United States Court of Appeals
FOR THE DISTRICT OF COLUMBIA CIRCUIT
Argued April 7, 2009 Decided June 9, 2009
No. 07-1369
CSX TRANSPORTATION, INC., ET AL.,
PETITIONERS
v.
SURFACE TRANSPORTATION BOARD AND UNITED STATES OF
AMERICA,
RESPONDENTS
AMERICAN CHEMISTRY COUNCIL, ET AL.,
INTERVENORS
Consolidated with 07-1370, 07-1371, 07-1372, 07-1410,
08-1194
On Petitions for Review of an Order
of the Surface Transportation Board
G. Paul Moates argued the cause for Railroad Petitioners.
With him on the briefs were Paul A. Hemmersbaugh, Peter J.
Shudtz, Paul R. Hitchcock, Louis P. Warchot, George A.
Aspatore, J Michael Hemmer, Louise A. Rinn, Samuel M. Sipe
Jr., Anthony J. LaRocca, Terence M. Hynes, and Michael L.
Rosenthal.
2
Nicholas J. DiMichael argued the cause for petitioners
The National Industrial Transportation League, et al. With
him on the briefs were Jeffrey O. Moreno, Andrew P.
Goldstein, and John M. Cutler, Jr.
Raymond A. Atkins, Associate General Counsel, Surface
Transportation Board, argued the cause for respondents. With
him on the brief were Robert B. Nicholson and John P. Fonte,
Attorneys, U.S. Department of Justice, and Ellen D. Hanson,
General Counsel, Surface Transportation Board. Anika S.
Cooper, Attorney, entered an appearance.
G. Paul Moates, Paul A. Hemmersbaugh, Peter J. Shudtz,
Paul R. Hitchcock, Louis P. Warchot, George A. Aspartore, J.
Michael Hemmer, Louise A. Rinn, Samuel M. Sipe, Jr.,
Anthony J. LaRocca, Terence M. Hynes, and Michael L.
Rosenthal were on the brief of Railroad Intervenors.
Nicholas J. DiMichael, Jeffrey O. Moreno, Andrew P.
Goldstein, and John M. Cutler, Jr. were on the brief for
Shipper Intervenors.
Before: ROGERS, TATEL, and GRIFFITH, Circuit Judges.
Opinion for the Court filed by Circuit Judge TATEL.
TATEL, Circuit Judge: In this case we consider a set of
challenges to a Surface Transportation Board regulation
establishing a simplified method for resolving rail rate
disputes too small to bring under ordinary procedures. The
Board’s new regulation gives shippers—the complainants in
rail rate disputes—a choice between using the usual
procedures or either of two cheaper and simpler “small
claims” alternatives better suited to uncomplicated cases.
3
Under each alternative, relief is capped due to the method’s
lower accuracy. A group of railroads challenges the Board’s
adoption of one of the alternative methods, and a group of
shippers challenges the other, as well as the relief caps on
both. Finding the Board’s balancing of the competing
interests in accuracy and simplicity well within its statutory
authority and neither arbitrary nor capricious, we deny the
petitions for review in all respects.
I.
The Surface Transportation Board regulates the rates
railroads charge shippers over which they have market
dominance. 49 U.S.C. §§ 10501, 10701(d)(1). Because such
captive shippers are unable to fend for themselves in the
market, Congress allows them to challenge a rail rate as
unjust or unreasonable before the Board, id. § 10707(b)–(c),
which can impose retrospective relief in the form of
reparations, id. § 11704(b), and prospective relief by
prescribing a new rate, id. § 10704(a)(1). To understand this
case, a brief whistle-stop tour through the history of the
Board’s procedures for resolving these rate disputes is in
order. All aboard!
In 1985, the Interstate Commerce Commission, the
Board’s predecessor, decided to resolve rate disputes under
“constrained market pricing” (CMP) principles, under which
the Commission would find reasonable a rate that (1) reflects
the amount a captive shipper would have to pay to receive
efficient service, (2) affords the railroad adequate revenues,
and (3) does so without cross-subsidizing any service or
facility from which the shipper receives no benefit. Coal Rate
Guidelines, Nationwide, 1 I.C.C.2d 520, 523–24 (1985), aff’d
sub nom. Consol. Rail Corp. v. United States, 812 F.2d 1444
(3d Cir. 1987). Under these principles, shippers able to
demonstrate that the railroad has market dominance had the
4
choice of one of several methods to prove that the challenged
rates were unreasonable. They could opt to examine the
railroad’s entire network for revenue adequacy or
management efficiency, or alternatively, they could choose to
examine only a subset of the network using the “stand-alone
cost” (SAC) test—the choice of most shippers.
