Distrigas Corporation, a subsidiary of the Cabot Corporation, imports liquefied natural gas (LNG) from Algeria. It sells the LNG to Distrigas of Massachusetts Corp. (DOMAC), another Cabot subsidiary, which pumps it through a terminal, stores it, and resells it to Boston Gas Company and other customers. In September 1976 the Federal Energy Regulatory Commission (then called the Federal Power Commission) determined, contrary to its prior view, that it had authority to regulate DOMAC and Distrigas sales under the Natural Gas Act. 15 U.S.C. §§ 717 et seq.; see Distrigas Corp. v. Federal Power Commission, 495 F.2d 1057 (D.C.Cir.), cert. denied, 419 U.S. 834, 95 S.Ct. 59, 42 L.Ed.2d 60 (1974).
The Commission did not immediately set rates. Rather, it allowed DOMAC and Distrigas to negotiate rates with their customers, see Distrigas Corp., 58 F.P.C. 2589 (1977) , “clarified” in Distrigas Corp., 1 FERC ¶ 61,163 (1977); and it then approved a ‘settlement’ agreement governing rates between 1978 and 1979, see Distrigas of Massachusetts Corp., 5 FERC ¶ 61,296 (1978) , modified, 6 FERC ¶ 61,253 (1979). In 1979, DOMAC (and Distrigas) applied for a rate increase under section 4 of the Natural Gas Act, 15 U.S.C. § 717c. The Commission began to consider whether these proposed rate increases were “just and reasonable.” Id. As section 4 provides, the Commission suspended the new rates for several months; they then took effect upon DOMAC's request, subject to refund after a final determination of their ‘reasonableness.’ (See Appendix for text of section 4.) An AU found that major portions of the proposed increase were not justified. And the Commission, for the most part, affirmed that decision. DOMAC and Distrigas now appeal the Commission’s decision to us for review. See 15 U.S.C. § 717r(b).
We note that in 1981, while the 1979 proceeding was still pending before the Commission, Distrigas and DOMAC filed for a further rate increase. The interested parties reached a settlement concerning this 1981 increase, conditioning its size on the resolution of several specified issues in dispute in the 1979 proceedings. The Commission has approved this settlement. Distrigas of Massachusetts Corp., 20 FERC ¶ 61,073 (1982). Thus, the rate decision that we here review governs only the period from 1979 to 1981 (which the parties call the “locked-in” period) with the exception of one issue concerning post-1981 refund levels to which we will turn in Part IV.
I
For the most part Distrigas and DOMAC are asking us to set aside certain subsidiary Commission findings — findings that, in part, led the Commission to its final conclusion about what rate was “just and reasonable.” We are asked to review these findings under traditional principles of administrative law. We must determine whether the findings of fact are supported by “substantial evidence,” 15 U.S.C. § 717r(b); see 5 U.S.C. § 706(2)(E); and we must make certain that the Commission’s policy judgments are not “arbitrary, capricious” or an “abuse of discretion.” 5 U.S.C. § 706(2)(A). As the Court of Appeals for the District of Columbia Circuit has stated, in a rate ease such as this one, applying these standards often comes down simply to insuring
that the Commission’s judgment is supported by substantial evidence and that the methodology used in arriving at that judgment is either consistent with past practice or adequately justified____
*1211City of Batavia v. FERC, 672 F.2d 64, 85 (D.C.Cir.1982) (quoting Public Service Commission v. FERC, 642 F.2d 1335, 1351 (D.C.Cir.1980), cert. denied, 454 U.S. 879, 102 S.Ct. 360, 70 L.Ed.2d 189 (1981)).
The case before us involves the application of classical public utility cost-of-service ratemaking principles. In applying these principles, a regulator traditionally will proceed as follows:
1. He selects a test year (t) for the regulated firm.
2. He adds together that year’s operating costs (OC), taxes (T), and depreciation (D).
3. He adds to that sum a reasonable profit determined by multiplying a reasonable rate of return (r) times a rate base (RB). The rate base typically consists of total historical investment minus total prior depreciation. The rate of return typically reflects the coupon rate for long-term debt plus a ‘fair’ return to shareholder equity.
'4. The total equals the firm’s revenue requirement (RR). The regulator then allows prices that will equate the firm’s gross revenues with this revenue requirement.
These four steps can be reduced to three formulae:
1. RR = OC + T + D + Profit
2. Profit = r(RB)
3. Price = RR/quantity sold
See generally A.E. Kahn, The Economics of Regulation (1970). This general account of ratemaking may help the reader understand to which of these formulae’s terms the petitioner’s claims and arguments relate and thus how they fit within the larger context of the ratemaking process. We now discuss each challenge in turn.
II
“Extra” Tax Expenditures
During the years 1979-81 (and thereafter) DOMAC will have to pay certain extra taxes (T) because it previously chose to take advantage of specially favorable tax code provisions in the early 1970’s, before it became regulated. Its early 1970’s tax choices (accelerated depreciation of plant and equipment, expensing instead of capitalizing of certain items) in effect deferred a portion of DOMAC’s current tax liabilities into later years. Who should pay these “deferred” taxes? Should they be included as part of the firm’s current revenue requirement and passed on to customers in the form of higher rates? Or should they be left out of the revenue requirement, effectively making DOMAC’s shareholder (Cabot) pay the extra tax liability out of the ordinary profit that the Commission will allow DOMAC?
The Commission answered these questions here by referring to an approach known as tax “normalization.” It treated DOMAC — which first became regulated after obtaining tax benefits — as it would have treated a firm always subject to regulation. It refused to include the extra tax expense in DOMAC’s revenue requirement; it required the shareholder to bear the cost; and it also required DOMAC to take an amount equal to the entire “extra” tax payment that it would eventually have to make, subtract it from its rate base, and attribute it to several special accounts representing “Accumulated Deferred Income Taxes.” See 18 C.F.R. Part 201, General Instruction 18 and Accounts 281-283. In determining rates, DOMAC, like other utilities, was not entitled to any return on the funds in these special accounts.
DOMAC and Distrigas challenge these decisions. We can best analyze their challenge by taking DOMAC’s “accelerated depreciation” liability as a representative example, by explaining how regulatory commissions typically treat this liability, and by then examining DOMAC’s argument for different treatment. The conclusions we reach concerning accelerated depreciation apply to the other deferrals here at issue as well.
1. The tax “normalization” problem. The tax normalization problem arises out of Congress’s decision to allow firms to *1212depreciate plant and capital equipment at a specially fast rate for income tax purposes. See 26 U.S.C. §§ 167-68. Suppose, for example, a firm buys a piece of equipment for $1 million; it has a twenty-year life. If the firm depreciates the equipment on a “straight line,” it will assume a depreciation expense of $50,000 each year for twenty years. Accelerated tax depreciation rules, however, allow it to depreciate over fewer years, taking a larger “depreciation” expense in the early years, but a correspondingly smaller depreciation expense in later years. If tax rules, for example, allow the firm to take a $100,000 expense (instead of the straight-line $50,000) in year one, the firm will keep whatever taxes it would otherwise have paid on the extra $50,000 (say, $25,000). The firm, however, will have to pay more tax in future years, when the depreciation expenses available for tax purposes will be correspondingly diminished. Assuming a stable tax rate, the lower early-year tax payments will be exactly counterbalanced by higher tax payments in later years; but, of course, firms find this system highly advantageous nonetheless, for they obtain the use of the “saved tax” money until the time it falls due.
