Orange and Rockland Utilities, a customer of Tennessee Gas Pipeline Company, complains of two aspects of the Federal Energy Regulatory Commission’s treatment of Tennessee’s 1982 rate filing. See Order Affirming and Reversing Initial Decision, Tennessee Gas Pipeline Co. (“Order”), 45 FERC ¶ 61,031, pet. for reh’g *322denied in pertinent part (“Order on Rehearing"), 45 FERC ¶ 61,470 (1988). The first is the Commission’s insistence on a so-called “100% load factor rate” (effectively a rate based on fully allocated or average costs) for interruptible sales; the second, its approval of a higher commodity charge for a special class of gas transmission. We uphold the Commission as to the former and remand as to the latter.
Interruptible Sales Rate
Tennessee’s filing proposed to reduce the rate it charged for interruptible gas sales. The proposed rate would have equalled the commodity component of its rates for ordinary firm customers — its contract demand or “CD” customers. Under the “modified fixed variable” rate design approved for Tennessee, this charge would include all its gas costs, the variable costs of transportation and storage, and part of the fixed costs for the latter — return on equity and related income taxes. Thus interruptible purchasers would not have borne a share of that part of Tennessee’s fixed costs allocated to its demand charge.
The administrative law judge concluded that under the proposed reduced rate firm shippers would subsidize the interruptibles, and that Tennessee had shown no competitive need for the proposed rate (as against competition from alternative energy sources). Instead the ALJ reaffirmed the existing, previously approved 100% load factor rate, which matches what a firm customer would pay per unit of gas (in both commodity and demand charges) if it took 100% of its contract demand. See Initial Decision on Reserved Issue, Tennessee Gas Pipeline Co. (“ALJ Opinion”), 31 FERC ¶ 63,001 at 65,004-07 (1985). The Commission affirmed. Order, 45 FERC at 61,102-04.
For pipeline filings under § 4 of the Natural Gas Act, 15 U.S.C. § 717c, the burden of proof is on the pipeline when it proposes a rate increase. See Sea Robin Pipeline Co. v. FERC, 795 F.2d 182, 183 (D.C.Cir.1986); ANR Pipeline Co. v. FERC, 771 F.2d 507, 513 (D.C.Cir.1985). The Commission kept it there, even though the change proposed was a reduction, on the ground that it carried a possibility of shifting costs to firm customers. See 45 FERC at 61,103. The prospect of cost-shifting would seem to turn on projected volumes. If Tennessee's proposal projected a high enough volume under the lower rate, interruptible service thereunder could make a greater contribution to fixed costs than under the 100% load factor rate, shifting costs away from the firm customers. It is not clear exactly how Tennessee handled the matter, but since no party has contested the allocation of the burden to Tennessee, we assume its validity-
In its petition for rehearing, Orange and Rockland rested primarily on what it viewed as the Commission’s failure to recognize the distinctions between the present case and prior Commission decisions applying a 100% load factor rate. That merely puts on the Commission a burden of explaining why it was extending the earlier decisions; it does not in itself make a case for the proposed alternative rate. Here we find that the Commission gave at least plausible explanations for extension of the prior cases. As neither Tennessee nor the customers seeking a lower interruptible rate undertook to show why the Commission’s approach was fundamentally wrong, we affirm.
Thus, Orange and Rockland points out that whereas in the previous instances in which the Commission imposed the 100% load factor rate interruptible sales weren’t truly interruptible, see, e.g., Public Service Comm’n of the State of New York v. FERC, 813 F.2d 448, 453 (D.C.Cir.1987), here they are “interruptible in the extreme.” To this the Commission had two kinds of answers. First, it invoked its prior decision in Texas Eastern Transmission Corp., 37 FERC ¶ 61,260 at 61,697-61,705 (1986), reh’g denied, 41 FERC ¶ 61,015 at 61,028-30 (1987), aff'd sub nom. Texaco, Inc. v. FERC, 886 F.2d 749 (5th Cir.1989), which acknowledged that quality differences called for rate differences but noted that a 100% load factor rate is different from a standard CD rate: a firm customer would pay as little per unit only if it *323bought its full contract quantity. See Order, 45 FERC at 61,104. In essence it appeared to be saying that as the 100% load factor rate was a genuine advantage, it was unwilling to create further refinements for degrees of interruptibility.
