The petitioners in these consolidated cases are wholesale customers of electric utilities whose wholesale rates are regulated by the Federal Energy Regulatory Commission.1 After notice and comment rulemaking, FERC ruled that electric utilities may generally include in their rate bases amounts equal to 50% of their investments in construction work in progress ("CWIP")2. We affirm in part, and vacate and remand in part.

A regulated utility is, of course, entitled to recover the cost of financing the construction of facilities used to produce and transmit electric power. The "cost" includes interest on debt and a reasonable return on capital investment. What is controversial is the timing of this cost recovery. Under one frequently used method, the utility's rate base does not reflect investment in new facilities until they commence commercial operation. During construction, the investment and accrued carrying charges are carried in an account called "Allowance for Funds Used During Construction" ("AFUDC"). The carrying charges on debt are recorded as an offset to interest expenses, and the carrying charges on equity are recorded as current income--but in fact ratepayers are not making cash payments to the utility. When the plant begins service, the entire value of the investment, including deferred financing charges, is added to the utility's rate base. Once added to rate base, both the investment and the accrued carrying charges earn a reasonable rate of return and are depreciated over the life of the facility.

Under the CWIP method, by contrast, capital investment is added to the utility's rate base when made. Ratepayers pay a return on that investment while the facility is under construction. Thus, under the CWIP method the carrying charges are not accrued and so are not added to rate base when the facility goes into service. Only the investment itself--for which ratepayers do not begin paying until service commences--is added to rate base at that time. Thereafter a return is paid on the investment and it is depreciated over the life of the facility, just as under the AFUDC method.

Until 1976, the Commission authorized use of the AFUDC method only. In Order No. 555, 56 F.P.C. 2939 (1976), the Commission created three exceptions to the rule against CWIP in rate base. The first two exceptions allowed use of the CWIP method for "socially beneficial" investment in (1) pollution control facilities and (2) conversion of plants fueled by oil and natural gas to coal. Id. at 2943-46. The third exception allowed a utility to include CWIP in rate base if the utility made a clear and convincing showing that it was experiencing severe financial problems. Id. at 2946. At the time of the rulemaking in this case, FERC had never issued an order allowing CWIP in rate base on the basis of financial distress. Order No. 298, 48 Fed.Reg. 24,323, 24,338 (1983) (to be codified at 18 C.F.R. Sec. 35.26), Joint Appendix ("J.A.") at 1298. We upheld the rule allowing these exceptional uses of CWIP in rate base without opinion in Oglethorpe Electric Membership Corp. v. FERC, 574 F.2d 637 (D.C.Cir.1978).

In July, 1981, FERC issued a Notice of Proposed Rulemaking "to amend its regulations concerning construction work in progress." 46 Fed.Reg. 39,445 (1981) (proposed July 27, 1981), J.A. at 1. The Commission published a proposed rule in the notice, but it also discussed and sought comments on "alternative changes to our existing rule relating to CWIP." Id., J.A. at 1. The proposed rule left intact the pollution control and fuel conversion exceptions for CWIP, and focused solely on revisions to the financial difficulty exception. The proposed revisions would have set out more detailed criteria for evaluating whether a utility was financially distressed and for determining the amount of CWIP that a distressed utility could include in rate base. Id., J.A. at 2. The Commission emphasized, however, that the rulemaking was not confined to evaluation of the proposed rule: "we intend to explore in this proceeding alternatives of broader scope than financial distress as a basis for allowing CWIP in rate base, as well as alternative formulations based on financial distress." Id., J.A. at 2. Among the alternatives FERC expressly stated it would consider were (1) maintaining the status quo; (2) allowing a straight percentage of CWIP in rate base regardless of financial condition; (3) allowing CWIP for a particular plant to go into rate base for a specified period prior to the time when the plant commenced operation; (4) allowing CWIP in rate base to the extent permitted by the utility's predominant state regulatory authority; and (5) requiring wholesale customers to pay contributions in aid of construction in lieu of CWIP. Id. at 39,425, J.A. at 36-37.

After receiving numerous comments from interested participants, FERC issued Order No. 298 as a Final Rule. The final rule "adopts a fixed percentage approach that allows any utility to file to include up to 50 percent of all CWIP in rate base in addition to any CWIP related to fuel conversion or pollution control facilities." 48 Fed.Reg. at 24,349, J.A. at 1350.

Order No. 298 is some 151 pages long. For now, we summarize the Commission's legal, economic, and factual presentation and postpone detailed discussion until we turn to petitioners' specific challenges.

FERC's new CWIP rule rests on a public interest rationale. FERC recognized that a literal interpretation of the principle that an asset must be "used and useful" before it can be included in rate base, see Smyth v. Ames, 169 U.S. 466, 18 S. Ct. 418, 42 L. Ed. 819 (1898), would foreclose allowing any CWIP in rate base under any circumstances. Acknowledging its general adherence to the "used and useful" principle, FERC noted that there are "widely recognized exceptions and departures from this rule, particularly when there are countervailing public interest considerations." 48 Fed.Reg. at 24,335, J.A. at 1286. Accordingly, FERC eschewed an inflexible understanding of the "used and useful" principle, believing it could depart to some extent from that concept "when the reliability of future service is in doubt," id. at 24,336, J.A. at 1287, and emphasizing that a rigid "used and useful" concept "may fail the interests of both the electric utility industry and its ratepayers." Id., J.A. at 1288.

The critical public interest consideration the Commission identified was its "responsibility to further the maintenance of an adequate and efficient electric utility industry," 48 Fed.Reg. at 24,332, J.A. at 1267. Hence FERC sought to strike "a reasonable balance between the principle of intertemporal cost responsibility and the need to create a regulatory setting within which the utility industry can supply the nation's need for electricity at the lowest reasonable cost." Id. at 24,329, J.A. at 1254. The Commission concluded that allowing CWIP in rate base would advance this least-cost strategy for the electric utility industry and hence promote the public interest in three ways:

--mitigat[ing] any bias which may discourage additional capital investment in needed facilities;

--enabl[ing] the need for those facilities to be more accurately evaluated by price signals which reflect today some of the costs associated with future facilities; and

--mitigat[ing] the sudden price increases which tend to result under an AFUDC policy thereby furthering the goal of rate stability.

Id. at 24,326, J.A. at 1237-38.

The Commission's three asserted purposes for its new CWIP policy, of course, rest on the Commission's assessment of the effects of an AFUDC-only rule on the electric utility industry, as well as its prediction of the effects of a 50% CWIP in rate base rule. Specifically, FERC found that an AFUDC policy tends, at least under current conditions, to create a bias against the creation of new capacity and a bias in favor of minimizing capital costs rather than total economic costs. 48 Fed.Reg. at 24,330, J.A. at 1258. An AFUDC policy has that effect because AFUDC earnings are paper earnings, not cash earnings. See id. at 24,339, J.A. at 1303. Consequently, "particularly when the magnitude of CWIP is large in relation to rate base," "financing difficulties arise because the internal cash flow generated by rate base investments is often insufficient to support the required sales of debt and equity securities" to finance the utility's construction program. Id. at 24,330, J.A. at 1260. This scenario, in the Commission's view, is increasingly likely, since by 1980 CWIP "constituted about 40 percent of net plant in service for Class A and B electric utilities," whereas CWIP averaged only about 20% of net plant in service between 1970 and 1974, and indeed only about 8% of net plant in service during 1950-1969. Id. at 24,329 n. 35, J.A. at 1255 n. 35. Furthermore, FERC found "convincing evidence in the record including a study prepared by the FERC and a study prepared for the Commission by EIA which shows an increase in cash flow from including 50 percent of CWIP which is sufficient to affect most utility financial indicators in a very positive fashion." Id. at 24,340, J.A. at 1310. Thus, the first ground for the Commission's contention that its CWIP rule would mitigate the perceived bias against optimal capital investment was that AFUDC led to cash flow problems that a CWIP rule would ameliorate.

In addition, FERC determined that its CWIP rule would reduce a second source of bias against capital investment. The Commission found that "[s]ince about 1974, utilities have not realized returns at levels approximating their cost of capital." 48 Fed.Reg. at 24,326, J.A. at 1241. This necessarily biases investor-owned utilities against capital investment, because "if a company earns less than its cost of capital, any additional investment constitutes a loss for its shareholders; they are poorer if the investment is made than if it is not." Id. at 24,327, J.A. at 1242.

Allowing CWIP in rate base alleviates this source of bias in two ways. First, "[i]nclusion of CWIP in rate base reduces a utility's investment," because it allows the utility to avoid investing in the carrying charges on the investment during construction. 48 Fed.Reg. at 24,327, J.A. at 1242. The utility's remaining investment is still an unattractive one, so long as the utility is earning less than its cost of capital, "[b]ut the loss is smaller than it would be if CWIP were excluded from rate base." Id., J.A. at 1243.

Second, the Commission had before it a number of studies concerning the effects of allowing CWIP in rate base on utilities' cost of capital. See 48 Fed.Reg. at 2340-41, J.A. at 1310-13. The Commission found methodological problems with all of these cost of capital analyses, but arrived at the "conclusion, which is admittedly no more than a matter of judgment, ... that we believe whatever the size of any effect that the inclusion CWIP in rate base will have on the cost of capital, the rule will generally provide a downward pressure on those costs." Id. at 24,341, J.A. at 1313. By reducing utilities' cost of capital, then, the CWIP rule would tend to reduce the shortfall between the utility's actual earnings and its cost of capital, and thereby weaken the disincentive to engage in capital investment.

FERC's second public interest rationale rests on the Commission's view that it is in the public interest "to enable the demand of customers for new capacity to be more accurately reflected in the planning forecasts of the utilities." 48 Fed.Reg. at 24,337, J.A. at 1294. FERC determined, on the basis of economic analysis and on its assessment of prevailing conditions in the industry, that the AFUDC method "mask[s] from the ultimate consumer the real cost of power during the period when a new plant is being constructed, thereby encouraging demand which serves to distort estimates of future demand and the attendant need for new capacity." Id. at 24,335, J.A. at 1284. FERC did not predicate its rule on the finding that "enormous new capacity is necessarily required." Id. at 24,334, J.A. at 1281. Rather, the Commission took the position that the need for new capacity should, to the extent practicable, be tested by the market. Id., J.A. at 1281. Current conditions, FERC concluded, made AFUDC increasingly unsuited to serving that goal.

The difficulty with AFUDC arises because that method does not allow current rates to reflect "the difference between the price of electricity at the time demand estimates are made and the price at which electricity must be sold when the planned facility comes on line." 48 Fed.Reg. at 24,331, J.A. at 1262. Whereas once new facilities could be expected to yield a decline in the price of electricity, "[t]oday the addition of a new facility is more likely to mean higher prices than lower ones." Id., J.A. at 1262. Consequently, estimates of future demand based on current prices tend to overpredict growth in demand, since, when new facilities commence service and the price of electricity rises, consumers will shift away from consuming electricity to other goods. "Then, when customers have adjusted electricity consumption decisions to the higher prices, it begins to appear--too late--that too much supply has been made available." Id. at 24,326, J.A. at 1237.

Including CWIP in rate base, FERC found, reduces this tendency to overpredict demand by passing on some of the cost impact attributable to a new plant to consumers throughout the entire construction period. 48 Fed.Reg. at 24,331, J.A. at 1263. Hence consumers will face higher prices sooner, and the extent of the resulting reduction in demand will allow the utility to avoid unnecessary capacity. Id., J.A. at 1263. "Thus inclusion of CWIP will generally allow utilities to pursue least total cost strategies to meeting their customers' electric power demands." Id., J.A. at 1263.

The Commission's third asserted purpose--rate stability--is closely related to the second. To some extent, the Commission appears to have viewed rate stability as desirable in itself because rate shocks may have a disruptive impact on consumers. See Order Granting in Part and Denying in Part Application for Rehearing, 48 Fed.Reg. 46,012, 46,019 (1983), J.A. at 1812-13. But, for the most part, the avoidance of the rate shocks that may result under an AFUDC-only rule when a new plant begins service was seen by FERC as a means to achieve a "substantially smoother" "price path" that would facilitate the assessment of need for new capacity. See 48 Fed.Reg. at 24,331, J.A. at 1263.

There is no doubt that these three public interest purposes played a decisive role in FERC's decision. FERC stopped short, however, of concluding that the importance of achieving these objectives would permit it entirely to disregard the used and useful principle. Instead, the Commission argued that its decision was largely though not completely consistent with that principle, understood as a principle of intergenerational equity. 48 Fed.Reg. at 24,33 6-37, J.A. at 1288-93.

