United Parcel Service , Inc . v. Postal Regulatory Commission

United States Court of Appeals
FOR THE DISTRICT OF COLUMBIA CIRCUIT
Argued January 22, 2018 Decided May 22, 2018
No. 16-1354
UNITED P ARCEL S ERVICE , INC .,
P ETITIONER
v.
P OSTAL R EGULATORY C OMMISSION,
R ESPONDENT
VALPAK FRANCHISE A SSOCIATION, INC ., ET AL.,
INTERVENORS
Consolidated with 16-1419
On Petitions for Review of Orders
of the Postal Regulatory Commission
Jeffrey A. Lamken argued the cause for petitioner. With
him on the briefs were James A. Barta, Steig D. Olson, and
Sara E. Margolis.
Bryan N. Tramont and Craig E. Gilmore were on the brief
for amicus curiae J. Gregory Sidak in support of petitioner.
Michael Shih, Attorney, U.S. Department of Justice,
argued the cause for respondent. With him on the brief were

-- 1 of 30 --

2
Michael S. Raab, Attorney, David A. Trissell, General Counsel,
Postal Regulatory Commission, and Christopher J. Laver,
Deputy General Counsel.
Morgan E. Rehrig and Eric P. Koetting, Attorneys, U.S.
Postal Service, Peter DeChiara, David M. Levy, John F.
Cooney, and James Pierce Myers were on the brief for
intervenors in support of respondent.
Before: TATEL, S RINIVASAN , and P ILLARD, Circuit Judges.
Opinion for the Court filed by Circuit Judge TATEL.
TATEL, Circuit Judge: The U.S. Postal Service holds
congressionally authorized monopoly power over the market
for some of its products, like first-class mail delivery, but for
other products, like parcel post, it competes with private
companies. To promote fair competition, Congress tasked the
Postal Regulatory Commission with ensuring that the Postal
Service sets competitive products’ prices high enough to cover
all “costs attributable to [those] product[s] through reliably
identified causal relationships.” 39 U.S.C. § 3631(b); see also
id. § 3633(a)(2). In two 2016 orders, the Commission directed
the Postal Service to include among the “costs attributable” to
competitive products those costs that would disappear were the
Postal Service to stop offering those products for sale. United
Parcel Service, Inc., which competes with the Postal Service,
petitions for review of both orders, arguing that the cost
attribution methodology the Commission embraced is both
inconsistent with the statute that gives the Commission its
regulatory authority and arbitrary and capricious. For the
reasons that follow, we deny the petitions.

-- 2 of 30 --

3
I.
Congress created what is now the Postal Regulatory
Commission (the “Commission”) in 1970 to oversee the U.S.
Postal Service’s efforts to set “reasonable and equitable rates
of postage and fees for postal services.” Postal Reorganization
Act, Pub. L. No. 91-375, § 3621, 84 Stat. 719, 760 (1970)
(codified as amended at 39 U.S.C. § 404(b)); see also id.
§ 3601, 84 Stat. at 759 (establishing the Commission). The
2006 Postal Accountability and Enhancement Act (the
“Accountability Act”), Pub. L. No. 109-435, 120 Stat. 3198
(2006), provides the framework within which the Commission
currently exercises this oversight authority.
Under the Accountability Act, all Postal Service products
are either “market-dominant” or “competitive.” See 39 U.S.C.
§ 3642(b)(1). Market-dominant products are those over which
“the Postal Service exercises sufficient market power that it can
effectively” raise prices or decrease quality “without risk of
losing a significant level of business to other firms offering
similar products.” Id. To prevent the Postal Service from
“improperly leverag[ing]” this market power, U.S. Postal
Service v. Postal Regulatory Comm’n, 785 F.3d 740, 744 (D.C.
Cir. 2015), the Act requires the Commission to limit rate
increases for market-dominant products, see 39 U.S.C.
§§ 3622(a), (d)(1); see also 39 C.F.R. §§ 3010.1–3010.66
(implementing this mandate).
Different concerns attend competitive products—products
over which “the Postal Service faces meaningful market
competition.” U.S. Postal Service, 785 F.3d at 744. For such
products, Congress wished to “ensure that the Postal Service
competes fairly,” S. Rep. No. 108-318, at 15 (2004) (“Senate
Report”)—that is, without using revenues from market-
dominant products subject to its monopoly power to defray
costs competitive products would otherwise have to be priced

-- 3 of 30 --

4
to cover. The Accountability Act therefore requires the
Commission to promulgate regulations that “prohibit the
subsidization of competitive products by market-dominant
products,” 39 U.S.C. § 3633(a)(1); “ensure that each
competitive product covers its costs attributable,” id.
§ 3633(a)(2), defined as “the direct and indirect postal costs
attributable to such product through reliably identified causal
relationships,” id. § 3631(b); and “ensure that all competitive
products collectively cover what the Commission determines
to be an appropriate share of the institutional costs of the Postal
Service,” id. § 3633(a)(3).
In effect, the Accountability Act subjects each competitive
product to a “price floor,” U.S. Postal Service v. Postal
Regulatory Comm’n, 842 F.3d 1271, 1272 (D.C. Cir. 2016)
(per curiam), which must be set high enough to cover both that
product’s “costs attributable,” 39 U.S.C. § 3633(a)(2), and a
portion of the Postal Service’s “institutional costs,” id.
§ 3633(a)(3), which the Commission construes to mean
“residual costs,” i.e., all costs that are not costs attributable, see
Order Concerning United Parcel Service, Inc.’s Proposed
Changes to Postal Service Costing Methodologies (UPS
Proposals One, Two, and Three), No. RM2016-2, at 10 (Postal
Regulatory Comm’n Sept. 9, 2016) (updated Oct. 19, 2016)
(“Order”).
This case concerns the Commission’s rules for
apportioning postal costs between “attributable” and
“institutional” costs. 39 U.S.C. §§ 3633(a)(2), (a)(3). Treating
the latter category as “residual” of the former, Order at 10,
Commission regulations focus on identifying which costs are
“attributable to [a specific] product through reliably identified
causal relationships,” 39 U.S.C. § 3631(b). In doing so, the
Commission distinguishes (albeit necessarily imperfectly)
between “fixed costs,” such as executive salaries, which remain

