We deal here with a novel effort by a pipeline and its principal customer to provide for retrospective adjustment of rates to reflect current information about the cost of service and the amount and expected lifetime of the reserves to be transported. Finding the Federal Energy Regulatory Commission not to have supported its interpretation of the controlling contract provision with any plausible theory, we grant the petition for review for want of reasoned decisionmaking and remand for further consideration by the Commission.

* * *

* * *

Tarpon Transmission Company owns 40.4 miles of pipeline located in the waters of the Outer Continental Shelf, off the Louisiana shore. The pipeline links reserves primarily owned by Trunkline Gas Company with the latter's offshore pipeline, and Trunkline has been Tarpon's principal customer since Tarpon began transporting natural gas in June 1978.1 A Transportation Agreement, effective since February 15, 1977, specifies the terms of the relationship between Tarpon and Trunkline and, with its amendments, serves as Tarpon's Tariff with the Commission.

The Agreement sets out to modify some of the conventional principles of cost-based ratemaking. Under the traditional approach, rates are based on projected volumes, with the aim that if service in fact proceeds at those volumes the projected revenue will cover the projected costs. For a pipeline in service, the estimation is derived from experience over a "test-year," with certain adjustments. See 18 C.F.R. Sec. 154.63(e)(2) (1988) (explaining the test-year methodology and adjustments). Where application of this method calls for a change in rates, the change is prospective only. Thus, if it appears in the third year of a project that the throughput has been and will likely be only half the anticipated level, entailing a higher unit rate (so long as all other things are unchanged), no adjustment will be made to enable the pipeline owner to catch up for the under-recovery of costs during the past period. Equally, of course, no retrospective adjustments will be made for excess recoveries due to unexpectedly high volumes. The rates for future volumes transported will, however, reflect the new information.

Article X of the Agreement establishes a special methodology. Section 10.1 establishes a "life-of-the-reserves" method for setting the "initial unit rate." The total cost of service for the estimated life of the reserves (including operating and maintenance expenses, taxes, depreciation and return on investment) is estimated and then divided by the estimated total reserves. This yields the initial rate per thousand cubic feet ("Mcf"). It is not clear whether this aspect of the parties' arrangement differs significantly from what FERC would typically require, and in any event it is not in dispute.

The clear innovation lies in Section 10.5, which governs changes in rates:

10.5 Rate Adjustment. Trunkline or Tarpon upon the giving of ninety (90) days written notice to the other prior to the end of the second (2nd), fourth (4th), sixth (6th) and eighth (8th) years of the primary term hereof, may request that the unit rate currently being utilized to determine the monthly charge and other charges or credits be decreased or increased to reflect changes in costs and/or gas reserves connected to the system. Such rate redetermination shall be based upon a cost of service for the entire life of the reserves transported and to be transported pursuant to this Agreement and other agreements Tarpon may enter into for the utilization of subject facilities, taking into consideration actual revenues collected to date or to be collected prior to the effective date of such unit rate charge; and such rate shall be calculated in the same manner as used in the calculation of the initial unit rate hereunder.

When Tarpon began operations in 1978, reserves were estimated to have a life of 8.25 years and a volume of 103,573,000 Mcf. These figures and other data as to Tarpon's costs yielded an initial rate of 31.64 cents per Mcf. The Commission approved the rate but required Tarpon to initiate a new rate proceeding within three years. See Tarpon Transmission Co., 59 FPC 1516, 1519 (1977).

Accordingly, in 1981 Tarpon and Trunkline set a new rate of 18.10 cents per Mcf. For no apparent reason, they did not employ Section 10.5, but rather the traditional test-year method. The Commission approved the new rate but again required Tarpon to file a notice of rate change within three years.