FERC's first argument is that present consumers in fact receive "significant benefits" from capital expenditures for plants that have not yet commenced service. Since much of new utility construction is needed to meet the future requirements of current customers, capital investment in new facilities "confer[s] a present benefit on customers, namely, the reasonable assurance of a continuing quality of service. It is reasonable for current ratepayers to pay for this benefit." 48 Fed.Reg. at 24,331, J.A. at 1266. Indeed, FERC pointed out that an analogous practice has been accepted as consistent with even a strict interpretation of the used and useful principle: "current ratepayers pay for facilities necessitated by past demand and for the capital costs of unused capacity." Id. at 24,336, J.A. at 1291. The Commission did not contend that assigning some of the current costs of capital investment to current ratepayers matched cost responsibility with benefit enjoyment as well as an AFUDC-only rule. Rather, its claim was that substantial ratepayer equity would be achieved even under a CWIP rule. See id. at 24,336-37, J.A. at 1292.

It is also clear that FERC assigned some weight to the recent financial problems many utilities have experienced. FERC did not conclude that CWIP was necessary as a form of emergency relief for the industry as a whole. See 48 Fed.Reg. at 24,334, J.A. at 1277. FERC did, however, find that "the record in this proceeding clearly shows that the electric utility industry has been in a weak financial position. While general economic conditions may have exacerbated the financial difficulties of some companies, there is ample support that the electric utility industry has experienced particular difficulty in financing construction programs." Id. at 24,333, J.A. at 1276. While reiterating that its ultimate concern was "the long-term outlook in light of the service demands that are anticipated," FERC clearly inferred, from the "particular difficulty" utilities have recently had in financing new plants, that "a regulatory remedy to offset the financial drain posed by construction is necessary." Id. at 24,334, J.A. at 1277.

Having in effect made a preliminary determination that the used and useful principle does not preclude allowing CWIP in rate base, FERC then proceeded to assess the impacts of its CWIP rule--both beneficial and adverse--on consumers, and wholesale customers as well as on investor-owned utilities. FERC balanced, in a very general way, these various effects, and concluded that its public interest rationale tipped the scales decisively in favor of allowing 50% CWIP in rate base. But this implies that had FERC's assessment of the various effects of its CWIP rule on wholesale customers and consumers been different, the eventual outcome might have varied as well. We therefore summarize FERC's "impact" findings to set the stage for review of petitioners' objections that FERC disregarded or slighted the CWIP rule's adverse effects on them.

FERC recognized that inclusion of CWIP in rate base "raise[d] current electric bills and lowers future ones." 48 Fed.Reg. at 24,327, J.A. at 1243. After receiving the available evidence, FERC concluded that it was possible, though unlikely, that CWIP-based rates would remain higher indefinitely, and reasonably likely that they would remain "minimally higher" over the next twelve years, id. at 24,328, J.A. at 1250-51, before CWIP-based rates fall below AFUDC-based rates around 1990. See id. at 24,345, J.A. at 1335. But, in any event, FERC stated that, in view of the "generally modest rate impact" the rule is expected to generate, the need to ensure adequate and reliable future service would justify proceeding even if rates remained slightly higher for the foreseeable future. Id., J.A. at 1335.

FERC's assertion that rate impact would be moderate rests on two Commission-sponsored studies as well as studies submitted by commenters. FERC found that, after taking into account the CWIP rule's 6% ceiling on initial rate increases, initial price increases would average 3.57%. 48 Fed.Reg. at 24,344, J.A. at 1333. FERC viewed this conclusion as consistent with studies by commenters predicting (without considering the 6% limitation) increases between 5% and 6%, and discounted a study prepared by wholesale customers, predicting an increase of 8.7% as "not representative of the industry." Id. at 24,343, J.A. at 1327.

In addition to deciding that its public interest goals outweighed the adverse effect of these modest rate increases on wholesale customers, the Commission gave considerable weight, as an equitable consideration, to the asserted fact that investor-owned utilities in most states have an obligation to serve the future needs of all customers within their territory of service. See 48 Fed.Reg. at 24,332 n. 51, J.A. at 1267 n. 51. "Since utilities are affected with the public interest, their obligation to serve drives the need to construct facilities even when such investment may not be economic from the investors' standpoint." Id. at 24,340, J.A. at 1312. Beside explaining why utilities would plan and construct new facilities at a time when they are not earning their allowed rate of return, this obligation to provide adequate future capacity, in FERC's view, made it equitable to require ratepayers to share some of the risks associated with construction of new facilities. Id. at 24,337-38, J.A. at 1297.

Having found only a modest rate impact on wholesale customers viewed solely as consumers, FERC went on to consider the impact of the CWIP rule on those customers viewed as competitors of their wholesale suppliers in state-regulated retail markets. The most important potential effect on wholesale customers in their capacity as competitors is "price squeeze." A price squeeze occurs when "a utility's price for wholesale service is higher in relation to wholesale costs than is the utility's price for retail service in relation to retail costs, resulting in actual or potential impairment of the wholesale customer's ability to compete with the utility for that retail service." 48 Fed.Reg. at 24,345, J.A. at 1338-39.

Wholesale customers argued that a FERC rule allowing CWIP in rate base will necessarily cause a price squeeze in states that do not allow CWIP in rate base, because wholesale rates, which FERC regulates, will rise as a result of CWIP, while retail rates, which the states regulate, will not rise wherever CWIP is not allowed. Even if CWIP should be allowed in rate base, these commenters urged that utilities should not collect CWIP subject to refund, as with other rate base items, because customers subjected to a price squeeze could be irreparably harmed before litigation leading to an eventual refund ran its course. 48 Fed.Reg. at 24,338, J.A. at 1300-01.

FERC rejected the contention that allowing CWIP in rate base would necessarily result in price squeeze in states that employ an AFUDC-only rule in regulating retail rates. It asserted that "[n]o substantive evidence establishing the occurrence or likely occurrence of a price squeeze was submitted in this record." 48 Fed.Reg. at 24,346, J.A. at 1340. To the contrary, FERC's own analysis confirmed that allowing 50% of CWIP in rate base "does not generally create price discrimination where it does not otherwise exist." Id., J.A. at 1340.

The same analysis also stated, however, that in some cases 50% CWIP in rate base might bring about price discrimination where none previously existed or increase the magnitude of preexisting price discrimination. Federal Energy Regulatory Commission, Office of Regulatory Analysis, Environmental Assessment Related to FERC Rule Amendment Regarding Inclusion of CWIP in Rate Base at 4-41 (1983), J.A. at 1461. But, it went on to say, the risk that either of those effects "is likely to occur in a given situation cannot be predicated in advance without having detailed information regarding both wholesale and retail rates and costs." Id. Therefore, FERC determined that price squeeze issues relating to CWIP in rate base should be resolved on a case-by-case basis. But the Commission also warned that it does not believe "that the rule must operate to automatically deny CWIP simply because a price squeeze is found to exist." 48 Fed.Reg. at 23,436, J.A. at 1340. That belief, in turn, rests on a legal premise: the Commission's "stated policy not to remedy a price squeeze if it is due solely to state regulatory action or inaction." Id. at 23,436 n. 97, J.A. at 1340 n. 97. Given that legal premise, it follows that a state's refusal to allow CWIP in rate base cannot, standing alone, justify denial of CWIP as a remedy for price squeeze.

FERC also determined that CWIP should be collected subject to refund, "in order that relief be immediately available to companies with larger construction programs and concomitant capital needs and that cash flow rates achieve more stability over time." 48 Fed.Reg. at 24,339, J.A. at 1301. Thus, the Commission's decision on the refund question explicitly rests on its price signal and rate stability rationales for allowing CWIP in rate base. See id., J.A. at 1301-02.

FERC was also concerned that, in some instances, the immediate effects of allowing CWIP in rate base would include the very rate shocks it was hoping to reduce through the new rule. 48 Fed.Reg. at 24,436, J.A. at 1342. To minimize this risk, and to reduce any adverse impact on wholesale customers. FERC's final rule includes a provision limiting rate increases resulting from addition of CWIP to rate base to 6% per annum for the first two years after the effective date of the rule. Id., J.A. at 1342.

Petitioners raise both procedural and substantive challenges to Order No. 298. Their procedural claims are:

1. that FERC failed to provide adequate notice and opportunity to comment on the CWIP rule it ultimately adopted;

2. that FERC erroneously determined that it was not required to provide an environmental impact statement ("EIS") pursuant to section 102(C) of the National Environmental Policy Act ("NEPA"), 42 U.S.C. Sec. 4332(2)(C);

3. that FERC erroneously determined that it need not consider the impact of the new CWIP rule on "small entities" pursuant to the Regulatory Flexibility Act ("RFA"), 5 U.S.C. Secs. 601-612 (1982);

4. that FERC improperly relied on studies that were not published until after Order No. 298 was adopted.

We address these procedural claims in Part III.

Petitioners' substantive objections to the validity of the new CWIP rule can be summarized as follows:

1. advancing the public interest purposes FERC identified is beyond FERC's statutory authority;

2. the record does not support FERC's findings that its CWIP rule will advance those public interest purposes;

3. even if FERC's asserted goals would be served by its CWIP rule, the rule violates the used and useful principle and is therefore contrary to law;

4. the rule is inconsistent with other recent FERC decisions;

5. FERC erred in concluding that the rule would not cause price squeeze and in adopting the rule despite its anticompetitive effects; and

6. FERC's errors in connection with anticompetitive effects resulted in numerous other errors, including failure adequately to consider alternative proposals that would have avoided the anticompetitive effects of its CWIP rule.

We address the first four of these challenges in Part IV, and the last two in Part V.

At the request of the court, the parties submitted additional briefs addressing the question of whether these claims are ripe for review. The issue of ripeness turns on two criteria--"the fitness of the issues for judicial decision and the hardship to the parties of withholding court consideration." Abbott Laboratories v. Gardner, 387 U.S. 136, 149, 87 S. Ct. 1507, 1515, 18 L. Ed. 2d 681 (1967). The fitness requirement is essentially one of finality, see id. at 149-51, 87 S.Ct. at 1515-17, and there is little doubt that it is satisfied in this case. As in Abbott, "[t]he regulation challenged here, promulgated in a formal manner after announcement in the Federal Register and consideration of comments by interested parties, is quite clearly definitive. There is no hint that this regulation is informal, ... or only the ruling of a subordinate official, ... or tentative." Id. at 151, 87 S.Ct. at 1517 (footnote and citations omitted).

The hardship criterion is satisfied when "the impact of the regulations upon the petitioners is sufficiently direct and immediate as to render the issue appropriate for judicial review at this stage." 387 U.S. at 152, 87 S.Ct. at 1517. We think petitioners' case meets that requirement as well. FERC informs us that under Order No. 298 "the Commission must accept CWIP rate filings and the petitioners must pay CWIP rates under the rule." Brief for FERC on Ripeness Issue at 2. Although utilities are not obligated to file for increased CWIP, "rates will be filed--and must be paid--that include 50 percent of CWIP in rate base." Id. at 5. Indeed, many such rates have already been filed and are being paid by petitioners in this case. See Supplemental Brief of Intervenors-Respondents at 4-5, 6. It seems clear, then, that as to inclusion of CWIP in rate base "the impact of the administrative action could be said to be felt immediately by those subject to it in conducting their day-to-day affairs." Toilet Goods Ass'n v. Gardner, 387 U.S. 158, 164, 87 S. Ct. 1520, 1524, 18 L. Ed. 2d 697 (1967).

While petitioners could also challenge this rule in an actual rate proceeding at a later date, we do not think that makes this case unripe. Because of the CWIP rule, petitioners are paying higher rates. This distinguishes this case from our decision in South Carolina Electric & Gas v. ICC, 734 F.2d 1541 (D.C.Cir.1984). The petitioners in South Carolina Electric, as that opinion pointed out, might never have to pay a higher rate. Id. at 1546-47. Moreover, a challenge to a specific rate would not add to the concreteness of the controversy now before us. The size of a rate increase might be relevant to the issue of competitive harm, which we are not deciding, but would not alter our view of the procedural and substantive issues which we do decide. Finally, one of the prudential values underlying the ripeness doctrine is the protection of agencies from judicial interference with their discretion. In South Carolina Electric, the agency retained discretion not to allow the rule to lead to higher rates.3 Here the Commission states that its rule binds it to accept rates calculated with CWIP in the rate base; its discretion has been exercised and no longer exists. Reflecting this difference, the ICC in South Carolina Electric argued that the case was not ripe while FERC here contends that the case is ripe. We therefore conclude that the CWIP rule is ripe for review.

