GINSBURG, Chief Judge.

BNSF Railway Co. petitions for review of an order of the Surface Transportation Board rejecting as unreasonable certain rates the railroad charged the Public Service Company of Colorado, d/b/a Xcel Energy, to ship coal from the Powder River Basin in Wyoming to Xcel's electric generating plant in Colorado. BNSF argues first the Board should have dismissed the rate proceeding three years after the complaint was filed, pursuant to the limitation in 49 U.S.C. § 11701(c). In the alternative BNSF argues we should, for a number of reasons, set aside the Board's order as arbitrary and capricious. We hold BNSF's first argument is forfeit and its other arguments are unpersuasive, wherefore we deny its petition for review.

I. Background

With the passage of the Staggers Rail Act of 1980, the Congress limited regulation of railroad rates to markets in which a single carrier exercises "market dominance," defined as "an absence of effective competition from other rail carriers or modes of transportation." 49 U.S.C. §§ 10701(c)-(d), 10707(a). Furthermore, it provided in the ICC Termination Act of 1995 that the Surface Transportation Board may begin an investigation into the reasonableness of a carrier's rates "only on [the] complaint" of an affected shipper. Id. § 11701(a). If the Board finds the carrier dominates the relevant market, then it must determine whether the rate charged the shipper is "reasonable." Id. § 10701(d)(1). If the rate is "unreasonable," id. § 10707(c), then the Board may prescribe the maximum lawful rate, id. § 10704(a)(1), and order the railroad to pay reparations to the complainant, id. § 11704(b). The Board is precluded, however, from finding market dominance and in turn regulating the rate if the revenue generated thereby does not exceed 180% of the carrier's variable cost of service. Id. § 10707(d)(1)(A).

The Board evaluates the "reasonableness" of rail rates in light of the standards promulgated by its predecessor, the Interstate Commerce Commission, see Coal Rate Guidelines, Nationwide, 1 I.C.C.2d 520 (1985), aff'd sub nom. Consol. Rail Corp. v. United States, 812 F.2d 1444 (3d Cir. 1987). In the Coal Rate Guidelines the Commission adopted the principles of Constrained Market Pricing (CMP) to set upper limits on the rates a railroad may charge its "captive shippers" — those customers who do not have practical access to an alternative carrier and who, because of their inelastic demand, the railroads may charge rates that significantly exceed the variable cost of service. Id. at 521. The Commission concluded that these principles would "meet [its] dual objectives of providing railroads the real prospect of attaining revenue adequacy while protecting coal shippers from `monopolistic' pricing practices." Id. at 524-25. Under CMP, rail carriers set their own rates for rail service, subject to three main constraints: revenue adequacy, management efficiency, and stand-alone cost. See id. at 534-46.

A shipper may challenge a rate either on a system-wide basis, by arguing that the rate charged exceeds the amount necessary for the railroad to achieve "revenue adequacy [as] adjusted for demonstrated management inefficiencies," id. at 534 n. 35, or as in this case, under the stand-alone cost (SAC) test, which is designed to prevent "cross-subsidization." Id. at 541. Regardless of a railroad's overall revenue adequacy, therefore, the rate charged a captive shipper is further constrained by the principle that a "captive shipper should not bear the costs of any facilities or services from which it derives no benefit," that is, should not be required to cross-subsidize other shippers. Id. at 523.

A shipper challenging a rate under the SAC test must hypothesize an efficient "stand-alone railroad" (SARR) that would serve the "captive shipper or a group of shippers who benefit from sharing joint and common costs." Id. at 528. The test assumes a "contestable market," that is, a market without any barriers to entry or exit. Id. If the rate being challenged is more than would be required by the hypothetical new entrant to cover its costs (including a reasonable return on investment), then that rate is unreasonable. See id. at 528-29.

