607 F. 2d 392 - National Association of Greeting Card Publishers v. United States Postal Service

Timothy J. May, Washington, D. C., for intervenor Reader's Digest Association, Inc. in Nos. 78-1448, 78-1449, 78-1453, 78-1483, 78-1484, 78-1509, 78-1683 and 78-1684.

Robert L. Kendall, Jr., Philadelphia, Pa., with whom Bernard G. Segal and John E. McKeever, Philadelphia, Pa., were on the brief, for intervenor, United Parcel Service of America, Inc. in Nos. 78-1448, 78-1449, 78-1453, 78-1483, 78-1484, 78-1494, 78-1509, 78-1517, 78-1518, 78-1531, 78-1532, 78-1683 and 78-1684.

Arthur B. Hanson, Washington, D. C., was on the brief, for intervenor, American Newspaper Publishers Association in No. 78-1448.

W. Terry Maquire, Washington, D. C., was on the brief, for intervenor, National Newspaper Association in No. 78-1448.

Paula A. Jameson, Princeton, N. J., was on the brief, for intervenor, Dow Jones and Co., Inc. in Nos. 78-1448, 78-1449, 78-1453, 78-1483, 78-1484, 78-1494, 78-1509, 78-1517, 78-1518, 78-1531, 78-1532, 78-1683 and 78-1684.

Richard M. Schmidt, Jr., Ian D. Volner and Ruth S. Baher, Washington, D. C., were on the brief, for intervenors, The Association of American Publishers, Inc. and Recording Industry Association of America, Inc. in Nos. 78-1448, 78-1449, 78-1453, 78-1483, 78-1484, 78-1683 and 78-1684.

Eugene E. Threadgill, Washington, D. C., entered an appearance for intervenor, The Council of Public Utility Mailers in Nos. 78-1448, 78-1449, 78-1683 and 78-1684.

J. Edward Day, Washington, D. C., entered an appearance for intervenor, Associated Third Class Mail Users in Nos. 78-1448, 78-1449, 78-1453, 78-1683 and 78-1684.

Kenneth Wells Parkinson and William J. Olson, Washington, D. C., entered appearances for intervenor, National Foundation March of Dimes in Nos. 78-1448, 78-1449, 78-1453, 78-1483, 78-1484, 78-1494, 78-1509, 78-1517, 78-1518, 78-1683 and 78-1684.

William H. Smith and Michael F. Crotty, Washington, D. C., entered appearances for intervenor, American Bankers Association in Nos. 78-1448, 78-1449, 78-1453, 78-1683 and 78-1684.

Dana T. Ackerly and David A. Levitt, Washington, D. C., entered appearances for intervenor, Direct Mail/Marketing Association, Inc. in Nos. 78-1448, 78-1449, 78-1453, 78-1483, 78-1484, 78-1494, 78-1509, 78-1517, 78-1518, 78-1531, 78-1532, 78-1683 and 78-1684.

David C. Todd, Washington, D. C., entered an appearance for intervenors, Parcel Shippers Association and Mail Order Association of America in Nos. 78-1449, 78-1453, 78-1483, 78-1484, 78-1494 and 78-1509.

Francis R. Cawley entered an appearance for intervenor, Agricultural Publishers Association, Inc. in Nos. 78-1517 and 78-1518.

Before TAMM and LEVENTHAL, Circuit Judges, and WILLIAM B. BRYANT,* Chief Judge, United States District Court for the District of Columbia.

Opinion PER CURIAM.

Opinions for the Court filed by LEVENTHAL and TAMM, Circuit Judges.

In these cases the court reviews and affirms orders of the Governors of the United States Postal Service approving a recommended decision of the Postal Rate Commission on increases in postal rates and fees. There are two opinions for the court. Judge Leventhal's opinion recounts the background of the litigation and considers the various objections of the parties to the cost allocation approaches of the Postal Service. Judge Tamm's opinion deals with claims that the Governors and Commission erred 1) in rejecting a proposed first-class "citizens' rate; " 2) in including in revenue requirement an item to recover past losses incurred by the Service; and 3) in imposing "constraints" on parcel post rates found to be otherwise cost-justified.

In this case the court again has occasion to consider the response of the United States Postal Service ("USPS" or "Postal Service") to the "special, and quite demanding, ratemaking requirements"1 of the Postal Reorganization Act of 1970 ("Act")2 and to this court's views concerning those requirements.

This case arises from the fourth general ratemaking proceeding under the Act. Our principal focus here, as in our cases reviewing earlier ratemaking proceedings,3 is on the methods by which the Postal Service, in setting the rates for the various classes of mail, allocates its costs among those classes. In particular, we must assess the Postal Service's "service related cost" methodology, by which it undertook in response to this court's mandate in National Association of Greeting Card Publishers v. USPS (NAGCP I )4 to assign to certain classes of mail fixed delivery costs deemed to result from the maintenance of a six-day-a-week, as opposed to a three-day-a-week, mail delivery schedule.

Secondarily, we also consider certain additional claims: By first-class users contending that the approved rate structure unduly discriminates against first-class mail; and by second-class users challenging certain specific cost attributions and rate decisions.

As to the cost allocation issues that are the subject of this opinion, we affirm. The ratemaking process under the Act has reflected a constructive interaction between the Postal Service and this court that has, to a large extent, developed and sharpened the governing concepts. While certain gaps remain that will be highlighted, the efforts of the Postal Service in this ratemaking proceeding represent a reasonable response to these governing concepts.

The decision that we review was the culmination of a proceeding, known as Docket No. R77-1, carried out in accordance with the Act's complex ratemaking procedures.5 The Postal Service initiated the process on July 13, 1977, by filing with the Postal Rate Commission (PRC) a request for a recommended decision on changes in postage rates and fees for postal services. As required by applicable regulations,6 the request contained detailed information on the Postal Service's current financial position, its anticipated revenue needs, and its proposed cost allocation and rate schedule. The request was accompanied by the direct testimony and workpapers of 11 supporting witnesses, and numerous exhibits.

The request, as subsequently amended, stated that without increases in rates and fees, USPS would incur a revenue deficiency in the "test year"7 of $2,410 million.8 USPS proposed increases in most rates and fees to eliminate the projected deficiency. Perhaps the most notable rate proposal was one to bifurcate the rate for first-class mail then 13 cents for the first ounce into a 16 cent regular first-class rate and a special 13 cent citizens' rate. USPS projected that its proposed rate schedule would substantially fulfill a revenue requirement of $17,642 million.9

Because of a statutory deadline mandating transmittal of a recommended decision within 10 months,10 the proceedings were conducted before the Commission en banc, without initial reference to an administrative law judge. A total of 62 intervenors and limited participants, together with an Officer of the Commission (OOC) appointed to represent the interests of the general public, took part. The Commission heard 75 witnesses in 62 days of hearings. The record was closed on February 22, 1978, briefs were submitted, and oral argument was heard for two days on March 28 and 29, 1978.11

The PRC issued its opinion and recommended decision on May 12, 1978. The decision modified slightly USPS's projected revenue requirement from $17,642 million to $17,585 million.12 The Commission also modified USPS's proposed rates in several respects. Most significant was its rejection of the bifurcated first-class rate in favor of a single 15 cent rate.

In two separate decisions, one addressed to the citizens' rate and one to the other determinations of the PRC, the Board of Governors of the Postal Service approved the recommended decision on May 19, 1978. The rate increases took effect on May 29, 1978. Numerous parties then filed the petitions for review that are now before us.

Full understanding of the specific issues raised in this case requires some discussion of the broader regulatory context: the manner in which this court has reviewed the Postal Service's efforts to develop and to apply workable principles of ratemaking under the Act. This court has stressed the congressional purpose to reduce the subjectivity inherent in the political ratemaking process that prevailed before enactment of the Act, and has required the Service to fashion methods that ensure that the rates for the various mail classes and postal services reflect as closely as possible the cost of providing these services.

In considering the Service's rate requests, the Postal Rate Commission faced the formidable task of developing ratemaking principles of nationwide applicability and importance on a virtually clean slate. In 39 U.S.C. § 3622(b) (1976), Congress enumerated the factors the Commission must take into account in formulating its recommended decision. The crucial criterion has emerged out of § 3622(b)(3), which establishes

the requirement that each class of mail or type of mail service bear the direct and indirect postal costs Attributable to that class or type plus that portion of all other costs of the Postal Service reasonably Assignable to such class or type.13

(Emphasis supplied). The central concern of prior ratemakings, as well as this one, has been to give content to the references to "attributable" and "assignable" costs.

In its first two rate proceedings under the Act, the Commission adopted a two-step approach to the allocation of the costs of operation of the Postal Service among the classes of mail. The Commission first "attributed" to the various mail classes and postal services only those costs demonstrably caused by providing the particular service, thereby establishing a rate floor for each class and service. The key to determining causation was a strict requirement of a showing of cost variability that a particular cost varied with a change in volume of the service provided. Excluded from this calculus were all cost allocations that might be derived by approximation or estimation but for which causation could not be conclusively demonstrated under the variability analysis. Thus, "(t)he stricture of this approach and the unavailability of complete data led in the first two rate proceedings to recommended decisions which, by the Commission's own account, did not attribute to each class all the costs that likely were the result of providing the particular service."14

In the second step, the costs remaining after the attribution in the first step were "assigned" to the various classes under a "value of service" or demand theory approach. Specifically, the Commission developed an "inverse elasticity rule," under which final rates were derived "by assigning to the various classes of mail different markups (above attributed costs) in inverse proportion to the relative elasticity of demand for each of the classes."15 In this way, the Commission purported to " 'assign costs in a manner that fully takes into account the noncost factors of the statute.' "16

In our first consideration of the ratemaking requirements of the Act, we affirmed the decision of the Governors approving the PRC's recommended decision in the first postal ratemaking proceeding, Docket No. R71-1. Association of American Publishers, Inc. v. Governors of USPS, 157 U.S.App.D.C. 397, 485 F.2d 768 (1973). Despite the affirmance, we took pains to express our concern over the ratemaking approach adopted by the PRC. Judge Bazelon's concurring opinion, which was joined by the other two members of the panel, noted that the Commission's attribution in that proceeding accounted for only 49 percent of the Postal Service's total costs. The remaining 51 percent was "reasonably assigned" by two USPS employees using what that opinion called a "vague formula" that consisted essentially of value-of-service considerations. Id., 157 U.S.App.D.C. at 406-07, 485 F.2d at 777-78. The concern was that the approach permitted unstructured and essentially unreviewable discretion in the USPS staff to allocate more than half the costs of the Service. The result was emphasis on non-cost considerations, the kind that Congress had intended to excise as much as possible from postal ratemaking. The court chose not to disturb the decision, however, in view of the presumption accorded initial efforts under a novel regulatory scheme,17 and the prospects for improved tracing of costs under PRC and USPS procedures. Id., 157 U.S.App.D.C. at 407-08, 485 F.2d at 778-79.

