjurisdictional retail rate. It is Edison that advocates an “airtight chamber” by insisting that “the entire locked-in period must be the framework for determining whether price discrimination is undue.” 40 F.E.R.C. at 62,-161. As the Commission stated: “This argument is based on an assumption that price discrimination is solely the product of the wholesale rate (which determines the locked-in period) without regard to the individual competing retail rates in effect over this period____” Id.
The Commission correctly reads Conway to focus the inquiry on the anticompetitive effects of the utility’s rates. See id. We agree with the Commission that “[t]he anti-competitive effect resulting from a price squeeze is a function of the magnitude and the duration of the price discrimination.” Id. Thus, price discrimination that exists over an entire locked-in period might not constitute price squeeze, while severe discrimination in a subperiod might. See id.; *33151 F.E.R.C. at 61,884. In making that determination, the Commission has authority to focus on subperiods that are defined by changes in the effective retail rate(s), as it did here. Edison’s approach, on the other hand, would “impos[e] an arbitrary time period,” 40 F.E.R.C. at 62,163, and could potentially thwart the price squeeze inquiry where the wholesale rate is “open-ended,” id. at 62,161.
Second, Edison takes issue with the Commission’s decision in Opinion 347 to include the FCBA credit in computing retail revenues. See 51 F.E.R.C. at 61,882-83. The CPTJC ordered the credit to reimburse retail customers for the overcollections of 1972-1976 at issue in Opinion 128. Edison paid the credit through a special tariff from 1979-1981. It totaled $133.2 million plus interest, of which $118.3 million was credited to the bills of retail customers during the locked-in period. See id. at 61,-882-83; Opinion 347-A, 53 F.E.R.C. at 61,316-17 & n. 38.
The Commission reasonably construed the credit as part of the effective rates against which the wholesale customers competed. The credit was included in the bills paid during the period of discrimination, and the amount and duration of the credit were substantial enough to affect “the consumption decisions of Edison’s retail customers.” 51 F.E.R.C. at 61,882; see 53 F.E.R.C. at 61,316-17. Thus, the Commission could reject the AU’s conclusion that retail customers would have known the credit was “only temporary.” 51 F.E.R.C. at 61,882 n. 28. Moreover, the Commission could reasonably conclude that Edison could not benefit from excluding the credit when the associated overcollections, at Edison’s urging, had been included in retail revenues in Opinion 128. See id. at 61,882.
Third, Edison claims the Commission erred in Opinion 347 in measuring price discrimination against Vernon on the basis of the Cities’ composite rate-of-return study, which Vernon had used to support its price squeeze claim. Because ninety percent of Vernon’s retail load was industrial, Edison argues, the Commission, like the AU, should have used Edison s comparison between the R-2 and LP/VLP rates of return, which indicated no discrimination. Instead, the Commission concluded that the LP/VLP study “ignore[d] the potential for competition for nonindustrial customers,” and that Edison had failed to rebut evidence that the study was flawed. See 51 F.E.R.C. at 61,893.
We will uphold the Commission’s ruling. We do so, however, without deciding whether a wholesale customer that makes a prima facie showing of competition only with respect to large industrial purchasers may nevertheless rely upon an analysis of price discrimination that includes a composite of all retail classes. Here, Edison conceded the existence of competition with Vernon, and Vernon was therefore allowed to join the Cities’ price squeeze case without a separate prima facie showing of competition. Furthermore, there is evidence in the record, in the form of expert testimony presented by Vernon, sufficient to support a finding of potential competition for nonindustrial service that could be affected by a broader rate disparity. See Joint Appendix (“J.A.”), No. 90-1515, at 69-73, 77-84, 252A-58A (testimony of Baker G. Clay).
