ORDER NO. 4228

UNITED STATES OF AMERICA POSTAL REGULATORY COMMISSION WASHINGTON, DC 20268-0001

Before Commissioners: Robert G. Taub, Chairman; Mark Acton, Vice Chairman; Tony Hammond; and Nanci E. Langley

Periodic Reporting Docket No. RM2017-10 (Proposal Six)

ORDER ON ANALYTICAL PRINCIPLES USED IN PERIODIC REPORTING (PROPOSAL SIX)

(Issued November 20, 2017)

I. INTRODUCTION

On July 28, 2017, the Postal Service filed a petition pursuant to 39 C.F.R. § 3050.11, requesting the Commission to initiate a rulemaking proceeding to consider a proposal (Proposal Six) to change an analytical method approved for use in periodic reporting.1 Proposal Six would modify the Parcel Select/Parcel Return Service (PRS) mail processing and transportation cost models and remove those portions of the cost

1 Petition of the United States Postal Service for the Initiation of a Proceeding to Consider Proposed Changes in Analytical Principles (Proposal Six), July 28, 2017, at 1 (Petition). A discussion of Proposal Six is attached to the Petition; Modification of the Parcel Select/Parcel Return Service Mail Processing and Transportation Cost Models (Proposal Six). Accompanying the Petition is a non-public annex. See Notice of Filing of USPS-RM2017-10/NP1 and Application for Nonpublic Treatment, July 28, 2017 (Library Reference USPS-RM2017-10/NP1). Docket No. RM2017-10 - 2 - studies related to discontinued Parcel Select/PRS price categories.2 In addition, the Postal Service proposes to correct certain minor errors that it discovered during the process of modifying the mail processing and transportation cost models. Id. The revised cost models also incorporate new data into their analyses and replace cost methodologies last presented in the FY 2016 annual compliance review proceeding.3 For the reasons discussed below, the Commission approves Proposal Six.

II. PROCEDURAL HISTORY

On August 1, 2017, the Commission issued a notice initiating this proceeding, providing for the submission of comments, and appointing a Public Representative.4 On August 15, 2017, the Commission granted United Parcel Service (UPS) and Amazon Fulfillment Services, Inc. (AFSI) access to Library Reference USPS-RM2017-10/NP1 and the FY 2016 ACR Library References.5 On August 18, 2017, the Public Representative filed a motion for issuance of an information request pursuant to 39 C.F.R. § 3001.21(a) and 39 C.F.R. § 3007.3(c).6 Chairman’s Information Request No. 1 was issued on August 23, 2017.7 The Postal Service responded to CHIR No. 1 on August 30, 2017.8 On September 6, 2017, the Public Representative filed a second motion for issuance of an information request.9 Chairman’s Information Request No. 2 was issued on September 7, 2017.10 The Postal Service responded to CHIR No. 2 on September 13, 2017.11

2 Proposal Six at 2, 5. The price categories for originating network distribution center (ONDC) and network distribution center (NDC) presort Parcel Select were discontinued and removed from the Price List during January, 2016. Id. at 1. During that same month, the price category for return network distribution center (RNDC) Parcel Return Service was also removed from the Price List. Id.

3 Id. at 1. See Docket No. ACR2016, Library Reference USPS-FY16-NP15, December 29, 2016 and Library Reference USPS-FY16-NP16, December 29, 2016 (FY 2016 ACR Library References).

4 Order No. 4023, Notice of Proposed Rulemaking on Analytical Principles Used in Periodic Reporting (Proposal Six), August 1, 2017.

5 On August 14, 2017, the Postal Service filed a notice of comments it had received from Canada Post Corporation (CPC) opposing, in part, UPS’s request for access to certain non-public information. United States Postal Service Notice of Filing Third Party Comments, August 14, 2017. The CPC comments do not address any substantive issue in this proceeding. See Order No. 4042, Order Granting Request for Access in Part, August 15, 2017. See also Order No. 4043, Order Granting Request for Access, August 15, 2017.

