Opinion for the court filed by Chief Judge MARKEY.
The proceedings consolidated here for review were initiated by three separate rate increase filings of New England Power Company (NEP), an investor owned electric utility, under section 205 of the Federal Power Act. In each instance, the Federal Energy Regulatory Commission (FERC) suspended the rate filing and ordered an investigation. NEPCO Municipal Customer Rate Committee (Committee) and numerous other intervenors participated in the extensive proceedings before FERC.1 In two proceedings (R-8 and R-10), FERC approved portions of the rate increases requested. The third proceeding (W-2) has not reached hearing stage, but is final in certain respects discussed below.2 We remand three and affirm the remainder of FERC’s determinations.
The R — 10 rate investigation is the major proceeding among those on review. The initial determination of an Administrative Law Judge (ALJ) was modified by FERC, in its Opinion 49.3 That Opinion affirmed the ALJ’s exclusion of NEP’s expenditures on a cancelled Salem Harbor construction project from the rate base. FERC departed in Opinion 49 from the ALJ’s determination: reducing NEP’s rate of return on common equity from 13.5% to 13.12%; modifying NEP’s capital structure to deduct the Yankee investment from NEP’s common equity; deducting the tax expense arising from an excess of book over guideline depreciation on the NEP/Narragansett facility; and reducing NEP’s charge for subtransmission service.
The R-8 rate determination 4 was before this court in New England Power Company v. FERC, et al. (D.C.Cir.1979) (unreported memorandum opinion), 605 F.2d 572, where the result was a remand for: (1) further review of FERC’s decision to exclude from the rate base NEP’s investment in four nuclear power companies (the Yankees); (2) for consideration of NEP’s income tax normalization in light of this court’s opinion in Public Systems v. FERC, 196 U.S.App.D.C. 66, 606 F.2d 973 (D.C.Cir.1979); and (3) to consider the justness and reasonableness of the rates allowed in view of FERC’s final determination on (1) and (2).
Because R — 8 was subjected to judicial review, FERC’s final determination was reached in R-8 after its determination in R — 10 and included revision of some conclusions reached in R — 10. On the remand of R-8,5 FERC reaffirmed its exclusion of the Yankee investments from the rate base, and revised the rate base to provide NEP a 13.28% return on common equity. FERC stated that its previous rate of return calculation did take into account the effects of tax normalization, but indicated that tax normalization rulemaking was still pending. An overall 9.26% rate of return was found just and reasonable in light of market costs of common equity, returns on similar risk investments, and NEP’s ability to attract capital.
In W-2, FERC made a final decision on only certain aspects of the rate change, and issued an order applying Opinion 49. In so *1332doing, FERC excluded from the rate base NEP’s investment in two other cancelled projects (Units 1 and 2); adjusted the depreciation treatment of certain facilities; and allocated investment tax credit funds to the cost of service.
NEP objects primarily to FERC’s excluding from its rate base its unamortized expenditures on the cancelled Salem Harbor project and on NEP Units 1 and 2; to the amount of return assigned to investment tax credit funds; to the depreciation treatment of the Narragansett facility; and to FERC’s refusal to make certain upward adjustments to NEP’s test year cost of service.
The Committee claims that FERC gave too much to NEP for the cancelled project, and should have made further downward adjustments to the test year cost of service. In addition, the Committee argues for a lower working capital allowance and rate of return, disputing FERC’s adjustment for the Yankee investments, and claiming consumers should benefit from a $30,000,000 debenture issued by NEP’s parent, New England Electric System (NEES).
ISSUES6
(1) Did FERC properly exclude NEP’s cancelled project expenditures from the rate base?
(2) Was FERC’s treatment of NEP’s tax credit funds proper?
(3) Did FERC err in calculating NEP’s interest deduction?
(4) Did FERC abuse its discretion in its treatment of test year estimates?
(5) Did FERC employ a permissible method in determining NEP’s reasonable rate of return on common equity?
(6) Did FERC err in rejecting an adjustment to NEP’s capital structure to reflect a debenture issued by NEES?
(7) Did FERC err in denying NEP’s proposed allocation of a Narragansett tax expense?
(8) Was a NEP bond issue properly included in its capital structure?
(9) Was FERC’s denial of consolidated tax benefits to consumers supported in the record?
(10) Did FERC err in interpreting a settlement agreement on subtransmission expenses?
(11) What is the appropriate disposition on the issue of NEP’s tax normalization and its effect on justness and reasonableness of NEP’s rates?
OPINION
(1) Cancelled Project Expenditures
In 1971, NEP decided to construct an 850 megawatt oil-fired electric generating facility, a project called Salem Harbor No. 5, in view of then current load forecasts of a need for additional generating capacity by 1977. Initial preparations, including site location, procurement, engineering, and licensing activity, were undertaken. In 1973, NEP’s plans for the project received a fatal blow when the Organization of Petroleum Exporting Countries imposed an oil embargo. Sky rocketing fuel costs and resultant energy conservation measures affected a major change in NEP’s anticipated load growth. Those and other factors persuaded NEP to cancel the project in 1975. A similar scenario led to cancellation of plans to build two nuclear power projects called NEP Units 1 and 2.
FERC found NEP’s expenditures for the Salem project to have been prudent in all respects and recoverable from ratepayers over a five-year amortization period. FERC denied NEP’s request to include unamortized expenditures on the Salem and Units 1 and 2 projects in the rate base because those expenditures did not result in *1333an operation of facilities “used and useful” in providing electric service.7
The question in this section is whether FERC’s refusal to include project expenditures in the rate base, while allowing their recovery as costs over time, is a valid approach to allocating the risks of project cancellation. NEP contends that FERC’s approach denies the opportunity to earn a return on prudent investments in the can-celled projects and thus deprives NEP of property without just compensation in violation of the Fifth Amendment.
NEP says capital prudently invested in a generating facility is taken for public use and therefore must be included in the rate base. That view had its genesis in a dissenting opinion of Mr. Justice Brandeis in Pacific Gas & Electric Co. v. San Francisco, 265 U.S. 403, 44 S.Ct. 537, 68 L.Ed. 1075 (1924). This court has recognized, however, that “Justice Brandéis’ formula for ascertaining rate base — the amount of capital prudently invested — was not to become the prevailing rule.” Democratic Central Committee v. Washington Metropolitan Area Transit Commission, 158 U.S.App.D.C. 7, 485 F.2d 786, 801 (1973), cert. denied sub nom., D.C. Transit System, Inc. v. Democratic Centra] Committee, 415 U.S. 935, 94 S.Ct. 1451, 39 L.Ed.2d 493 (1974).
The general rule recognized by this court is that expenditure for an item may be included in a public utility’s rate base only when the item is “used and useful” in providing service; that is, current rate payers should bear only legitimate costs of providing service to them. This court has further held that the Federal Power Commission (FPC, now FERC) “may adopt any method of valuation for rate base purposes so long as the end result of the rate order cannot be said to be unjust or unreasonable”. Washington Gas Light Co. v. Baker, 88 U.S.App. D.C. 115, 188 F.2d 11 (1950), cert. denied, 340 U.S. 952, 71 S.Ct. 571, 95 L.Ed. 686 (1951). That view accorded with the guideline expressed by the Supreme Court:
The Constitution does not bind rate-making bodies to the service of any single formula or combination of formulas. Agencies to whom this legislative power has been delegated are free, within the ambit of their statutory authority to make the pragmatic adjustments which may be called for by particular circumstances.