-- 4 of 30 --

5
constant regardless of overall product volume, and “variable
costs,” such as wage labor or raw materials, which vary with
the Service’s production levels. Order at 6; see also Responses
of the United States Postal Service to Questions 1–4 of
Chairman’s Information Request No. 2, No. RM2016-2, at 11
n.9 (Postal Regulatory Comm’n Dec. 10, 2015) (“Postal
Service Responses”) (acknowledging that “fixed costs can be
difficult to identify in practice”). Except for certain product-
specific costs not at issue, fixed costs are not attributed to
particular products and so are considered institutional. See
Order at 9 & n.12 (attributing only those fixed costs “that are
uniquely associated with an individual product,” id. at 9, such
as product-specific advertising costs). The issue here is what
portion of the Postal Service’s variable costs can be reliably
attributed.
Broadly speaking, the Postal Service, in implementing
Commission regulations, attributes variable costs on an
activity-by-activity basis. After drawing up a list of the discrete
production activities, such as highway transportation, that
collectively account for its total variable costs, the Postal
Service calculates what share of each activity’s costs can be
attributed to each product. See Order App’x A at 13–14 (laying
out this process); Postal Regulatory Comm’n, FY16 Public
Cost Segments and Components Report (2016),
https://go.usa.gov/x54x2 (listing production activities). To
perform this calculation, it first identifies an activity’s “cost
driver,” defined as the unit of measurement that best captures
the activity’s “essen[ce].” Order App’x A at 14. For example,
highway transportation is measured in cubic-foot-miles, such
that one “unit” of cost driver in this context represents one
cubic foot of mail being transported one mile. See id. Then, the
Postal Service determines the share of each activity’s cost-
driver units that each product is responsible for generating,
typically by conducting worksite observations in order to

-- 5 of 30 --

6
produce a “distribution key” that, like a pie chart, illustrates an
activity’s product-by-product breakdown. See id. at 9; see also
Order at 9 n.14; Office of Inspector General, U.S. Postal
Service, A Primer on Postal Costing Issues 17–18 (Mar. 20,
2012), https://go.usa.gov/x54Dd (explaining the role of
distribution keys).
The present dispute stems from the uncertainty inherent in
translating this product-by-product breakdown of activity
quantity into a similar breakdown of activity costs, given the
cost savings that accrue as total production volume increases.
If every cost-driver unit were equally costly, the distribution
keys could be used to apportion all an activity’s costs to
specific products: a product responsible for 5% of the cubic-
foot-miles accrued in highway transportation, for example,
could be linked to 5% of that activity’s costs. But not all cost-
driver units are created equal. Under the principle of
diminishing marginal costs, the cost of adding each new unit—
in economic parlance, that unit’s “marginal cost”—decreases
as production quantity increases, due to the efficiency gains
that result from scaling up operations. See Order at 35 (“As a
result of economies of scale and scope, the marginal cost of
individual units of volume . . . decreases with volume.”); see
also Order App’x A at 2 (defining marginal cost). To transport
one cubic foot of mail, for instance, the Postal Service must
make an initial outlay to hire a driver and maintain a truck. But
throwing a second cubic foot of mail onto the truck carries
fewer additional costs, and a third cubic foot carries fewer still.
Given this variability, introducing a new product line that
increases the Postal Service’s total cubic-foot-mileage by 5%
may well increase highway transportation costs by something
less than 5%. Due to diminishing marginal costs, therefore, the
share of cost-driver units a particular product generates might
not determine the share of costs that can be reliably linked to
that product.

-- 6 of 30 --

7
Historically, the Commission dealt with this uncertainty
by directing the Postal Service to attribute to specific products
only that portion of an activity’s costs that would result if every
cost-driver unit cost only as much as the unit with the lowest
marginal cost. Put into agency lingo, the Commission had the
Postal Service attribute only an activity’s “volume-variable
cost[s],” defined as the marginal cost of the “last,” i.e.,
cheapest, cost-driver unit, multiplied by the total number of
units accrued. Order at 36 n.56; see also id. at 9. A Commission
graph, reproduced below as Figure 1, illustrates volume-
variable costs. The downward-sloping curve shows a
hypothetical activity’s diminishing marginal cost (marked on
the vertical axis) as production quantity (marked on the
horizontal axis, and measured in cost-driver units) increases.
The shaded rectangle represents this activity’s volume-variable
costs—the $1 marginal cost of the twentieth cost-driver unit,
applied to all twenty units.
Figure 1
Order App’x A at 15 fig. A-7.
Given that every cost-driver unit contributes an identical
dollar amount to an activity’s volume-variable costs, the Postal
Service, in attributing only these costs, could securely rely on

-- 7 of 30 --

8
its distribution keys and assign each product a share of volume-
variable costs equivalent to that product’s contribution to cost-
driver quantity. For example, consider a truck carrying six
cubic feet of mail—two cubic feet each of letters, postcards,
and parcels—for one mile. Imagine too that the marginal cost
of the first cubic-foot-mile is $60, the marginal cost of the
second is $50, the marginal cost of the third is $40, and so forth.
The activity’s volume-variable costs are $60, or the marginal
cost of the “final” cubic-foot-mile ($10) multiplied by the total
number of cubic-foot-miles (six). Because letters, postcards,
and parcels each account for one-third of the cost-driver units,
volume-variable costs can be apportioned among them in like
manner, with one-third of those costs ($20) attributed to each
product.
As this example shows, the Commission’s historic
approach left some variable costs unattributed to any one
product. Although the volume-variable costs in this example
amount to only $60, total highway transportation costs are $210
($60 plus $50 plus $40 plus $30 plus $20 plus $10). The
remaining $150 left unattributed represents “variable costs that
are not volume-variable costs.” Order at 35. The Commission
calls these “inframarginal costs.” Id. These costs can be
visualized as the white space in Figure 1 that lies between the
downward-sloping marginal cost curve and the shaded
rectangle that represents volume-variable costs. Historically,
the Commission classified all inframarginal costs as
institutional costs, only a limited share of which competitive
products are obliged to cover. See id. at 10; 39 C.F.R.
§ 3015.7(c) (setting competitive products’ minimum collective
share of the Postal Service’s institutional costs at 5.5%).
Dissatisfied with this approach, United Parcel Service, Inc.
(UPS), which runs a parcel delivery service that competes with
the Postal Service’s, petitioned the Commission in 2015 “to