Regulators have generally had to decide whether the advantages of this “loan” from the government should accrue to the regulated firm’s shareholders or to its customers; they have had to choose between “flow-through” methodologies, which pass the immediate tax benefits on- to the ratepayers, and “normalization” ^methodologies, which preserve more of the immediate tax benefits for the owners. See Federal Power Commission v. Memphis Light, Gas & Water Division, 411 U.S. 458, 465-67, 93 S.Ct. 1723, 1728-29, 36 L.Ed.2d 426 (1973); Public Systems v. FERC, 709 F.2d 73 (D.C.Cir.1983); Memphis Light, Gas & Water Division v. FERC, 707 F.2d 565, 567-69 (D.C.Cir.1983). But see Economic Recovery Tax Act of 1981, Pub.L. No. 97-34, § 201(a), 95 Stat. 172, 203, 208 (codified at 26 U.S.C. § 168(e)(3)) (requiring normalization approach as condition for accelerated depreciation by public utilities of post-1981 properties); 26 U.S.C. § 167(l) (imposing similar condition on use of accelerated depreciation for some post-1969 utility properties).
Flow-through: Under the flow-through approach, the firm is not allowed to collect from the ratepayer on account of tax expenses any more than the actual amount of tax that it will have to pay in the current year. In other words, if the firm saves $25,000 in taxes in equipment year 1 because of accelerated tax depreciation, it is to collect $25,000 less from its ratepaying customers. Of course, in, say, year 18, when it has a tax bill that is $25,000 higher than it otherwise would be, the firm can collect the additional $25,000 from its customers by charging them higher rates. The Commission in essence gives the customers the government “loan” to use as they see fit until the time the deferred taxes come due. Then the customers will have to pay the deferred bill.
Normalization: By contrast, the normalization approach allows the firm to collect the same tax money from its customers in early years that it would have collected in the absence of any accelerated tax advantage. The firm keeps the $25,000 that its use of accelerated tax depreciation saved it in equipment year 1. The firm then pays the money that it has kept to the government in, say, year 18 when taxes are higher. In the meanwhile, it may use the funds it has already collected from the ratepayers as it sees fit.
This system does not usually allow the firm to obtain full advantage of the “loan,” however, for the regulator typically deducts from the firm’s rate base the amount of the funds that the firm has collected from the ratepayers “early.” This prevents the firm from charging the ratepayers to generate a return on the funds that it has, in effect, borrowed at no cost from them. The firm then adds the deducted amount back into the rate base in later years as the extra tax owed is paid to the government.
*1213Suppose, for example, that a firm possesses $10 million in undepreciated plant, equipment, and working capital. And assume that it receives $25,000 from its ratepayers reflecting tax liabilities that (because of accelerated depreciation) are not currently due. Suppose that it invests this $25,000 either in a new machine or in added working capital. Were no further adjustment made, the firm’s rate base would now stand at $10,025,000; and the firm would be allowed to earn a reasonable return (say 13 percent) on that amount. But instead the regulator deducts $25,000 from the rate base and attributes it to a special account for deferred tax receipts (“Account 282” under FERC’s system of accounts). So, the firm is allowed to earn 13 percent only on $10 million, not on $10,025,000. (This adjustment has no impact on the depreciation charges that the company is allowed to include in its revenue requirements; although it is only entitled to a return on a part of its plant and equipment, it is nonetheless entitled to current, straight-line depreciation on all depreciable assets.) Later, when the $25,000 is paid to the government, the rate base is increased to reflect the fact that the firm has now made the $25,000 investment out of its own funds. (See Figure 1.)
FIGURE 1
Assets Liabilities
Plant and other utility-related capital $10 Million
New Machine (utility-related) $25,000
Shares (Investment) $10 Mill
Deferred Tax Account $25,000
(Rate Base = $10,025,000 - $25,000 = $10 Million)
Of course, the firm may take the extra $25,000 and simply invest it in a savings account or a municipal bond, or put it to some other non-utility related use. In that case, the firm’s rate base, absent further adjustments, would remain unchanged at $10 million, but the shareholders would be earning an extra return from the non-utility use of the .$25,000. So, in this case, too, the rate base is reduced by subtraction of the $25,000 (leaving $9,975,000 in our example), and the return that the firm earns on the bond or savings account is assumed to compensate the shareholders for their “missing” $25,000 of plant and equipment. (See Figure 2.)
FIGURE 2
Assets Liabilities
Plant and other utility-related capital $10 Million
Municipal Bond (not utility-related) $25,000
Shares (Investment) $10 Million
Deferred Tax Account $25,000
(Rate Base = $10,000,000 - $25,000 = $9,975,000)
In a nutshell, the adjustment to the rate base reflects the fact that, under the normalization approach, the $25,000 was given to the company by its customers to pay taxes not yet due. One might alternatively view the $25,000 as being “loaned” to the company by the Internal Revenue Service. Either way, the firm at no cost to itself has obtained funds which it can invest as it chooses. The return the company is usual*1214ly allowed to recover on its rate base compensates it for its costs in obtaining the requisite capital. So, in the regulators’ view, the company should not be allowed to charge the ratepayers for a “return” on this $25,000 (temporary) addition to the firm’s capital, because it was obtained by the company without cost.
In the thirty years since Congress introduced accelerated depreciation, the choice between flow-through and normalization has been the subject of much controversy, and the regulatory agencies, including FERC, have not held to consistent positions. See, e.g., Public Systems v. FERC, supra. The advocates of normalization have pursued the issue before both Congress and the agencies. They argued to Congress that flow-through compounds the tax revenue losses attendant upon accelerated depreciation and convinced Congress to condition accelerated depreciation for public utilities on the use of normalization. See Tax Reform Act of 1969, Pub.L. No. 91-172, § 441(a), 83 Stat. 487, 625-28 (codified at 26 U.S.C. § 167(/)), discussed in Federal Power Commission v. Memphis Light, Gas & Water Division, supra; see also 26 U.S.C. § 168(e)(3). And they have argued to the regulators that utility shareholders should receive the benefits of accelerated depreciation, since Congress enacted it as an incentive to company owners to induce companies to increase investment. While FERC for a time required use of flow-through, see Alabama-Tennessee Natural Gas Co., 31 F.P.C. 208 (1964), aff'd, 359 F.2d 318 (5th Cir.), cert. denied, 385 U.S. 847, 87 S.Ct. 69, 17 L.Ed.2d 78 (1966), for the past decade the Commission has mandated the use of normalization. See Texas Gas Transmission Corp., 43 F.P.C. 824 (1970), vacated sub nom. Memphis Light, Gas & Water Division v. Federal Power Commission, 462 F.2d 853 (D.C.Cir.1972), reversed 411 U.S. 458, 93 S.Ct. 1723, 36 L.Ed.2d 426 (1973), and affirmed on remand, 500 F.2d 798 (D.C.Cir.1974); Public Systems v. FERC, supra.