A SAC proposal must be comprehensive, taking into account a host of variables from capital expenses for trains and track to the operating plan and routing of traffic on the SARR. The complaining shipper has "broad flexibility" to design the route of the SARR in order "to lower costs by taking advantage of economies of density." Id. at 543. Although there are no "restrictions on the traffic that may potentially be included in a stand-alone group," the proponent "must identify, and be prepared to defend, the assumptions and selections it has made." Id. at 544. There is a rebuttable presumption that "non-issue" traffic, that is, the traffic of non-complaining shippers, will contribute revenue "at the level of their current rates." Id. When such traffic is routed over a SARR for only a part of its through movement, the method for allocating the revenue from this "cross-over traffic" may be hotly disputed, as it is in this case — of which more later.

Xcel filed its complaint with the Board in December 2000 challenging rates BNSF charged for the transportation of coal from the Powder River Basin to Xcel's Pawnee electric generating station near Brush, Colorado. In January 2003 Xcel submitted its opening evidence, including its proposed SARR, which replicated a section of the traffic handled by BNSF's rail lines between the Eagle Butte mine in Northern Wyoming and the Pawnee plant. Cross-over traffic, which would move on the SARR for only a part of its overall movement before reaching an interchange point where it would be transferred to BNSF for carriage to its destination, accounted for more than 90% of all traffic on the SARR. See Appendix (map showing route of SARR and residual BNSF lines that handle cross-over traffic).

BNSF moved in February 2003 to dismiss the complaint on the ground that Xcel's operating plan was infeasible and it had therefore failed to make out a prima facie case. The Board denied the request, holding BNSF had "not demonstrated that the alleged errors in Xcel's evidence are so large in magnitude or so egregious as to warrant dismissing the complaint at this early stage in the proceeding." BNSF then submitted its own evidence, which focused upon four points: (1) BNSF must be allowed to charge shippers with highly inelastic demand, such as Xcel, high rates in order to achieve revenue adequacy; (2) the use of cross-over traffic, upon which Xcel's SARR so heavily relied, distorted the results of the SAC analysis by allocating excessive revenues to the SARR's portion of the overall movement; (3) the single largest movement on the SARR, coal destined for Western Resources' Jeffrey Energy Center (the "Jeffrey traffic"), was unreasonably diverted from its present route to a longer route on the SARR; and (4) Xcel's operating plan was infeasible.

In June 2004 the Board ruled in favor of Xcel, rejecting BNSF's challenges to Xcel's SAC presentation and holding BNSF's rates unreasonable. See Pub. Serv. Co. of Colo. d/b/a Xcel Energy v. Burlington N. Santa Fe Ry., STB Docket No. 42057, 2004 WL 1428724 (STB served June 7, 2004) ( Decision I). BNSF petitioned for reconsideration, which the Board denied in relevant part, see Pub. Serv. Co. of Colo. d/b/a Xcel Energy v. Burlington N. Santa Fe Ry., STB Docket No. 42057, 2005 WL 126476 (STB served Jan. 19, 2005) ( Decision II), and then petitioned this court for review.

II. Analysis

As a threshold matter, BNSF argues Xcel's complaint should have been dismissed pursuant to 49 U.S.C. § 11701(c) in December 2003, three years after it was filed. In the alternative the carrier claims the Board's decision is, in a number of respects, arbitrary and capricious.

A. Three-Year Time Limit

Subsections 11701(a) and (c) of Title 49 provide in pertinent part:

(a) Except as otherwise provided in this part, the Board may begin an investigation under this part only on complaint.

. . .

(c) A formal investigative proceeding begun by the Board under subsection (a) of this section is dismissed automatically unless it is concluded by the Board with administrative finality by the end of the third year after the date on which it was begun.

BNSF contends this rate proceeding was "begun by the Board under subsection (a)," 49 U.S.C. § 11701(c), when Xcel filed its complaint on December 20, 2000. Therefore, BNSF urges, the case was dismissed "automatically" on December 20, 2003, nearly six months before the Board issued its decision.