Finally, we affirm the Commission’s use of test year data in calculating the wholesale rate of return. See 40 F.E.R.C. at 62,154-55; 51 F.E.R.C. at 61,-880 & n. 17. This ruling fits with the Commission’s court-approved practice of relying on test year estimates in determining just and reasonable rates, even when the actual data for the projected test year become available. See Cities of Batavia, 672 F.2d at 74; see also Boroughs of Ellwood City, 731 F.2d at 965-66. The Commission relies on test year data unless they are shown to be substantially in error or to yield unreasonable results; Edison made no such showing here. See 40 F.E.R.C. at 62,155. Given the Commission’s considerable discretion in constructing an appropriate wholesale rate for price discrimination purposes, just as in defining the “effective” retail rate, we find its reasoning on this issue neither arbitrary nor capricious.
*3322. Anticompetitive Effects
Edison’s challenge to the findings of anti-competitive effect is more substantial. The Commission stated in Opinion 284 that “anticompetitive effect is the alpha and omega of the price squeeze inquiry.” 40 F.E.R.C. at 62,165. We agree. As we have previously pointed out, “discriminatory effect ... is the very raison d’etre of the price squeeze doctrine” and must be the Commission’s “paramount consideration ].” Boroughs of Ellwood City, 731 F.2d at 978. Regrettably, in these cases the Commission failed to make a sufficient finding of anticompetitive effects.
Edison does not attack head-on the legality of the presumption of anticompetitive effects. Rather, Edison claims that the Commission, once it had decided that the presumption is invalid and that complaining customers will henceforth have the burden of showing competitive harm, should have applied the new rule to the cases at hand, absent manifest injustice. Cf. Clark-Cowlitz Joint Oper. Agency v. FERC, 826 F.2d 1074, 1081-82 (D.C.Cir.1987) (en banc), cert. denied, 485 U.S. 913, 108 S.Ct. 1088, 99 L.Ed.2d 247 (1988), cited in Opinion 284, 40 F.E.R.C. at 62,183 n. 65; Illinois Power Co., 52 F.E.R.C. ¶ 61, 162, at 61,624 (1990). In the alternative, Edison argues that its evidence in both cases was sufficient to overcome the presumption. Because we agree on the evidence, we do not reach the retroactivity issue.
The Commission’s approach to “potential” anticompetitive effects springs from the agency’s view that it must “consider the broad policies expressed in the various antitrust statutes” to “determine whether the objectives of these statutes are being hindered” by price discrimination. Connecticut Light, 8 F.E.R.C. at 61,653. One objective of antitrust policy is “to prohibit practices whose effect is or could be to harm competition.” Id. at 61,654 (emphasis added). Thus, the Robinson-Patman Act, which prohibits price discrimination between purchasers of like commodities where the effect “may be substantially to lessen competition,” 15 U.S.C. § 13(a), “does not require that the discriminations must in fact have harmed competition, but only ... a reasonable possibility that they ‘may’ have such an effect,” Corn Prods. Refining Co. v. FTC, 324 U.S. 726, 742, 65 S.Ct. 961, 89 L.Ed. 1320 (1945). See 8 F.E.R.C. at 61,654. Similarly, section 7 of the Clayton Act, 15 U.S.C. § 18, requires only a “reasonable probability” that an acquisition will result in competitive harm. See 8 F.E.R.C. at 61,654 (quoting Brown Shoe Co. v. United States, 370 U.S. 294, 323 n. 39, 82 S.Ct. 1502, 1523 n. 39, 8 L.Ed.2d 510 (1962)).
The Commission took its evidentiary presumption directly from FTC v. Morton Salt Co., 334 U.S. 37, 68 S.Ct. 822, 92 L.Ed. 1196 (1948). See 8 F.E.R.C. at 61,654. In Morton Salt, which dealt with the practice of providing substantial quantity discounts to purchasers of carload lots of salt, the Court ruled that enforcement of the RobinsonPatman Act would be hampered by requiring the government to prove what was, under the circumstances, “self-evident”: “that there is a ‘reasonable possibility’ that competition may be adversely affected by a practice under which manufacturers and producers sell their goods to some customers substantially cheaper than they sell like goods to the competitors of these customers.” 334 U.S. at 50, 68 S.Ct. at 830.