The standard of review applicable whenever we review orders of the Commission (whether issued after adjudication or rulemaking) is, under the Federal Power Act, the "substantial evidence" standard. See 16 U.S.C. Sec. 825 l (a) (1982). However, as we recently stated in Association of Data Processing Service Organizations v. Board of Governors, 745 F.2d 677, 686 (D.C.Cir.1984), where we review a rule issued after informal rulemaking under a review provision providing for substantial evidence review of "orders," it is ordinarily true that the " 'substantial evidence' requirement applicable to our review ... demands a quantum of factual support no different from that demanded by the substantial evidence provision of the APA, which is in turn no different from that demanded by the arbitrary or capricious standard." Indeed, we noted that our prior decisions construing the "substantial evidence" provisions of the Federal Power Act and the Natural Gas Act stand for the proposition that that provision "did not have the effect of requiring increased factual support beyond that demanded by the normal 'arbitrary or capricious' rulemaking standard of review." Id. at 686 (discussing Public Systems v. FERC, 606 F.2d 973, 980 n. 34 (D.C.Cir.1979), and American Public Gas Association v. FPC, 567 F.2d 1016, 1028-29 (D.C.Cir.1977), cert. denied, 435 U.S. 907, 98 S. Ct. 1456, 55 L. Ed. 2d 499 (1978)). Our duty under that standard of review, as we said in Weyerhaeuser Co. v. Costle, 590 F.2d 1011 (D.C.Cir.1978), is

to insist upon an explanation of the facts and policy concerns relied upon by the Agency in making its decisions; second, to see if those facts have some basis in the record; and finally, to decide whether those facts and those legislative considerations by themselves could lead a reasonable person to make the judgment that the Agency has made.

Id. at 1027 (citations omitted). In conducting that inquiry, however, "we keep in mind the Supreme Court's statement that ratemaking studies are not bound to 'any single formula or combination of formulas' and that if the Commission's order 'produces no arbitrary result' our inquiry is at an end." Public Systems v. FERC, 709 F.2d 73, 79 (D.C.Cir.), (quoting FPC v. Natural Gas Pipeline Co., 315 U.S. 575, 586, 62 S. Ct. 736, 743, 86 L. Ed. 1037 (1942)).

Petitioner Mid-Tex asserts that FERC's Notice of Proposed Rulemaking ("NPR") failed to give adequate notice that FERC might adopt a rule allowing a utility to include CWIP in rate base on the basis of an efficiency rationale rather than a financial hardship rationale. Thus, Mid-Tex frames its claim in terms of "the need to renotice the new rationale developed in the final rule," Brief for Mid-Tex at 30 (emphasis in original). We find no merit in this claim. The Commission's NPR specified a straight percentage CWIP rule as one of a number of alternatives that would be considered and on which comments were solicited. J.A. at 36. Moreover, the NPR stated that these were "alternatives of broader scope than financial distress," id. at 2, and referred both to a general efficiency rationale and to the effects of CWIP in rate base on cash flow, cost of capital, and rates. Id. at 30. The Commission's reasoning undoubtedly evolved significantly as a result of the notice and comment rulemaking. As FERC said, "[i]n the ensuing year and a half [since issuance of the NPR] ... we have gained a much better understanding of the effects of AFUDC treatment on utility financing, the investment bias created by inadequate returns and the unnecessary future costs which may be imposed by decisions reflecting that bias." 48 Fed.Reg. at 24,338, J.A. at 1299. But this does not establish that "the original notice did not adequately frame the subjects for discussion." Connecticut Light & Power Co. v. NRC, 673 F.2d 525, 533 (D.C.Cir.1982). Petitioners had actual notice that FERC might adopt the alternative it ultimately chose, and FERC's eventual rationale seems to us to "follow logically," Connecticut Light & Power, 673 F.2d at 533, from the NPR's announced concern with capital costs and minimizing the aggregate cost of new construction to consumers. Therefore, renoticing the rule is not warranted.

Mid-Tex's NEPA claim is that the Commission erroneously determined in its environmental assessment ("EA") that it need not prepare an environmental impact statement. Mid-Tex asserts that the Commission's finding that the CWIP rule would not significantly affect the environment is flawed because the EA itself demonstrates that the rule would cause a widespread increase in coal usage and consequently have a significant--and adverse--environmental impact.

In deciding whether an agency's decision not to prepare an environmental impact statement is arbitrary and capricious, this court ordinarily engages in four related inquiries:

(1) whether the agency took a "hard look" at the problem; (2) whether the agency identified the relevant areas of environmental concern; (3) as to the problems studied and identified, whether the agency made a convincing case that the impact was insignificant; and (4) if there was an impact of true significance, whether the agency convincingly established that changes in the project sufficiently reduced it to a minimum.

Sierra Club v. Peterson, 717 F.2d 1409, 1413 (D.C.Cir.1983); see also Kleppe v. Sierra Club, 427 U.S. 390, 410 n. 21, 96 S. Ct. 2718, 2730, n. 21, 49 L. Ed. 2d 576 (1976). Because Mid-Tex challenges only the validity of FERC's determination that the CWIP rule would not result in a significant increase in coal consumption for generation of electricity, we need focus only the third of the above inquiries. We find that FERC's predictive finding regarding coal usage is adequately supported by substantial evidence in the record.

Mid-Tex asserts that "Environmental Assessment Table 4.9 demonstrates the type of widespread coal usage resulting from the final CWIP rule." Brief for Mid-Tex at 36. Table 4.9, however, contains only generic information--it compares the typical atmospheric emissions from a 1000-MWe generating station of each of the five generating types. While Table 4.9 indicates that coal produces more atmospheric emissions than oil or gas, and that all three fossil fuels produce vastly more emissions per unit of energy than do nuclear or hydroelectric power, Table 4.9 reveals nothing about the effects of the CWIP rule.

However, Mid-Tex also asserts that "[t]he Commission is simply wrong in asserting that, by lowering demand for electricity, the new CWIP rule will yield a positive environmental impact." Brief for Mid-Tex at 36. Since the EA found that the CWIP rule would trigger a shift from oil and gas to coal, nuclear, or hydroelectric power, because the latter have the lowest operating costs, see J.A. at 1445, Mid-Tex concludes that FERC should have prepared an EIS to examine the increase in atmospheric emissions and other adverse environmental effects of the shift to coal.

Mid-Tex challenges neither the assumptions nor the findings contained in the EA, and those findings refute its argument. The EA found that the CWIP rule would cause a drop in consumption of electricity that would more than offset the effect of a shift away from oil and gas to coal. As Table 4.7 shows, the EA found that, compared to the nationwide base case, a 50% CWIP rule would reduce the amount of electricity generated by coal by about 1.03% in 1985 and by about .61% in 1990. J.A. at 1445. For 1995 the base case and the 50% CWIP rule would produce virtually identical amounts of coal-generated electricity. See id. In some regions, the EA found that coal usage would increase, while in others it would decrease. But, according to Table 4.8, the largest increase in regional coal usage was a one-half of 1% increase in the Northwest in 1995. We need not decide whether the Commission could completely ignore regional increases in coal usage on the grounds that those increases were offset by regional decreases elsewhere, because we agree with the EA that "[a]llowing for some range of error in the ... models and assumptions, these small impacts may not be reliably distinguished from no (or zero) impacts." Id. at 1446. Nor, given the pattern of declining overall consumption for each type of fossil fuel predicted under the CWIP rule, do we see any basis for disputing the EA's conclusion that the rule's environmental impact would be negligible but predominantly beneficial. See id. at 1452.

Petitioners Mid-Tex and Public Systems argue that FERC failed to comply with the Regulatory Flexibility Act ("RFA"), 5 U.S.C. Secs. 601-612 (1982). The RFA provides that

all agencies, as part of the rulemaking process, must conduct a "regulatory flexibility analysis" for any rule that has a "significant economic impact on a substantial number of small entities." 5 U.S.C. Sec. 605(b). The flexibility analysis must, among other things, discuss how a rule will affect small entities, describe "significant alternatives" that would "minimize any significant economic impact of the rule on small entities," and explain "why each one of such alternatives was rejected." Id. Sec. 604(a)(3) (describing the "final regulatory flexibility analysis" to be issued with the final rule); see id. Sec. 603 (describing the "initial regulatory flexibility analysis" to be issued with the proposed rule).

Small Refiner Lead Phase-Down Task Force v. EPA, 705 F.2d 506, 537 (D.C.Cir.1983) ("SRTF" ).

Section 611(b) of the RFA expressly precludes judicial review of agency compliance with these requirements, but goes on to say that "[w]hen an action for judicial review of a rule is instituted, any regulatory flexibility analysis for such rule shall constitute part of the whole record of agency action in connection with the review." In SRTF, this court, after examining the legislative history of the RFA, concluded that Congress intended to preclude judicial review of regulatory flexibility analyses except as part of judicial review of the underlying rule. 705 F.2d at 537-38. We accordingly held that "a reviewing court should consider the regulatory flexibility analysis as part of its overall judgment whether a rule is reasonable and may, in an appropriate case, strike down a rule because of a defect in the flexibility analysis." Id. at 539. We emphasized, however, that "a major error in the regulatory flexibility analysis may be, but does not have to be, grounds for overturning a rule." Id. at 538.

Section 605(b) of the RFA provides that an agency is not required to prepare either an initial or a final regulatory flexibility analysis "if the head of the agency certifies that the rule will not, if promulgated, have a significant economic impact on a substantial number of small entities." The certifying agency must publish the certification in its NPR and include "a succinct statement explaining the reasons for such certification." 5 U.S.C. Sec. 605(b) (1982). In its NPR, the Commission certified that its proposed rule would not have a significant economic impact on a substantial number of small entities. J.A. at 42-43. FERC explained that virtually all of the utilities it regulates do not fall within the meaning of the term "small entities" as defined in the RFA. Petitioners do not contest the accuracy of this explanation.

Instead, petitioners argue that FERC's determination that the utilities it regulates are not "small entities" is legally insufficient to satisfy the certification requirement. Petitioners say that an agency must consider whether its proposed rule will have a significant economic impact on wholesale customers as well as regulated utilities. Before the Commission, the Small Business Administration ("SBA") advocated a similar interpretation of the RFA, arguing that FERC "should have considered the impact of the proposed rule on wholesale and retail customers of the jurisdictional entities subject to rate regulation by the Commission." 48 Fed.Reg. at 24,350, J.A. at 1360. Indeed, SBA maintained that FERC should also consider the possible impact on ultimate retail electric consumers, many of which are small businesses, in the event that state regulatory authorities elect to follow FERC's new CWIP policy. FERC responded that "the RFA does not require the Commission to consider the effect of this rule, a federal rate standard, on non-jurisdictional entities whose rates are not subject to the rule. The Commission, therefore, is not under a statutory obligation to study the impact of the CWIP rule upon wholesale or retail customers of any jurisdictional utility or upon other ultimate retail electric consumers." Id. at 24,350, J.A. at 1361. FERC adheres to that interpretation in this court.

FERC's case in support of its interpretation of the RFA relies on the language of the statute and its legislative history. FERC points out that the RFA's Congressional Findings and Declaration of Purpose state in part that

(2) laws and regulations designed for application to large scale entities have been applied uniformly to small businesses ...;

(3) uniform Federal regulatory and reporting requirements have in numerous instances imposed unnecessary and disproportionately burdensome demands including legal, accounting and consulting costs upon small businesses ...;

(4) the failure to recognize differences in the scale and resources of regulated entities has in numerous instances [produced undesirable economic effects];

* * *

(6) the practice of treating all regulated business ... as equivalent may lead to inefficient use of regulatory agency resources....

Act of Sept. 19, 1980, Pub.L. No. 96-354, Sec. 2, 94 Stat. 1164 (emphasis added). Having made these findings, Congress declared that the purpose of the RFA was to require agencies to endeavor, "consistent with the objectives of the rule and of applicable statutes, to fit regulatory and informational requirements to the scale of the businesses ... subject to regulation." 94 Stat. 1165 (emphasis added). According to FERC, this evidence demonstrates that Congress intended agencies to limit their consideration of the economic impact of proposed rules to small entities that would be directly regulated by those rules.

Some language in the Senate Report supports this interpretation of the RFA. The Report referred to the problem Congress perceived in terms of the tendency of "regulations of general and uniform applicability" to place "disproportionate burden[s] upon small businesses." S.Rep. No. 878, 96th Cong., 2d Sess. 3, reprinted in 1980 U.S.Code Cong. & Ad.News 2788, 2790. And the Report described the aim of the bill as ensuring that agency rules will be framed taking into account the size and nature of the regulated businesses. Id. at 2. Moreover, the instances of praiseworthy regulatory flexibility cited in the Senate Report confirm that Congress was primarily concerned about the high costs of compliance with regulations by small businesses bound to conform their conduct to those regulations. See id. at 6-7. Congress viewed those costs as relatively more burdensome to small businesses than their larger counterparts, and believed this disproportion could have anticompetitive effects.