The Board counters that BNSF forfeited this argument because, even if the three-year limitation applied, BNSF did not raise the point until long after three years had elapsed and the Board had ruled; indeed BNSF first made the argument in a footnote to its petition for reconsideration. On the merits the Board reasons that because the Congress "cannot have intended to punish a complainant for agency inaction," the three-year limit must be read to apply only to investigations begun by the Board "on its own initiative." This it does by reading the phrase "formal investigative proceeding," as used in § 11701(c), to refer not to an investigation begun "on complaint" of a captive shipper, pursuant to the second clause of § 11701(a), but rather to a Board-initiated investigation "otherwise provided in this part" and thus within the first or exception clause of § 11701(a). See, e.g., 49 U.S.C. § 722(c) (Board may reopen an investigation); id. § 10704(b) (Board may extend an investigation). The Board contends its interpretation of "formal investigative proceeding" is consistent with the Commission's reading of the preceding version of § 11701. See 49 U.S.C. § 11701 (1978) (amended 1995). The Commission read the three-year limit in that version of § 11701(c) to apply only to Commission-initiated investigations under § 11701(a), which at the time provided the Commission could begin an investigation not only on complaint but also "on its own initiative," id. § 11701(a). See Complaints Filed Pursuant to the Savings Provisions of the Staggers Rail Act of 1980, 367 I.C.C. 406 (1983). According to the Board, because by 1995 "the term `formal investigative proceeding' had an established meaning," the Congress "is presumed to have been aware of [that interpretation] when it retained that term." See Lorillard v. Pons, 434 U.S. 575, 580-81, 98 S.Ct. 866, 55 L.Ed.2d 40 (1978). Further, the Board argues that reading the revised statute differently than the Commission read the preceding version would produce an "absurd, unfair," and perhaps unconstitutional result because it would "depriv[e] Xcel of a decision on the merits of its rate complaint where the delay was not Xcel's fault"; more generally, it would, quite perversely, reward any railroad that managed to prolong a rate proceeding beyond the three-year time limit.

The Board's concern with due process for shippers may be well-founded. See Logan v. Zimmerman Brush Co., 455 U.S. 422, 428, 433-34, 102 S.Ct. 1148, 71 L.Ed.2d 265 (1982) (holding "a cause of action is a species of property protected by the Fourteenth Amendment's Due Process Clause" and therefore could not constitutionally be extinguished by expiration of 120-day period for state agency to convene fact-finding conference). We need not resolve the issue of the three-year limit, however, because BNSF failed to raise the argument in a timely manner. A reviewing court generally will not consider an argument that was not raised before the agency "at the time appropriate under its practice." United States v. L.A. Tucker Truck Lines, Inc., 344 U.S. 33, 37, 73 S.Ct. 67, 97 L.Ed. 54 (1952). BNSF raised this argument when, after three and one half years of proceedings, the Board had ruled against it on the merits and the carrier was petitioning for reconsideration. Assuming its relegation of the argument to a footnote was not itself fatal, cf. United States v. Whren, 111 F.3d 956, 958 (D.C. Cir. 1997) ("absent extraordinary circumstances . . . we do not entertain an argument raised for the first time . . . in a footnote"), the timing surely was.

Without identifying the exact moment the argument was forfeited, we are confident it could not have been later than when the Board decided the case because the criteria for granting reconsideration are limited by statute; the Board may not grant a petition for reconsideration except for "material error, new evidence, or substantially changed circumstances." 49 U.S.C. § 722(c). The three-year limitation obviously was not new evidence or a changed circumstance, and if it was a "material error," the error was induced by BNSF's own failure to raise the argument in good time. Cf. Canady v. SEC, 230 F.3d 362, 364 (D.C. Cir. 2000) (agency decision that statute of limitations defense was forfeited by failure to raise argument until motion for reconsideration held not arbitrary or capricious); see also Tex. Mun. Power Agency v. Burlington N. Santa Fe Ry., STB Docket No. 42056, 2004 WL 2619767, 3 (STB served Sept. 27, 2004) (Board "generally does not consider new issues raised for the first time on reconsideration where those issues could have and should have been presented in the earlier stages of the proceeding"). In sum, BNSF's argument came too late to command the attention of the Board, let alone that of this court.