-- 8 of 30 --

9
initiate rulemaking proceedings to change how the United
States Postal Service accounts for the costs of competitive
products.” Petition of United Parcel Service, Inc. for the
Initiation of Proceedings to Make Changes to Postal Service
Costing Methodologies, No. RM2016-2, at 1 (Postal
Regulatory Comm’n Oct. 8, 2015); see also 39 C.F.R.
§ 3050.11(a) (authorizing “any interested person” to submit
such a petition). By classifying inframarginal costs as
institutional costs, UPS argued, the Postal Service had been
shifting “nearly all of the cost savings of [its] economies of
scale and scope” to competitive products, see Proposal One—
A Proposal to Attribute All Variable Costs Caused by
Competitive Products to Competitive Products Using Existing
Distribution Methods, No. RM2016-2, at 15 (Postal Regulatory
Comm’n Oct. 8, 2015), enabling it to “compete unfairly”
against private companies, like UPS, that are unable to offset
their competitive products’ inframarginal costs by wielding
monopoly pricing power elsewhere, id. at 14. Seeking to spur
the Postal Service to increase its competitive products’ prices,
UPS urged the Commission to require that all inframarginal
costs be attributed to specific products. See id. at 1.
UPS proposed a two-step process for performing this
attribution. The Postal Service would first calculate each
production activity’s inframarginal costs, and then apportion
those inframarginal costs among products according to the
distribution keys that show what proportion of cost-driver units
each product generates. See id. at 19–21. By way of example,
recall our mail truck that carries two cubic feet each of letters,
postcards, and parcels, and that incurs $60 in volume-variable
costs and $150 in inframarginal costs. Because the three
products account for equal quantities of cost driver, UPS’s
proposal would attribute inframarginal costs, like volume-
variable costs, equally among them. In this scenario, the
highway transportation costs attributable to each product

-- 9 of 30 --

10
would be $70—one-third of the $60 in volume-variable costs
($20) plus one-third of the $150 in inframarginal costs ($50)—
with no remaining variable costs left to be classified as
institutional. With inframarginal costs thus attributed, the
Postal Service would need to raise competitive products’ rates
to comply with its duty to “ensure that each competitive
product covers its costs attributable,” 39 U.S.C. § 3633(a)(2),
leaving UPS in a stronger market position.
The Commission rejected UPS’s proposal in a September
2016 order, finding that it relied on “unverifiable assumptions”
for both “the calculation and allocation of inframarginal costs.”
Order at 55–56. As to calculation, the Commission explained
that a central assumption underlying UPS’s model for
estimating a production activity’s total inframarginal costs
“lack[ed] an empirical basis.” Id. at 39. As to allocation, the
Commission faulted the proposal’s use of distribution keys to
determine any given product’s share of an activity’s
inframarginal costs, believing that such an approach relied on
the unsupported assumption “that the proportion of
inframarginal costs incurred by that product is identical to the
proportion of the cost driver [generated by] that product.” Id.
at 51. Accordingly, the Commission concluded, UPS’s
proposal was inconsistent with the Accountability Act’s
directive to attribute only those costs that can be linked to a
particular product “through reliably identified causal
relationships.” 39 U.S.C. § 3631(b); see also Order at 56
(“[UPS’s proposal] fails to reliably identify a causal
relationship . . . between all of the inframarginal costs it seeks
to attribute and products.”).
Having rejected UPS’s request that all inframarginal costs
be attributed to individual products, the Commission then
considered whether some such costs could nonetheless be
reliably attributed. In particular, the Commission observed that

-- 10 of 30 --

11
in the course of fulfilling its separate statutory obligation to
“prohibit the subsidization of competitive products by market-
dominant products,” 39 U.S.C. § 3633(a)(1), it had earlier
approved a method that could be used to calculate a
competitive product’s “incremental cost,” defined as “the
difference between the [Postal Service’s] total costs . . . and the
total costs without [that] product,” Order at 58; see also id. at
12–14 (describing the development of the incremental-cost
methodology). Because a product’s incremental cost is the
amount by which total costs would decrease had the cost-driver
units associated with that product never been accrued, it
encompasses not only that product’s share of volume-variable
costs, but also the “inframarginal costs that would be removed”
if the product “were not to be provided.” Order App’x A at 19.
In effect, the incremental-cost methodology attributes to a
product responsible for 5% of an activity’s cost-driver units the
total cost—both volume-variable and inframarginal—of the
“last,” i.e., cheapest 5% of those units.
To illustrate, consider one last time the truck that carries
six cubic feet of assorted mail and incurs $60 and $150 of,
respectively, volume-variable and inframarginal costs. What
happens if the driver removes the two cubic feet of parcels
before the truck sets off? In that case, the truck would carry
only four cubic feet of mail, for a total cost of $180—$60 plus
$50 plus $40 plus $30. The incremental cost of parcels is $30,
or the difference between the $210 in total costs incurred when
parcels are included and the $180 incurred when they are not.
This $30 includes parcels’ one-third share of highway
transportation’s volume-variable costs, or $20, as well as the
$10 in inframarginal costs that would not have been incurred
but for the fifth and sixth cubic feet of mail.
Here, the Commission concluded that because “the portion
of inframarginal costs” included within a product’s incremental

-- 11 of 30 --

12
cost has “a causal relationship” with that product, Order at 55,
the Accountability Act “require[s] the Postal Service to
attribute” it, id. at 61. In December 2016, the Commission
adopted final rules formalizing this requirement. See Changes
to Attributable Costing, 81 Fed. Reg. 88,120, 88,123 (Postal
Regulatory Comm’n Dec. 7, 2016). Those rules define a
competitive product’s “attributable costs” as “the sum of its
volume-variable costs, product-specific costs, and those
inframarginal costs calculated as part of [its] incremental
costs.” Id. (codified at 39 C.F.R. § 3015.7(b)). All other costs,
including all remaining inframarginal costs, remain classified
as institutional.
Figure 2
Petitioner’s Br. 37 (adapting Order App’x A at 18 fig. A-9).
The parties have produced a helpful graphic depiction of
the Commission’s new incremental-cost approach, reproduced
above as Figure 2. The shaded area represents the incremental
cost of a product that is responsible for a share of a hypothetical
activity’s cost-driver units. This area includes not only a
corresponding share of the activity’s volume-variable costs—
the only costs that would have been attributed to the product

