United States Court of Appeals
FOR THE DISTRICT OF COLUMBIA CIRCUIT
Argued February 5, 2008 Decided May 20, 2008
No. 06-1372
BNSF RAILWAY COMPANY,
PETITIONER
v.
SURFACE TRANSPORTATION BOARD AND
UNITED STATES OF AMERICA,
RESPONDENTS
WESTERN COAL TRAFFIC LEAGUE, ET AL.,
INTERVENORS
Consolidated with
06-1373, 06-1374, 06-1398, 06-1399, 06-1401, 06-1404,
06-1409, 06-1421
On Petitions for Review of an Order of the
Surface Transportation Board
John H. LeSeur argued the cause for Shipper Petitioners.
With him on the briefs were Christopher A. Mills, William L.
Slover, Kelvin J. Dowd, C. Michael Loftus, Andrew B.
Kolesar III, and Peter A. Pfohl.
2
Samuel M. Sipe, Jr. and Michael L. Rosenthal argued the
cause for Railroad Petitioners. With them on the briefs were
Richard E. Weicher, Anthony J. LaRocca, Peter J. Shudtz,
Paul R. Hitchcock, George A. Aspatore, John M. Hemmer,
Louise A. Rinn, Terence M. Hynes, G. Paul Moates, and Paul
A. Hemmersbaugh.
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.
Richard E. Weicher, Samuel M. Sipe, Jr., and Anthony J.
LaRocca were on the brief for intervenor BNSF Railway
Company.
C. Michael Loftus, Andrew B. Kolesar III, and Peter A.
Pfohl were on the brief for intervenor Western Coal Traffic
League.
Before: GINSBURG, ROGERS, and KAVANAUGH, Circuit
Judges.
Opinion for the Court filed by Circuit Judge
KAVANAUGH.
KAVANAUGH, Circuit Judge: In a recent rulemaking, the
Surface Transportation Board changed aspects of its rail rate-
setting methodology. Railroads and shippers both petition for
review – railroads arguing that certain changes improperly
benefit shippers and shippers arguing that certain changes
improperly benefit railroads. We conclude that the Board’s
3
changes are reasonable and reasonably explained. We
therefore deny the petitions.
I
Since Congress enacted the Interstate Commerce Act in
1887, the Federal Government has regulated the rates of
interstate railroads. Until 1995, the Interstate Commerce
Commission regulated the rates; since then, the Surface
Transportation Board has done so. See ICC Termination Act
of 1995, Pub. L. No. 104-88, §§ 101, 201, 109 Stat. 803, 804,
933-34.
Under federal law, a party may file a complaint with the
Board challenging a railroad’s rate. See 49 U.S.C.
§ 10704(b). After receiving a complaint, the Board first must
determine whether it has jurisdiction over the challenged rate.
The Board’s jurisdiction covers only those railroads that
possess “market dominance.” See §§ 10701(d)(1), 10707(b)-
(c). To have market dominance, a railroad must have revenue
that meets or exceeds 180 percent of its variable costs for the
traffic to which the rate applies. See §10707(d)(1)(A).
(Variable costs are those costs that increase as traffic over the
railroad increases – for example, the cost of fuel.)
After the Board determines that it has jurisdiction over a
challenged rate, the Board must decide whether the rate is
reasonable. See § 10701(d)(1). If the Board finds the rate
unreasonable, it sets the maximum rate the railroad may
charge. See §10704(a)(1). In setting that rate, the Board must
permit the railroad to cover its costs “plus a reasonable and
economic profit or return (or both) on capital employed in the
business.” §10704(a)(2).
Part of what makes railroad rate regulation complex is
that a railroad incurs many costs that cannot be attributed to
4
any one shipper – costs that the Board has appropriately
termed “unattributable costs.” See Rate Guidelines – Non-
Coal Proceedings, 1 S.T.B. 1004, at 2-5 (1996) (Non-Coal
Guidelines). For example, how does the railroad allocate the
cost of a railroad terminal shared by multiple shippers?
Allocation is difficult, moreover, because railroads serve a
mix of “competitive” shippers and “captive” shippers –
competitive shippers can secure alternative transportation
relatively cheaply but captive shippers cannot. See id.
Therefore, a railroad cannot simply charge each shipper a pro
rata share of the unattributable costs without the risk of losing
competitive shippers to other carriers. See id.
In 1985, the Board promulgated guidelines to calculate
rates for shipping coal. See Coal Rate Guidelines,
Nationwide, 1 I.C.C.2d 520 (1985) (Guidelines). The
Guidelines approach, which has since been extended to non-
coal rates, established certain principles to resolve rate
disputes. Those principles sought to approximate “Ramsey
pricing,” which sets rates for individual shippers in inverse
proportion to those shippers’ demand elasticities. See Non-
Coal Guidelines, at 2-5. Ramsey pricing enables a railroad to
collect a higher share of unattributable costs from captive
shippers than from competitive shippers. Because captive
shippers have inelastic demand, the railroads can charge them
higher rates with a lower risk of losing their business.