Still, BNSF protests, the statutory provision for "automatic" dismissal is a "mandatory directive" and therefore leaves no discretion to the agency to treat its claim as having been forfeited. Even a defect in the jurisdiction of an agency, however, when not timely raised before that agency is forfeit, see USAir, Inc. v. DOT, 969 F.2d 1256, 1259-60 (D.C. Cir. 1992) (challenge based upon 90-day deadline for agency action forfeit when not raised before agency), unless it "concerns the very composition or `constitution' of [that] agency," Mitchell v. Christopher, 996 F.2d 375, 378 (D.C. Cir. 1993), which BNSF's objection does not. Compare Arbaugh v. Y H Corp., ___ U.S. ___, ___, 126 S.Ct. 1235, 1244, 163 L.Ed.2d 1097 (2006) ("subject matter jurisdiction, because it involves the court's power to hear a case, can never be forfeited or waived") (citation omitted).

B. The Merits

Because the investigative proceeding initiated by Xcel's complaint was not dismissed, we shall go on to consider BNSF's arguments concerning the merits of Xcel's case. As usual, we review the Board's findings of fact for substantial evidence and ask whether its decision is "arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law," 5 U.S.C. § 706(2)(A), (E), bearing in mind that "[w]here an agency has rationally set forth the grounds on which it acted, . . . this court may not substitute its judgment for that of the agency," McCarty Farms v. Surface Transp. Bd., 158 F.3d 1294, 1301 (D.C. Cir. 1998). As detailed below, we find no fault with the Board's reasoning and therefore leave its decision undisturbed.

1. Revenue Adequacy

BNSF first argues the Board's decision to lower the carrier's rates when, according to the Board's own calculations, BNSF's revenues were not adequate to provide a reasonable return on its investment, violated the Board's statutory duty to look out for the adequacy of the carrier's revenues. See 49 U.S.C. § 10704(a)(2) (the Board "shall make an adequate and continuing effort to assist . . . carriers in attaining revenue levels" that are "adequate, under honest, economical, and efficient management, to cover total operating expenses, including depreciation and obsolescence, plus a reasonable and economic profit or return (or both) on capital"); see also id. § 10101(3) (policy in regulating railroad industry "to promote a safe and efficient rail transportation system by allowing rail carriers to earn adequate revenues"). In order to attain revenue adequacy, reports BNSF, it must be allowed to "charg[e] relatively high rates to coal shippers, like Xcel, with highly inelastic demand." See Coal Rate Guidelines, 1 I.C.C.2d at 526-27 (owing to significant production economies "the cost structure of the railroad industry necessitates differential pricing of rail services" based upon diverse shippers' sensitivities to price).

Although the Board explained that the SAC test "inherently addresses" a railroad's need for adequate revenues, see Decision II, at 6, BNSF argues the Board must "address the revenue adequacy mandate in the context of individual cases." The Board responds that it is charged with seeking not only adequate revenues for carriers but also reasonable rates for shippers, see 49 U.S.C. §§ 10101(6), 10702, and that it seeks both via the SAC test, which is designed to "accommodate the[se] dual objectives" by assuring "captive shippers that they are not cross-subsidizing other parts of the defendant's network, while assuring railroads that any [given] rate prescription will provide a reasonable return on the replacement of facilities needed to serve the shipper."

The Board is on solid ground here. Regardless whether BNSF as a system is revenue-adequate, system-wide revenue inadequacy is not a basis upon which a carrier may defend an unreasonable rate over a segment of its system. See Coal Rate Guidelines, 1 I.C.C.2d at 536 ("[A] rate may be unreasonable even if the carrier is far short of revenue adequacy"). As the Board explained in denying BNSF's petition for reconsideration, the SAC test is designed to take into account the railroad's need for revenue adequacy "on the portion of its system that is included in the system of the SARR." Decision II, at 6; see also Burlington N.R.R. v. ICC, 985 F.2d 589, 597 (D.C. Cir. 1993) ("CMP explicitly builds in the idea of revenue adequacy (subject to the SAC constraint)"). The test therefore reasonably "excludes revenue needs associated with other traffic" traveling over other parts of the system. Decision II, at 6. To be sure, a railroad may still charge a captive shipper more than it charges noncaptive shippers for the use of shared facilities. The SAC test, however, is designed to ensure the carrier does not cross-subsidize revenue-inadequate portions of the system by charging its captive shippers "more than they should have to pay for efficient rail service," Coal Rate Guidelines, 1 I.C.C.2d at 524, and thereby recovering from them "the costs of . . . facilities or services from which [they] derive no benefit," id. at 523.