FPC v. Natural Gas Pipeline Co., 315 U.S. 575, 586, 62 S.Ct. 736, 743, 86 L.Ed. 1037 (1942).
Accordingly, we find no constitutional infirmity in the challenged action.
NEP also argues that FERC’s decision to exclude cancelled project expenditures from the rate base: is internally inconsistent; lacks rational explanation for, and discussion of policy implications in, exclusion of the cancelled projects from the rate base; and arbitrarily and capriciously fails to follow relevant precedent.
NEP’s claim of internal inconsistency rests on the view that FERC, having allowed recovery for cancelled project expenditures, must include expenditures for cancelled projects in the rate base. We disagree.
A balancing of policy objectives is not synonymous with inconsistency. Because it would be inequitable to place on the utility the entire loss of expenditures prudent when made, FERC allowed recovery thereof over time. At the same time, FERC is charged with a duty to protect the consumer against unreasonably high rates. Far from an inconsistency, FERC here struck a reasonable balance between the interests of investors and ratepayers. This court has recognized the value and propriety of such a practical approach:
This path between competing policy objectives — as one of those “ ‘pragmatic adjustments which may be called for by particular circumstances,’ ” Permian Basin Area Rate Cases, supra, 390 U.S. [747] at 777 [88 S.Ct. 1344, at 1365, 20 L.Ed.2d 312] (quoting FPC v. Natural Gas Pipeline Co., 315 U.S. 575, 586 [62 S.Ct. 736, *1334743, 86 L.Ed. 1037] (1942)) — is clearly within the province of the Commission, which “ ‘must be free ... to devise methods of regulation capable of equitably reconciling diverse and conflicting interests,’ ” Mobile [sic] Oil Corp. v. FPC, [417 U.S. 283, 331, 94 S.Ct. 2328, 2345, 41 L.Ed.2d 72 (1974)] (quoting Permian Basin Area Rate Cases, supra, 390 U.S. at 767 [88 S.Ct. at 1360]).
Ashland Exploration Inc. v. FERC, 203 U.S. App.D.C. 436, 631 F.2d 1018, 1023 (1980), cert. denied, sub nom., Tema Oil Co. v. FERC, 450 U.S. 917, 101 S.Ct. 1358, 67 L.Ed.2d 340 (1981).
NEP’s assertions of a lack of rational explanation and policy discussion merely express its dissatisfaction with FERC’s determination. The opinions of the ALJ and FERC make plain that the record evidence was reviewed and the involved equities were weighed. No more is required. See United States ex rel. Chapman v. FPC, 345 U.S. 153, 171, 73 S.Ct. 609, 619, 97 L.Ed. 918 (1953).
Because FERC has on occasion appeared to stray from the “used and useful” standard, NEP contends that application of that standard here fails to follow precedent; Those allegedly aberrant occasions, however, are not comparable.
In Order No. 555, Order Adopting in Part Construction Work in Progress Rulemaking and Terminating Proceeding, 10 FPS 5— 1133 (1976), FERC included expenditures for Construction Work In Progress (CWIP) in rate base, pointing to the limited circumstances envisaged in 18 C.F.R. § 2.16:8
Thus we will allow the inclusion of CWIP in rate base where the construction is of facilities to be used for the conversion to the burning of other fossil fuels of plants which now burn oil or gas.
Because none of NEP’s cancelled projects involved the conversion described in Order No. 555, application of the standard here does not conflict with FERC’s determination in that Order. See Public Service Commission of N.Y. v. FPC, 151 U.S.App. D.C. 307, 467 F.2d 361 (1972).
Similarly, nothing in Washington Gas Light Co. v. Baker, supra, or Democratic National Committee v. Washington Metropolitan Area Transit Commission, supra, conflicts with FERC’s decision in this case. Those cases involved property used and retired from service before investors had been fully compensated. Neither case involved the issue presented here — how expenditures should be allocated when a project is cancelled before any use of the facility begins.
Another exception to the used and useful doctrine was discussed but not applied in Tennessee Gas Pipeline Co. v. FERC, 196 U.S.App.D.C. 187, 606 F.2d 1094 (1979), cert. denied, 445 U.S. 920, 100 S.Ct. 1284, 63 L.Ed.2d 605 (1980), cert. denied, 447 U.S. 922, 100 S.Ct. 3012, 65 L.Ed.2d 1113. In that case, this court recognized FERC’s inclusion of expenditures for alternative gas supply projects in the rate- base. The included expenditures were those for research and development of synthetic natural gas (SNG) production, as outlined in 18 C.F.R. 201(103) (1978).9 In Tennessee Gas, however, as here, the petitioner’s expenditures at issue were found outside the scope of recognized exceptions to the used and useful standard and thus ineligible for rate *1335base treatment. In affirming FERC’s rejection of petitioner’s claims for rate base treatment this court noted:
The Commission had established such an exception for research and development costs, but the Transco projects failed to qualify for special R & D treatment. These expenditures were prudent investments, argues Transco; however, for rate base inclusion expenditures must satisfy not only the necessary condition of prudent investment but also must be “used and useful” in providing service. The Commission did not abuse its discretion when it applied a policy “that SNG expenditures which do not qualify as R & D can be recovered, if at all, only through the price paid for actual SNG production sold in interstate commerce. [Footnotes omitted.]
606 F.2d at 1123-1124.
We therefore find FERC’s refusal to include the contested expenditures in rate base to be consistent with its previous decisions and with judicial precedent. NEP has set forth no compelling reason for departing from that approach in this case.
(2) Investment Tax Credit Funds
The investment tax credit operates to reduce federal income taxes in relation to the amount a utility invests in new plant and equipment. For those investments, the utility takes a credit known as an Accumulated Deferred Investment Tax Credit (AD-ITC). The credit is measured by a statutory percentage of the investment in new qualifying facilities during the tax year. The objective is to stimulate business investment in new plants and equipment and thereby increase employment, productivity, and general economic activity. As an incentive, the credit has been turned on and off, and its statutory percentage adjusted from time to time, as economic conditions change.
The precise issue presented in this section involves the mechanics of FERC’s rate base/rate of return determination. As a general proposition, a regulated utility is allowed to recover from ratepayers all of its expenses, including income taxes, plus a reasonable return on capital invested in the enterprise and allocated to public use. More precisely, the “return” is set by FERC at the level which allows the utility to recover the cost of various types of capital which it has raised and dedicated to public use. The capital so dedicated is derived from three sources: (1) long term bonds; (2) preferred stock equity; and (3) common stock equity. The “cost” of (1) and (2) is readily ascertained by reference, respectively, to the required interest on bonds and the required dividend on preferred shares. The “cost” of (3), common stock, is often subject to argument and debate.
