The Staggers Rail Act gave railroads substantial freedom to set rates but restricted this freedom for common carrier rates when the service is supplied in markets lacking “effective competition.”1 These rates are not regulated directly, but they can be challenged by a shipper after the fact. The law states that in markets in which a railroad has “market dominance,” its common carrier rates must be “reasonable.”2 Market dominance is defined as the absence of effective competition from other railroads or modes of transportation.3 A rate is automatically considered reasonable if it does not exceed 180 percent of its “variable cost,” as determined by the Surface Transportation Board (STB).4 If a disputed rate exceeds this percentage and is found to be in a market lacking effective competition, STB can rule on whether the rate is reasonable.5 If STB finds the rate to be unreasonable, it must order the railroad to compensate the shipper for overpayments, and it may prescribe the maximum rate the railroad can charge for future movements.6
In ruling on the reasonableness of a rate, STB is directed to be respectful of the law’s overarching policies (see Box 1-2),7 including the policy that railroads must be able to earn “adequate revenues.” Adequate revenues are defined as those “sufficient—under
1 49 USC §10101(4).
2 49 USC §10701(d)(1), §10702.
3 49 USC §10707(a).
4 49 USC §10707(d)(1)(A).
5 When a complaint is filed, STB may investigate the reasonableness of the challenged rate or dismiss the complaint if the complaint does not contain reasonable grounds for investigation and action [49 USC §10704(b), §11701(a), 11701(b)].
6 49 USC §11704(b), §10704(a)(1).
7 49 USC §10101.
honest, economical, and efficient management—to cover operating expenses, support prudent capital outlays, repay a reasonable debt level, raise needed equity capital, and otherwise attract and retain capital in amounts adequate to provide a sound rail transportation system.”8
In 1995, when Congress last amended the Interstate Commerce Act (ICA) to terminate the Interstate Commerce Commission (ICC) and create STB, it added a new policy calling for the “expeditious handling and resolution of all proceedings.”9 It further instructed STB to ensure the prompt handling of rate challenges in particular by adopting appropriate measures for “avoiding delay in the discovery and evidentiary phases” of proceedings and by establishing “a simplified and expedited method for determining the reasonableness of challenged rail rates in those cases in which a full stand-alone cost [SAC] presentation is too costly, given the value of the case.”10
Thus, STB’s implementation of the law’s rate relief provision involves the following three steps:
- Estimate the variable cost of a priced unit of traffic to determine whether its rate exceeds the 180 percent statutory threshold.
- Determine whether a market subject to a rate challenge lacks effective competition and qualifies as being dominated.
- Establish standards for determining whether a disputed rate that passes these eligibility screens is unreasonable and the shipper is entitled to relief.
The methods used by STB in implementing each of the three steps are assessed in this chapter. Most originated with ICC and have been modified and added to by STB over the past 20 years. Before the methods are assessed, a brief review of their origins and the historical concerns that shaped them is provided. In addition, how the steps interrelate is explained. For example, the law’s “revenue-to-variable-cost” (R/VC) formula determines initial eligibility to challenge a rate.
8 49 USC §10701(d)(2), §10704(a)(2).
9 49 USC §10101(15).
10 49 USC §10704(d), §10701(d)(3).
Because this is a highly imperfect screen, regulators have come to depend on market dominance inquiries and elaborate cost calculations to ensure that rate relief is granted in a measured way that does not conflict with the law’s interest in protecting revenue adequacy.
After the STB methods for granting rate relief are reviewed, consideration is given to alternative approaches for implementing the law’s maximum rate protections. The rate dispute resolution process used in Canada is examined, and an alternative method for identifying unusually high rates is described.
Rate Reasonableness: Efficiency, Fairness, and Revenue Adequacy
As discussed in Chapter 1, the policy interest in “reasonable” rates dates back to the ICA of 1887 and even further to the common law principles that shaped the legal doctrine of common carriage. Railroads were subject to the long-standing common law duty to offer “just and reasonable rates” and to respond to all reasonable requests for transportation service without “discrimination” (Scharfman 1915, 191). The ICA specifically prohibited as unjust discrimination any preferential treatment of a “like or contemporaneous service in the transportation of a like kind of traffic under substantially similar circumstances and conditions” (Scharfman 1915, 117). Efforts to bring about a “just and reasonable” railroad pricing system led to requirements, including the posting of generally applicable tariff rates, and to rate structures that limited the ability of railroads to charge shippers of the same commodity rates that differed according to competitive circumstances.