The Service and the Commission failed to heed the cues of American Publishers. In Docket No. R74-1, the second general postal ratemaking, the Commission again applied the two step approach, "attributing" only 52.5 percent of all costs and "assigning" the remainder under the "inverse elasticity" principle. In doing so, the Commission rejected entirely the initial decision of the Administrative Law Judge. His initial decision had attributed 70.6 percent of the total costs by means of cost accounting principles that permitted allocation of costs to particular classes even where direct cost variability could not be demonstrated. It assigned the remaining costs primarily in accordance with cost-of-service, rather than value-of-service, principles.18

On judicial review, we held that this two step approach of the PRC did not comply with the requirements of the Act. National Association of Greeting Card Publishers v. USPS (NAGCP I ), 186 U.S.App.D.C. 331, 569 F.2d 570 (1976), Vacated as to other issues, 434 U.S. 884, 98 S.Ct. 253, 50 L.Ed.2d 169 (1977). In the language of § 3622(b)(3) and the legislative history of the Act, we discerned an obligation to apply cost-of-service ratemaking principles to the greatest extent possible. Id., 186 U.S.App.D.C. at 346-50, 569 F.2d at 585-89. Thus, of the ratemaking criteria enumerated in § 3622(b), only § 3622(b)(3) was framed as a "requirement," emphasizing the special role played by cost considerations. The requirement that each class or type of service bear the "direct and indirect costs attributable" to it necessitated an "extended attribution" that would take into account indirect costs that were not measurably variable but that could, by employing cost allocation formulae based on accounting principles, "be determined with reasonable confidence to be the consequence of providing the service." Id., 186 U.S.App.D.C. at 347, 569 F.2d at 586. Further, by referring to "all other Costs of the Postal Service reasonably assignable to such class or type," the section required that even the process of assignment was to be carried out in accordance with cost-of-service principles. Finally, we found that the proper locus for allocation on non-cost principles was to be found in the language of § 3622(b)(3), which contemplated that not all postal costs were to be attributed or assigned to a particular class or service. As to the residuum of costs, the other, "discretionary" criteria of § 3622(b) would come into play.

We found this reading of the statutory language supported by the legislative history of the Act. The central purpose of the Act was to "get politics out of the Post Office" to eliminate the discretion to set rates that had resulted in discrimination against certain classes of mail. We noted:

Prior to the Act the Service enjoyed broad discretion in the allocation of postal costs, a discretion which in the past had made the setting of postal rates susceptible to political bartering and the frequently abusive influence of lobbyist efforts. . . . (I)t would be anomalous to construe subsection 3622(b)(3) as permitting a grudging use of cost-of-service principles which, by expanding the residuum of costs subject to discretionary allocation, simply preserves the potential for continuing the very same discriminatory treatment that the Act so clearly intended to remedy.

Id. 186 U.S.App.D.C. at 350, 569 F.2d at 589. Under this standard, we found a deficiency in the Postal Service approach, which employed cost-of-service principles only where direct cost variability could be demonstrated, leaving about one half of total costs subject to discretionary allocation. We summarized the statute's requirements as follows:

(T)he Postal Rate Commission must first of all attribute to each mail class or postal service all postal costs which may reasonably be determined, through variability theory as well as through other reasonable inferences of causation, to be the consequence of providing the service. It must then distribute among the mail classes and services that significant portion of all remaining costs of the Postal Service that may reasonably be assigned to each on the basis of best available cost-of-service estimates. The residuum of costs is subject to discretionary allocation in accord with the noncost factors set forth in the Act.

Id. (footnote omitted).

Our principal function in the previous postal rate cases lay in ascertaining the broad approaches mandated by Congress for the PRC's exercise of its ratemaking authority. Sound principle bids us accompany any further judicial review of the specifics of these approaches and tracing methodologies with diffidence and restraint.19

While cost allocations often possess a patina of scientific precision, they remain ultimately matters more of judgment and appraisal than of mathematical certainty.20 Recognizing the crucial play of judgment in this area, reviewing courts have traditionally deferred to agency expertise and have abstained from imposing unrealistic technical requirements or standards of precision on the administrative process. As we have pertinently summarized the matter, selection of ratemaking theories is more a question of policy than of fact: "A theory of ratemaking must be reasonable, explained, and supported, but is not subject to the same substantiation principle as the substantial evidence test applicable to fact-finding."21 Similarly, a method of cost allocation may not be rejected simply because difficulties can be identified. Justice Brandeis long ago remarked that "experience teaches us that it is much easier to reject formulas presented as being misleading than to find one apparently adequate."22

In NAGCP I, we recognized our obligation to defer to the PRC's expert judgment in the selection of cost allocation methodologies. 186 U.S.App.D.C. at 353-54, 569 F.2d at 592-93. We noted that "to be acceptable, a methodology need only be reasoned, nonarbitrary and permissible under the statute." Id., 186 U.S.App.D.C. at 352, 569 F.2d at 591. Our deference is particularly great where the PRC has gone beyond even the "reasonable inferences of causation" that permit "extended attribution" into the zone of "assignment." While both attribution and assignment involve inferences of causation, we observed in NAGCP I that "the latter concept permits a greater degree of estimation and connotes somewhat more judgment and discretion than the former." Id., 186 U.S.App.D.C. at 348 n.59, 569 F.2d at 587 n.59.B. Marginal Cost Pricing

Several parties to Docket No. R77-1 urged the PRC to reject the holding of NAGCP I.23 The PRC properly declined the invitation. But it did lament that NAGCP I foreclosed its continued adherence to what it referred to as the "most talked about new development in rate regulation today" so-called marginal cost pricing.24 See PRC Op. at 75-82.

We refer to this passage, although it is not the occasion of specific protest to us, for several reasons. We are concerned that it may reflect a hostility to our opinion that has resulted in an application first grudging and then inconsistent. We are also concerned that our ruling may have been carried too far, beyond what was originally envisioned. We would not wish our rulings to result, by virtue of misapprehension, in a loss of efficiency that was not required by or implicit in the statutory mandates.

As to "marginal cost pricing," traditional economic theory, if we may sketch our general understanding, holds that in a system of "pure" or "perfect" competition the most efficient allocation of resources occurs when the price of a good or service equals its marginal cost the cost to the economy of producing one additional unit. The marginal cost concept includes variable costs but not historic fixed costs ("sunk" costs), though it does include the cost of capital if capital investment is required to increase capacity for additional production. Traditional economic theory concludes that a price set at marginal cost achieves the optimum equalization of the current cost to society of employing scarce resources and the value to consumers of using those resources. In a system of perfect competition, an individual producer will employ his productive capacity to the degree that he can recover his variable costs, thereby minimizing average cost and maximizing attainable profit at the market price.

We do not presume to venture at large into economic theory, but it seems pertinent to observe even at this point of discussion, that the foregoing traditional theory presumes that a firm operates in an environment where market price is determined independently of the cost structure or management decisions of any individual enterprise, and where the firm tends to halt production when its marginal cost exceeds the market price.

The transfer of marginal cost pricing theory to regulated monopolies presents obvious difficulties, including the inapplicability of the basic free-market assumptions. However, it is appropriate to note that in the view of some economists, economic efficiency is similarly enhanced when the prices charged by regulated utilities are set at the long run marginal costs of the enterprise, including not only the variable cost of producing an additional unit with existing capacity, but the costs of adding additional capacity to meet increases in volume.25 Traditional ratemaking principles require, however, that total revenues generated by the regulated utility cover the historic costs of the enterprise, including a return on useful investment, as well as current operating and future capacity costs. In return for the guarantee of such a return the utility foregoes opportunities to charge the "market prices" that might be exacted from consumers. There is not a free market but a market under the constraint of what is, in effect, a long-range regulatory bargain. There is no certainty that the total revenues guaranteed by the regulatory understanding will equal the revenues that would be generated by pricing at marginal cost which is prospective in nature even though long run marginal cost embraces capital additions. (Of course the risk implicit in that lack of certainty is taken into account by the regulator in determining permissible rate of return.)

What economists frequently advance as the solution is the selective use of price discrimination permitting deviation from marginal costs in inverse proportion to the elasticity of demand for the utility service of various classes of users. The concept is that this minimizes inefficient allocation of resources, because those consumers who would, relatively speaking, consume the same amount of resources regardless of price pay the greatest differential from marginal costs, while those whose level of consumption is most susceptible to price considerations are charged the price closest to marginal costs.26

This is the kind of thinking that apparently informed the PRC in its first two ratemaking decisions, although it did not elaborate on its view. The PRC first established an initial rate floor in the process of determining "attributed" costs by ascertaining variable (or marginal) costs. Because strictly variable costs generated only about half the revenues necessary to recover the costs of the Postal Service, the PRC then sought to "assign" the remaining unattributed costs in inverse proportion to the elasticities of demand of the various classes.

In terms of economic efficiency, the advantage of marginal cost pricing for regulated utilities is that it serves as a means of "encouraging the maximum economic use of a company's services consistent with the so-called 'full-cost' requirement."27 A corollary is its potential for close control of consumer demand, by assuring that consumption choices reflect the current costs to society of providing the resource. Marginal cost pricing has attained some currency in electric power ratemaking, where efficient utilization of increasingly scarce energy resources is a matter of primary concern.28

In the context of postal ratemaking, however, the dominant objective of Congress, as ascertained by the court in NAGCP I, was not so much the regulation of demand for postal services, as the prevention of discrimination among the mail classes. In any event, the concern for maximization of use of capacity is less compelling where demand is inelastic. The court in NAGCP I noted that the Postal Service had conceded that demand for all classes of mail was essentially inelastic at foreseeable rates.29 It rejected the effort to assign almost half the costs of USPS on the basis of relative inelasticities of demand as inappropriate, as in effect permitting a discretion barred by the Act, and as unduly burdening first class mail, the most inelastic of the classes. NAGCP I certainly did not command a complete jettison of economic principles. It approved cost variability as an appropriate starting point for cost determinations. The path to be followed from that point on differed from that of the PRC in view of the congressional intent. This is not the only context in which a concern for equal or fair treatment yields results different from those obtainable if economic efficiency in the allocation of resources were the exclusive or even the dominant goal. The choice of goals and objectives is a policy choice of the legislature, and the court's function is to ascertain the legislature's choice and to apply it, including the assurance of faithful application by agencies which make decisions subject to judicial review.