6 Public Representative Motion for Issuance of Information Request, August 18, 2017 (First PR Information Request).

7 Chairman’s Information Request No. 1, August 23, 2017 (CHIR No. 1). Docket No. RM2017-10 - 3 -

The Public Representative and UPS filed comments on September 15, 2017.12 The Postal Service filed reply comments on September 22, 2017.13 On September 26, 2017, the Public Representative responded to USPS Reply Comments.14 Finally, on October 2, 2017, the Postal Service responded to Chairman’s Information Request No. 3.15

III. PROPOSAL SIX

Mail Processing Cost Model

Summary. The proposed Parcel Select/PRS mail processing cost model removes worksheets and portions of worksheets specific to discontinued price

8 Responses of the United States Postal Service to Questions 1-13 of Chairman’s Information Request No. 1, August 30, 2017 (Responses to CHIR No. 1). Responses to CHIR No. 1 was accompanied by a non-public annex. See Notice of Filing of USPS-RM2017-10/NP2 and Application for Nonpublic Treatment, August 30, 2017 (Library Reference USPS-RM2017-10/NP2). On September 5, 2017, the Commission granted AFSI and UPS access to Library Reference USPS-RM2017-10/NP2. Order No. 4089, Order Granting Requests for Access, September 5, 2017.

9 Public Representative Second Motion for Issuance of Information Request, September 6, 2017 (Second PR Information Request).

10 Chairman’s Information Request No. 2, September 7, 2017 (CHIR No. 2).

11 Responses of the United States Postal Service to Questions 1-12 of Chairman’s Information Request No. 2, September 13, 2017 (Responses to CHIR No. 2). Responses to CHIR No. 2 was accompanied by a non-public annex. See Notice of Filing of USPS-RM2017-10/NP3 and Application for Nonpublic Treatment, September 13, 2017 (Library Reference USPS-RM2017-10/NP3). On September 18, 2017, the Commission granted AFSI access to Library Reference USPS-RM2017-10/NP3. Order No. 4097, Order Granting Request for Access, September 18, 2017.

12 Public Representative Comments, September 15, 2017 (PR Comments); Comments of United Parcel Service on Notice of Proposed Rulemaking on Analytical Principles Used in Periodic Reporting (Proposal Six), September 15, 2017 (UPS Comments).

13 Reply Comments of the United States Postal Service Regarding Proposal Six, September 22, 2017 (USPS Reply Comments). The USPS Reply Comments were filed together with the Motion of the United States Postal Service for Leave to File Reply Comments Regarding Proposal Six, September 22, 2017. The Postal Service’s motion is granted.

14 Public Representative Response to Reply Comments of the United States Postal Service, September 26, 2017 (Response to USPS Reply Comments). The PR Response to USPS Reply Comments was filed together with Public Representative Motion for Leave to File a Response to Reply Comments Submitted by the United States Postal Service, September 26, 2017. The Public Representative’s motion is granted. Docket No. RM2017-10 - 4 - categories. Proposal Six at 2. The volumes of discontinued Parcel Select price categories are incorporated into Parcel Select Ground volume. Id. Likewise, Full Network PRS mail flow models have been added for machinable, non-machinable outside (NMO), and oversized parcels and now incorporate parcel volume that used to be in the discontinued RNDC PRS price categories. Id. at 2-3. These mail flow models are similar to those used for Parcel Select Ground, with the exception that they incorporate the fact that all return parcels are entered at delivery units and include additional operations to account for scanning and postage due costs specific to PRS parcels. Id. at 4-5. In addition, operations were added to or removed from the return delivery unit (RDU) PRS machinable, NMO, and oversize mail flow models to more accurately reflect the actual mail processing costs incurred by the Postal Service for these price categories. Id. at 5. Impact. The mail processing cost model is used to de-average estimated mail processing costs by shape for Parcel Select/PRS and to estimate separate costs for each price category. Given that the product-level mail processing cost by shape estimates are identical to those in Docket No. ACR2016, the mail processing unit cost estimates for some price categories increased while the mail processing unit cost estimates for other price categories decreased. Id. at 14. Full Network PRS cost estimates were added to the analysis. The Postal Service states that this change contributed to a lower Cost and Revenue Analysis (CRA) proportional adjustment factor, which normally would have resulted in lower return sectional center facility (RSCF) and RDU PRS mail processing unit cost estimates. Id. at 15. While the decreases to the RSCF mail processing unit cost estimates all exceeded 8 percent, the RDU machinable and NMO cost estimates increased over 4 percent and the RDU oversize cost estimate decreased by more than 30 percent. Id. The RDU results were caused by changes that were made to the cost models. The

15 Responses of the United States Postal Service to Questions 1-4 of Chairman’s Information Request No. 3, October 2, 2017 (Responses to CHIR No.3). Chairman’s Information Request No. 3 had previously been issued on September 25, 2017. See Chairman’s Information Request No. 3 and Notice of Filing under Seal, September 25, 2017 (CHIR No. 3). Docket No. RM2017-10 - 5 - container loading costs were removed from the models, and postage due costs were added to the models to reflect the fact that the RDU prices now vary by weight increment. Id. A full list of impacts by price category is shown in Table 1. Id. at 18.