Recently, however, the Board decided that the Guidelines
approach had become increasingly complex and costly, and in
some respects contrary to congressional intent. To address
those problems, it began a rulemaking proceeding in early
2006. The Board completed the rulemaking later that year,
changing how to determine its jurisdiction and how to
evaluate rate reasonableness. Both railroads and shippers
5
filed timely petitions for review challenging various aspects
of those changes.
We review Board decisions under the deferential
standards of the Administrative Procedure Act. As relevant
here, we will set aside a Board decision if it is “arbitrary,
capricious, an abuse of discretion, or otherwise not in
accordance with law.” 5 U.S.C. § 706(2)(A). The Board may
depart from its own precedent, moreover, so long as it
provides a reasoned explanation. PPL Mont., LLC v. STB,
437 F.3d 1240, 1246 (D.C. Cir. 2006). 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.
II
We first consider the Board’s new method for
determining whether it possesses jurisdiction over a
challenged rate.
As a general matter, the Board has jurisdiction over a rate
if the railroad’s ratio of revenue to variable costs (R/VC) for
the traffic to which that rate applies is at least 180 percent.
Therefore, to determine whether it has jurisdiction, the Board
must have a method to calculate variable costs. The statute
requires that the Board use a method called the Uniform Rail
Costing System, referred to as URCS, or an adequate
substitute. See 49 U.S.C. § 10707(d)(1)(B); Adoption of the
Uniform R.R. Costing Sys., 5 I.C.C.2d 894 (1989). The
railroad submits various data to the Board, and the Board, via
a computer program, plugs the data into URCS to produce a
figure for system-wide average variable costs. See generally
Surface Transp. Bd., Industry Data – Economic Data: URCS,
6
http://www.stb.dot.gov. The amount of revenue from the
relevant traffic is then divided by a figure incorporating the
system-wide average variable costs and a number of operating
characteristics of the shipment to arrive at the R/VC ratio. If
the R/VC ratio is less than 180 percent, the Board has no
jurisdiction.
In the past, the Board has permitted parties to propose
“movement-specific adjustments” to the average variable
costs figure produced by URCS. In other words, parties could
argue that a higher or lower figure better reflected the variable
costs of a particular movement. Shippers, of course, propose
adjustments that would lower the variable-costs figure,
because that would result in higher R/VC ratios and thus
make Board review more likely. Railroads favor adjustments
that would raise the variable-costs figure, thereby lowering
R/VC ratios and making Board review less likely.
In the rulemaking at issue here, the Board eliminated the
ability of parties to suggest movement-specific adjustments.
Both the railroads and the shippers challenge that change as
an unreasonable departure from agency precedent. The Board
acknowledged that permitting movement-specific adjustments
has been its “longstanding practice,” but nevertheless
concluded that “these adjustments may not serve a useful
public purpose.” Major Issues in Rail Rate Cases, STB Ex
Parte No. 657, at 48 (Oct. 30, 2006). The Board gave seven
interrelated reasons for the change:
First, the analysis of proposals for movement-
specific adjustments is complex, expensive, and time
consuming. Second, the Board believed that
Congress intended, in adopting the 180% R/VC
limitation on Board rate review, to create an
administratively quick and easy-to-determine
7
regulatory safe harbor for the railroads. Third, the
URCS program already tailors the variable cost
calculation to the movement at issue. Fourth,
disallowing movement-specific variable cost
adjustments would eliminate substantial uncertainty
in the current rail rate adjudication process. Fifth,
railroads do not consistently keep certain types of
information that shippers have relied on for
favorable movement-specific adjustments. Sixth,
adjustments to URCS may not provide more reliable
results than using the system-average expenses.
Finally, piecemeal or incomplete adjustments to
URCS are suspect.
Id. (emphases added). The Board ultimately concluded that it
“must balance the costly burden and complexity created by
movement-specific adjustments against any improvements in
the resulting variable cost,” and it found that “notwithstanding
[its] past allowance of these adjustments, such expense and
complexity are not justified.” Id. at 50.
The railroads, except BNSF, challenge the Board’s
decision on statutory grounds. Section 10707 of Title 49
directs that “variable costs for a rail carrier shall be
determined only by using such carrier’s unadjusted costs,
calculated using the Uniform Rail Costing System cost
finding methodology . . . with adjustments specified by the
Board.” 49 U.S.C. § 10707(d)(1)(B). The railroads claim
that the last phrase – “with adjustments specified by the
Board” – means that the Board may not eliminate all
movement-specific adjustments. We disagree. To begin
with, the railroads did not raise this argument before the
Board, so it is forfeited. See Univ. of D.C. Faculty Ass’n v.
