Opinion for the Court PER CURIAM.
Separate concurring opinion filed by Senior Circuit Judge MacKINNON.
PER CURIAM:Petitioners, the Electricity Consumers Resource Council (“ELCON”), an organization of large industrial users of electricity, and the Cities, Villages and Towns of Cedarburg, et al. (“the Cities”), a group of wholesale customers, challenge two orders of the Federal Energy Regulatory Commission (“FERC”). The first order, Opinion No. 186,1 accepted and adopted in full a new rate design proposed by the Wisconsin Electric Power Company (“WEPCO”), reversing the initial decision of the Administrative Law Judge (“AU”) on all issues.2 The second order, Opinion No. 186-A,3 denied rehearing on the ground that the rehearing applications raised no new issues. The principal issue presented by these consolidated appeals is whether FERC's finding — that WEPCO’s proposed new rate design .adopting two new pricing methods is just, reasonable, non-preferential and nonprejudicial — is based upon substantial evidence in the record and is within a zone of reasonableness.
This case arises under the Federal Power Act, 16 U.S.C. § 791a, et seq. (1982), (“the Act”), and associated regulations. Section 205(a) of the Act4 requires that all rates subject to the Commission’s jurisdiction be “just and reasonable,” and declares rates that do not meet these standards to be unlawful. In addition, section 205(b) mandates that rates not be unduly preferential *399or prejudicial. Section 205(e) imposes on the public utility the burden of proving that any proposed rate is just and reasonable.
For the reasons set forth below, we affirm FERC’s implementation of the time-of-day pricing component of the new rate design but reverse FERC’s adoption of the marginal pricing component. We find that, with regard to the marginal pricing component, the Commission failed to comply with its statutory mandate to ensure that rates are just, reasonable, non-discriminatory and non-preferential. Moreover, we find that FERC both failed to offer a reasoned explanation for its decision and failed to develop record support, based on the facts of this case, for the adoption of the new marginal pricing ratemaking methodology. Accordingly, we remand this case for further proceedings consistent with this opinion.
I. BACKGROUND
On July 30, 1980, WEPCO filed a wholesale electric rate increase which was made the subject of a FERC investigation under section 205(e). All aspects of WEPCO’s rate proposal were satisfactorily resolved by settlement negotiations prior to hearing except for the two new concepts in the rate design that are at issue here: the substitution of time-of-day energy charges for unitary energy charges, and the use of marginal cost pricing concepts 5 instead of the traditional embedded or average cost pricing. The new rate design still recognizes two discrete types of costs: (1) demand or fixed costs, i.e., building and maintenance expenses, and (2) energy or variable costs, i.e., the cost of generating electricity.
Petitioners do not challenge the adoption of time-of-day rates in principle. Brief of ELCON, 32-33; Brief of the Cities, 59-60. They argue only that WEPCO has not offered sufficient evidence to support cost differentials in the periods chosen, and that neither Exhibit 5 nor Exhibit 32, offered as record support for the choice of peaking periods, suggests that a 12-hour peak period is justified. Id., J.A. 60-61. FERC, on the other hand, found the selection of peak periods to be reasonable and to be supported by substantial evidence on the record. J.A. 45. We agree and affirm this portion of FERC’s order.
Petitioners challenge the adoption of marginal cost pricing in this case on several grounds, the most important of which is that substantial adjustments to the theory were made in order to apply it to this case. These adjustments were necessary because of the “revenue constraint,” representing the maximum revenue collectable by the utility in a given period. In this case, the revenue constraint was determined according to the traditional ratemaking method of average costs. Consequently, if WEPCO collects the fixed costs allocated to wholesale customers by a demand charge and the variable costs by an energy charge under marginal cost pricing instead of average pricing, the total revenues generated will exceed the revenue constraint. Therefore, FERC approved the following reconciliation methodology: WEPCO would collect the entire marginal cost of energy and would consider any residual revenue collectable under the revenue constraint to be the demand charge. J.A. 22, 76. The resulting charges are not pegged to the demand or energy costs assigned to the customers in WEPCO’s cost-of-service study, and there are no demand charges for off-peak customers. J.A. 22-23.