It is to the credit of USPS and the PRC that they have departed from their initial approach to postal ratemaking and have sought with reasonable fidelity to apply the legislative intent as delineated by this court. The task that confronted them in this ratemaking proceeding was formidable. Yet at the present juncture the primary dispute over cost allocation revolves about the allocation of only 7.14 percent of the Postal Service's total revenue requirement.30 That is the percentage the PRC found "reasonably assignable" to certain classes of mail as "service related costs" calculated as the fixed delivery costs incurred in maintaining a six-day-a-week, as opposed to a three-day-a-week, delivery schedule. The parties raise numerous challenges to the rationality of this assignment pattern. With the deference we accord the PRC's expert judgment in developing cost allocation methods, we find this method of allocation to be reasoned, nonarbitrary, and congruent with the legislature's objective.

Further discussion of the issue requires that we delve in some detail into the mechanics of the cost allocation process. The PRC essentially followed the approach of USPS's initial presentation, although it developed its own figures in many instances. For convenience, we shall refer in our discussion only to the PRC.

A. The Domestic Mail Classification Schedule

It is useful to start with a brief sketch of the various mail classifications as they are defined in the Domestic Mail Classification Schedule.31 The four general classifications of mail are probably familiar to most mail users. First-class consists of letters and sealed parcels weighing 12 ounces or less, plus postal cards and post cards.32 Priority mail is a high priority service akin to first-class for sealed parcels weighing more than 12 ounces.

Second-class and Controlled circulation are available for magazines, newspapers and other periodicals. Second-class is divided into several subclasses and subcategories. Second-class regular rate is the one that encompasses the vast majority of second-class mailings. Second-class is essentially for bulk, not single piece, mailing.

Third-class is a service for single pieces and bulk mail, consisting of matter that is not mailed or required to be mailed as first-class, is not entered as second-class or controlled circulation, or weighs too little to qualify as fourth-class. Single piece service is used by individuals for small parcels and by organizations for minor distribution of parcels, catalogs and other printed matter. Bulk mail service is used by businesses, institutions or government agencies to send brochures, catalogs, samples or other matter in mailings of 200 pieces or more or with a total weight of 50 pounds or more.

Fourth-class includes parcel post (merchandise weighing 16 ounces or more); bound advertising matter; books, films, sound recordings and other matter entitled to a "special rate;" and library rate mail.

USPS also offers Express Mail (an expedited delivery service to and from some cities) and many other special services, such as certification, registration and C.O.D.

As its first step, the PRC determined USPS's revenue requirement by projecting its costs for each of 20 "cost segments"33 based on estimates of the volume of mail that would be generated at the proposed rates.34 The PRC determined the revenue requirement to be $17,585 million,35 modifying USPS's calculation slightly.

Within each cost segment, the PRC then determined the percentage of costs that could be said to vary with volume over either the short or long run. This percentage was then available for attribution to the various classes by application of an appropriate "distribution key." In the PRC's opinion, cost variability with volume remained the key to attribution. It stated that it could not have "reasonable confidence" that costs were in fact the "consequence of providing the service" unless

there is at least some showing of volume variability over the long run. If costs do not vary with volume over either the short or long run, there is still the possibility of cost assignment under (NAGCP I ), but we see no basis for attribution, if we are to maintain any distinction between attribution and assignment.36

In view of USPS's improved data on long run variable costs, the Commission found NAGCP I 's requirement of "extended attribution" satisfied. The PRC called for further improvements in data to lessen the need for reliance on "mere inferences of causation."37

The PRC apparently takes the position that "mere inferences of causation" beyond volume variability are not an appropriate basis for attribution.38 This is not entirely congruent with that part of the language of the NAGCP I opinion that prescribed Attribution of costs "through variability theory as well as through other reasonable inferences of causation."39 Since no party has challenged the PRC decision on this point, we do not pass on it.40 We do observe, however, that the PRC was struggling with the NAGCP I court's rather murky distinction between "attribution" (for reasons other than volume variability) and "assignment," both of which were to proceed from "inferences of causation."41 Taking into account the PRC's expansion of the scope of attributable costs based on variability, its relegation of costs derived from inferences of causation to the assignable category alone may well have been an appropriate application of NAGCP I. The context is one in which USPS's enhanced ability to attribute costs on the basis of volume variability resulted in the attribution, after PRC modification, of almost 65 percent of total costs.42

The PRC's assignment of service related costs proceeded from differences in service accorded the various classes of mail. The Postal Service Manual commits the Service to different standards of service for different types of mail, based on USPS's perception of customer needs, and transportation and distribution capabilities.43 For first-class mail and for newspapers (a "service category" within the second-class regular rate classification, also known as "red tag" mail44), the Manual requires delivery on "the first delivery trip" if the mail is received at a "distribution facility" prior to the cut-off time.45 For third- and fourth-class parcels, USPS is required to deliver on the next delivery trip all parcels received prior to the cutoff,46 but the post offices have more flexibility in determining that cutoff. For all other mail, including third-class mail and second-class publications not entitled to red tag service, the regulations provide for "delivery not later than the second day after receipt."47

Two witnesses provided the main evidence before the Commission. Postal Service witness Jellison, the Assistant Postmaster General for Mail Processing, testified on the differences in service standards. Jellison first ranked the priorities of treatment given the various classes of mail, in order of decreasing significance: (1) first-class mail, including priority mail and post or postal cards; (2) newspapers or red tag mail; (3) parcels (both third- and fourth-class); (4) ordinary papers (all second-class not entitled to red tag service and controlled circulation); and (5) circulars (fourth-class bound printed matter and all third-class except single piece parcels). Jel lison then characterized first-class and newspapers as "preferential" because of the requirement of immediate delivery upon receipt at the distribution facility; parcels as "borderline" preferential; and the other classes as "nonpreferential" or "deferrable" because of the ability to defer delivery of that mail for two days after it reached the distribution facility. The witness differentiated the levels of service afforded preferential and nonpreferential mail at each stage of the mail process between receipt and delivery. He asserted that while preferential and nonpreferential mail were often delivered together, deferrability of some classes permitted workload levelling, and avoidance of the need for overtime or part-time employees, at peak periods, with resultant efficiencies. Jellison concluded that because of the deferrability feature, nonpreferential service standards could be met by delivery every other day, as opposed to the current six-day delivery schedule. In his view, were it not for the existence of preferential mail, a three-day delivery system with recipients receiving mail every other work day would be feasible.48

Based on Jellison's conclusion as to the feasibility of a three-day delivery schedule, USPS witness Hume, a private consultant, sought to calculate the costs resulting from the need to maintain six-day delivery to meet preferential service standards.49 Although transportation and processing costs would likely be affected by initiation of a three-day schedule, Hume focused only on the delivery function. He found that in the other two functions, service related costs either were volume variable (and therefore would be accounted for in the attribution step), or could not be determined with sufficient accuracy from available data.50

Hume estimated the delivery costs that would be incurred in a three-day delivery system, assuming no variations in total mail volume. By subtracting this figure from the projected delivery costs in the six-day delivery system, he obtained the delivery costs occasioned by maintenance of a six-day schedule. Because variable costs would be attributed under either schedule, he then determined the proportion of delivery costs that would be fixed under each system. The difference in fixed costs he assigned as service related costs to the classes receiving preferential service.

The PRC approved the methodology proposed by the USPS witnesses. It found that nonpreferential mail is deferrable under the Postal Service's standards and that, in fact, such mail is frequently deferred.51 It agreed with the Postal Service that "it can reasonably be inferred that the preferred classes cause the differential between 3-day and 6-day delivery costs."52 Adjusting somewhat USPS's calculations, the PRC made an "assignment" of $1,256 million in test year costs, or 7.14 percent of the total revenue requirement.53

While asserting that, as a first effort, the use of service-related costs was satisfactory, the Commission urged further refinement in three areas: (1) the analysis used in determining the cost differentials resulting from service priorities at the delivery stage; (2) the development of data and methodology to reflect more accurately the actual hierarchy of delivery standards in the assignment of service related costs; and (3) more complete investigation into the possibility of extending service related costs to functions other than delivery.54

D. Objections to Assignment of Service Related Costs

In considering the various challenges to the assignment of service related costs, we recall that we judge a particular ratemaking methodology by whether it is reasoned, nonarbitrary and congruent with the statutory mandate. We do not insist on mathematical precision in developing new ratemaking methods; mere articulation of imperfections in a new scheme will not alone defeat it. This is especially true in cost Assignment, where, as NAGCP I recognized, there is a wide scope for the exercise of judgment and discretion by the PRC. With this overview, we conclude that the service related cost method reflects a reasonable inference of causation, and despite the imperfections identified by the parties, represents an acceptable first effort at meeting the strictures of NAGCP I. Our conclusion is supported by PRC's avowed intention to require refinement of the service related cost method in future rate proceedings.

The newspaper publishers55 attack the assignment of service related costs as arbitrary and unreasonable on the ground that nonpreferential mail is rarely, if ever, deferred in the delivery function. According to these parties, the current delivery system is designed to ensure that each carrier can deliver each piece of mail he receives at the delivery unit each working day. To further this design, the Postal Service conducts an annual review of its routes and adjusts them to guarantee delivery of all mail that comes into the delivery unit.56 On heavy volume days, delivery service managers may use overtime or auxiliary help, but deferral is discouraged.57 Instead, it is argued, most deferral occurs not at the delivery stage, but in mail processing units before reaching the delivery unit.58 Based on these facts, the publishers submit that "it is arbitrary and unreasonable to apportion costs upon the basis of a theoretical differential between preferential and nonpreferential mail in the delivery function."59

The publishers do not dispute that deferral can and does occur relatively frequently at some point in the mail handling process. The deferrability gives rise to differences between preferential and nonpreferential mail in the overall level of service provided. Accordingly, it was not unreasonable for the PRC to infer that the preferential classes may be called upon to bear the cost of providing the capacity I. e., six-day delivery necessary for this preferential service. The PRC reasoned that if all mail were nonpreferential, then three-day delivery would be feasible, and that the additional fixed costs resulting from maintenance of the six-day schedule could be determined. The primary source of such costs was found in the delivery unit, but the PRC recognized that, with improved data, such costs could be determined for other functions as well. In drawing this inference of causation, the salient feature was overall level of service, not whether, under current practice, the deferral that does take place occurred within any particular function.