A SAC presentation simulates a “stand-alone railroad,” a
fully efficient hypothetical competitor railroad that serves the
complaining shipper and other traffic sharing common
facilities. BNSF Ry. Co. v. STB (“BNSF I”), 453 F.3d 473,
477 (D.C. Cir. 2006). A challenged rail rate is unreasonable
to the extent it exceeds the costs (including a reasonable
profit) of running the stand-alone railroad. Id. A SAC
presentation thus furthers CMP principles by promoting
efficiency and eliminating cross-subsidization. It
accomplishes the former by forcing the railroad to bear the
cost of any inefficiencies, and the latter by preventing the
shipper from paying for any facilities from which it receives
no benefit. Due largely to the difficulty of modeling an
efficient stand-alone railroad, however, this process is both
expensive and time-consuming—each “full SAC” case can
cost a shipper up to $5 million to litigate. Simplified
Standards for Rail Rate Cases (“Decision”), STB Ex Parte
No. 646 (Sub-No. 1), at 31 (served Sept. 5, 2007). In fact,
coal companies are virtually the only shippers who deliver
sufficiently large loads along fixed routes to justify using full
SAC procedures. See Rate Guidelines—Non-Coal
Proceedings (“1996 Guidelines”), 1 S.T.B. 1004, 1008 n.7
(1996) (noting “prevalence” of coal rate challenges).
Recognizing the expense of full SAC cases, the
Commission soon began searching for a simplified
alternative. Throughout the 1980s and early 1990s, it
considered but ultimately discarded several alternatives. One
5
proposal, intended to create a simplified SAC procedure,
came in the form of a computerized model from the
Association of American Railroads (AAR). Because the
AAR refused to provide the proprietary source code for its
computer program (known as AAR-SSAC), the Commission
ran sample cases through the program to test it. AAR-
SSAC’s days were numbered when it labeled reasonable a
rate set at 5000 percent of the railroad’s variable costs.
By 1995, when Congress replaced the Commission with
the Board, the Commission still had not settled on a
simplified alternative. As a result, when Congress passed the
ICC Termination Act of 1995, it gave the newly-created
Board a year to “establish a simplified and expedited method
for determining the reasonableness of challenged rail rates in
those cases in which a full stand-alone cost presentation is too
costly, given the value of the case.” Pub. L. No. 104-88, §
102(a), 109 Stat. 803, 810 (1995) (codified as amended at §
10701(d)(3)). Responding to this directive, the Board issued
a set of simplified guidelines, which rejected AAR-SSAC and
introduced a “three benchmark” system, 1996 Guidelines, 1
S.T.B. at 1041, whereby the reasonableness of a challenged
rate was assessed not by simulating any alternative railroad,
but simply—at least as “the starting point for a rate
reasonableness analysis”—by comparing it to similar existing
rates, id. at 1022. When this approach went unused for years,
the Board held hearings to find out why. During those
proceedings, shippers testified that the three benchmark
guidelines were too vague and that the question of whether a
case was even eligible for resolution under the three
benchmark system was so uncertain as to require litigation.
Simplified Standards for Rail Rate Cases (“NPRM”), STB Ex
Parte No. 646 (Sub-No. 1), at 3 (served July 28, 2006) (notice
of proposed rulemaking).
6
In 2006, the Board issued a notice of proposed
rulemaking, proposing (1) the retention of a slightly modified
three benchmark system for the smallest cases, (2) the
creation of a simplified SAC procedure more complicated
than the three benchmark system but simpler than full SAC,
for use in medium-size cases, and (3) clear eligibility
thresholds for each procedure. Id. After reviewing comments
submitted by railroads and shippers, the Board in 2007 issued
its final rule—the rule challenged here—which gave the
shippers the choice of a modified three benchmark system
intended for the smallest cases, a new simplified SAC
procedure intended for medium-size cases, or full SAC.
Decision at 5–6. Absent from the final rule are the eligibility
thresholds and with them the prospect of litigation over which
method to use. Instead, the new rule allows each shipper to
elect the three benchmark method, simplified SAC, or full
SAC for any case. To channel larger cases to the more
accurate methods, the rule limits the relief available to $1
million over five years for a three benchmark case and $5
million over five years for a simplified SAC case. Id. at 5;
see also id. at 27–28. This limit applies to whatever
combination of retrospective and prospective relief the Board
imposes. Id. at 28.
The three benchmark system compares the challenged
rate to three benchmark figures, each expressed as a
relationship between revenues and variable costs, id. at 10,
i.e., “those costs that increase as traffic over the railroad
increases,” BNSF Ry. Co. v. STB (“BNSF II”), 526 F.3d 770,
773 (D.C. Cir. 2008). Calculating one of these benchmarks
involves comparing the rail movement at issue with a group
of similar movements. The Board selects a comparison group
from groups proposed by the parties, who choose the
comparison movements from the four most recent years of
data in the “Waybill Sample.” Decision at 18. That sample,
7
compiled by the Board, is a survey of information from rail
movements across the nation. Id. at 78.
The Board’s simplified SAC procedure is similar to the
full SAC method, but with a crucial difference. In a full SAC
presentation, the stand-alone railroad is hypothetical and fully
efficient. In a simplified SAC presentation, the stand-alone
railroad is instead a portion of the actual railroad with limited
modifications not relevant here. Id. at 15–16.