Nor are we persuaded by BNSF's argument that it was arbitrary and capricious for the Board to lower its rates below the rates indicated by the Board's Revenue Shortfall Allocation Method (RSAM). The RSAM is the Board's way of calculating the average percentage by which revenues received from captive shippers must exceed the variable costs (R/VC) of serving those shippers if the railroad is to achieve revenue adequacy. See Rate Guidelines — Non-Coal Proceedings, 1 S.T.B. 1004 (1996); see also Ass'n of Am. R.Rs. v. Surface Transp. Bd., 146 F.3d 942, 944-45 (D.C. Cir. 1998). BNSF contends the rates prescribed by the Board in this case yield a R/VC ratio of not more than 273%, significantly below its RSAM figure of about 316% for the relevant time period. Because a railroad in order to cover its fixed costs must be allowed to charge its highest rates to shippers with the least elastic demand, BNSF claims it must be able to charge them rates equal to or greater than the rates indicated by the RSAM in order "to have any chance of achieving revenue adequacy."

The RSAM figure is not dispositive, however, it is but one of three "benchmarks" established by the Board for use in smaller rate proceedings, where a full-blown SAC analysis would be prohibitively expensive. See Ass'n of Am. R.Rs., 146 F.3d at 944-45. As the Board points out, the RSAM figure merely provides a test of "system-wide revenue need" and therefore "provides no guidance on the rates Xcel should be charged for the particular facilities and services Xcel uses." In contrast, the Board has "consistently affirmed that CMP, with its SAC constraint, is the preferred and most accurate procedure available for determining the reasonableness of rates in markets where the rail carrier enjoys market dominance." Burlington N.R.R., 985 F.2d at 596 (internal quotation marks and citation omitted). Of course, a railroad does need to recover a higher percentage of its fixed costs from shippers with relatively inelastic demand, but that is not to say it may charge a price that cross-subsidizes other shippers. The SAC test constrains rates precisely to that end. As a result, where fixed costs are relatively low, even a shipper with inelastic demand may be charged less than the average derived by the RSAM; indeed, this will inevitably occur to some extent because the average derived by the RSAM is the average for captive shippers only, so the ratios for some captive shippers must be above and some below that figure. Here, the rates prescribed by the Board were well in excess (indeed, 273%) of variable cost.

In sum, BNSF has not shown us that the Board arbitrarily applied its SAC test. We will not disturb its decision on this ground.

2. Cross-over Traffic

BNSF next objects to the heavy reliance of Xcel's SARR upon cross-over traffic. According to the railroad, the allocation of revenues between the SARR and the off-SARR portions of a through movement "is distorted because a railroad does not charge rates for discrete portions of a through movement," as a result of which there must be "an arbitrary allocation of through revenue between the two portions of the through movement." Likewise, the "cost side of the comparison is distorted because the costs of the off-SARR portions of the movements are ignored altogether." Therefore, contends BNSF, a properly performed SAC test must use only end-to-end movements and must "examine the full costs of all facilities used to provide service to the shipper group and the total revenues generated by that traffic."

In response the Board acknowledges that the use of cross-over traffic "introduces . . . imprecision into the SAC analysis" but argues that excluding such traffic would "risk being intractable." Decision I, at 16. The SAC analysis must reflect the cost sharing and production economies derived from sharing facilities on the SARR. See Decision I, at 14 (quoting Coal Rate Guidelines, 1 I.C.C.2d at 544: "Without grouping, SAC would not be a very useful test, since the captive shipper would be deprived of the benefits of any inherent production economies"); Nevada Power, 10 I.C.C.2d at 265 n. 12. Therefore, to exclude cross-over traffic from a SAC analysis "would dramatically enlarge the geographic scope of a SARR" needed to serve the group of shippers selected for the SARR by the complainant. Decision I, at 14. In this case, the Board estimated that in order "to serve the same 37 shippers without any cross-over traffic, the SARR would need to be at least 10 times larger than the [SARR that Xcel proposed]." Id. The complexity of the consequent proceeding, the Board concluded, "would expand exponentially" beyond what is already "a dauntingly large and detailed task." Id. at 16.