Recognition that overall service levels, and not deferrability within one function, provide the key to service related costs also forms the basis for distinguishing the PRC's rejection of the citizens' rate. As explained by Judge Tamm in the companion opinion, that special rate was premised on deferrability in the delivery function Only.60 The PRC concluded that because such deferral was unlikely to occur in practice, there would be no difference in overall service that would justify an unequal allocation of delivery costs as between citizens' rate and regular first-class mail. On the other hand, in the case of preferential versus nonpreferential mail, there are conceded differences in overall service that do justify disparate cost assignments.

The newspaper publishers, joined by the first-class mail users,61 make a further argument. For this purpose they assume, in effect, the reasonableness of the PRC model in determining costs that are caused by the need to maintain a six-day delivery schedule. Although the non-preferential classes may not require six-day delivery, in practice they are often delivered without deferral, and so receive the benefit of six-day delivery capacity. By assigning "capacity" costs entirely to the preferential classes, they argue, the PRC gives the nonpreferential classes a "free ride." These parties contend that such a "free ride" is contrary to legal precedent and sound economics. They point to Colorado Interstate Gas Co. v. FPC, 324 U.S. 581, 65 S.Ct. 829, 89 L.Ed. 1206 (1945), as prohibiting such a "free ride" by theoretically "interruptible" customers who get the benefit of capacity provided to serve peakload customers. While Colorado Interstate does lean against free riders, it does not establish a legal prohibition against assigning all capacity costs to peakload customers. That case rather affirmed the FPC's discretion to require interruptible customers to bear some portion of capacity costs.

In general we agree with the first-class users that there is good reason, in economics and common sense, to call upon nonpreferential classes to contribute to the capacity costs used in providing them a benefit.62 However, as applied to the costs entailed in maintaining a six-day system, the question is not a simple one. There is a question whether and to what extent the users of this nonpreferential mail receive a benefit of value. On the record, we cannot hold that the PRC abused its discretion in choosing as of this ratemaking proceeding not to require the nonpreferential classes to contribute to the differential in capacity costs ascribable to a six-day system.

Various parties attack the PRC for failing to justify fully the operational feasibility of its hypothetical three-day delivery schedule. They also assert that the PRC erred in assuming that all other aspects of postal operations would remain relatively constant under a three-day system and in failing to consider the effects on costs of distortions in the postal system that would result from three-day delivery. The PRC did provide a discussion of the feasibility of the three-day system.63 While some of its assumptions are, to put it mildly, optimistic, we cannot say that the system is so unrealistic a model as to call into question the rationality of the conception underlying ser vice related costs. A model may be useful for purposes of analysis, though it premises a condition that has not been and indeed cannot be exactly duplicated in real life. A familiar instance is the model of Newton's laws of motion with its (unrealistic) premise of a world without friction.

More substantial is the contention that the Postal Service had an obligation to consider the effect of a three-day system on other functions of the postal system. However, the PRC did not reject the need to make its model as realistic as feasible. Instead, it found as a fact that the other functions of the Postal Service would remain relatively stable, without significant distortion from the institution of a three-day system.64 That conclusion finds support in the record.65 It must be remembered that we deal here with inferences of causation. The PRC may draw reasonable inferences of causation without resolving all subsidiary issues, especially those that would require extensive and disproportionate effort.

The magazine publishers66 challenge the reasonableness of the service related cost concept on the ground that the service standards on which it was based do not reflect any determination as to which classes actually "need" six-day delivery and which can therefore be said to "cause" the six-day delivery schedule. They argue that the existing service standards were adopted as the basis for service related costs without investigation into whether the classes receiving preferential service under the standards actually required that preference. Specifically, the magazine publishers contend that magazines require delivery only two days a week; hence, magazines, at least, cannot be said to "cause" six-day delivery.

The issue raised by this contention is substantial, but in the last analysis it does not warrant rejection of the PRC's methodology. While the mail system is bounded by constraints against undue preferences or discrimination, it need not be, and probably could not be, tailored to match perfectly individual customers' perceptions of their own service needs. In mail classification, judgments are properly made that similarities between types of mail or mail users justify their inclusion within a single class, even where there might be differences among them in the service desired. Thus, daily newspapers, which presumably demand prompt delivery, are included in second-class with magazines of various publication frequencies.67 Their similarities are evident, especially in the context of their joint entitlement to a congressionally mandated preference based on their key characteristics of regularity of publication and type of information conveyed.68 The Postal Service must be afforded broad discretion to manage its operations efficiently by developing service priorities. In making these decisions, USPS need not restrict itself only to considerations of customer need, but may also take into account such operational factors as transportation availability and distribution capabilities. So long as these service priorities are not manifestly unrealistic, USPS is not obliged to undertake a comprehensive review of its service priorities before those priorities may form the basis of cost allocations. In this case, it was not manifestly unrealistic for USPS to conclude that magazines published weekly or more often (and therefore entitled to "red tag" service) would, like daily newspapers, require the prompt delivery accorded by a six-day schedule.

In a similar vein, the magazine publishers and mailers of third- and fourth-class parcels69 contend that it was arbitrary to assign costs on an all-or-nothing basis depending on whether a class was designated pref erential or nonpreferential, when the service priorities as described by witness Jellison form not a simple dichotomy but rather a hierarchial array with first-class taking precedence over all other classes. The result, these parties contend, is discrimination the assignment of equal cost responsibility for unequal degrees of service. While there is some merit to this objection, we do not find it fatal in the context of this ratemaking proceeding. USPS and the PRC, in developing the service related cost method, endeavored to meet this court's mandate in NAGCP I. In the initial implementation of a new ratemaking approach, some leeway for approximation and estimation must remain. The courts often uphold regulatory actions on the premise that the approximations will be subject to refinement.70 The rough categorization of service priorities into preferential and nonpreferential was a reasonable first approximation in tracing the costs of providing the various levels of service. That this dichotomy may not suffice in future proceedings is highlighted by the PRC's exhortation to USPS to develop data and methodologies that will reflect more accurately the hierarchy of service priorities. We prefer to leave such efforts at refinement of methodologies in the hands of the PRC.

This approach also serves to dispose of the parcel mailers' further claim of error in the PRC's failure to determine actual operating priorities before assigning costs on the basis of "hypothetical" service priorities. Such a study, they believe, would prove that "borderline" classes I. e., parcels are actually afforded nonpreferential treatment. We do not find it unreasonable to draw inferences of causation without detailed studies of actual operating practices, at least in the context of an initial effort. We again note the PRC's request that USPS provide more accurate data on service priorities in future proceedings.

Finally, we do find substance to the claim of American Business Press, Inc. (ABP), an association of publishers of specialized business publications. Despite the fact that preferential or "red tag" second-class mail makes up only 40 percent of all second-class mail, the PRC assigned to all second-class, both preferential and nonpreferential, the service related costs caused by providing preferential service to red tag. We agree with ABP that assignment of service related costs to mail not receiving preferential service raises serious concerns of discrimination.71 In our preceding discussion, we have accepted, at least for this case, the PRC's adoption of a model based on a dichotomy between preferential and nonpreferential mail, even though that distinction may not entirely reflect actual service priorities. But the failure to distinguish in the assignment of costs between preferential and nonpreferential mail goes contrary to the PRC's own model.

Despite our concern, we decline to disturb the PRC's decision at this time. There is some doubt whether the issue was fully ventilated in the proceedings before the PRC.72 In any event, the Commission has chosen to deal with the issue as a classification matter, rather than in the context of a ratemaking proceeding. In Docket No. MC76-2, the Commission recently rejected a proposed surcharge on red tag mail, as well as a proposal to eliminate the current restriction on the availability of red tag to periodicals with a frequency of once a week or more.73 The Governors affirmed this aspect of the Commission's recommended decision, but rejected a recommendation that red tag be incorporated into the Domestic Mail Classification Schedule as a separate subclass of second-class.74 Both decisions indicated that a reexamination of the question would be appropriate in light of the adoption subsequent to the development of the record in that proceeding of the service related cost methodology.75 The PRC has recently initiated a proceeding, Docket No. MC79-3, to undertake that reexamination.76 Although, as this discussion reveals, mail classification and postal rate issues are often intertwined, the Act establishes separate, though parallel, procedures for considering each of them.77 While it is feasible, and perhaps desirable, to consolidate the consideration of classification and rate issues where appropriate,78 the PRC did not act arbitrarily in choosing to explore this matter in the context of classification proceedings. We therefore defer to the PRC's pending proceeding where the issue will receive further analysis, with this court's concern in mind.79

A. Undue Discrimination Against First-Class Mail

Several petitioner and intervenor states, large users of first-class mail, mount a generalized attack on both the rate schedule adopted by the Board of Governors in approving the PRC decision, and the cost allocations underlying that rate schedule. The cornerstone of their position is a comparison of the rates charged for allegedly similar pieces of first- and second-class mail receiving identical service. Thus, the rate for mailing a piece of presorted first-class mail weighing four ounces would be 52 cents, while the rate for a piece of second-class red tag mail weighing four ounces would 10.8 cents or less, depending on the degree of presortation.80 The States point to por tions of Postal Service witness Jellison's testimony supposedly proving that first-class and red tag receive identical service.81

The States claim that this disparity demonstrates discrimination Per se, since unequal rates are being charged for identical services.82 They also infer that first-class users are subsidizing the costs of providing second-class service. Adjusting for differences in "cost coverage"83 between first- and second-class rates, the States surmise that the actual cost under the PRC's cost calculation of mailing the four ounce first-class piece is approximately 40 cents and the cost of the comparable second-class red tag service is 10 cents. To the states, "simple common sense" suggests that, since both classes receive identical service, the rate for second-class red tag service cannot be covering the true cost of providing that service.

This argument does not undercut the reasonableness of the rate schedule. In the first place, it is not at all clear from the record that the service accorded red tag mail is identical to that given first-class. Jellison's testimony was that while second-class red tag mail is processed, transported and delivered concurrently with first-class mail the majority of the time, red tag was deferred as against first-class in heavy volume periods.84 This deferrability undercuts any easy conclusion that the services are identical, even though its cost consequences are unclear, since relative deferrability within the preferential category did not enter into the PRC's model for assigning service related costs.85 Further, in handling first-class mail, USPS must perform a number of functions such as pickup at a multitude of locations (mailboxes, office buildings) that are not required for handling second-class, which must be delivered presorted to the post office. It is accordingly likely that the cost of handling second-class is lower than the cost of handling first-class mail.

A more fundamental difficulty is that the States' charge of a gross disparity in the rates charged first-class and second-class pieces of identical weight presumes that cost responsibility is a function only of weight, and that weight alone is the proper basis for comparison. Yet, depending on the particular cost segment, cost responsibility may result from any one or a combination of a number of factors, including not only weight but such factors as cubic volume and number of pieces mailed. The PRC employs these factors as "distribution keys" to distribute the costs of any one segment among the classes based on an inference as to causation of the costs of providing that segment of service to that particular class.