More important, the Commission asserted that because interruptible service used the facilities for which the fixed costs were incurred, it properly bears some of those fixed costs. See id.; see also Order on Rehearing, 45 FERC at 62,471. In fact this does not seem necessarily so. In its regulation of interstate sales of electricity, the Commission takes the view that “those creating the critical need for power [i.e., the peak period users] are assigned the cost responsibility for all capacity-related costs.” Union Electric Co., 40 FERG ¶ 61,046 at 61,141 (emphasis in original), modified on reh’g, 41 FERC ¶ 61,343 (1987), rev’d on other grounds sub nom. Union Electric Co. v. FERC, 890 F.2d 1193 (D.C.Cir.1989). This view embodies the principle that regulators should allocate the costs of facilities to those whose demand causes them to be incurred, in order to provide users with correct incentives in their decisions about whether to invest in alternatives and about what quantities to take and when. See Union Electric Co., 890 F.2d at 1198-1201. If peak-period use determines the size of gas transmission facilities, as one would expect by analogy to electricity generation, then it is not clear why interruptible service — at least if it is genuinely interruptible — should bear any portion of the fixed costs. On the other hand, the Commission has for various reasons qualified its general policy of loading the fixed costs of electricity generation onto peak-period users, see discussion at 890 F.2d at 1198-1200, and similar ones may apply, perhaps even more forcefully, for gas pipelines.
In a policy statement issued after the decision under review here, however, the Commission appeared to adopt a view extremely sympathetic to peak-period pricing. Noting that regulations for unbundled transportation rates called for interruptible and off-peak rates that would “maximize throughput,” it observed that this principle should extend to the transportation portion of bundled sales rates as well. Policy Statement Providing Guidance mth Respect to the Designing of Rates, 47 FERC ¶ 61,295 at 62,052 (1989). Although “maximize throughput” is surely hyperbole (a zero rate would do so, but would not cover variable costs), the policy statement appears to reflect an intention to apply to gas pipeline service something at least approximating the cost causation principles that it uses for electricity. But as neither Tennessee nor Orange and Rockland shouldered the burden of establishing the wisdom or necessity of those principles, we accept the Commission’s adherence to its former view.
Before the Commission, Tennessee and Orange and Rockland also asserted that the higher price seriously impaired the marketability of the interruptible gas. Although there was some vagueness in the AU’s response, which may mirror faults in Tennessee’s and petitioner’s own arguments, he did find that the 100% load factor rate was “competitive with the cost of alternative fuels in Tennessee’s market area.” 31 FERC at 65,006. Of course this is far from a finding that a lower rate would not increase sales — that would imply no price elasticity, which seems highly improbable. But neither Tennessee nor petitioner appears to have quantified the increase that would follow from the proposed reduction. Compare Ozark Gas Transmission System v. FERC, 897 F.2d 548 (D.C.Cir.1990); Natural Gas Pipeline Co., 27 FERC ¶ 61,235 (1984) (allowance of off-system sales at commodity rate based on demonstration of net benefit to system customers). Under the regulatory viewpoint then prevailing for gas sales rates, and not challenged by the parties at any fundamental level, the Commission weighed an uncertain gain in gas sales against a concern about wrongly shifting fixed costs to other customers. See id. On such a view, its decision to hold the line at the 100% load factor rate was not arbitrary.
Although the ALJ made no express finding on the effect of the lower rate on Tennessee’s competitive position with respect to other gas suppliers, again the pre*324vailing policy left the Commission free to decide as it did, at least where the proponent of a change failed to make a strong record on the subject. For one thing, gas-against-gas competition raises special issues not present for .competition against alternative fuels; increased gas usage may not be large enough to justify the reduced collective recovery of fixed costs by gas pipelines. Cf. Associated Gas Distributors v. FERC, 824 F.2d 981, 1019 (1987). And the Commission itself went on to observe that if Tennessee wished to develop a program for discounted interruptible sales, it could file for a certificate of convenience and necessity under § 7 of the Natural Gas Act, under which it could charge different rates for different customers. See Order on Rehearing, 45 FERC at 62,471. Although such a proposal would not have the procedural advantage of taking effect automatically on filing (subject to the Commission’s § 4 power to suspend for five months), the suggestion manifests a Commission recognition of the appropriateness of sales below the 100% load factor rate for purposes of meeting competition. It also provided Tennessee with a device for improving marketability with less impact on interruptible service’s contribution to fixed costs (because of the selectivity of the discounts), thereby assuaging the Commission’s then strong anxiety on the subject.
