United Parcel Service, Inc. v. United States Postal Service

184 F.3d 827, 337 U.S. App. D.C. 247, 1999 U.S. App. LEXIS 16922, 1999 WL 518841

• Court: Court of Appeals for the D.C. Circuit
• Decided: July 23, 1999
• Docket: 98-1310, 98-1320 and 98-1336
• Status: Published

Opinion

Opinion for the Court filed PER CURIAM.

PER CURIAM:

The petitioners raise five challenges to the May 11, 1998 Opinion and Recommended Decision of the United States Postal Rate Commission (Commission), as approved by the United State Postal Service Board of Governors (Governors) on June 29, 1998. For the reasons set out below, we reject each of the challenges and deny the petitions for review.

I. Background

Under the Postal Reorganization Act (Act), “the Governors are authorized to establish reasonable and equitable classes of mail and reasonable and equitable rates of postage and fees for postal services” subject to the over-all “break even” limitation that “[pjostal 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 Postai Service.” 39 U.S.C. § 3621 (1994). The United States Postal Service (Postal Service, Service or USPS) initiates a rate-making proceeding by requesting that the Commission “submit a recommended decision on changes in a rate or rates of postage or in a fee or fees for postal services.” Id. § 3622(a).

The Commission is then required to

make a recommended decision on the request for changes in rates or fees in each class of mail or type of service in accordance with the policies of this title and the following factors:

(1) the establishment and maintenance of a fair and equitable schedule;

(2) the value of the mail service actually provided each class or type of mail service'to both the sender and the recipient, including but not limited to the collection, mode of transportation, and priority of delivery;

(3) 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;

(4) the effect of rate increases upon the general public, business mail users, and enterprises in the private sector of the economy engaged in the delivery of mail matter other than letters;

(5) the available alternative means of sending and receiving letters and other mail matter at reasonable costs;

(6) the degree of preparation of mail for delivery into the postal system performed by the mailer and its effect upon reducing costs to the Postal Service;

(7) simplicity of structure for the entire schedule and simple, identifiable relationships between the rates or fees charged the various classes of mail for postal services;

(8) the educational, cultural, scientific, and informational value to the recipient of mail matter; and

(9) such other factors as the Commission deems appropriate.

Id. § 3622(b).

The Commission has construed section 3622(b) to establish a “two-tier approach to allocating the Postal Service’s total revenue requirement” under which the Commission “first must determine the costs caused by (‘attributable to’) each class of mail, § 3622(b)(3), and on that basis establish a rate floor for each class” (the “attributable” costs) and “then must ‘reasonably assign,’ see § 3622(b)(3), the remaining costs to the various classes of mail on the basis of the other factors set forth in § 3622(b)” (the “institutional” costs). National Ass’n of Greeting Card Publishers v. USPS, 462 U.S. 810, 814-15, 103 S.Ct. 2717, 77 L.Ed.2d 195 (1983). The Commission then issues its recommended decision setting rates in accordance with the combined attributable and institutional costs for each class of mail and with the statutory mandate that the Postal Service’s rates and fees “equal as nearly as practicable total estimated costs of the Postal Service,” 39 U.S.C. § 3621 (1994). Upon receiving the Commission’s decision, the Governors “may approve, allow under protest, reject, or modify that decision.” Id. § 3625(a). 1

The Commission issued its Opinion and Recommended Decision allocating attributable and institutional costs for each class of mail on May 11, 1998 (PRC Op. R97-1). See Joint Appendix (JA) vol. ii. On June 29, 1998 the Governors issued their decision accepting the Commission’s rates with “minimal exceptions.” See JA vol. i. 708. We address below the petitioners’ challenges to the Commission’s decision as accepted by the Governors.

II. DISCUSSION

As noted above, the petitioners challenge the Commission’s ratemaking decision on five grounds. We examine each ground separately.

A. The Overall Rate Increase

1.

During the three years (1995-1997) since its last rate increase in Docket No. R94-1, the Postal Service has experienced revenue surpluses after decades of deficits. The Service feared, however, that its net income would be insufficient to cover planned increases in capital spending on several management-initiated projects designed to improve the Postal Service’s performance and infrastructure. The Service initially estimated that its total revenue requirement for Fiscal Year 1998 would be $61.6 billion, including $60,564 billion in incurred costs, $605.6 million for a one-percent contingency fund, and $446.9 million to recover one-ninth of the Service’s $4,022 billion in accumulated debt. On this basis, it projected that it would need over $2.4 billion in additional revenue. The Service filed its request with the Commission in July 1997, based on data from FY 1996, using 1998 as a “test year” — a year that is to be “representative of the period for which the proposed rates are to be in effect.” PRC Op. R97-1 at 12; see also 39 C.F.R. § 3001.54(f)(2) (1998).

While the request was pending before the Commission, subsequent data indicated that the Postal Service’s original revenue estimates had been overly pessimistic. For example, although it had initially projected a surplus of only $636 million for 1997, in fact the Service received a net income of $1,264 billion. In addition, although it originally projected a $1.4 billion shortfall in revenues for FY 1998, 2 in the first seven accounting periods of FY 1998, the Service received a $1.36 billion net income and would have to lose $2.6 billion over the remainder of the year to experience the initial estimated losses. As a result of these discrepancies, the Commission took the apparently unusual step of asking the Governors to provide updated estimates for FY 1998 based on 1997 actual results; although this request -would delay the proceeding, the Commission observed that “no pressing need for new rates” existed at the time. The Governors declined the Commission’s request, rejecting an extension of the ten-month deadline and stating that they did not wish to “comment ... on the state of the evidentiary record” and that the Governors could use their discretion as to the timing of implementing rates “to provide for the best transition to new rates.”

After it became apparent that its original revenue estimates were overly pessimistic, the Postal Service reported to the Commission that it would face more costs than it had initially predicted. Specifically, it requested a new contingency figure of 1.5% instead of 1%, noting that in prior years the figure had been as high a 3.5%. In addition, the Service predicted that it would need $300 million more than it had initially requested for discretionary programs, such as automated data processing. The Commission rejected what it viewed as attempts to avoid the full impact of the Service’s bright economic situation, labeling the new 1.5% contingency number “a plug figure” used by the Service to counterbalance the decrease in the size of its contingency fund in light of 1997’s actual data. Further, the Commission dismissed as “speculative” the Postal Service’s claims that it would spend even more money than it had initially projected in FY 1998, even though it continued “to spend significantly less than its rate case forecasts” during the first half of the test year. The Commission pointed to- a Postal Service document — inadvertently included as evidence and initially disavowed by the Service as inauthentic — that identified the Service’s updating “strategy” as “providing] updated information on cost increases to offset the decreases” resulting from 1997’s actual figures. The Commission found that although the document “may not demonstrate an intent to mislead.... it indicates that the Service was looking for potential cost increases.”