2. The Commission's decision here. The Commission followed its normalization principles in this case. The amount of “extra” future tax liability that DOMAC has accrued as a result of accepting tax benefits in the early 1970’s amounted, not to $25,000 as in our example, but to $4.6 million. The Commission held that DOMAC would not be allowed to raise revenues from its ratepayers to satisfy this liability; it would have to pay these “extra” taxes from money that would otherwise go to its parent corporation as part of its profit. The Commission also held that DOMAC’s rate base would be reduced by $4.6 million; that amount would be allocated to separate deferred tax accounts on which DOMAC would not be allowed to earn a return. If we imagine, for example, that DOMAC’s rate base—the land, equipment, working capital, and so forth—was valued for rate-making purposes at, say, $30 million when it first became regulated (we do not have the exact figures before us), the Commission in effect required that the $30 million be reduced to $25.4 million. The “profit” that DOMAC would be allowed to raise from its customers would amount, not to 13 percent of $30 million (or $3.9 million), but 13 percent of $25.4 million (or roughly $3.3 million) instead, reducing the company’s return by something over half a million dollars per year. In essence, the Commission has treated DOMAC as if DOMAC, in the years before it became regulated, had already collected from its customers the $4.6 million tax bill it will have to pay in coming years.
3. The reasonableness of the Commission’s approach. DOMAC argues that the Commission’s decision is unreasonable both in requiring its shareholder to pay the $4.6 million future tax liability and in deducting that sum from its rate base until the deferred taxes become due. In DOMAC’s view, this unreasonableness stems from the fact that it is not an ordinary regulated utility whose rates have been regulated throughout the period subject to normalization. Rather, it is a utility that first became regulated in 1976, and it incurred the $4.6 million tax liability before it became regulated. In evaluating DOMAC’s argu*1215ment, we consider the two related aspects of the Commission’s decision in turn.
a. Who should pay the $¡¡..6 million1 DOMAC essentially makes two arguments for the proposition that its customers should pay the $4.6 million future tax liability. First, it points out that as of 1976, when it first became regulated, it had already incurred this liability, just as it might have incurred a liability, for example, to repay borrowed money. The liability was a fixed feature of the company, and the ratepayers, as of 1976, should have taken the company as they found it — “warts and all,” so to speak. Second, unlike a regulated firm that “normalizes” its tax account from the beginning, DOMAC has never (prior to becoming regulated) collected from its customers the money needed to fund this future tax liability. Rather, its revenues pri- or to 1976 reflected its efforts to charge its customers all the market would bear; and those pre-1976 rates were still not high enough during the early 1970’s to save DOMAC from enormous losses. Since no one can argue that the customers have payed for this liability in the past, they should do so in the future.
The difficulty with DOMAC’s arguments rests in the fact that the Commission has an equally plausible, though differing, view of the matter. In the Commission’s view whether or not past customers paid DO-MAC money to fund this future tax liability is beside the point. In either case, the extra tax liability was incurred in order to obtain tax benefits (in the past) that flowed directly to DOMAC’s parent corporation. Those benefits in principle meant higher past profits for DOMAC (in actuality, they meant lower past losses). Why should future ratepayers in effect pay the check for a meal they never ate? Presumably future ratepayers would not have to pay, in say 1979, for gas which DOMAC had purchased and delivered to customers in, say, 1972 but for which it had neglected to pay its bills before 1979 (or for which it had arranged not to be billed until 1979). And this result would not hinge on whether or not the 1972 customers had paid DOMAC for the services in question. Why should the Commission have to treat this future obligation — likewise flowing from a pre-regulation benefit — any differently?
We need not evaluate the comparative merits of these two sets of arguments, for the Commission must receive the benefit of the doubt. As long as its argument is logical, consistent with traditional ratemaking principles, supported (insofar as it is based on fact) by substantial evidence in the record, and not otherwise “arbitrary,” we must uphold the result. See, e.g., Permian Basin Area Rate Cases, 390 U.S. 747, 767, 790-92, 88 S.Ct. 1344, 1372-73, 20 L.Ed.2d 312 (1968); Public Service Commission v. FERC, 642 F.2d at 1342. Since the Commission’s argument is strong enough to satisfy these criteria, we affirm its decision.
b. Should the Commission have reduced DOMAC’s rate base? We have far greater difficulty accepting the Commission’s ground for reducing DOMAC’s rate base. The Commission argues that the tax benefit that DOMAC received (leading to the $4.6 million future liability) essentially amounted to an “interest free loan” from the government of $4.6 million (just as in the case' of an ordinary regulated firm). And the future ratepayers should not have to pay for a “return” to that portion of DOMAC’s rate base that might (in theory) be traced to purchase with this “loan.” Of course, when DOMAC eventually pays the $4.6 million. “back” to the government, its rate base will, automatically be increased by that amount (just as in the ease of an ordinary regulated firm).
We see three serious difficulties with this argument. First, in the case of the ordinary regulated firm, the deduction from the rate base reflects the fact that the firm’s customers have paid the future tax liability in advance, thus making available to the firm funds that can (in theory) be traced to a portion of the rate base. See, e.g., Public Systems v. FERC, 709 F.2d at 83; Memphis Light, Gas & Water Division v. FERC, 707 F.2d at 568; Alabama-Tennessee Natural Gas Co. v. Federal Power *1216Commission, 359 F.2d 318, 327 (5th Cir.), cert. denied, 385 U.S. 847, 87 S.Ct. 69, 17 L.Ed.2d 78 (1966). Indeed, it seems to be this fact that makes the company’s temporary enjoyment of the funds look like a “loan” from the government. (Where the customers do not pay — for example, under ‘flow-through’ treatment — one is not tempted to say the government has ‘loaned’ the company, any money.) Here, since the firm lost money in the early 1970 s and since it simply charged its customers as much as the traffic would bear, it is difficult to find any basis for assuming that the firm’s customers advanced the $4.6 million. And if the Commission’s position is that the government’s deferral of DOMAC’s tax liability should be treated as an interest-free source of capital that must be excluded from the rate base even where the customers have made no advance payments, then why, in the ordinary regulated firm’s case, would the Commission not make two subtractions from the rate base — one reflecting the “loan” from the customers and another reflecting the “loan” from the government?