On May 25, 1984 Tarpon filed a notice of a rate change with FERC, proposing to reduce its rate to 16.88 cents per Mcf. Again Tarpon used the test-year approach rather than that of Section 10.5. Joint Appendix ("J.A.") 473.2 On this occasion, however, the Commission found that Tarpon had not demonstrated that the proposed rate was just and reasonable as required by the Natural Gas Act, 15 U.S.C. Sec. 717c(a) (1982). It allowed the rate to take effect on July 10, 1984, subject to a refund if Tarpon could not prove the rate was just and reasonable in subsequent rate proceedings. Tarpon Transmission Co., 28 FERC p 61,027 (1984). Dispute over details in the application of the traditional method were mooted, however, when all concerned concluded that Section 10.5 of the Agreement should control.3

An administrative law judge accepted the interpretation of Section 10.5 advanced by Tarpon and ruled that the proposed rate of 16.88 cents was just and reasonable. Tarpon Transmission Co., 32 FERC p 63,020 (1985). Although the Section 10.5 approach (as construed by Tarpon) yielded a rate of about 25 cents per Mcf,4 Tarpon claimed only the rate originally asserted in its 1984 Notice of Rate Change, and accordingly that was what the ALJ approved. Here, too, Tarpon claims only the 16.88 cents.

The Commission's staff filed exceptions to the ALJ's decision, and the Commission reversed. Tarpon Transmission Co., 41 FERC p 61,044 (1987). Although the Commission agreed that the rate should be determined according to Section 10.5, it accepted the staff's interpretation of the contract provision. Id. at 61,136. Under this interpretation, Tarpon's proposed rate was not just and reasonable, and FERC ordered Tarpon to recalculate the rate using the staff's interpretation. That approach yielded a rate of 4.02 cents per Mcf.

The Commission denied Tarpon's motion for a rehearing, Tarpon Transmission Co., 42 FERC p 61,050 (1988), and this appeal followed.

Although the interpretation of Section 10.5 is a matter of contract law, we nonetheless owe deference to the Commission's view. In National Fuel Gas Supply v. FERC, 811 F.2d 1563, 1569 (D.C.Cir.), cert. denied, --- U.S. ---- 108 S. Ct. 200, 98 L. Ed. 2d 151 (1987), this court read Chevron U.S.A. Inc. v. NRDC, 467 U.S. 837, 104 S. Ct. 2778, 81 L. Ed. 2d 694 (1984), as requiring judicial deference to the Commission's interpretation of language in a settlement agreement resolving certain rate disputes. The court identified two factors militating in favor of deference. First, Congress explicitly delegated to FERC broad powers over ratemaking, including the power to analyze relevant contracts. 811 F.2d at 1569. In this case, the Commission was involved with the Agreement from the time of its initial approval of the contract as Tarpon's tariff.

Second, the Commission has greater technical expertise relevant to the interpretive problem. Id. at 1570; Southern California Edison Co. v. FERC, 805 F.2d 1068, 1072 (D.C.Cir.1986). Under National Fuel the Commission's expertise calls for deference even though Section 10.5 is novel and the agency is not experienced in interpreting its language. There the court rejected the idea that deference was appropriate to an agency's contract interpretation only when it was clear that the interpretation rested on expertise. "We think that the better view ... is that deference should be given because the congressional grant of authority to the agency indicates that the agency's interpretation typically will be enhanced by technical knowledge." National Fuel, 811 F.2d at 1570.

Nonetheless, the courts still play a vital role in reviewing an agency interpretation of a technical contract provision. To fulfill that role, this court must ask whether the interpretation is "amply supported both factually and legally." Vermont Dep't of Public Service v. FERC, 817 F.2d 127, 134 (D.C.Cir.1987). We cannot accept an agency determination unless it is the result of reasoned and principled decisionmaking that can be ascertained from the record. Columbia Gas Transmission Corp. v. FERC, 628 F.2d 578, 593 (D.C.Cir.1979). To do otherwise would transform judicial review into a process of rubberstamping unsupported, and perhaps unsupportable, agency decisions.