FERC determines an “overall” rate of return to be applied to the company’s rate base by taking the three elements of a company’s overall capital structure, i.e., the amount of bonds, preferred equity and common equity capital outstanding, weighing each in accordance with its proportionate representation of total capital, and applying the appropriate cost rate to each element to produce an overall cost rate or overall rate of return.
FERC concluded here that tax credit funds should be treated as though they were obtained from bonds, preferred equity, and common equity in the same proportions as those items are present in NEP’s overall capital structure. NEP contends that the tax credit funds should be treated as though derived solely from common equity funds, thereby generating the higher rate of return earned by contributions of common shareholders.
Section 46(f) of the Internal Revenue Code (Code)10 reflects the principle that regulatory agencies should not defeat the credit’s intended purpose by requiring a utility to “flow through” to ratepayers the savings resulting from the credit.11 Specifi*1336cally, Section 46(f)(2), the provision under which NEP has elected to be covered, disallows the credit in two circumstances:
(A) Cost of service reduction. — If the taxpayer’s cost of service for ratemaking purposes or in its regulated books of account is reduced by more than a ratable portion of the credit allowable by section 38 (determined without regard to this subsection), or
(B) Rate base reduction. — If the base to which the taxpayer’s rate of return for ratemaking purposes is applied is reduced by reason of any portion of the credit allowable by section 38 (determined without regard to this subsection).
Thus, Section 46(f)(2)(A) allows a limited (a “ratable” portion) reduction of the cost of service and consequent flow through of that portion to ratepayers.12 Section 46(f)(2)(B) specifies that a utility will lose the credit if the rate base is reduced by any part of the credit. As discussed below, the controversy regarding the application of Section 46(f)(2) revolves around the interpretation of the “ratable” portion of the credit, and whether the flow through of that portion to ratepayers impermissibly reduces the rate base.
I.R.S. has issued regulations implementing Section 46(f)(2). I.R.S. Regulation § 1.46-6(b)(3)(ii) states:
(ii) In determining whether, or to what extent, a credit has been used to reduce rate base, reference shall be made to any accounting treatment that affects rate base. In addition, in those cases in which the rate of return is based on the taxpayer’s cost of capital, reference shall be made to any accounting treatment that affects the permitted return on investment by treating the credit in any way other than as though it were capital supplied by common shareholders to which a “cost of capital” rate is assigned that is not less than the taxpayer’s overall cost of capital rate (determined without regard to the credit). What is the overall cost of capital rate depends upon the practice of the regulatory body. Thus, for example, an overall cost of capital rate may be a rate determined on the basis of an average, or weighted average, of the costs of capital provided by common shareholders, preferred shareholders and creditors.
In the proceedings below, FERC determined that the foregoing regulation requires that tax credit funds be assigned the established overall rate of return. As above mentioned, NEP challenges that determination noting that the regulation requires treatment of a tax credit fund “as though it were capital supplied by common shareholders.” NEP thus focuses on the first portion of the regulation, arguing that FERC’s reliance on the last sentence of the regulation results in a reduction of the rate base by the amount of the credit involved and thus results in a violation of Section 46(f)(2).
The assignment of a rate of return level to tax credit funds was extensively analyzed in this court’s recent opinion in Public Service Company of New Mexico v. FERC, 653 F.2d 681 (D.C.Cir.1981). There this court held that FERC’s assignment of the overall rate of return to tax credit funds does not violate section 46(f)(2). Noting an inconsistency of that result with a statement in the legislative history of section 46(f), this court nonetheless concluded that it was consistent with the language and overall design of Section 46(f)(2) and IRS regulation § 1.46-6(b)(3)(ii). New Mexico, at 687. Reexamination of the language and legislative history of section 46(f) in light of the present facts discloses no reason to depart here from the holding in New Mexico.
*1337In Kansas-Nebraska Natural Gas Company, Opinion No. 731, 5 FPS 5-600 (FPC 1975), FPC allowed a common equity return level on tax credit funds, applying the interpretation of section 46(f) here advanced by NEP. Kansas-Nebraska, however, was decided before IRS promulgated the investjnent tax credit regulation (§ 1.46-6(b)(3)(H)) relied on here by FERC. Indeed, the Kansas-Nebraska opinion includes a statement that regulations then under consideration (i.e., 26 C.F.R. No. 1.46-6(b)(3)(H)) would allow a rate of return on tax credit funds equal to the overall cost of capital as determined without regard to the credit. Id. at 7. Those regulations were adopted in 1979 by the IRS and included § 1.46 — 6(b)(3)(H) quoted above.13
In Carolina Power and Light Company, Opinion No. 19, 15 FPS 5-619 (FERC 1978), FERC relied on the same proposed IRS regulation as “support[ing] the contention by staff and intervenors that ADITC should be reflected proportionately in CP&L’s capital structure and that the return allowed on ADITC should be measured by the overall rate of return rather than the higher common equity rate of return.” Id. at 5-625. Since that decision, FERC has summarily denied all requests similar to that made here by NEP.
The introduction appearing in the Federal Register in relation to the 1979 amendment of regulation § 1.46-6(b)(3)(ii) states:
The Committee reports also state that the limitations of section 46(f) were intended to achieve two goals: a sharing of benefits between consumers and investors and a limitation on Federal revenue losses. Under certain circumstances, the common shareholder equity rule [granting ADITC the return on common equity] would deny consumers any of the benefits of a credit and could force ratemaking authorities to set rates higher than the rates that would have been established had no credit been available. Under such circumstances, Federal revenue losses would not be merely limited to the amount of the credit, but would be reduced to an amount less than the credit. Congress did not intend to force consumers to subsidize the cost of the investment tax credit.
44 Fed.Reg. 17666, 17667 (1979). That explanation of the amendment is entitled to great deference, E.P.A. v. National Crushed Stone Asso., 449 U.S. 64, 101 S.Ct. 295, 307, 66 L.Ed.2d 268 (1980), and is consistent with FERC’s treatment of tax credit funds here. New Mexico, supra at 689. Accordingly, we conclude that application of the overall rate of return is an acceptable interpretation of “ratable portion” and a flow through of that portion does not impermissibly reduce the rate base. Hence FERC’s treatment of NEP’s investment tax credit funds was proper.
(3) Calculation of NEP’s Interest Deduction
In determining a utility’s federal income tax expense (as an element of cost of service) for rate-making purposes, FERC must determine NEP’s interest expense, a deductible item. To the extent that interest expense decreases, tax expense increases, and customer rates must be increased to pay the tax expense.
In its December 31, 1979 order in W-2, FERC determined that NEP’s interest deduction should be determined without regard to the tax credit. NEP argues that because FERC has already decreased cost of service by a ratable portion of the tax credit funds, the inclusion of those funds in the calculation of the interest deduction further reduces cost of service, producing in turn, says NEP, an overall return lower than the minimum mandated by the Code and IRS regulations.