Second, the Commission in making this deduction from the rate base is, in essence, examining the sources that had previously contributed to DOMAC’s current value as of the time that the Commission first began to regulate DOMAC (apparently 1976). And such a source examination — stretching back prior to the time DOMAC first became regulated — seems an arbitrary departure from regulatory practice. The Commission agrees that as of 1976 DOMAC’s rate base consisted of a certain value — call it $30 million as in our example. Some of that value can be traced to debt and the rest, presumably, makes up the shareholder’s equity. Does the Commission, or does any regulator, ever look back prior to the time of regulation, and- seek to separate the value of shareholder equity into “legitimate” and “illegitimate” parts, depending on the source of the value of what the shareholders own prior to regulation? Suppose, for example, that the $30 million of plant, equipment, working capital, and so forth was purchased in part with a gift to the corporation from an aged aunt — one mythical Mrs. Cabot, who wished to help her nephews. Or suppose it was purchased in part through special tax credits granted by Massachusetts because an Historical Commission insisted that the appearance of DOMAC’s warehouse remain as it was in the time of Paul Revere. Or suppose it was purchased in part out of monopoly profits enjoyed by the firm before it was regulated. Would the Commission similarly deduct that portion of the $30 mfflion traceable to such sources as these — though they too might be considered “zero-cost capital?”
0f course, it makes sense to ask such questions about contributions to capital that take place during regulation, for a regulator seeks to impose upon customers only the legitimate “cost” to the firm of the capital that is contributed, and under regulation the source will be highly relevant to legitimate costs. But the worth of an unregulated firm as seen by its investors— and hence the return they expect from their investment in it — is not closely tied to the sources of the firm’s value, so to try to disaggregate the sources when the firm becomes regulated is a very different matter. In any case, we have never heard of a regulatory agency stretching its examination of sources of capital back beyond the time regulation begins, to consider the origins of the value that the firm “contributes” (or of the value that makes up the firm) at the advent of regulation. Regulators might have tried to do so. The I.C.C., for example, might have tried to determine how much of a railroad’s rate base was purchased from monopoly profits earned prior to regulation; and the I.C.C. then might have tried to deduct that amount from the rate base on which it would allow a profit. (Of course, the practical difficulties would have been considerable.) But we know of no regulator that has done this, FERC provides us with no precedent. Nor does FERC suggest that it similarly deducts from the rate base any other portions that might be traced to other “low-cost” or otherwise “special” sources predating the *1217advent of regulation. The few cases that we have found involving scrutiny of pre-regulation transactions seem aimed at ascertaining the value of the firm’s property, not the source of that value. See, e.g., Colorado Interstate Gas Co. v. Federal Power Commission, 324 U.S. 581, 606-08, 65 S.Ct. 829, 841-42, 89 L.Ed. 1206 (1945) (examination under 15 U.S.C. § 717e to see if company had improperly inflated total cost of property). Under these circumstances, the Commission’s deduction of a portion of the rate base theoretically traceable to a kind of “government loan” during the preregulatory period seems arbitrary.
Third, while it seems reasonable for a regulatory agency to decide that a regulated firm’s customers, not its shareholders, should enjoy the tax advantages Congress provided in its “accelerated depreciation” provisions, see, e.g., Alabama-Tennessee Natural Gas Co. v. Federal Power Commission, supra, (affirming 31 F.P.C. 208 (1964)), it does not seem reasonable to take those advantages away from the shareholders of an M^regulated firm. It is difficult to see why the parent of unregulated DO-MAC should not have received tax benefits in the early 1970’s. It may be required to pay the subsequent, post-1976, bills related to those benefits, but it seems unreasonable to deprive it of the pre-1976 benefits as well. Yet, to refuse to allow it a return on the investment in DOMAC plant, etc. that it (in theory) acquired with the tax benefit money is, in effect, to deprive it of the value of that pre-1976 tax benefit, FERC does not claim a right to regulate the firm’s pre-1976 behavior, but its requirement of a $4.6 million reduction of DOMAC’s rate base has precisely that effect.'
Because of the complexity of this area, and the risk that we may have overlooked possible responses to some of these arguments, we state explicitly that each of these three reasons taken separately, in our view, makes the Commission’s decision in this respect “arbitrary.” Together they are more than sufficient to convince us that this portion of the FERC decision fails Ba-tavia’s test. It is neither “consistent with past practice” nor otherwise “adequately justified.” The decision to deduct the $4.6 million from DOMAC’s rate base must be set aside.
111
We turn next to a series of DOMAC claims that are somewhat less difficult.
1. DOMAC argues that FERC erred in deducting from its balance sheet $6.5 million that DOMAC had loaned to Cabot, its parent, in return for a set of 8% demand notes. The treatment of the $6.5 million is not directly related to the size of DOMAC’s rate base; rather, it affects the calculation of the rate of return to which DOMAC is entitled on its rate base. As we noted above, the rate of return is determined as a weighted average of (i) the actual return the firm is obligated to pay on its outstanding long-term debt (here 8% on $10 million of debt) and (ii) a fair return on its equity (here 16.5%). The weighting of these two rates of return depends on the proportions of the firm’s worth represented respectively by long-term debt and by equity. See El Paso Natural Gas Co. v. Federal Power Commission, 449 F.2d 1245, 1249-50 (5th Cir.1971).
To calculate these proportions, FERC looks to the firm’s balance sheet. By excluding the $6.5 million in question here from DOMAC’s assets, FERC shrinks the balance-sheet value of DOMAC’s equity by a like amount (from roughly $21.5 million to roughly $15 million), while leaving the $10 million of long-term debt unchanged. The effect is to reduce equity’s share of the firm’s total capitalization from about 68% ($21.5 million out of $31.5 million) to roughly 60% ($15 million out of $25 million) and to correspondingly increase debt’s share. Since this change reduces the weight given to the higher rate of return for equity, it shrinks the overall, averaged rate of return FERC countenances (from 13.81% to 13.11%).
The parties apparently agree that the propriety of the $6.5 million deduction depends on whether these funds were, or were not, available to DOMAC for use in its regulated activities during the test year. *1218See El Paso Natural Gas Co. v. Federal Power Commission, 449 F.2d at 1250-51. Only if they are related to DOMAC’s regulated activities is it fair to count them as a ‘regulatory’ balance sheet asset for purposes of apportioning regulation-related assets among equity, long term debt, and other liabilities. DOMAC points to uncon-tradicted testimony stating that these funds were available to DOMAC on demand. They were given to Cabot simply because Cabot ordinarily used a “demand note” system to manage its subsidiaries’ cash efficiently. Hence, says DOMAC, these funds should be treated no differently than if a regulated firm, with $6.5 million cash, had deposited the funds in a bank demand account where they would remain available for use. And in such a case, no one would argue for elimination of the funds from the firm’s balance sheet or exclusion of them when apportioning assets into equity and debt components.