-- 12 of 30 --

13
under the Commission’s prior approach—but also the
inframarginal costs associated with the “final,” lowest-priced
share of cost-driver units, which are included among the
product’s costs attributable under the new approach.
The approach the Commission adopted under the 2016
orders is responsive to UPS’s complaint that the historic
approach, by attributing no inframarginal costs, left
unattributed some costs that could be reliably linked to specific
products. It differs from UPS’s proposed approach, however,
in that it attributes to a product responsible for x% of a given
activity’s cost-driver units only those inframarginal costs
associated with the lowest-priced x% of units, rather than, as
UPS would prefer, x% of that activity’s total inframarginal
costs.
Unhappy with its partial victory, UPS petitioned this court
for review of the 2016 orders, arguing that the Commission’s
decision not to require the Postal Service to attribute all
inframarginal costs to specific products was both inconsistent
with the Accountability Act and arbitrary and capricious. See
39 U.S.C. § 3663 (establishing that Commission orders are
reviewed under 5 U.S.C. § 706, which directs courts to set
aside agency action that is “arbitrary, capricious, an abuse of
discretion, or otherwise not in accordance with law,” id.
§ 706(2)(A)). Economist J. Gregory Sidak has filed an amicus
brief supporting UPS, and a quartet of intervenors—Amazon
Fulfillment Services, Inc.; National Association of Letter
Carriers, AFL-CIO; Parcel Shippers Association; and the U.S.
Postal Service—has filed a brief supporting the Commission.
In Part II, we consider whether the challenged orders are,
as UPS claims, contrary to the Accountability Act. In Part III,
we consider UPS’s argument that the orders reflect arbitrary
and capricious decision-making. We are grateful to counsel for

-- 13 of 30 --

14
both sides for their excellent briefs and fine oral argument,
which have helped us considerably.
II.
UPS presses two statutory arguments as to why, in its
view, the challenged orders conflict with the Accountability
Act. We reject both.
A.
UPS first argues that the Commission’s classification of
all inframarginal costs not included in a product’s incremental
cost as “institutional costs,” 39 U.S.C. § 3633(a)(3), is
inconsistent with that term’s unambiguous meaning, see
Chevron, U.S.A., Inc. v. Natural Resources Defense Council,
467 U.S. 837, 842–43 (1984) (“If the intent of Congress is
clear, that is the end of the matter; for the court, as well as the
agency, must give effect to the unambiguously expressed intent
of Congress.”). According to UPS, “institutional costs”
unambiguously refers to “costs, such as overhead and
executive compensation, associated with operating the Postal
Service as an establishment, independent of production,”
Petitioner’s Br. 35, and so excludes all variable costs, including
inframarginal costs.
Even though the Accountability Act nowhere defines
“institutional costs,” it does define the complementary category
of “costs attributable.” 39 U.S.C. § 3631(b). Because UPS
never disputes the Commission’s view that “[a]ll Postal Service
costs are . . . either attributable or institutional,” Order at 9, it
must believe that all variable costs—in its view,
unambiguously excluded from “institutional costs”—are
“attributable” under the statute. But UPS offers no basis for
believing that the Accountability Act unambiguously compels
the Commission to treat each variable cost as a “cost[]
attributable” without first considering whether it possesses the

-- 14 of 30 --

15
statutorily requisite “reliably identified casual relationship[]”
with any one product. 39 U.S.C. § 3631(b).
Instead, UPS hinges its argument on three pieces of
evidence that, it says, establish unambiguously that
“institutional costs” exclude variable costs. First, it cites a
dictionary that defines “institutional” to mean “of, relating to,
involving, or constituting an institution.” Petitioner’s Br. 34–
35 (quoting Webster’s Third New International Dictionary
1171 (2002)). This definition, however, is fully consistent with
classifying some variable costs as institutional. Variable postal
costs, such as the hourly wages of employees who deliver the
mail, “relate to” the Postal Service no less than do fixed postal
costs, such as the Postmaster General’s annual salary. UPS next
cites its own amicus’s statement in a law review article that
“[i]nstitutional costs are fixed overhead and capital costs that
are not volume-sensitive.” Id. at 36 (emphasis omitted)
(quoting J. Gregory Sidak & Daniel F. Spulber, Monopoly and
the Mandate of Canada Post, 14 Yale J. on Reg. 1, 56 (1997)).
But this lone characterization—which itself cites no
authority—falls far short of demonstrating that UPS’s
definition of “institutional costs” is so universally accepted that
Congress must have adopted it. Finally, UPS cites the Act’s
Senate Report, which refers to “salaries for management and
other overhead costs” as examples of “institutional costs.” Id.
at 35 (quoting Senate Report at 9). That Congress intended the
term to include some fixed costs, however, hardly compels the
conclusion that it intended the term to exclude all variable
costs.
With no indication that the statute requires UPS’s reading,
we are left to ask whether the Commission’s own interpretation
is “permissible,” deferring to the agency under Chevron,
U.S.A., Inc. v. Natural Resources Defense Council, 467 U.S.
837 (1984), if it is. U.S. Postal Service, 785 F.3d at 750