To meet the requirement of reasoned decisionmaking, the agency must set forth the basis of the decision "with such clarity as to be understandable. It will not do for a court to be compelled to guess at the theory underlying the agency's action; nor can a court be expected to chisel that which must be precise from what the agency has left vague and indecisive." SEC v. Chenery Corp., 332 U.S. 194, 196-97, 67 S. Ct. 1575, 1577-78, 91 L. Ed. 1995 reh'g denied, 332 U.S. 783, 68 S. Ct. 26, 92 L. Ed. 367 (1947). Here we conclude that the absence of an adequate analysis by the Commission renders meaningful review impossible.

Tarpon views Section 10.5 as establishing a two-step process to adjust its rates. The first step is a recalculation of the cost of service for the entire life of the reserves that either have been transported or are predicted to be transported through the pipeline since the date of the initial deliveries. The recalculation takes account of all current information concerning costs and expected revenues. For our purposes, the most critical changes are in (1) the volumes expected to be carried and (2) the period of the pipeline's useful life. Although the original estimate of volumes had been about 103 million Mcf in 1978, in fact 148 million Mcf had been transported by 1984 and Trunkline estimated that over 105 million still remained to be transported. And although in 1978 the reserves were expected to be exhausted in 8.25 years, by 1984 they were expected to last until 1999, making a total of 21.5 years from the start of service.

At this stage of the analysis, both these changes tend to increase the total cost to be recovered by the pipeline operator. The increased reserves tend to increase total cost because higher volume increases some expenses, such as the energy costs of transmission. More important, for purposes of the present dispute, is the tendency of the increase in the reserves' expected lifetime to increase the total amount to which Tarpon is entitled as a return on investment. The stretch-out to 21.5 years slows the rate of depreciation; as Tarpon is entitled to a return on undepreciated rate base, this increases the aggregate dollars to be collected as return on that base.

So far as this second effect is concerned, we note two points. First, the increase is in aggregate dollars recovered, not discounted for delay in recovery. Because the change in expected life slows the rate of depreciation, it delays the pipeline company's recovery of its initial investment. Unless the allowed return on investment is sufficient, there is no reason to suppose that the stretch-out, even with its increased flow of undiscounted dollars, will improve Tarpon's economic position. Second, precisely because of this point, we asked Tarpon's counsel at oral argument whether the stretch-out could truly enhance its position unless the allowable rate of return exceeded its true cost of capital. Counsel denied that this was necessary, but we confess ourselves uncertain as to why.5

The second step, as Tarpon reads the Agreement, is to apply Section 10.5's revenue-crediting provision. Tarpon would subtract the sum of revenues already collected prior to the effective date of the rate change from the newly-computed total cost of service. The goal of this provision is to give Trunkline credit for any excess rates paid because of imperfect information in 1978. J.A. at 262-63. Finally, to obtain a new unit transportation rate, Tarpon would divide the resulting figure (i.e., the portion of the estimated total cost of service not already received by Tarpon) by the estimated remaining recoverable reserves. In this last computation, of course, the increase in reserves tends to decrease the resulting unit rate. (Thus the 25-cent unit rate yielded by Tarpon's analysis is lower than its original 31.64-cent rate.)

A table submitted by one of Tarpon's experts may help elucidate its approach:

      1.        Total Cost of Service
                6/10/78-12/31/99                $ 66,291,334
      2.        Less Revenues from Inception
through 7/9/84  41,601,015
      3.        Cost to be Recovered              24,690,319
                                                ------------
                                                ------------
      4.        Remaining Recoverable Reserves
 7/10/84--Mcf   105,691,526                     ------------
                                                ------------
      5.        Unit Cost--cents per Mcf               23.36
                                                ------------
                                                ------------

J.A. at 291.

Through Frank S. McGee, Tarpon's Executive Vice-President and the only witness who had participated in the negotiation of the Agreement, J.A. at 232, Tarpon offered a theory of Section 10.5. He testified that the drafters intended it to allow the parties to alter the terms of their relationship periodically so the Agreement would approximate the arrangement they would have reached had they possessed perfect foresight in 1978. J.A. at 262-63. The parties realized that their initial estimates concerning costs and reserves were "not very firm," J.A. at 184; thus they constructed a method by which they could adjust the rates as new information became available. Obviously the fact that the pipeline was originally intended only for the exploitation of a single set of reserves made the risks of misestimation acute.