This case involves two sets of challenges to the Board’s
rule. A group of shippers argues that the Board set the relief
caps too low and adopted simplified SAC with neither
justification nor testing. Several railroads challenge the
Board’s adoption of the three benchmark method, claiming
that the Board sanctioned the use of stale data and barred
railroads from presenting certain types of evidence in those
cases. All railroad petitioners intervene to oppose the
shippers’ petition, and most shipper petitioners intervene in
opposition to the railroads’ petition. We review the Board’s
orders using the Administrative Procedure Act’s standards,
under which we will set aside agency action that is
“‘arbitrary, capricious, an abuse of discretion, or otherwise
not in accordance with law.’” BNSF II, 526 F.3d at 774
(quoting 5 U.S.C. § 706(2)(A)). “In the rate-making area, our
review is particularly deferential, as the Board is the expert
body Congress has designated to weigh the many factors at
issue when assessing whether a rate is just and reasonable.”
Id.
II.
We begin with the shippers’ claims. They argue that the
Board acted arbitrarily in setting the relief cap levels and in
adopting simplified SAC.
8
Relief Caps
The shippers bring an intriguing but ultimately unavailing
challenge to the Board’s decision to set the relief limits at $1
million under the three benchmark method and $5 million
under simplified SAC. Although embracing the overall
approach of setting relief caps, the shippers claim that the
Board failed to make the findings necessary to ensure it
complied with section 10701(d)(3)’s requirement that it
“establish a simplified and expedited method for determining
the reasonableness of challenged rail rates in those cases in
which a full stand-alone cost presentation is too costly, given
the value of the case.” According to the shippers, the Board
failed to find that in those cases in which full SAC is too
costly, the relief caps still allow simplified SAC to generate
reasonable rates. Similarly, they claim that the Board failed
to assure that in those cases in which simplified SAC is too
costly, the relief caps still allow the three benchmark method
to generate reasonable rates.
The shippers start from the premise that at some point a
limit on relief might be so low as to produce an unreasonable
rate, either by making it infeasible to bring a case or by falling
too far below the rate to which the shipper would otherwise
be entitled. This premise follows from the shippers’ belief
that any rate the Board prescribes as relief “must be
reasonable” under section 10701(d)(1). See Shippers’
Opening Br. 12. If unduly low relief caps produce an
unreasonable rate in a case in which full SAC is too costly,
the shippers continue, then the shipper is left without a
meaningful way to get a reasonable rate. If too many cases
fall into this category—that is, if too many cases in which full
SAC is too costly are also cases in which simplified SAC fails
to produce a reasonable rate—then, they conclude, simplified
SAC would not constitute “a simplified and expedited method
for determining the reasonableness of challenged rail rates in
9
those cases in which a full stand-alone cost presentation is too
costly, given the value of the case.” § 10701(d)(3).
Thus, according to the shippers, the Board should have
identified that subset of cases in which full SAC is too costly
and ensured that in such cases simplified SAC produces
reasonable rates. They take no issue with the Board’s
estimate that full SAC cases cost $5 million to litigate. But
according to them, the Board should have determined the
minimum permissible potential recovery ratio—i.e., the ratio
of available relief to litigation cost, a ratio the shippers
confusingly call a “risk factor”—below which full SAC
becomes too costly, and then ensured that for cases falling
under the threshold produced by that ratio, simplified SAC
provides enough relief to produce a reasonable rate.
Shippers’ Opening Br. 20–21. For example, suppose the
Board had picked a potential recovery ratio of 3.0.
Multiplying that ratio by full SAC’s $5 million litigation cost
would indicate that cases with anticipated relief of $15
million or less are those “in which a full [SAC] presentation is
too costly, given the value of the case,” § 10701(d)(3). After
picking that ratio the Board would then have to set the
simplified SAC relief caps high enough that, in cases with
anticipated relief of less than $15 million, the rate produced
by simplified SAC would still be reasonable. The shippers
insist that the Board did none of this, but we think they ask
too much.
Section 10701(d)(1) requires that a rate “must be
reasonable” only if established by a rail carrier with market
dominance, not if prescribed by the Board. § 10701(d)(1) (“If
. . . a rail carrier has market dominance over the transportation
to which a particular rate applies, the rate established by such
carrier for such transportation must be reasonable.”
(emphasis added)). Therefore, contrary to what the shippers
10
believe, unlike railroad-set rates, the rates the Board
prescribes as relief in simplified SAC cases, though subject to
different requirements, need not themselves be “reasonable”
within the meaning of section 10701(d)(1). Compare §
10704(a)(2) (requiring revenue adequacy for rates prescribed
by the Board) with § 10701(d)(2) (requiring revenue adequacy
and setting out criteria for the Board to consider in
“determining whether a rate established by a rail carrier is
reasonable for purposes of [section 10701(d)(1)]” (emphasis
added)). That said, section 10701(d)(3)’s requirement of a
“simplified and expedited method for determining the
reasonableness of challenged rail rates in those cases in which
a full stand-alone cost presentation is too costly, given the
value of the case,” clearly contemplates a method that may
substitute for a full SAC proceeding in low-value cases—that
is, a method for determining the reasonableness of rail rates
not in the abstract, but for the purpose of awarding some relief
to shippers. Thus, section 10701(d)(3) requires the
“simplified and expedited method” to function as a
meaningfully effective way to seek some degree of redress for
unreasonable rail rates, and so excessively stingy relief caps
could in theory render a method ineffective. In that sense,
then, the shippers are correct: the Board was obliged to
determine that the relief caps were sufficiently high to satisfy
section 10701(d)(3).
