Section 1801(a) of the Energy Policy Act of 1992 ("EPAct") directed the Federal Energy Regulatory Commission to promptly issue a final rule establishing "a simplified and generally applicable ratemaking methodology for oil pipelines." Pub. L. No. 102-486, 106 Stat. 2776, 3010 (1992), reprinted in 42 U.S.C. § 7172 note. Congress also declared that oil pipeline rates that had not been contested or opposed for one year before October 24, 1992 should be deemed "just and reasonable" for purposes of the controlling statutory mandate, § 1(5) of the Interstate Commerce Act. EPAct, § 1803(a)(1). Under this provision, "the vast majority" of then-prevailing rates were available as a baseline for a "rate cap" system, under which a pipeline could charge any rate within the previously established ceiling. See Revisions to Oil Pipeline Regulations Pursuant to Energy Policy Act of 1992, Order No. 561-A, FERC Stats. & Regs. (CCH) ¶ 31,100 (1994) ("Order No. 561-A") at 31,092. But to protect both pipelines and their customers from being adversely affected by cost changes over time, the Commission provided that the cap would change annually in accordance with a pre-determined formula or index, which would be independent of each pipeline's own cost experience. Revisions to Oil Pipeline Regulations Pursuant to the Energy Policy Act of 1992, Order No. 561, FERC Stats. & Regs. (CCH) ¶ 30,985 (1993) ("Order No. 561") at 30,946. In contrast to traditional rate ceilings based on each regulated firm's own cost experience, use of such a formula gives the pipelines incentives to pursue cost-saving innovations. See id. at 30,948 & n. 37. And if the annual adjustment approximates the likely experience of reasonably efficient pipelines, there will be relatively little need to entertain applications for exceptions based on a divergence of pipeline cost experience from the evolving cap. See Order No. 561-A at 31,092. Accordingly the Commission seeks an index based on the pipelines' aggregate past experience.

In its initial implementation of the EPAct in 1993-94, the Commission applied a methodology that led it to an index of PPI-1, i.e., the annual change in the producer price index, minus one percentage point. See Order No. 561 at 30,951-52. (Technically, it was PPI-FG-1, or producers' price index — finished goods, minus one percentage point.) Pipeline owners challenged the methodology in court, but we upheld the Commission. Association of Oil Pipe Lines v. FERC, 83 F.3d 1424 (D.C.Cir.1996) ("AOPL I"). In 2000 FERC fulfilled an earlier promise to revisit the issue. Although it arrived again at PPI-1, it did so via a new methodology. Again the pipelines sued. This time, finding that the Commission had deviated without adequate explanation from the earlier methodology, and had failed to answer various criticisms of its new approach, we remanded the case for further consideration. We expressly refrained from vacating, noting that PPI-1 might prove sustainable even after FERC's reexamination. Association of Oil Pipe Lines v. FERC, 281 F.3d 239, 247-48 (D.C.Cir. 2002) ("AOPL II"). On remand, rather than defend the methods used in the 2000 approach, FERC re-adopted those used in 1993-94 (with some changes), and chose an index of plain PPI. See Order on Remand: Five-Year Review of Oil Pipeline Pricing Index, 102 FERC ¶ 61,195, 2003 WL 21436759 (February 24, 2003) ("Remand Order"). The present petitioners are shippers, who argue that the Commission was arbitrary and capricious in failing to stick to the innovations made in 2000. They have not made their case, however.

* * *

The shippers attach great importance to our decision merely to remand the 2000 decision, rather than to vacate and remand. From this they infer that the Commission was obliged to really try to justify its 2000 methodology for estimating the rate of change in aggregate pipeline costs, rather than beating a retreat to its 1993-94 approach. We rejected exactly such a claim in Southeastern Michigan Gas Co. v. FERC, 133 F.3d 34 (D.C.Cir. 1998), saying that while further explanation of the approach triggering remand was one of the Commission's options, "once FERC reacquired jurisdiction [via the remand], it had the discretion to reconsider the whole of its original decision." Id. at 38. It would make little sense to force the agency to struggle to uphold a methodology that it had failed to justify, just because it had once tried to do so.