Tarpon's interpretation of Section 10.5 reflects this justification. All costs of service, including the depreciation rate applied to its capital investments, are recalculated so as to approximate the agreement that the parties would have reached at the outset had they operated with perfect information about the future. By virtue of the revenue-crediting provision, Tarpon recovers the same sum in undiscounted dollars that it would have recovered if the parties had initially estimated a 21.5-year life and 250 million Mcf in total reserves. In economic effect, of course, the equivalence is not exact. The revenue-crediting provision contains no adjustment for the value enjoyed by Tarpon by virtue of its collecting a greater proportion of the aggregate dollar amount in the early years of the pipeline's total lifetime. Tarpon's theory allows it to enjoy an interest-free loan on such acceleratedpayments. Estimation errors in the other direction, of course, would have had the opposite effect.

The ALJ accepted Tarpon's interpretation, noting "the consistency and equity underlying [Tarpon's] theory of Section 10.5 and the fairness of the result produced." 32 FERC p 63,020, at 65,060.

The Commission, however, rejected the ALJ's view and found that "Tarpon's tariff is reasonably susceptible to only one interpretation"--that espoused by the Commission's staff. 41 FERC p 61,044, at 61,135.6 It agreed with a general form of Tarpon's interpretation of Section 10.5, finding that its purpose was to enable Tarpon to collect "Tarpon's actual cost of service for the life of the project." Id. Thus FERC gave effect to the revenue-crediting portion of the Section and reduced the rate that Tarpon was entitled to receive by the excess of Tarpon's pre-1984 revenues over its recalculated pre-1984 costs--an excess evidently due primarily to the excess of actual over anticipated throughput. J.A. at 383.

The critical difference, however, was that the Commission did not allow any retrospective calculation of depreciation to account for the stretch-out in the pipeline's useful life. Whereas under Tarpon's calculation there would be undepreciated plant investment of about $9.5 million in 1984, J.A. at 285, the Commission calculates undepreciated investment at that date as $700,000, id. at 313.

A generous reading of the record suggests three possible justifications for the Commission's holding, none of which rises to the level of reasoned decisionmaking. First, the Commission's staff appear to have been guided by traditional ratemaking principles in interpreting Section 10.5. For example, an economist in the Depreciation Branch of FERC's Office of Pipeline and Producer Regulation's Division of Pipeline Rates, Patrick Crowley, testified that changes in the depreciation rate must be made prospectively only. "[I]f we were to go back in time to 'correct' our earlier estimates and their associated effects on the cost of service, we might find ourselves in a constant state of flux." J.A. at 405. Putting aside the obvious exaggeration implicit in the allusion to constant flux, the observation totally overlooks the self-evident intention of Section 10.5 to depart from traditional ratemaking conventions. Although the staff approach gave effect to part of Section 10.5's innovation, namely the revenue-crediting provision, it and the Commission switched to a conventional approach with respect to the treatment of depreciation. It offered no explanation for this bifurcated application of Section 10.5.

Second, the Commission may have believed its approach was necessary to prevent Tarpon from recovering its capital investment twice. Although its order did not advance this rationale, counsel for the agency did so at oral argument, and the Commission did so in response to Tarpon's earlier request for a stay from this court, Answer of Federal Energy Regulatory Commission in Opposition to Emergency Motion for Stay Pending Judicial Review 7. Such a justification is patently incorrect; both Tarpon's and FERC's applications of Section 10.5 result in the same amount of return of investment. The only difference between the two is the speed at which Tarpon's capital investment will be depreciated and thus the return on investment.