The Commission’s use of the Morton Salt inference must be qualified in two important respects. First, the inference is not irrebutable; as the Commission itself recognized, it merely shifts to the defending party the initial burden of producing the minimum quantum of proof necessary to support a finding of the nonexistence of the presumed fact. See Opinion 284, 40 F.E.R.C. at 62,171 (citing Pennzoil Co. v. FERC, 789 F.2d 1128, 1138 (5th Cir.1986)). “Specific, substantial evidence” showing an absence of the presumed competitive injury is “sufficient to rebut what is, after all, only an inference.” Boise Cascade Corp. v. FTC, 837 F.2d 1127, 1144 (D.C.Cir.1988). Thus, Edison may overcome the presumption of potential effects by presenting substantial evidence that there was no reasonable probability the price discrimination harmed competition. See City of Groton *333v. Connecticut Light & Power Co., 662 F.2d 921, 935 (2d Cir.1981).
Second, the competition at risk in a Robinson-Patman case fundamentally differs from that in a price squeeze case, especially one arising in the fully regulated electric power industry. Morton Salt’s inference was meant to apply where discriminatory pricing directly harms competition between the buyers. See 334 U.S. at 49-50, 68 S.Ct. at 829-30. Price squeeze occurs when a firm with monopoly power on the primary, or wholesale, level engages in a prolonged price increase that drives competitors out of the secondary, or retail, level, and thereby extends its monopoly power to the secondary market. See Town of Concord v. Boston Edison Co., 915 F.2d 17, 18, 23 (1st Cir.1990), cert. denied, — U.S.-, 111 S.Ct. 1337, 113 L.Ed.2d 268 (1991); cf. United States v. Aluminum Co. of Am., 148 F.2d 416, 437-38 (2d Cir.1945). The anticompetitive effect of price squeeze is indirect. The squeeze must last long enough and be severe enough to produce effects in the secondary market.
The addition of regulation makes it “less likely that a price squeeze will actually drive independent distributors from the marketplace.” Town of Concord, 915 F.2d at 26. The retail service franchises granted by the state are themselves monopolies in which a single distributor of electricity supplies “relatively immobile customers.” Id. If a utility is able to squeeze a competing distributor out of the retail market, it still cannot take over the competitor’s franchise without the state’s permission. Id. In Connecticut Light, the Commission took notice of these factors: “[T]he regulated industries have not been viewed as industries in which competition can be seen in the form of active rivalry between alternative suppliers of the same product.” 8 F.E.R.C. at 61,653. Rather, competition between a utility and its resale customers is strongest for those purchasers, like industrial and other commercial firms, that can choose a supplier when deciding where to locate; although there is also competition for service areas, this competition is “on a long term basis.” See id. (quoting Borough of Ellwood City v. Pennsylvania Power Co., 462 F.Supp. 1343, 1346 (W.D.Pa.1979)).
With these qualifications in mind, we conclude that Edison made a sufficient showing to rebut the presumption in both cases. In the first proceeding, Edison presented testimony of Dr. Joe D. Pace, an industrial organization and public utility economist, concerning the probable effects of the eleven-and-a-half months of price discrimination on competition to attract and retain large industrial, or A-8, customers, and on “fringe area” competition for A-8 customers located near the borders of competing service areas. Because price discrimination was confined to the A-8 rate, the evidence properly focused on competition for large industrial customers. Dr. Pace testified that price discrimination of the magnitude and duration at issue would be unlikely to affect industrial customers’ locational decisions. These decisions are made “with the objective of minimizing costs over a period of years,” and industrial firms that are particularly sensitive to the cost of electricity “must be concerned with expected long-term rate relationships when choosing a site for new or expanded operations.” Testimony of Joe D. Pace, July 15, 1983, J.A. No. 90-1236, at 396. Choosing an industrial location on the basis of short-term rate differentials, especially ones that are only mildly unfavorable, would be irrational.