The Commission viewed the matter differently. The ALT noted that the funds were not in DOMAC’s hands at the close of the test year. And he found that the funds were not “available” to DOMAC at that time. He concluded that these funds did not “represent utility capital or capital employed in the public service.” DOMAC agrees that the funds were not in its hands in fact, but insists that they were “available.” The Commission, however, is not required, as a matter of logic, to treat a Cabot Corporation demand note as if it were a bank account. After all, Cabot is DOMAC’s sole shareholder. One might reasonably believe that a right to call for money from a demand deposit with a bank depends solely on the will of the depositor, while the exercise of a similar right to call for money on deposit with a sole shareholder depends in large part, as a practical matter, on the desires of the shareholder. Testimony that it was Cabot which chose to use the demand note system as a method for consolidating and managing its subsidiaries’ funds efficiently tends to support, not to undercut, this distinction between funds on deposit with a bank and funds on deposit with a sole shareholder. DOMAC has not convinced us that the Commission was arbitrary in deciding that these funds were not readily enough available to DO-MAC to count as regulated utility balance-sheet capital.
2. DOMAC attacks the Commission’s decision to allow it a working capital allowance based on five days’, rather than forty-five days’, worth of adjusted annual expenses. The Commission has explained its working capital allowance as follows:
A utility is permitted to include in its rate base an allowance for the cash needed to meet operating expenses for the period during which the utility has provided services to its customers and has not been paid for those services. Since all the operating expenses will eventually be paid out of revenues received by the utility, the need for working capital arises largely from the time lag between the utility’s payment of expenses incurred in the rendition of service and the receipt of payments therefor.
Pennsylvania Power Co., 12 FERC ¶ 61,-049, at 61,078 (1980). DOMAC points to the Commission’s rule that the cash working capital allowance is to include “an amount equivalent to one-eighth of annual operating expenses, as adjusted____” 18 C.F.R. § 154.63(f) (Statement E). (“One-eighth” of annual operating expenses equals forty-five days’ worth.) DOMAC also argues that the Commission follows a fixed rule of departing from its “45-day rule” only where “a fully developed and reliable lead-lag study is available in the record” to establish a different gap between receipt of revenues and payment of expenses, Carolina Power & Light Co., 6 FERC ¶ 61,154, at 61,296 (1979). The parties concede the absence of a fully developed study here. Hence, DOMAC, noting that an agency must follow its own rules, see, e.g., Service v. Dulles, 354 U.S. 363, 77 S.Ct. 1152, 1 L.Ed.2d 1403 (1957); Arizona Grocery Co. v. Atchison, T. & S.F. Ry. Co., 284 U.S. 370, 52 S.Ct. 183, 76 L.Ed. 348 (1932), concludes that the Commission must apply its “45-day rule.”
*1219An agency, however, is free to interpret, to supplement, to revise, even to depart from, a previously existing rule, provided that it offers an adequate, and adequately supported, explanation. See, e.g., United States v. Caceres, 440 U.S. 741, 754 n. 18, 99 S.Ct. 1465, 1472 n. 18, 59 L.Ed.2d 733 (1979); American Farm Lines v. Black Ball Freight Service, 397 U.S. 532, 538-39, 90 S.Ct. 1288, 1292-93, 25 L.Ed.2d 547 (1970); Mississippi Valley Gas Co. v. FERC, 659 F.2d 488, 500-02 (5th Cir.1981); cf. National Conservative Political Action Committee v. Federal Election Commission, 626 F.2d 953, 958-59 (D.C.Cir. 1980) (per curiam) (“Agencies are under an obligation to follow their own regulations, procedures, and precedents, or provide a rational explanation for their departures.”). In this case, DOMAC’s customer, Boston Gas, argued to the AU that DO-MAC should receive no working capital allowance at all, for, unlike most firms, DO-MAC bills and receives revenues in advance. The record reveals factual controversy concerning the impact of this practice on DOMAC’s requirements for working capital. But the AU expressly stated that he relied, as to the relevant facts, upon DOMAC’s own witnesses, and he resolved all factual controversies in DOMAC’s favor. DOMAC here points to no evidence, nor to any offer of evidence that was (or might be) made, suggesting that the AU’s particular factual findings were wrong (unless perhaps wrongly favorable to DO-MAC). DOMAC points to no fact suggesting that it needs more than a five-day working capital allowance to cover any gap between the time its bills fall due and the time it receives sufficient revenues from customers to cover those bills. What need is there, then, from its perspective, for a fully developed “lead-lag” study? Or, more to the point, what prejudice did it suffer from the study’s absence?
It does not seem unreasonable to us for the Commission to modify, or to interpret, its rules to allow dispensing with a lead-lag study where evidence in the record, produced by the regulated firm, shows that a lesser working capital allowance is all that is needed, where all significant factual disputes are resolved in the regulated firm’s favor, and where the regulated firm does not point to facts suggesting significant prejudice flowing from the failure to conduct the study. Cf. Pennsylvania Power Co., 12 FERC at 61,079 (“The Commission ... has never considered the 45-day formula to be an ‘irrefutable approximation’ of the utility’s cash working capital needs. Where the utility’s actual cash working capital needs are shown, the allowance will be set accordingly.”) (footnotes omitted). Nor do we see prejudice to DOMAC in the Commission’s applying this interpretation or modification of its basic rules in this case. For these reasons, the five-day allowance is not arbitrary.
3. On appeal from the AU to the Commission, Boston Gas argued that the AU wrongly included a figure of $393,768 when he calculated another component of DOMAC’s need for working capital. This figure represented an insurance premium that DOMAC had prepaid on Oct. 1, 1978— one day after the thirteen-month period relevant to calculation of the firm’s need for working capital for insurance prepayments had expired. See 18 C.F.R. § 154.-63(f) (Statement E) (firm may include in working capital “an allowance for the average of 13 monthly balances of ... prepayments”). The Commission agreed with Boston Gas and excluded the sum, effectively lowering the figure reflecting DO-MAC’s monthly working capital needs (for insurance purposes) by roughly $30,000 (from $230,870 to $200,580).
DOMAC argues that the Commission was required to include the October 1 payment in its calculations by a different provision of the same regulation, which creates a test year of
12 consecutive months of most recently available actual experience, adjusted for changes in ... costs which are known and are measurable with reasonable accuracy at the time of the filing, and which will become effective within nine months after the last month of available *1220actual experience utilized in the filing
18 C.F.R. § 154.63(e)(2)(i). DOMAC says that the October 1 payment was obviously “known and ... measurable with reasonable accuracy” at the time of filing; thus it should have been included in the calculation.
The thirteen-month “prepayment calculation” period seems to us, however, designed as an inevitably somewhat arbitrary — but simple and reasonable — way of trying to estimate a firm’s likely monthly working capital needs. Viewed as such, it is unreasonable to think of the more general “known and measurable” provision as automatically giving a firm an extra nine months’ worth of payments to throw into the calculation simply at the firm’s request. At a minimum, the firm should provide some reason for believing that looking at months after the original thirteen will create a more accurate average. DOMAC provided the Commission no reason at all. (And its argument on appeal here — that the October 1 payment would have been made on September 30 but for an oversight— comes late in the day. Cf. 15 U.S.C. § 717r(b) (“No objection to the order of the Commission shall be considered by the court unless such' objection shall have been urged before the Commission____”).) At the least (and on the basis of the cursory discussion of the issue here and before the Commission), we believe the Commission could reasonably have viewed DOMAC as failing to justify reliance on insurance expenses incurred outside the test year.