D.C. Fin. Responsibility & Management Assistance Auth.,
163 F.3d 616, 625 (D.C. Cir. 1998). In any event, it is
8
meritless. The statute does nothing more than broadly
delegate to the Board the authority to make reasonable
adjustments to the variable-costs figures produced by URCS.
It does not require the Board to adopt any adjustments. The
Board’s interpretation is therefore consistent with the
statutory text.
The railroads also claim that the Board did not give
adequate consideration to alternative proposals that would
allow the Board to take into account certain categories of
adjustments. We reject that argument as well. The Board
explained that it had considered the alternatives and found
none of them preferable in light of the seven considerations
listed above. The Board said that the elimination of
movement-specific adjustments would save up to $1 million
per party, per case. Moreover, the Board cited its years of
experience in dealing with those adjustments as the basis for
concluding that they are not especially accurate. In short, the
Board made a policy judgment that the cost savings and
increase in predictability of the Board’s jurisdiction, among
other factors, outweigh any gains in accuracy from the
railroads’ or shippers’ adjustment proposals. That kind of
judgment call, which balances inherently incommensurable
costs and benefits, falls within the expertise of the agency,
and we will not disturb it. Cf. Central & Southern Motor
Freight Tariff Ass’n v. United States, 757 F.2d 301, 321-22
(D.C. Cir. 1985) (“Deference is particularly appropriate when
– as here – the delegation of . . . power is very broad and
necessarily involves the administrative weighing of the costs
and benefits of regulation.”).
For the same reason, we reject the shippers’ arguments
that the Board’s decision to eliminate movement-specific
adjustments was unjustified. The shippers contend that the
Board placed too much emphasis on the expense of litigating
9
movement-specific adjustments and that the Board
underestimated the increase in accuracy effected by those
adjustments. Again, the Board possesses the responsibility to
balance those kinds of competing considerations. The
shippers have not demonstrated that the Board’s decision was
unreasonable or unsupported by substantial evidence.
The fact that both the railroads and shippers contest the
Board’s elimination of movement-specific adjustments is not
enough to persuade us that the Board’s decision was arbitrary
and capricious. The Board has an institutional interest in
reducing the cost for parties litigating rate cases. And the
Board has discretion to consider the interests of the railroads
and shippers that could not afford to participate in the
rulemaking proceeding.
III
We turn now to petitioners’ challenges to the changes in
the Board’s rate-evaluation methodology. To provide
necessary context for our discussion, we begin with a brief
overview of how the Board evaluates railroad rates.
As we have said, railroads serve a mix of competitive and
captive traffic. Because of the varying demand elasticities of
the different shippers, a railroad has no interest in
apportioning costs evenly among the shippers for facilities or
services that the shippers share. If it imposes a pro rata share
of unattributable costs on each shipper, competitive shippers
with lower-cost transportation alternatives may opt for those
alternatives, and the railroad would lose revenue. Despite that
problem, the railroads cannot go too far in the other direction
and overload captive shippers with excessively high rates.
Even though captive shippers do not have practical access to
alternative carriers, they do have access to Board review, and
10
the Board has a statutory duty to ensure that their rates are
reasonable.
The Board’s solution to the railroads’ problem, adopted
in Guidelines, has been the principle of Constrained Market
Pricing. See Coal Rate Guidelines, Nationwide, 1 I.C.C.2d
520 (1985) (Guidelines). Constrained Market Pricing sets
three constraints on a railroad’s rates, including the Stand-
Alone-Cost constraint, which ensures that a captive shipper
does not pay for services that provide it no benefits – in other
words, that it does not cross-subsidize other shippers. See
BNSF Ry. Co. v. STB, 453 F.3d 473, 476-77 (D.C. Cir. 2006);
Guidelines, 1 I.C.C.2d at 523-24.
To determine whether a complaining captive shipper is
paying for only those services that benefit it, the Board uses
an approach called the Stand-Alone-Cost test. The Stand-
Alone-Cost test posits a hypothetical railroad that serves a
subset of the movements in the railroad’s network, including
the route used by the complaining shipper. That hypothetical
railroad is called a Stand-Alone Railroad, known as a SARR,
and it is designed to be optimally efficient. The Stand-Alone-
Cost test determines the rate that the shippers using the SARR
(the “traffic group”) would be charged by taking into account
the costs of running the SARR, including a reasonable return
on investment, (the “Stand-Alone Costs”). See PPL Mont.,
LLC v. STB, 437 F.3d 1240, 1242 (D.C. Cir. 2006). The
amount of those costs becomes the maximum amount that the
railroad may collect from the traffic group. See id.