Transportation Cost Model

Summary. The proposed transportation cost model removes worksheets and portions thereof specific to discontinued price categories. Id. at 5. The ONDC and NDC presort volumes were incorporated into Parcel Select Ground and RNDC volume was incorporated into Full Network PRS. Id. at 11. The transportation cost model estimates four types of transportation costs: (1) local; (2) intermediate; (3) long distance non-zone-related; and (4) long distance zone- related. These separate types are used to develop price category-level transportation costs. The proposed model uses a new methodology to estimate long distance zone- related costs. In the current model, these cost estimates are based on the proportion of Docket No. RM2017-10 - 6 -

stop-days16 at NDC facilities on InterNDC contracts.17 The Postal Service proposes that this statistic now be calculated using the TRACS percentage of cubic foot-miles18 for mailpieces traveling between NDC service areas. Proposal Six at 6. The proposed methodology uses data drawn from TRACS on transportation legs, each of which is defined as the direct trip of a mailpiece from one postal facility to another. If a mailpiece on one leg of transportation originates and destinates in different NDC service areas as determined by their 3-digit ZIP Codes, then that mailpiece is considered to have traveled a long distance. Id. While the current model only computes long distance zone-related costs for InterNDC contracts, the proposed methodology is used to estimate long distance percentages for the other three contract types as well. Id. at 6-7. Under the proposal, the long distance percentages described above are used to split the total attributable costs of each contract type into long distance zone-related costs and either local costs (for IntraSCF contracts) or intermediate costs (for InterNDC, IntraNDC, and InterSCF contracts). Id. at 11. The treatment of long distance non-zone-

16 Stop-days are the primary sampling unit of the Transportation Cost System (TRACS) Surface statistical sampling subsystem. Each stop-day in the TRACS sample represents the stop of a truck (van or vehicle) at one facility for a specific trip and day combination. A TRACS Surface test consists of a sample drawn from the mail that is unloaded from the truck during each stop-day. See Docket No. ACR2016, Library Reference USPS-FY16-36 Transportation Cost System (TRACS) Documentation December 29, 2016, file USPS_FY16_36_TRACS.Preface.pdf, at 3.

17 Id. at 6; Responses to CHIR No. 1, question 11. Highway service contracts are agreements between the Postal Service and external contractors to transport mail between postal facilities. There are four types of highway service contracts: (1) InterNDC; (2) IntraNDC; (3) InterSCF; and (4) IntraSCF. InterNDC contracts “carry mail between NDCs; however, InterNDC trips often stop at SCFs and may not necessarily connect two NDCs.” United States Postal Service, Handbook F-65, Transportation Cost Systems, October 2012, at 22-23. IntraNDC contracts “carry mail within an NDC service area; the route may be from an NDC to an AO [associate office] or SCF [sectional center facility] (outbound) or to an NDC from an AO or SCF (inbound).” Id. at. 22. InterSCF contract “carry mail between SCFs (some trips stop at AOs).” Id. IntraSCF contracts “carry mail between an SCF and the AOs with its service area.” Id. at 23.

18 Percentage of cubic-foot-miles represents the percentage of vehicle floor space occupied (or cubic-feet utilized) by a mailpiece or container and the miles traveled by the vehicle. See Docket No. ACR2016, Library Reference USPS-FY16-36 file USPS_FY16_36_TRACS.Preface.pdf, at 5. Docket No. RM2017-10 - 7 - related costs, which consist mainly of domestic air costs, remains the same as that of the current version of the model. Under both the proposed and current versions of the model, local, intermediate, and long distance zone-related costs are distributed to individual price categories based on the relative numbers of local, intermediate, and long distance legs incurred by parcels in each price category.19 The current methodology relies on operational assumptions about the number of legs that are expected to be incurred. Under this methodology, the current version of the model assumes that the average number of intermediate and long distance legs for destination delivery unit (DDU) and destination sectional center facility (DSCF) parcels and the average number of long distance legs for DNDC parcels are zero. Id. at 7. However, the Postal Service states that “[e]mpirical data suggests that each of the destination-entered price categories…incur[s] costs for modes of transportation in which one might not expect to find any costs. Id. The Postal Service proposes to estimate these values using the observed frequencies of destination-entry parcels on “unexpected” transportation legs using TRACS Surface test data. Id. at 7-9. An identical approach replaces assumptions about the number of average intermediate and long distance legs for RDU and RSCF parcels and local legs for RDU parcels, which are also assumed to be zero in the current version of the model. Id at 9. Impact. The transportation cost model is used to estimate costs per cubic foot for each price category and these costs are estimated by zone where applicable. Id. at 15. Cost Segment 14 costs are a major input to the transportation cost model. Id. In Docket No. RM2016-12, the Commission approved a change to the methodology used to estimate Cost Segment 14 highway costs.20 The impacts on individual price categories from the changes made in Docket No. RM2016-12 are shown in the second