The Commission therefore rejected the Service’s effort to increase its original estimate by $362 million. As to the initial spending program estimates, however, the Commission observed that, despite having “serious doubts about the Postal Service’s forecasts in the area of other programs expense, ... [the Commission] does not scrutinize the wisdom of Postal Service spending plans.” Lacking sufficient grounds to reduce this initial estimate of other programs expenses, the Commission reasoned that, “[w]hile a proportional amount of spending has not occurred in the first half of the test year, no party has presented evidence suggesting that the Postal Service will not spend funds for any particular program during the remainder of 1998.” It rejected, however, the Postal Service’s position that “it does not matter when the money is spent because it will eventually be spent,” on the ground that it was “antithetical to the test year ratemak-ing process.”

In revising the revenue request, the Commission observed that it could “not estimate the degree to which the error in forecasting 1997 results will continue into the test year, primarily because it lacks the Cost and Revenue Analysis for 1997 (CRA),” after the Governors declined to delay the proceedings to allow time for final FY 1997 data to be compiled. It did, however, adjust the Postal Service’s original request based on the 1997 figures it had, with reductions for corrections provided by the Postal Service ($67 million), a cost-of-living adjustment ($511.1 million, the largest single change), and corrections for cost reduction and other programs estimates ($101 million). 3 It retained the 1% contingency figure and, in keeping with the Postal Service’s nine-year amortization plan, it reduced by $69.9 million the amount the Service could ascribe to prior year losses, or “one ninth of the difference between actual and estimated 1997 profits.” The Commission noted that “[t]he nine year amortization period is standard, having been used in Docket Nos. R80-1, R84-1, R87-1, R90-1, and R94-1,” and that “[t]he Service still believes it is appropriate.” These figures, combined with attribution and miscellaneous adjustments adding $4 million to the total revenue requirement, led the Commission to reduce the proposed rate increase by approximately $745 million.

At the same time, in light of the breakeven requirement, 39 U.S.C. § 3621, the Commission urged the Governors to delay implementing the new rates “until additional revenues are needed to offset actual (as opposed to planned) expenditures.” In sum, despite the recent surpluses, the Commission approved an increase of $1.6 billion in overall rates on the ground that these changes “will provide added funds to enable the Postal Service to proceed with its plans to spend $5.6 billion on equipment and service enhancement programs in the 1998 fiscal year.” 4

A month later, the Governors adopted most of the Commission’s recommendations. See 39 U.S.C. § 3625. In their view, “[t]he revenue requirement was driven in large part by the need to fund specific management initiatives and programs, many of which have been approved by the Board of Governors to maintain and improve service for the public, as well as by the usual need to cover expenses and repay prior years’ losses.” At the same time, they acknowledged that in FY 1998 they expected a gain in net income. Although criticizing the Commission’s rejection of certain costs and the 1.5% contingency figure, the Governors accepted the revenue requirement portion of the Commission’s decision. They added, however, that, under their Resolution No. 95-9, 5 the Postal Service could recover for prior years’ losses

at a more rapid rate, if possible, than that based on the amount included in the revenue requirement. Continued surpluses above and beyond those anticipated will allow for the complete restoration of equity in the near future, obviating the need to include this provision in subsequent revenue requirements, and thus relieving the ratepayers of a burden they have carried for many years.

Finally, in light of comments by mail customers to the Governors and the Commission’s request of a delay, the Governors postponed implementing the rate changes until January 10, 1999. Among the factors influencing the delay were the Service’s current financial situation, as reflected in the annual report for FY 1997 and reported expectations for FY 1998, and the fact that January marked the four-year anniversary of the last general rate increase. The Governors also concluded that applying the increase in January was “consistent with the Postal Service’s goal for equity restoration through FY 1998, in accordance with Resolution No. 95-9 and the Commission’s recommendation for the recovery of prior years’ losses.”

2.

In 1970, Congress enacted the “break even” requirement, see 39 U.S.C. § 3621, as part of the Postal Reorganization Act, Pub.L. No. 91-375, § 3621, 84 Stat. 719, 760 (1970) (codified as amended at 39 U.S.C. § 101, et seq. (1994)), following years of deficits by the then-Post Office Department. See National Ass’n of Greeting Card Publishers v. USPS, 607 F.2d 392, 425 (D.C.Cir.1979) (NAGCP IIP; see also H.R.Rep. No. 91-988 at 3, 6, 13 (1970). The House Committee on Post Office and Civil Service explained that “the ‘break-even’ requirement of H.R. 17070 represents a commitment that the Postal Service no longer rely on massive annual infusions of general revenues of the Treasury at the taxpayers’ expense.” H.R.Rep. No. 91-1104 at 17 (1970); see also H.R.Rep. No. 91-988 at 13. Even so, that version of the bill did not contemplate the Postal Service becoming “self-sustain ing — [i.e.] eliminating the postal deficit”— until January, 1978. H.R.Rep. No. 91-1104 at 10. The final version of the legislation, however, replaced the 1978 target date with a requirement that “revenue from rates and fees, plus annual appropriations for public service, debt service, and revenue foregone should cover full costs.” H.R.Rep. No. 91-1363 at 87 (1970) (conference report).

Despite the restructuring of the postal system, however, the Service continued to operate budget deficits in all but nine of the 26 years from 1971, when it became an independent agency, to 1997. See PRC Op. R97-1, at i, 11; Postal Rate Commission, Opinion & Recommended Decision, Docket No. R94-1, at 11-24 to 11-26 (1994) (“PRC Op. R94-1”); see also NAGCP III, 607 F.'2d at 425, 431. As the Commission observed in the 1994 rate case, “[w]hile reorganization led to improvements in the cumulative deficit trend, it has not livéd up to the expectations of break-even operations.” PRC Op. R94-1 at 11-34. This problem began to change- in' the three years following the 1994 rate case when the Service experienced “unprecedented operating surpluses totaling $4.6 billion.” PRC Op. R97-1 at i.

The Service’s improved fortunes, however, lie at the heart of the Alliance’s challenge to the overall rate increase. The Alliance contends that the Governors’ decision flies in the face of evidence that the Service was operating at a surplus at the time the Governors approved the Commission’s recommended rate increase and therefore the Governors’ decision violated the “break-even” requirement of § 3621. The Service responds that the Commission (and therefore the Governors, who adopted the Commission’s recommendation) carefully considered its request, reducing the proposed increase when new data became available, and that “[n]o party filed factual evidence controverting the Postal Service’s revenue requirement presentation.”

The plain language of § 3621, that total estimated income and expenses be “equal as nearly as practicable,” suggests that Congress did not contemplate the breakeven provision to require a strict dollar-for-dollar match when the Service presents its budget proposal to the Commission. The legislative history also recognizes that income and costs could be “approximately in balance” and. that the Commission should recommend a decision balancing the two “as nearly as possible.” S.Rep. No. 91-912, at 14-15 (1970). The Senate Committee on Post Office and Civil Services reported that the Governors were to notify the Commission if estimated costs and estimated income were “significantly different,” at which time they were to request a change in the rate structure. Id. at 14.