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. See Burlington N.R.R., 985 F.2d at 597 ("the Commission is free to make reasonable trade-offs between the quality and cost of possible regulatory approaches. . . . and of course we owe the Commission's judgment on the point great deference" so long as it "intelligibly explained why the trade-off chosen was reasonable").

Here, the Board's explanation for its decision to allow cross-over traffic as a simplifying mechanism, which the Board has described as "now a standard feature of SAC cases," Decision I, at 17, was both reasonable and intelligibly explained. The Board must balance, among other concerns, the need for a reasonably accurate methodology and the need to avoid unduly protracting already complex and expensive SAC proceedings. See 49 U.S.C. § 10101(15) (Board must provide for "expeditious handling and resolution of all proceedings"); Ass'n of Am. R.Rs. v. Surface Transp. Bd., 306 F.3d 1108, 1111 (D.C. Cir. 2002) ("[I]t is up to the Board to arrive at a reasonable accommodation of the conflicting policies set out in the Staggers Act"). In view of the Board's estimate that presentation and analysis of the SARR, which already involved "dozens of volumes of evidence," would have burgeoned tenfold without the simplifying mechanism of cross-over traffic, Decision I, at 16, it was not unreasonable for the Board to conclude that barring cross-over traffic from the SARR would be not only inefficient but infeasible. See Decision II, at 7 ("We remain concerned that, without cross-over traffic, captive shippers could lack a practicable means by which to prosecute rate complaints").

Our view of this matter might be different if BNSF had presented evidence to establish that the imprecision implicit in the use of cross-over traffic tends to overestimate the revenues generated by a SARR to a degree that outweighs any efficiency gains. Instead, lacking such evidence, we are struck by the irony of BNSF calling for a dramatic increase in the complexity of the SAC proceeding even as it argues the case should be dismissed because the Board failed to resolve it more speedily.

In sum, we do not think the Board unreasonably concluded that the "value of this evidentiary tool outweighs its limitations." Decision II, at 7.

More persuasive, but also ultimately unconvincing, is BNSF's argument that the specific method by which the Board allocates revenue to cross-over traffic is flawed. The appropriate allocation of revenue from cross-over traffic is a perennial issue in SAC proceedings and one the Board even now has not resolved definitively. See, e.g., PPL Mont., LLC v. Burlington N. Santa Fe Ry., STB Docket No. 42054, 2002 WL 1905118, 7 n. 14 (STB served Aug. 20, 2002) ("We have not adopted a single preferred procedure for developing revenue divisions on cross-over traffic"). The Modified Straight-Mileage Prorate (MSP) procedure, which was applied in this and several other recent proceedings, "is a refinement of a mileage-based formula long used in SAC cases to allocate cross-over traffic revenues." Decision II, at 8; see, e.g., Duke Energy Corp. v. CSX Transp., Inc., STB Docket No. 42070, 2004 WL 250254 (STB served Feb. 4, 2004); Duke Energy Corp. v. Norfolk S. Ry., STB Docket No. 42069, 2003 WL 22673026 (STB served Nov. 6, 2003); Carolina Power Light Co. v. Norfolk S. Ry., STB Docket No. 42072, 2003 WL 23109610 (STB served Dec. 23, 2003). Under MSP, revenues from a movement are allocated to the SARR based upon the movement's "proportionate share of the combined mileage" and upon the assumption that "average costs are a continuous function of distance (holding other factors constant)." Decision II, at 8. In recognition of the proportionally higher costs associated with originating and terminating traffic, for each movement that originates or terminates on the SARR, a 100-mile additive is included in the calculation "as a surrogate in the absence of any better evidence as to the costs of those functions." Id.