19 Local transportation legs are defined as trips between a Delivery Unit (DU) and a Processing and Distribution Center (P&DC). Id. at 12. Intermediate transportation legs are defined as trips between a P&DC and an NDC. Id. Long distance transportation legs are defined as trips between NDCs. Id.

20 Id. See Docket No. RM2016-12, Order No. 3973, Order On Analytical Principles Used in Periodic Reporting (Proposal Four), June 22, 2017. Docket No. RM2017-10 - 8 - column of Table 2 of Proposal Six. See Proposal Six at 19. The estimated transportation cost per cubic foot for individual price categories decreased by between 13.0 percent and 16.7 percent. Id. The impacts on estimated transportation costs for individual price categories due to this proposal are listed in the third column of Table 2 of Proposal Six. Id. The changes in estimated costs for Ground price categories vary by zone distance, ranging from a decrease of 40.9 percent for Zone 1-2 parcels to an increase of 8.4 percent for Zone 9 parcels. Id. Estimated cost impacts for DNDC price categories increase by zone distance. Id. Among PRS price categories, the costs of RSCF parcels decrease by 9.7 percent as a result of the proposal. Id. The percentage change in cost for RDU is not reported because these parcels were assumed to have zero transportation legs in the original model, and consequently no transportation costs were distributed to this price category. Id. at 16-17. The percentage change in cost for Full Network is also not reported because the current transportation cost model does not contain this price category. Id. at 16. In its Petition, the Postal Service provided the estimated percentage change in transportation costs, including both the combined effects of this proposal and the changes approved in RM2016-12, and these changes are shown in Table 2. Id. at 19.

1. CHAIRMAN’S INFORMATION REQUESTS

Mail Processing Cost Model

The mail processing cost model initially filed with the Commission prompted the issuance of CHIR No. 1 and CHIR No. 2 to resolve a number of data issues.21 As part of its response to CHIR No. 1, question 1, the Postal Service identified various discrepancies in links to source data and filed a revised Excel file “Prop.6.ChIR.1.NP15.xlsx” in Library Reference USPS-RM2017-10/NP2 with corrections. In its Responses to CHIR No. 2, the Postal Service provides an operational explanation for certain mail processing cost estimates. Responses to CHIR No. 2, question 6. In that connection, the Postal Service filed a revised Excel file

21 See Responses to CHIR No. 1, questions 1-5; Responses to CHIR No. 2, questions 1, 3, 6-7, 9-10. Docket No. RM2017-10 - 9 -

“Prop.6.ChIR.2.NP15.xlsx” in Library Reference USPS-RM2017-10/NP3 with corrected mail flow models. Also in its Responses to CHIR No. 2, the Postal Service offers justifications for the Full Network PRS mail flow models and the incorporation of additional operations such as scanning and postage due operations into these models. Responses to CHIR No. 2, question 9.