It may be enlightening to compare the rates charged the average pieces of first-class mail and second-class red tag mail. The average weight of all first-class letters and sealed parcels is 0.5 ounce;86 the rate for this average piece would be 15 cents. The average weight of second-class regular rate publications is six ounces,87 and using a piece comparable to that hypothesized by the states,88 would be charged a rate of approximately 13.1 cents or less. After adjustment for cost coverage similar to that made by the States in their comparison, the cost to mail a first-class letter is 12 cents, to mail the second-class piece 13.1 cents. When comparison is made on a per piece basis, the claims of discrimination and subsidization lose their force. The point is not that this comparison proves absence of discrimination, but that reliance on any one key as the only test is inadequate.

We are sensitive to the first-class users' concern that the historical discrimination against first-class mail a concern that permeated the passage of the Act and the decision in NAGCP I89 may persist. We do not advance these defects in the States' model as demonstrating decisively that first-class is not the subject of discrimination and does not subsidize other classes. But they do expose in high relief the States' failure to relate their claims to the extensive cost data relied upon by USPS and the PRC. Artificial comparisons cannot substitute for a showing that the PRC's specific cost allocations were not supported by substantial evidence.

The States make an additional argument that the PRC relied upon "value of service" criteria to an impermissible extent, in contravention of the command of NAGCP I. Under a "value of service" approach costs are allocated not by tracing of cost causality, but by judgments as to the value of the service to the user or "what the traffic will bear."

We find the States' complaint hard to comprehend. It is directed at the testimony of Postal Service witnesses Sobin and McCaffrey. Sobin testified about projections of mail volume in the test year and estimates of "price parameters," or elasticities of demand, for the various classes of mail and mail services. These latter estimates of elasticity both entered into the projections of volumes in the test year before rate changes, and provided a basis for estimating the reduction in volume that would result from the increased rates recommended by the rate design witness.90 That witness was McCaffrey, who determined the appropriate cost coverages for the various mail classes and subclasses, applying the rate criteria of 39 U.S.C. § 3622(b). According to McCaffrey, "Based on my proposed rates, witnesses Sobin and Kluttz (the Postal Service's revenue requirement witness) computed after-rates volumes and revenues."91

In determining the appropriate cost coverages, which measure the allocation of residual costs after attribution and assignment, McCaffrey applied the discretionary criteria of § 3622(b), including "the value of the mail service actually provided each class or type of mail service to both the sender and the recipient." 39 U.S.C. § 3622(b)(2) (1976). We do not understand the States to challenge to the use of "value of service" in the allocation of residual costs to the various classes. Rather, they seem to challenge the reliance, in the initial determination USPS' costs, on projections of the volumes that will be mailed at the proposed increased rates, because these volume projections are derived by applying estimated elasticities of demand to the projected test year volumes at the rates prevailing before the increases. Since any estimate of demand elasticity will incorporate judgments as to the value to the consumer of obtaining the service, reliance on these estimates permits "value of service" to creep somehow into the process of cost attribution and assignment.

Giving the States' argument the most generous cast, we nonetheless find it without merit.

Contrary to the States' suggestion, it is not at all clear that volume projections based on demand elasticity enter into the process of cost attribution and assignment. Conceptually, attribution and assignment are functions discrete from determination of the revenue requirement or designation of the appropriate rate schedule. By definition, attributable and assignable costs vary, at least by inference, with volume. Unless there is an assumption of significant economies of scale, not here posited, the process of attribution and assignment of costs is not materially affected by total volume projection; whatever the volume, a rate will recover attributable and assignable costs. Total volume projections enter the calculation because it is necessary to design the rate schedule to recover not only "variable" (attributed and assigned) costs but also "institutional" (residual) costs. It can only be determined whether a given rate schedule will recover all institutional costs when there is an accurate projection of the volume that will be generated under that schedule. Demand elasticity is pertinent only in the stage of designing rates to recover residual costs, and that is precisely the area where the statute permits "value of service" to be considered.

Even if, for purposes of discussion, one were to accept the States' supposition that demand elasticity enters into cost attribution and assignment, such an effect would be incidental to analysis based on volume estimates. Volume estimates are vital to sound ratemaking. It is difficult to envision a ratemaking scheme that does not require estimates of volume, and it would be absurd to preclude the use of volume projections because their ascertainment requires some consideration of the relative value to consumers of the services.

Nothing in NAGCP I requires such a result. The court condemned the overt use of "value of service criteria," I. e., assigning almost half of the costs of the Postal Service on the basis of the relative inelasticities of demand of the mail classes.92 The court commended the use of cost-of-service principles. But whether the costs so ascertained, and rates based thereon, will suffice to cover revenue needs naturally requires volume projections, even though these involve some ascertainments of demand elasticity. This is inherent in any rational scheme of cost allocation.

Having been concerned to this point with the broad sweep of the PRC's performance under the Act, we now consider the lacunae of three specific decisions concerning second-class mail.

As noted above, Supra note 80, the second-class regular rate (for mailings of 5,000 or more copies, mailed outside the county of publication) consists of three separate charges: a piece rate, a poundage rate for the editorial portion of the publication, and a poundage rate for the advertising portion that is "zoned" according to distance. In Docket No. R77-1, the Commission adopted for the first time a three-tier piece rate that varies according to the degree of presortation.93 Mail presorted to carrier route is charged a piece rate of 4.4 cents. Mail presorted to five digit zip code (or three digit ZIP code for certain cities) is charged a piece rate of 5.4 cents. All other second-class regular rate is charged a piece rate of 7.0 cents.94 Although Postal Service regulations require that all second-class mail be presorted to the greatest extent possible,95 the two presort "discounts" are available only where the presorted mail is sorted into sacks either at least one-third full or weighing at least 20 pounds.

American Business Press (ABP) attacks the limitation on eligibility for the presort discount as discriminatory against smaller publications that cannot fill one-third sacks when the presorting is to five digit zip codes, either because of low circulation or wide geographic dispersion. Record evidence suggested that only a small percentage of all publications would be eligible for the discounts,96 although approximately 77 percent of regular rate mail, by volume, would be eligible.97 ABP argued before the Commission that the proper measure of eligibility for the presort discounts should be the six-copy bundle, the standard prescribed by postal regulations for separating bulk mailing from individual mailings.

We find the Commission's limitations on eligibility supported by the evidence. Postal Service witness McCaffrey stated that the one-third sack requirement in presortation was a standard USPS requirement.98 Witness Madison found the presort discount justified by cost savings.99 His study was based, however, on the assumption of a one-third sack limitation. The Commission noted that there was "no correlation between the Madison cost and volume data and the presortation of six piece bundles which is the eligibility requirement sought by ABP."100 ABP attacks as "not explained" the reliance on the Madison study.101 Its position presumes that the effect on USPS costs of presortation to six-copy bundles would not differ from the cost effects of presortation to one-third sacks. We cannot ignore or gainsay the likelihood that the significantly smaller six copy bundle would entail greater handling costs than the one-third sack. Since precise cost data was lacking with regard to the six copy bundle, the PRC was justified, at least in its first attempt at devising a second-class presort discount, in choosing to limit the availability of the discount to those mailings for which it had reliable data.

We find no basis for ABP's claim that, in adopting the second-class rate structure, the PRC focused on one ratemaking criterion the degree of preparation performed by the mailer102 to the exclusion of other factors, particularly the effect of rate increases on mail users.103

We read the PRC's opinion as reflecting a full consideration of the pertinent factors. The Commission expressly took into account the interests of small publications though not to the degree ABP desires by modifying the presort discount, limited to mailers using five digit zip codes, so as to be availa ble as well to those using three-digit city zip codes (though still subject to the one-third sack restriction).104 The Commission also rejected a zoned editorial charge, because small publications are least able to avoid the highest zone charges through the device of multiple points of entry.105

B. Average Weight of Second-Class Publications

The magazine publishers attack as contrary to the evidence the PRC's adoption of the Postal Service's estimate that the average weight of a second-class regular rate publication in the test year would be 6.1 ounces.106 They assert that this figure represents an unjustified extrapolation from a recent downward trend in the average weight of second-class publications. Invoking a survey of Magazine Publishers Association (MPA) members, the publishers argue that the downward trend has been reversed, and that average weight has actually increased. The result is a $29 million underestimation of the revenues that will be generated by the two poundage charges, which together account for 52 percent of second-class regular rate revenues.107

It is not our function to weigh again evidence that was carefully considered by the PRC.108 The Commission discounted the MPA study as "inconclusive" because its sample MPA members was not necessarily representative of all second-class publications. The Commission further noted that the seeming downward adjustment of average weights in the test year to 6.1 ounces from the average weight of 6.2 ounces found in fiscal years 1976 and 1977 was not a downward extrapolation at all, but was merely the product of "rounding the detail actually used in the projection."109 Finally, the Commission concluded that any increase in revenues resulting from increased weight would be offset, at least partially, by increased costs, and that, even assuming such a revenue excess, the resulting second-class regular rate cost coverage of 105 percent would not be unreasonable.110 Though the evidence is in conflict, a substantial evidentiary basis for the PRC's decision is present, and we have no warrant to substitute the court's assessment of the evidence for the considered judgment of the Commission.

C. Attribution of "Individual Load Time"

The magazine publishers also attack the PRC's conclusion that "individual load time" varies entirely with volume and therefore is 100 percent attributable. "Load time" generally is "the time spent depositing mail in delivery receptacles."111 Load time is divided into "batch load time" and "individual load time." Batch load time is the "time spent at delivery points placing batched letter size mail in receptacles or otherwise delivering mail in batches."112 Individual load time is the "time spent at delivery points placing individual pieces of mail into receptacles or otherwise delivering individual pieces."113

Since letter-size mail is "batched," individually loaded mail consists primarily of second-class publications. Because, in theory, individual loading requires a separate step for each piece, the Commission has consistently found it 100 percent variable with volume, and has attributed individual load time costs, primarily to second-class. The magazines allege that this attribution is patently arbitrary because it proceeds from an assumption that a carrier at a delivery point will load batches and magazines in separate steps, rather than by first combining the magazine with the batch and placing them into the receptacle together.

