Southwestern Bell Telephone Co. v. Federal Communications Commission

Opinion for the Court filed by Circuit Judge STEPHEN F. WILLIAMS. STEPHEN F. WILLIAMS, Circuit Judge: Certain local exchange carriers (“LECs”) challenge an order of the Federal Communications Commission on the ground that the Commission arbitrarily and capriciously disregarded its own rule when it denied “exogenous cost” treatment for cost increases that the LECs experienced as a result of mandated changes in their accounting for certain post-retirement worker benefits. Treatment of Local Exchange Carrier Tariffs Implementing Statement of Financial Accounting Standards, “Employers Accounting for Post-retirement Benefits Other than Pensions ”, 8 FCC Red 1024 (1993) (“OPEB Order”). We reverse and remand. The concept of “exogenous costs” is an outgrowth (at least for regulatory purposes) of the Commission’s decision to shift from conventional rate-of-return methods to the use of price caps for some of the firms subject to its rate regulation. Among the hopes for price caps is that they will improve incentives for innovation on the part of the regulated firms. Under rate-of-return regulation the Commission projects future costs on the basis of immediate past history, and sets rates calculated to recover such costs, with the result that a firm’s only benefit from a cost-saving innovation is the advantage gained in the period before new rates become applicable — i.e., the firm enjoys additional profits only during the “regulatory lag”. National Rural Telecom Ass’n v. FCC, 988 F.2d 174, 178 (D.C.Cir.1993). With price caps, the initial base rates (here, the rates prevailing on July 1, 1990) are for the most part adjusted solely for reasons independent of the regulated firm’s actual behavior, notably (1) an annual adjustment for general price inflation, measured by the Gross National Product Price Index (“GNP-PI”) see Policy and Rules Concerning Rates for Dominant Carriers, 5 FCC Red 6786, 6792-93 ¶¶ 47-64 (1990)(“LEC Price Cap Order”), modified on recon., 6 FCC Red 2637 (“LEC Price Cap Reconsideration”), further recon. dism’d, 6 FCC Red 7482 (1991), and (2) an automatic annual downward adjustment for expected improvements in firm productivity, see LEC Price Cap Order, 5 FCC at 6793-6801 ¶¶ 65-119. The Commission also provided, however, for adjusting the price caps on the basis of “exogenous costs”, which it described as “in general those costs that are triggered by administrative, legislative or judicial action beyond the control of the carriers.” Id. at 6807 ¶ 166. Because of the carriers’ lack of control, adjustments for such changes presumably do not undermine the price caps’ incentive structure. The Commission considered in advance quite a number of possible candidates for exogenous cost treatment, including, for example, changes in amounts paid or received under certain pooling arrangements, see LEC Price Cap Order, 5 FCC Red at 6807 ¶¶ 169-70, and, most pertinently here, changes in accounting rules. For accounting changes, it specified automatic exogenous cost treatment for changes made by the Commission itself in its Uniform System of Accounts (“USOA”), explaining that “such changes are imposed by this Commission and — 168 — are outside the control of carriers.” Id. ¶ 168; see also 47 CFR § 61.45(d)(1). It said, however, that changes in generally accepted accounting principles (“GAAP”) ordered by the Financial Accounting Standards Board (“FASB”), were not to be the basis of automatic price cap adjustments, explaining: As explained in the Second Further Notice, certain GAAP changes may require amendment to the USOA while others may not. Carriers must notify us of their intention to apply a change in GAAP and we will allow such change if we find it to be compatible with our regulatory accounting needs. No carrier may adjust its price caps to reflect a change in GAAP until we have approved the carrier’s proposed change. Furthermore, we wish to clarify that no GAAP change can be given exogenous treatment until the Financial Accounting Standards Board has actually approved the change and it has become effective. The cap mechanism is intended to reflect changes in costs that have occurred, not anticipated cost changes. LEC Price Cap Order, 5 FCC Red at 6807 ¶ 168 (footnotes omitted).1 The treatment of GAAP changes thus appeared to differ from that of USOA changes only in that GAAP changes would originate in the FASB and would become mandatory in the pertinent sense only after the Commission