Last week the United Parcel Service lost its appeal in U.S. Circuit Court challenging a September 2016 ruling (Order No. 3506) by the Postal Regulatory Commission concerning the Postal Service’s cost allocation methodologies. The Court’s decision represents another setback for UPS in its efforts to force the Postal Service to raise prices on the shipping services with which UPS competes.
While the winner in the case is technically the PRC, whose previous rulings were upheld, the other winners include those who intervened on behalf of the PRC — the Postal Service, National Association of Letter Carriers, AFL-CIO, Parcel Shippers Association, and Amazon. (Their brief to the Court is here.)
Because the Court’s ruling will not force the Postal Service to raise parcel rates — which would allow UPS and FedEx to raise their rates as well — the ultimate winners are the small businesses that won’t pay higher shipping rates and consumers who shop online.
As the following analysis will suggest, if UPS had prevailed in court, prices on parcels might have gone up about 7 percent. The price on a medium Flat Rate Box could have gone up nearly $1.
The Court’s ruling is also of interest because it relates to Trump’s claims that the Postal Service is undercharging on Amazon deliveries. Those claims were based on a Wall Street Journal op-ed that was based on a flawed Citigroup report that was based on a UPS petition to the PRC concerning cost methodologies. That petition involved the “appropriate share” that competitive products (shipping services) should contribute to institutional (fixed) costs. (For more on that, check out this earlier post.)
The case decided last week by the Court didn’t involve this “appropriate share” issue. Instead, it involved another aspect of the UPS petition on costing methodology — namely, how the Commission interprets “attributable costs,” and in particular, how it handles what’s called “inframarginal costs.”
Back in October 2015, UPS filed a petition with the Commission making several proposals for changing costing methodologies, addressing both the appropriate share issue and inframarginal costs. (These proposals were discussed by Mark Jamison in an excellent post on this website entitled “When Titans collide: UPS petitions the PRC to change USPS costing methodologies.”)
In September 2016, the Commission issued an order rejecting most of the UPS argument on inframarginal costs while recognizing that a relatively small costing adjustment would be appropriate. In December 2016 PRC Order No. 3641 worked out the details on this adjustment. UPS wasn’t happy about the direction things were going and appealed to the US Circuit Court.
The Court’s ruling last week examines some very technical issues concerning postal costing in great detail, but ultimately it came down to Chevron, a famous Supreme Court case. According to Chevron, the courts should defer to the administrative agency to which Congress gave authority to decide cases, so long as the agency’s decision is “reasonable and permissible.”
The PRC didn’t have to prove its costing methodology was absolutely correct or even more correct than the one UPS had proposed. It just had to show its order was reasonable. In the Court’s view, the Commission’s judgment was in fact reasonable and neither arbitrary nor capricious. In accordance with the doctrine of Chevron deference, the Court rejected the UPS appeal.
Attributable and institutional costs
The rules set by the 2006 Postal Accountability and Enhancement Act (PAEA) apportion the costs on postal products between “attributable” and “institutional.”
Attributable costs are those for which there’s a causal relationship between the product and the expense. The greater the volume, the greater the cost; if the mail piece didn’t exist at all, the attributable cost wouldn’t exist either. The institutional costs, on the other hand, are those that would exist regardless of volume; they’re sometimes called fixed costs or just “overhead.”
Here’s a chart illustrating the two main cost categories. The chart comes from a classic text on pricing, the OIG’s Primer on Postal Costing Issues (2012). The numbers are from 2010.
Due to increases in parcel shipping, the numbers have changed somewhat since 2010. In 2017, total costs were almost $70 billion; attributable costs were 41.6 billion, and institutional costs were about $28.4 billion — about 60/40 percent-wise.
Here’s a another graphic illustrating how USPS costs are allocated. This one is from a report by Dr. Kevin Neels submitted to the PRC in support of the UPS petition.
As the OIG Primer explains, attributable costs consist of two components: (1) Volume variable costs are those that vary directly with the volume; variable costs go to zero if the activity stops. (2) Product-specific fixed costs don’t vary with volume, but they are directly caused by the product. Product-specific costs are typically very small, about 3 percent of total costs, so the chart doesn’t even include product-specific costs at all.