In response, petitioners claim that the RFA is intended to apply to all rules that affect small entities, whether the small entities are directly regulated or not. Mid-Tex notes that the RFA refers, without limitation, to "any rule which the agency expects to propose or promulgate which is likely to have a significant economic impact on a substantial number of small entities." 5 U.S.C. Sec. 602(a)(1) (emphasis added). Since the RFA does not distinguish between the direct "impact" of a regulation on regulated small entities and the indirect "impact" of a regulation on small entities that do business with or are otherwise dependent on the regulated entities, petitioners find no basis in the RFA for the limitation on which FERC relies. In support of their analysis, petitioners refer to a section-by-section analysis of the RFA by Senator Culver--an analysis to which this court gave substantial weight in SRTF, see 705 F.2d at 538--suggesting that in determining the impact of regulations on small entities agencies should broadly evaluate the direct and indirect effects of the proposed regulation. 126 Cong.Rec. 21,458-59 (1980). Finally, petitioners urge us to give weight to the fact that the SBA argued in favor of the broader reading of the RFA before the Commission, because the SBA is charged with monitoring agency compliance with the RFA. See 5 U.S.C. Sec. 612.

We think FERC's interpretation of the RFA is well founded. The problem Congress stated it discerned was the high cost to small entities of compliance with uniform regulations, and the remedy Congress fashioned--careful consideration of those costs in regulatory flexibility analyses--is accordingly limited to small entities subject to the proposed regulation. We find a clear indication of this limitation in section 603 of the statute, which specifies the contents of initial regulatory flexibility analysis. These initial analyses are to include "a description of and, where feasible, an estimate of the number of small entities to which the proposed rule will apply," 5 U.S.C. Sec. 603(b)(3) (emphasis added), and "a description of the projected reporting, recordkeeping and other compliance requirements of the proposed rule, including an estimate of the classes of small entities which will be subject to the requirement." 5 U.S.C. Sec. 603(b)(4) (emphasis added). Since the scope of the final regulatory flexibility analysis is limited to the issues raised by the initial analysis, it is clear that Congress envisioned that the relevant "economic impact" was the impact of compliance with the proposed rule on regulated small entities. Reading section 605 in light of section 603, we conclude that an agency may properly certify that no regulatory flexibility analysis is necessary when it determines that the rule will not have a significant economic impact on a substantial number of small entities that are subject to the requirements of the rule.

Petitioners' arguments do not persuade us that this reading of the Act, which comports with Congress' declared purposes, is wrong. Indeed, FERC clearly has the better of the dispute over the appropriate inference to be drawn from the legislative history of the RFA. For the legislative history cited by FERC, which shows that the costs of compliance with uniform regulations to small businesses were the focus of congressional concern, also gives rise to an inference that Congress did not intend to require that every agency consider every indirect effect that any regulation might have on small businesses in any stratum of the national economy. That is a very broad and ambitious agenda, and we think that Congress is unlikely to have embarked on such a course without airing the matter. The only statement petitioners point to from the legislative history would not persuade us otherwise even if it were not ambiguous: concern with "the direct and indirect effects of the proposed regulation" may very well mean concern with effects, direct or indirect, on the regulated entities. Accordingly, we hold that FERC correctly determined that it need not prepare a regulatory flexibility analysis in connection with its proposed CWIP rule.

Petitioner Cities asserts that Order No. 298 unlawfully relies on a study prepared by the Energy Information Institute ("EIA"), and that the Commission's EA also unlawfully relied on the EIA study and on studies prepared by the Argonne National Laboratory and by Southwest Energy Associates. The gravamen of Cities' claim is that none of these three studies is part of the record or was made available for comment. Hence, Cities says that this case falls within the condemnation of our statement in Connecticut Light & Power, 673 F.2d at 530-31, that "[a]n agency commits serious procedural error when it fails to reveal portions of the technical basis for a proposed rule in time to allow for meaningful commentary."

FERC's brief does not respond to Cities' argument. However, FERC did address this alleged defect in its order on rehearing. It stated that "[a]ny non-record studies that the Commission considered in reaching its decision complemented and supported the findings of the on-the-record studies and the Commission's policy analysis contained in the final rule." 48 Fed.Reg. at 46,017, J.A. at 1801. Specifically, FERC said, it relied on the EIA study only as support for the finding of its own study that the disparity between CWIP-based and AFUDC-based rates would be "greatest in the early years and diminish over time." Id., J.A. at 1802. Hence, "[w]ith regard to the initial rate impacts, primary reliance is placed by the Commission on the FERC study which was issued for public comment and which forms part of the rulemaking record. There is thus an adequate basis for the rule without the EIA study." Id., J.A. at 1802.

We think there is no need for us to determine the extent of FERC's reliance on the EIA study, because we think FERC was entitled to rely on that study. Cities conveniently neglects to mention that the EA expressly noted that "the following documents are available in the office of Public Information (Room 1000, 825 North Capitol Street, N.E., Washington, D.C. 20426): (a) Results from the EIA study; (b) EIA's 1981 Annual Report to Congress (four volumes); (c) User manuals for the [National Utility Financial Statement] model used in the EIA study (three volumes); (d) the FERC computer model; and (e) Results from the FERC study." J.A. at 1424 n *. It seems certain that the documents listed as (a), (b) and (c) above supply all the information referred to by FERC as the EIA study. Certainly Cities has not suggested any respect in which they do not. Hence the EA clearly identified the relevant materials and supplied notice of their availability in a public document room.

Since "[t]he APA does not prescribe any procedure for assembling a 'record' for judicial review of informal rulemaking," SRTF, 705 F.2d at 519, we have in the past sustained an agency's reliance on materials available in a public documents room, so long as interested persons had good reason to believe those documents were relevant to the rulemaking proceedings. See Connecticut Light & Power, 673 F.2d at 531 & n. 7. Indeed, in Connecticut Light & Power we did so notwithstanding the fact that those documents were not specifically identified by the agency "until long after the comment period had closed." Id. at 531. Here, by contrast, FERC specifically identified the EIA study in its EA, stated the manner in which that study could be obtained, and noted that the EA extensively relied on the EIA study. See J.A. at 1423-24. Since the EA was made public at about the same time as Order No. 298, interested persons had some thirty days in which to obtain and review the EIA study and apprise FERC of their objections to that study. There is no procedural infirmity here.

As for Cities' objection to the EA's use of the Argonne and Southwest Energy studies, we think it evident that those studies played little if any part either in the development of the EA or in the formulation of the final rule. Cities has not directed us to any finding or conclusion in the EA or in Order No. 298 that purports to be based on these studies even in part. Hence any alleged procedural error is manifestly harmless.

Petitioner Alabama Electric flatly asserts that the Commission's public interest rationale for allowing CWIP in rate base is beyond its statutory authority. Brief for Alabama Electric at 9. Alabama Electric correctly notes that the Federal Power Act denies the Commission the power to order a utility to enlarge its generating facilities for the purpose of selling or exchanging electric energy with another entity. See 16 U.S.C. Sec. 824a(b) (1982). Alabama Electric characterizes the Commission's stated objective of removing impediments to capital investment in needed facilities as an attempt to achieve by indirection what the Federal Power Act prohibits the Commission from doing directly--compelling the enlargement of generating facilities.

The problem with this argument, of course, is the false equivalence it posits between a Commission order directing a utility to add capacity to serve a wholesale customer and a Commission regulation addressing the timing of recovery by the utility of costs that indisputably belong in rate base at one time or another. There is no compulsion here. Accepting, as we must for purposes of this analysis, that a CWIP-in-rate-base rule will have the effects the Commission attributes to it, the most that can be said is that the CWIP rule reduces certain disincentives to add new generating facilities. Accordingly, we reject Alabama Electric's claim.

We have searched the briefs of the other petitioners in these cases to determine whether they raise any other challenge to the Commission's statutory authority to invoke the various public interest considerations on which the CWIP rule is based. We have found no suggestion that, standing alone, the need for an adequate and reliable supply of electricity, or the need for least total cost facilities, are impermissible criteria for the Commission to employ in rulemaking. Rather, petitioners contest the extent of FERC's reliance on these considerations at the expense of other criteria which they believe FERC is obliged to weigh in the balance. Accordingly, we shall accept the first two purposes FERC announced for its CWIP rule--mitigating bias against capital investment and ensuring more accurate price signals--as valid regulatory objectives. Our acceptance does not foreclose considerations of the extent to which those goals may override other regulatory objectives. We take up that question in Part V infra, in connection with some of petitioners' other objections.

Petitioner Public Systems, however, asserts that FERC's third purpose--rate stability--"plays an important role in the Commission's decision, but the legitimacy of that role is extremely questionable." Brief for Public Systems at 66. Public Systems correctly notes that the Federal Power Act requires that rates be "just and reasonable," without requiring that they be stable, and asserts that "the Commission had an obligation to explain why the new-found goal of rate stability was given any weight in its decision." Brief for Public Systems at 68.

We think FERC has reasonably determined that "[r]atepayers should not face rates that fluctuate radically according to the completion of base load units," 48 Fed.Reg. at 46,019, J.A. at 1812, and that this concern is not inconsistent on its face with FERC's mandate to ensure that rates be just and reasonable. The Commission suggests rate shock can have "adverse effects on consumers," id., J.A. at 1813, and we think the Act's command that rates be "just" as well as "reasonable" gives FERC discretion to give at least some weight to the goal of protecting consumers "from financial shock or economic dislocation" as a result of precipitous and unanticipated increases in rates. 48 Fed.Reg. at 24,346, J.A. at 1342.

Petitioner Public Systems attacks FERC's two principal justifications for allowing CWIP in rate base on the grounds that each undercuts the other. The existence of a bias against capital investment, Public Systems says, suggests that absent CWIP utilities will not build sufficient capacity to serve future needs. The understatement of that future cost of electricity in current price signals, on the other hand, suggests that, without CWIP, utilities will build excess capacity. "What," Public Systems asks, "is wrong with a regulatory scheme that leads utilities to underprepare for exaggerated anticipated future demand?" Brief for Public Systems at 53.

This argument rests on an oversimplification of FERC's analysis. Consider the case of a utility that begins construction of a plant that is somewhat too large due to overprediction of demand, and subsequently cancels the plant due to inability to meet mounting financing costs. In this example, the two effects FERC has identified skew the utility's decisionmaking at different times and, far from "offsetting" one another, these effects reinforce one another. Under the CWIP rule, FERC predicts that the utility will construct a smaller plant and will be more likely to complete it. That result is both more efficient and more adequate to meet future demand for electricity.

Or consider a utility that decides to build a facility that is somewhat cheaper to construct but far more expensive to operate than another facility of equal capacity. That decision is inefficient, and the inefficiency persists regardless of whether the two facilities are larger than necessary due to imperfect price signals. Conversely, the prosperous utility may have no trouble financing any plant it thinks needed, yet without CWIP it will tend to overbuild.

Even in instances where the two effects somewhat offset one another, moreover, Public Systems asks us to assume that the two effects are equal as well as opposite. That assumption is quite implausible, and presumably unverifiable. We think that while in a particular case underpreparation for overpredicted demand may yield the same results as the CWIP rule is said to produce, that result is a fortuity rather than a necessity. Accordingly, we reject the contention that FERC's rationale is truly at war with itself.

Petitioners contend that even if the CWIP rule will achieve the purposes FERC advances, and even if FERC can legitimately pursue those goals, the used and useful principle precludes inclusion of CWIP in rate base. According to petitioners, CWIP in rate base violates that principle because "[u]tility plant under construction is not 'used and useful;' hence, the costs of that plant typically may not be assessed against current ratepayers. It is only when the plant goes into commercial operation that a legitimate claim may be made against ratepayers for financial support." Brief for Mid-Tex at 6.

FERC responds that it reasonably determined that utility customers derive a present benefit--assurance of adequate future service--from construction work in progress. Furthermore, FERC argues that our decision in Goodman v. Public Service Commission, 497 F.2d 661 (D.C.Cir.1974), which endorsed the proposition that "[f]unds are not necessarily 'used and useful' only when they are currently invested in completed plants," id. at 668 (emphasis added), supports affirmance of this aspect of Order No. 298. We agree.