II. DISCUSSION
A. Standard of Review
The scope of our review in this case is limited. We defer to the agency’s expertise, particularly where the statute prescribes few specific standards for the agency to follow, so long as its decision is supported by “substantial evidence” in the record and reached by “reasoned decision-making,” including an examination of the relevant data and a reasoned explanation supported by a stated connection between the facts found and the choice made. Burlington Truck Lines v. United States, 371 U.S. 156, 168, 83 S.Ct. 239, 245, 9 L.Ed.2d *400207 (1962); Memphis Light, Gas and Water Division v. FPC, 504 F.2d 225, 230 (D.C.Cir.1974); 16 U.S.C. § 825l (1982). As this court recently stated in a case arising under the Natural Gas Act, which prescribes a similar “just and reasonable” standard:
This paucity of statutory guidance has resulted in judicial deference to administrative ratesetting. However deferential the judiciary may be, courts have never given regulators carte blanche ____ [T]he judiciary has inferred certain requirements from the “just and reasonable” standard [and] ... “[w]hat is basic is the requirement that there be support in the public record for what was done” ____ In reviewing the record, this Court engages in an inquiry akin to the “substantial evidence” inquiry mandated by the Administrative Procedure Act, 5 U.S.C. § 706(2)(E) (1976) .... Indeed, this Court has required the Commission to specify the evidence on which it relied and to explain how that evidence supports the conclusion it reached.
City of Charlottesville v. FERC, 661 F.2d 945, 949-50 (D.C.Cir.1981) (citations omitted).
B. FERC’s Order Fails to Comply With the Statutory Mandate to Determine Whether Rates are Just, Reasonable, Non-Discriminatory and Non-Preferential
The Commission relied exclusively on economic theory to justify the adoption of a rate design based on marginal costs. But, especially in a case of this sort— where the agency has given lip service to a theoretical design and then inexplicably distorted the theory in its application — mere reliance on an economic theory cannot substitute for substantial record evidence and the articulation of a rational basis for an agency’s decision. While we might defer to FERC’s application of this new theory in another case, we cannot do so here where the agency failed to address potentially serious problems with the new rate design, especially with WEPCO’s reconciliation methodology.
1. The Reconciliation Methodology
FERC asserts that the new rate design “will serve the public interest by more closely tracking the utility’s cost incurrence than the traditional non time-differentiated average cost methodology.” J.A. 46. This contention is hardly surprising because FERC has previously considered cost tracking to be the primary feature of a just and reasonable rate structure.6 Nevertheless, it is clear in this case that, under WEPCO’s proposal of modified marginal cost pricing, costs will be tracked less accurately than under average cost pricing.
For instance, WEPCO’s Exhibit 20, J.A. 572, shows that demand costs approximate 41% and energy costs approximate 59% of the actual cost of service. Under WEP-CO’s proposal, however, demand charges will equal about 20% of the revenues and energy charges will equal 80%. WEPCO’s own witness acknowledged the total lack of correlation between the demand cost and the demand charge, J.A. 146, stating under cross-examination that “it is difficult to calculate an accurate marginal demand cost and inasmuch as our philosophy is as it is to shrink the demand cost first, we didn’t pay a lot of attention to the theoretical niceties of that calculation.” J.A. 241 (emphasis added).
*401WEPCO’s witness also testified that the reconciliation necessitated by the revenue constraint should be made where it will have the least adverse impact on the allocation of resources, and argued that the demand charge represents that part of the rate structure. The Commission accepted two of three reasons put forth by WEPCO to support the reconciliation methodology chosen, placing primary reliance on WEP-CO’s stated excess capacity:
The excess capacity argument is the key factor in our conclusion. Because there is currently excess capacity, WEPCO’s shortrun marginal cost of capacity is zero. The logical conclusion of this fact, all other things being equal, would be a zero demand charge both on-peak and off-peak. But even with energy costs priced at their margin, that would produce a revenue shortfall below the company’s revenue requirement. This revenue short-fall [sic] should be avoided by increasing the demand charge.
J.A. 58. In accepting this rationale, however, the Commission appears to treat the demand charge like modeling clay in order to force the application of marginal cost pricing to the energy component.
The Commission’s second rationale is equally conclusory. Again noting the present excess capacity, the Commission simply asserts that “it is legitimate in setting a demand charge to focus on the short-term.” J.A. 58. The Commission fails to explain why a short-term focus is acceptable, or to discuss the potential effects of this short-term approach. We believe that this approach poses serious potential problems, including the future possibility of a sudden rate surge when additional capacity is required. FERC has totally failed to address these problems in any meaningful fashion.7
Accordingly, we find that the record lacks substantial evidence to support the reconciliation methodology chosen, and that the Commission’s stated reasons for its approval are almost wholly conclusory, largely short-sighted and patently unpersuasive.