The -statutory language and the legislative history recognize that ratemaking inherently involves some degree of imprecision and, as this court has previously observed, it is not an exact science. See Association of Am. Publishers v. Governors of the USPS, 485 F.2d 768, 773 (D.C.Cir.1973). The Service makes projections about its costs and revenue that may or may not come to .pass;, projections are no more than educated guesses. The use of projections for future costs and revenues necessarily will involve some imprecision when actual data becomes available. Of course, the Service must make its estimates in good faith. In addition, the Commission has a duty to evaluate the Service’s proposal independently. See Mail Order Ass’n of Am. v. USPS, 2 F.3d 408, 422 (D.C.Cir.1993). Nevertheless, the Postal Service’s request for a rate change “shapes the Commission’s power to recommend.” Dow Jones & Co. v. USPS, 110 F.3d 80, 83 (D.C.Cir.1997).

The Alliance’s challenge focuses, therefore, on the Postal Service’s estimate of its costs, noting that even the Commission expressed some doubts as to whether the Postal Service would spend all of the money in the test year that it initially projected. The court has previously rejected efforts to define the term “cost” under § 3621 too restrictively, lest we “clamp the shackles of a narrow rule onto the Postmaster General’s attempt to return the Postal Service to financial stability.” NAGCP III, 607 F.2d at 428. Although the Service is not free to define “total estimated costs” so broadly as to make the term meaningless, the court accepts the Service’s determination as to costs “unless it lies outside the range of permissible choices contemplated by the statute.” Id. at 430 (quoting Hardin v. Kentucky Utilities Co., 390 U.S. 1, 8, 88 S.Ct. 651, 19 L.Ed.2d 787 (1968)). As the Supreme Court observed in New York v. United States, 331 U.S. 284, 328, 67 S.Ct. 1207, 91 L.Ed. 1492 (1947), “[t]he appraisal of cost figures is itself a task for experts, since these costs involve many estimates and assumptions and, unlike a problem in calculus, cannot be proved right or wrong. They are, indeed, only guides to judgment.” The Alliance does not challenge the type of expenses the Service proposes to count as costs, but only the amount of those expenses.

In reviewing the record, the court must determine whether there was substantial evidence for the Commission to rely on the Service’s original cost estimates in calculating the revenue required for the Service to break even. See 5 U.S.C. § 706(2)(E) (1994); Mail Order Ass’n, 2 F.3d at 420. Such evidence need not be “overwhelming,” and the agency “must have latitude to draw permissible inferences from ... the record.” Mail Order Ass’n, 2 F.3d at 421. Here, the Commission noted, first, that the need for a revenue increase arose from the Service’s plans to increase capital spending to $2.5 billion on ambitious management-initiated programs to improve customer service. When data became available indicating that the Service’s financial performance was better than expected in 1997, the Commission adjusted the Service’s revenue requirements, while noting that the factors causing 1997’s stellar performance might not continue into the test year. It also rejected the Service’s claim that its costs would be even greater than it had initially projected. Thus, the Commission did not, as the Alliance suggests, set rates without regard to actual data. By contrast, in West Ohio Gas v. Public Utilities Comm’n of Ohio, 294 U.S. 79, 55 S.Ct. 324, 79 L.Ed. 773 (1935), the agency “shut [its] eyes” when presented with actual revenue figures for 1930 and 1931, instead relying on estimates based on 1929 data. Id. at 81, 55 5.Ct. 324. Here, the Commission adjusted its figures as new data became available and was not required to delay indefinitely the ratemaking process until all 1997 data had been compiled, particularly in light of its statutory obligation to make its recommendation within 10 months. See 39 U.S.C. § 3624(a), (c)(1).

Second, although expressing doubts about whether the Service could actually spend all the money it initially planned for during the test year, the Commission found that it had no basis to reduce this estimate, observing that its role was not to pas's judgment on the wisdom of the Service’s proposed spending. See Governors of USPS v. United States Postal Rate Comm'n 654 F.2d 108, 115 (D.C.Cir.1981). The Service offered evidence that it had plans in place to make sure its managers timely spent these funds, and it noted that a number of contracts had already been signed. The Alliance does not seek disal-lowance of any specific expenditure. Although the Alliance challenges the Service’s claim that it would in fact spend the millions of dollars during the test year, reversing early fiscal year performance, it provided no evidence that such spending would not occur. The Commission could reasonably presume, therefore, that the Service’s initial estimates and managerial efforts reflected- a good faith forecast of its spending needs. 6 See FTC v. Owens-Corning Fiberglas Corp., 626 F.2d 966, 975 (D.C.Cir.1980); cf. West Ohio Gas v. Public Utilities Comm’n of Ohio, 294 U.S. 63, 72, 55 S.Ct. 316, 79 L.Ed. 761 (1935).

Because we conclude that the Commission’s recommended decision to approve a revenue increase was “based on such relevant evidence as a reasonable mind might accept as adequate to support [the] conclusion” that the Service would spend its initial estimates on new programs during the test year, making the rate increase necessary, see Mail Order Ass’n, 2 F.3d at 420 (citations and internal quotation marks omitted), we affirm the Governors’ decision. The Governors could reasonably rely on the Commission’s conclusions. Although the Service’s failure in the early part of the fiscal year to keep pace with its initial spending projections might suggest an inability to meet its spending targets for the remainder of the test year, this is not the only conclusion reasonably to be drawn from the evidence. Reviewing courts “may not overturn an agency finding simply because evidence existed supporting an alternative finding.” Direct Marketing Ass’n v. USPS, 778 F.2d 96, 108 (2d Cir.1985) (quoting Newsweek, Inc. v. USPS, 663 F.2d 1186, 1210 (2d Cir.1981)). The Service’s witnesses testified, and the Commission accepted, that a number of contracts for such spending had been signed and that the Service was taking steps to ensure that managers would be accountable for spending the money in their budgets on the new programs. 7 These programs were consistent with the Service’s obligation to maintain and develop the postal system and to improve its service to customers. See generally 39 U.S.C. § 3621. The Alliance’s requested relief, complete “disallowance of the proposed rate increases in their entirety,” would seriously interfere with the Governors’ determination that additional funds were needed to improve service, and hence the Commission could reasonably reject its request. We also reject the notion that the Postal Service could implement rates only once its profits were exhausted: the Service can rely on the test-year estimates, so long as the Commission has substantial evidence with which to support those calculations. See Mail Order Ass’n, 2 F.3d at 420.