Transportation Cost Model

In its Responses to CHIR No. 1, questions 1, 9, the Postal Service confirms the presence of inconsistencies in the version of the transportation cost model that was initially filed with the Commission. The Postal Service addresses those questions in its Responses to CHIR No. 1, CHIR No. 2, and CHIR No.3.22 It also filed a revised Excel file “Prop.6.ChIR.1.NP16.xlsx” in Library Reference USPS-RM2017-10/NP2 with these corrections. In CHIR No. 1, question 11, the Commission asks the Postal Service to discuss how the proposed method for classifying long distance transportation improves upon the current methodology and provide data to support this claim. The Postal Service asserts that the proposed methodology is superior because it accounts for long distance trips that terminate in facilities other than NDCs, and because it uses parcel transportation data directly instead of relying on indirect inferences based on the structure of the postal transportation network. Responses to CHIR No. 1, question 11. The Postal Service provided data on average leg lengths by contract type to support its assertion that the proposed methodology provides a more reasonable classification of long distance legs. Id. In its Responses to CHIR No. 2, the Postal Service states that these long distance percentages will be updated annually and confirms that its methodology does not include any estimates of the actual distances between the origin and destination facility (where each Parcel Select mail piece is unloaded) and the NDC that serves it. Responses to CHIR No. 2, questions 11(a), 11(c). In its Responses to CHIR No. 1, the Postal Service seeks to justify the need for the proposed classification of parcel transportation legs as either expected or unexpected,23 based on explaining how this methodology reflects actual postal operations and defining expected and unexpected legs together with examples. Responses to CHIR No. 1, question 13. The Postal Service asserts that the addition of unexpected transportation legs to the model is necessary to account for its observation

22 See Responses to CHIR No. 1, questions 6-13; Responses to CHIR No. 2, questions 2, 4, 5, 8, 11-12; Responses to CHIR No. 3, questions 1-4.

23 Expected transportation legs are trips taken by parcels that are consistent with the mail flow models used to represent the movement of parcels through the postal transportation network. For example, a DNDC parcel is expected to have one intermediate transportation leg (DNDC to DSCF) and one local transportation leg (DSCF to DDU) but no long distance transportation legs because DNDC parcels enter at the delivery NDC. Accordingly, transportation legs observed in the TRACS data involving DNDC parcels moving between NDCs are unexpected. Docket No. RM2017-10 - 10 - of destination-entry parcels using transportation modes that are assumed not to occur under the current version of the model. Responses to CHIR No. 1, question 13(a). The Postal Service also confirms that additional types of unexpected transportation legs are possible other than those addressed by the proposed methodology, but that the TRACS data are not able to provide estimates of additional unexpected legs that occur on top of expected transportation legs. Responses to CHIR No. 1, question 13(c). In its Responses to CHIR No. 2, questions 4, 5, the Postal Service attempts to justify its use of the unexpected legs methodology instead of using the observed frequencies of different parcel price categories on each transportation mode in the TRACS/Product Tracking and Reporting (PTR) data to distribute costs directly. Responses to CHIR No. 2, questions 4, 5. The Postal Service responded that their methodology is superior both because it accounts for differences in the cubic footage of parcels and because price categories with low volumes and low usage of contract types would have unstable cost estimates caused by small sample sizes. Id. The Postal Service also cites stability concerns stemming from small sample size to support maintaining its assumption that unexpected transportation legs are the same across price categories for each transportation mode despite acknowledging that relaxing this assumption would otherwise be preferable. Responses to CHIR No. 3, question 4. In its Responses to CHIR No. 3, the Postal Service discusses the feasibility of using a larger TRACS/PTR sample to mitigate the concerns about estimate instability discussed above but asserts that at least a five-fold increase in the number of TRACS Surface tests would be necessary to mitigate these concerns sufficiently to implement less restrictive alternatives such as those mentioned above. Responses to CHIR No. 3, question 3. The Postal Service asserts that such a large increase would be so expensive as to be infeasible. Id.

2. COMMENTS

United Parcel Service

UPS supports those provisions of Proposal Six that would rectify inaccurate assumptions regarding transportation costs incurred by destination-entry parcels. UPS Comments at 1. It asserts that the use of untested assumptions appears to lead to systematic under-attribution of costs to competitive products and supports basing cost attribution on data that reflects actual experience rather than untested assumptions. Id. at 2. UPS states that the Postal Service continues to use approximations in other parts of Proposal Six when more accurate information might be available. It offers the Postal Service’s methodology for calculating cubic foot-miles by zone as a specific example, arguing that using the actual route lengths between Postal facilities would be more accurate than the great circle distance measurement currently used by the Postal Service. Id. UPS suggests that the Commission encourage the Postal Service to use actual data in place of assumptions that mail moves via a great circle route. Id. Docket No. RM2017-10 - 11 -