found them, under its standard procedure, consistent with the agency’s regulatory objectives. See 47 CFR § 32.16. For both types of accounting changes, the Commission’s mandate brings about the change and demonstrates that the carriers lacked control. See also LEC Price Cap Reconsideration, 6 FCC Red at 2664-65 ¶ 63 (referring, for a GAAP change, to issue of “whether the change is outside the control of the carrier”). In a parallel proceeding relating to AT & T the Commission set forth a second criterion that a GAAP change would have to satisfy — a demonstration that the change “will not be double counted in the Price Cap Index, once in the GNP-PI and once as an exogenous cost.” Memorandum, Opinion and Order on Reconsideration, 6 FCC Red 665, 674 ¶ 75 (1991). And on reconsideration of the price order for LECs the Commission made clear that this second criterion also applied to them. See LEC Price Cap Reconsideration, 6 FCC Red at 2665 ¶ 63. Thus it appeared that changes in GAAP were to receive exogenous cost treatment if they were mandated by the Commission (the “control” test) and were shown not to involve double counting with the GNP-PI adjustment. In December 1990 the FASB adopted Statement of Financial Accounting Standards — 106 (“SFAS-106”), altering the way in which companies adhering to GAAP account for “other postemployment benefits” for fiscal years beginning after December 15, 1992. The “other”, which explains the “0” in the OPEB acronym, is intended to exclude pension benefits; what is left generally consists of retirees’ life insurance and medical and dental care benefits. Before SFAS-106, firms accounted for these benefits on a “pay as you go” or cash basis, recognizing them when the costs were paid rather than when the firm received the services for which the benefits were compensation. SFAS-106 adopts an accrual method, requiring recognition of OPEB costs as they are earned by current employees. See OPEB Order, 8 FCC Red at 1025 ¶3. Besides requiring accrual treatment for ongoing OPEBs, SFAS-106 required businesses to recognize a transition benefits obligation, i.e., a reflection of the accumulated obligation accrued for work done in the past. Firms were to recognize this expense either — 169 — all at once or to spread it out — either by using the average remaining service period of active plan participants or, if the average remaining service period were less than 20 years, by using a 20-year period. Id. at ¶ 4. On application by Southwestern Bell, the Commission found SFAS-106 consistent with Commission objectives and authorized the LECs to adopt it on or before January 1, 1993. Southwestern Bell, 6 FCC Rcd 7560 (1991). Noting that the effect of recognizing the transition obligation immediately “would be so large as to seriously distort the carriers’ operating results,” the Commission directed the companies to amortize that obligation. Id. at ¶4. In 1992 Bell Atlantic, US West, and Pacific Bell each filed tariff revisions which hiked their Price Cap Index (“PCI”) levels and their rates to reflect the change wrought by SFAS-106 (i.e., the increase in their current-year OPEB costs over what they would otherwise have charged), asserting that these increments were exogenous costs. The Commission’s Common Carrier Bureau suspended the tariffs and initiated an investigation into whether the LECs had demonstrated “that implementing SFAS-106 results in an exogenous cost change under the Commission’s price cap rules”. Treatment of Local Exchange Carrier Tariffs Implementing Statement of Financial Accounting Standards, “Employers’ Accounting for Postre-tirement Benefits Other Than Pensions ”, 7 FCC Rcd 2724, 2725 ¶ 10 (1992). The Bureau noted that the order in Southwestern Bell authorized carriers to adopt SFAS-106 “as a mandatory practice for purposes of the USOA [Uniform System of Accounts].” Id. at 2724 ¶ 3. Given the complex issues, the FCC made all LECs subject to price caps parties and requested them to submit information as to whether exogenous cost treatment should be accorded. The Commission denied the claims for exogenous cost treatment, but used different rationales for the ongoing and the transitional elements. As to ongoing costs, it recognized that the accounting change was “not within the carriers’ control”, OPEB Order, 8 FCC Red at 1033 ¶ 53, but denied exogenous cost treatment on the ground that they had considerable “control over the present and future benefit plans they set with their employees and the costs of these plans”. Id. As