2. The Rate Tilt
There is a similar absence of evidence and explanation for the approval of a “rate tilt” that results from the application of WEPCO’s rate design. Under the Act, a rate design must be non-discriminatory and non-preferential as well as being just and reasonable. 16 U.S.C. § 824d(b); see also Alabama Elec. Co-op., Inc. v. FERC, 684 F.2d 20, 27 (D.C.Cir.1982) (quoting United States v. Illinois Central R.R. Co., 263 U.S. 515, 524, 44 S.Ct. 189, 193, 68 L.Ed. 417 (1924)).8 If a rate design has different effects on charges for similar services to similar customers, the utility bears the burden of justifying these different effects. It can satisfy this burden by “offering a valid reason for the disparity or by demonstrating that the gap is as small as practicable under the circumstances.” Id. at 29. Not all discrimination is necessarily unlawful. Section 205(b) proscribes “any unreasonable difference in rates” and any “undue preference or advantage.” Id. at 28 (emphasis added); see also Second Taxing Dist. v. FERC, 683 F.2d 477, 486-88 (D.C.Cir.1982); Pub. Serv. Co. v. FERC, 575 F.2d 1204, 1211 (7th Cir.1978); Southwestern Pub. Serv. Co., 22 F.E.R.C. (CCH) ¶61,341 (1983).
Petitioners have shown, and the record makes clear, that the proposed rate design *402results in a cross-subsidization, charging high-load factor customers part of the costs of service to low-load customers. Brief of ELCON, 36; J.A. 297-99.9 The utility has put forth no legally sufficient reason for charging high-load factor customers a rate that does not accurately reflect the cost of serving them. Indeed, at oral argument before this court neither FERC’s counsel nor WEPCO’s counsel even attempted to explain or justify the rate tilt.
In light of the record before us, it is plain that WEPCO has not met its • burden of showing that such a rate tilt is reasonable. In addition, FERC has made absolutely no attempt to outline the factors justifying the rate tilt. Instead, the Commission’s decision offers only frivolous and circular reasoning in an effort to deny the existence of a tilt, and the brief from FERC fails to address the issue.
In its opinion, FERC stated:
Admittedly, these results are not likely to be beneficial from the perspective of certain customer classes. This explains the source of the opposition of the high load factor municipal and industrial intervenors. Other customer groups, however, are likely to benefit from WEPCO’s proposal. This trade-off of customer gains and losses is a fact of life ____ The important question is whether there is a cost justification for the rate structure adopted. We find that there is. WEPCO’s proposal comports with the well-reasoned theory of marginal cost pricing ____ [Ojnce marginal costs are chosen as the standard to be applied, it cannot be complained that the result lacks a rational basis.
Nor can it be complained that the result promotes an unlawful rate tilt. In point of fact, no “tilt” exists as that term is commonly understood because the result obtained here comports entirely with the pricing and reconciliation methodologies which underly [sic] it. There has been no deviation from marginal cost pricing theory. If a “tilt” can be said to exist, it is a “tilt” from average cost pricing. But that is no tilt at all because WEPCO never intended to use average cost pricing.
J.A. 46-48 (emphasis added). FERC’s reasoning is totally circular and utterly without meaning. It is uncontested that, under the proposed rate design, part of the demand cost is being collected through the energy charge. Yet, the Commission accepts the reconciliation methodology as consistent with marginal price theory. Therefore, the tilt logically arises from marginal pricing, not from average pricing as the Commission asserts. Even assuming, arguendo, that the tilt does arise from “average cost pricing” (as claimed by FERC), it still must be shown that the proposed rate design is non-discriminatory and non-preferential. See note 8 supra. In this case, neither WEPCO nor FERC has produced any evidence showing factual differences to justify the tilt or to show that it is de minimis. Such a rate tilt, without legal justification, is unlawful. Columbia Gas Transmission Corp. v. FERC, 628 F.2d 578, 592-93 (D.C.Cir.1979).
C. The Alleged Application of Marginal Price Theory
Petitioners complain that WEPCO’s record evidence is limited to "recitations and quotations of marginal cost pricing theory ... and a few broad, unsubstantiated assertions that the same design employed *403at retail has had beneficial results.” Brief of the Cities, 23; see also Brief of ELCON, 20-21. The AU was struck by the lack of a record justification for the new model:
The adoption of any particular mode of pricing can hardly be an end in itself, no matter what the mode may be. The results of a particular mode, its effects upon the adopting utility and upon the consuming public, those are the ends that give the mode value, that justify its adoption. Those results may be seen only through analysis of the actual facts and through comparisons with the probable results of the adoptions of different modes. It is only by comparisons with other modes that the value of a particular mode may be identified.