The underlying logic is that there are cost savings when
the portion of the railroad that constitutes the SARR is
combined with the rest of the real railroad; therefore, the costs
of that segment as part of the real railroad could never exceed
the costs of that segment if it stood alone. With a Stand-
11
Alone-Cost ceiling, “no shipper (or shipper group) subsidizes
others, at least in a strict sense of the term: though some bear
a higher share of fixed costs than others, they still pay no
more than what they would for a facility designed to serve
only them.” Burlington Northern R.R. Co. v. ICC, 985 F.2d
589, 596 (D.C. Cir. 1993).
The Board’s rulemaking changed various aspects of the
Stand-Alone-Cost test. Petitioners challenge three of those
changes: (i) the method the Board uses to determine
maximum reasonable rates; (ii) the degree to which
productivity gains are taken into account when forecasting the
SARR’s operating expenses; and (iii) the allocation of
revenue to the SARR from shippers that use both the SARR
and other, off-SARR facilities. Shippers also challenge the
Board’s application of its new revenue-allocation rule to a
case that was pending when the Board issued its notice for
proposed rulemaking.
A
We first address the Board’s change to its method of
determining a complaining shipper’s maximum reasonable
rate.
Under the Stand-Alone-Cost test, if the hypothetical
SARR’s total revenue from the Stand-Alone-Cost traffic
group exceeds the Stand-Alone Cost, then the traffic group in
real life is covering more of the costs of the real railroad than
are attributable to it, and the rates of the shippers in the traffic
group are reduced. Once the Board decides to reduce the
rates of the traffic group (for purposes of the test), it then
must determine how to allocate that reduction among various
members of the traffic group to set maximum reasonable
12
rates. In the rulemaking, the Board changed the way that it
performs that allocation.
In the past, the Board reduced the excessive rates in a
relatively straightforward way. Using the Percent Reduction
Method, the Board would reduce the rate of each shipper in
the traffic group by the same percentage: the percentage by
which the revenue from the traffic group exceeded the Stand-
Alone Costs. Thus, if revenue exceeded the Stand-Alone
Costs by 20 percent, the Board lowered the rate of each
shipper, including the complaining shipper, by 20 percent.
The rationale for that approach was that it maintained the
same proportion of rates among members of the traffic group.
For example, if one shipper initially paid twice the rate of
another shipper, that would continue to be true after the
reduction. The underlying assumption was that the existing
rate structure reflected the varying demand elasticities among
members of the traffic group. Under the Ramsey pricing
principle discussed above, which sets shippers’ rates in
inverse proportion to their demand elasticities, the Board
thought it important to maintain that rate structure – even
though the rates are entirely within the control of the railroad.
The railroads, of course, favor that assumption: In their view,
the Board should assume that the rates they set adequately
reflect differences in demand between the complaining
captive shipper and the other shippers.
In recent Stand-Alone-Cost cases, however, the Board
realized that railroads can easily manipulate the Percent
Reduction Method. In particular, a railroad can game the
system by initially setting an exceedingly high rate for a
captive shipper. If the shipper then challenges the rate and the
Board uses the Percent Reduction Method to reduce it, the
new rate will still be a function of the initial rate; the higher
the initial rate, the higher the final rate.
13
To prevent “gaming,” the Board adopted a new method
to correct excessive rates: the Maximum Markup
Methodology. Rather than requiring an across-the-board cut
for every shipper, this new methodology lowers only the rates
of those shippers that make excessive revenue contributions
relative to the variable costs that they impose on the railroad.
And it requires that those shippers’ ratios of revenue to
variable cost be the same. The railroads cannot manipulate
this methodology because the higher they set the initial rate of
a captive shipper, the higher that shipper’s revenue
contribution relative to the variable costs it imposes on the
railroad – and the bigger the percentage cut for that shipper.
The railroads argue that the Board failed to sufficiently
explain what it meant by “gaming.” We, however, have no
trouble understanding the Board’s concern: A railroad could
charge any rate, including an inefficient monopoly rate,
simply by setting the rate incrementally higher than the rate it
wanted prior to the SAC proceeding.
The railroads further argue that the Board’s decision is
arbitrary and capricious because there is no evidence of
gaming by railroads. They cite our decision in National Fuel
Gas Supply Corp. v. FERC, which vacated a prophylactic rule
aimed at preventing market manipulation. See 468 F.3d 831
(D.C. Cir. 2006). In that case, FERC had specifically relied
on a supposed record of abuse to justify a rule, yet FERC had
not produced any evidence of abuse. See id. at 841. The
Court’s order instructed FERC to either compile the record of
abuse or “try to support [its rule] by setting out its best case
for relying solely on a theoretical threat of abuse.” Id. at 844.