for the transitional obligation, the FCC skipped the control criterion and held that the LECs had failed to demonstrate that the effects of SFAS-106 were not already reflected in the GNP-PI adjustment. See id. at 1034-35 ¶¶ 61-66. It then went on to find that SFAS-106 failed to satisfy a number of hitherto unmentioned tests. The LECs filed the instant petition with us. ****** Both sides agree that the FCC’s statement of its criteria for exogenous cost treatment constituted a rule, not a policy statement. See, e.g., FCC Brief at 30-31, 32 n. 31 (characterizing the issue as one of interpretation of the Commission’s rules). Accordingly the Commission was bound to follow those statements until such time as it altered them through another rulemaking. See National Family Planning v. Sullivan, 979 F.2d 227, 231-32 (D.C.Cir.1992). Compare McLouth Steel Products Corp. v. Thomas, 838 F.2d 1317, 1320 (D.C.Cir.1988) (policy statements distinguishable as not establishing binding norms).2 Thus the key question posed by petitioners is whether the FCC adhered to those criteria in evaluating the LECs’ filings on SFAS-106. We conclude that it did not. Ongoing expenses. The Commission frankly recognized that the accounting change was “not within the carriers’ control”. OPEB Order, 8 FCC Rcd at 1033 ¶ 53. Yet it denied exogenous cost treatment, saying that because the carriers “exercise substantial control over the level and timing of OPEB expenses”, such treatment would “give the LECs undue power to influence their PCI levels, and would undermine the incentive structure of price caps.” OPEB Order, 8 FCC Rcd at 1033 ¶53. There simply is not a hint of such a control test in the Commission’s discussion of accounting changes in either the LEC Price — 170 — Cap Order or the LEC Price Cap Reconsideration. The key passage of the LEC Price Cap Order indicated that GAAP would be on a par with changes in the Commission’s Uniform System of Accounts once the Commission reviewed the FASB change to see whether it was compatible with the Commission’s regulatory accounting needs: Carriers must notify us of their intention to apply a change in GAAP and we will allow such change if we find it to be compatible with our regulatory accounting needs. 5 FCC Red at 6807 ¶ 168. That the Commission meant for the “control” test to be satisfied simply by the fact of exogenous imposition of the accounting rule, without concern for the underlying costs covered by the rule, is perhaps even more clearly shown in the Commission’s earlier formulation of the point (in the AT & T context): We also agree that there is no difference in principle between a cost change caused by a USOA change and a cost change caused by a GAAP change. We do not, however, authorize carriers automatically to adjust price caps to reflect changes in GAAP. Our current procedures for implementing GAAP in the context of the USOA require carriers to notify us of their intention to apply a change in GAAP. They may make the change only if we find it to be compatible with regulatory accounting needs. Some changes in GAAP which are compatible with regulatory needs can be carried out within our existing rules, while others may require amendment of the USOA. Policy and Rules Concerning Rates for Dominant Carriers, Report and Order and Second Further Notice of Proposed Rule-making, 4 FCC Rcd 2873, 3017-18 ¶295 (1989) (emphasis added). The fact that a USOA change is adopted by the Commission obviously tells us nothing about how much or little the carrier may control the cost that is to be accounted for differently. Thus, the Commission’s view that the two types of accounting change were “no different in principle” confirms the natural meaning of the rest of the language: an FASB change adopted by the Commission is not a change under control of the carrier, and, once mandated by the Commission, the change satisfies the control criterion. Recall also that in the LEC Price Cap Reconsideration the Commission referred to the issue of “whether the change is outside the control of the carrier”. 6 FCC Red at 2664-65 ¶ 63 (emphasis added). Like many accounting changes, SFAS-106 simply altered the time as of which a cost would be recognized, and that shift was indisputably outside the carriers’ control. To be sure, SFAS-106 required much more estimation of expenses than was necessary for accounting under a cash basis. In discussing the transition obligations, the Commission referred to evidence that “a one percentage point reduction in the health care trend from the value assumed by Pac Bell would reduce the accrual amount by 15.3 percent”. 