Dr. Pace quantified the total rate-of-return disparity over the period as 0.67 percent on a test year basis. Eliminating the disparity would have produced a wholesale rate decrease that, if fully passed on by the Cities, would in turn have lowered the Cities’ industrial retail rates by an amount equal to two to five percent of the average A-8 customer’s revenue per kilowatt-hour. Because electricity costs are less than five percent of a firm’s total costs, even in high-energy industries, this rate differential would amount to an overall cost savings of less than 0.25 percent.
Dr. Pace concluded that such a differential would not cause a price-sensitive firm to ignore long-term rate relationships. He *334testified that this is especially true in southern California, where strict local air quality standards impose prohibitive costs on energy-intensive industries. Turning to specific industrial customers identified by the Cities, he explained why, in his opinion, there was no reasonable probability that any had been lost because of the rate relationship.
Finally, Dr. Pace discussed fringe area competition. He explained that the annexation of fringe areas is “more or less permanent” and involves “a host of economic and social factors in a long run context.” The R-2/A-8 rate disparity could not have a rational influence on any annexation decision by the Cities, unless the area in question contained predominantly A-8 customers. Id. at 407. Given the length of the period at issue, the disparity would only have caused the Cities to delay any such extensions of service until 1977. Again, Dr. Pace went through the Cities’ examples and explained why he believed the rate relationship had not been a determining factor in any annexation decision.
In the second proceeding, Edison presented testimony from Dr. Pace as well as from another industrial organization economist, Dr. John H. Landon. Based on his review of evidence produced by the Cities, Dr. Landon testified that during the twenty-three months of the locked-in period, there was no indication of any effect on franchise competition or of any consideration given by the Cities to selling their franchises. He found no attempt by the Cities to match Edison’s rates, and, reviewing the profit margins and financial records of the municipal electric systems, he concluded that all of them, with the exception of Banning’s (whose distribution costs were three times higher), were in sound financial shape and could have sustained rates comparable to Edison’s.
Dr. Landon, like Dr. Pace, further testified that the rate relationship would not have affected the locational decisions of industrial customers. On the issue of fringe competition, he could find no evidence of border customers switching to Edison; nor had the rate relationship impaired the Cities’ ability or willingness to expand their service areas, because they had actively done so. He also found nothing to indicate adverse effects to “yardstick” competition, which he defined as pressure from customers for a change in franchise ownership and pressure for more efficient management of the municipal system.
Dr. Pace also testified that no anticompetitive effects were reasonably probable. Because the disparity involved all retail rates, the Cities viewed franchise competition as the most fundamental. Dr. Pace testified that a rate differential of the magnitude and duration claimed by the Cities could not reasonably threaten the disenfranchisement of any of the municipal systems because rational franchise decisions are made over a much longer period and “tend to be once-and-for-all, all-or-nothing decisions.” Id. at 651. The same conclusions, in his opinion, applied to yardstick competition, which at bottom is simply the threat of disenfranchisement. Finally, his testimony supported Dr. Landon’s views concerning potential effects on locational decisions of retail customers and on fringe area competition.
This testimony constitutes substantial evidence to support a finding that no adverse competitive effects were reasonably probable. The Commission offered no basis for concluding that such testimony “consisted] mainly of conjecture,” see 40 F.E.R.C. at 62,172. In each case, the Commission’s finding that Edison failed to rebut the presumption rested principally on the judgment that “witnesses have testified persuasively in other proceedings that the only valid basis for analyzing price squeeze is to evaluate what was confronting the wholesale customer from day-to-day.” 40 F.E.R.C. at 62,172; see 51 F.E.R.C. at 61,-890. If this statement is meant to be a rule of general application, it conflicts with the Commission’s proper focus on the “magnitude and duration” of price discrimination, see 40 F.E.R.C. at 62,166 (emphasis added), and would seem to impose an impossibly heavy burden on the utility, see id. at 62,186 (Comm’r Sousa, dissenting in part). Under this rule, competitive harm *335would seem to result from a one-day rate disparity, unless it could be shown that not a single skittish customer made a judgment that day that Edison was more economical than the municipal system.