The record now before the Commission shows that WEP [sic] has made no analysis of the type just described, has made no such comparison.
J.A. 9. The Commission’s reversal of the AU’s order, by stark contrast, merely states that “once marginal costs are chosen as the standard to be applied, it cannot be complained that the result lacks a rational basis.” J.A. 48.
We conclude that, according to the applicable law in this circuit, mere invocation of theory is an insufficient substitute for substantial evidence and reasoned explanations. This is especially true in the context of this case where the theory has been severely compromised by the revenue constraint and the resulting modification in marginal pricing methodology.
Apart from the problems discussed above, petitioners have pointed out that pure marginal pricing theory in this case is limited in at least two additional ways. First, the theory assumes a state of pure competition which even FERC concedes does not exist in the wholesale electric utility industry in Wisconsin.10 Second, there is evidence in the record suggesting that some effects of the new design might be economically inefficient.11 The Commission’s order did not address this latter point.
The principal point to be made here is not any opposition to marginal price theory (in its pure or a modified version), but rather that mere economic theory may not take the place of record evidence and reasoned decision-making. This point was made clear by this court in City of Charlottesville v. FERC, 661 F.2d 945 (D.C.Cir.1981), where FERC approved an increase in rates for two interstate pipeline companies purely on the basis of an incentive theory. We found that the record lacked any meaningful evidence of a causal relationship between the rate and the theoretical design, vacated FERC’s decision and remanded, quoting Judge Leventhal’s opinion in Greater Boston Television Corp. v. FCC, 444 F.2d 841, 852 (D.C.Cir.1970), cert. denied, 403 U.S. 923, 91 S.Ct. 2229, 29 L.Ed.2d 701 (1971):
An agency’s view of what is in the public interest may change, either with or without a change in circumstances. But an agency changing its course must supply a reasoned analysis ... and if an agency glosses over or swerves from prior precedents without discussion it may cross the line from the tolerably terse to the intolerably mute.
City of Charlottesville, 661 F.2d at 951 n. 35 (emphasis added). In her concurrence, Judge Wald added:
That evidence is not to be found in this record, and ... I do not think a purely theoretical incentive, the effectiveness of which is belied by the facts in the specific case, is sufficient to justify a deviation *404from normal ratemaking practices ---- Though this court in the past has approved agency use of predictive economic models ... we have always stressed that the agency must make “a conscientious effort to take into account what is known as to past experience and what is reasonably predictable about the future” to monitor whether the model’s assumptions work in practice.
Id. at 954-55 (quoting American Pub. Gas Ass’n v. FPC, 567 F.2d 1016, 1037 (D.C.Cir.1977), cert. denied, 435 U.S. 907, 98 S.Ct. 1456, 55 L.Ed.2d 499 (1978)).12
III. CONCLUSION
The heart of our problem with FERC’s approval of distorted marginal cost pricing is that the Commission relied solely on theoretical postulates that, so far as we can divine from the record and FERC’s orders, bear no relation to present realities in this case. Although we strongly prefer to defer to the expertise of the agency with regard to rate design, we cannot do so where FERC has failed to consider relevant factors and to articulate a reasonable basis for its decision, and where FERC has not complied with its statutory mandate to set rates at a just, reasonable, non-discriminatory and non-preferential level. We repeat that FERC is not foreclosed from relying on marginal price theory if there is a basis for its adoption in this case. We vacate FERC’s order, not because it relied on pure theory, but because the theory was applied in this case despite the fact that the theory upon which FERC bases its approval was not actually used here.
In sum, we unequivocally state that we are not hereby expressing our opposition to the adoption of marginal cost based rate designs in any form. We are concerned only with the total lack of record support for FERC’s position and with the absence of reasoned decision-making on the part of FERC. There are a number of suggested post-hoc rationalizations in FERC’s and the intervenors’ briefs on review but those cannot substitute for record evidence or reasoned decision-making. The agency should consider and explain the effect of the reconciliation methodology on cost-tracking, the long-term problems inherent in ignoring the demand component at this point and the reasons that justify the acknowledged rate tilt. The agency should also explain and spell out in more than purely theoretical terms how the alleged economic efficiencies flow from the modified or adapted version of the theory applied in this case.