In light of the inherent mismatch that can occur when using a test year and estimates to project revenue requirements, the Commission necessarily faces the prospect that some of the data initially provided to it by the Service may later prove to be inaccurate. The Commission considered the data before it, rejected the Service’s late-breaking spending increase projections, and reduced the initial estimates based on what data it had. Although the Commission requested a three-month delay to allow time for the submission of updates to its FY 1997 data, once the Governors rejected this request, the Commission was within its discretion to proceed based on the evidence before it and to decline to reopen the record and thereby endanger its statutory obligation to complete the rate proceeding within ten months. See 39 U.S.C. § 3624(c)(1); see also Direct Marketing Ass’n, 778 F.2d at 107 (citing City of San Antonio v. Civil Aeronautics Bd., 374 F.2d 326, 329 (D.C.Cir.1967)). The Alliance acknowledges that the record before the Commission need not be “continually” updated to reflect the latest, most accurate data. Indeed, in enacting the Postal Reorganization Act, Congress was concerned that “protracted disputes over rates and classifications not block the adequate flow of revenues to the Postal Service.” Mail Order Ass’n, 2 F.3d at 419 (citing H.R.Rep. No. 91-1104 at 19).

The Alliance contends, however, that whatever evidence was before the Commission, the Governors in effect admitted that a rate increase in FY 1998 would violate the break-even provision when they stated in their decision adopting the Commission’s recommendation a month later that “[i]n FY 1998, the Board once again expects the Postal Service to gain a net income.” This statement, divorced from any specific data in the record to support it, is somewhat ambiguous, in that it indicates nothing about the size of the expected surplus or whether it would be more than the $19 million surplus projected by the Commission for 1998 if the new rates were implemented. Furthermore, the Governors made this observation in the context of explaining why they were delaying implementation of the rates until January 1999. Hence, the statement is hardly a precise calculation of the Service’s revenue requirements based on evidence in the record. 8

B. Nonprofit Standard A Mail

The Alliance also challenges the Postal Service’s increased rate for “nonprofit Standard A mail,” 9 arguing that the Service improperly allocated costs. The Postal Service initially proposed a rate increase of 11.3% for nonprofit Standard A mail. Before the Commission, however, the Alliance’s witness, John Haldi, came forward with evidence that part of the increase in nonprofit costs resulted from a “mismatch” between two Postal Service methods for tracking nonprofit mail. Specifically, Haldi testified that volume data collected through the Revenue-Piece-Weight (“RPW”) system might be out-of-synch with cost data measured through the In Office Cost System (“IOCS”), leading to the costs of nonprofit mail being overstated. Due to changes in nonprofit mail eligibility requirements, some mail “bearing nonprofit indicia of postage” were “entered at commercial rates or later charged back postage based on'commercial rates.” Haldi estimated that 7.85% of mail with nonprofit markings paid commercial rates and that therefore 7.85% of total attributable costs for Standard A nonprofit mail should shift to commercial mail.

The Postal Service’s rebuttal study challenged Haldi’s findings. The Service conceded that the mismatch documented by Haldi was possible, but it found that only 0.061% of commercial mail had nonprofit indicia, with a net effect that only $400,000, or 0.18%, of nonprofit costs should be assigned to commercial costs. Given the small figure, the Postal Service argued that no adjustment was necessary.

The Commission found that both surveys “are significantly flawed and may not be relied upon for quantitative assessment.” For example, one third of the responses in the Haldi study came from one organization, the American Association of Museums, while the Postal Service’s study relied extensively upon the memories of its employees. A problem clearly existed — “that some nonprofit mail may be correctly reported in the RPW system as commercial mail, but recorded as nonprofit in the IOCS system.” But the Commission concluded that quantifying this problem presented a “greater challenge,” and in the end determined “[ajfter examination of the record evidence, including nonprofit mail volume, the Commission estimates that one percent of total nonprofit attributable costs should have been associated with Standard A commercial mail.” The resulting rate recommended for Standard A Nonprofit Mail was an increase of 9.6%, which the Governors adopted.

On appeal, the Alliance contends that the rate increase for Nonprofit Standard A mail violates the requirements of § 3626(a)(3), which provides that rates for this category of mail include the “estimated costs attributable” to it plus an additional markup, which for FY 1998 equaled % of the commercial subclass markup. 10 See 39 U.S.C. § 3626(a)(3)(A)-(B) ■ (1994). The statute defines “costs attributable” as “the direct and indirect postal costs attributable to such class of mail or kind of mailer (excluding any other costs of the Postal Service).” Id. § 3626(a)(2)(A). This language parallels § 3622(b)(3), which requires “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.” In the Alliance’s view, because the cost data used to create the nonprofit rate was “corrupted,” and the Postal Service had failed to provide sufficient evidence on the extent of cost-shifting, the Commission had to reject the rate increase or toll the deadline until the Postal Service provided better data. 11 Selecting the one-percent figure as a compromise number between the competing parties’ estimates did not constitute, the Alliance contends, reasoned decision-making.

In examining the Commission’s decision, context is significant. The Commission viewed the Service’s proposed request to be “the most technically complex” rate case ever presented to it. See PRC Op. R.97-1 at iii. Yet the statute required the Commission to make its recommendations on the Service’s request “no later than 10 months after receiving” it. 39 U.S.C. § 3624(a), (c)(1). To reach that point, the Commission first had to conduct hearings to allow the Service, mail users, and a Commission officer appointed to represent the general public the opportunity to comment on the Service’s request. Id. § 3624(a). Congress required that the Commission reach its decision promptly in recognition that “the Postal Service is a labor-intensive organization,” S.Rep. No. 91-912, at 16, which needed sufficient income to operate efficiently, see H. Rep. 1104 at 19, and therefore Congress implicitly anticipated that the Commission would have to make its recommendations based on the data that might suffer from analytical flaws or, with time, prove inaccurate.

Consequently, although the Commission has an obligation to explain its reasoning and to support its position with substantial evidence, see Mail Order Ass’n, 2 F.3d at 420; Direct Marketing Ass’n, 778 F.2d at 100, it will not always be possible for it to provide an extended explanation of every issue addressed in its recommendation, particularly an issue raised relatively late in the proceedings. So viewed, the Commission’s decision is sufficient if the court can discern the path that the Commission followed in reaching its one-percent figure. See Motor Vehicle Mfrs. Ass’n of the United States v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43, 103 S.Ct. 2856, 77 L.Ed.2d 443 (1983) (quoting Bowman Transp., Inc. v. Arkansas-Best Freight Sys., Inc., 419 U.S. 281, 286, 95 S.Ct. 438, 42 L.Ed.2d 447 (1974)); see also Greater Boston Television Corp. v. FCC, 444 F.2d 841, 852 (D.C.Cir.1970).