On the merits, the shippers insist that the three innovations that elicited the remand in AOPL II are not merely preferable, but that the alternatives chosen by the Commission on remand are so inferior that for it to return to them was arbitrary. In measuring the changes in pipeline costs between 1994 and 1999, the Commission adopted in 2000 (and abandoned on remand) the following methods: First, it moved from some form of "fixed-weight" method to a "floating-weight" approach. See AOPL II, 281 F.3d at 241-45, for a discussion of these two techniques. Second, it turned away from its decision in Orders Nos. 561/561-A to drop the top and bottom 25% of all observations as "outliers" and instead used all observations. See id. at 245-46. Third, it stopped using changes in "net plant" to proxy otherwise unobserved changes in capital costs due to returns on investment and income taxes. See id. at 246-47. The shippers assert that had FERC maintained these three innovations following remand, it would have chosen a price-cap index of PPI-1.

The shippers present this argument only in aggregate terms, positing that all three changes made by FERC, taken together, would yield a rate-of-change estimate closer to PPI-1 than to PPI. They fail to bolster this claim with any analysis describing the incremental impact of each change. Such information is at a minimum useful, and its absence can prove fatal; without it, the shippers have provided no reason to think that their justification for the use of PPI-1 can survive on only some of its legs. (We note that no one in the current litigation seems to advocate use of an index differing from PPI by fractions of a whole number.) This implies that a defeat on one prong can defeat the whole claim. In this particular case, however, the shippers fail to show FERC's arbitrariness on any of the three issues.

First, the Commission's decision to revert to a fixed-weight methodology was reasonable. To show the superiority of the floating-weight methodology, the shippers argue that changes in industry averages capture the effect of shifts from crude oil to petroleum products pipelines. But the Commission was trying to predict the likely rate of change in the costs to be experienced by any given pipeline. For that task, the change in industry averages is irrelevant and leads directly to the risk of material distortions that we identified in AOPL II, 281 F.3d at 242-43, a subject on which the shippers are silent. At oral argument shippers' counsel advanced an additional theory — namely, that a fixed-weight index is inferior to its floating-weight analog because the former fails to account for substitution from high-priced to low-price goods over time. We're not sure that this adds much conceptually to the crude-products argument, but in any event we do not reach it. We noted the argument in AOPL II, id. at 244-45, but didn't evaluate it then as FERC had not relied on it. This time around, although the shippers raised the point before FERC on remand, they failed to do so in their briefs here, so it is not properly before us. Compare Carducci v. Regan, 714 F.2d 171, 177 (D.C.Cir.1983) (citing Fed. R.App. Pro. 28(a)(4)).

Second, the shippers fail to show arbitrariness in the Commission's rejection of their preferred method for dropping outlying observations. Their method involves two differences from that of the Commission, which dropped the 25% highest and lowest observations of cost changes. The shippers (1) would use 95% of the observations rather than 50%, and (2) would drop instances of outlying cost levels, rather than, as did the Commission, outlying changes in costs. See Reply Declaration of Alfred E. Kahn, Revisions to Oil Pipeline Regulations Pursuant to the Energy Policy Act, Docket No. RM00-11-000 (October 2, 2000) ("Kahn Reply") at 15. Although the relevant question for the price-cap index is how to compensate firms for changes in their costs, not for the costs per se, the shippers focus on the latter. To the extent that the shippers' submissions can be said to address the point at all (and only by giving their statements a generous reading can one say that they did so), they evidently seek to justify the difference with an analysis that rests in essence on the same ground as their claims in 2000 for use of a floating-weight system, i.e., inter-pipeline competition. See Further Statement of F. M. Scherer, Revisions to Oil Pipeline Regulations Pursuant to the Energy Policy Act, Docket No. RM00-11-000 (September 27, 2000) ("Scherer Reply") at 13-14; see also AOPL II, 281 F.3d at 243 (discussing pipeline-competition argument for floating-weight method). Having reasonably chosen to return to the fixed-weight methodology, the Commission might well have contradicted itself if it had adopted the shippers' proposal for handling outliers. Curiously, if one removes outlying changes in costs, keeping 95% rather than only 50% of the observations no longer helps the shippers but in fact appears to support an index as high as PPI + 1. See Kahn Reply at 15-16 and Table 1.