In Goodman, the Public Service Commission of the District of Columbia ("PSC") granted Potomac Electric Power Company ("PEPCO") a rate increase based on a systemwide rate base that included CWIP. The Goodman court followed precedent holding that PSC's "authority to regulate rates is identical to the authority of the Federal Power Commission to set rates under the Natural Gas Act." 497 F.2d at 665. That makes Goodman directly relevant here, because FERC's authority to regulate under the Federal Power Act is in all relevant respects identical to its authority under the Natural Gas Act. See, e.g., Arkansas Louisiana Gas Co. v. Hall, 453 U.S. 571, 577 n. 7, 101 S. Ct. 2925, 2930, n. 7, 69 L. Ed. 2d 856 (1981).

The petitioner in Goodman argued that "if plant under construction is included in the rate base, the ratepayers will be forced to assume a responsibility to pay for property which does not now benefit them and that this duty properly should fall on those who invest in the utility." 497 F.2d at 667. That argument combines the arguments petitioners raised before the Commission in this proceeding--that CWIP violates the used and useful principle both in terms of intergenerational equity and by shifting risk from investors to ratepayers.

This court upheld PSC's choice of the CWIP method as one reasonable alternative among several. We said that "[s]o long as capitalized interest is not included in the rate base, the inclusion of plant under construction provides no excessive compensation to the utility." 497 F.2d at 668. We then found PSC's reason for choosing CWIP over AFUDC sufficient: that reason was that funds invested in construction work in progress " 'are being used for the benefit of the public just as much as funds invested in plant in service,' " especially where the record established " 'a continuing need for permanently financing a large construction program.' " Id. (citation omitted). As already noted, we suggested that the benefit to the public from current investment could be seen as bringing CWIP within the ambit of the used and useful principle and noted that there was some support for the theory that the CWIP method "is, in the long term, the least expensive to the ratepayers." Id. That, we commented, would make the CWIP method "a desirable alternative," and in any event we saw no legal reason why the CWIP method could not be used so long as it achieved a reasonable result with regard to the rate under review. Id. at 668-69.

Petitioners argue that Goodman is inapposite for several reasons. First, they assert that the fact that FERC regulates rates for sale to distributors of electricity implies that allowing CWIP in rate base has anticompetitive effects that were absent in Goodman. Since courts have repeatedly held that FERC has an obligation to assess the anticompetitive effects, if any, of its orders and regulations, petitioners suggest that the used and useful principle is violated here although it was not in Goodman.

This argument, we think, properly goes not to whether FERC violated the used and useful principle but to whether FERC gave adequate consideration to the alleged anticompetitive effects of allowing CWIP in jurisdictional rate base. Accordingly, we postpone analysis of this objection until Part V, where we consider it in connection with petitioners' price squeeze claims.

Next, petitioners correctly note that the Goodman court alluded to the fact that CWIP has been allowed in PEPCO's rate base for some twenty years. But the Goodman court did not give decisive weight to this feature of that case, and we see no reason to strip Goodman of its persuasive force on that account. The Goodman court perceived CWIP in rate base as a reasonable alternative to the AFUDC method that is consistent with the used and useful principle. That rationale clearly would have justified affirmance even if the orders under review in Goodman had been the first to allow CWIP in PEPCO's rate base.

Petitioners also ask us to distinguish Goodman on the grounds that it involved a single rate case, in which the utility's needs could be balanced against the impact of CWIP in rate base on customers, whereas Order No. 298 applies to hundreds of utilities without regard to need or the extent of adverse impact on their customers. But that difference is inevitable given FERC's unquestioned authority to decide the CWIP issue by informal rulemaking rather than adjudication. Beyond that, there was no finding in Goodman that the PSC had engaged in an explicit balancing of the needs of the utility and the impact on ratepayers. Hence, we cannot distinguish Goodman on this basis.

Finally, petitioners dispute FERC's claim on brief that there is a nationwide need for new utility construction that is analogous to the finding, to which the Goodman court gave some weight, that the record there established a continuing need for financing a large construction program. See Brief for FERC at 20. Yet petitioners' contention is inapposite insofar as FERC did not rest its conclusion on such a claim.

FERC stated that "the present rule is not predicated on an assessment that enormous new capacity is necessarily required," 48 Fed.Reg. at 24,334, J.A. at 1281, and purported to "express no views on the future growth of the demand for electricity, or on the question whether new generating facilities may be economically justified even in the absence of demand growth." Id. at 24,330, J.A. at 1257. Both statements were made to highlight FERC's overall purpose--"to create a regulatory framework within which utilities can make an unbiased assessment of the need for new capacity and of the best means of meeting that end." Id., J.A. at 1257.

And yet neither statement can fairly be read as a refusal to predict that utilities will continue to find it necessary to engage in large, long-term construction projects. To begin with, old plants will continue to wear out even if demand stays level. Moreover, the principal uncertainty FERC identified as to future demand was not whether demand would continue to increase but whether "demand will grow faster than the economy." 48 Fed.Reg. at 24,330, J.A. at 1261. In fact, FERC stated that "realistic current estimates" of growth in demand for electricity are "in the neighborhood of three percent [per annum]." Id. at 24,328 n. 28, J.A. at 1251 n. 28. We think it clear, then, that FERC anticipated a continuing need for financing large construction programs on the part of many if not most utilities. We do not read Goodman as requiring a finding to that effect, but in any event, given that we deal in this case with rulemaking rather than adjudication, we think the analogous finding is apparent throughout FERC's decision. We hold, therefore, following Goodman, that FERC's decision to allow CWIP in rate base is consistent with the used and useful principle.

We turn now to petitioners' challenges to the adequacy of the record evidence supporting several of FERC's key findings in Order No. 298. Our task as a reviewing court is unfortunately made much more difficult by FERC's tendency, in the lengthy text of Order No. 298, to discuss the views of commenters without citations. To make matters worse, the Commission's brief pays very little attention to the evidentiary basis for FERC's various findings. Yet those findings are no less critical to Order No. 298 than is the Commission's economic analysis or its legal reasoning.

Petitioner Cities asserts that "there is no evidence in the record which can lawfully support the conclusion that the Commission's jurisdictional ratemaking creates a bias against the maintenance of reliable service at the lowest reasonable costs." Brief for Cities at 26. Cities' argument runs as follows: the bias against capital investment, according to FERC's own reasoning, can occur only if the utility is earning a return below its cost of capital, or if the utility has insufficient cash flow. As to the first source of bias, then, "it was necessary for the Commission to find that utilities are not able to earn their cost of equity." Id. at 21. Cities recognizes that FERC did make a finding that utilities were not able to earn their cost of equity but attacks that finding as insufficient because the data on which FERC relied were based on total company earnings, rather than on that segment of company earnings attributable to sales within FERC's jurisdiction. Thus, unless it was reasonable for FERC to assume that a utility's overall inability to earn its allowed rate of return generally traces both to its retail business and its FERC jurisdictional business, FERC's finding does not rest on substantial evidence. Cities maintains that FERC's assumption was clearly unreasonable, because Commission practice--allowing one-day suspensions on filed rates, employing a forecasted test year period for determining rate base, and allowing "pancaked" rate filings--"virtually permits utilities the total power to determine whether they will earn their cost of equity on their jurisdictional business." Id. at 22.

FERC addressed this argument on rehearing. Without denying that there might be some validity to the position that "the primary cause of any bias against capital investment attributable to earnings attrition is ... inadequate retail rate regulation by the various state commission," 48 Fed.Reg. at 46,017, J.A. at 1803, FERC stated that its review of the comments showed that "at least on the basis of presenting available information and analysis, it cannot be shown that wholesale rates are immune from earnings attrition." Id., J.A. at 1804. FERC conceded that further study was called for on this point, and indicated that its staff is now engaged in such a study. Cities urges that we should therefore remand to the Commission to defer the effectiveness of Order No. 298 until this study is completed and subjected to comment by interested persons.

We think remand is uncalled for. Cities has not pointed to anything in comments submitted to FERC that would have allowed a reliable causal analysis of earnings attrition along state and federal jurisdictional lines. One obviously relevant component of that inquiry is the extent to which the Commission practices Cities points to are or are not followed by state regulatory commissions. Cities has not even substantiated its claim by presenting such a comparison. Thus we cannot say that FERC's working hypothesis--that unchallenged systemwide earnings attrition is attributable both to the state- and federally-regulated branches of a utility's business--is unreasonable.

Nor is FERC's willingness to commission further study of this question a basis for remand. The rulemaking process cannot be an endless one. FERC sees a need to intervene now on the basis of what it does know, and we think it outside our competence to second-guess its judgment on that point. This is particularly true because there is another basis for the general finding that the AFUDC method creates a bias against capital investment.

Cities challenges the evidentiary support for that basis--insufficient cash flow--as well. According to Cities, cash flow is utterly irrelevant to a utility's decision about whether to build a more or less capital-intensive plant, so long as the utility can market new stock at book value. That is because the ability to market new stock at book value ensures that the utility can raise the capital needed for new construction without diluting the value of the stock held by current investors. And, Cities asserts that any utility that is earning its cost of capital will be able to market stock at book value. Consequently, there is no possibility, and "no evidence, that utilities earning their costs of equity have difficulty in financing or are encouraged to build less capital-intensive facilities." Brief for Cities at 26.

FERC contends, to the contrary, that "even if earnings attrition is controlled, the utility faced with a cash flow problem, will be encouraged to favor less capital-intensive facilities." 48 Fed.Reg. at 46,017, J.A. at 1806. The crux of FERC's position is the claim, amply supported by comments in the record, that AFUDC earnings are viewed by investors as low-quality earnings because they are non-cash earnings. See id., J.A. at 1805. FERC discussed, and generally credited, numerous studies that detailed the adverse effects this discounting of AFUDC earnings tends to have on a utility's ability to finance new construction. See 48 Fed.Reg. at 24,339-40, J.A. at 1303-10. Specifically, FERC found that a high proportion of AFUDC earnings tends to reduce a utility's interest coverage ratio (earnings before taxes and interest expenses divided by interest expenses). That ratio "is critical to financing because trust indenture provisions applicable to the utilities commonly prohibit the issuance of new long-term debt unless the interest coverage ratio is maintained at or above a specified minimum level, typically 2.0." Id. at 24,339 n. 65, J.A. at 1303 n. 65. Hence the discounting of AFUDC earnings can result in a situation in which the utility is unable to finance through new debt offerings. FERC views that cost--the loss of flexibility in financing--as giving rise to a bias against capital investment even if the utility is earning its allowed rate of return. And there was reason to believe this problem had been growing worse in recent years, because the percentage of utility earnings represented by AFUDC "paper earnings" apparently quadrupled, from 11.8% to 46.3%, from 1970 to 1980. Id. at 24,333, J.A. at 1272.

Moreover, FERC credited comments suggesting that the discounting of AFUDC earnings by investors tends to increase utilities' need for external financing. See 48 Fed.Reg. at 24,339, J.A. at 1305. That, too, can reasonably be seen as a cost to the utility, and hence a source of bias against capital investment. In this connection, we may note that while Cities' argument posits capital markets that can absorb all-equity financing by a multitude of utilities, FERC accepted as accurate the depiction by several commenters of the "growing capital requirements of the electric industry in a contracting capital market." Id. at 24,333, J.A. at 1272. We think, then, that FERC reasonably concluded that poor cash flow attributable to AFUDC could bias utilities against optimal capital investment in new facilities.

Petitioner Cities maintains that Order No. 298 interprets the used and useful principle in a manner that is totally inconsistent with FERC's rule on tax normalization, which we upheld in Public Systems v. FERC, 709 F.2d 73 (D.C.Cir.1983) (Public Systems II ). Tax normalization is one of two alternative methods for determining a utility's proper tax allowance when ratemaking rules "require that a current utility expense not be borne entirely by current ratepayers but rather be changed to customers over time, while tax rules ... permit the utility to deduct the entire expense in the current year." Public Systems II, 709 F.2d at 75. Under the normalization method, which is now FERC's general policy, the benefit of that immediate tax deduction is spread over the period during which ratepayers are charged for the expense under ratemaking rules. Under the flow-through method, which was formerly favored by FERC, the entire tax deduction is used to reduce rates for the year in which the deduction is taken.

The Commission's primary justification for its decision to adopt tax normalization was "the matching principle: as a matter of fairness, customers who pay an expense should get the tax benefit that accompanies the expense.... To do otherwise would subsidize present customers at the expense of future ones." Public Systems II, 709 F.2d at 80 (footnote and citation omitted). In turn,[u]nderlying the matching principle is the concept of "used and useful" property. Property or equipment can usually be included in the rate base only if the property is providing service to current ratepayers. For example, construction work in progress is generally not included in rate base. Normalization is consistent with the used and useful property concept in that it allocates the tax benefits of expenses to the periods when ratemaking policy recognizes the expenses.