The Supreme Court has cautioned, moreover, that “[allocation of costs is not a matter for the slide-rule. It involves judgment on a myriad of facts. It has no claim to an exact science.” National Ass’n of Greeting Card Publishers, 462 U.S. at 825, 103 S.Ct. 2717 (quoting Colorado Interstate Co. v. FPC, 324 U.S. 581, 589, 65 S.Ct. 829, 89 L.Ed. 1206 (1945)). Discussing postal rate cost allocation in the context of § 3622(b)(3), the Court observed that “Congress did not dictate a specific method for identifying causal relationships between costs and classes of mail” and that the Commission’s interpretation of the statute, including the method to choose to comply with § 3622(b)(3), was due deference. Id. at 826, 103 S.Ct. 2717. The Court upheld the Commission’s construction of the Postal Reorganization Act as establishing a “two-tier rate-setting structure” to allocate costs, first identifying “all costs that in the judgment of the Rate Commission are the consequence of providing a particular class of service” and second “assigning] remaining costs reasonably on the basis of the other eight factors set forth by § 3622(b).” Id. at 833-34, 103 S.Ct. 2717. The Court concluded that “[t]he statute requires attribution of any cost for which the source can be identified, but leaves it to the Commissioners, in the first instance, to decide which methods provide reasonable assurance that costs are the result of providing one class of service.” Id. at 833, 103 S.Ct. 2717.

When an agency does not “entirely disregard two experts, but [finds] each somewhat in error,” the court has permitted the agency to take “as its own solution a point somewhere between the two expert figures. When neither of two suggested adjustments applied to inaccurate data is completely satisfactory a rate-making body may fashion its own adjustments within reasonable limits.” Association of Am. Publishers, 485 F.2d at 773. Accord Direct Marketing Ass’n, 778 F.2d at 102, 110-11. In Association of American Publishers, the Commission decided to split the difference between two competing figures, observing that “[s]ince Solomon’s day, to split the difference or to come close thereto has been thought wise, if only because it makes parties more likely to disclose to tribunals the truth.” 485 F.2d at 773. Here, the Commission elected a figure at the low-end of a narrow range, between the Service’s estimate of 0.18% and the Alliance’s estimate of 7.85%.

The dilemma for the Commission was to quantify the extent, of the mismatch between the two Service methods for tracking nonprofit mail. Although the Alliance and the Service’s surveys acknowledged some mismatch existed, survey-design flaws made it impossible for the Commission to know the exact amount of mismatching that had occurred. As in Association of American Publishers, “[t]he only available figures were inaccurate, but were susceptible of rough adjustment,” id., because the Commission could reasonably assume that the mismatch was no worse than the Alliance’s estimate of 7.85%, given its incentive to find the most favorable sample possible, 12 nor better than the Service’s estimate of 0.18%, given its interests in minimizing the problem so that its cost allocation estimates would remain undisturbed. 13 Viewing the surveys, though flawed, to represent the outer limits of the true nature of the mismatch, with 7.85% as the ceiling and 0.18% as the floor, the Commission, given time constraints, could reasonably select a figure somewhere between that range and choose a number at the low end of the range. First, the results of the Alliance’s survey were questionable because the Alliance failed to show that they reflected the experiences of non-profit mailers as a whole, particularly in light of the fact that so many of the responses came from one group. 14 Second, the Commission expressed concern that the survey relied on a volume growth rate figure that it had not attempted to quantify. Third, although the Service’s survey suffered from flaws as well, in that it failed to consider a source of volume (mail voluntarily entered at commercial rates) that could not readily be quantified, and it relied in part on the memory of Postal employees, the Commission could still reasonably assume the mismatch problem was relatively small in view of the lack of reliable evidence presented by the Alliance that the mismatch was significant enough to warrant a major adjustment. Cf. Mail Order Ass’n, 2 F.3d at 438. Had the Commission adopted the Alliance’s figure without having reliable evidence to back it up, the Commission would be creating a solution to a problem that may be relatively small, thereby unnecessarily penalizing commercial subclasses and opening itself up to accusations of arbitrary decision-making by subclasses negatively affected by its “solution.” Although the Commission must address a petitioner’s allegations of error in the Service’s calculations, the Commission need not assume that simply because an objection is raised that a problem exists, when the petitioner fails to provide reliable data to support its position. Here the Commission’s choice of the one-percent figure avoided penalizing the commercial subclasses, in that the small shift in costs had almost no effect on their unit attributable cost, yet it also took into account the Alliance’s concern that nonprofit mailers were being penalized with “unjustifiably high rates.”

Admittedly, the choice of the one-percent figure (as opposed to some other point between 0.18% and 7.85%) is somewhat mysterious, but the general path is clear enough. The record indicated that a mismatch problem existed, but that it might not be very large. Although with more time, the Commission might have been able to get better information, see Association of Am. Publishers, 485 F.2d at 773, a “judgmental approach” selecting a figure between these two estimates, though favoring the low end of the spectrum, was within the Commission’s authority.

Direct Marketing Association, 778 F.2d 96, presents an analogous situation. The Second Circuit upheld the Commission’s recommendation of a four-cent discount for certain presorted first class mail. Id. at 109. Finding fault with both the Service’s recommended three-cent discount and another witness’ 4.5 cent figure, the Commission treated these estimates as the outer range of possible discounts, and selected a figure in between the two. Id. at 110. The Commission found the Service’s estimate approach “novel” and “very conservative,” while the other witness’ approach took into account costs unrelated to presort savings. Id. In developing its own approach, based on the Service’s Revenue and Costs Analysis Report, the Commission noted that “[t]he record does not contain sufficient information to develop a more precise estimate.” Id. at 110-11 (quoting PRC Op. R84-1 at 368-69). Thus, as in the instant case, the Commission’s approach was “diseernable from the evidentiary record upon which the recommendation [was] based.” Id. at 111.

Schurz Communications, Inc. v. FCC, 982 F.2d 1043 (7th Cir.1992), is distinguishable. There, the Seventh Circuit criticized the Federal Communications Commission for “throwing up [its] hands and splitting the difference,” rather than assessing who had the stronger case, when it enacted new rules governing television syndication rights. Id. at 1050. The court concluded that the agency had overlooked key evidence and ignored arguments that it previously had accepted, id., while the Commission here considered the evidence before it. San Antonio, Texas v. United States, 631 F.2d 831 (D.C.Cir.1980), clarified by 655 F.2d 1341 (D.C.Cir.1981), rev’d on other grounds sub nom. Burlington Northern, Inc. v. United States, 459 U.S. 131, 103 S.Ct. 514, 74 L.Ed.2d 311 (1982), is also distinguishable; the Interstate Commerce Commission offered no evidence or rationale for its “seven percent solution,” and in fact its reasoning could support any percentage, whether one or 99 percent. Id. at 852. Here, the Commission chose a percentage between two competing numbers, representing a relatively small range, and although the Commission could have explained in greater detail why it chose the low-end rather than the high-end of this range, its path was reasonably clear.