Although not using the methodology the shippers urge,
the Board did just that: it clearly recognized its section
10701(d)(3) obligation and made the necessary findings. It
made clear that it understood the shippers’ precise concerns,
describing them as complaining of a “Hobson’s choice” for
certain case values and focusing on a hypothetical where a
shipper who, pursuing relief under the simpler method, would
“relinquish over half the value of its case,” while under the
more complex method would stand to make only twice the
11
litigation costs. Simplified Standards for Rail Rate Cases
(“Rehearing Decision”), STB Ex Parte No. 646 (Sub-No. 1),
at 7 (served Mar. 19, 2008). Though the precise example the
Board mentioned compared the three benchmark method to
simplified SAC, rather than simplified SAC to full SAC, it
discussed both cases together, and its reasoning applies
equally to the comparison between simplified SAC and full
SAC.
After correctly identifying the shippers’ concerns, the
Board addressed them. It began by stating that cases which
might net the same relief have different prospects for success
and explained the desirability of encouraging a shipper who
was “more confident of its prospects for obtaining greater
relief” to use the more precise (and more costly) methods. Id.
at 8. Next, the Board stated that according to the table of case
values and net relief submitted by the shippers, “the $1
million and $5 million limits provide every shipper with a
potential case with sufficient net relief after litigation costs to
justify bringing a complaint under Three-Benchmark or
Simplified-SAC method[s].” Id. The Board accordingly
found that “every complainant will have a vehicle to pursue
its complaint regardless of the value of the case.” Id. That is,
the Board found that shippers subject to the relief caps retain
a sufficient amount of relief, even after the cost of litigation,
to make it feasible to bring their cases.
The shippers insist that a too-low relief cap could violate
section 10701(d)(3) in another way: even if not too close to
the litigation cost of the case, a relief cap might be too far
below the actual amount to which the shipper is entitled, thus
requiring the shipper to forgo such an unreasonably large
amount of relief as to prevent simplified SAC from serving as
an effective “simplified and expedited method.” Perhaps so,
but the Board addressed this possibility. It acknowledged that
12
wherever relief caps might be set, some shippers would face a
difficult choice due to the effect of the cap. Id. The Board
then concluded:
Ultimately, we do not think it is improper for
there to be some trade-off involved in using a
simpler, faster, and less costly method that is
inherently less precise. We believe the limits
we have set strike the appropriate balance so
that we do not open the door to excessive
litigation under methods that are not justified
for the amount at dispute.
Id. This discussion clearly represents the Board’s assessment
that the “trade-off” does not require shippers to forgo too
much relief in order to get the benefit of a simpler proceeding.
The Board thus fully responded to the shippers’ concern.
To be sure, the Board’s analysis was qualitative instead
of quantitative, but it rested on the Board’s expertise, as well
as an assessment of the relief cap levels. Not every problem
is appropriate for qualitative analysis, but this one is: the
interest in channeling larger disputes into more accurate
forums is “inherently incommensurable” with the interest in
giving shippers meaningful access to a simpler forum, such
that there is no way to balance the two without making a
“judgment call,” BNSF II, 526 F.3d at 776 (upholding the
Board’s refusal to adopt certain rate adjustments). Even had
the Board explicitly considered a wider range of cases, at the
end of the day it would still have had to make a policy
judgment as to when full SAC is “too costly” and when a
relief cap renders the rate produced by simplified SAC
unreasonable. Cf. FCC v. Fox Television Stations, Inc., 129
S. Ct. 1800, 1813 (2009) (“It is one thing to set aside agency
action under the Administrative Procedure Act because of
13
failure to adduce empirical data that can readily be obtained.
It is something else to insist upon obtaining the unobtainable.”
(citation omitted)). Given this reality, we have no difficulty
concluding that the Board explained itself adequately for us to
“reasonably discern” its path as to the relief caps under both
the three benchmark method and simplified SAC. ACS of
Anchorage, Inc. v. FCC, 290 F.3d 403, 408 (D.C. Cir. 2002).
Hinting at an additional challenge to the Board’s rule, the
shippers argue in a single sentence in their opening brief that
the relief caps are in fact too low to comply with section
10701(d)(3) or the general requirement that rates charged to
captive shippers “must be reasonable,” § 10701(d)(1).
Shippers’ Opening Br. 12. Yet the shippers make no attempt
to demonstrate this. They never pick a certain amount or
percentage of relief forgone by relief caps and claim that such
an amount renders the resulting rate unreasonable. Nor do
they pick a particular potential recovery ratio and claim that
cases falling under that ratio are necessarily those for which
full SAC is too costly. They argue only that the Board
arbitrarily failed to find that simplified SAC would provide a
reasonable rate in cases in which full SAC was too costly, not
that the Board’s findings on that point represent an
unreasonable interpretation of the statute. Even assuming
they have adequately raised the latter argument, the shippers
have failed to convince us of its merit.