4. As the third formula in Section I of this opinion makes clear, after a regulator determines the firm’s revenue requirement, it typically sets a price that, when multiplied by the number of units to be sold, will produce that revenue for the firm. Obviously, it is of great importance to have a reasonably accurate estimate of how many units will likely be sold; otherwise, rates may be set too high (or too low) in respect to the revenue requirement. DOMAC, on the basis of its test year (October 1977 — September 1978) sales of 11.8 trillion btu’s and its substantially higher level of sales in the following nine-month adjustment period (at an annual rate of about 22.6 trillion btu’s), argued that it would likely sell 22.1 trillion btu’s of natural gas annually. Boston Gas argued that DOMAC would likely sell 37 trillion btu’s. The Commission’s staff argued for a figure of 32 trillion btu’s. By the time the AU reached his decision on this issue, he had before him DOMAC’s sales figures for the year ending June 30, 1980. This period, which began immediately after the nine-month adjustment period, substantially overlapped the “locked-in” period to which the rates he was considering were to apply. The actual sales figure for that year almost exactly matched the FERC staff’s estimate. With the guidance of this more current information, the AU adopted the staff’s recommendation.
DOMAC argues that the evidence was inadequate to justify a departure from its lower sales estimate (and consequent higher rate), which everyone admits was reasonably supported by the actual test-year and correction-period sales. We do not agree. The later 1980 figures — arising after the test year — provide an adequate basis for the Commission to conclude that DOMAC’s estimate was not a reasonably accurate indicator of likely actual sales during the locked-in period. Case law does not rigidly tie a regulator to the use of test-year figures, when' later information reveals that the estimates based on those figures are likely to be seriously in error. See Indiana & Michigan Municipal Distributors Association v. FERC, 659 F.2d 1193, 1198 (D.C.Cir.1981) (citing Indiana Municipal Electric Association v. FERC, 629 F.2d 480, 485 (7th Cir.1980)). Indeed to fail to adjust past figures may well lead to serious mistakes, creating rates radically different from those that would replicate costs or serve other valid regulatory purposes. In the present case, use of DO-MAC’s 22.1 trillion btu figure would have produced rates nearly 45 percent higher than those based on the 32 trillion btu figure the Commission selected. The Com*1221mission was not unreasonable in using the later information to determine likely sales during the period covered by the proposed rates.
5. DOMAC complains of the Commission’s decision that it must “share” with its customers revenues it obtained from selling “cool-down” services to LNG tankers — services that involve sea-testing the LNG tankers’ cryogenic systems. DOMAC evidently provided these services to four tankers, the Leo, the Libra, the Taurus, and the Virgo, in December 1978, April 1979, August 1979, and December 1979, respectively. Contrary to DOMAC’s argument, we believe the Commission could reasonably have found that DOMAC provided these services in part with its “natural gas facilities,” namely its docking, terminalling, and mooring facilities and its “gas” personnel. Since the full cost of these “jurisdictional” facilities is charged to its “jurisdictional service” customers when they buy gas, it is reasonable for the Commission to deduct from the gas rates an amount that reflects revenues that DOMAC obtains from using these facilities for the “non-jurisdictional” cool-down work. See, e.g., Public Utilities Commission of Colorado v. FERC, 660 F.2d 821, 826 (D.C.Cir.1981), cert. denied, 456 U.S. 944, 102 S.Ct. 2009, 72 L.Ed.2d 466 (1982); Public Service Co. of New Mexico, 16 FERC ¶ 63,040 (1981), aff'd, 18 FERC ¶ 61,276 (1982). The Commission essentially decided that the cool-down service revenues should be split, 50-50, between DOMAC and its customers, in order to give DOMAC an incentive to sell the use of temporarily idle gas facilities while also sharing the proceeds of those sales with its gas customers. We find adequate support in the record for this determination.
The Commission acted on this determination by ordering a “refund” to gas customers of half the revenues obtained from DOMAC’s previous sale of cool-down services for the four ships just mentioned. And here we agree with DOMAC’s objections. We do not understand the legal basis for the order — at least in the pre-July 1979 cases of the Leo and the Libra. DO-MAC filed for a rate increase under section 4-of the Natural Gas Act, 15 U.S.C. § 717c, in January 1979; its new rates took effect subject to refund in July 1979. In determining whether the new rates were justified, the Commission could properly consider DOMAC’s income from cool-down services for the Leo and the Libra, which occurred during the test-year adjustment period. And the Commission has the power to award refunds to gas customers paying under the new rates insofar as the increases the utility proposed turn out not to be justified. But how can it award a refund for rates paid prior to the time the increase took effect, i.e. prior to July 1979? Section 4 does not authorize a refund of payments made before the time of the rate increases. See Part IV, 1, infra, and Appendix.
Nor can the refund order rest on section 5 of the Natural Gas Act, 15 U.S.C. § 717d. (The text of section 5 is reproduced in the Appendix to this opinion.) Section 5 gives the Commission the power to set just and reasonable rates on its own initiative. But section 5 only authorizes the Commission to set rates prospectively. This provision does not allow the Commission to order a refund of payments made under a rate that was lawful at the time of payment. Cf. Panhandle Eastern Pipe Line Co. v. FERC, 613 F.2d 1120, 1127-33 (D.C.Cir.1979) (FERC cannot circumvent section 5 by requiring adjustment of established rates as condition on approval of requested certificate for different services), cert. denied, 449 U.S. 889, 101 S.Ct. 247, 66 L.Ed.2d 115 (1980).
Neither section 4 nor section 5 authorizes the Commission to make adjustments or to order refunds to compensate customers for past excessive charges by the utility — i.e., charges made prior to the time a proposed rate increase took effect; both are directed only at the reasonableness of prospective or increased rates. And the Commission cites no other authority for its action. Though it is conceivable that something in the ‘settlement’ that governed DOMAC’s *1222pre-July 1979 rate created a special circumstance, this point has not been argued at length. Thus, we believe it appropriate to set aside the Commission’s decision on the four-ship refund and to order reconsideration of the point in light of the principles noted here.
IV
On the basis of its revenue requirement and “reasonable rate” determinations, the Commission has ordered DOMAC to make certain refunds to its customers. DO-MAC’s customer Boston Gas, and DOMAC, attack different aspects of the. Commission’s refund determinations. To understand their attacks, the reader should recall three different time periods and five different rates that are at issue (See Figure 3):
1. The “settlement period,” (prior to July 1979). A ‘settlement agreement’ among the parties governed the rates (“the settlement rates”) charged during this time. The Commission approved the agreement in Distrigas of Massachusetts Corp., 5 FERC ¶ 61,296 (1978), modified, 6 FERC ¶ 61,253 (1979) (Docket No. CP77-216).