Id. at 80 (footnotes omitted).

Cities regards this rationale for tax normalization as entirely inconsistent with allowing CWIP in rate base. We disagree. In the tax allowance context, it was undisputed that ratemaking policy properly recognized certain expenses (of which we listed CWIP as but one of several examples) at a different time or over a longer time than tax policy recognized those expenses. Consequently, there was no dispute over when ratepayers bore the burden of those expenses. The matching principle was thus invoked to support matching the benefits attributable to tax policy (deduction of those expenses) with the burden of the associated expenses. As applied to the orders under review, the matching principle would imply only that since some CWIP will now be allowed in rate base, the deductions associated with that CWIP must now be passed through to the ratepayers who are currently paying CWIP carrying charges. That result follows from the normalization method itself, and does not demonstrate in any way that allowing CWIP in rate base is inconsistent with tax normalization.

Of course, we also gave some credence in Public Systems II to FERC's claim that the matching principle is a means of giving effect to the used and useful principle. But that claim, properly understood, was only that prior determinations as to the proper ratemaking treatment of various expenses--determinations based in part on an interpretation of the used and useful principle--would be undercut to some extent by the flow-through method, because ratepayers paying a charge that had previously been determined to relate to used and useful property would not be given the benefit of tax deductions associated with that property. In giving that argument some weight in Public Systems II, we certainly did not suggest that we believed FERC's discretion to interpret the used and useful principle in the public interest had thereby been extinguished. Indeed, we specifically noted that FERC had "voted in principle" to adopt a final rule allowing 50% CWIP in rate base. See 709 F.2d at 75 n. 2.

We therefore conclude that Order No. 298 is not inconsistent with the rationale underlying FERC's tax normalization policy. FERC has determined that construction work in progress benefits both current and future ratepayers, and accordingly that allowing CWIP in rate base does not violate the used and useful principle. We have already upheld these determinations as reasonable. It follows that this is not a case in which FERC is violating the matching principle by allowing part of the CWIP carrying charges to be included in rate base when those charges are incurred by the utility.

Petitioner Mid-Tex argues that FERC's price signal rationale is flawed because the improved fuel efficiency of future generating plants will generally offset higher capital costs during operation. Since the savings resulting from improved fuel efficiency will not be reflected in rates while new plant is under construction, while under the new CWIP rule higher capital costs will be reflected in current prices, "CWIP could yield rates in excess of the unit cost of future generation and thus send inaccurate price signals to consumers." Brief for Mid-Tex at 42.

FERC, however, believes that "[t]oday the addition of a new facility is more likely to mean higher prices than lower ones." 48 Fed.Reg. at 24,331, J.A. at 1262. As the examples of rate shock FERC relied on demonstrate, see id. at 24,331 n. 47, J.A. at 1263 n. 47, there was a reasonable basis for this belief. See also 48 Fed.Reg. at 46,019 n. 31, J.A. at 1812 n. 31. While we would agree with petitioner that greater attention to the relative magnitude of fuel savings and capital cost increases would have put the Commission's price signal rationale on a sounder footing, we are not prepared to hold that this aspect of Order No. 298 is irrational. Mid-Tex's general assertion that fuel savings can outweigh capital cost increases may well be correct; but Mid-Tex does not dispute the proposition that "[c]apacity costs in real terms have been increasing over the last decade." 48 Fed.Reg. at 24,345, J.A. at 1338. Hence Mid-Tex's argument shows only that the price signal rationale will not support the CWIP rule in each particular instance in which that rule may be applied. That does not make the rule, or the price signal rationale, irrational. It is enough that the rule will generally have the effect on price signals that FERC envisions.

We now address petitioners' claims concerning the anticompetitive effects of the new CWIP rule. In Part V-A, we consider the price squeeze issue; in Part V-B, we consider the "double whammy" issue; and in Part V-C, we consider the contention that FERC erroneously rejected alternative rules that would have achieved the Commission's stated purposes without anticompetitive effects or as severe an adverse impact on wholesale customers.

In FPC v. Conway Corp., 426 U.S. 271, 96 S. Ct. 1999, 48 L. Ed. 2d 626 (1976), the Supreme Court held that the Commission has jurisdiction to consider the allegations of a utility's wholesale customers that proposed wholesale rates are discriminatory and anticompetitive in relation to the utility's state-regulated retail rates. The Conway Court rejected the Commission's position that "a just and reasonable wholesale rate can never be a contributing factor to an undue discrimination," because it rejected the underlying assumption that "there is only one level at which a wholesale rate can be said to be just and reasonable and that any attempt to remedy a discrimination would always result in an unjustly low rate that would fail to recover fully allocated wholesale costs." Id. at 277-78, 96 S. Ct. 2003-04. Rather, "there is no single cost-recovering rate, but a zone of reasonableness." Id. at 278, 96 S.Ct. at 2004. Thus, the Court adopted the conclusion this court advanced in its decision below that the Commission is empowered, when confronted with a price squeeze, "to put wholesale rates in the lower range of the zone of reasonableness, without concern that overall results would be impaired, in view of the utility's own decision to depress certain retail revenues in order to curb the retail competition of its wholesale customers." Id at 279, 96 S. Ct. at 2004 (quoting Conway Corp. v. FPC, 510 F.2d 1264, 1274 (D.C.Cir.1975)).

Conway, of course, arose out of a particular ratemaking proceeding, and involved allegations that the wholesale rate increase, which would have raised some wholesale rates above the filing utility's retail rates, see 510 F.2d at 1266, was an attempt by the utility to squeeze the customers out of competition. See Conway, 426 U.S. at 274, 96 S.Ct. at 2002. The presence of these features of Conway meant that the Supreme Court had no occasion directly to address two important questions. First, is FERC obliged to consider the potential anticompetitive effects of a rule it proposed to adopt through informal rulemaking, or should FERC address price squeeze allegations only in the context of setting rates within the zone of reasonableness? Second, is FERC obliged to take account--either in rulemaking or in setting rates within the zone of reasonableness--of a price squeeze that is created solely by a disparity between the ratemaking policies of state regulatory agencies and the Commission?

Unfortunately, FERC has yet to supply coherent answers to these questions.4 4] Because, as we shall explain, the validity of the CWIP rule and also of several subsidiary decisions FERC made in Order No. 298 turns on how those questions are answered, we remand to the Commission for a full explanation of its position and further proceedings if necessary in light of that position. We do not decide these questions ourselves because we think that, in substantial part, their resolution calls for the exercise of FERC's expertise and discretion.

We begin by examining the reasons FERC gave for its decision to consider CWIP-related price squeeze claims only on a case-by-case basis. First, as FERC said in its Notice of Proposed Rulemaking, "[m]any factors can contribute to a Commission finding of price squeeze and undue discrimination, including differences between federal and state commission policies." 46 Fed.Reg. at 39,452, J.A. at 26. FERC adhered to that position in Order No. 298. See 48 Fed.Reg. at 24,345, J.A. at 1339. As we understand it, the thrust of this point is that only a full exploration of the particular facts of each case can determine which of the many disparities between state and federal regulatory practice should be deemed to cause a price squeeze.

Second, FERC also made an effort here to gauge the likelihood that CWIP in rate base would cause a price squeeze. Commission staff performed a simulation analysis based on information from five cases in which price squeeze had been alleged. J.A. at 1461. Eleven different price discrimination comparisons were made in light of the fact that some of the cases involved multiple rates or time periods. The simulation study found that in none of these cases did CWIP create a price squeeze, although in three of the eleven comparisons inclusion of CWIP increased the magnitude of preexisting price discrimination. Id. at 1462. In addition, FERC found that petitioners had offered no evidence "establishing the occurrence or likely occurrence of a price squeeze" as a result of allowing CWIP in rate base. 48 Fed.Reg. at 24,346, J.A. at 1340.

Finally, FERC stated that even if a price squeeze is found to exist in a particular case as a result of CWIP in rate base, its CWIP rule would not "operate to automatically deny CWIP simply because a price squeeze is found to exist." 48 Fed.Reg. at 24,346, J.A. at 1340. FERC based this decision on "our stated policy not to remedy a price squeeze if it is due solely to state regulatory action or inaction." Id. at 24,346 n. 97, J.A. at 1340 n. 97.

We now examine the implications each of FERC's reasons has for the two threshold questions we have identified. FERC's first reason--that many disparities between state and federal ratemaking practice can contribute to or cancel out a price squeeze effect--clearly implies that a price squeeze created solely by state regulatory action or inaction is in principle remediable by FERC when it sets rates within the zone of reasonableness. But that zone is determined by the facts of each particular case; thus, it is possible that all or a portion of a price squeeze would be irremediable because the remedy would result in unreasonably low wholesale rates. In addition, the net effect of the various regulatory disparities, and the causal contributions of each to a price squeeze, obviously depend on the facts of each particular case. Hence FERC invokes its discretion not to take potential price squeeze effects into account in formulating the substantive content of a rule of general applicability. Case-by-case treatment, however, will ensure that, as applied, the rule will not operate to create a price squeeze.

FERC's second reason displays two sides of the same coin: petitioners did not show that the disparity between FERC's CWIP rule and the AFUDC-only rules of some state agencies would actually and frequently cause price squeezes, and FERC's own study suggested that allowing 50% CWIP in rate base would generally not have this effect. Again, the necessary implication is that FERC must consider price squeeze resulting solely from disparities between state and federal regulatory action. Otherwise the dispute as to the likely impact of FERC's CWIP rule vis-a-vis the less permissive rules of some states would be utterly irrelevant. At the same time, because FERC is invoking its conclusion that allowing 50% CWIP in rate base will generally not, standing alone, create a price squeeze, the implication is that FERC might be required to modify its substantive rule if it were shown always or generally to have that effect.

FERC's final reason seemingly contradicts each of the others. FERC states that it will generally not remedy price squeeze resulting solely from state regulatory action or inaction, even on a case-by-case basis. The basis for that general policy is that federal ratemaking policy ordinarily should not yield to competing state policies. See Pennsylvania Power Co., 21 F.E.R.C. (CCH) p 61,313 (Dec. 21, 1982). It arguably follows that FERC need not consider the potential price squeeze effects of a proposed rule of general applicability, where those effects arise solely because some state agencies have adopted different rules.

Thus, FERC has in effect said (1) that it need consider the price squeeze effects of regulatory disparities only when FERC's rules are applied in particular cases, (2) that it may be required to consider those same price squeeze effects at the rulemaking stage if they are sufficiently widespread, and (3) that it may refrain from considering those price squeeze effects at the rulemaking stage and will generally not remedy them by setting rates within the zone of reasonableness.

We obviously cannot affirm a decision based on three different and inconsistent answers to the same fundamental questions. In its brief, FERC elides this inconsistency by ignoring its second and third answers and urging only the first, which it says we accepted as sufficient in a closely analogous context in Public Systems II. This, we think, is post hoc rationalization--though by subtraction of old reasons rather than addition of new ones. Unless we can agree that FERC would necessarily have reached the same decision on the basis of the first reason (the multiple regulatory disparities rationale), we would in effect be affirming on a ground different from the one on which the agency based its decision, in contravention of the Chenery principle. See SEC v. Chenery Corp., 318 U.S. 80, 63 S. Ct. 454, 87 L. Ed. 626 (1943). We need not decide this Chenery question, however, for we find that our decision in Public Systems II is not controlling here and that FERC's first rationale is deeply flawed.

In Public Systems II, wholesale customers argued that disparities between the normalization policy FERC had adopted and the tax allowance policies of state regulatory agencies would inevitably lead to price squeezes. 709 F.2d at 84. That assertion precisely parallels the claim made here that disparities between FERC's CWIP rule and state AFUDC-only rules will lead to price squeezes. In its tax normalization rulemaking, FERC had decided to deal with normalization-related price squeeze issues on a case-by-case basis,

because no single ratemaking principle necessarily creates price squeeze.... There are many differences between the principles the Commission utilizes in setting wholesale rates and those that the states utilize in setting retail rates. A given state may treat rate of return, cost allocation, rate base valuation, and normalization differently than the Commission. It is possible that the divergent policies have offsetting effects that minimize or eliminate any rate disparity otherwise created by conflicting normalization practices. Therefore, price squeeze can be shown to exist only by examining the circumstances in a particular rate proceeding.

Id. (citations omitted). This policy, we held, was "rational and adequately explained." Id.