There is some force to the magazines' position. The Commission concluded that load time includes "all time involved handling mail at the receptacle."114 Under this view, even if the magazines were correct in their point, load time would include the time spent combining the batches and the individual pieces. This does not entirely jibe with the definitions of load time, which seem to restrict that function to the act of placing mail in the receptacle. The problem appears to stem from a deficiency in accurate data as to the proportion of second-class regular rate mail that is loaded individually,115 and uncertainty as to where in the cost model to account for the steps of combining the individual pieces and batches.116

Despite some misgivings, we are not inclined to remand on this point alone, in view of its relative lack of consequence and the fact that the PRC has insisted upon improved data in future proceedings. Postal ratemaking continues to evolve and we cannot yet expect USPS and PRC to have gleaned every bit of data and developed every model to perfection. Considering the PRC's immense task in digesting huge quantities of data and meeting the requirements of the Act, we believe it a sound approach to forgive such minor transgressions.

We reemphasize our general satisfaction with the Postal Service's and the PRC's performance in the cost allocation aspects of this ratemaking proceeding. In the face of a difficult task, they have proved responsive and responsible. By this we do not intimate, nor do we read the PRC opinion so as to suggest, that the process may remain static. Further improvements are feasible, and expected. That is not only our hope but our premise, and it is on that premise that we affirm the cost allocation decisions.

We are called upon to review decisions of the Governors of the United States Postal Service (Governors)1 approving the Postal Rate Commission's (PRC) recommended decision2 in docket no. R77-1, a ratemaking proceeding.3 The petition for review gives rise to numerous issues involving the Postal Reorganization Act of 1970 (Act)4 and interpretations thereof.5 We have considered challenges to postal cost methodology6 and certain additional claims7 in the companion opinion penned by Judge Leventhal. We now turn to three final areas of the PRC's and Governors' actions which are questioned on review: (1) the rejection of the citizens' rate mail (CRM) proposal; (2) the provision in the revenue requirement for recovery of accumulated losses; and (3) the imposition of constraints on fourth-class mail rates found to be cost-justified. We treat these challenges Seriatim.

The Board of Governors initiated the CRM proposal by authorizing8 the United States Postal Service (Postal Service) to request a recommended decision on a change in the classification schedule that would lead to the adoption of a new subclass for citizens' rate mail, with appropriate rates.9 CRM, as proposed by the Postal Service, would be a separate first-class service to be used by individuals for the mailing of their personal correspondence. Users of CRM would be entitled to a lower rate of thirteen cents in exchange for service inferior, to some degree, to regular first-class service, which would be available at a higher rate of sixteen cents. CRM would have to comply with standard size restrictions and be fully ZIP-coded, and would be denoted by a special stamp.

According to the Postal Service's proposal, all first-class mail would be treated alike prior to the delivery stage of the mail process. However, CRM would be "deferrable" in the delivery function.10 If exigencies required, mail carriers would separate the CRM mail from the first-class mail at the delivery unit, and set aside sufficient CRM to alleviate the immediate delivery problem.11 ] While delivery of the remaining regular first-class mail would be accomplished on the first delivery trip, delivery of the segregated CRM would be deferred until the second delivery trip.

The PRC held prolonged hearings on the CRM proposal, and considered briefs, comments filed, and oral argument by the parties to the proceedings and numerous intervenors.12 The PRC concluded that the evidentiary record did not support the implementation of a CRM service,13 as requested by the Postal Service, and did not establish that the CRM proposal would meet the service, ratemaking and classification requirements14 of the Act. As a result, the PRC recommended to the Governors that the CRM proposal not be adopted, and that the regular first-class mail rate be increased only to fifteen cents rather than sixteen cents, as proposed. One week later, the Governors approved the PRC's recommended decision.15

The National Association of Greeting Card Publishers (NAGCP)16 makes a multi-pronged attack on the administrative decisions declining to adopt the CRM proposal. NAGCP contends that the Governors abused their discretion in approving the PRC's recommended decision after finding it not fully acceptable. NAGCP also directly challenges the PRC decision on narrower grounds. NAGCP alleges the PRC erred in determining that: (1) the proposed separate citizens' classification was not justified on the record before the agency; and (2) CRM must bear service-related costs, and the revenue and cost estimates before the PRC were too uncertain.

NAGCP bases its general objection to the Governors' ratification of the PRC rejection of the CRM proposal on qualifying language employed in the Governors' decision. Although the Governors found that the Commissioner's various opinions did not "add up . . . to a fully acceptable analysis"17 and that the PRC was "less than wholly persuasive,"18 the Governors nevertheless approved the PRC decision "with reservations on part of the Commission's analysis."19 According to NAGCP, the Governors abused their discretion by approving an unacceptable decision.

However, contrary to NAGCP's assertion, the Governors did not approve a decision they found unacceptable. The mere fact that the Governors found the PRC decision to be not "totally acceptable" or "less than wholly persuasive" does not vitiate the Governors' approval. Adherence to NAGCP's line of argument would mandate the Governors' rejection of any PRC decision with which the Governors were not totally satisfied. Such a mandate is unsupported by logic or by statute. The statutory provision applicable to the Governors' action, 39 U.S.C. § 3625 (1976), specifically allows for varying degrees of acceptance or rejection of a PRC recommended decision.20 The Governors were not forced to choose between total acceptance or total rejection of the PRC decision. They could have, under protest, allowed the recommended decision to take effect and either sought judicial review or returned the recommended decision to the PRC for reconsideration.21 The Governors alternatively could have chosen to modify the PRC decision themselves. The Governors obviously were not sufficiently dissatisfied with the PRC decision to reject it, allow it under protest, or modify it. We decline to add the judicial veneer of a non-statutory requirement that the Governors find a PRC decision "wholly acceptable" in order to approve it.22

NAGCP attacks the PRC's rejection of the CRM proposal as violative of the classification23 and ratemaking24 provisions of the Act. The Governors' order, which is the subject of our review, interprets the PRC rejection of the CRM proposal as resting on two basic grounds, only one of which they question as less than wholly persuasive. The Governors read the PRC decision as boiling down to a "recommendation that (1) the record does not support a finding that 'individuals' is a proper grouping for Classification purposes, and therefore, the proposal violates (39 U.S.C.) § 403(c), and (2) the proposed Rates for citizens' rate mail do not meet the requirements of (39 U.S.C. s) 3622(b)(3)."25

The PRC's conclusion that the CRM proposal creates an impermissible classification under section 403(c) is based on two findings: (1) the record reveals no material service distinctions from regular first-class mail justifying a separate classification; and (2) even assuming Arguendo that a material service distinction exists, the record does not support a finding that household users are a special group qualified for separate classification.26

The PRC split">split over the question whether deferrability of CRM justified a separate classification.27 The Governors found that the deferrable feature of the CRM proposal distinguished it from regular first-class mail for classification purposes.28 The Governors therefore considered the alternate PRC ground for its determination that CRM may not be separately classified, I. e., a subclass for use only by individuals is unjustified.29

According to the Governors, the PRC inappropriately phrased the crucial classification question as whether 39 U.S.C. § 101(a) (1976)30 accords a "special status" to "personal correspondence from individuals."31 Rather, the issue is whether a distinction might be drawn between business originated mail and mail sent by individuals. The Governors found that section 101 was hospitable to such a distinction.32

The Governors found the PRC's reasons for its rejection of the CRM proposal as an impermissible classification not wholly persuasive, but did not find the classification issue to be dispositive. Thus, we need not endorse or condemn the PRC classification conclusion. As noted by the Governors, the statutory scheme may well be hospitable to CRM as a separate classification, and our declination to reach this question does not, by any means, presage a future rejection of a future CRM proposal.

After reviewing the PRC classification decision, the Governors considered the remainder of the issues that attended the CRM rate proposal issues concerning the assignment of service related costs (SRC) to CRM and the record's uncertainties in the subclass' volume, revenue, and effect on postal operations.33

a. Assignment of Service Related Costs to CRM

In determining the unit costs for CRM, the Postal Service did not assign to CRM any service related costs.34 The PRC decided that exemption of CRM from SRC assignment would be proper only if "CRM could be deferred on an operational basis to meet Postal Service needs should appropriate occasion therefore arise."35 Based on the evidence, the PRC concluded that CRM is not easily separable from "preferential mail," which requires immediate delivery,36 and thus, not readily susceptible to actually being deferred at any point in the mailstream.37 The PRC decided that insufficient operational diminution in first-class service for CRM existed, due to the likely infrequency of deferral of CRM38 and the resultant minimal delay in delivery, to warrant the total exemption of CRM from SRC.39

NAGCP claims that the same material service differences which persuaded the PRC to endorse assignment of SRC to preferential mail and nonassignment of SRC to nonpreferential mail40 support assignment of SRC to regular first-class mail and nonassignment of SRC to deferrable CRM. Therefore, NAGCP asserts, the PRC discriminatorily assigned SRC across the board to all first-class mail, while drawing finer and more accurate distinctions in its assignment of SRC to preferential as opposed to nonpreferential mail.41

b. Uncertainties in Revenue and Cost Estimates

The PRC determined that uncertainties in the record on CRM's volume, revenue, and effect on postal operations, factors which go to the heart of a rate decision under 39 U.S.C. § 3622(b)(3) and (4),42 made informed consideration of the CRM proposal impossible. For example, the PRC determined that there was insufficient evidence in the record to estimate CRM volume; and, therefore, it could not estimate the loss in regular first-class mail revenues that would result from CRM, or the revenue that would be produced by CRM itself.43 The PRC also determined that the Postal Service's cost estimates for CRM were questionable.44 In general, the PRC concluded that the evidence revealed an extreme range of unlimited uncertainty as to revenue and cost estimates.45

According to NAGCP, the PRC's alleged requirement for certainty in estimates was arbitrary and capricious. More specifically, NAGCP points to the inherent uncertainty in estimates for new utility services and to the possible formulation of a compromise estimate based on the existing evidence.

c. Judicial Review

It is the court's function, at this juncture, to determine whether substantial evidence in the record justifies the PRC's rejection of the CRM proposal on the basis of uncertainties in volume, revenue and cost estimates which would render application of the ratemaking criteria in section 3622(b) a futile exercise. It is clear that the PRC did not create an insuperable barrier of certainty for the CRM proposal. It merely concluded that adequate information was not available in the record by which a reasonable estimate could be made. The lack of information is understandable in light of the late stage at which CRM was introduced.46 The PRC concluded that the Postal Service's revenue estimates were based on questionable studies,47 and that questions concerning assignment of SRC to CRM eroded the Postal Service's cost estimates. In light of these uncertainties in the record, we conclude that the PRC's rejection of CRM was based on substantial evidence.