Public Representative

The Public Representative supports the adoption of proposed modifications to the mail processing model advanced by Proposal Six to improve the model’s consistency and to better reflect actual postal operations. She agrees that the proposed weight-based classification for characterizing Parcel Select volume as machinable is reasonable from an operational standpoint. PR Comments at 4. She also agrees that disaggregating Parcel Select Ground volume into machinable, NMO, and oversize subcomponents and separately estimating the processing costs of these subcategories should improve the accuracy of the model. Id. at 5. She supports incorporating ONDC and NDC presort volume into Parcel Select Ground volume to maintain consistency with the CRA. Id. at 4. Finally, she agrees that using the Parcel Select Ground mail flow models as the bases for the mail flow models of Full Network PRS is reasonable. Id. at 6. The Public Representative suggests certain modifications and clarifications to improve the mail processing model. She recommends that parcels should be grouped consistently by price category in the “Summary” and “Volumes” worksheets in future filings. Id. at 3. She also suggests that the Postal Service clarify the interaction between the RNDC and Full Network PRS price categories and volumes reported in the current and proposed versions of the model to improve transparency. Id at 5. Finally, she encourages the Commission to request the Postal Service perform additional checking of the model for errors. Id. at 6. The Public Representative expresses concerns about the proposed methodology for calculating long distance percentages in the transportation cost model. She notes that the proposed methodology does not consider actual distances between facilities, which can result in transportation legs near NDC service area boundaries being classified as long distance despite having low actual mileage. Id. at 9. She recommends that the Postal Service provide a clear definition of long distance, intermediate, or local transportation modes. Id. The Public Representative also expresses serious concerns about the proposed methodology for calculating long distance percentages in the transportation cost model and, although she affirms that operational circumstances warrant consideration of unexpected transportation legs in the model, she questions whether the unexpected transportation legs and long distance cost percentages have been accurately estimated. Id. at 9-10.

Postal Service Reply to Public Representative Comments

In its reply, the Postal Service seeks to clarify the record regarding several points made in the Public Representative’s comments regarding the mail processing model. USPS Reply Comments at 1-2. The Postal Service states that machinable and irregular Parcel Select Lightweight (PSLW) volumes are disaggregated on the “Volumes” worksheet of the mail processing model because these categories have different processing costs which are combined into a weighted average on the “Summary” Docket No. RM2017-10 - 12 - worksheet. Id. at 2. The Postal Service also states that the Full Network price category is not a replacement for the RNDC price category; rather, Full Network and RNDC volumes were combined to allow total PRS volume to match the CRA. Id. at 3-4. Finally, the Postal Service asserts that the model’s contents have been reviewed and verified and that additional checks of the model are unnecessary. Id. at 4. The Postal Service also clarifies the distinctions between local, intermediate, and long distance transportation used in the transportation cost model.24 The Postal Service responds to the Public Representative’s critique of the proposed long distance classification methodology for not relying on actual mileage measures by noting that the current methodology does not rely on actual mileage either and that, while long distance designations do not rely on mileage measures, the distribution of long distance zone- related costs to price categories does. Id. at 6. The Postal Service also responds to the Public Representative’s assertion that the unexpected transportation percentages are too high considering the rarity of operational circumstances relied upon to develop those percentages by noting that the number of unexpected transportation legs of destination- entry price categories may be large even if the fraction of such parcels using unexpected transportation modes is very small if the overall volume of parcels in these categories is high. Id at 5. The Postal Service states further that, while actual distances between facilities are not used to determine whether a mailpiece is long distance, cubic foot-mileage estimates based on actual distance measurements are subsequently used to distribute long distance costs. Id. at 6. The Postal Service also states that it disagrees with the Public Representative’s statement that the proposed methodology moves costs from intermediate to long distance without any reasonable justification; rather, it asserts that the proposed methodology is based on empirical data and hence reflects operational realities more accurately. Id. at 7.

Public Representative Response to Postal Service Reply

The Public Representative states that she finds the Postal Service’s explanation regarding its handling of Full Network PRS volume reasonable. Response to USPS Reply Comments at 1. She concludes that the Postal Service has responded to all major concerns regarding the mail processing model, but notes that it appears possible that this model still contains discrepancies. Response to USPS Reply Comments at 1- 2. Although the Public Representative is generally satisfied with the proposed changes to the mail processing model, she continues to oppose the Postal Service’s