The institutional costs also include two components. (1) Common fixed costs are those that cannot be attributed to a specific product or service and that would be incurred even if there were no volume. (2) Inframarginal costs are variable costs that do not vary directly with the volume of a particular mail product because they are caused by two or more products in common.
According to Dr. Neels’ analysis, about half of institutional costs are fixed and half are inframarginal. UPS wanted the PRC to shift all these inframarginal costs from the institutional cost category to the attributable category.
The dispute over “inframarginal costs” is esoteric, technical, and impossible for a layperson to fully grasp, but here’s a rough picture.
Consider how economies of scale affect pricing. Say it costs $100 to deliver a single letter; two letters might cost $188 (so the second letter costs $88); three letters might cost $268 (the third letter costs $80). Eventually, the cost per letter will approach a point where each additional letter has very little impact on price per piece. In this example that cost is $.05. This table from the OIG Primer illustrates the example.
In this example, the Postal Service uses the marginal cost of $.05 — the cost of the “last piece” — as the volume variable cost, and all the extra costs that were incurred by the lower-volume pieces are considered “inframarginal” costs.
Here’s how these inframarginal costs are described in the PRC’s 2016 ruling on the UPS petition: “Inframarginal costs are costs that result from economies of scale and scope. They are variable costs that do not vary with volume, and they are unique to postal costing. They contain both common variable costs and the costs that result from economies of scale. These costs have effectively been treated as institutional costs. In a marginal cost curve, these are the costs remaining after volume-variable costs have been calculated, as shown in the green area in Figure A-8 below.”
While inframarginal costs are variable, they would not be avoided if the product were eliminated, and in the real world of the postal system, where the mail is all mixed together in plants and on trucks, they are also difficult — or perhaps impossible — to determine. Historically, these inframarginal costs have therefore been treated as if they were like a fixed cost and classified as institutional costs.
It’s this practice that UPS has challenged as inappropriate. UPS argues that it is in fact possible to assign inframarginal costs to specific products.
UPS proposes, PRC rejects
Here’s how UPS explained the rationale for assigning inframarginal costs to particular products in its 2015 petition to the PRC:
“The Postal Service’s cost accounting methodologies assume that the cost of delivering each unit of its competitive products is equivalent to the cost of delivering only the last unit of those products. This assumption is reasonable only when the marginal cost of producing the last unit is constant for all levels of volume. But when marginal cost varies with the level of volume being produced, such as when economies of scale and scope make producing additional units of output cheaper, this assumption is unreasonable and inaccurate, and it leads to a dramatic understatement (and thus under-attribution) of variable costs.”
By treating inframarginal costs as institutional, argued UPS in its comments on the proposal, the Postal Service has been shifting “nearly all of the cost savings of [its] economies of scale and scope” to shipping services. According to UPS, competitive products are thus “being allowed to ride nearly for free on a delivery network paid for almost entirely by market dominant products.”
Many of the commenters on the docket disagreed with UPS. For example, the National Association of Letter Carriers characterized UPS’s petition as its “latest ploy . . . to use rhetoric about supposed unfair postal subsidies to try to gain a competitive advantage for itself in the parcel market.” Regarding Dr. Neels’ method for calculating inframarginal costs on each product, NALC described it as “based on arcane game theory,” which “has little connection to the reality of the cost of mail delivery.”
In September 2016, the Commission rejected UPS’s proposal to reclassify all inframarginal costs as attributable, finding that UPS relied on “unverifiable assumptions” for both “the calculation and allocation of inframarginal costs.”
The way the Commission read PAEA, costs can be attributed to a product only if a “reliably identified causal relationship” exists between the costs and the product. As NALC, Amazon, and the other intervenors put it in their brief to the Court: “Allocating common costs to an individual product fails this test because eliminating the product would not avoid the costs. Decades of Commission and judicial precedent and the consensus of mainstream economics support both of these propositions. UPS offers no valid counterargument.”