In this case, the Board reasonably explained that the
undetectable nature of the problem plainly justifies the
Board’s reliance on the theoretical threat. To discern whether
an initial rate is set because it reflects a railroad’s perception
14
of relative demand or a railroad’s effort to game the system
would require the Board to either divine the motives of the
railroad in setting the challenged rate or undertake the costly
task of estimating the railroad’s marginal costs and the
complaining shipper’s demand elasticity. The Board
reasonably concluded that either endeavor would be utterly
impracticable.
In addition to the anti-gaming rationale, the Board
offered another justification for adopting the Maximum
Markup Methodology: By statute, railroads must maximize
revenue from competitive shippers before increasing captive
shippers’ rates. See 49 U.S.C. § 10701(d)(2)(B); Guidelines,
1 I.C.C.2d at 539 (Under Constrained Market Pricing, “a
carrier must charge its competitive traffic as much of the
unattributable costs as the demand will permit.”). According
to the Board, this “reflects a Congressional directive” that
captive shippers “not bear a differentially larger share of the
joint and common expenses” until the railroad has charged its
competitive shippers “as much of the unattributable costs as
demand will permit.” STB Ex Parte No. 657, at 18. The
Board determined that the Maximum Markup Methodology
better implemented that statutory directive: Unlike the
Percent Reduction Method, it allows captive shippers, which
tend to contribute more revenue relative to the variable costs
they impose on railroads, to receive a disproportionately
higher share of a rate reduction.
The railroads counter that giving a disproportionately
higher share of a rate reduction to captive traffic runs directly
counter to the Ramsey pricing principle that the Guidelines
approach adopted. Those principles instruct that rates should
be set in inverse proportion to shippers’ demand elasticities.
The railroads argue that once the rate structure has been
established in that way, it should be maintained. The Percent
15
Reduction Method preserved the rate structure because the
Board would reduce the rates of all shippers in the traffic
group by the same percentage when it would conclude that a
railroad was receiving excessive revenue.
The Board’s Maximum Markup Methodology is not a
departure from Ramsey pricing principles as reflected in
Guidelines unless one assumes that railroads set the initial
rate structure in inverse proportion to the shippers’ demand
elasticities. The Board’s conclusion that rate structures are
susceptible to gaming rejects that assumption. Moreover, the
Maximum Markup Methodology preserves demand-based
differential pricing to a significant degree: Shippers that pay
low rates relative to the variable costs attributable to them will
still bear considerably less of the railroad’s unattributable
costs than shippers that pay high rates relative to the variable
costs attributable to them.
There is therefore no contradiction between the
Maximum Markup Methodology and the Board’s goal under
Guidelines: Under both approaches, the objective is for
railroads to “ensure that competitive traffic contributes as
much as possible toward [unattributable] costs,” which
includes ensuring that competitive traffic does not leave the
railroad for transportation alternatives. 1 I.C.C.2d at 524. As
the Board put it, “Congress envisioned that captive shippers
would be the residual suppliers of capital, but only where the
competitive traffic cannot provide a sufficient share of the
contribution needed to support the rail infrastructure that it
uses.” STB Ex Parte No. 657, at 18. The Board has simply
changed its mind about how best to achieve that goal. It no
longer assumes that whenever it finds a railroad to be
receiving excessive revenue in a Stand-Alone-Cost case,
every shipper’s rate is too high and the railroad must lower all
of its shippers’ rates by the same percentage to maximize
16
revenue from competitive traffic. Now the Board believes
that it makes the most sense to lower the rates of only those
shippers that are paying a high rate relative to the variable
costs attributable to them. The Board has license to change
how it implements its statutory duties, “either with or without
a change in circumstances,” so long as it supplies “a reasoned
analysis.” Motor Vehicle Mfrs. Ass’n v. State Farm Mut.
Auto. Ins. Co., 463 U.S. 29, 57 (1983) (internal quotation
marks omitted). The Board has met that requirement here,
and we find no reason to overturn its decision.
Finally, the railroads argue that the Maximum Markup
Methodology violates this Court’s decision in Burlington
Northern Railroad Co. v. ICC, 985 F.2d 589 (D.C. Cir. 1993).
But that decision addressed a substitute for the entire Stand-
Alone-Cost analysis, not a different method of reducing rates
after performing the Stand-Alone-Cost test. We found
multiple defects in the approach at issue in Burlington,
including a flaw that deprived the ICC’s approach of “any
glimmer of supporting principle or intellectual coherence.”
Id. at 597. By contrast, the Board here responded to a flaw in
its existing Percent Reduction Method, and adopted the
Maximum Markup Methodology to correct the problem. This
new methodology, as explained above, furthers the Board’s
goals under Guidelines and § 10701(d)(2)(B). Therefore, we
are satisfied that the Board’s decision is consistent with
Burlington.