The question in this section is whether FERC may properly treat tax credit funds in relation to interest deduction in the same way it treats tax credit funds in relation to rate of return determination. FERC assumes that the company would, absent the credit, have acquired financing by using a combination of debt and equity in the same proportion they bear in present capitalization (less tax credit funds). That assump*1338tion creates an “alternative utility” that might have evolved in the absence of tax credit funds. The capital structure so created is speculative but permissible, in view of the impossibility of determining exactly what financing, and consequent interest payment, the utility might have arranged absent the investment tax credit. As FERC noted:
The only administratively feasible approach is to assume that nothing except the source of cash would have changed if the credits had not been available. The Commission has thus determined that, for ratemaking purposes, it will be assumed that in the absence of investment tax credits a utility would have financed the ADITC portion of its rate base with the same proportions of debt and equity, reflected in its capital structure by which it has financed the remainder of its rate base.14
NEP and FERC offer contrasting and complex economic scenarios and relate them and their alleged results to § 46(f)(2)(A). As this court indicated in New Mexico, supra, however, our role does not entail economic speculation:
As a matter of economic theory, the competing theories advanced by FERC and PNM raise questions that are undoubtedly complex. The question before this court, however, is not. The function of this court is not to speculate on the manner in which a utility would finance investment in a world devoid of the investment tax credit. Since no issue of statutory interpretation is presented, the question before this court is simply whether the factual assumption relied on by the Commission in applying the statute is arbitrary, capricious, or an abuse of discretion. We hold that it is not.
Id. at 689.
This court went on to discuss application of the arbitrary/capricious standard to the issue presented here:
As discussed above, we believe that the Commission’s assumption of capital acquisition in the absence of the credit is not arbitrary, capricious, or an abuse of discretion. As a result, we affirm the Commission’s decision to exclude ADITC from capitalization. Assuming that the capital otherwise provided by the credit would be supplied by debt as well as equity, the exclusion of ADITC from capitalization does not significantly affect the debt ratio. Consumers bear the same cost of service, excluding the ratable reduction permitted by the statute, as they would pay in the absence of the credit. No violation of section 46(f)(2) exists.
Id. at 691.
Accordingly, we find no reversible error in FERC’s approach to calculation of NEP’s interest deduction.
(4) Test Year Estimates and Adjustments
The R — 10 proceeding was based on test year ratemaking concepts, under which, as set forth at 18 C.F.R. § 35.13, a utility seeking to raise wholesale electric rates must file cost of service data for two periods. Period I must contain actual data for the most recent 12 months available. Period II, the test year, must contain estimated cost of service data projected over 12 consecutive months beginning after the end of Period I but not later than three months after the proposed effective date of the rate filing. 18 C.F.R. § 35.13(d)(3)(ii). Here, Period I is the year 1975 and Period II is 1976.
Because rate proceedings are protracted, actual Period II data often becomes available before or during the administrative hearing. Use of actual data does not necessarily lead to more just and reasonable rates, however. Actual costs may reflect unique, non-recurring situations, and substitution of actual data for test year estimates may cause interminable delays and dilatoriness in the hope that history might undermine a utility’s estimates. Indiana Municipal Electric Association v. FERC, 629 F.2d 480, 483 (7th Cir. 1980).
*1339A utility must present a full explanation of the bases for test year cost estimates, establishing the validity and accuracy of each, 18 C.F.R. § 35.13(d)(4), (5) American Public Power Association v. FPC, 173 U.S.App.D.C. 36, 522 F.2d 142 (1975), and the utility bears the burden of showing reasonableness in the increase requested. 16 U.S.C. § 824d(e). Once substantiated, those estimates become the bases for rate-making unless a challenger can prove the projections unreasonable when made or that subsequent events indicate their use would yield unreasonable results. Indiana Municipal, supra at 485. NEP and the Committee contest not that regulatory framework but its application by FERC. Hence the questions in this section concern FERC’s acceptance of certain test year estimates and its adjustments of others.
NEP claims that its over-estimate of revenue from contract demand sales 15 and its under-estimate of its purchased power expense should be corrected to adjust the cost of service in its favor. NEP also challenges as inconsistent and excessive FERC’s adjustment to its coal handling estimate.
NEP provided testimony before the ALJ on each challenged item. It described its estimate of contract demand revenue, in light of actual data, as inherently inaccurate because the number of customers and sales amounts could not be precisely estimated. That claim is simply insufficient to support an adjustment to the test year estimate. NEP made no showing that its estimate was unreasonable when made, aside from implying that the claimed shortfall demonstrates unreasonableness per se. But estimates are inherently subject to error. Hence an error in estimation cannot of itself demonstrate unreasonableness in the estimate, nor does it vitiate use of the estimate for testing the justness and reasonableness of the rates. Accordingly, the requested adjustment to contract demand sales was properly denied. Indiana Municipal, supra.
Concerning NEP’s requested adjustment for its purchased power expense, NEP never established the level of increased expense with a required degree of accuracy. Its initial estimate was between $500,000-$600,000, followed by an estimate of $400,000-$800,000, and then by an estimate of about $850,000. NEP’s failure to supply a reliable figure prevented FERC from making a reasonable determination. Having failed to meet its burden of demonstrating that its requested adjustment in its purchased power estimate was just and reasonable,16 NEP cannot be granted that adjustment on appeal.
Similarly, NEP’s protest of FERC’s adjustment to its coal handling expense was rejected for failure to present persuasive or probative evidence. NEP’s coal handling estimate was based on its assumption that it would burn coal for six months in place of oil. Because NEP did not get approval from environmental authorities, it burned no coal. The ALJ considered it unreasonable to assume coal handling costs without assurance of permission to burn coal. The ALJ correctly concluded, and FERC agreed, that the likelihood that coal would be used was too speculative and uncertain to justify inclusion of its cost in NEP’s estimated cost of service. We consider that conclusion reasonable on this record.
Nor is there the contradiction NEP alleges between FERC’s adjustment of the coal handling estimate and its refusal to adjust the contract sales estimate. The former was found unreasonable when made while the latter was found reasonable when made.
There was no error in FERC’s disposition of the coal handling issue.17
*1340The Committee attacks FERC’s treatment of NEP’s steam maintenance expense, working capital allowance, and treatment of income from Braintree Electric System (Braintree) and Boston Electric Company (BECO).
The Committee contends that the ALJ’s $4,344 million downward adjustment to NEP’s coal handling and associated steam maintenance expense was insufficient in light of NEP’s actual Period I expenses. In response to that suggestion, the ALJ reviewed NEP’s actual steam maintenance expenses from 1973 through 1976, and an estimated 1977 steam maintenance expense.18 The ALJ concluded that, in view of increasing cost trends, the Committee’s suggested use of 1975 actual expense as a base for test year (1976) steam maintenance expenses would produce an unduly low result. Instead, the ALJ applied this formulation:
It is concluded that the appropriate adjustment for this overstatement is the difference between the originally estimated test year [1976] cost of $16.614 million for steam maintenance and the more recent estimate used by the company for 1977, of $12.39 million for this expense, resulting in an overstatement of $4.224 million for steam maintenance costs. After adjustment for the scheduled Bray ton Point overhaul of $2.13 million, the steam maintenance overstatement is determined to be $2.094 million. Upon adding the $2.25 million estimated coal handling cost, the total overstatement for coal handling and steam maintenance is determined to be $4.344 million.