FERC urges that the identical policy underlies its decision to resolve CWIP-related price squeeze issues on a case-by-case basis. Viewing each case in isolation, we would agree with counsel for the Commission that there is no basis for distinguishing FERC's price squeeze policy concerning tax normalization from its price squeeze policy concerning CWIP in rate base. In each instance, the potential for price squeeze exists because wholesale customers (who are competing with the utility in the retail market) are paying rates that include a cost of service item that some state regulatory commissions forbid the utility from including in its retail rate base. In the case of tax normalization, that cost is the increased tax expense that results when a utility's tax deduction is not flowed-through to ratepayers at the time the IRS allows the deduction. In the case of CWIP in rate base, that cost is the carrying charge on investment in construction work in progress. Thus, the potential price squeeze effects are, in principle if not necessarily in magnitude, identical.

Public Systems asks us to distinguish Public Systems II on the grounds that at the time that case was decided, because FERC had a "comparatively conservative CWIP policy ... it could state with some semblance of plausibility that requiring tax normalization would not necessarily tip the scales so as to create a likelihood of price discrimination. Now the Commission is liberalizing its last major conservative rule, and the likelihood of discriminatory impact cannot sensibly be denied." Brief for Public Systems at 47-48 (citations omitted).

This argument has much appeal. If, in one rulemaking after another, FERC adopted policies that invariably tended to increase wholesale rates and that were generally not followed by state regulatory agencies, at some point the cumulative effect of FERC's decisions would make price squeezes inevitable. Indeed, the fact that FERC justified its decision to address price squeeze issues on a case-by-case basis on the alternative grounds that its CWIP rule will not generally lead to price squeeze confirms that FERC recognized that this danger exists. FERC implicitly conceded that if allowing CWIP in rate base would increase wholesale rates by a sufficiently high percentage--FERC did not say how high--then an inference that the rule would generally lead to price squeezes would arise. But, on the basis of its simulation study, of other studies showing initial rate increases in the 5% range, and of its 6% limitation on CWIP-related increases under the new rule, FERC concluded that this point had not been reached.

Thus, we think that petitioners correctly assert that we cannot mechanically accept, in one rulemaking case after another, the rationale we upheld in Public Systems II. What underlies that rationale is a general judgment that the disparities between federal and state regulation tend to balance out and that no one rule is likely to have effects that swamp the effects of all other rules. In Public Systems II, we did not require FERC to detail the basis for that judgment. However, judging from our opinion in that case, the petitioners had not argued, as they strenuously do here, that a pattern of FERC rules tending to increase wholesale rates on the basis of costs not allowed by many state agencies has emerged. We think FERC was obliged to address that claim--unless it chose to take the position, suggested by its general policy of not remedying "regulatory" price squeezes, that it could ignore any resulting price squeezes because they are caused by regulatory inconsistencies rather than utility conduct.

FERC failed to do either of these things in Order No. 298. FERC did not take the position to which its general policy tends--that price squeeze caused solely by disparities in state and federal regulation should not be a factor in determining the content of federal ratemaking policy--although FERC may well have been influenced by that policy in concluding that the CWIP rule would not result in "undue price discrimination." 48 Fed.Reg. at 24,34 6, J.A. at 1340 (emphasis added). Nor did FERC claim that, in its judgment, prevailing state practices and policies, on balance, are not so much less permissive with respect to allowable costs, rates of return, and cost allocation that any incremental rise in wholesale prices would probably cause price squeezes in some states. We recognize that FERC might well conclude that any such conclusion is necessarily impressionistic; the agency might also conclude that detailed comparison of state and federal policies is both feasible and warranted. We do not presume to make that decision. To require FERC to engage in a comprehensive inquiry into the differing practices of the state regulatory commissions and into the individual and aggregate effects of each of those practices seems to us to intrude much too far into a province reserved to the expertise and discretion of the Commission. Nonetheless, we think FERC must at least consider whether such an investigation is needed if it wishes to rely on the rationale it argued for in Public Systems II.5

There is a second reason why we do not believe Public Systems II is controlling here. FERC traces its general policy against remedying "regulatory" price squeezes to its decision in Pennsylvania Power Co., 21 F.E.R.C. (CCH) p 61,313 (Dec. 21, 1982). Public Systems II was argued in February 1983, and our opinion contains no indication that we were aware of FERC's general policy. But that policy obviously affects the assumption on which Public Systems II was decided: as FERC explained its decision to resolve normalization-related price squeeze claims on a case-by-case basis, the plain implication was that if FERC found that a price squeeze existed in a particular case, it would remedy that price squeeze whether caused by the utility's conduct or by a disparity between state and federal regulatory rules. Hence FERC's decision was justified on the grounds that the tax normalization rule would not ordinarily be applied so as to create a price squeeze.

We cannot possibly make that assumption here. If anything, we must assume that FERC will not remedy a price squeeze caused solely by the fact that FERC allows CWIP in rate base while a particular state agency does not. FERC has not announced that it is reconsidering or abandoning that general policy, which it adhered to in Boroughs of Ellwood City v. FERC, 731 F.2d 959, 971-72 (D.C.Cir.1984). There, in dicta, we expressed some doubt as to the consistency of FERC's general policy with its obligations under Conway. See id. at 974-75.6 Absent a retraction of FERC's general policy, we must nonetheless assume that FERC will continue to apply that policy despite the questions raised in Boroughs of Ellwood City. Plainly, then, Public Systems II is not controlling on this point. The Public Systems II court proceeded in the belief that there was a case-by-case remedy for "regulatory" price squeezes; as it turns out, there generally is not.

Indeed, we think this analysis demonstrates that it is largely futile for FERC to continue to evade the question whether its general price squeeze policy also implies that "regulatory" price squeezes are not a factor to be considered in fashioning federal ratemaking policy. FERC seems in reality to have two related policies: it will consider price squeeze issues only on a case-by-case basis, and it generally will not remedy a "regulatory" price squeeze even where one is found to exist. Combined, these two policies appear to us to come very close to a policy against considering the price squeeze effects of disparities between state and federal ratemaking policy in either the rulemaking or the ratemaking context.7

We have therefore considered the possibility that we should rule on the question whether FERC's general policy concerning "regulatory" price squeezes is consistent with Conway and the Federal Power Act. We have concluded that it would be inappropriate to do so, because that question was not briefed or argued in these cases, and because the Commission should have an opportunity to bring its expertise to bear on this question before we decide it. However, unless the Commission decides on remand to modify the CWIP rule itself to take account of the rule's potential "regulatory" price squeeze effects, or finds that the rule will have no such effects, we believe the Commission must be prepared to justify its restrictive view of FERC's role in preventing price squeezes in the course of judicial review of the validity of the CWIP rule.

We have also considered one final ground for affirmance. Were we to conclude that FERC's second rationale (no actual price squeeze impact) for its price squeeze decision was sound, we would not hesitate to affirm because that is the rationale on which FERC primarily relied. FERC's finding of minimal price squeeze impact rests on two props: the absence of evidence establishing likelihood of price squeeze as a result of allowing CWIP in rate base, and FERC's simulation study. The first prop is clearly insufficient. Allowing CWIP in rate base concededly increases the chances of price squeeze in AFUDC-only states, and some finding as to the size of the initial risk is therefore necessary before it can be assumed that the increase in that risk will not amount to a likelihood. As earlier noted, FERC made no such finding. Hence the critical question is whether FERC could validly generalize from the results of its simulation study.

Public Systems assails FERC's simulation study, arguing that all but one of the cases in which the study found that CWIP would not cause price squeeze involved retail rates in Illinois, which petitioner says has a generous retail CWIP policy, and that in the one remaining case the retail rate of return was extraordinarily high. See Brief for Public Systems at 44-45. These are serious criticisms, yet FERC did not respond either in its order on rehearing, see 48 Fed.Reg. at 46,015, J.A. at 1792, or its brief in this court. FERC is not entitled to give dispositive weight to a small and selective simulation study in the face of these methodological weaknesses. We therefore cannot affirm FERC's price squeeze decision on the basis of the simulation study. However, it is open to the Commission on remand to establish that the simulation study is representative and hence a valid basis for generalizations about the price squeeze impact of the CWIP rule.

We have determined that FERC's decision to resolve CWIP-related price squeeze issues as they arise in particular ratemaking proceedings, rather than to refrain from allowing CWIP in rate base or otherwise modify its new CWIP rule, must be set aside. That decision was prompted by inconsistent rationales, no one of which we can sustain on this record. On remand, FERC may seek to show that its CWIP rule will not have any significant price squeeze effects. Alternatively, FERC may determine that its general policy against remedying "regulatory" price squeezes is paralleled by a general policy against considering the potential that a proposed rule will cause such price squeezes in deciding whether to adopt the rule or in tailoring its content. In that event, FERC must explain the basis, in law and policy, for this restrictive view of its price squeeze responsibilities. Finally, FERC may conclude that it must consider the price squeeze effects of a proposed rule at the rulemaking stage. If FERC then finds that its CWIP rule will have such effects, it must explain what importance is being assigned to those effects in comparison to the importance of the public interest goals on which we have determined FERC may validly rely as general support for allowing CWIP in rate base.

Petitioners allege that allowing CWIP in rate base will produce a second kind of anticompetitive effect in addition to price squeezes. This effect, which FERC referred to as the "double whammy," arises when a wholesale customer embarks on a construction program of its own in order to supply itself with all or part of its future power requirement, thereby reducing its dependence on FERC-regulated utilities. Forcing this wholesale customer to pay CWIP-based rates to its supplying utility, the argument goes, "is inherently anticompetitive, since it would cause a wholesale customer building generation [facilities] to subsidize its competitor." Brief for Cities at 36 (quoting Request for Rehearing by Municipal Electric Distribution Group at 5, J.A. at 1490).

In its order on rehearing, FERC acknowledged that the "double whammy" is inequitable and anticompetitive and that steps needed to be taken to minimize the possibility that the CWIP rule would have this effect. FERC's position is that wholesale customers should not be required to pay CWIP-based rates related to facilities which it is shown will not be used to serve them. FERC recognized that in Public Service Co. of New Mexico, 23 F.E.R.C. (CCH) p 61,218 (May 12, 1983), it "assumed that non-receipt of power from a plant is sufficient to absolve a wholesale customer from responsibility for CWIP." 48 Fed.Reg. at 46,014, J.A. at 1788. But FERC now says that because utilities "decide to plan and construct new plant on the basis of their and their customers' load forecast," "a wholesale customer who subsequently decides to obtain power from an alternative source may also be at least partially responsible for the decision to construct a plant for which CWIP is sought." Id., J.A. at 1788. Accordingly, the Commission stated that it would "allow wholesale customers to escape liability for CWIP if they can prove that they bear no responsibility for the decision to build a new plant and will, in fact, purchase no full or partial firm power requirements which involve the new plant." Id., J.A. at 1788.

To implement this decision, Order No. 298 amends FERC's filing regulations to require utilities "to identify and functionalize" CWIP facilities that are "dedicated to a particular customer or group of customers, such as unit sales transactions." 48 Fed.Reg. at 46,014, J.A. at 1787 (citations omitted). That requirement is intended to provide information from which it can be determined whether any given construction project will serve wholesale customers subject to the filed rates. Id., J.A. at 1787. Moreover, FERC reiterates that utilities must adequately support requests for CWIP in rate base, and demonstrate that the proposed allocation of CWIP plant to wholesale customers is reasonable. But while the utility still bears the ultimate burden of persuasion as to the reasonableness of its proposed rates, FERC has shifted to wholesale customers "the burden of going forward to establish both that the customer's demand was not a significant factor in the utility's decision to build a facility and that the customer will definitely leave the filing utility's system by the time the plant is operational or during its operational life." 48 Fed.Reg. at 46,014, J.A. at 1789.

Petitioners raise several specific objections to this method of resolving "double whammy" issues. Mid-Tex argues that since it is the utilities and not their customers who decide whether new facilities are needed, the utilities should bear the burden of establishing that service to their wholesale customers was instrumental in the decision to build. Furthermore, Mid-Tex asserts that FERC has impermissibly shifted the burden of proof to utility customers to show that CWIP-based rates are unreasonable by demonstrating that the new plant was not built for them. Brief for Mid-Tex at 53. Finally, Cities objects that requiring the customer to establish that it will not purchase power that involves the new plant is inconsistent with FERC's general approach to cost allocation. Generally, Cities says, FERC has allocated plant costs according to how much capacity the utility was required to build in order to save the customer's load, not according to how much power the customer actually takes. Brief for Cities at 37.