The issue whether Vel non CRM should bear SRC further obfuscates the uncertain CRM cost estimates. We are unpersuaded by NAGCP's attempt to analogize nonassignment of SRC to nonpreferential mail, with that of nonassignment of SRC to CRM. The overall level of service received by nonpreferential mail justifies disparate SRC assignments,48 but there are no comparable differences in overall levels of service between the proposed CRM and regular49 first-class mail. In contrast to its finding on nonpreferential mail, the PRC found that the infrequency of deferral of CRM and the resultant minimal delay, did not support the necessary finding that CRM could be deferred on an operational basis. Thus, it concluded that CRM as proposed should bear at least a portion of SRC. This is not to say that an across-the-board assignment of SRC to all first-class mail, both to CRM and regular first-class, is warranted. We have noted our concern, in the context of the treatment of "red tag" mail and other regular rate second-class, that the assignment of SRC reflect service priorities within a class.50

We need not decide whether CRM should be totally assigned, or exempt from, SRC. Rather, we find that if CRM is to bear SRC, such assignment should be finely tuned. The record clearly does not provide complete data to compute a ratio of assignable SRC to CRM. The fact that the PRC declined to take the initiative to formulate such a ratio based on the Postal Service's estimates and those of other parties, does not constitute an abuse of discretion nor arbitrary or capricious action.

In sum, we affirm the Governors' order which accepts the PRC recommendation to reject the CRM proposal. However, we, like the Governors, do not rest our decision on the PRC classification conclusion. Rather, we agree with the Governors and the PRC that the wide range of uncertainties in the record surrounding the CRM proposal rendered rational and informed application of the ratemaking provision51 a near impossibility. However, we emphasize that our decision is strictly limited to the CRM proposal and the record in this ratemaking proceeding. Our affirmation of the PRC's rejection of this CRM proposal does not imply that future CRM proposals, supported by a fuller, more substantial record, should be rejected. By contrast, we echo Chairman DuPont's encouragement to the Postal Service to restudy the CRM concept and offer a new proposal in future proceedings.52

We now turn to the question whether the Postal Service may include a provision for the recovery of prior years' operating losses in its revenue requirement under 39 U.S.C. § 3621 (1976).53 The PRC54 and the Governors55 first approved this practice in the 1976 ratemaking proceeding (docket no. R76-1), review of which was dismissed as moot.56 The PRC's and Governors' decisions in the proceedings now under review approved the recovery of this item in the revenue requirement, relying almost entirely on their 1976 decisions. We hold that the language of 39 U.S.C. § 3621 and the legislative intent underlying the Act57 allow the Postal Service, in the exercise of its sound economic judgment, to recover these accumulated operating deficits through the revenue requirement in these proceedings.

The fiscally sound operation of any organization rests upon the maintenance of a balance between costs and income. The statutory mandate of 39 U.S.C. § 3621, which applies this economic principle to the Postal Service, provides in part:

Postal rates and fees shall provide sufficient revenues so that the total estimated income and appropriations to the Postal Service will equal as nearly as practicable total estimated costs of the Postal Service. For purposes of this section, "total estimated costs" shall include (without limitation) operating expenses, depreciation on capital facilities and equipment, debt service (including interest, amortization of debt discount and expense, and provision for sinking funds or other retirements of obligations to the extent that such provision exceeds applicable depreciation charges), and a reasonable provision for contingencies.

Pursuant to this "break-even" standard, the Postal Service must set its postal rates and fees at a level which ensures that costs are met by income from rates and fees and by appropriations. The Postal Service's "revenue requirement," which represents the overall amount of income which must be collected from rates and fees, thus forms the basis for determining appropriate rates for postal services.

The PRC decision before us states "that the requested provision for recovery of prior year losses is reasonable and in accord with precedent established in Docket No. R76-1."58 Due to the relatively cursory treatment of this issue in the PRC decision now under review, we have looked to the PRC decision in docket no. R76-1 for the explanation of this item and the rationale supporting its inclusion in the revenue requirement.59 Our scope of review is limited to whether a fair reading of section 3621 and the legislative intent of the Act provide authority for the Postal Service's election to recoup prior years' losses through the revenue requirement, rather than through borrowing or appropriations.C. Computation of Prior Years' Losses

Operating deficits have plagued the Postal Service for decades.60 The Act was passed to solve this problem, but unfortunately the Act's goals have failed to reach fruition. The Postal Service has continued to incur operating losses.61 It has attempted to offset these losses by drawing down working capital,62 incurring debt,63 reducing services, and increasing rates.64 The Postal Service has also turned to Congress seeking appropriations. See Postal Reorganization Act Amendments of 1976 (1976 Amendments).65

In the proceeding now under review, the Postal Service computed the total amount of unrecovered prior years' losses at $2.336 billion,66 which is the excess of net operating losses and deficits from 1971 to 1978 over the $1 billion subsidy appropriated by Congress in the 1976 Amendments.67 The PRC proposes to recover the $2.336 billion over a seven year period, from 1978 to 1985, thus yielding an annual recovery of $334 million.68 It is the inclusion of this $334 million in the revenue requirement as recovery of prior years' losses which petitioners challenge.

Although the PRC decision under review fails to provide us with a detailed breakdown of the composition of the $334 million, it is apparent that these losses have a hybrid nature: (1) a portion of the prior years' losses are represented by operating debt; (2) the remaining segment is not represented by debt obligations, but was incurred by drawing down working capital.69

Under 39 U.S.C. § 2005(a) (1976),70 the Postal Service is authorized to borrow money to meet operating expenses and to issue obligations therefor. "The Postal Service may . . . pledge and use its revenues and receipts for the payment of the principal of or interest on such obligations." 39 U.S.C. § 2005(b) (1976). The resultant debt obligations, according to the PRC, are includible in the revenue requirements under section 3621, as "debt service (including . . . provision for sinking funds or other Retirements of obligations to the extent that such provision exceeds applicable depreciation charges) . . . ."71

Petitioners assert that the term "total estimated costs," as used in section 3621, limits costs that may be included in the revenue requirement to those that will be incurred in a future period. They also contend that the phrase, "to the extent that such provision exceeds applicable depreciation charges," limits debt service that may be included in the revenue requirement to those debts incurred for capital purposes. They conclude that debt incurred for operating expense, which is nondepreciable, is thus nonincludible in the revenue requirement under section 3621.

The PRC decision in docket no. R76-1, upon which the PRC and the Governors relied in this case, considered sections 2005 and 3621 together and interpreted section 3621 as permitting recovery of an amount for payment of obligations issued pursuant to section 2005.72 The PRC decision also rejected the "capital debt" argument that petitioners advance in this court. The PRC explained that the limitation on recovery of debts "to the extent that such provision exceeds applicable depreciation charges" merely prevents double recovery in the revenue requirement, once through depreciation charges and again through charges for debt service.73

We find that the plain language of section 3621 clearly supports the recovery, through the revenue requirement, of prior years' losses represented by debt. A common-sense reading of the statute refutes the restriction of "estimated costs" to costs that are incurred in the future.74 The stat utory language permitting recovery of an amount for "retirement of debt obligations" clearly denotes recovery for costs (debt obligations) incurred in prior years.75

We also conclude that the language of section 3621 does not support a narrow reading of "debt service" that would restrict it to debt incurred for capital expenses. Congress specifically provided that the Postal Service could incur debt both for capital improvements and for operating expenses. It differentiated between the two outlays in section 2005(a) solely for the purpose of setting a debt ceiling in connection with operating expenses. Congress declined to distinguish between the capital or noncapital purposes for which debt had been incurred when it provided for the inclusion of debt service in the revenue requirement under section 3621. The fact that depreciation charges must be deducted when including a provision for capital debt in the revenue requirement does not mean that a provision for noncapital debt cannot be included in the revenue requirement at all.

As a broad challenge to the recovery of prior years' losses, whether Vel non represented by debt, petitioners also cite the conventional public utility rule which eschews recovery of past losses in current rates. Galveston Electric Co. v. Galveston, 258 U.S. 388, 393, 42 S.Ct. 351, 66 L.Ed.2d 678 (1922). Recognizing that its interpretation may represent a departure from this conventional rule of regulatory practice, the PRC questioned the rule's applicability to a nonprofit">nonprofit organization, such as the Postal Service, which operates on a statutory break-even standard and may not recoup prior years' losses through later profits.76

We do not find that the recovery of prior years' losses is barred by a public utility rule barring recovery of prior losses in current rates. See Galveston Electric Co. v. Galveston, 258 U.S. at 393, 42 S.Ct. 351; Communications Satellite Corp. v. FCC, No. 75-2193, slip op. at 22 n. 19 (D.C. Cir. Oct. 14, 1977). Even if the Postal Service were treated as a conventional utility, the rule might be inapplicable in view of the restrictive circumstances surrounding the Postal Service's financial decline.77 In any event, it would be incongruous for us to clamp the shackles of a narrow rule onto the Postmaster General's attempt to return the Postal Service to financial stability, especially in light of the congressional intent to provide the Postal Service with the "unfettered authority and freedom it has been denied for years to maintain and operate an efficient service."78 Therefore, section 3621 statutorily authorizes the recovery of prior years' operating losses represented by debt through inclusion in the revenue requirement.

E. Recovery of Losses Not Represented by Debt Obligations

The issue whether the Postal Service is authorized to recover prior years' losses not represented by debt obligations through inclusion in the revenue requirement is not as easily decided. The PRC relied primarily on congressional intent and the policy of the Act to justify inclusion of these losses under section 3621. The PRC pointed out that one of the major policy determinations of the Act was to give postal management the flexibility to operate an efficient and economical postal system.79 It would be inconsistent with this policy, the PRC reasoned, to limit recovery of losses only to those represented by debt, because this would compel the Postal Service to borrow, and thereby incur attendant interest expense, when the cash needs of the Postal Service would permit drawing down existing funds.80 The PRC concluded that " 'the general principles of postal self-sufficiency which underlie the Postal Reorganization Act contemplate the recovery of prior year losses through rates and fees,' to an extent not limited to those losses represented by debt."81

The Postal Service points out in this court that Congress, in directing that " 'total estimated costs' shall Include (without limitation) operating expenses, depreciation . . . , debt service . . . , and a reasonable provision for contingencies,"82 clearly created a nonexclusive list of costs to be included in the revenue requirement. The Postal Service also asserts that principles of statutory interpretation accord weight to the PRC's application of a broad statutory term83 and that the PRC's decision was within the range of permissible choices contemplated by statute.84 The Postal Service contends that congressional intent justifies the inclusion in the revenue requirement of losses not represented by debt because these prior years' losses were incurred as an alternative to borrowing, and allowing their inclusion in the revenue requirement would serve a purpose similar to the provision for retirement of debt obligations.