B
We next address the Board’s change to its method for
forecasting the SARR’s future operating expenses.
To calculate the costs that the hypothetical SARR would
likely incur over the 10-year Stand-Alone-Cost analysis
17
period, the Board must estimate the operating expenses that
the SARR would face. Since 1980, the Board has used some
form of the “Rail Cost Adjustment Factor,” established by
statute, as an index to track changes in railroad costs. Before
this rulemaking, the Board’s operating-expense forecasts did
not take into account the possibility that the SARR could
experience productivity gains – gains in efficiency that would
reduce operating expenses. The Board figured that, because
“the SARR is designed to be an efficient replacement for the
railroad, it would not be able to realize the same productivity
gains as the rest of the industry, particularly in the early
years.” STB Ex Parte No. 657, at 40. In its Stand-Alone-Cost
tests, the Board thus had used the Rail Cost Adjustment
Factor-U index, which measures “the change in the prices of
inputs, such as labor and fuel, used to produce railroad
services,” but does not factor in anticipated industry-wide
productivity gains. Id. at 39. A separate index, the Rail Cost
Adjustment Factor-A index, takes into account the industry’s
productivity gains. Shippers have urged the Board to adopt
that index because, if there are productivity gains, then
operating expenses will be lower. And the lower the
forecasted operating expenses of the SARR, the lower the
revenue needed to cover the SARR’s costs, and the lower the
maximum permissible rate for shippers. The railroads, by the
same logic, have favored the status quo.
In the rulemaking, the Board settled on a hybrid
approach. Based on its special expertise in rail regulation, the
Board posited that a new hypothetical railroad would not
immediately experience the same level of productivity growth
as the anticipated industry average: “[A] SARR is presumed
to begin the analysis period at a higher productivity level than
the industry as a whole,” and as a result, in the early years, it
would not have as much room to increase productivity in
certain areas. Id. at 43. For example, “railroads realize
18
productivity gains in locomotives as they replace old
locomotives with newer technologies. The SARR would not
experience those same productivity gains in the short term,
because it would begin its operations with all new
locomotives.” Id. at 40. The Board, however, concluded that
a SARR would experience some productivity increases
“where gains derive from more efficient use of existing assets
such as improved management techniques, more flexible
work rules and learning by doing.” Id. at 43. Also, the SARR
could experience productivity gains for “short-lived assets”
whose replacement would “introduce the latest available
technology.” Id.
The Board posited that, within 20 years, the SARR’s
productivity growth rate would match that of the industry
because at that time the SARR would have about the same
mix of old and new assets as the industry generally. “[A]s the
SARR approaches the industry’s vintage of technology over
time, both the productivity level and the rate of growth for the
industry and the SARR would converge.” Id. at 44.
Therefore, the Board decided to phase in the Rail Cost
Adjustment Factor-A index – the measure of costs that takes
into account productivity gains – into its operating-expense
forecast gradually over a 20-year span. To accomplish this,
the Board calculates operating expenses based solely on the
Rail Cost Adjustment Factor-U index (no productivity gain)
for year 1 and factors in the Rail Cost Adjustment Factor-A
index (full productivity gain) at a rate of 5 percent per year
until it is fully phased in at year 20.
Unsurprisingly, both the shippers and the railroads object
to the Board’s hybrid approach, with each favoring an
opposite end of the spectrum. The shippers argue that the
Board ignored “substantial evidence of record before the STB
demonstrating the rapid rate at which the railroad industry
19
renews its assets and its technology.” Shippers’ Br. 36. The
shippers note that the average age of rail assets in 2002 was
only seven years. For their part, the railroads argue that
“there was no evidence that any [productivity] improvements
would occur in equal amounts over a 20-year period.”
Railroads’ Br. 34. The railroads believe that most
productivity gains would not be realized until many years in
the future. As a result, the railroads argue, even if the Board
is correct that the SARR would converge with the industry in
20 years, because the Stand-Alone-Cost analysis period
covers only the first 10 years, underestimating productivity
gains in years 11 through 20 will not balance out the
inaccuracy created by overestimating productivity gains in
years 1 through 10.
We decline to enter this hyper-technical fray. “It is well
established that an agency’s predictive judgments about areas
that are within the agency’s field of discretion and expertise
are entitled to particularly deferential review, so long as they
are reasonable.” Wis. Pub. Power, Inc. v. FERC, 493 F.3d
239, 260 (D.C. Cir. 2007) (internal quotation marks omitted);
see also Nuvio Corp. v. FCC, 473 F.3d 302, 306 (D.C. Cir.