8 FCC Red at 1035 ¶ 65. Whatever that may imply in terms of how to calculate the amount accrued, it obviously does not mean that the requirement to accrue was under the carriers’ control. The Commission noted in the OPEB Order that its rule — the LEC Price Cap Order and the LEC Price Cap Reconsideration — had denied exogenous cost treatment for changes in depreciation rates, on the ground that even though such changes were set by regulatory agencies, “carriers still exercise control over their depreciation costs with their decisions to deploy or retire equipment.” LEC Price Cap Reconsideration, 6 FCC Rcd at 2672 ¶ 74 (cited at OPEB Order, 8 FCC Rcd at 1033 ¶ 53). Thus, it reasoned, similar inquiry into control over the underlying costs was appropriate here. But whatever the Commission’s treatment of depreciation rate changes, it held that view at the same time as it set forth its rule on the treatment of GAAP changes, yet, in the GAAP context, it said nothing about control over underlying costs. And if the disparate treatment of depreciation rates were extended to GAAP changes, it would belie the Commission’s statement that they and USOA changes were “no different in principle”. Accordingly, it provides no basis for injecting the issue of control over underlying costs into the classification of GAAP changes. — 171 — Transitional obligations. The FCC did not decide whether the SFAS-106 change satisfied the control prong as to the transitional obligation because it concluded that even if it passed that test, the carriers had failed to show the necessary absence of double counting under GNP-PI. OPEB Order, 8 FCC Red at 1033-35 ¶¶ 57-66. We should perhaps start by explaining how SFAS-106 cost increases might be duplicated in the LECs’ GNP-PI adjustment. If (1) the SFAS-106 cost increase represented the same fraction of total costs for all employers as for LECs (which would depend on such matters as (a) whether the average firm offered OPEBs of the same cost and character as LECs, (b) whether the demographic profile of workers as a whole were the same as that of LEC workers, and (e) whether labor costs were the same fraction of total costs for the average firm as for the average LEC), and (2) all SFAS-106-induced cost increases were passed forward to consumers in price increases, then a 1% SFAS-106 increase in LECs’ OPEB costs might be matched by a 1% increase in prices generally.3 Thus, exogenous cost treatment for the LECs’ SFAS-106 costs would result in complete double counting. None of these assumptions appears to be valid. The most obvious difficulty is that a far lower fraction of private sector employees is eligible for OPEBs compared to telephone company employees. One of the studies submitted by the LECs quoted General Accounting Office figures to the effect that only 30.7 million out of 95.8 million private sector employees were eligible for OPEBs, or less than 30%, see Godwins Study, Joint Appendix (“J.A.”) at 83, as opposed to 100% for telephone companies, see id. at 110 (not GAO figures). That huge discrepancy is, to be sure, somewhat offset by the lesser role of directly employed labor in telephone companies, where (under some estimates anyWay) labor accounts for 38.5% of value added as against 64.3% in the economy as a whole. Id. To support their claim that any double counting would be very limited, some LECs offered the Godwins study already referred to and others offered one by the National Economic Research Associates (“NERA”). The Godwins study assumed that before SFAS-106 firms offering OPEBs were not taking OPEB costs into account in selecting output levels or prices. (In perfect competition, of course, the firm is a price taker and price emerges from the interaction of demand with all firms’ output decisions.) The study concluded that about 85% of the cost increase would not be reflected in an increase in GNP-PI. Id. at 68. The NERA study took the opposite tack, arguing that non-regulated firms would already have been taking accrued OPEB costs into account, so that SFAS-106 would produce no direct change in their conduct. It reasoned that in hiring an extra worker such a firm would have to take into account all resulting costs—money wages plus the present value of all future expenses including OPEBs. See NERA Study, J.A. at 32-34. Thus the only impact of SFAS-106 on GNP-PI would be through its impact on the prices of regulated firms, which (certainly under rate-of-return regulation) are based on booked costs. Pursuing this reasoning, the NERA study concluded that if SFAS-106 caused a 1.1% increase on the booked expenses of an average firm, it would increase prices generally by only 0.12%. J.A. at 24, 54. If the SFAS-106-induced increase for NERA’s client, Pacific Bell, were 1.92%, then only .12/1.92, or 6.26% of Pacific Bell’s SFAS-106 cost increase needed to be deducted from its SFAS-106 cost increase to avoid double counting. Id. at 54-55. The Commission attacked the God-wins and NERA studies on a variety of grounds. First it observed that neither study proved that its initial assumptions were correct, OPEB Order, 8 FCC Red at 1034 ¶ 63, noting caustically that the sets of assumptions were in “sharp contrast”, id. at ¶ 62. The claim of complete want of support is in fact false, for the NERA study pointed — 172 — to econometric evidence that accounting changes generally have no effect on stock prices, see J.A. at 49-50, which tends to support the proposition that the market sees through such conventions. But quite apart from that, any analysis of whether an exogenous change will be reflected in GNP-PI will involve some unproven — and likely unprovable — assumptions. Indeed, the Commission’s own brief characterized the assumptions as “impossible to verify”. FCC Brief at 22. If an agency can reject an econometric study merely by observing that it employed unproven assumptions (and that the outside party bore the burden of proof), then no party with the burden can ever prevail. “[Ajssigning the burden of proof is not a magic wand that frees an agency from the responsibility of reasoned decision-making.” Kansas Gas & Elec. Co. v. FERC, 758 F.2d 713, 721 (D.C.Cir.1985). To reject such a study, the Commission must at least express a reason for doubting some critical assumption. Moreover, to the extent that the FCC concluded that because the studies began with different assumptions, neither could be relied upon, its decision was quite illogical. Given the difficulty of verifying the assumptions that must underlie any such analysis, it was natural for the LECs to cover a range of possibilities. The substantial identity of results in the face of widely varying assumptions tended simply to show that the outcome was insensitive to this variation. That rendered the conclusions more robust, not less. Equally troubling is the Commission’s pointing to the number of “parameters” for which the Godwins study suggested ranges of possible values, such that under the most extreme (i.e., anti-LEC) assumptions the lowest portion of the SFAS-106 increase not reflected in GNP-PI would be 60.1%. OPEB Order, 8 FCC Red at 1035 ¶ 64. Only 60.1%! If only 60.1% was clearly free of overlap, the proper response would seem to be to limit exogenous cost treatment to that percentage. This is especially so as the Common Carrier Bureau, when designating the issues for investigation, had separated the question of the propriety of exogenous treatment from the issue of the size of cost to receive such treatment. See In the Matter of Treatment of Local Exchange Carrier Tariffs Implementing Statement of Financial Accounting Standards, “Employers Accounting for Post-retirement Benefits Other Than Pensions”, 7 FCC Rcd 2724, 2725 ¶ 10 (1992). In the same vein, the Commission also based its rejection of exogenous cost treatment on a concern that SFAS-106 required the carriers to make numerous assumptions about the costs of future benefits. These estimates were deemed “highly speculative”, as even small changes in certain assumptions could lead to drastic swings in the projected costs. OPEB Order, 8 FCC Red at 1035 ¶ 65. Given the division of the proceeding, it would seem that this problem should lead to complete rejection only if there was no way of obtaining even conservative estimates, which the Commission does not claim. Apart from imposing impossible burdens as to GNP-PI double counting, the Commission invoked several altogether new criteria in rejecting the LECs’ claim for exogenous cost treatment of the transition obligation. None of these was in the faintest way foreshadowed in the rules the Commission had adopted to handle such issues. First, it introduced a criterion called inter-temporal double-counting. As we understand it, this referred to the possibility that LECs would effectively collect twice, once on a cash basis, once on an accrual basis. See OPEB Order, 8 FCC Rcd at 1035 ¶ 67-68. The Commission acknowledged that the