Finally, the Commission responded to Tarpon's theory of Section 10.5 in purely conclusory terms. Tarpon had argued that the parties had intended to spread "the 'recovery of the cost of the facility evenly over the life of the reserves to be transported.' " 41 FERC p 61,044, at 61,135. The Commission's response was twofold. First, it declared that "prospective adjustment to the depreciation rate," i.e., spreading the $700,000 resulting from prior depreciation over the newly estimated 15 remaining years of the facility, would prevent "under or over recovery of Tarpon's investment." Id. at 61,140 n. 14. But this in no way comes to grips with Tarpon's view that Section 10.5 endeavors to replicate what would have resulted if the volume and lifetime of reserves had been known at the outset. Second, the Commission claimed to find internal inconsistency in Tarpon's view, in that where early out-of-pocket costs have been misestimated,

the tariff would not reallocate a portion of the early expenses to later years to even up their recovery over the life of the project. The tariff merely provides for prospective adjustment of the rates to reflect such a change in estimated future expenses and through the use of past actual expenses and the revenue crediting mechanism for under or over recoveries of out-of-pocket expenses incurred in the past.

Id. at 61,136. We can detect no inconsistency. So far as we can discern, Tarpon advocates (and the Commission allows) complete adjustment for improved information as to out-of-pocket expenses. This fails to explain why it should not allow complete adjustment for new information as to the probable lifetime of the facility.

Intervenor Sun Exploration and Production Company advances a slightly different textual argument in its brief to this Court to prove that Tarpon's interpretation of Section 10.5 is faulty. See Response Brief of Intervenor Sun Exploration and Production Company in Support of Respondent Federal Energy Regulatory Commission 18-19. Sun argues that Tarpon's use of "imputed depreciation expenses"--a rate that differs from the actual depreciation expenses reflected both in the company's books and in its previous rates--is contrary to the requirement in Section 10.5 that any rate adjustment take account of "actual revenues." Sun appears to believe that Tarpon's proposed treatment of Section 10.5 involves "ignor[ing] the actual depreciation expenses collected prior to a rate determination." Id. at 18. Moreover, Tarpon's interpretation allows it the "unprecedented opportunity to continue to recover return on previously recouped investment." Id.

As the Commission has not addressed these contentions, they could not fill the reasoning vacuum. See City of Charlottesville v. FERC, 661 F.2d 945, 950 (D.C.Cir.1981) ("[T]his Court has required the Commission to specify the evidence on which it relied and to explain how that evidence supports the conclusion it reached."). In any event, we note that to characterize Tarpon's method as "ignoring" depreciation collected is highly questionable in light of the revenue-crediting provision. Further, the last clause of Section 10.5 requires that the adjusted rate "be calculated in the same manner as used in the calculation of the initial rate hereunder." In its brief, Tarpon argues that its interpretation of Section 10.5 would give effect to that clause because "it applies a single depreciation rate, based on the total reserves estimated at the time of the rate revision, to Tarpon's initial investment to get appropriate depreciation allowances and return on investment for both past and future periods." Brief of Petitioner 20. Thus, according to this analysis, Tarpon's approach would better serve the goal of allowing the parties to reshape their relationship so as to approximate that which they would have constructed at the outset had they possessed perfect information.

If on reconsideration FERC chooses to rely on Sun's theory, or on a similar analysis, it will have to explain how such an interpretation accords with the parties' desire to adjust the unit transportation rate as to reflect updated information. In other words, any textual argument adopted by FERC should be consistent with some theory of the goals of Section 10.5, a theory that in turn must be grounded in discussion of the text and any relevant evidence.

By requiring the Commission to explain its decision with regard to whether Tarpon's rate proposal is just and reasonable, we do not hold that the outcome is unacceptable. Indeed, given the novelty of this contractual provision, there may well be a number of textual arguments and policy justifications that would support the Commission's interpretation. However, none of them can be found in the record before us; in fact, the only reasoned interpretation of Section 10.5 is offered by Tarpon. We therefore grant the petition for review and remand the case to FERC so that it may re-examine the issue mindful of its obligation to provide a reasoned explanation for any outcome.

Judgment Accordingly.