2. The “locked-in period, ” (July 1979 to August 1981). This is the period between the time that DOMAC’s 1979 proposed rate increase (“the locked-in rate”) took effect and the time that its still higher, proposed 1981 rate increase took effect. The Commission considered this 1979 proposed increase in Docket RP79-23. It issued its decision, No. 178, on June 23, 1983, specifying a rate for the locked-in period (“the just and reasonable rate”). That “just and reasonable” rate was substantially lower than both the interim “locked-in rate” and the prior “settlement rates.” Distrigas of Massachusetts Corp., 23 FERC ¶ 61,416, rehearing denied, 24 FERC ¶ 61,250 (1983). It is this decision that we have here reviewed so far.
3. The “future period, ” (after August 1981). Most of the issues concerning DOMAC’s post-August 1981 rates were settled by an agreement and stipulation approved by the Commission in July 1982. Distrigas of Massachusetts Corp., 20 FERC ¶ 61,073 (1982). The parties agreed to a rate level (“the stipulated future rate”) to be charged until the Commission decided RP79-23. And they agreed further that the decision in RP79-23 would resolve all but one (rate of return) of the five remaining disputed issues for the “future period,” as well as for the “locked-in period.” When the Commission decided RP7923, DOMAC would change its current charges accordingly (to the “authorized future rate,” an approximation of a “just and reasonable” rate) and make any required refunds.
Both Boston Gas’s and DOMAC’s disagreements with the Commission’s handling of the refund questions concern the extent to which the original 1978 “settlement rates” constitute a “floor” below which the Commission cannot order DOMAC to make refunds. (See Figure 3.)
*1223FIGURE 3
[[Image here]]
The Commission’s decision of the issues before it in RP79-23 — i.e., the issues reviewed in the bulk of this opinion — led to the conclusion that the “just and reasonable” rate for the “locked-in” period (1979— 81) was lower than the settlement rate, i.e., the 1978-79 rate. That is to say, DO-MAC’s 1979 rate increase was not justified because rates should have been lower, not higher. Still in ordering a rate refund, the Commission only required DOMAC to return charges in excess of the pre-1979 rates. Boston Gas challenges this use of the settlement rate as a floor. It believes the refund should have reflected the larger difference between the “locked-in rate” and the newly determined “just and reasonable” rate.
In its subsequent denial of rehearing, the Commission “clarified” its prior decision to dictate a different approach to refunds for the “future” post-1981 period. With regard to that period, the Commission ordered DOMAC to make refunds reflecting the full difference between the authorized future rate and the stipulated rate DOMAC had actually been collecting, despite the fact that the authorized future rate was below the settlement rate “floor.” It is this decision that DOMAC challenges. We consider the two attacks in turn.
*12241. For the locked-in period, the Commission held that the original settlement rates provided a floor beneath which DOMAC could not be required to make refunds, despite the Commission’s conclusion that the “just and reasonable” rate for that period was substantially below that floor. The limitation has the effect of allowing DOMAC to retain 1979-81 earnings in excess of the just and reasonable rates that the Natural Gas Act prescribes. Boston Gas attacks this limitation. It argues that the Commission should have ordered a refund of all that DOMAC collected in 1979-81 above what was “reasonable.” The difficulty with Boston Gas’s argument, however, is that the relevant statute provides, and the Supreme Court has held, to the contrary.
The Commission’s power to order refunds comes from section 4(e) of the Natural Gas Act, 14 U.S.C. § 717c(e). (See Appendix for full text of section 4(e).) That section deals with rate increases that a regulated firm proposes. It provides that, after five months, these increases shall take effect on the firm’s motion. While Commission review proceedings continue, the Commission can require the firm to “keep accurate accounts in detail of all amounts received by reason of such increase ____” Upon completion of the review proceedings, the Commission may order the firm “to refund, with interest, the portion of such increased rates or charges by its decision found not justified.” The Supreme Court has specifically stated that the pre-existing lawful rate provides a refund floor in a section 4 proceeding. Federal Power Commission v. Sunray DX Oil Co., 391 U.S. 9, 22-25, 88 S.Ct. 1526, 1533-1535, 20 L.Ed.2d 388 (1968). It derived this conclusion from the language of the statute and from the fact that, otherwise, a firm asking for an increase could end up considerably worse off than if it had not requested one. And, ironically, the firms least likely to be earning monopoly profits would be the firms most exposed to past rate scrutiny (for they are the firms most likely to see a need to ask for a rate increase).
We find Boston Gas’s efforts to distinguish Sunray unconvincing. The Supreme Court’s language clearly interprets section 4’s refund authority as limited to the “amounts received by reason of such increase.” 15 U.S.C. § 717c(e) (emphasis added). We also find the logic of Sunray convincing and note that it is broadly consistent with the traditional language of similar statutes in the ratemaking area. Cf. Commonwealth of Massachusetts, Department of Public Utilities v. United States, 729 F.2d 886 (1st Cir.1984) (listing similarly structured statutes). But, of course, we would be bound by Sunray even were this not so.
2. With regard to the future period, the Commission declined to follow the Sunray approach. It ordered refunds of the full difference between the stipulated future rate, which DOMAC had been charging, and the authorized future rate, which was below the original settlement rate. The reason for this difference in treatment is a simple one. The Commission found that DOMAC had agreed to the larger refund; it had agreed to set aside the pre-existing floor. It did this in the future-period settlement agreement itself, which includes the following provision:
C. Rate Reductions and Refunds
Within forty-five (45) days of a final Commission order on any of the issues reserved ...,
(1) without regard to the rate levels which have been established in Docket Nos. CP77-216, et al., or may be established in Docket Nos. RP79-23, et al., DOMAC shall file a rate reduction (to be effective August 2, 1981) and shall make refunds, with interest, to customers ... as may be indicated by the Commission’s determination on the reserved issue____
Stipulation and Agreement in Settlement of Rate Proceeding RP81-34, Art. I, § 5 (March 19, 1982) (emphasis added).
DOMAC argues that this language should not be read to waive the Sunray *1225restriction of refunds to the level fixed by the settlement of Docket No. CP77-216. It says that the words “without regard to the rate levels which have been established” is meant only to recognize that the Commission was to decide whether the CP77-216 rate (the settlement rate) would act as a floor in the RP79-23 decision (for the locked-in period), and that the Commission’s decision concerning the applicability of Sunray to the locked-in period would then apply to the future period as well. But the parties noted specifically in the Agreement what issues remained to be decided by the Commission. They listed demurrage, cool-down revenues, deferred tax liabilities, minimum bills, and rate of return. See Distrigas of Massachusetts Corp., 20 FERC ¶ 61,073, at 61,155 & n. 1 (1982). They did not mention anything about a “refund floor” issue. Moreover, it makes sense that parties to a settlement, uncertain whether the settlement rate is actually “just and reasonable” might leave open the possibility of a lower rate in the future and refunds for the past, should the Commission find that a lower rate was “just and reasonable.” At least one party to the agreement presumably would wish such a provision and would wish it not to be limited by reference to a pre-existing rate. Since the language of the Agreement favors the Commission’s interpretation, and that interpretation is consistent with a plausible purpose of the agreement, we accept the Commission’s view of the matter.