Over time, regulatory prohibitions on varying rates in response to demand contributed substantially to railroads falling short of the revenues needed to maintain their capital-intensive systems and to the loss of large amounts of traffic to trucks. Hence, Congress made it clear to ICC that the Staggers Rail Act’s protections from unreasonable rates should not be interpreted as an opportunity to reregulate rates and hinder the ability of railroads to earn adequate revenues.
This [rate reasonableness] provision sets forth for the first time a standard for the Commission to use in determining if a rate is reasonable, and that standard goes to ensuring that railroads can continue to operate as private enterprises. Previous admonitions by the Congress that the Commission assist carriers in earning adequate revenue levels have not achieved their goals. As a result, the Committee is establishing a more straightforward mandate. This is a clear directive to ensure financially sound railroads, and the Commission is not to misuse the term “reasonable” to circumvent this directive.11
Before the Staggers Rail Act, rail rates had been kept too high in many markets in which shippers had nonrail competitive options; this caused potentially profitable traffic to be priced out of the rail market. By allowing railroads to adjust rates according to demand, the act ended this inefficiency of the previous regulatory system.12 In making an exception to these pricing freedoms, the act’s maximum rate provisions required regulators to determine what constitutes an “unreasonable” rate in markets in which a railroad lacked effective competition and could exercise significant market power. However, even in noncompetitive rail markets, concern over high rates causing efficiency losses should be minimal, for reasons that are explained next.
A traditional rationale for regulating prices in markets in which firms have substantial market power is the prevention of inefficiencies caused by supracompetitive pricing. Typically, a monopolist that is unable to price discriminate will raise prices generally to levels that maximize profits from customers with high willingness to pay. In the process it will price some other customers with a lower willingness to pay out of the market even though they could have been served profitably with a lower price. The priced-out traffic represents an efficiency loss. However, as
11 H.R. 1035, 96th Congress, Second Session, 54 (1980), as cited by ICC 1985, 10–11.
12 By keeping rates artificially high and thus pricing some traffic out of the market, the regulatory system created inefficiencies as the affected shippers turned to second-best transportation options, reduced their output, or took other action more costly than would have occurred if railroads had been allowed to adjust their rates to lower but still profitable levels.
discussed in Chapters 1 and 2, the Staggers Rail Act gave railroads the freedom to contract with shippers for service and ended the requirement that railroads post tariff rates that apply to all shippers. In markets where it had substantial market power, a railroad could therefore raise tariff rates for shippers with the fewest options and the highest willingness to pay without concern for losing price-sensitive customers, who could be retained by negotiating discounted contract rates. Because little traffic would be priced out of the market, efficiency losses would be minimal.
Price discrimination, if applied perfectly, leads to no efficiency losses. While contracting may not provide railroads with the opportunity to engage in perfect price discrimination, the ability to negotiate rates with individual shippers should be sufficient to ensure that almost all profitable traffic is served.13 Accordingly, there can be little or no efficiency rationale for the law’s requirement that tariff rates be kept reasonable. Other purposes for the maximum rate protections in the Staggers Rail Act might therefore be surmised. The most obvious would be conformance to the common law principle, discussed above, that seeks to promote fairness by protecting shippers who lack competitive options from paying tariff rates that are unusually high compared with those of similar shippers who have more options.14
As a practical matter, ICC was thus charged with developing a standard for rate reasonableness that was concerned with fairness, not efficiency. In fulfilling this charge, the agency had to respect the law’s interest in revenue adequacy by ensuring that rate relief was not granted so liberally that railroads would once again fall short of their revenue needs. Thus, in introducing its first set of guidelines for ruling on rate reasonableness, the 1985 Coal Rate Guidelines,15 ICC set a precedent whereby all evidentiary standards for judging the fairness of a rate would need to be linked, at least ostensibly, to the statutory goal of ensuring revenue adequacy. This link was stated explicitly in the
13 As noted in Chapter 1, Grimm and Winston (2000) have quantified minimal deadweight losses from rates charged to captive shippers.
14 Concern over fairness is predominant in the field of rate regulation. Schmalensee (1979) has observed that to the extent that utility regulators in the United States have been concerned with rate structures, they have tended to be motivated and informed by considerations of equity or fairness rather than efficiency.