BNSF criticized the MSP approach for its failure to take into account economies of density, that is, the principle that as the density of traffic increases over a stretch of rail, average costs diminish, see Coal Rate Guidelines, 1 I.C.C.2d at 526, at least initially. BNSF therefore proposed an alternative method it called the "Density Adjusted Revenue Allocation" (DARA). Under this approach, revenues are allocated between the SARR and off-SARR segments of a cross-over movement "in proportion to each segment's relative variable cost, distance, and density." Decision I, at 17. The Board rejected DARA because, although it does allocate a higher proportion of revenues to lower density lines, it "ignor[es] the well-accepted principle that economies of density will vary with different levels of output." Decision II, at 8-9. Thus, even where "economies of density have been, for practical purposes, exhausted, DARA would continue to allocate greater revenue to the part of the movement using the lighter-density line." Id. at 11. The Board therefore concluded that DARA had "not been shown to be superior" to the MSP approach ordinarily used in SAC cases. Decision II, at 11.

Although we take BNSF's point that the MSP method of allocating revenue to cross-over traffic does not take into account economies of density, we believe the Board gave an adequate reason for rejecting the DARA method, namely, its failure to take into account the diminishing nature of those economies. Each method has a limitation and, faced with a choice between them, the Board reasonably stayed on the course it had long ago adopted. See Atchison, Topeka Santa Fe Ry. v. Wichita Bd. of Trade, 412 U.S. 800, 807-08, 93 S.Ct. 2367, 37 L.Ed.2d 350 (1973) ("A settled course of behavior embodies the agency's informed judgment that, by pursuing that course, it will carry out the policies committed to it by Congress. There is, then, at least a presumption that those policies will be carried out best if the settled rule is adhered to").

BNSF correctly points out that the Board has not adopted a "single preferred procedure," PPL Mont., at 7 n. 14, but we see that it has for more than a decade used a mileage-based allocation of revenue. In its order denying rehearing in this case, the Board recognized there "may well be a better revenue allocation procedure that could be practical for SAC cases" and invited proposals, whether submitted in future rate proceedings or as requests for rulemaking. See Decision II, at 11. Were 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 its MSP model would be on shaky ground indeed. But that day is yet to come.

3. Challenges to Xcel's Evidence

BNSF also challenges the Board's reliance upon certain evidence in Xcel's SAC presentation. As detailed below, we do not find its arguments persuasive.

a. Operating Plan

BNSF contends the Board should have dismissed Xcel's complaint either (1) when BNSF, in its motion to dismiss, identified what it described as "obvious, elementary, and fundamental errors" in Xcel's operating plan, which assumed trains would travel at unrealistic speeds on the SARR, or (2) in its decision on the merits, when the Board instead substituted BNSF's proposed operating plan, with slower estimated train speeds and higher estimated costs, for Xcel's flawed plan.

First, the Board denied BNSF's threshold motion to dismiss after reviewing Xcel's SAC presentation and concluding the errors in Xcel's operating plan, upon which the motion to dismiss was based, appeared to be "readily correctable without a significant redesign of the SARR" and were not "so large in magnitude or so egregious as to warrant dismissing the complaint" at that early stage. Pub. Serv. Co. of Colo. d/b/a Xcel Energy v. Burlington N. Santa Fe Ry., STB Docket No. 42057, 2003 WL 1788666, 2 (STB served April 4, 2003). The Board later explained that Xcel had made out a prima facie case by virtue of its "good faith effort to present reasonable evidence on all of the basic components of the SAC test." Decision II, at 6. BNSF argues that by permitting Xcel to proceed with an admittedly flawed operating plan, the Board relieved Xcel of its burden of proving every element of its claim.

Although Xcel does bear the burden of persuasion, see Coal Rate Guidelines, 1 I.C.C.2d at 547 ("[T]he complainant must demonstrate that the challenged rate is unreasonable"), the complainant's initial presentation need not be flawless in order to resist a motion to dismiss. See 49 U.S.C. § 11701(b) (requiring dismissal of the complaint if it does not state "reasonable grounds for investigation and action"); McCarty Farms v. Burlington N., Inc., ICC Docket No. 37809, 1995 WL 55449, 8 (ICC served Feb. 1, 1995) ("Unless the model is patently incapable of meeting the shipper's needs, we will presume that the stand-alone system is feasible unless and until its feasibility is challenged in the railroad's case-in-chief"). Because of the sheer size of a SAC presentation, it will almost inevitably have some flaws to which the carrier can point. Therefore, we do not think the Board unreasonably refused to dismiss Xcel's complaint merely because its presentation was less than perfect. See Decision II, at 5 ("Were we to entertain only those rate complaints where the railroad could not poke holes in the operating plan devised by the shipper for its SARR, almost every rate challenge [would have to be dismissed]").