While denying UPS petition to make all inframarginal costs attributable, the Commission did agree to making some of these costs attributable. Here’s how the UPS proposal might have worked out and how things actually changed.
Changing all inframarginal costs
Fully distributing all inframarginal costs among attributed costs, as UPS proposed, would have impacted pricing for competitive products significantly.
In its petition, UPS observes that Neels’ report “confirms that the variable costs the Postal Service disregards are staggering — in FY 2014, the Postal Service failed to attribute over $2.68 billion to its competitive products business due to this methodological flaw alone” (a reference to treating intra-marginal costs as institutional).
Here’s a table from Neels’ report showing how he came to that number.
As the table shows, the attributed costs for both market dominant and competitive mail would be increased by UPS’s proposal. Over $13 billion would be shifted from institutional costs to attributed costs. But because market dominant products cover a larger share of institutional costs, the net effect of shifting costs like this is that costs for market dominant products go down while costs for competitive products go up.
That shouldn’t come as a surprise, considering that the whole point of the UPS petition is that market dominant products are overpriced and thereby cross-subsidizing competitive products, which are underpriced.
Overall, according to Dr. Neels’ analysis, attributable costs for competitive products would go up by $2.68 billion, an increase of 25 percent. But because their institutional costs would be reduced at the same time, total costs for competitive products wouldn’t go up by 25 percent. Costs would go up, though, and so would prices.
Estimating the impact on prices
In its filings with the PRC, neither Dr. Neels nor UPS took the step of figuring out how actual prices might be impacted by their proposal. Below is a table based on Neels’ table 1-1 that shows how total costs could be impacted if the UPS proposal were adopted.
Additional numbers for the table come from the PRC’s 2014 Annual Compliance Determination Report, p. 83, which notes that total institutional costs were $34.2 billion and competitive products’ contribution was $4.3 billion, or 12.9 percent. For the numbers under the UPS proposal, the table assumes that competitive products would contribute the same 12.9 percent to institutional costs, i.e., $2.6 billion out of $20.3 billion. (It should be noted that as part of its “appropriate share” proposal, UPS also wanted that percentage changed, to about twice this amount, which would increase competitive prices even more.)
|Potential Impacts of UPS Proposal on Inframarginal Costs|
|FY 2014 ($ billion)||Under 2014 rules||Under UPS Proposal|
|Attribu-table||Institu-tional||Total||Attribu-table||Institu-tional||Total||Differ-ence $||Differ-ence %|
As the table shows, shifting $13 billion from institutional costs to attributable costs, as UPS proposed, ends up decreasing total costs on market dominant products by $1.4 billion, or 2.5 percent, while increasing total costs on competitive products by about $1 billion, or 6.8 percent.
Roughly speaking, if those cost changes were reflected in the final price, the price of a First Class stamp would go down by one cent while the price of a medium Priority Flat Rate Box (currently $13.65) would go up $0.93. It’s easy to see why UPS would like that result.
These numbers also explain why Amazon intervened in the case against UPS. If, as discussed in this earlier post, Amazon’s deal with the Postal Service involves paying about $2 per parcel on 1.2 billion Parcel Select deliveries per year, an increase of 6.7 percent would come to $160 million a year.
In its comments on the UPS proposal, Amazon doesn’t get into numbers like these, but it does say this: “The relief ultimately sought by UPS through its petition for review could lead to higher postal rates and shipping costs for Amazon, its customers, and the customers of independent merchants (manufacturers, wholesalers and retailers, large and small) who sell goods on Amazon and other online commerce websites.” (Law360)
Changing some inframarginal costs
While the PRC did not accept the UPS proposal to reallocate all inframarginal costs, the Commission was somewhat responsive. The Commission agreed to change the costing methodology to include some inframarginal costs, namely those that it felt could be reliably attributed to individual products. In December 2016 the Commission issued Order No. 3641 implementing this adjustment.
The effect on attributable costs, however, was very small. The PRC’s FY 2017 Financial Analysis (p. 36) notes that “the change in methodology only impacts the comparison to FY 2016 results slightly because the attributed inframarginal costs make up less than 3 percent of each product’s attributable costs.”