DOMAC also argues that the Commission does not have the power to enforce the agreement as interpreted, for, to do so, in DOMAC’s view, would give it the power to set aside the “rate floor” that section 4 provides. “Can the parties’ private agreement empower the Commission to ignore the law?” asks DOMAC. Put so pejoratively, one is tempted to say, “Of course not,” but the true answer is, “It depends.” Obviously the parties can, by agreeing on rates, allow the Commission to dispense with certain statutory requirements, those for a hearing, for example. See, e.g., Placid Oil Co. v. Federal Power Commission, 483 F.2d 880, 893-94 (5th Cir.1973), affirmed sub nom. Mobil Oil Corp. v. Federal Power Commission, 417 U.S. 283, 94 S.Ct. 2328, 41 L.Ed.2d 72 (1974); Pennsylvania Gas & Water Co. v. Federal Power Commission, 463 F.2d 1242, 1245-51 (D.C.Cir.1972); City of Lexington v. Federal Power Commission, 295 F.2d 109, 119-22 (4th Cir.1961). There is ample authority to the effect that a utility and its customers can agree upon rates, that the parties can embody the agreed rates in a tariff, and that a commission can subsequently enforce them. See, e.g., In re Hugoton-Anadarko Area Rate Case, 466 F.2d 974 (9th Cir.1972); Pennsylvania Gas & Water Co. v. Federal Power Commission, supra; Texas Eastern Transmission Corp. v. Federal Power Commission, 306 F.2d 345 (5th Cir.1962), cert. denied sub nom. Manufacturers Light & Heat Co. v. Texas Eastern Transmission Corp., 375 U.S. 941, 84 S.Ct. 347,11 L.Ed.2d 273 (1963). We see nothing in the law that would prevent the parties from agreeing to a rate refund, conditioned upon specified future events, should they choose to do so (at least where the refund is not otherwise unlawful, as, for example, when the resulting rate is discriminatory, 15 U.S.C. § 717c(b); see, e.g., Wight v. United States, 167 U.S. 512, 17 S.Ct. 822, 42 L.Ed. 258 (1897) (construing section 2 of Interstate Commerce Act, 49 U.S.C. § 2, recodified at 49 U.S.C. § 10741(a)), or unreasonably low). Thus, we conclude that the Commission can enforce DOMAC’s agreement to pay the larger refund — and that the Commission is reasonable in interpreting the agreement here so to provide.
In sum, we find that the Commission’s rulings in this case are reasonable and adequately supported by the record, with two exceptions. First, the Commission cannot reduce DOMAC’s rate base by the amount of the firm’s deferred tax liabilities. And second, it must reconsider and properly justify its treatment of DOMAC’s revenues from providing cool-down services to LNG tankers. We remand the case to the agency for further proceedings in accordance with our discussion of these two issues.
*1226 Affirmed in part; vacated in part; remanded to the Commission for further proceedings.
APPENDIX
Excerpts from the Natural Gas Act
1. Section 4, 15 U.S.C. § 717c
Rates and charges; schedules; suspension of new rates
(a) All rates and charges made, demanded, or received by any natural-gas company for or in connection with the transportation or sale of natural gas subject to the jurisdiction of the Commission, and all rules and regulations affecting or pertaining to such rates or charges, shall be just and reasonable, and any such rate or charge that is not just and reasonable is declared to be unlawful.
(d) Unless the Commission otherwise orders, no change shall be made by any natural-gas company in any such rate, charge, classification, or service, or in any rule, regulation, or contract relating thereto, except after thirty days’ notice to the Commission and to the public. Such notice shall be given by filing with the Commission and keeping open for public inspection new schedules stating plainly the change or changes to be made in the schedule or schedules then in force and the time when the change or changes will go into effect. The Commission, for good cause shown, may allow changes to take effect without requiring the thirty days’ notice herein provided for by an order specifying the changes so to be made and the time when they shall take effect and the manner in which they shall be filed and published.
(e) Whenever any such new schedule is filed the Commission shall have authority, either upon complaint of any State, municipality, State commission or gas distributing company, or upon its own initiative without complaint, at once, and if it so orders, without answer or formal pleading by the natural-gas company, but upon reasonable notice, to enter upon a hearing concerning the lawfulness of such rate, charge, classification, or service; and, pending such hearing and the decision thereon, the Commission, upon filing with such schedules and delivering to the natural-gas company affected thereby a statement in writing of its reasons for such suspension, may suspend the operation of such schedule and defer the use of such rate, charge, classification, or service, but not for a longer period than five months beyond the time when it would otherwise go into effect; and after full hearings, either completed before or after the rate, charge, classification, or service goes into effect, the Commission may make such orders with reference thereto as would be proper in a proceeding initiated after it had become effective. If the proceeding has not been concluded and an order made at the expiration of the suspension period, on motion of the natural-gas company making the filing, the proposed change of rate, charge, classification, or service shall go into effect. Where increased rates or charges are thus made effective, the Commission may, by order, require the natural-gas company to furnish a bond, to be approved by the Commission, to refund any amounts ordered by the Commission, to keep accurate accounts in detail of all amounts received by reason of such increase, specifying by whom and in whose behalf such amounts were paid, and, upon completion of the hearing and decision, to order such natural-gas company to refund, with interest, the portion of such increased rates or charges by its decision found not justified. At any hearing involving a rate or charge sought to be increased, the burden of proof to show that the increased rate or charge is just and reasonable shall be upon the natural-gas company, and the Commission shall give to the hearing and decision of such questions preference over other questions pending before it and decide the same as speedily as possible.
2. Section 5, 15 U.S.C. § 717d
Fixing rates and charges; determination of cost of production or transportation
*1227(a) Whenever the Commission, after a hearing had upon its own motion or upon complaint of any State, municipality, State commission, or gas distributing company, shall find that any rate, charge, or classification demanded, observed, charged, or collected by any natural-gas company in connection with any transportation or sale of natural gas, subject to the jurisdiction of the Commission, or that any rule, regulation, practice, or contract affecting such rate, charge, or classification is unjust, unreasonable, unduly discriminatory, or preferential, the Commission shall determine the just and reasonable rate, charge, classification, rule, regulation, practice, or contract to be thereafter observed and in force, and shall fix the same by order: Provided, however, That the Commission shall have no power to order any increase in any rate contained in the currently effective schedule of such natural gas company on file with the Commission, unless such increase is in accordance with a new schedule filed by such natural gas company; but the Commission may order a decrease where existing rates are unjust, unduly discriminatory, preferential, otherwise unlawful, or are not the lowest reasonable rates.
(b) The Commission upon its own motion, or upon the request of any State commission, whenever it can do so without prejudice to the efficient and proper conduct of its affairs, may investigate and determine the cost of the production or transportation of natural gas by a natural-gas company in cases where the Commission has no authority to establish a rate governing the transportation or sale of such natural gas.