In opposition, petitioners maintain that only an "open-ended, virtually unlimited construction" of the "without limitation" language in section 3621 would justify the inclusion in the revenue requirement of prior years' losses not represented by debt.85 Finally, opposing parties make the economic policy arguments that blanket recovery of prior years' losses through rates is a strong disincentive to the efficient and economical management mandated by the Act, and that inclusion of prior years' losses will permit rates to be set at levels resulting in a statutorily impermissible profit.

Upon consideration of the parties' numerous arguments, and after examination of the legislative history of the Act, we hold that the Postal Service's election in these proceedings to recover prior years' losses not represented by debt through inclusion in the revenue requirement under section 3621 comports with the legislative intent of the Act and is statutorily permissible. Although losses not represented by debt are not explicitly enumerated as includible in the revenue requirement under section 3621, we do not find this to be a bar. The plain language of the statute provides that " 'total estimated costs' shall Include (Without limitation )"86 the enumerated items. We are hard-pressed to envision language less restrictive than this.87

This is not to say, however, that the term "total estimated costs" is so broad as to be without meaning, or that the phrase "without limitation" grants the Postal Service Carte blanche to include any amount as costs in the revenue requirement. The question whether a particular cost is includible "is one of specific application of a broad statutory term in a proceeding in which the agency administering the statute must determine it initially." NLRB v. Hearst Publications, Inc., 322 U.S. 111, 131 (1944) 64 S.Ct. 851, 860, 88 L.Ed. 1170. Its determination "is entitled to acceptance unless it lies outside the range of permissible choices contemplated by the statute." Hardin v. Kentucky Utilities Co., 390 U.S. 1, 8, 88 S.Ct. 651, 656, 19 L.Ed.2d 787 (1968).

In specifying items as "costs" within the meaning of section 3621, the "range of permissible choices contemplated by the statute" includes costs which are demonstrably related to those specifically enumerated in section 3621. The prior years' losses in question are demonstrably related to operating expenses, which are enumerated in section 3621, because they were incurred to pay for operating expenses. Moreover, prior years' losses not represented by debt are of the same character as those losses represented by debt, which the plain language of the statute clearly authorizes,88 because the former were incurred as an alternative to the latter. To hold that the recovery of the losses in question is barred because they are not represented by formal debt would be unduly formalistic and in contravention of the specific statutory purpose of maintaining a financially stable enterprise that operates on a "break-even" basis.

Consideration of the legislative history of the Act convinces us that the Postal Service's determination has "reasonable support in relation to the statutory purpose." Hardin v. Kentucky Utilities Co., 390 U.S. at 9, 88 S.Ct. at 656. In enacting the legislation in question, Congress responded to the severe and imminently catastrophic financial crisis with which this nation's postal system was faced.89 Prior to the Act, the postal system was caught in a seemingly hopeless situation of chronic debt and disorganization. Its descent into bankruptcy and financial chaos appeared inevitable, and the then restrictive laws and archaic postal organization made relief unlikely.90

Congress then moved affirmatively to rescue the system from impending disaster. It reorganized the postal system and created an independent United States Postal Service. Recognizing this country's need for a viable system of communication by mail, and acknowledging the cumbersome shackles of existing postal regulation, Congress approved the new statutory scheme to meet "the obvious requirement that postal management . . . be given the unfettered authority and freedom it has been denied for years to maintain and operate an efficient service."91 This freedom included giving the Postal Service control over costs as well as the tools with which to exercise that control.92 The tools include the power to borrow,93 to raise revenues from ratemaking,94 and to seek appropriations.95

When continually mounting deficits and the sluggish movement of ratemaking proceedings96 prompted the Postal Service to seek relief from its losses through appropriations in 1976, the congressional spirit of optimism and revitalization which had accompanied the 1970 Act was tempered with the realization that the Postal Service's struggle with operating deficits would continue.97 To alleviate part of this pressure, Congress enacted the 1976 Amendments,98 thereby appropriating $1 billion to be applied to the operating losses for 1976 and 1977. Congress left intact the remaining amount of loss, recovery of which the Postal Service has determined essential to achieve the goal of a financially stable enterprise.99

The Postal Service has selected, as its method of financing in this ratemaking proceeding, the inclusion of these losses in the revenue requirement. We do not find its choice an impermissible one in this proceeding, nor do we find it inconsistent with the "central purpose of the (Act) to place the Postal Service on a self-sufficient basis." Direct Mail Advertising Association v. USPS, 147 U.S.App.D.C. 394, 400, 458 F.2d 813, 817, Cert. denied, 409 U.S. 843, 93 S.Ct. 43, 34 L.Ed.2d 83 (1972). Upon consideration of the legislative history of the Act, we agree that recovery of prior losses not represented by debt in this proceeding has reasonable support in relation to the statutory purpose. See Hardin v. Kentucky Utilities Co., 390 U.S. at 9, 64 S.Ct. 851.

We find the broad economic policy arguments raised by those challenging the recovery of prior years' losses unpersuasive. The various economic stratagems and their effects are matters entrusted to the wise discretion and expertise of the Postal Service.100 We will not substitute our judgment concerning the economic wisdom of its decision to recoup these past operating losses through section 3621, because its decision is consistent with the statutory purposes of the Act the Postal Service should be financially self-sufficient101 and should "break even."102

The recovery of past losses in the revenue requirement is not to be construed as a mechanism for the approval of costs imprudently incurred.103 In its recovery of prior years' losses in general, the Postal Service must comply not only with the statutory policy of financial self-sufficiency, but also with the congressional mandate of honesty, efficiency, and economical management.104 In the context of these ratemaking proceedings, we find that the language and purpose of the Act support the Postal Service's, the PRC's, and the Governors' decisions to allow the recovery of the past operating expenses not represented by debt through inclusion in the revenue requirement under section 3621.

A final challenge to the PRC's and Governors' decisions focuses on whether the PRC's imposition of "constraints" on fourth-class parcel post rates found to be cost-justified is arbitrary and in violation of 39 U.S.C. § 3622(b) (1976).105 We affirm the constrained rate schedule for parcel post in this ratemaking proceeding, because the constraints temper the possibly deleterious impact of restructured parcel post rates and advance the statutory goals of the Act. We rely on the PRC's assurance that this interim path of moderation, which is the product of considering the best available evidence in light of section 3622(b), is not a dead-end but will inevitably lead to a finer tuned compliance with the statutory mandates for ratemaking.106

Parcel post rates have traditionally rested on weight and distance factors. Eight distance-based categories and sixty-nine weight categories107 intersect to form a matrix of 552 zones for computing parcel post rates. Within each zone, the rate is based on a discrete per piece and per pound charge. In the past, parcel post rate increases have consisted of across-the-board escalations for all of the 552 weight-distance rate cells.

In an attempt to comply with NAGCP I, the Postal Service proposed a new methodology of rate design, which would recover all transportation costs for parcel post through the per pound charge and all remaining costs through the per piece charge.108 One of the basic assumptions underlying this proposal is that all costs, with the exception of transportation costs,109 are insensitive to changes in weight up to thirty pounds, and again from thirty pounds to the seventy pound limit.110

The PRC was not convinced that the Postal Service's rate design had been verified by empirical methods. It was concerned, moreover, that the drastic rate increases and decreases that would be caused by the proposal would have a harmful effect on users, competitors, and the Postal Service. The PRC thus undertook an independent analysis of the rate design.111

The PRC's analysis revealed that insufficient evidence supported the proposal's premise that non-transportation costs were insensitive to weight112 and that the Postal Service incorrectly applied the proposed rate component for transportation costs.113 The PRC therefore proposed a modified cost-based rate design based on data from the record.114 The rates that resulted from the PRC's cost-based development would have effectuated dramatic increases and decreases in the current rates in many rate cells. To diminish the potentially deleterious effects of such drastic changes, the PRC placed the following restrictions on its recommended rates: (1) increases in each rate cell are not to exceed fifty percent of the former rate; and (2) no rate cell may decrease to a rate lower than the old rate.115 The PRC expressed its hope that the constrained rates would "move (the PRC) and the Postal Service in the direction of more closely cost-based rates."116 For the interim period, the PRC believed that its constrained rates would more closely reflect costs than the old rates, without creating radical market disturbances.

B. Arguments Regarding the PRC's and Governors' Decisions

Petitioners Growers and Shippers League of Florida and Florida Gift Fruit Shippers Association (Fruit Shippers) argue that the PRC should have adopted a cost-based rate schedule without constraints.117 They assert that the imposition of arbitrary constraints, which preclude a decrease below former rates, disregards 39 U.S.C. § 3622(b)(3) and is therefore not "fair and equitable," in violation of 39 U.S.C. § 3622(b)(1). Moreover, they allege that using the old rates as a reference point for the constraints is fatal error, because those former rates are unsupported by the evidence and are not tied to cost incurrence. According to the Fruit Shippers, the arbitrary constraints coerce intraclass subsidization, in direct contravention of the mandates of NAGCP I.

United Parcel Service of America, Inc. (UPS) intervened to oppose the Fruit Shippers' attack on the PRC's and Governors' decisions.118 UPS argues that the imposition of constraints is essential to mitigate the potentially deleterious impact of the recommended rate structure, impacts which have not been sufficiently studied and which may result in a radical tilting of the entire parcel post rate structure to the detriment of users, competitors, and the Postal Service alike. Furthermore, UPS argues, the focusing of the constraints on former rates is not unreasonable or arbitrary when one considers the cost-basis underlying the old rates.

We agree with the Fruit Shippers that the PRC's recommended rates, without constraints, were formulated in the spirit of NAGCP I and section 3622(b)(3) and, therefore, are an improvement over the former rates. However, section 3622(b)(4)-(5)119 requires the PRC to consider the effect of rates on the public and competitors.120 It is clear that insufficient empirical data existed, at the time of the PRC's decision, to assess the impact that the new cost-based rates would have on the public and Postal Service competitors.121 NAGCP I and section 3622(b)(3) do not require myopic and exclusive adherence to principles of cost-justification, while recklessly ignoring the impact of rates on the other factors enumerated in section 3622(b). The constraints proposed by the PRC are not arbitrary. They are based on former rates which historically have been more tied to cost incurrence than other rates122 and the impacts of which are known.

We conclude that the interim imposition of constraints, until data is available to assess the impact of cost-based rates, is a reasonable compromise at this time. In affirming the PRC's and Governors' decisions, we not only hope, but fully expect, "that the current rates will move (the PRC) and the Postal Service in the direction of more closely cost-based rates."123

Affirmed.