24 The Postal Service states that “[l]ocal costs are the transportation costs that are incurred in transporting parcels between facilities that are within the service area of a processing and distribution center (P&DC), primarily between delivery units and P&DCs. Intermediate costs are the transportation costs incurred in transporting parcels between facilities that are within the service area of a network distribution center (NDC), primarily between P&DCs and NDCs. Long distance costs are the transportation costs incurred in transporting parcels between facilities that are in different NDC service areas, primarily between two NDCs.” Id. at 5. Docket No. RM2017-10 - 13 - proposed changes to the transportation cost model and continues to question whether unexpected transportation costs reflect operational realities. Id. at 2-3. She asserts that although neither the current nor the proposed methodology uses actual distances between facilities, the current methodology for classifying long distance transportation is rational because it ties transportation modes to contract types, whereas, the proposed methodology allows transportation costs from any contract type to be classified as long distance. Response to USPS Reply Comments at 2-3. She maintains that an expansion of the long distance travel designation is unjustified and would likely require a change in the definitions of highway transportation contracts, which is outside the scope of Proposal Six. Id. at 3. The Public Representative does not, however, explain why a change in the definitions of highway contracts would be necessary.

3. COMMISSION ANALYSIS

Based upon a review of the Postal Service’s filing, supporting workpapers, and the comments of the participants, the Commission approves Proposal Six. Pursuant to 39 C.F.R § 3050.42, the Commission finds that the proposed modifications to the Parcel Select/PRS mail processing cost and transportation cost models would improve the quality and reliability of the Postal Service’s costing methodology for these products.

Mail Processing Cost Model

The Commission finds that all of the modifications to the revised mail processing model that were filed in Library Reference USPS-RM2017-10/NP3 will improve the quality of the cost estimates produced by this model. In disaggregating Parcel Select Ground volume into machinable, NMO, and oversize volumes, Proposal Six produces cost estimates that better reflect differences in the operations required to process these parcel types. In addition, the Commission agrees that updating the price categories included in both models to reflect changes in the Price List will result in cost estimates that better reflect current postal operations.

Transportation Cost Model

The Commission finds that the proposed changes to the transportation cost model should improve the quality and accuracy of the cost estimates. The Commission agrees with the Postal Service that using data to update the fixed assumptions made in the current version of the model is an improvement. The Commission notes that two aspects of the transportation cost model could be improved: (1) the calculation of long distance percentages; and (2) the calculation of unexpected transportation legs.

Calculation of Long Distance Percentages

The Commission finds that the proposed methodology for estimating long distance percentages improves the current methodology. The proposed modifications allow for the possibility that transportation on contract types other than InterNDC may Docket No. RM2017-10 - 14 - also entail travel over long distances. The Commission finds that the proposed methodology designates long distance transportation more reasonably than the current methodology. The average mileage statistics presented in the Postal Service’s Responses to CHIR No.1, question 11, which show that IntraNDC and InterSCF transportation legs are frequently hundreds of miles long, support the Postal Service’s assertion that it is reasonable to treat some of the costs of these contract types as long distance. The Commission does not agree with the Public Representative’s contention that an expansion of the long distance travel designation is unjustified and would likely require a change in the definitions of highway transportation contracts, which is outside the scope of Proposal Six. Response to USPS Reply Comments at 3. In particular, the Commission notes that long distance costs in the model differ from other costs solely because they are distributed to price categories using cubic foot-miles instead of cubic- foot; accordingly, the Commission agrees with the Postal Service that the costs of all long distance transportation legs should be distributed consistently, regardless of the contract type into which they fall. The Commission also agrees in principle with the Public Representative’s comment that a long distance classification methodology based on the actual distances traveled by parcels would be superior to the proposed methodology. PR Comments at 8. In its Responses to CHIR No. 1, question 11, the Postal Service demonstrated the ability to measure the actual miles traveled of parcels observed in the TRACS/PTR data. The Commission requests the Postal Service to research the feasibility of replacing the long distance classification methodology proposed in this docket with one that explicitly uses the actual mileage traveled by parcels in the TRACS/PTR data sample to calculate long distance percentages and report its findings to the Commission within 90 days of the date this order is issued.