LECs had asked only for the increment resulting from SFAS-106, id. at ¶ 68, but proceeded to express concern that an overestimation of accrued OPEBs might lead to distortions. It is not at all clear why that cannot be resolved by reasonable conservatism in the accrual estimates. Finally, seeming to acknowledge that the issue was one of timing only, the Commission argued that exogenous cost treatment would require annual calculation of both the accrued amount and the cash amount, with only the difference (plus or minus) counting as an exogenous cost. Id. at 1035-36 ¶ 69. This realization may justify a change in the Commission’s rule, but we fail to see how it could justify — 173 — refusal to apply its rule while that still governs. Second, the FCC suggested that if LEC investors knew that LECs would not be able to raise their rates upon implementation of SFAS-106, they might have demanded a higher rate of return; thus, unexpectedly permitting such raises would allow them to recover twice — once in the rate of return, once for the exogenous cost hike. Id. at 1036 ¶¶ 70-71; cf. Duquesne Light Co. v. Barasch, 488 U.S. 299, 311-12, n. 7, 109 S.Ct. 609, 618 n. 7, 102 L.Ed.2d 646 (1989) (expressing assumption that the allowed rate of return will reflect degree of risk implicit in regulators’ approach to application of used-and-useful rule). The reasoning here appears to stretch an insight to the outermost reaches, to the point where it may justify any arbitrary and capricious resolution of any issue: so long as investors can anticipate the caprice, no matter. That is not, however, our current legal system. Finally, the Commission suggested the SFAS-106 cost might in some way have been already counted in calculation of the productivity offset. Id. at 1036 ¶ 72.4 We note that each of these three issues, if adopted as a basis for rejecting exogenous cost treatment for GAAP changes, would drive a still greater and more puzzling wedge between them and USOA changes. The fundamental difference between the two, as we said, is that GAAP changes are initiated by the FASB; each becomes mandatory only when mandated by the Commission. In any event, whatever the intrinsic merits of these three possible bases for rejecting exogenous cost treatment, the Commission is free to consider them as a basis for amending its current rule, not for concocting a new rule in the guise of applying the old. Accordingly, we remand to the FCC to consider the LECs’ request for exogenous cost treatment of their SFAS-106 incremental costs in a manner consistent with this opinion and with the LEC Price Cap Order and the LEC Price Cap Reconsideration. So ordered. Footnotes [1] Readers may wonder why a change in accounting rules can be regarded as having changed real costs at all. No party discusses the point, and it is common ground that this feature is no obstacle to exogenous cost treatment for the changes at issue here. Certainly accounting changes may have material economic impact. A change in recorded earnings will change the company’s price-earnings ratio, and thus possibly the market price of the stock, effectively altering the company's real cost of capital. But see National Economic Research Associates Study, Joint Appendix at 49-50 (citing econometric evidence that accounting changes generally have no effect on stock prices). In extreme cases, the accounting change may impose a binding limit on earnings available for payment of dividends or even push a marginal company over the edge into bankruptcy. [2] Of course the Commission would have to explain a deviation even from a policy statement. See, e.g., Greater Boston Television Corp. v. FCC, 444 F.2d 841, 852 (D.C.Cir.1970). [3] This account leaves out serious complications. For example, if the price level is a function of the quantity and velocity of money and the supply of real goods and services (P = MV/Q), see, e.g., Paul A. Samuelson & William D. Nordhaus, Economics 323 (11th ed.1985), it is not clear just how implementation of SFAS-106 might have changed any of these. [4] Some LECs had adopted methods of prefund-ing OPEBs in the 1980s, thus presumably retarding their productivity improvement rates. In estimating likely productivity improvements, the Commission had not adjusted for this; any adjustment would presumably have yielded a higher estimate of annual productivity improvement. The Commission appears agnostic on whether this fact means that exogenous cost treatment of SFAS-106 increases would result in double counting. See id.