C. Alaskan Parcel Post Air Costs

United Parcel Service (UPS) first challenges the amount of the Commission’s attributable costs for Parcel Post mail 15 on the ground that the Commission improperly excluded from them a substantial portion of air transportation costs attributable to delivering Parcel Post mail to the remote Alaskan “bush country.” Because the Alaskan bush country is accessible only by air, all mail delivered there, inelud-ing nonpreferential Parcel Post mail, which is usually carried by ground transport, must be delivered by air, inflating considerably the costs of delivering Parcel Post mail to the area. The Commission elected to attribute only a portion of the air delivery costs to the Parcel Post subclass, 16 however, concluding the remainder was attributable to the Act’s “universal service obligation,” which the Commission found to be the primary cause of the air costs. See 39 U.S.C. § 101(a) (1994) (providing Postal Service “shall provide prompt, reliable, and efficient services to patrons in all areas and shall render postal services to all communities”) (emphasis added). Because the statutory term “attributable” is ambiguous, we defer to the Commission’s reasonable interpretation of it and uphold its consequent decision to attribute only a portion of Alaskan air costs to the Parcel Post subclass.

As noted above, section 3622(b)(3) requires “that each class of mail or type of mail service bear the direct and indirect postal costs attributable to that class or type.” Id. § 3622(b)(3). Thus, “all-costs that in the judgment of the Rate Commission are the consequence of providing a particular class of service must be borne by that class.” National Ass’n of Greeting Card Publishers, 462 U.S. at 833, 103 S.Ct. 2717. In this ratemaking, as in past rate-makings, the Commission generally attributed costs under the “volume variability” methodology, which classifies a cost as “volume variable” and therefore attributable to a particular class if the cost rises as the volume of the particular class of mail rises. See Mail Order Ass’n, 2 F.3d at 427 (“Traditionally, access costs have been attributed to mail subclasses based on a ‘volume variability’ formula that related ‘access costs’ to a particular subclass’s mail volume. Generally, the greater the volume of the subclass’s mail, the greater the attributed access costs.”); see also Newsweek, Inc. v. USPS, 663 F.2d 1186, 1207-08 (2d Cir.1981), aff'd and remanded, 462 U.S. 810, 103 S.Ct. 2717, 77 L.Ed.2d 195 (1983). Here, however, the Commission elected to deviate from strict volume variable causation because of the unusual and constraining geographical circumstances of Alaskan Parcel Post service.

In its decision the Commission applied a “premium costing approach” under which the attributable costs of delivering Alaskan Parcel Post were “calculated based on the nationwide average costs of [] highway transportation,” while “[t]he remaining portion, approximately $70 million for the last year, is transferred to the institutional cost pool and recovered through the markup procedure pursuant to the Act.” PRC Op. R97-1 at 220. The Commission’s decision explained this attribution only briefly:

The costs of serving areas without road access, the so-called Bush Country of Alaska, are considerably higher than the costs of providing service to other areas in the United States. Since the Postal Service’s universal service obligation extends to citizens of all regions of the United States, it would not be appropriate to recover all these costs from the nonpreferential classes carried by intra-Alaska-Air.

Id. The Commission explained its reasoning more clearly and extensively in the 1990 postal ratemaking decision in which, as the Commission specifically noted here, for the first time “a portion of the costs of intra-Alaskan transportation costs ... ha[d] been considered institutional, although they are recognized as being volume variable in nature.” Id.; see Opinion and Recommended Decision of the United States Postal Commission in Docket No. R90-1 (January 4, 1991), III — 194 to -237 (JA vol. i 814).

In the 1990 ratemaking the Commission determined:

The record supports a finding that nonp-riority Alaska air costs are attributable only to the extent that they substitute for the surface costs that would be incurred if that transportation service were available. The remaining costs, which we refer to as the “universal service obligation premium,” are institutional. These costs are caused by the Postal Service’s statutory obligation to serve the entire nation.

Id. at III-195. The Commission defended its use of the “premium costing approach” as reasonable under the circumstances:

Our approach is the one supported by the record before us. The evidence shows that the costs are being overattri-buted, and it is our statutory duty to be as accurate as possible in attributing costs. Over-attribution can be just as much an error as the under-attribution proscribed by section 3622(b)(3). Our approach is a better reflection of reality. And, as this record shows, the potential to support the rate design and rate schedules of two subclasses, parcel post and Priority Mail, requires that the costing method be improved.

Id. at III-212. The Commission also explained why it considered the Alaskan air costs caused by and therefore attributable to the Postal Service’s universal mail obligation:

In considering these costs and the mail which is being carried on both mainline and bush transportation, we look for the true causal connection. Regardless of how these costs might actually vary with volume, we find that the premium is caused by the statutory obligation to provide universal service rather than the mail volumes. It is true that if none of this mail existed, the costs would not be incurred. It is difficult to believe, however, that this nonpreferential mail would be incurring these very high air costs in the absence of a statutory mandate to serve the entire nation. The Postal Service interprets its duty as one to offer its basic services to every part of the country, and not to deny the lower priced parcel post service to people who live in remote areas which have only expensive transportation available.

Id. at III-213 to -14 (footnote & record citation omitted). The Commission’s reasoning adequately supports its bifurcated attribution of Alaskan air costs.

Nevertheless, UPS contends the Commission’s use of the premium cost approach violates the Act because it either (1) fails to allocate to Parcel Post the Alaskan air costs that the Commission has found attributable to that subclass or (2) fails in the first instance to find that such costs are attributable to Parcel Post even though the Commission acknowledged the costs “are recognized as being volume variable in nature.” PRC Op. R97-1 at 220. UPS’s first objection is easily answered: the Commission specifically found that the “premium” air delivery costs are attributable not to Parcel Post service but to the statutory universal service obligation. As for the second, although the Commission has generally used volume variability to attribute costs, the Act itself does not require any specific cost method or define the term “attributable,” which, as the Commission’s analysis demonstrates, can have various meanings that support various attribution methods. See National Ass’n of Greeting Card Publishers, 462 U.S. at 825-26, 103 S.Ct. 2717. (“We agree with the Rate Commission’s consistent position that Congress did not dictate a specific method for identifying causal relationships between costs and classes of mail, but that the Act ‘envisions consideration of all appropriate costing approaches.’ ”) (quoting Commission’s decision). Instead, the Act “leaves it to the Commissioners, in the first instance, to decide which methods provide reasonable assurance that costs are the result of providing one class of service.” Id. at 833, 103 S.Ct. 2717; see also id. at 827, 103 S.Ct. 2717 (“On its face, there is no reason to suppose that § 3622(b)(3) denies to the expert rateset-ting agency, exercising its reasonable judgment, the authority to decide which methods sufficiently identify the requisite causal connection between particular services and particular costs.”). Because “the statute is silent or ambiguous” on which cost method to use, “the question for the court is whether the agency’s answer is based on a permissible construction of the statute.” Chevron USA, Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 843, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984); see also National Ass’n of Greeting Card Publishers, 462 U.S. at 814-15, 103 S.Ct. 2717. Based on its analysis in the 1990 ratemaking decision, we conclude that the Commission’s choice of the premium methodology reflects a reasonable construction of the Act and must therefore be upheld.