15 ICC 2d 520, 1985 WL 56819 (ICC), August 8, 1985.
guidelines: “The maximum rate guidelines we are adopting here culminate several years of research and effort to develop an economically efficient and equitable methodology for determining the reasonableness of rates. … We believe [the guidelines] will provide the necessary protection for captive shippers, while providing railroads the opportunity to earn adequate revenues” (ICC 1985, 2).
Thirty years have passed since the Coal Rate Guidelines were introduced, and STB has remained committed to linking rate fairness and revenue adequacy. Many shippers have been critical of the evidentiary standards used to make this linkage. They believe that the standards have unnecessarily precluded access to the law’s maximum rate protections. Railroads generally defend the linkage as crucial in safeguarding their ability to continue producing and reinvesting.
Exacting Standards for Rate Reasonableness and the R/VC Formula
Shipper criticisms of STB’s granting of rate relief have tended to center on the procedures used in making qualitative assessments of market dominance and in ruling on the reasonableness of rates by using evidentiary standards characterized as being excessively costly and burdensome. However, any critical assessment of these procedures must begin by examining their respective roles in the overall process for granting rate relief. Market dominance and rate reasonableness inquiries are the second and third steps in a system that begins with a shipper having to show that its rate exceeds the statutory threshold of 180 percent of the shipment’s variable cost. Because of the congressional directive that regulators not grant rate relief so permissively that it threatens railroad revenue adequacy, the confidence that regulators have in this statutory formula as a screen for limiting rate relief cases is critical. In the absence of such confidence, the introduction of exacting and imposing standards for assessing market dominance and rate reasonableness might be expected. Indeed, the 180 percent formula is an unreliable screening tool for reasons explained and documented next. That unreliability, in turn, may have prompted regulators to institute complex evidentiary procedures, such as a SAC test, that function as the primary safeguard for railroad revenue adequacy.
Over the past two decades, STB has tried with limited success to address shipper concerns about the accessibility of the rate relief process by making piecemeal changes in the procedures used during the latter stages of the process. System-level reforms that place greater emphasis on the role of the R/VC standard have been largely neglected, although its significance suggests that reforms centered on it are an essential first step.
When the Staggers Rail Act introduced the R/VC formula for screening rates for relief eligibility, ICC had long been using accounting-based cost allocation systems for costing railroad services and activities.16 The law directed the agency to develop an updated method to determine “economically accurate railroad costs directly and indirectly associated with particular movements of goods, including the variable costs associated with particular movements.”17 To comply, ICC developed the Uniform Railroad Costing System (URCS), which shares a methodological approach with earlier cost accounting schemes and remains in use today.18 For reasons explained next, estimation of variable cost with the URCS or any other cost allocation scheme is inherently problematic. No cost allocation scheme can yield economically valid relationships for assessing a railroad’s rate levels or market power.
Insoluble Problem of Allocating Variable Costs
The Staggers Rail Act did not provide regulators with clear guidance on how a “variable cost” should be defined or computed. The law refers to the need to establish variable costs for “particular movements.” That requirement may seem straightforward, but railroads produce many kinds of freight service. The services, or products, vary in many dimensions that are relevant to their individual pricing, such as volume,
16 The Staggers Rail Act, §10705a(m)(1), required ICC to determine variable costs by using its Rail Form A costing method or to adopt an alternative method.
17 Cost accounting principles in Title III, Section 301, §11162 of the Staggers Rail Act of 1980.
18 ICC decided that Rail Form A’s data structure and statistical techniques did not reflect the operation of the modern railroad industry (STB 2010, 2).
location, commodity, and time of travel. A coal unit train priced at one time of year for one routing in a railroad’s network is a product different from a coal unit train priced at another time of year for another routing. Inasmuch as the R/VC formula was meant to inform decision making, regulators need to know how product specific and precise a variable cost estimate must be to achieve its purpose. For example, they need to know whether systemwide averages would suffice or whether a variable cost must be fully traceable to the specific unit of traffic to which the rate applied.