True, as the shippers point out, at some case values
neither full SAC nor simplified SAC affords much relief. For
example, under full SAC a case with total relief of $7.5
million would net only $2.5 million. This modest recovery,
just half of the $5 million litigation costs, makes bringing the
case under full SAC a risky venture. Under simplified SAC
the case would net $4 million ($5 million capped relief minus
$1 million litigation costs), but the relief cap would require
14
the shipper to forgo $2.5 million, a substantial sum equal to
more than half of the shipper’s net relief. Even so, that hardly
means the Board erred in finding simplified SAC to be a
“simplified and expedited method of determining the
reasonableness of challenged rail rates in those cases in which
a full stand-alone cost presentation is too costly, given the
value of the case,” § 10701(d)(3). In light of the obvious
ambiguity inherent in this statutory language, we must uphold
the Board’s interpretation unless it is unreasonable. See Ass’n
of Am. R.Rs. v. STB, 237 F.3d 676, 680 (D.C. Cir. 2001)
(citing Chevron U.S.A. v. Natural Res. Def. Council, 467 U.S.
837, 843 (1984)). It isn’t. For one thing, as the Board points
out, the $4 million recovery under simplified SAC is more
valuable than would be the same amount obtained under full
SAC. After all, it comes more quickly and, thanks to reduced
litigation costs, with less downside risk, unquantifiable yet
real benefits to expedited procedures. Respt.’s Br. 46. For
another, as discussed above, the prescribed rate the Board
imposes—that is, the relief limited by the cap—isn’t itself a
rate that must qualify as reasonable under section
10701(d)(1). Far from limiting the reasonableness of a rate,
the relief cap represents a procedural mechanism necessary to
the reliable adjudication of a challenged rate’s reasonableness.
The very phrase “simplified and expedited method”
contemplates that the method will impose some costs. And
the costs imposed by the simplified SAC relief cap are
unlikely ever to exceed the litigation costs of full SAC;
because any case worth more than $9 million will yield more
potential profit under full SAC, simplified SAC’s relief cap
will presumably cause shippers to forfeit only $4 million, less
than the $5 million cost to litigate full SAC. Perhaps the
Board should treat relief caps differently than other
procedural mechanisms which impose litigation costs, but the
shippers make no such argument. We are thus unpersuaded
15
by their cursory suggestion that the Board’s interpretation of
section 10701(d)(3) is unreasonable.
Extending this argument to the choice between simplified
SAC and the three benchmark method, the shippers claim that
section 10701(d)(3) obligates the Board to identify the case
value at which simplified SAC is too costly and to ensure that
in those cases the three benchmark procedure produces
reasonable rates. We disagree. Although the statute requires
the Board to devise “a simplified and expedited method for
determining the reasonableness of challenged rail rates in
those cases in which a full stand-alone cost presentation [i.e.,
full SAC] is too costly, given the value of the case,” §
10701(d)(3), it nowhere requires the Board to provide any
procedure for those cases in which a simplified SAC
presentation is too costly. So long as simplified SAC
qualifies as a “simplified and expedited method,” nothing in
the statute requires the Board to promulgate the three
benchmark method at all. True, some cases will be too small
to bring under simplified SAC, which costs an average of $1
million, but that’s fully consistent with simplified SAC
qualifying as a “simplified and expedited method.” Under
any procedure, some cases will always be too small to be
worth bringing.
That said, having decided that it needed to implement a
three benchmark system, the Board had to do so
nonarbitrarily. See, e.g., Eagle Broad. Group v. FCC, 563
F.3d 543, 551 (D.C. Cir. 2009). We are convinced it did. The
Board analyzed the effect of the three benchmark relief caps
together with that of the simplified SAC relief caps. Its
findings that the relief caps “strike the appropriate balance”
and afford “sufficient net relief after litigation costs,”
Rehearing Decision at 8, apply to the three benchmark caps as
well as to the simplified SAC caps. For reasons similar to
16
those discussed above, these findings amply justify the
Board’s promulgation of the three benchmark caps.
Simplified SAC
In addition to challenging the relief caps, the shippers
object to the Board’s adoption of simplified SAC itself. They
claim that the Board’s decision not to require the simplified
SAC stand-alone railroad to be optimally efficient guts the
entire rationale for adopting simplified SAC in the first place.
They also claim that the Board was required to evaluate
simplified SAC using test data. We disagree on both counts.
In promulgating the simplified SAC method, the Board
eliminated the search for inefficiencies in the stand-alone
railroad. The shippers argue that this constitutes so
significant a deviation from CMP principles as to render the
Board’s reliance on those principles to justify simplified SAC
arbitrary and capricious. But the Board explained that the
efficiency inquiry is precisely what renders full SAC
presentations so costly, and that in its view modern railroads
suffer from too little inefficiency to justify this expense for
the purpose of simplified SAC. Decision at 55–56.