D. Priority Mail Institutional Costs

UPS next challenges the Commission’s allocation of Priority Mail institutional costs. 17 The Commission assigns institutional costs by establishing a separate markup for each class of mail and then applying the markup to the class’s attributable costs. In this ratemaking the Commission recommended an institutional markup for Priority Mail of 66.1%. UPS contends this markup is artificially low and shifts to First Class mail institutional costs reasonably assignable to Priority Mail in violation of section 3622(b)(3). We conclude the institutional costs for Priority Mail are, as the statute requires, “reasonably assignable” to the subclass and we therefore uphold them.

As we noted above, once the Commission has established attributable costs under its two-tier cost methodology, it must then allocate institutional costs by “ ‘reasonably assigning]’ the remaining costs to the various classes of mail on the basis of the other factors set forth in § 3622(b).” National Ass’n of Greeting Card Publishers, 462 U.S. at 816, 103 S.Ct. 2717 (internal citation omitted). In assigning Priority Mail institutional costs, the Commission relied heavily on the second statutory factor included in section 3622(b): “the value of the mail service actually provided each class or type of mail service to both the sender and the recipient, including but not limited to the collection, mode of transportation, and priority of delivery.” 39 U.S.C. § 3622(b)(2). The Commission cited testimony that Priority Mail has a high “intrinsic value of service,” which might justify a higher share of institutional costs, but also noted that it has a “high own-price elasticity,” meaning that rate increases might drive away customers despite the high intrinsic value, therefore calling for a lower markup. PRC Op. R97-1 at 369. The Commission further pointed to testimony questioning the value of Priority Mail’s service because (1) it often falls short of one- and two-day delivery benchmarks, (2) its service will deteriorate further with implementation of a new processing network service, (3) its market share has been decreasing, and (4) it lacks enhancements available with private priority delivery services, such as automatic insurance coverage, billing, payment and rate options and guaranteed delivery. Id. at 360-62. Based on its perception of the value of Priority Mail service and of the deleterious effect a price increase might have, the Commission concluded that a “reduction in the proportional contribution by Priority Mail is not unreasonable,” especially since even the lower markup the Commission recommended led to a rate increase for Priority Mail that exceeded the system-wide average.' Id. at 362. UPS challenges the 66.1% Priority Mail markup primarily on two grounds. We find neither one persuasive.

First, UPS contends that in the 1997 ratemaking the Commission impermissibly “changed course” without explanation because for the first time it assigned a lower markup to Priority Mail than to regular First Class mail. This argument misapprehends the Commission’s institutional cost assignment process. The Commission has not, in this ratemaking or previous ones, assigned the Priority Mail markup based on its relationship to the First Class markup, as is manifest from the widely varying gaps between the two markups in each of the ten ratemakings conducted under the Act. See JA vol. i 706. Instead, the Commission assigned the markup here, as before, based on consideration of the mandatory statutory factors. See PRC Op. R97-1 at 371. 18 It was these factors, and the second one in particular, that led the Commission to assign lower institutional costs to Priority Mail.

UPS also argues that consideration of the fourth and fifth statutory factors (“the effect of rate increases upon the general public, business mail users, and enterprises in the private sector of the economy engaged in the delivery of mail matter other than letters” and “the available alternative means of sending and receiving letters and other mail matter at reasonable costs,” 39 U.S.C. § 3622(b)(4), (5)) requires that the Commission assign lower universal costs to the monopoly regular First Class mail than to Priority Mail because those factors were intended to protect the interests of First Class customers, who have no private alternative, and of Priority Mail competitors, each of which will be harmed by higher First Class and lower Priority Mail rates. We disagree. While the Commission must “ ‘take into account all the relevant factors and no others,’ ” Mail Order Ass’n, 2 F.3d at 426 (quoting Association of Am. Publishers, 485 F.2d at 775), it need not give each factor equal weight. “ ‘[U]nder familiar jurisdictional principles,’ ” we “ ‘may not, and under human limitations generally could not, reassess the weights given by a rate-making agency to different factors, absent a legislative direction as to precisely what gravity each factor bears.’ ” Id. (quoting Association of Am. Publishers, 485 F.2d at 774-75). Given that the Act provides no such direction, we cannot fault the Commission’s determination that the second factor is the decisive one here. In any event, UPS’s reading of the statutory provisions it invokes is unduly narrow. By its terms, § 3622(b)(4) allows the Commission to consider lowering rates in order to protect “the general public [and] business mail users,” as well as raising them in the interests of “enterprises in the private sector ... engaged in the delivery of mail matter.” As to § 3622(b)(5), the Commission has consistently, and reasonably, held that it authorizes a reduction in rates to maintain the position of the Postal Service as a competitor in the mail delivery industry.

E. “Local Only” Mail

Finally, Niagara Telephone Company (Niagara) challenges the Commission’s rejection of Niagara’s proposed separate rate for mail deposited in “local only” mail boxes located at individual post offices. Niagara maintains that, because this mail is sorted by the sender, it costs the Postal Service less to deliver than other First Class mail and that this savings should be passed on to customers. The Commission rejected Niagara’s proposal because “the record remains undeveloped on matters critical to a determination on the merits, such as its impact on net revenues.” PRC Op. R97-1 at 345. We see no defect in the Commission’s determination.

In Mail Order Ass’n, we upheld the Commission’s decision not to establish a separate classification and rate, proposed by Niagara, for “non-transported” mail that never leaves the post office where deposited but is placed directly into on-site post office boxes (specifically, utility bills Niagara sent to customers). The court reasoned:

Even though 39 U.S.C. § 3622(b)(3) requires each “class of mail or type of mail service” to recover its attributable costs, that section does not require creation of a separate class of mail for every single cost characteristic. As we noted before, § 3622(b)(7) allows the Commission to consider the simplicity of the rate structure, and a separate rate for every group of mailers with special cost savings, no matter how small the group, would produce a hopelessly complicated rate schedule. This does not mean the Commission may always reject proposed cost-based classifications in order to avoid complexity in the rate schedule; in some cases the facts might be compelling enough to require a new classification. Here, however, given the complete absence of evidence establishing the existence of a substantial category of mail systemically involving lower costs, the Commission’s rejection of Niagara’s proposal was not arbitrary or a violation of § 3622(b)(3).

Mail Order Ass’n, 2 F.3d at 426. In this case too there is no record evidence to compel creation of the mail subclass Niagara proposes and we therefore conclude the Commission reasonably declined to do so.

For the preceding reasons, the petitions for review are

Denied.