-- 15 of 30 --

16
(quoting Chevron, 467 U.S. at 843). As we have explained, the
Commission understands the undefined category of
“institutional costs” to consist of all “residual” Postal Service
costs, fixed or variable, that fall outside the statutory definition
of “attributable” costs. Order at 9–10. In addition to its
consistency with statutory structure, this reading gains support
from the established meaning “institutional costs” held in the
postal ratemaking context long prior to the Act’s 2006
enactment. As early as 1975, the Commission observed that
“the Postal Service considers certain costs . . . to be
attributable” and that “[a]ll other costs are classified as
institutional.” Opinion and Recommended Decision, No. R74-
1, at 99 (Postal Rate Comm’n Aug. 28, 1975). And since then,
the Commission has continued to conceive of “the institutional
cost pool” as “[t]he remaining portion” of total costs after
attributable costs are subtracted. See United Parcel Service,
Inc. v. U.S. Postal Service, 184 F.3d 827, 842 (D.C. Cir. 1999)
(per curiam) (alteration in original) (internal quotation marks
omitted) (quoting Opinion and Recommended Decision, No.
R97-1, at 220 (Postal Rate Comm’n May 11, 1998)).
The Commission’s interpretation of the Accountability
Act in line with this longstanding usage is perfectly reasonable
under Chevron. We typically presume that Congress is “aware
of established practices and authoritative interpretations of the
coordinate branches,” United States v. Wilson, 290 F.3d 347,
357 (D.C. Cir. 2002), and here the Act’s legislative history
confirms that the enacting Congress knew the Commission
took “institutional costs” to mean those costs that “cannot be
attributed to any specific product,” Senate Report at 9. One
could reasonably infer that, in employing a known term of art
in the statute, “Congress intended it to have its established
meaning.” McDermott International, Inc. v. Wilander, 498
U.S. 337, 342 (1991).

-- 16 of 30 --

17
UPS challenges the idea that the meaning the Commission
now assigns to “institutional costs” has such a consistent
pedigree in the postal ratemaking context that the agency could
reasonably construe the Accountability Act to accommodate it.
UPS’s evidence of inconsistency is underwhelming. It first
points to a Postal Service publication stating that institutional
costs “can be considered common costs or overhead costs
needed for overall operations,” but that same publication,
consistent with longstanding practice, defines institutional
costs as those “[p]ostal costs that cannot be directly or
indirectly assigned to any mail class or product,” and then
expressly contrasts such costs with “attributable cost[s].” U.S.
Postal Service, Glossary of Postal Terms 104 (2013),
https://about.usps.com/publications/pub32.pdf. Next, UPS
pulls a sentence from a 2012 Commission order saying that
“institutional costs do not vary with volume.” Order Reviewing
Competitive Products’ Appropriate Share Contribution to
Institutional Costs, No. RM2012-3, at 23 (Postal Regulatory
Comm’n Aug. 23, 2012) (“2012 Order”). But this remark,
which appeared well after the Accountability Act’s passage
and so could not have informed Congress’s meaning, was made
in passing in connection with an issue that had “not [been]
raised by the parties” to that agency proceeding. Id. Even if
UPS is correct that this stray sentence in a single agency order
conflicts with the meaning the Commission has long and
repeatedly assigned “institutional costs,” it hardly follows that
it was unreasonable for the Commission to interpret the
Accountability Act to be consistent with that longstanding
meaning.
Finally, UPS emphasizes that the Commission’s approach
leaves nearly half the Postal Service’s costs in the “institutional
costs” category. True enough, but UPS has failed to show why
reading “institutional costs” to permit this outcome is
unreasonable under the statute. Indeed, in passing the

-- 17 of 30 --

18
Accountability Act, Congress found “no reason for changing”
existing attribution standards, Senate Report at 10, under
which, it recognized, institutional costs made up “40 percent of
the Postal Service’s costs,” id. at 9; see also Newsweek, Inc. v.
U.S. Postal Service, 663 F.2d 1186, 1200 (2d Cir. 1981)
(“There is nothing in the legislative history [of the
Accountability Act’s predecessor statute] to suggest that
attribution of fifty percent of postal costs is inadequate.”).
Given our conclusion that the Commission’s reading of
“institutional costs” is reasonable and so merits our deference,
we need not consider the Commission’s argument that, under
Chevron, its reading is not only permissible, but also
unambiguously correct.
B.
UPS next argues that the Commission’s orders give no
effect to the word “indirect” in the Accountability Act’s
requirement that a product’s “costs attributable” include the
“direct and indirect postal costs attributable to such product
through reliably identified causal relationships.” 39 U.S.C.
§ 3631(b). Contending that “[i]ndirect costs are costs that are
jointly caused by multiple products,” Petitioner’s Br. 39, UPS
argues that because the Commission’s methodology attributes
only those costs that are “caused by providing a specific
product,” Order at 52, that methodology will attribute no
“indirect” costs and so is “not in accordance” with the Act, 5
U.S.C. § 706(2)(A). In addressing this argument, we assume
without deciding that, as UPS contends, the statute’s reference
to “direct and indirect” costs means that the Commission may
lawfully adopt an attribution formula only if the formula
assigns at least some “indirect” costs to specific products.
Even if UPS has correctly interpreted “indirect postal
costs” to mean joint costs, the Commission has reasonably

-- 18 of 30 --

19
concluded that its approach in fact attributes some such costs.
See U.S. Postal Service, 785 F.3d at 750 (where statute’s
meaning is not at issue, court asks whether “the Commission’s
exercise of its authority [was] ‘reasonable and reasonably
explained’” (quoting Manufacturers Railway Co. v. Surface
Transportation Board, 676 F.3d 1094, 1096 (D.C. Cir. 2012))).
Observing that both volume-variable and inframarginal costs
“contain common costs,” Order at 50, defined as “costs that are
shared by multiple products but do not directly vary with any
of those individual products,” id. at 7, the Commission
explained that its new cost-attribution methodology “do[es] not
exclude all common costs, but only those without a reliably
identified causal relationship to [a specific] product,” id. at 52.
For example, the cost of fueling a truck that delivers letters
and parcels may, we think, be viewed as a common cost. It is
“shared by” the two products that contribute to it, but it “do[es]
not directly vary” with either product: the amount by which it
rises (or falls) when mail is added to (or taken from) the truck
is unaffected by whether that mail consists of letters, parcels,
or some combination. Id. at 7. The mere fact that the
Commission is capable of calculating how much the truck’s
fuel costs would decrease in the absence of parcels (or,
importantly, in the absence of an identical volume of letters)
does not change the characteristics that make those fuel costs
“common.” The orders, therefore, lay out an attribution
methodology that the Commission reasonably understands to
be consistent with even UPS’s own view of the statute.
In any event, the Commission does not agree that “indirect
postal costs,” 39 U.S.C. § 3631(b), refers only to joint costs. In
rejecting UPS’s “perceived definition of indirect costs,” Order
at 32, the Commission contemplated that indirect costs can
include the costs of those single-product production activities
“that contain support activities,” id. at 103, and that—in