The Committee strenuously objects to use of the $12.39 million 1977 estimate as a basis for determining the 1976 expense, though the Committee itself submitted the 1977 estimate, supra, n. 18. The reasonableness of the 1977 estimate, however, is indicated by its falling more than $4 million below the original $16.614 million 1976 estimate, and by its in-line relationship with NEP’s actual steam maintenance expenses from 1973 to 1976.19
Moreover, as the Supreme Court has recognized, FERC need follow no single method in arriving at just and reasonable utility rates. FERC v. Pennzoil Producing Co., 439 U.S. 508, 99 S.Ct. 765, 58 L.Ed.2d 773 (1979), accord, American Public Power Association v. FPC, supra.
The record establishes that the ALJ gave careful consideration to the evidence in arriving at an appropriate steam maintenance adjustment. The ALJ’s determination, adopted by FERC, is supported by substantial evidence, 16 U.S.C. § 8251(b), and has not been shown to result in unjust and unreasonable rates, 16 U.S.C. § 824d(a). Accordingly, FERC’s steam maintenance adjustment is upheld.
The Committee’s argument concerning NEP’s working capital allowance is similarly flawed. The Committee loses sight of its burden, arguing the weight of the evidence, by attacking the accuracy of testimony and associated FERC-adopted findings of the ALJ, rather than showing a lack of evidence to support those findings.
FERC adjusted NEP’s cash working capital downward to reflect a fuel purchase arrangement between NEP and a subsidiary of Prudential Insurance Company. Under that arrangement (Pru-lease), the subsidiary purchases portions of NEP’s fuel inventory, selling it back to NEP as it is burned. Because a reduction resulted in the working capital needed for fuel expense, NEP’s requested 45 day capital allowance was reduced to a 20 day allowance. The ALJ found, and FERC agreed, that “it would be clearly unreasonable to authorize *1341a 45 day working capital for fuel when NEP is required to supply funds to finance this cost for a period of only 20 days”. The adjustment was supported by a NEP study disclosing a 20 day lag between NEP’s payment for fuel and its receipt of revenue from its customers. That study constitutes substantial evidence supporting FERC’s decision. 16 U.S.C. § 8257(b). The Committee having failed to show that decision unreasonable, it will not be upset merely because another was possible. See Consolo v. Federal Maritime Commission, 383 U.S. 607, 86 S.Ct. 1018, 16 L.Ed.2d 181 (1966).
The Committee further says FERC acted inconsistently with its long-standing practice and policy20 of “synchronizing” related items affecting cost. Having used actual data for determining the cash working capital allowance, FERC must, says the Committee, use NEP’s actual income from the Pru-lease arrangement, must apply the actual prime rate to any claimed NEP expense in the Pru-lease arrangement, and must count the actual income from sale of surplus power to Braintree and the actual rental income from BECO, in determining the cost of service.
The ALJ rejected the Committee’s proposal to relate the capital allowance under the Pru-lease arrangement, a rate base issue, to the cost of service. He determined that NEP’s prime rate estimate was reasonable when made and rejected the Committee’s requested adjustment to reflect the Period II actual prime rate as unsupported by evidence. The ALJ also concluded that there was no basis on which to have anticipated sale of surplus power to Braintree and rejected the Committee’s proposed adjustment as an attempt to substitute actual data for estimated data. Because NEP’s estimate included no rental income from BECO, the ALJ entered an amount, rejecting the actual amount sought by the Committee as not reasonably foreseeable at the time the estimates were developed. See Indiana Municipal, supra at 483. FERC adopted the findings and conclusions of the ALJ on these items.
FERC cannot arbitrarily or discriminatorily change its policy.21 We do not, however, view the 20 day capital allowance entered here by FERC and its treatment of the prime rate, Braintree and BECO matters, to have reflected an arbitrary or discriminatory change of policy. The capital allowance determination was supported by evidence. The challenge to the determinations on the prime rate and Braintree matters lacks a showing that those estimates were substantially in error because of subsequent events not reasonably foreseeable when the estimates were made. Indiana Municipal, supra, at 486. Substantiality is required to permit a degree of latitude “because overstated estimates would almost certainly be balanced by offsetting understatements.” Id. Rectifying, in the light of actual data, a complete omission of a clearly identifiable income source (the BECO rental income) does not require that FERC adjust numerous other estimates reasonable when made.
NEP and Committee having failed to carry the challengers’ burden, FERC’s determinations respecting test year estimates and adjustments are affirmed.
(5) Rate oí Return on Common Equity
The rate of return computations for NEP are complicated because NEES owns all the stock in NEP, and because NEP partially owns the Yankees. As a wholly owned subsidiary of NEES, NEP’s cost of capital cannot be directly measured. Under these circumstances, FERC elected to use NEES’ cost of equity capital as a substitute for NEP’s in R — 10. In Opinion 49-A, resulting from remand of R-8, supra, note 5, FERC determined that NEP was entitled to a 13.28% return on common equity, and affirmed its exclusion of NEP’s *1342Yankee investment from the rate base. The questions presented in this section are whether FERC’s method of determining NEP’s rate of return on equity in this case was permissible, and whether the resultant rate was supported by evidence.
A. Exclusion of Yankee Investment From Rate Base
In response to NEP’s insistence that its investment in the Yankees be included in the rate base, FERC explained that the return on that investment is separately determined in proceedings relating to the Yankees. The Commission stated:
After reconsideration, we still find the answer to the first question is that the NEP Yankee investment should be excluded from NEP’s capital structure. This is so because the NEP Yankee investment is already reflected as common equity in the capital structures of the Yankee plants. Thus, it would clearly be inequitable for NEP, through the Yankee companies, to earn a return on that NEP investment in the Yankee rate proceedings and then for NEP to earn a return on the same investment in the NEP proceedings.4 In other words, unless the Yankee investment is excluded from NEP’s capital structure, the same investment would be taken into account again for ratemaking purposes. To avoid this dual impact, the Commission has consistently excluded the Yankee investments from the capital structures of the sponsoring companies.
As above indicated, FERC’s initial determination to exclude NEP’s Yankee investment entirely from common equity was remanded by this court. New England Power Co. v. FERC, supra. In its slip opinion at 5, this court noted a “meagre, if indeed not non-existent evidentiary foundation” for FERC’s determination, noting further that NEP bears the burden of proving the constituent elements of its capital structure.
In Opinion 49-A, FERC reviewed the policy considerations supporting its determination to exclude Yankee investments and its consequences, that is, NEP will recover the cost of capital invested in its own facilities via the rate base, and will recover the cost of capital invested in the Yankees via its share of those companies’ profits. FERC also noted that NEP had failed to make a persuasive showing in support of its burden of proof on including the Yankee investments. FERC thus articulated a rational basis for its determination, and NEP has not demonstrated that its capital structure necessitates the judicial substitution of a determination contrary to that of FERC.