1 . For a more detailed exegesis of the statutory scheme, see Mail Order Ass’n of Am. v. USPS, 2 F.3d 408, 413-16 (D.C.Cir.1993).

2 . The Commission later identified this figure as $1.2 billion, without explaining the discrepancy.

3 .The Governors criticized this last decrease of “assumed supervisor cost savings” as “based on one party's unsupported speculations that such costs were overlooked.” Despite this complaint, the Governors elected not to challenge this reduction; nor do the petitioners raise it as an issue.

4 . This spending increase represented the first portion of a five-year plan to invest $17 billion in the Postal Service’s operations.

5 . Resolution 95-9, a policy statement by the Governors, provides that the "Postal Service will plan for cumulative net income ... to equal or exceed the cumulative prior years' loss recovery target."

6 . Although the Alliance makes much of the fact that the Postal Service earned a $550 million net income during FY 1998 and $611 million during the first four months of FY 1999, these figures were not available to the Commission and the Governors when they made their respective decisions. See 39 U.S.C. § 3628 (1994); see also Commercial Drapery Contractors, Inc. v. United States, 133 F.3d 1, 7 (D-C.Cir.1998); Direct Marketing Assn v. USPS, 778 F.2d 96, 109 (2d Cir.1985).

7 . The Alliance is therefore mistaken when it contends that the Service is engaging in post hoc rationalization by' suggesting that the' delays in spending at the beginning of the fiscal year were only temporary. See SEC v. Chen- ery Corp., 332 U.S. 194, 196, 67 S.Ct. 1575, 91 L.Ed. 1995 (1947).

8 . The Commission urged the Governors to delay implementing the rates until the new revenues were actually needed “to offset actual (as opposed to planned) expenditures.” The Commission did not, however, condition its recommendation of the rates upon the Board’s acceptance of a delay. Had the Commission required the Service to delay rates until January, it would be making rates on the basis of something other than the test year, in that it would be acknowledging that the Service could not spend all the money it proposed during the test year. Although the decision when to start new rates is within the Governors' discretion, see Mail Order Ass'n, 2 F.3d at 419-20; see also 39 U.S.C. § 3625(f), we do not reach the separate inquiry of whether the Commission could approve an increase contingent upon the Governors’ delaying its implementation until actual spending needs arise.

9 . .The Alliance describes this subclass as "the primary medium for nonprofit organizations to raise funds and disseminate information.” Commercial Standard A mail, in contrast, is "the primary subclass for commercial bulk ad vertising. mail.''’

10 . In 1993 Congress passed the Revenue Forgone Reform Act, as part of the larger Treasury, Postal Service, and General Government Appropriations Act, 1994, Pub.L. No. 103-123, 107 Stat. 1226, §§ 701-708 (1993). This statute amended 39 U.S.C. § 3626(a), phasing in a series of mark-ups based on year. See 107 Stat. at 1268. For fiscal years after 1998, the rate for nonprofit Standard A mail was to "reflect one-half the markup of the comparable commercial subclass.” PRC Op. 97-1, at 458; see also 39 U.S.C. § 3626(a)(3)(B)(ii)(VI).

11 . For support of this proposed delay, the Alliance cites 39 U.S.C. § 3624(c)(2), which permits the Commission to extend the 10-month deadline by one day for each day that the Postal Service has unreasonably delayed a response to a lawful order by the Commission. The Postal Service responds that at no time did it fail to comply with a specific Commission order. The Alliance contends, however, that the Postal Service's failure "to provide adequate documentation of the costs attributed to nonprofit Standard (A) mail” violated the Service’s obligation to provide complete documentation on cost attribution for individual classes of mail. Reply Br. at 15 n.7 (citing 39 C.F.R. § 3001.54(b)). Despite the Alliance's assertions, the statute provides that the delay may be imposed when "the Commission determines that the Postal Service has unreasonably delayed consideration.” See 39 U.S.C. § 3624(c)(2) (emphasis added). Although the Commission was critical of the Service's failure to "expend significant efforts to evaluate the matter until after [being] directed ... to do so,” the Commission made no such determination on this issue.

12 . The Alliance contended during oral argument that there might have been a more biased sample of interested parties out there but that it did not "have access to the whole universe of nonprofit mailers.” Even so, the survey itself attempted to elicit a favorable response from those participants that the Alliance was able to reach. Specifically, one version of the survey used by the Alliance alerted participants that "the ongoing postal rate case litigation before the Postal Rate Commission threatens to hit nonprofit Standard A mailers with substantial increases ... as high as 15-18%” and urged them to respond "[i]n order to best protect your interests and the interests of your colleagues.” Such language could potentially-discourage at least some respondents from replying if their own figures would not aid the nonprofit mailers’ cause.

13 . Indeed, the Service argued before the Commission that no adjustment to nonprofit costs or rates was warranted.

14 . One-third of the Alliance's survey responses came from members of a single association, the American Association of Museums, even though this group did not represent one-third of all non-profit mailers. Two of the fundamental goals of designing a sample is to "choos[e] a sample that reflects relevant characteristics of the population, and [to] ac-hiev[e] a certain level of precision for the statistical results. A major influence on the precision of estimates is the size of the sample.” David W. Barnes & John M. Conley, Statistical Evidence in Litigation 253 (1986). The validity of a survey’s results is undermined if the sample is not representative of the population it purports to represent or is not selected in a sufficiently random manner. See Frazier v. Consolidated Rail Corp., 851 F.2d 1447, 1452 (D.C.Cir.1988).

15 . "Parcel Post includes mailable matter weighing 16 ounces or more, but not exceeding 70 pounds in weight or 108 inches in combined length and girth. In general, Parcel Post is used for matter not eligible for mailing in any other Standard Mail subclass, and consists primarily of merchandise.” PRC Op. R97-1 at 476.

16 . Although the Act directs the Commission to recommend “changes in rates or fees in each class of mail” 39 U.S.C. § 3622(a) (emphasis added), the Commission has carved out discrete subclasses of mail classes that it deems warrant separate consideration.

17 . "Priority mail is a service available for all mailable items up to 70 pounds in weight that offers somewhat more expedited delivery than First-Class Mail. On this basis, it competes in the two-day document and package market. Priority Mail also constitutes the extension of First-Class-Mail services to pieces weighing 11 ounces or more. Consequently, Priority Mail consists both of monopoly letter mail and items that could be delivered by a competing carrier-” PRC Op. R97-1 at 352.

18 . The Commission did remark that, because Priority Mail is "an extension of First Class Letters and Sealed Parcels,” assigning it "a markup similar to First-Class letters is justified,” PRC Op. R97-1 at 362, but only after it assigned the markup-apparently as back-up justification and in answer to UPS's claim that the relationship between the two markups should stay the same as in the previous ratemaking. See PRC Op. R97-1 at 359-60.