More generally, petitioners also argue that the risk of "double whammy" establishes that allowing CWIP in rate base is inherently anticompetitive and, together with the rule's potential price squeeze effects, demonstrates that Order No. 298 is unlawful. Brief for Cities at 36; Brief for Alabama Electric at 40. In view of our decision to remand on the price squeeze issue, we think it clear that, on remand, FERC must also reconsider its treatment of potential "double whammy" effects. FERC concedes that its mandate to consider anticompetitive effects includes "double whammy" as well as price squeeze effects. That being so, the same question that we raised in Part V-A in connection with price squeeze recurs here: whether and to what extent FERC must consider those effects in the course of rulemaking as well as ratemaking. FERC has yet to clearly answer this question. Similarly, FERC's decision to allow wholesale customers to escape CWIP liability on proof that they are not responsible for the decision to build and will not take power from the facility under construction is essentially a reprise of its decision to consider price squeeze issues only on a case-by-case basis. See Brief for FERC at 37. We cannot affirm this decision as it stands. FERC offered no explanation for its refusal to consider "double whammy" as an objection to its CWIP policy rather than as a defense to its application. Nor did FERC offer its estimate of the scope of the "double whammy" problem, or respond to petitioners' assertions that efforts by wholesale customers to develop their own generating plants, or engage in joint ventures with utilities or other distributors, are increasingly common. As Public Systems correctly says, FERC simply never "examin[ed] the competitive forces at work in the bulk power market." Brief for Public Systems at 34. While it is open to FERC on remand to explain why an assessment of prevailing market conditions is feasible or unnecessary, we cannot upheld FERC's altogether cursory treatment of "double whammy" effects.

On remand, if FERC concludes that its initial decision to treat "double whammy" issues on a case-by-case basis is justified, FERC must also respond to petitioners' specific criticisms of the manner in which FERC has structured trial of those issues. For example, FERC offered no reasoned basis for its imposition of the burden of going forward with issues of allocating "responsibility" for a utility's decision to build a plant on the utility's customers. Nor did FERC offer even a hint as to what kind of "responsibility" is decisive. Does FERC mean to focus on the subjective decisionmaking of the utility's managers or rather reconstruct an "objective" decision on the basis of the data concerning customer load growth that was available to the utility? These and related questions must be resolved before a reviewing court can say whether or not FERC's "double whammy" doctrine is rational.

In addition, FERC's "double whammy" doctrine includes an element of extreme discontinuity. FERC appears to make the "double-whammy" defense an all-or-nothing matter. Thus, a wholesale customer who can show that its load played a role in the utility's decision to build that was disproportionately small in relation to its current share of the utility's power, and that it will take only that much smaller quantum of power from the new facility, would apparently not be able to avoid any portion of its CWIP liability. FERC may have made an implicit judgment that any rule that required allocation of CWIP liability on a basis other than allocation of current costs of service is administratively unworkable. But that is merely a guess on our part, and we cannot review, or affirm, on the basis of guesswork.

Finally, it appears that utilities' "obligation to serve" was a decisive consideration in FERC's disposition of petitioners' "double whammy" claims. As we noted in Part I, FERC gave that asserted obligation some weight at several junctures in Order No. 298. We have thus far avoided discussion of the "obligation to serve" because we think it unlikely that this factor was dispositive on the issues addressed in Part IV. Here, however, FERC argues that "the wholesale customers surely would hold the utilities to their obligation to serve ten years down the road in the event those customers did not invest in their own generating capacity. They should be able to commit in advance to not needing the utilities' capacity if they are permitted to avoid paying a return on investment made in capacity to serve them." Brief for FERC at 37.

If, on remand, FERC chooses to rely on this reasoning in support of its case-by-case "double whammy" doctrine, it must explain the source and extent of this "obligation to serve." In FERC's fullest discussion of this point in Order No. 298, the Commission conceded that utilities under its jurisdiction do not have "an explicit obligation to serve requirements customers," but stated that utilities "would not be allowed to terminate such service unless they could show that it was in the public interest." 48 Fed.Reg. at 24,332 n. 51, J.A. at 1267 n. 51. We are perfectly willing to take FERC at its word, but its word is unfortunately open-ended: depending on the content of the "public interest" exception, the practical effect of FERC's policy may be that utilities will rarely, sometimes, or almost always be able to avoid this obligation. The weight that can reasonably be attached to that obligation, therefore, obviously depends on the content of the "public interest" exception, as to which FERC has left us altogether uninformed.

Petitioners claim that FERC's errors in connection with the potential price squeeze effects also infected FERC's decisionmaking in several important respects. We must address this possibility in order to determine which of FERC's decisions in the course of Order No. 298, beyond those directly relating to price squeeze and "double whammy," must be reconsidered on remand.

First, Public Systems correctly suggests, Brief for Public Systems at 49, that FERC's decision to allow CWIP in rate base subject to refund was based, as a "[m]atter of policy, on the importance of immediate relief for utilities with large construction programs" and on the finding that "CWIP does not generally cause price discrimination where it does not otherwise exist." 48 Fed.Reg. at 24,338 & n. 64, J.A. at 1301 & n. 64. FERC, in short, was balancing the harm to utilities from delaying CWIP relief against the harm to wholesale customers from allowing CWIP subject to refund. Since we have held that FERC must reconsider its assessment of potential price squeeze effects on wholesale customers, that balancing obviously must be redone unless FERC validly determines that CWIP will indeed not have price squeeze effects or justifies giving those effects scant weight.

Second, Public Systems argues that because FERC conceded that "as the percentage of CWIP in rate base increases, the likelihood of some price discrimination also appears to increase," 48 Fed.Reg. at 24,346, J.A. at 1340, a reassessment of the magnitude of CWIP-related price squeeze effects must logically have some effect on FERC's choice of 50% as the proper amount of CWIP allowed in rate base. FERC chose the 50% figure because "it more evenly balances the equities due present and future ratepayers," who are alike in "shar[ing] an interest in continued service reliability." Id. at 24,348, J.A. at 1351. We agree with Public Systems that FERC must reexamine its choice of the 50% figure in the event that FERC's reassessment of CWIP-related price squeeze effects results in a finding that those effects are significant.

Finally, petitioners urge that had FERC properly assessed the extent of the price squeezes that would result from its CWIP rule, and given the proper weight to these and other anticompetitive effects, the agency might well have adopted one of several alternative proposals submitted by petitioners or raised by FERC itself. Petitioners also assert that, even apart from FERC's errors in connection with anticompetitive effects, FERC failed to give a reasoned explanation for rejecting these alternative proposals.

The principal alternatives to which petitioners direct our attention are:

(1) requiring contributions in aid of construction ("CIAC") in lieu of allowing CWIP in rate base;

(2) requiring joint ventures between investor-owned utilities and their wholesale customers for constructing and operating new plants; and

(3) including CWIP in rate base only to the extent that the state regulatory agency with responsibility for regulating the utility's retail rates would do so.

We address these alternatives seriatim. We first consider whether FERC's reasons for rejecting each alternative would be sufficient without regard to the extent of anticompetitive effects or the weight to be assigned those effects, and then ask whether FERC's reassessment of price squeeze and double whammy issues is likely to require reconsideration of that alternative.

The CIAC alternative requires ratepayers to make a nonrefundable loan to the utility to supply it with capital to finance new construction. 48 Fed.Reg. at 24,348-49, J.A. at 1353. FERC rejected this alternative for two reasons: first, it found that the tax treatment of CIAC (which may be less costly to customers, if tax-exempt, than CWIP in rate base) is unresolved. Id. at 24,349, J.A. at 1356. Second, FERC believed that since capital collected through CIAC is not revenue to utilities, this method would reduce utilities' external financing needs but would not improve the quality of earnings or interest coverage ratios. Id., J.A. at 1356.

We have grave doubts about FERC's second reason, because we have difficulty seeing why investors would not look through this loan arrangement and treat it as economically equivalent to converting AFUDC paper earnings into current cash receipts. Our perception may be wrong, but we would need to know a good deal more about how investors would perceive CIAC before we could sustain this reasoning.

FERC's first reason, however, supplies a rational basis for its rejection of CIAC. FERC was not rejecting an alternative that would unquestionably do everything FERC expected from its CWIP rule far better than the CWIP rule itself. More important, FERC reasonably determined that there was a need for prompt action. Even if the uncertainty about CIAC's tax status can be resolved, there is no reason to believe it can be resolved quickly. The IRS is not required to resolve difficult issues quickly to accommodate FERC.

We do not think FERC is required to reconsider this proposal on remand, though it is free to do so. There was evidence in the record from which FERC could perhaps have concluded that CIAC, if tax-exempt, is less costly to wholesale customers than is CWIP, see 48 Fed.Reg. at 24,349, J.A. at 1354. But unless we misunderstand the nature of CIAC, it appears that the costs CIAC would impose on wholesale customers could also create price squeeze and "double whammy" effects. Especially in the face of uncertainty as to the tax treatment of CIAC (and hence its true cost to customers) we see no purpose in requiring reconsideration of this alternative.

FERC rejected Public Systems' proposed joint venture requirement as (1) insufficiently specific, (2) beyond FERC's role, and (3) a desirable form of voluntary cooperative arrangement rather than an "industry-wide solution to the problems that prompted this rule." J.A. at 1820-21. FERC's reasoning is concise but not indecipherable or irrational. Public Systems has not explained why it believes FERC has the authority to order utilities to enter into joint ventures. That is an enormous power, and we are not willing to assume that Congress conferred it on the Commission in the face of its suggestion to the contrary.

Thus, if FERC is indeed satisfied that it lacks authority to order joint ventures, we see no basis for further consideration of this proposal as an alternative to the CWIP rule. However, it is clear from Public Systems' comments during the FERC proceedings that the "proposal" is also an objection to the CWIP rule itself--on the grounds that allowing CWIP will reduce the incentives for utilities voluntarily to enter into joint ventures. See 48 Fed.Reg. at 46,020, J.A. at 1072. That objection is, like the other alleged anticompetitive effects of the CWIP rule, within the scope of the inquiry we expect FERC to undertake on remand.

The final alternative we consider is that CWIP be allowed in wholesale rate base only to the extent that the responsible state agency allows CWIP in retail rate base. FERC rejected this approach, which eliminates any possibility of CWIP-related "regulatory" price squeeze, on the grounds that it must formulate federal ratemaking policy rather than defer to state policy, and that there would be no uniform federal policy were it simply to defer to each of the many different approaches to CWIP financing costs that command the support of at least one state regulatory agency. We think it self-evident that, except to the extent that avoidance of anticompetitive effects is itself an element of federal policy, these reasons are compelling. They may possibly be compelling even when set against the pro-competitive thrust of the Federal Power Act and federal antitrust law. For no one seriously contends that the sole purpose of the Federal Power Act is to maximize competition in markets that will inevitably be dominated by natural monopolists. Nonetheless, we have yet to learn how FERC believes that its responsibilities to preserve competition interrelate, in the rulemaking context, with other aspects of its responsibility to regulate in the public interest under the Federal Power Act. Consideration of this alternative is inevitably part of that issue, and we have held that FERC must address itself to this question unless it validly determines that CWIP-related anticompetitive effects, of whatever kind, are so insignificant that there is no need to do so.

We conclude with a recapitulation of our decision. We have rejected petitioners' procedural challenges to the Commission's CWIP rulemaking. As to their substantive challenges to the CWIP rule, we have determined that FERC's stated purposes for adopting the rule are valid and that FERC reasonably concluded that the rule would indeed serve those purposes. However, we have found that FERC's consideration of the potential price squeeze and "double whammy" effects of the rule was both inconsistent and inadequate. This error requires that we remand to FERC for reconsideration of each of the material decisions that were based on or clearly influenced by those errors. The affected decisions include (1) the decision to allow CWIP in rate base, (2) the choice of a 50% CWIP allowance, (3) rejection of a CWIP rule that tracks the regulatory policies of the several states, (4) the decision to resolve CWIP-related price squeeze issues only on a case-by-case basis, (5) the decision to resolve CWIP-related "double whammy" issues only on a case-by-case basis, and (6) the decision to allow CWIP in rate base subject to refund. Unless, on remand, FERC reasonably determines that its CWIP rule will have no significant anticompetitive effects, or unless FERC convincingly explains why it can properly consider those effects exclusively on a case-by-case basis (and, in the case of "regulatory" price squeeze effects, can generally refuse to consider them even in particular cases), each of these decisions must be reconsidered.

Accordingly, we vacate those portions of Order No. 298 that relate to the decisions we have identified, and remand to the Commission for further proceedings consistent with this opinion.