If the statement represents a finding, it is not supported by substantial evidence on the record. The “day-to-day” concept originated with the testimony of an expert witness presented by intervenors in an unrelated price squeeze proceeding. See Pennsylvania Power Co., 11 F.E.R.C. 11 63,009, at 65,054 (1980) (subsequent history omitted), quoted in 34 F.E.R.C. at 65,-297. This witness expressed the opinion that in a price squeeze the wholesale buyer is harmed because he cannot compete with his retail customer’s “evaluation of what the competitor supplier is charging for like service right now. ” Id. (emphasis added). This focus on day-to-day perceptions and fleeting rate disparities, borrowed from the record in another proceeding, cannot sweep away the evidence put forth by Edison in the present cases. Moreover, as an approach to the effects inquiry, it is inconsistent with the Commission’s principle that “the subjective plans of individual customers are irrelevant,” 40 F.E.R.C. at 62,172 (quoting City of Groton, 662 F.2d at 934).
In Opinion 284, the Commission also based its finding, in part, on Dr. Pace’s response to the claim that Edison’s price discrimination had caused the City of Riverside to delay extending service to an area known as “Lily-Tulip,” which it had annexed in 1975. The Commission noted that “witness Pace conceded that there is ‘some probability’ that Riverside’s extension of service to Lily-Tulip would have gone forward in 1976 had the wholesale/retail rate disparity not existed.” Id. Dr. Pace’s recognition of a possible link is not sufficient to support a finding of nonrebuttal. It does not undercut his general conclusion that no fringe area competition had been affected or his specific opinion that no connection between the rate disparity and Riverside’s decision was reasonably probable. He stated that although “elimination of the A-8/wholesale rate disparity in 1976 would have made the economic comparison of the costs and benefits of extending service much closer from the City’s point of view, the evidence does not suggest that the City would have acted at that time. In reality, Riverside chose to extend service only after observing a several-year period of extremely favorable A-8/wholesale rate relationships.” J.A., No. 90-1236, at 412-13.
As further support for finding against Edison in Opinion 347, the Commission pointed out that the competitive disadvantage a wholesale customer suffers as a result of a rate disparity is itself an anti-competitive effect. See 51 F.E.R.C. at 61,-890. This conclusory judgment does not suffice as a finding. It simply restates the inference of competitive harm, which Edison overcame.
The Cities claim that Edison made no attempt to rebut evidence of actual effects that the Cities introduced in Phase I of the first proceeding as part of their prima facie showing of competition. See 34 F.E.R.C. at 65,306 n. 35. It is not clear that Edison had an obligation specifically to rebut all Phase I findings on the issue of competition that might also relate to effects. In any event, the Commission based its findings of competitive harm in both Opinion 284 and Opinion 347 squarely on the presumption of potential, not actual, anticompetitive effects. We conclude only that the utility rebutted that inference.
Edison goes further and argues that in each case the record conclusively proves that no adverse effects were reasonably probable. That finding is not ours to make. Rather, we must remand each order to the Commission to determine whether, absent the presumption, the evidence suffices to support a finding of anticompetitive effects.
The Cities and Edison also challenge the extent and the scope of the refunds ordered by the Commission. Given our conclusion that Edison rebutted any inference of anticompetitive effects, we need not reach such issues. If, on remand, the Commission once again finds that Edison’s rate disparities led to a price squeeze, the Commission may revisit the question of remedy. *336If the parties still are not satisfied, they may reassert their arguments in any subsequent petitions for review.
III. Conclusion
We affirm the Commission’s rulings on price discrimination in all respects, but reverse its findings on anticompetitive effect and remand both orders for further proceedings consistent with this opinion. The petitions for review in Nos. 90-1236 and 90-1515 are denied; the petitions in Nos. 90-1369 and 90-1566 are denied in part and granted in part.
So ordered.