2007) (We owe “substantial deference [to an agency’s]
predictive judgments.”). That maxim is especially true here,
where we are reviewing the Board’s predictive judgment
about hypothetical railroads. The agency has adopted a
straight-line, phase-in approach that is routinely used to
estimate the depreciation of assets, and we cannot conclude
that the approach is unreasonable. Although the parties have
submitted evidence that they claim supports their conflicting
views on how a SARR would experience productivity gains,
“[p]articularly where, as here, an agency issues a regulation
reflecting reasoned predictions about technical issues, logic
suggests that the record may well contain evidence sufficient
to support more than one possible outcome.” Ass’n of Pub.-
20
Safety Communications Officials-Int’l Inc. v. FCC, 76 F.3d
395, 398 (D.C. Cir. 1996). And as for the railroads’ claim
that imperfections in the 20-year phase in may inure to the
benefit of the shippers, at some point simplicity outweighs
accuracy, and the Board “is free to make reasonable trade-offs
between the quality and cost of possible regulatory
approaches.” Burlington Northern, 985 F.2d at 597.
C
We now consider the Board’s change to its method of
allocating to the SARR the revenue from shippers that use
both the SARR and other, off-SARR parts of the railroad.
As we have said, the Stand-Alone-Cost analysis posits a
hypothetical railroad – the SARR – that would serve the route
that the complaining shipper uses. The Stand-Alone-Cost
analysis then determines the total costs that the SARR would
incur – the Stand-Alone Costs – and what percentage of those
costs is attributable to the complaining shipper. If the total
revenue that the railroad collects from the SARR’s services
(calculated based on the real-world rates that the railroad
charges the traffic group that uses the SARR) exceeds the
Stand-Alone Costs, then the rate of the complaining shipper
may be lowered in accordance with the Maximum Markup
Methodology discussed above.
In determining the total revenue that a SARR generates, a
problem arises: Unlike the complaining shipper, the other
shippers do not necessarily use only the SARR. In the real
world, other shippers may use both on-SARR and off-SARR
parts of the railroad. For those shippers, the Board must
allocate to the SARR only a portion of the revenue that they
contribute in the real world. If the Board attributed all of their
revenue contribution to the SARR, it would overestimate the
21
SARR’s revenue because some of those shippers’ revenue
contributions go to covering off-SARR costs. The Board has
termed the traffic that uses both on-SARR and off-SARR
facilities “cross-over traffic.”
In this rulemaking, the Board changed the way that it
allocates the revenue of cross-over traffic between on-SARR
and off-SARR facilities. The Board previously allocated
revenue based essentially on the percentage of miles the
shipper used the SARR. Thus, if 60 percent of a shipper’s
route was on-SARR and 40 percent was off-SARR, roughly
60 percent of its revenue contribution would be allocated to
the SARR.
Although the old approach had the virtue of simplicity, it
had a critical flaw, which we identified in BNSF Railway Co.
v. STB, 453 F.3d 473 (D.C. Cir. 2006). The mileage-based
approach did not take into account “economies of density” –
the principle that the more traffic on a given stretch of rail, the
lower the average cost (and hence the lower the cross-over-
traffic revenue that should be attributed to it).
To take an example, imagine a toll road that five drivers
use. If the annual upkeep for the road costs $100, each driver
would need to contribute $20 annually. If those drivers pay
$40 per year in taxes, then 50 percent of their tax contribution
is attributable to the road. Now imagine that 50 drivers use
the road – that is, that its density has increased tenfold. Each
driver would need to contribute only $2 annually. Of their
$40 tax liability, only five percent would be attributable to the
road. The same logic applies here. For cross-over traffic, the
higher the density of the on-SARR facilities, the smaller the
proportion of their overall revenue contribution should be
attributed to the SARR. In other words, more of their revenue
22
contribution must be going to cover costs for off-SARR
facilities.
In BNSF, the complaining railroad proposed a method of
allocating revenue from cross-over traffic that would have
taken into account economies of density. We concluded,
however, that the Board had reasonably declined to adopt that
alternative because the proposal ignored the diminishing
nature of economies of density – that is, the fact that at some
point, higher density no longer results in lower average costs.
See id. at 483-84. As the Board summarized the principle,
“the railroad industry is characterized by economies of
density, meaning the average total cost for a network of a
given size initially decreases with increases in output. But
economies of density also diminish with higher output and at
some point are exhausted.” STB Ex Parte No. 657, at 26.
Although we were not convinced in BNSF that the Board
had acted unreasonably in rejecting the incomplete alternative
proposed by the railroad, we stated that “[w]ere the Board
presented with a model that took account both of the
economies of density and of the diminishing returns thereto, a
decision to adhere to [the old] model would be on shaky
ground indeed. But that day is yet to come.” BNSF Ry., 453
F.3d at 484.