-- 19 of 30 --

20
consequence—are only “indirectly linked to the volume of the
product that cost was incurred to produce,” Respondent’s Br.
50; see also Intervenors’ Br. 16 (“At least since the mid-1970s,
the term ‘indirect postal costs’ has referred . . . [to] costs that
vary with other costs.”). These costs would include, for
example, the cost of hiring supervisors to oversee employees
who sort parcels: the number of supervisors needed depends on
the number of employees, which in turn depends on the volume
of parcels. See, e.g., Direct Testimony of Joe Alexandrovich on
Behalf of U.S. Postal Service, No. 97-1, at 3 (Postal Rate
Comm’n July 10, 1997) (“Alexandrovich Testimony”); Direct
Testimony of Howard S. Alenier on Behalf of U.S. Postal
Service, No. R80-1, at 6 n.1 (Postal Rate Comm’n Apr. 21,
1980) (“Alenier Testimony”). Supervisors’ wages are thus a
product-specific cost that varies only indirectly with volume—
the sort of cost that the Postal Service calls a “piggyback” cost.
See Postal Regulatory Comm’n, Financial Analysis of United
States Postal Service Financial Results and 10-K Statement 49
(Mar. 31, 2017) (“Financial Analysis”),
https://go.usa.gov/x5kWz (describing a product’s “indirect
cost[s]” as those that are “piggybacked to the direct cost”).
The Commission’s reading of “indirect postal costs” to
include this sort of single-product piggyback cost is reasonable.
Past testimony before the Commission has, after all, repeatedly
confirmed that “indirect costs,” in the specific context of postal
accounting, has long included costs that vary only indirectly
with product volume due to the presence of an intermediate
factor. See, e.g., Alexandrovich Testimony at 3 (“Direct and
indirect variable costs are terms distinguishing whether or not
there is at least one intervening link between cost and
volume.”); Alenier Testimony at 6 (“The terms direct and
indirect [cost] indicate whether or not at least one intermediate
element links cost to volume.” (footnote omitted)).

-- 20 of 30 --

21
UPS argues that the statute forecloses the Commission’s
reading, but here too its evidence is insufficient. It first notes
the Supreme Court’s observation that a study upon which
Congress relied in enacting a predecessor statute defined
indirect costs as “[t]hose elements of cost which cannot
unequivocally be associated with a particular output or
product.” National Ass’n of Greeting Card Publishers v. U.S
Postal Service (“NAGCP”), 462 U.S. 810, 827 n.21 (1983)
(alteration in original) (internal quotation marks omitted). This
definition, however, is not necessarily inconsistent with the
Commission’s view that indirect costs include those that are
associated only indirectly with product volume, or “output,”
even if they can be associated with a single “product.” Id. Next,
UPS cites a federal accounting regulation, which actually
supports the Commission’s reading because it defines
“[i]ndirect cost” as a cost “identified with two or more final
cost objectives or with at least one intermediate cost
objective,” 48 C.F.R. § 9904.418-30(a)(3) (emphasis added),
such as the employees who report to the supervisor in the
example above. Finally, UPS cites an accounting textbook that
defines indirect costs as those “incurred in providing benefits
to several different cost objects.” Petitioner’s Br. 40 (emphasis
omitted). UPS, however, never explains why single-product
costs that support intermediate cost objects (such as
subordinate employees) as well as final cost objects (i.e., end
product) fall outside this definition.
Put simply, UPS has failed to show that the Accountability
Act unambiguously compels a reading of “indirect postal
costs” that includes only those costs that are shared across
products. Under Chevron, we therefore defer to the
Commission’s reasonable view that the term can include those
single-product costs that vary indirectly with volume.

-- 21 of 30 --

22
C.
UPS argues that even if the Commission’s interpretations
of “institutional costs” or “indirect postal costs” are
permissible, Chevron deference is inappropriate because the
Commission made “no ‘reasonable attempt to grapple’ with or
even refer back to the statutory text.” Petitioner’s Br. 45
(quoting BP Energy Co. v. FERC, 828 F.3d 959, 965 (D.C. Cir.
2016)). In our view, the Commission had no need to say
anything more. “Institutional costs” and “indirect postal costs”
have established meanings in the postal ratemaking context,
see supra at 14–21, and the orders are faithful to these
meanings, see Order at 10 (describing “institutional costs” as
“residual costs”); id. at 103 (explaining that “the piggyback
method” applies to “cost components that contain support
activities”). To be sure, “no amount of historical consistency
can transmute an unreasoned statutory interpretation into a
reasoned one.” Southeast Alabama Medical Center v. Sebelius,
572 F.3d 912, 920 (D.C. Cir. 2009). Here, however, the
longstanding definitions upon which the Commission relied
create no anomalies and flow sensibly from text, history, and
statutory structure. Cf. id. at 919–20 (requiring agency to
explain its historical view that a hospital’s “wage-related costs”
include postage costs where, even in litigation, the agency
offered but “one somewhat opaque rationale” that was itself
apparently inconsistent with the agency’s treatment of certain
other costs). The Commission therefore had no duty to
expressly justify its decision to continue embracing them. See
Hall v. McLaughlin, 864 F.2d 868, 872 (D.C. Cir. 1989)
(“Where the reviewing court can ascertain that the agency has
not in fact diverged from past decisions, the need for a
comprehensive and explicit statement of its current rationale is
less pressing.”).
UPS believes that this rationale cannot save the orders’
treatment of either “institutional costs” or “indirect postal