Second, the Board's substitution for Xcel's flawed operating plan of a modified version of BNSF's own plan, did not relieve Xcel of the need to prove BNSF's rates were unreasonable. Rather the Board concluded that Xcel could and did meet its burden by using evidence submitted by (and more favorable to) BNSF. So long as the record supports that conclusion, BNSF has no cause to complain about the source of the evidence. Cf. Consol. Edison Co. v. FERC, 165 F.3d 992, 1008 (D.C. Cir. 1999) ("[T]he burden of proof requirement . . . relates to the burden of persuasion . . ., not to the burden of production, and thus the identity of the party submitting evidence is not dispositive").

b. Rerouting of Jeffrey Traffic

BNSF also argues the Board should have excluded the largest movement on the SARR — the movement of coal to Western Resources' Jeffrey plant, which currently moves on a shorter and less congested route — because Xcel did not submit competent evidence that the rerouting was "reasonable and would meet the shipper's transportation needs." Tex. Mun. Power Agency v. Burlington N. Santa Fe Ry., STB Docket No. 42056, 2003 WL 1523335, 21-24 (STB served Mar. 24, 2003). The Board used the data in BNSF's operating plan to compare the travel times and lengths of the two routes and concluded they would provide comparable service; it also added $150 million for additional capital investment in order to cover the costs of any congestion created by moving the traffic on the SARR. Decision I, at 20-22, 30. In so doing, the Board, we think, reasonably applied its own expertise to fill a minor gap in the record. See Balt. Ohio R.R. v. United States, 386 U.S. 372, 430, 87 S.Ct. 1100, 18 L.Ed.2d 159 (1967) (Board "is not the prisoner of the parties' submissions" but rather has a duty "to weigh alternatives and make its choice according to its judgment how best to achieve and advance the goals of the National Transportation Policy") (Brennan, J., concurring).

c. EIA Rate Forecast

Finally, BNSF objects to the Board's reliance, in estimating revenues available to the SARR, upon a rate forecast produced by the Energy Information Administration (EIA) of the United States Department of Energy in preference to either of the forecasts proffered by the parties. BNSF, invoking "[f]undamental principles of administrative law" and due process, argues it was entitled to advance notice that the Board would take official notice of extra-record evidence so it could "parry its effect." See Union Elec. Co. v. FERC, 890 F.2d 1193, 1202-04 (D.C. Cir. 1989) (citation omitted). BNSF protested generally the use of the EIA data in its motion for reconsideration, but unlike the petitioner in Union Electric, did not make "a good showing it [could] contest the evidence." 890 F.2d at 1203 (citing Market St. Ry. v. R.R. Comm'n, 324 U.S. 548, 562, 65 S.Ct. 770, 89 L.Ed. 1171 (1945)). In fact, the carrier failed to identify any flaw in the evidence or even to request an additional opportunity in which to do so.

Due process requires only a "meaningful opportunity" to challenge new evidence, Mathews v. Eldridge, 424 U.S. 319, 349, 96 S.Ct. 893, 47 L.Ed.2d 18 (1976), which opportunity BNSF failed to take in its application for rehearing. Cf. Opp. Cotton Mills, Inc. v. Adm'r of Wage Hour Div., 312 U.S. 126, 152, 61 S.Ct. 524, 85 L.Ed. 624 (1941) ("The demands of due process do not require a hearing, at the initial stage or at a particular point or at more than one point in an administrative proceeding so long as the requisite hearing is held before the final order becomes effective"); Gutierrez-Rogue v. INS, 954 F.2d 769, 773 (D.C. Cir. 1992) (an opportunity to rebut officially noticed facts satisfies due process). We have no occasion, therefore, to overturn the Board's decision on this ground.

III. Conclusion

For the foregoing reasons, BNSF's petition for review is

Denied.

APPENDIX