Specifically, while acknowledging that in its 1996 rulemaking
it had rejected the AAR-SSAC proposal because it failed to
account for inefficiencies, the Board explained its change in
position by observing that “rail capacity and traffic conditions
have changed,” and railroads are no longer “burdened by
substantial excess capacity.” NPRM at 14; see also Decision
at 55–56 (referencing NPRM). Given this, the Board
concluded that “railroads, in most instances, are likely
operating at a sufficiently efficient level so that it would not
be worth the time and considerable expense required to
attempt to measure the amount of inefficiency that could be
eliminated” by a more efficient stand-alone railroad.
Decision at 56. It further noted that railroads have little
17
incentive to build unnecessary facilities given the market
realities governing most rail rates, NPRM at 14, that
simplified SAC allows limited modifications based on certain
easy-to-detect inefficiencies, Decision at 56, and that in any
event simplified SAC conforms more closely to CMP
principles than does the three benchmark approach, id. at 56.
In our view, the shippers are simply second-guessing the
Board’s determination of how closely simplified SAC must
track CMP principles—principles the Board itself chose to
retain. True, simplified SAC does nothing to serve CMP’s
objective of eliminating inefficiencies, but it still perfectly
serves CMP’s other objective: eliminating cross-subsidization
of facilities for which shippers see no benefit. See id. at 55.
The Board reasonably concluded that this was enough. As we
have said: “[t]he pursuit of precision in rate proceedings, as in
most things in life, must at some point give way to the
constraints of time and expense, and it is the agency’s
responsibility to mark that point.” BNSF I, 453 F.3d at 482.
The shippers next take issue with the Board’s explanation
for its change in position from its 1996 rejection of the AAR-
SSAC proposal. Although “[a]n agency may not . . . depart
from a prior policy sub silentio or simply disregard rules that
are still on the books,” Fox, 129 S. Ct. at 1811, here the Board
did no such thing. As we have pointed out, it adequately
explained the turnabout by noting the decline in excess
capacity and overall increase in railroad efficiency since the
1996 rulemaking. See NPRM at 14; Decision at 55–56. The
Board’s inference that excess capacity signals inefficiency,
and the consequent finding that inefficiency had decreased to
the point that uncovering it was no longer cost-effective,
provide a sufficient explanation for jettisoning that inquiry in
simplified cases. According to the shippers, in 1996 the
Board treated certain capacity constraints (i.e., railroads
18
running at full capacity in certain situations) as indicative of
inefficiency, which they think contradicts its current treatment
of capacity constraints as a sign of efficiency. The simple
answer to this cryptic objection is that in 1996 the Board did
not treat capacity constraints as indicative of inefficiency.
The only evidence the shippers cite on this point is the Board
making the very different point that existing double-track
infrastructure was less efficient than more modern computer-
controlled single-track architecture. 1996 Guidelines, 1
S.T.B. at 1015 n.33. But this remark has nothing at all to do
with capacity constraints, so it can hardly conflict with the
Board’s current reasonable treatment of capacity constraints
as signaling lower excess capacity and greater efficiency.
Next, the shippers complain that the Board failed to test
its simplified SAC procedure with sample data.
Acknowledging that nothing in the statute requires such
testing, the shippers nonetheless maintain that since the Board
tested AAR-SSAC and found that it produced unreliable
results, the Board acted arbitrarily by failing to test its own
simplified SAC proposal. For its part, the Board explained
that the Commission only tested AAR-SSAC because the
AAR refused to provide the source code for the program, thus
preventing the Commission from learning its precise details.
Decision at 54. And although AAR-SSAC was similar to
simplified SAC in disregarding inefficiencies, it also had very
significant differences, as it expanded the stand-alone railroad
to include other profitable traffic and excluded certain
insufficiently profitable traffic, steps absent from the Board’s
simplified SAC. Compare id. at 13 n.18 (describing AAR-
SSAC method) with id. at 15–16 (describing the Board’s
simplified SAC). Given this difference between the two
methods, we see nothing arbitrary about the decision to test
the first but not the second, even after the first’s poor
showing.
19
III.
This brings us to the railroad’s challenges to the Board’s
adoption of the three benchmark system. They make four
arguments, which we consider in turn.
Notice
The three benchmark system involves a comparison of
the challenged rates to a group of rates taken from the waybill
sample. Under the Board’s initial proposal, parties could
propose comparison groups drawn from the most recent year
of waybill sample data. Under the final rule, however, the
parties could draw from the four most recent years of data.
Relying on APA section 553, the railroads argue that the
Board failed to provide notice that it was considering this
change. See 5 U.S.C. § 553 (requiring agencies to give notice
of proposed rules). We needn’t address the merits of this
argument, however, because the railroads failed to present it
to the Board. As a general rule, “‘courts should not topple
over administrative decisions unless the administrative body
not only has erred but has erred against objection made at the
time appropriate under its practice.’” Advocates for Highway
& Auto Safety v. Fed. Motor Carrier Safety Admin., 429 F.3d
1136, 1148 (D.C. Cir. 2005) (quoting United States v. L.A.