-- 22 of 30 --

23
costs.” With respect to “institutional costs,” UPS argues that
the interpretation reflected in the orders represents an
unexplained deviation from the Commission’s prior reading of
the term. See Encino Motorcars, LLC v. Navarro, 136 S. Ct.
2117, 2125 (2016) (agency must provide “a reasoned
explanation” for a change in policy position). By including
some variable costs among institutional costs, UPS argues, the
Commission has departed from its previous statements that
“institutional costs do not vary with volume,” 2012 Order at
23, and that “variability with volume should be sufficient to
establish causality” and thus attribution, Appendices to
Opinion and Recommended Decision, App’x B, No. R80-1, at
26 (Postal Rate Comm’n Feb. 19, 1981). Notwithstanding these
sentences, cherry-picked and shorn of context, the Commission
has never taken the view that all variable costs, including all
inframarginal costs, bear an adequate causal relationship with
specific products to be counted among costs attributable or—
what amounts to the same thing—that no variable costs may be
considered institutional costs. Indeed, it was the Commission’s
previous classification of all inframarginal costs—which UPS
accepts are variable costs, see Petitioner’s Br. 44—as
institutional that prompted UPS to petition the Commission in
the first place.
With respect to “indirect postal costs,” UPS argues that the
orders failed to make clear what meaning the Commission
assigned to the term because, as the Commission
acknowledges, they “‘declined’ to pass on whether ‘indirect
costs’ include joint costs.” Reply Br. 13 (quoting Respondent’s
Br. 53). But the Commission had no need to opine on whether
“indirect postal costs” include joint costs in addition to single-
product costs that vary indirectly with product volume:
recognizing that the term includes at least the latter, as the
Commission has consistently done, was sufficient to defeat
UPS’s argument that no indirect costs would be attributed

-- 23 of 30 --

24
under the Commission’s newly adopted cost-attribution
scheme.
III.
This brings us to UPS’s argument that the challenged
orders are “arbitrary, capricious, [or] an abuse of discretion.” 5
U.S.C. § 706(2)(A). In considering this argument, we
emphasize our “‘reluctan[ce] to interfere with [an] agency’s
reasoned judgments’ about technical questions within its area
of expertise.” Alliance of Nonprofit Mailers v. Postal
Regulatory Comm’n, 790 F.3d 186, 197 (D.C. Cir. 2015)
(second alteration in original) (quoting NRG Power Marketing,
LLC v. FERC, 718 F.3d 947, 953 (D.C. Cir. 2013)). As the
Supreme Court has recognized, Congress vested postal
ratemaking authority in the Commission out of a desire to
harness “the educated and politically insulated discretion of
experts.” NAGCP, 462 U.S. at 823. Consequently, Congress
has not “dictate[d] or exclude[d] the use of any method of
attributing costs,” id. at 820, leaving it to “the expert ratesetting
agency, exercising its reasonable judgment . . . to decide which
methods sufficiently identify the requisite causal connection
between particular services and particular costs,” id. at 827. In
considering whether the orders suffer from arbitrary and
capricious decision-making, then, we ask only whether “the
Commission’s exercise of its authority [was] ‘reasonable and
reasonably explained.’” U.S. Postal Service, 785 F.3d at 750
(quoting Manufacturers Railway Co., 676 F.3d at 1096).
A.
UPS first argues that the Commission failed to “reasonably
explain[]” the adoption of its incremental-cost methodology.
Id. (quoting Manufacturers Railway Co., 676 F.3d at 1096).
Given that we have already considered and rejected UPS’s
claim that the Commission inadequately explained its statutory

-- 24 of 30 --

25
interpretation, see supra at 22–24, UPS is left with its argument
that the Commission failed to explain how its chosen approach
“serve[d] the [Accountability Act’s] objectives,” Northpoint
Technology, Ltd. v. FCC, 412 F.3d 145, 151 (D.C. Cir. 2005).
We find the Commission’s explanation perfectly adequate.
Though recognizing that the Accountability Act was
“intended to ensure that the Postal Service competes fairly in
the provision of competitive products,” Order at 121 (quoting
Senate Report at 19), the Commission rejected UPS’s
complaint that attributing only incremental costs fails to fulfill
this goal, see id. at 57. In the Commission’s view, “[t]he
purpose of the incremental cost test is not to ensure that the
Postal Service is competing fairly,” but rather, as used here, to
“ensure that products cover all of the costs the Postal Service
incurs in providing them,” which in turn plays but a
contributing role in the statute’s overall pro-competitive aims.
Id. at 58.
The Commission properly recognized that its role is to
carry out the particulars of the scheme Congress created, not to
engineer specific market outcomes. The Supreme Court, while
acknowledging that “Congress’ concern about . . . cross-
subsidies, of course, was one motive for including [a] rate
floor” in a predecessor statute, observed that Congress also
took care to provide that cost attribution be methodologically
sound. NAGCP, 462 U.S. at 829 n.24. Indeed, before the
Commission, UPS itself recognized that “[t]he relevant
inquiry” in selecting a cost attribution approach “is whether the
Postal Service’s . . . practices comply with [the Accountability
Act],” Reply Comments of United Parcel Service, Inc.
Regarding UPS Proposals One and Two, No. RM2016-2, at 33
(Postal Regulatory Comm’n Mar. 25, 2016), not the approach’s
effects on “market conditions,” id. at 36. This is correct.
“Congress,” as UPS explained, “did not direct the Commission

-- 25 of 30 --

26
to consider prevailing market conditions in connection with”
cost attribution. Id. at 37. In any event, despite its fear that
leaving some inframarginal costs unattributed “might allow the
Postal Service to monopolize otherwise competitive markets,”
Petitioner’s Br. 49, UPS offers no reason to doubt that the
Accountability Act’s prohibition on cross-subsidization, 39
U.S.C. § 3633(a)(1), and requirement that competitive
products cover a share of institutional costs, id. § 3633(a)(3),
will adequately ameliorate any competitive deficit left by the
Commission’s approach to cost attribution, cf. id. § 3633(b)
(requiring Commission to consider “the prevailing competitive
conditions in the market” when setting competitive products’
share of institutional costs); Competitive Postal Products, 83
Fed. Reg. 6758, 6774 (Postal Regulatory Comm’n Feb. 14,
2018) (proposing a new, formula-based method for
determining competitive products’ share of institutional costs
that will be “more responsive to changing market conditions”).
B.
Next, UPS argues that the Commission’s adoption of an
incremental-cost approach to attribution was itself arbitrary
and capricious, insisting that this approach suffers from the
very same features that led the Commission to reject UPS’s
proposal that all inframarginal costs be attributed. See U.S.
Postal Service, 785 F.3d at 753 (“The agency fails to
reasonably explain its decision if it gives ‘differential treatment
of seemingly like cases.’” (quoting LePage’s 2000, Inc. v.
Postal Regulatory Comm’n, 642 F.3d 225, 232 (D.C. Cir.
2011))). In support, it offers three arguments, none persuasive.
It begins by challenging the Commission’s rejection of the
assumption that any given product is just as likely to be
responsible for “early,” more expensive cost-driver units as it
is for “later,” less expensive units. This assumption, which the
Commission deemed “empirically unverifiable,” Order at 46,