As above indicated, resolution of conflicting economic issues is precisely within the expertise of FERC. It determined that NEP’s rate of return on equity should be based on cost of capital invested in its own facilities commensurate with the attendant risks, noting that the risk and allowed rate of return differ on NEP’s different investments.23 That determination, which forms the basis for the exclusion of the Yankee investment, is neither devoid of reason nor lacking support in the record. City of Cleveland v. FPC, 174 U.S.App.D.C. 1, 525 F.2d 845 (1976). Accordingly, we affirm FERC’s exclusion of the Yankee investment from common equity.
B. Inclusion of Yankee Investment in Total Capital Structure.
Using a capital market-oriented approach (the Discounted Cash Flow or DCF method) in R-10, FERC determined the rate of return which investors required before investing their money in the NEES enterprise, in light of available alternative investment op*1343portunities, to be 12.75%. FERC also found that rate an appropriate proxy for the NEP composite rate,24 noting that the testimony on derivation of NEP’s rate of return value was primarily premised on the capital structure of NEP composite rather than on that of NEP operating alone.25 FERC explained:
In this case, NEP witnesses argued against excluding NEP-Yankee investments from total (composite) NEP capital ' structure. The implication of this observation is that when they used NEESbased information to estimate a rate of return value, it was a value intended to be applied to both the capital invested in NEP’s own rate base and the capital invested in the Yankees. Consequently, although the witnesses did not explicitly state this, such a rate of return value probably is intended to reflect NEP-composite rather than just the NEP operating component. We note, for example, that NEP Witness Benderly’s direct testimony listed common equity at $288,984,-425 which is (rounded) the $289 million figure we used for NEP-composite investment in Opinion No. 49. [Emphasis in original.]
FERC’s method took account of the integration of Yankee data into NEP’s overall capital structure and investment risk. Recognizing that NEP would earn approximately 10% on its Yankee investments, supra, note 23, FERC determined that NEP must be afforded the opportunity to earn a 13.28% return on its investment in its own operating rate base facilities, if it were to have the opportunity to earn 12.75% on a composite basis. FERC stated:
The Commission’s approach, which falls somewhere between that of NEP and the customers, is really quite straightforward: we find that NEES more closely resembles NEP-composite than it does NEP-operating. The investment in and earnings from the Yankee investment is a component of NEES and a component of NEP-composite; it is not a component of NEP-operating. Thus, we conclude that the rate of return we have determined appropriate for NEES is also appropriate for NEP-composite. The rest of the analysis is simply arithmetical: if NEP-composite is to have an opportunity to earn 12.75%, and the portion of NEP-composite reinvested in the Yankees is earning approximately 10%, then the return from NEP-operating must be in excess of 12.75%. This approach does not, as customers contend, result in an equalization of return on investment for NEP-operating and NEP-Yankees. Neither does it flow from a perception that the respective risks are equal. To the contrary, implicit in the approach is a recognition that the risk and appropriate returns are different. Furthermore, by this process, the Commission has not improperly guaranteed earnings to shareholders, as the customers have argued. Instead, the rate of return determination represents no more than a measure of NEP-operating’s cost of capital, given what we know about NEES shareholders’ risks and re-’ quired returns. [Footnote omitted.]
The Committee and intervenors argue that FERC is allowing NEP, as a shareholder, to collect rates for Yankee companies’ power which exceed those companies’ FERC-approved rate schedules. Although FERC’s method does not change the rate which Yankee companies collect from their customers, the Committee says the method effectively allows FERC to structure rates whereby NEP is able to earn a return on its Yankee investments greater than that envisaged in the Yankees’ filed rate schedule and thus violates the “filed rate” doctrine.
The Supreme Court, in originating the filed rate doctrine, stated that “[a utility] can claim no rate as a legal right other than the filed rate, whether fixed or merely accepted by the Commission . . . . ” Montana-Dakota Utilities Co. v. Northwestern Public *1344Service Co., 341 U.S. 246, 251, 71 S.Ct. 692, 695, 95 L.Ed. 912 (1951). “The considerations underlying the [filed rate] doctrine . . . are preservation of the agency’s primary jurisdiction over reasonableness of rates and the need to insure that regulated companies charge only those rates of which the agency has been made cognizant.” City of Piqua v. FERC, 198 U.S.App.D.C. 8, 610 F.2d 950, 955 (1979) quoting City of Cleveland v. FPC, supra, 525 F.2d at 854.
FERC’s predecessor agency recognized that differences in rates allowed individual utilities and associated returns are justified when predicated on differences in facts or management conditions. St. Michaels Utilities Commission v. FPC, 377 F.2d 912 (4th Cir. 1967). Although St. Michaels Utilities Commission deals with rate classes of one utility, not with a relationship between rates of two separate utilities, its underlying rationale is applicable here. The difference in return levels of NEP’s Yankee investment and its investment in its own generating facilities is attributable to residence of those investments in two distinct sets of separate utilities, namely NEP-operating and the Yankee companies. Differences in rates and returns associated with those two sets of investments result from independent ratemaking processes, where the circumstances of each specific utility are considered. The return authorizable for NEP is therefore not necessarily limited to that authorized the Yankee facilities or to that authorized any other regulated industry in which NEP may have an interest.26
Moreover, FERC methodology is more accurately characterized as separating, rather than as creating an upward adjustment of return on, NEP’s Yankee investment. Hence FERC’s method does not violate the filed rate doctrine. Nor do we consider the difference between the 12.75% rate of return and the 13.28% NEP-operating rate of return to have resulted from an “adjustment” in the latter to take further account (beyond that resulting from DCF analysis) of a shift in risk; rather, the difference comes about through a separation of the components of NEP-composite, and a calculation of a weighted average of the respective components’ rates of return to deduce the implied 13.28% rate of NEP-operating.
The Committee claims that the return on equity rate is not just and reasonable when considered under the “end result” test of Permian Basin, supra, and other authorities. To upset FERC’s determination, however, the Committee bears a heavy burden. FPC v. Hope Natural Gas, 320 U.S. 591 (1944) at 602, 64 S.Ct. 281 at 287, 88 L.Ed. 333. Ratemaking is a complicated process involving many factors, e.g., money market conditions, financial health of the utility, and financial risks. As the ALJ noted,
Mathematical computations used to support desired results relating to rates of return have the ring of authority, but simplistic formula approaches cannot be relied upon as sole determinants of a reasonable level of earnings for a regulated utility. Prudent judgment, consideration of the earnings of comparable companies and the requirements of the traditional Bluefield and Hope doctrines must be the basis for the determination of fair levels of earnings for regulated utilities. [Footnote omitted.]
The approaches used by witnesses for both NEP and other parties in deriving their recommendations for an equitable return for NEP are predicated upon the same basic data relating to earnings of comparable companies, historic earnings patterns of NEES, market prices of NEES securities, and related factors. In addition, these witnesses have made liberal use of various judgmental and subjective factors, including arbitrary adjustments relating to the estimated cost of *1345equity, differences between cost of equity and bond yields, stock prices and other measures.