Tucker Truck Lines, Inc., 344 U.S. 33, 37 (1952)). And as we
recognized in Petroleum Communications, Inc. v. FCC, 22
F.3d 1164 (D.C. Cir. 1994), and reaffirmed just last month in
Globalstar, Inc. v. FCC, 564 F.3d 476 (D.C. Cir. 2009), the
objection that an agency violated the APA’s notice
requirement is a “classic example” of an issue that should be
raised before the agency. Petroleum Commc’ns, 22 F.3d at
1171; see also Globalstar, 564 F.3d at 484.
20
The railroads respond with two arguments. First, they
claim that they had no way of objecting to any lack of notice
until the Board promulgated its final rule. Fair enough, but
they never dispute the fact that they could then have sought
reconsideration, as did the shippers on other grounds.
Second, the railroads think that cases like Petroleum
Communications and Globalstar are distinguishable because
the FCC’s organic statute, unlike the Board’s, expressly
requires petitioners to present their objections to the agency.
Railroads’ Reply Br. 13 n.6. But this distinction makes no
difference because the FCC’s statute only “codifies the
judicially-created requirement of exhaustion,” Petroleum
Commc’ns, 22 F.3d at 1170, and we have described the statute
as “requiring the same degree of exhaustion for the FCC as
for other agencies,” Wash. Ass’n for Television & Children v.
FCC, 712 F.2d 677, 682 (D.C. Cir. 1983). Given that the
railroads argue neither that their claim falls within any of the
“recognized exceptions” to the issue exhaustion doctrine,
Petroleum Commc’ns, 22 F.3d at 1170, nor that the Board
was “afforded a fair opportunity to pass on the argument in
question,” Globalstar, 564 F.3d at 484 (internal quotation
marks and brackets omitted), they may not raise their claim
here.
The railroads insist that courts have no authority to
require parties to exhaust administrative procedures where a
statute imposes no such requirement. They are correct that in
Darby v. Cisneros, 509 U.S. 137 (1993), the Supreme Court
held that where no statute or regulation requires a party to
pursue administrative remedies, courts are without authority
to require parties to exhaust administrative procedures as a
precondition of seeking judicial review. Id. at 144–45. But in
Darby, the only question before the Court was whether
agency action was “final” for the purposes of judicial review.
21
See id. at 143–47. Because Darby says nothing at all about
other reasons courts might find certain claims barred, it leaves
intact the general requirement that parties give the agency a
chance to rule on all their objections. See ExxonMobil Oil
Corp. v. FERC, 487 F.3d 945, 962 (D.C. Cir. 2007)
(“Petitioners believe that the absence of a rehearing
requirement in the [Interstate Commerce Act] means that they
were not required to raise their complaints with FERC.
Petitioners miss the point: Their error was not failing to seek
rehearing, but rather failing to raise the issue at all.” (citations
omitted)). Even where, as here, presenting a claim to the
agency requires seeking reconsideration, nothing in Darby
permits parties to obtain judicial review of a claim they never
gave the agency a chance to address.
Regulatory Lag
The railroads argue that the Board arbitrarily failed to
account for the “regulatory lag” caused by the delay inherent
in using waybill sample data. Due to the time it takes the
Board to gather the data, the most current waybill samples are
at least one year old, and the ability to draw comparison
movements from the most recent four samples compounds the
problem. Citing the rapid change in rail rates over time, the
railroads argue that the use of outdated samples converts the
three benchmark system into a comparison between current
rates and historical rates. The Board recognized the problem
of regulatory lag and established a mechanism for addressing
it on a case-by-case basis. Although the three benchmark
procedure uses waybill sample data to run the benchmarks
and determine a presumed maximum lawful rate, it gives the
parties an opportunity to present evidence of “other relevant
factors” to rebut the presumption of lawfulness and seek to
modify the maximum allowable rate. Decision at 17, 21–22.
The railroads insist that this mechanism is insufficient for
three reasons, none of which has merit.
22
First, they argue that the opportunity to modify the
presumed maximum lawful rate is illusory because it requires
rebutting a presumption. But the Board has represented—and
the railroads nowhere meaningfully dispute—that the
presumption simply shifts the normal burden of persuasion to
the party seeking a modification. Respt.’s Br. 29; Railroads’
Reply Br. 6–7. This clearly allows the railroads a reasonable
opportunity to seek a modification.
Second, the railroads complain that when they submit
evidence of other relevant factors, the Board requires them to
quantify the impact of the factors on the overall rate. True,
this requires more of the railroads than would a rule allowing
them to simply dump evidence in the Board’s lap without
explanation, but it hardly poses an insurmountable hurdle.
Even under the railroads’ preferred alternative, they would
still need to present data sufficiently precise to have a
quantifiable impact—the only difference is that under the
Board’s system this process of adjustment occurs after
calculating the benchmarks, not before. The railroads offer
no reason to believe that quantifying the impact of changing
conditions is feasible at the outset of the benchmark analysis
but impracticable as an adjustment to the result of that
analysis. According to the railroads, in a recent set of cases
brought under the new guidelines, the Board was unpersuaded
by their proffered “other relevant factors” evidence. E.g., E.I.
du Pont De Nemours & Co., STB No. 42099 (June 30, 2008).