Rate for Transmission of Storage Gas
Tennessee’s storage service customers buy CD gas from Tennessee during the summer, store it with Tennessee at a point along the way, and then withdraw it during the winter. The gas travels the same distance as unstored gas. The arrangement benefits customers by enabling them to spread their purchases out over the year; they may as a result nominate a level of contract demand closer to the level of gas actually taken, thereby reducing their demand charges (effectively the charges made for claims on pipeline capacity). The pipeline presumably also benefits by being enabled to increase the total usage of its facilities.
At issue here is the way in which storage customers are charged for the transmission involved in the downstream leg of the journey, from storage to point of ultimate delivery. While the arithmetic is unclear to us, it appears undisputed that for this transmission, the storage customers pay a higher unit rate for transportation than do CD customers for firm service supplied in the peak period.
The record is particularly garbled as to the first possible element of the higher rate. The pipeline proposed to allocate the transmission costs of the downstream leg to the storage schedule rather than to the CD sales schedule, and the Commission agreed. This separation may appear a formality, with no real effect on the allocation of costs between storage and other customers. But the Commission itself appears to have believed that more was at stake, for in adopting Tennessee’s approach it rested on a substantive finding “that Tennessee’s pipeline and compressor facilities located downstream from storage were constructed in part to provide capacity for these storage services and that these are additional costs associated with moving gas from storage to the customer that are not associated with service to CD customers who do not use storage service.” Order, 45 FERC at 61,112.
It may be that the Commission approach leads to a kind of double charge for storage customers. Tennessee apparently eliminated any double counting as to its total revenues by subtracting from the CD schedule the costs added to storage service, see ALJ Opinion, 31 FERC at 65,017, but that would not necessarily solve the issue. If the storage customers' initial CD charge covers transmission costs all the way to the point of ultimate-delivery to them, as parts of the record suggest, see 45 FERC at 61,111 (noting Orange and Rockland argument that “deliveries out of storage actually represent the completion of CD sales”); 31 FERC at 65,018 (“the [storage] customer is entitled to final delivery, whether out of storage or not, by virtue of its purchase under the CD rate”), the storage customers would be paying twice for the downstream leg. Not quite twice, of course, because as CD customers they would (along with the *325other CD customers) receive some benefit from the CD rate reduction.
If this is in fact the case, the Commission has not justified it. The fact that the downstream transmission facilities were constructed “in part” for the storage customers does not explain why they should pay more than other customers, for whom the facilities were also constructed “in part.” It is no reason to impose a burden on one class to point to a characteristic shared by the other.
Tennessee also secured Commission approval of a specific method of allocating costs — on the basis of average peak-day deliveries rather than on the basis of average annual deliveries, as it does for transmission costs under the CD sales schedule. Again the arithmetic is obscure, but again it is clear that this causes or contributes to the higher unit rate that storage customers pay for downstream transmission. Here the Commission justification was simply that “storage withdrawal service is clearly rendered on a peak winter basis and warrants an allocation on the basis of peak (not annual) deliveries.” Order, 45 FERC at 61,112.
The theory seems to be customers whose usage is heavily or exclusively loaded in the peak periods should pay a higher average rate than ones whose usage is more evenly spread out. But there are two problems here. First, that result can readily be attained by adopting a differential between peak and off-peak charges, an approach that also has the effect of imposing incentives on all peak-period users that reflect the role of peak usage in cost incurrence. Second, even if for some (as yet undisclosed) reason the Commission is reluctant to use a peak-period differential, in this particular case it has imposed special burdens on the storage customers for a reason — low load factor (i.e., an unevenly distributed load) — that evidently applies to many of the regular CD customers. See ALJ Opinion, 31 FERC at 65,017-18.
We realize that it is hard for the Commission to write simply enough to make its views comprehensible to inexpert judges. But whether because we are inexpert or obtuse, or because the Commission has not expressed itself clearly enough, we are unable to detect reasoned decision-making on this issue. Accordingly, although we affirm as to the interruptible rate, we must remand the case to the Commission as to the treatment of storage costs.
So ordered.