For competitive products, these attributed inframarginal costs may be even smaller than that.
The FY 2017 Public Cost and Revenue Analysis (PCRA) Report shows both the old way of calculating attributable costs and the new way approved by the PRC. The report therefore provides some idea of the share of inframarginal costs the PRC shifted to attributable costs (click on the table to enlarge it).
In this report, the “attributable cost” column reflects the change made by the PRC in 2016; it represents the sum of the volume variable cost, the product specific cost, and the portion of inframarginal costs approved by the PRC. In the next column, the report shows just the volume variable cost. The difference between the attributable cost and volume variable cost represents the sum of the product specific cost and the inframarginal cost that the PRC shifted to attributable costs.
Overall, for Competitive Products, volume variable costs totaled $13.538 billion, and attributable costs were $13.319 billion, a difference of about $220 million. The Annual Compliance Report for FY 2017 (p. 73) indicates that $163 million of this $220 million were product specific costs and $56 million were inframarginal costs.
For another version of the cost analysis, there’s Appendix A of the PRC’s FY 2017 Financial Analysis, p. 85. It shows that inframarginal costs on domestic competitive products were $194.3 million and on international competitive products, 24.7 million, for a total of $219 million.
For a third version of the numbers, the Postal Service’s Annual Compliance Report for FY 2017, p. 68, includes a table showing that total inframarginal costs for competitive products were $359 million; how that total was determined is unavailable since the costs for individual products were submitted as a nonpublic library reference.
In other words, depending on which of these numbers one uses, as a result of the changes in costing methodology approved by the PRC in 2016, attributable costs on competitive products were $56 million or $220 million or $359 million greater than they would have been using the earlier methodology. That represents an increase in attributable costs of about 0.4 percent, 1.6 percent, or 2.6 percent, respectively.
All of these estimates are obviously a lot less than the 25 percent increase UPS had proposed. Such a small change in costing methodology probably had little or no impact on pricing for parcels and on UPS’s competitive position.
Dissatisfied with the PRC’s modest concession to reclassify a small portion of inframarginal costs, UPS petitioned the Circuit Court of Appeals for review of the PRC’s 2016 orders. UPS argued that the Commission’s decision not to require the Postal Service to attribute all inframarginal costs to specific products was both inconsistent with PAEA and “arbitrary and capricious.”
The Court’s ruling reviews and analyzes in great detail several highly technical issues involved with costing methodologies, but in some respects the Court’s decision was ultimately based not on technical matters but rather on a well-known legal doctrine called Chevron deference.
According to the landmark 1984 Supreme Court case, Chevron U.S.A., Inc. v. Natural Resources Defense Council, Inc., the courts are supposed to defer to the judgment of the administrative body that Congress has charged with authority to make decisions. A court is not supposed to substitute its judgment for that of the agency so long as the agency’s decision is “reasonable or permissible.”
In deciding on the case, the Circuit Court did not have to determine whose interpretion of PAEA was more reasonable or correct. It was just a matter of determining if the Commission’s order was reasonable. “Given our conclusion that the Commission’s reading of ‘institutional costs’ [in PAEA] is reasonable and so merits our deference,” wrote the Court, “we need not consider the Commission’s argument that, under Chevron, its reading is not only permissible, but also unambiguously correct.”
As the Court states in the conclusion to its ruling, Congress expected that “the expert ratesetting agency, exercising its reasonable judgment” would “decide which methods sufficiently identify the requisite causal connection between particular services and particular costs.” The Commission “did exactly that.” Because the Commission’s exercise of its authority was “reasonable and reasonably explained,” the Court denied UPS’s petitions for review.
The Court’s ruling may not be the end of the matter. Conceivably UPS could appeal to the U.S. Supreme Court. It might find a receptive audience, considering that Judge Gorsuch has expressed a desire to revisit Chevron.
(Photo: Victor J. Blue | Bloomberg | Getty Images. UPS driver pushes a dolly of packages in New York.)