As we have noted, in arriving at just and reasonable rates, FERC is not bound “to the service of any single formula or combination of formulas”. FPC v. Natural Gas Pipeline Co., supra, 315 U.S. at 586, 62 S.Ct. at 743. The ALJ found a 13.5% return on equity reasonable. FERC reduced that rate to 13.28%. Further, the testimony and evidence indicating a required rate of return on equity ranging from 11.21% to 13.8% supports FERC determination of a 13.28% return. That determination, supported by substantial evidence, falls within the zone of protection from the authority of courts to set it aside. FERC v. Penzoil Producing Co., supra, 439 U.S. at 517, 99 S.Ct. at 771; Permian Basin, supra, 390 U.S. at 767, 88 S.Ct. at 1360. It cannot on this record be gainsaid that FERC’s rate of return on equity determination may reasonably be expected to maintain NEP’s financial integrity while providing appropriate protection to the relevant public interests, both existing and foreseeable. Permian Basin, supra, 390 U.S. at 792, 88 S.Ct. at 1373. FERC’s rate of return on equity determination must therefore be upheld.
C. Rate Determination in R-8.
In Opinion 49-A, FERC contrasted its evaluation of NEP’s rate of return in R-8 with its evaluation in R-10. Because the former employed only NEP-operating information, it was not necessary in R-8 to deduce NEP-operating’s rate of return from NEP-composite’s value as was done in R-10. NEP argues that the R-8 determination was actually based on NEP-composite data and therefore the R — 10 methodology should be applied to R — 8.
Neither NEP nor FERC point to definitive support in Opinion 803 for their respective views regarding the methodology employed in R-8. The repeated distinction in Opinion 803 between NEP and NEES data, and the specific exclusion therein of the Yankee investment from NEP’s common equity, are, however, consistent with FERC’s view that only NEP-operating information was employed in R — 8.
As noted above, one who seeks to overturn a FERC determination bears a heavy burden of demonstrating error therein. FPC v. Hope Natural Gas, supra, 320 U.S. at 602, 64 S.Ct. at 287. NEP has not met that burden. Indeed the evidence to the contrary is more convincing. Accordingly, we affirm FERC’s rejection of NEP’s suggestion that the R-8 rate of return determination be subjected to the method employed in R — 10.
(6) NEES Debentures
The question in this section is whether FERC, in Opinion 49, properly rejected the Committee’s proposal to adjust NEP’s capital structure to reflect proceeds of a NEES-issued debenture.
The proposed debenture adjustment rests on the contention that about $30 million of NEP equity was derived from debentures issued by NEES, at an interest rate of 3.25%. The Committee argues that that portion of NEP’s equity should be treated as long term debt for ratemaking purposes. FERC viewed that argument as directed to the sources of funds invested by NEES in NEP equity, and noted that NEES’ 3.25% interest cost was well below the 9.26% cost ascribed to NEP’s common equity.
The ALJ, in originally denying the proposed adjustment, observed that the meagre and conflicting testimony on the source of the funds is subject to varying interpretations. In commenting on the Committee’s proposed adjustment, FERC stated its intention to use NEP’s (not NEES’) capital structure where NEP’s structure would produce a reasonable end result, and indicated that any adjustment based on NEES’ debt holding would require a compensating adjustment to reflect NEES’ expenses. FERC found the record inadequate to establish that adjustment but concluded that the net effect of the proposed adjustment to NEP’s capitalization “would not be substantial,” and rejected the proposal.
FERC’s rejection of the proposed adjustment must be upheld: (1) the Committee cites no precedent for its proposed method*1346ology; (2) FERC’s overall capital structure determination is reasonable and based on substantial evidence in the record; (3) FERC’s factual assumption on sources of capital structure vidlates no statutory mandate; (4) that assumption is not arbitrary, capricious, or an abuse of discretion, New Mexico, supra, at 689; and (5) FERC’s approach to determination of capital structure should not be judicially reversed in favor of another that produces substantially the same result. Indiana Municipal, supra.
(7) Narragansett Tax Expense
The Narragansett tax expense arose from application of liberalized tax depreciation to generating facilities of Narragansett Electric Company (Narragansett), a retail affiliate of NEP. Those generating facilities, during the early 1970’s, passed the so-called “cross-over” point and straight line book depreciation thereafter exceeded the depreciation available for tax deduction purposes. Before 1967, Narragansett used the facilities primarily in intrastate service, selling their output at rates set by the Rhode Island Division of Public Utilities and Carriers (Division). Those rates resulted in a flow through of the liberalized depreciation benefit to Narragansett’s Rhode Island customers.
In 1967, Narragansett and NEP entered an agreement under which NEP controlled Narragansett’s generating and transmitting facilities and supplied all of Narragansett’s power requirements. In effect, NEP rented Narragansett’s generating and transmission facilities, paying rent in the form of a credit on the monthly purchased power bill it submits to Narragansett. The rent becomes a part of NEP’s overall wholesale or “interstate” cost of service; and all NEP customers, including Narragansett, pay a proportionate share of the rental amount. The agreement had the effect of integrating Narragansett with NEP as a supplier of power to customers both intrastate and interstate.
In a recent retail rate case, Narragansett sought to allocate to its retail customers the additional tax expense arising from an excess of book depreciation over guideline tax depreciation.27 Narragansett claimed that its retail customers in Rhode Island had derived benefits from a flow-through during the earlier period when its facilities were dedicated to intrastate service. The Division rejected the requested allocation, but held that the portion of increased tax expense attributable to facilities actually providing intrastate service was properly allocatable to Narragansett’s intrastate operations. Re Narragansett Electric Co., Docket No. 1172, slip op. at 38-39. (R.I.DPUC 1975), aff’d sub nom., Narragansett Electric Co. v. Harsch, 117 R.I. 395, 368 A.2d 1194, 1204-06 (1977). The Rhode Island Supreme Court affirmed that decision *1347as being within the Division’s discretion. 368 A.2d at 1205-06.
Thereafter, NEP attempted to allocate the nonallocated interstate portion of the additional tax expense arising from the excess of book over guideline depreciation to its interstate operations in the R — 10 • proceeding, relying on the integration of Narragansett’s facilities with NEP’s wholesale power supply system serving all NEP customers. NEP also maintained that its inability to allocate all of the tax expense to intrastate operations, among other factors, made equitable the' allocation of the remaining tax expense to interstate operations.
In Opinion 49, FERC rejected NEP’s proposed allocation, citing NEP’s pre-1967 normalization of the tax effect of' the difference between book and guidelines depreciation of Narragansett’s facilities. Under that normalization, NEP’s interstate wholesale customers did not benefit from the additional tax deduction when guideline exceeded book depreciation. FERC reasoned that it would be inappropriate for interstate wholesale customers to be burdened with additional expense when the tables were turned and book exceeded guideline depreciation. Asserting that the increased tax expense should be treated as any other cost of service item associated with NEP’s interstate service, NEP says FERC’s reliance on the past allocation method is an abuse of discretion, resulting in a patently inequitable situation in which NEP is unable to recover the additional tax expenses from either wholesale or retail customers.