Calculation of Unexpected Transportation Legs

The Commission agrees with the Postal Service that the proposed introduction of unexpected transportation legs improves upon the current version of the transportation cost model. The Postal Service explains that the introduction of unexpected legs into the model was a new concept intended to correct empirically false assumptions about the estimated number of legs traveled by destination entry products made under the current approved methodology by basing these estimates on observed data rather than incorrect operational assumptions. Responses to CHIR No. 1, question 13. While the assumption that price categories have identical proportions of unexpected transportation legs conflicts with the TRAC /PTR data, the Commission acknowledges the Postal Service’s concern that replacing additional fixed assumptions might result in unstable cost estimates. Responses to CHIR No. 3, question 4. The Postal Service’s quantification of the implied sampling error in the TRACS/PTR data suggests that the small sample size of these data will cause estimates made with these data to have large margins of error. Responses to CHIR No. 3, question 3. However, this evidence by itself does not establish how sampling error in the TRACS/PTR data impacts the standard errors of the cost estimates produced by the model or Docket No. RM2017-10 - 15 - demonstrate that the proposed methodology diminishes this impact more than alternatives that rely on fewer a priori assumptions. When it provides the FY 2017 estimates developed using the FY 2017 TRACS/PTR data, the Postal Service should include a comparison of the resulting change in estimated transportation costs in the proposed model due to the data update with the change in estimated costs that would have resulted if unexpected transportation legs were computed separately for each price category.

Final Mail Processing and Transportation Cost Impacts

The Commission approves Proposal Six based on the foregoing analysis of the revised mail processing model presented in Excel file “Prop.6.ChIR.2.NP15.xlsx” that was filed in Library Reference USPS-RM2017-10/NP3 in its Responses to CHIR No. 2, questions 1, 2, and the revised transportation cost model presented in Excel file “Prop.6.ChIR.1.NP16.xlsx” that was filed in Library Reference USPS-RM2017-10/NP2 in its Responses to CHIR No. 1, question 1. The impact on cost estimates after adjusting for the differences between the revised versions of the models and the initial versions filed by the Postal Service in Library Reference USPS-RM2017-10/NP1 is listed in Table 1 and Table 2 below. Docket No. RM2017-10 - 16 -

Table 1: Final Mail Processing Cost Impact Percent Price Category Change Parcel Select Ground Machinable -3.2% Ground NMO --- Ground Oversize --- DNDC Machinable -0.8% DNDC NMO -0.8% DNDC Oversize -0.8% DSCF 5-Digit Machinable -0.6% DSCF 5-Digit NMO -0.7% DSCF 5-Digit Oversize -0.7% DSCF 3-Digit NMO -0.7% DDU Machinable -0.7% DDU NMO -0.7% DDU Oversize -0.7%

Parcel Select Lightweight No Destination Entry MNDC 1.1% No Destination Entry NDC 0.1% DNDC Entry NDC 0.4% DNDC Entry SCF 2.5% DNDC Entry 5-Digit 0.8% DSCF Entry SCF 1.5% DSCF Entry 5-Digit -0.7% DDU Entry 5-Digit -0.8%

Parcel Return Service Full Network Machinable --- Full Network NMO --- Full Network Oversize --- RSCF Machinable -8.1% RSCF NMO -8.3% RSCF Oversize -10.9% RDU Machinable 6.1% RDU NMO 4.7% RDU Oversize -34.8%

Barcode Savings Estimate 0.0%

Table 2: Final Transportation Cost Impact (Revised Additional Docket No. RM2017-10 - 17 -

CS/14) ACR2016 Proposal Total Percent Percent Percent Price Category Change Change Change

Parcel Select Ground Zones 1 and 2 -15.2% -40.9% -56.1% Zone 3 -15.2% -37.1% -52.3% Zone 4 -15.2% -32.1% -47.3% Zone 5 -15.1% -27.4% -42.5% Zone 6 -15.1% -17.6% -32.6% Zone 7 -15.0% -5.7% -20.7% Zone 8-9 -14.9% 8.4% -6.4% DNDC Zones 1 and 2 -16.2% 1.0% -15.1% Zone 3 -16.6% -64.2% -80.8% Zone 4 -16.7% -73.9% -90.5% Zone 5 -16.7% -77.9% -94.6% DSCF -13.2% 206.4% 193.2% DDU -13.0% 2.9% -10.1%

Parcel Return Full Network Zones 1 and 2 ------Zone 3 ------Zone 4 ------Zone 5 ------Zone 6 ------Zone 7 ------Zone 8-9 ------RSCF -16.0% -9.7% -25.7% RDU ------

Docket No. RM2017-10 - 18 -

4. ORDERING PARAGRAPH

It is ordered: 1 For purposes of periodic reporting to the Commission, the changes to analytical principles proposed by the Postal Service in Proposal Six are approved.

2. Within 90 days of the date this Order is issued, the Postal Service should report to the Commission the feasibility of replacing the long distance classification methodology proposed in this docket with one that explicitly uses the actual mileage traveled by parcels in the TRACS/PTR data sample to calculate long distance percentages.

By the Commission.

Stacy L. Ruble Secretary