In this rulemaking, the Board determined that the day had
arrived. It adopted an approach that takes into account both
economies of density and their diminishing nature. The
Board’s new approach – called the Average-Total-Cost
method – allocates revenues based partly on the average total
cost of a segment rather than just on mileage. Because
average total cost for a given segment of rail decreases as
density increases (up to a point), basing the revenue allocation
in part on average total costs solves the problem that we
23
identified in BNSF. As the Board recognizes, our decision in
BNSF strongly suggested that the Board would be required to
adopt an appropriate density-based approach if one were
presented to it. The Average-Total-Cost method “takes
account of both economies of density and diminishing
returns.” STB Ex Parte No. 657, at 34. The Board thus
concluded that continued use of the mileage-based approach
“would be on shaky ground.” Id.
The shippers nonetheless claim that the Board’s new
revenue-allocation formula arbitrarily departs from
Guidelines. Their argument can be summarized in the
following syllogism: Guidelines does not permit the Board,
when setting rates, to allocate a percentage of fixed costs to a
given shipper, but rather requires the Board to set rates on the
basis of shipper demand. The new revenue-allocation
formula for cross-over traffic, which is a fundamental
component of the Stand-Alone-Cost analysis, is based on the
average total costs of the on-SARR and off-SARR segments,
not shipper demand. Therefore, the cost-based, revenue-
allocation formula violates Guidelines.
We do not agree that the Board’s change to the Average-
Total-Cost method was unreasonable or contrary to precedent.
We have already held that the Board may allocate revenue
between on-SARR and off-SARR facilities without taking
into account shipper demand. In BNSF, we upheld the
mileage-based approach because the Board had reasonably
assumed that “average costs are a continuous function of
distance.” BNSF Ry., 453 F.3d at 483 (internal quotation
marks omitted). The new method simply refines that
approach by taking into account economies of density.
Although Guidelines may favor a demand-based approach
generally for setting rail rates, the Board has acted reasonably
in using a cost-based approach, for the Stand-Alone-Cost test,
24
to estimate the costs that cross-over traffic imposes on the
SARR. “The 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. Our role is limited to
determining whether the balance it struck is arbitrary.” Id. at
482. In this case, it follows from our decision in BNSF that
the Board’s action was reasonable.
D
The shippers contend that the Board’s application of the
Average-Total-Cost method to a case that was pending when
the Board issued its notice for proposed rulemaking was
impermissibly retroactive and otherwise arbitrary and
capricious. See Western Fuels Ass’n, Inc. v. BNSF Ry. Co.,
2007 WL 2590251 (STB Sept. 7, 2007). The shippers argue
that the Board should not have applied its new Average-Total-
Cost revenue-allocation formula because they had relied on
the mileage-based approach in incurring significant costs to
design and defend a SARR for the Stand-Alone-Cost analysis.
We reject the shippers’ argument. “A new rule may be
applied retroactively to the parties in an ongoing adjudication,
so long as the parties before the agency are given notice and
an opportunity to offer evidence bearing on the new standard,
and the affected parties have not detrimentally relied on the
established legal regime.” Consol. Edison Co. v. FERC, 315
F.3d 316, 323 (D.C. Cir. 2003) (internal citations omitted).
Here, there was no established legal regime on which the
parties litigating before the Board could have reasonably
relied: They were on notice that the Board had not settled on
any one method for allocating the revenue contribution of
cross-over traffic. As we said in BNSF, “[t]he appropriate
allocation of revenue from cross-over traffic is a perennial
25
issue in [Stand-Alone-Cost] proceedings and one the Board
even now [in 2006] has not resolved definitively.” 453 F.3d
at 483; see also, e.g., Duke Energy Corp. v. Norfolk Southern
Ry. Co., 2003 WL 22673026, at *10 (STB Nov. 5, 2003)
(“The Board has long recognized, however, that this
methodology may not work in all cases, and it has been open
to suggestions for other methods to allocate cross-over
revenues.”). The shippers do not respond to the Board’s
argument that, before adopting the Average-Total-Cost
method, the Board had repeatedly warned that it sought to
adopt a methodology that would take density into account.
As the Board made clear both in the rulemaking and in
Western Fuels, the shippers had no basis for relying on the
prior revenue-allocation formula. See STB Ex Parte No. 657,
at 75; Western Fuels, 2007 WL 2590251, at *20.
Nevertheless, the Board gave the shippers an opportunity to
redesign or defend their SARR using the new formula. See
Western Fuels, 2007 WL 2590251, at *20.
Moreover, given that the new methodology was
“designed in large part to improve the reliability of [the
Stand-Alone-Cost] analysis, and given the possibility of rate
prescriptions of nearly 20 years,” it was reasonable for the
Board to immediately discard the flawed procedure and apply
its new rule to pending cases when the parties were on notice
of the potential change. STB Ex Parte No. 657, at 76.
***
For the reasons stated above, we deny the petitions for
review.
So ordered.