NEP cites no support for its assumption that FERC lacks authority to consider previous related incomes and sources in determining the proper treatment to be given current liabilities and expenses. NEP further assumes that FERC’s lack of discussion of several grounds cited in support of NEP’s position affirmatively demonstrates a failure to consider those grounds. Neither assumption has merit. We cannot fault FERC’s allocation of increased tax expense to NEP, which had reaped the benefits of the corresponding tax credit. FERC reached that result on undisputed evidence concerning the dollar amounts and allocation of service between wholesale and retail customers. As noted above, FERC is not bound to the service of any single methodology. FPC v. Natural Gas Pipeline Co., supra; FPC v. Memphis Light, Gas and Water Division, 411 U.S. 458, 78 S.Ct. 430, 2 L.Ed.2d 420 (1973). We find no basis for overturning FERC’s resolution of the tax expense allocation issue.
(8) Inclusion of Bond Issue in NEP’s Capital Structure
The question in this section is whether FERC confused actual with estimated data in approving NEP’s capital structure. NEP acknowledges that it included a $50 million bond issue in its estimate of year-end (December 31, 1976) capital structure, but notes that the bonds themselves did not issue until January 27, 1977, five days before the rate period ended. NEP contends that FERC intended to approve a capital structure based only on actual data as of December 31, 1976, which would not have included the bonds issued. Alternatively, NEP contends that if inclusion of the bonds was intended, the inclusion is arbitrary and irrational because it does not accurately reflect the cost of capital and interest expense deductions available during the period the rate was in effect.
In Opinion 49, FERC indicated that it was using “NEP’s actual capital structure as of December 31, 1976 adjusted to reflect [its] determination on deferred taxes, the Yankee investments, the NEES debentures, and the $50 million bond issue . . .. ” (Emphasis supplied.) Thus it cannot be said that FERC, having “adjusted” the actual capital structure to reflect a number of items, including the bond issue, intended to approve a capital structure based only on actual data.
Nor can we agree that the inclusion of the bonds was arbitrary or irrational. NEP’s arguments are an attempt to abandon its original estimate, which included the bond issue, in favor of actual data which did not. As indicated above, use of estimated data, once substantiated, is proper unless the estimate was unreasonable when made, or where subsequent events would cause *1348the estimates to yield unreasonable results. Indiana Municipal, supra at 485. NEP has made no such showing. Accordingly, FERC’s inclusion of NEP’s bond issue in capital structure must be affirmed.
(9) Consolidated Tax Benefits
The question presented here is whether FERC erred in ignoring tax benefits associated with the NEES debenture. In so claiming, the Committee envisions a benefit to consumers of $794,000.
In his original determination, the ALJ identified a tax savings to NEP of $794,000 arising out of the consolidated tax return filed by NEES.28 To conform NEP’s authorized tax expense, the ALJ suggested elimination of the excess.
In its Opinion 49, FERC purported to resolve the matter in summary fashion, stating:
NEP is directed to revise its calculations on this issue in accordance with Columbia Gulf Transmission Company, et al., Docket Nos. RP75-105 and RP75-106, Opinion No. 47, issued July 2, 1979.29
Although FERC and the Committee disagree here on the impact of Columbia Gulf on the present facts, FERC has chosen to apply an interpretation denying the consumer any benefits from the alleged consolidated tax savings.
In City of Charlottesville, supra, n. 29, this court remanded FERC’s determination in Columbia Gulf because the order in that proceeding denied the consumer any benefit of a tax saving on the basis of a purported investment incentive effect, which effect was without the support of substantial evidence in the record. Nothing presented in the case before us fills the evidentiary void that required remand in City of Charlottesville, or in any manner supports FERC’s benefit-denying interpretation in light of the ALJ’s determination.
FERC allocated consolidated tax savings to shareholders in Columbia Gulf because those savings were generated by exploration and development investments, and because a need exists to encourage such investments. Remand was necessitated by the lack of evidence that such allocation would produce such investments. Nothing of record here indicates that the present tax savings were generated by such investments, or that allocation of those savings to shareholders would produce such investments. Accordingly, we remand to FERC the question of allocation of consolidated tax savings.30
(10) Subtransmission Expenses
NEP contests FERC’s allocation of certain subtransmission expenses in light of certain settlement agreements, contending for an opportunity to recover costs not recovered as a result of a settlement agreement filed in an earlier rate proceeding (R-9).31 The ALJ described the situation:
*1349It appears that this rate design dispute has resulted, in part, from varying interpretations of the R-9 Settlement Agreement, in which it was stipulated that the former monthly surcharge of 46$ per kilowatt for subtransmission service would be reduced to 5$.
NEP contends that by agreeing to this reduced surcharge, it did not agree to forego recovery of the full amount of the lease costs it has incurred as the result of use of Mass Electric facilities. Thus, NEP’s proposed revised rates are designed to charge the deficiency between this cost and the proceeds from the 5$ surcharge, against all NEP customers, whereas [Committee] would charge this deficiency against only subtransmission customers. This [Committee] proposal would have the effect of a second or an additional charge against Mass Electric customers.
FERC viewed the omission of any provision in the R-9 settlement agreement on sub-transmission expenses not recovered by the surcharge as a waiver of those expenses. NEP argues that the R — 9 settlement was limited to setting the surcharge and has no effect on its right to recover its additional subtransmission expenses, pointing to an earlier settlement agreement (R — 5) on sub-transmission expenses which included a specific clause waiving NEP’s recovery of the remaining expenses.
Though the interpretation of the agency which approved the settlement cannot be discounted, that interpretation must have some rational basis beyond a mere omission of a clause to the contrary. The question here is one of contract interpretation, a function peculiarly within the expertise of courts. Interpretation of the R — 9 settlement contract on this appeal, however, is precluded by the absence from the record of the circumstances and acts of the parties in the making and performance of that contract. At the same time, to affirm the administrative interpretation on the present record would be to substitute judicial abdication for judicial review. See National Labor Relations Board v. Truck Drivers, etc., 630 F.2d 505, 507 (7th Cir. 1980); cert. denied sub nom., Gilmer v. Truck Drivers, etc., 450 U.S. 1030, 101 S.Ct. 1740, 68 L.Ed.2d 225 (1981); Northwest Publications Inc. v. National Labor Relations Board, 656 F.2d 461 (9th Cir. 1981). Accordingly, we vacate FERC’s determination on NEP’s subtransmission expenses and remand that question for further consideration.
(11) Tax Normalization
As noted above, FERC was directed by this court’s remand order in R-8 to reconsider the tax normalization issue in light of this court’s previous opinion in Public Systems v. FERC, supra. FERC admits, however, that it did not on remand decide that issue, because normalization rulemaking was still pending, and that it did not, as a consequence, determine the remanded issue of the justness and reasonableness of finally allowed rates in the light of its conclusions on normalization.
Precluded from review of an inchoate tax normalization determination, and of its effect on the justness and reasonableness of the overall rate of return, , we remand those issues to FERC.32
CONCLUSION
FERC’s determinations respecting can-celled project expenditures, investment tax credit funds, interest deduction, test year estimates and adjustments, rate of return on common equity, NEES debentures, Narragansett tax expense, and inclusion of the bond issue in NEP’s capital structure, are affirmed.
*1350FERC’s determinations respecting consolidated tax benefits, subtransmission expenses, and tax normalization are remanded.
Modified.