But whatever its propriety, the Board’s decision in those
cases, which we remanded to the agency unopposed, CSX
Transp., Inc. v. STB, No. 08-1246 (D.C. Cir. Jan. 15, 2009)
(remanding STB Nos. 42099, 42100, 42101), hardly
impeaches the entire rule.
23
Third, the railroads complain that the Board forbids
parties from submitting as “other relevant factors” evidence
either of movement-specific adjustments to the cost estimates
or of product or geographic competition. But these
objections, to which we turn in the next two subsections, have
no special force when applied to the regulatory lag problem.
If, as we conclude in that discussion, the Board may exclude
evidence of movement-specific costs or of competition
generally, it may certainly exclude evidence of change in such
costs or competition.
Evidence of Movement-Specific Adjustments
The three benchmarks evaluate ratios of revenues to
variable costs. Although revenues generated by a specific rail
movement are easy to measure, variable costs directly
associated with that movement are not. See Adoption of the
Uniform Railroad Costing System as a General Purpose
Costing System for All Regulatory Costing Purposes, 5
I.C.C.2d 894, 904 (1989) (“Given the degree of aggregation
in the accounting data reported to the Commission, it is
impossible for either Rail Form A or URCS [i.e., two
alternative costing systems] to produce true marginal costs for
particular movements . . . .”). To estimate these costs, the
Board uses the Uniform Rail Costing System (URCS), a
procedure that generates a statistical estimate of each
railroad’s variable costs based on its “system-wide average
variable costs.” BNSF II, 526 F.3d at 774. For years, the
Board has used URCS to answer the threshold question
whether a railroad has enough market dominance to allow the
Board to regulate the rate in the first place. In that context,
the Board originally allowed the parties to argue for
movement-specific adjustments to the cost estimates, but
recently decided to bar them, finding that “the cost savings
and increase in predictability . . . outweigh any gains in
accuracy from the railroads’ or shippers’ adjustment
24
proposals.” Id. at 776. We upheld the Board’s decision as a
permissible exercise of its judgment. Id.
Here the railroads argue that although the Board could
permissibly exclude movement-specific adjustments from
cost estimates used in the threshold market dominance
determination, it acted arbitrarily in barring them from
estimates used in the three benchmark procedure. None of the
railroads’ three arguments for this point has merit.
First, they claim that the Board failed to address
commenters’ proposals to use only certain such adjustments.
But the Board specifically considered and rejected these
intermediate proposals, explaining that it would be unfair to
allow only certain adjustments without granting the opposing
party an opportunity to “submit counter-adjustments,” as well
as “access to broad discovery” for that purpose. Decision at
97.
Second, they claim that the need for accurate cost
estimates is particularly acute in three benchmark cases given
the methodology’s inherent crudeness. The Board, however,
relied on its experience with such adjustments to conclude
they are too costly in light of their limited effect on accuracy.
Id. at 84. And although crude, the three benchmark method is
used for small disputes where efficiency is paramount.
Further, using movement-specific adjustments in a three
benchmark presentation would be even more cumbersome
than in the threshold market dominance determination, as it
would require calculating movement-specific adjustments for
every movement in the comparison group, not just the
challenged movement. Given this, the Board’s conclusion
that, as in the case of the threshold market dominance
determination, movement-specific adjustments are too costly
for three benchmark presentations, represents the “kind of
25
judgment call” that “balances inherently incommensurable
costs and benefits” and “falls within the expertise of the
agency,” BNSF II, 526 F.3d at 776.
Third, the railroads argue that comparing movement-
specific revenues with system-average costs is inherently
arbitrary. But if that were true, then doing so in determining
market dominance would have been equally impermissible.
And in any event, there are sound reasons for the mismatch:
getting movement-specific revenues is easy, whereas
calculating movement-specific variable costs is difficult, and
in the Board’s expert judgment not much more accurate.
Evidence of Product or Geographic Competition
Finally, the railroads complain about the Board’s
preclusion of evidence of product or geographic competition.
They argue that the Board never provided notice that it was
considering barring such evidence, but given that the Board
initially proposed barring all evidence offered to disturb the
result of the three benchmark calculation, the Board’s
eventual rule barring some evidence represented a “logical
outgrowth” of the proposal. Ass’n of Battery Recyclers, Inc.
v. EPA, 208 F.3d 1047, 1058 (D.C. Cir. 2000) (internal
quotation marks omitted); see also id. at 1059 (“EPA
proposed allowing alternative standards for remediated soils.
. . . One would logically conclude that EPA could have ended
up allowing alternative standards for all soils as the proposal
suggested, for no soils, or—as it turned out—for some
soils.”).
On the merits, the railroads argue only that the Board
failed to consider the possibility that parties could present
such evidence without the need for discovery. But no
commenter suggested to the Board that such evidence, which
the Board had previously excluded from full SAC cases due
26
to the discovery and other burdens it caused, Market
Dominance Determinations—Prod. & Geographic
Competition, 3 S.T.B. 937, 946–47 (1998), could be presented
without discovery. Given that, the Board hardly acted
arbitrarily in failing to consider that point.
V.
For the reasons stated above, we deny the petitions for
review in their entirety.
So ordered.