The third criticism leveled by the shippers is that FERC improperly used net plant as a surrogate for unobserved capital costs. As FERC notes, capital costs consist of depreciation, amortization, return on investment, and income taxes, and the forms pipelines are required to file with FERC contain data only on the first two components; in 1993-94, and again on remand, FERC used changes in net plant as a proxy for the latter two. Before us the shippers challenge this on the ground that the data submitted to FERC have become increasingly accurate, thus eliminating the need for a proxy. While it is evidently uncontested that reported data have become more accurate, the matters reported on have remained unchanged, as the shippers themselves acknowledge. See Petitioners' Opening Br. at 42. As long as the filings fail to account for returns on investment and income taxes, some sort of proxy for these is suitable regardless of how accurate the other data may be.

The shippers go on to attack the use of net plant as being contrary to the Generally Accepted Accounting Practices. But as they do so only in their Reply Brief, the issue isn't properly before us. See Carducci, 714 F.2d at 177. (Moreover, even if the net plant issue had been properly raised, it likely could not help the shippers: FERC concluded that use of the data had no impact on the final decision, see Remand Order, 102 FERC at 61,541, a point the shippers have not challenged.)

Since the shippers fail to show that FERC's decisions were unreasonable on any of the three issues comprising this triad, their argument clearly fails.

The shippers advance a second cluster of three arguments, citing evidence in the record supporting the view that if they prevailed on all three, the Commission's choice of PPI would be exposed as arbitrary. In this cluster they assume the acceptability of the fixed-weight methodology but (1) repeat their "outliers" position, advocating exclusion only of firms with extreme cost levels, (2) argue that volumes for 1999 (rather than for 1994) should be used for weighting the observations, and (3) say that FERC should have included five pipelines that reported data in 1994 and 1999 but not in some of the intervening years. Again the shippers fail to point to any calculations in the record suggesting that error on fewer than all three would undermine the Commission's conclusion.

We need not repeat our analysis of the argument recycled here from the first triad, namely the claim that FERC should have excluded outliers based on levels of costs rather than changes in costs. Coupling it with two new criticisms doesn't make it any more compelling.

The attack on the Commission's choice of 1994 volumes for weighting purposes, rather than those of 1999, addresses an issue that is inherent in cost and price indices. At its heart is the difference between the Laspeyres and Paasche methods, with the former weighting elements on the basis of their share of the total in the initial period, and the latter assigning weights based on end-period shares. Neither index is a perfect measure of changes, however. See, e.g., P. R. G. Layard and A. A. Walters, Microeconomic Theory 156-57 (1978). The shippers make the well-established point that an initial-period index overstates the true change in cost. See Scherer Reply at 12. Indeed. In standard welfare or utility analysis, the Laspeyres index tends to overstate the change and acts as an upper bound on the actual rate of change. By keeping the original weights, it tends to conceal the extent to which customers have preserved their satisfaction by substituting away from goods with rising relative prices towards those with declining relative prices. Layard & Walters at 157. The Paasche conversely serves as a lower bound; neither index is a priori superior to the other. Id. As the shippers' perfectly orthodox critique of an initial-period index is matched by the equally orthodox critique of an end-period index, it is not enough to render the Commission's choice unreasonable. (We assume in the shippers' favor that although the usual object of concern is a measurement of effects on utility, the concerns for over and under-statement are substantially the same here despite the different focus in the Commission's exercise.) No one appears to have advocated year-by-year weighting, which tends to split the difference.

The shippers' final argument is that FERC should have included several pipelines that reported data in 1994 and 1999 but not in some of the intervening years and two other pipelines apparently excluded from the study. But the shippers offer no data on the implications of this change independent of their treatment of outliers and their choice of base years for weighting. See Reply Comments of Sinclair Oil Corp. et al., Revisions to Oil Pipeline Regulations Pursuant to the Energy Policy Act, Docket No. RM00-11-000 (October 2, 2000) at 9-11. Given that they do not discuss the proposed change's incremental impact under FERC's assumptions of a 50% sample and a 1994 base year, they haven't made a sufficient claim that the inclusion of these firms, however sensible, could alone render PPI-1 a better estimator than PPI.

Since the shippers have failed to establish that any of FERC's methodological choices (or combination of choices) was both erroneous and harmful, the petition for review is

Denied.