-- 26 of 30 --

27
underlay UPS’s proposal to distribute inframarginal costs
among products according to each product’s contribution to
cost-driver quantity. The Commission’s rejection of this
assumption was arbitrary, UPS argues, because the
Commission’s own incremental-cost approach was “based on
[the] even more unverifiable assumption” that all products use
only the latest, lowest-priced cost-driver units and so bear the
minimum possible inframarginal costs. Petitioner’s Br. 53.
UPS misunderstands the Commission’s statutory task,
namely to attribute only those costs that can be linked to a
product “through reliably identified causal relationships.” 39
U.S.C. § 3631(b). Contrary to UPS’s claim that the
Commission untenably assumed that “every product comes
last” in the production chain, Petitioner’s Br. 53, the
Commission simply declined to assume that any given product
incurred more than the minimum cost that could reliably be
assigned to it. Attributing more than this amount, after all,
necessitates guesswork, and the Commission sensibly
concluded that such guesswork was inconsistent with its
statutory obligation to base attribution on only “reliably
identified causal relationships.” 39 U.S.C. § 3631(b).
UPS’s second argument—that the Commission acted
arbitrarily in rejecting distribution keys as a means of
apportioning inframarginal costs—fails for the same reason. As
the Commission saw it, attributing inframarginal costs on the
basis of distribution keys, which measure only the share of
cost-driver units for which a given product is responsible,
would rely on the “unverifiable assumption that the proportion
of inframarginal costs incurred by [a] product is identical to the
proportion” of cost-driver units generated by that product.
Order at 51. Here, too, the Commission reasonably declined to
make this assumption absent supporting evidence. Nor,
contrary to UPS’s argument, did the Commission act

-- 27 of 30 --

28
inconsistently by using distribution keys as part of its
incremental-cost approach. After all, the Commission uses
them only for their intended purpose—to determine how many
of any given activity’s cost-driver units derive from any one
product. From there, the Commission calculates “the marginal
cost of providing [each of these] specific unit[s]” without
further recourse to the distribution keys. Id. at 50.
The third and final argument takes aim at the
Commission’s rejection of UPS’s proposed method for
estimating an activity’s total inframarginal costs in the first
place, prior to any question of attribution. To arrive at an
estimate of an activity’s total costs, the Postal Service would
have to estimate the cost of each cost-driver unit, even the
earliest, most costly units associated with low levels of volume.
As the Commission explained, though, “[a] real-world multi-
product firm does not have the information necessary” to
estimate costs at such volume levels because “it has not
experienced” them. Id. at 8. UPS proposed to get around this
problem by employing an assumption of constant elasticity—
i.e., an assumption that each added unit of product quantity
corresponds to an equal decrease in marginal cost—that would
allow the Postal Service to extrapolate backwards from present,
observed volumes. Id. at 38; see also Order App’x A at 4–5
(explaining constant elasticity). The Commission rejected this
approach as untenable because “[a]pplying the constant
elasticity assumption to levels of volume far beyond the range
of actual experience produces results that are inadequately
supported and unreliable.” Order at 39.
Here, too, UPS responds that the Commission itself relies
on a constant-elasticity assumption when extrapolating
backward from present values to estimate a product’s
incremental cost. The Commission, however, explained that
the incremental-cost test “avoids the issues facing UPS’s

-- 28 of 30 --

29
proposed method by restricting itself to limited amounts of
volume” and by “estimat[ing] inframarginal costs in a very
small range of [an activity’s] cost curve where the constant
elasticity assumption has been empirically verified.” Id. at 42.
UPS challenges the Commission’s claim that it applies the
assumption over only a small range of volumes. As the
Commission points out, however, competitive products make
up a small fraction of the Service’s total business, see Financial
Analysis at 92, making it reasonable to believe that any one
competitive product represents a comparatively small range of
any given activity’s marginal cost curve. UPS further argues
that, even within that range, the Commission’s empirical
justification for the constant-elasticity assumption rests on a
20-year-old article that predated “significant changes in the
Postal Service’s competitive parcel business in the 21st
century.” Petitioner’s Br. 61. But UPS has identified no
substantive deficiency in the article or any way intervening
events have undermined its conclusions. At any rate, it was
hardly arbitrary for the Commission to find the constant-
elasticity assumption sufficiently reliable to make a limited
extrapolation from present conditions but insufficiently
reliable to estimate cost at all levels of production volume.
Alternatively, UPS suggests that the Commission could
calculate inframarginal costs by simply subtracting one known
quantity—an activity’s volume-variable costs—from
another—that activity’s total costs. The Commission, however,
reasonably concluded that this method of calculation would
“result[] in an overstatement of the inframarginal costs of that”
activity because it disregards the fact that fixed costs—neither
volume-variable nor inframarginal—comprise part of an
activity’s total costs. Order at 39. UPS responds that “the Postal
Service can subtract those fixed costs from the [activity’s] total
costs.” Petitioner’s Br. 62. But the Postal Service informed the
Commission that it does not currently “determine which of its

-- 29 of 30 --

30
costs are fixed,” and that doing so would require it to “answer[]
difficult counterfactual questions” about “which costs would
remain if the Postal Service handled no volume.” Postal
Service Responses at 11 & n.9. UPS has proposed no reliable
means of calculating fixed costs, merely claiming without
support that additional data from the Postal Service would, if
made available, suggest a way forward.
IV.
The Accountability Act requires a competitive product to
cover only those costs that can be attributed to the product
“through reliably identified causal relationships.” 39 U.S.C.
§ 3631(b). In establishing this causal requirement, 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.” NAGCP, 462 U.S. at
827. Here, the Commission did exactly that, settling on a cost-
attribution methodology that implements its statutory mandate
and falls well within the scope of its considerable discretion.
Because “the Commission’s exercise of its authority [was]
‘reasonable and reasonably explained,’” U.S. Postal Service,
785 F.3d at 750 (quoting Manufacturers Railway Co., 676 F.3d
at 1096), we deny UPS’s petitions for review.
So ordered.

-- 30 of 30 --