Accordingly, we remand the case to allow FERC and interested parties to supplement the record on the issue of marginal cost pricing. A modified version of marginal cost pricing theory may indeed be perfectly acceptable, but not without record support and reasoned decision-making by FERC that address the particular nuances of the modified theory. If the adoption of marginal cost theory to WEP-CO’s wholesale rates is justified by the further proceedings, we have no doubt that the agency’s decision will be upheld.
We affirm FERC’s implementation of the time-of-day pricing component of the new rate design but reverse FERC’s adoption of the marginal pricing component. The case is remanded for further proceedings consistent with this opinion.
So ordered.
. Opinion and Order Setting Just and Reasonable Rates, Docket No. ER 80-567-000 (September 19, 1983), Joint Appendix ("J.A.”) 17-61.
. Initial Decision, Wisconsin Electric Power Company, Docket No. ER 80-567-000 (February 22, 1982), J.A. 1-16.
. Opinion and Order Denying Rehearing, Docket No. ER 80-567-000 (November 17, 1983), J.A. 62-64.
. 16 U.S.C. § 824d(a) (1982). All statutory references herein are to the Federal Power Act.
. Marginal costs represent the cost of producing an additional unit at a given time.
. See Brief for Appellee FERC, 18; Northwest Pipeline Corp., 23 F.E.R.C. (CCH) ¶ 63,079 (1983). This Court has stated:
While neither statutes nor decisions of this court require that the Commission utilize a particular formula or a combination of formulae to determine whether rates are just and reasonable, it has come to be well established that electrical rates should be based on the costs of providing service to the utility’s customers, plus a just and fair return on equity. FERC itself has stated that ”[i]t has been this Commission’s long standing policy that rates must be cost supported. Properly designed rates should produce revenues from each class of customers which match, as closely as practicable, the costs to serve each class or individual customer."
Alabama Elec. Co-op., Inc. v. FERC, 684 F.2d 20, 27 (D.C.Cir.1982) (quoting Carolina Power and Light Co., Docket No. ER76-495, By Direction Letter, 2 Oct. 1979 (emphasis in original)).
. See Brief of ELCON, 28-31, for additional potential problems that were not considered including the disruptive impact on its customers of the skewed marginal cost pricing proposal.
. In Alabama Elec. Co-op., Inc. v. FERC, supra, this Court held that section 205(b) can be violated by the application of the same rate design to two groups of customers which are largely similar. In that case, the rate design yielded a different rate of return on the capital devoted to serving the municipalities from that devoted to serving the cooperatives: the difference amounted to 0.45%. The court, in remanding, stated:
But even if we cannot at this point hold that the discrimination at issue here is per se unreasonable neither can we say, as a matter of law, that the disparity is so small as to be per se reasonable. Translated into dollars, it is not insignificant.
684 F.2d at 28.
. As to the "tilt," FERC’s staff witness stated:
I oppose WEPCO’s [sic] proposed margin cost based energy rates for several reasons. First, as stated by Company witness Abdoo, there cannot be a complete reconciliation between the theories of marginal cost and average cost. Since WEPCO’s filed cost of service derives an allocated revenue requirement utilizing average cost, using marginal cost based rates creates some inconsistencies. As can be seen from page 15 of Exhibit No._(S-2), using the Company's costs from its case-in-chief, over $10 million in demand related costs are to be recovered through the proposed marginal cost based rates thereby causing a severe tilt.
This tilt causes an inappropriate price signal to be sent to ratepayers by the demand portion of the rate structure, because the demand charge is being understated ....
J.A. 297-98.
. Petitioner claims that a switch to a different supplier would require three years in WEPCO’s case. Brief of the Cities, 26 n. 9.
. The City of Cedarburg’s manager testified that Cedarburg shut down its peaking plant and changed from a partial to an all-requirements customer as a result of the rate design charge. This occurred because of the low demand component charge, removing the incentive for wholesale customers to operate their own generational facilities for peak load control. In fact, there is evidence that the city can generate its own peak load at a lower real cost. Thus general economic efficiency might be reduced and WEPCO's peak load factor increased. J.A. 196-201, 212-14.
. This court articulated a similar standard in Columbia Gas Transmission Corp. v. FERC, 628 F.2d 578 (D.C.Cir.1979), and Arizona Pub. Serv. Co. v. United States, 742 F.2d 644, at 653 (D.C.Cir.1984).