Furthermore, whether the term “variable cost” was meant to be the incremental cost of a priced shipment (e.g., the added fuel use or wear and tear that one additional shipment creates) or something else is not clear. Regulators collect large amounts of expense data from railroads. However, they do not collect expense data at the shipment-specific level. Regulators have traditionally used the expense reports to assign costs to segments of rail traffic, essentially by dividing total reported expenses (or some subset of these expenses) among portions of a railroad’s output. Of course, railroads incur many costs that are unreported in expense ledgers, both at the firm- and the movement-specific level, such as system congestion and delays that are added by the operation of another train or the switching of another car. It is often argued that the omission of such unreported costs from expense-based cost allocation schemes can be misleading because railroads must take all costs into account. While that is true, minimizing such reporting omissions would not make the cost allocations substantially more valid or relevant, for reasons that become evident when the cost allocation process itself is reviewed.
The cost allocation process begins by characterizing a railroad’s total expenses as being more or less variable with levels of traffic output.19 These characterizations are made by observing how a railroad’s total expenses change as a function of total traffic. For example, 95 percent of fuel expenses and 50 percent of road maintenance costs may be characterized as variable with traffic output. In performing this exercise, the time period selected is critical. A decline in traffic output
19 The common use of cost allocation methods for rail regulatory purposes has been well documented. An early study of the processes was conducted by Meyer (1958).
of 10 percent in a month, for example, would likely lead to the finding that a railroad’s expenses changed less than if traffic had declined by 10 percent from one year to the next. That is because individual railroad cost items can be adjusted differently over time. For example, labor costs may be fixed from week to week but variable from year to year. Thus, the time period used for the “variable-fixed” allocation will have a significant impact on the items defined as variable and thus on the total amount of variable costs to be allocated across traffic. Because the time period is likely to be based solely on when expense data are collected, there will be an inherent arbitrariness in the distinction between fixed and variable costs.
The cost allocator thus has to establish rules for dividing the subset of a railroad’s reported expenses characterized as “variable” for an arbitrarily determined time period. These divisions must then be assigned to units of traffic that are inevitably going to be crude versions of the actual priced units of traffic. In the process, many relevant costs could be omitted because they are not recorded on expense ledgers. The vagaries, omissions, and arbitrariness of these steps alone are intimidating. The process of dividing costs, however, is even more problematic and adds another large degree of arbitrariness to the process. The reason is that relatively few costs that a railroad incurs can be directly attributed to specific units of traffic. With good data, the cost allocator may be able to attribute a few costs to a specific movement—for example, the movement’s incremental contribution to the fuel used by a train or wear and tear on specific equipment. However, there will be few opportunities for such unambiguous attributions because most railroad costs are shared by all traffic or multiple groupings of traffic, such as those associated with operating and maintaining track, locomotives, terminals, and yards. Division of these cost items in any economically meaningful way among individual units or narrow segments of traffic is simply not possible.
The common cost problem arises even in allocating costs associated with the movement of an individual train. Consider a train containing a 10-carload shipment of wheat and a 10-carload shipment of coal. The total variable cost of the train may be readily defined as all of the costs such as crew wages and fuel that can be avoided by not
operating it.20 The incremental cost of each set of cars is also readily defined. It is the difference between the cost of operating the train with and without each set—for example, the fuel saved by not having to move as much weight. However, the locomotive must be used even if only one set of cars is moved. Some of its operating costs, such as crew wages, are included in the train’s total variable cost but not in the incremental cost of each set of cars. Any allocation of the total variable cost of the train among individual shipments must divide all the operating costs of the locomotive among the shipments, a division that has no basis in fact and is inherently arbitrary.
In summary, regulators face insurmountable challenges in estimating a variable cost to compare with the rate charged for a unit of traffic. Nevertheless, their fundamentally arbitrary cost allocation results have meaning because they are used in making regulatory decisions, most significantly in calculating the R/VC percentages for assessing eligibility for rate relief. The problems arising from use of the URCS for this purpose are described next.
Fundamental Flaws of URCS
ICC introduced URCS with the idea of defining a railroad’s fixed and variable costs more precisely by dividing expenses into more discrete categories and defining units of traffic more narrowly, such as by car type, shipment size, and length of haul.21 Regressions were run to determine whether certain cost items were more or less fixed with respect to traffic volume changes (e.g., by showing that annual fuel used to run locomotives was 96 percent variable with traffic). Expert judgments were made about certain other cost relationships (e.g., that 50 percent of a railroad’s annual expenditures on capital and road property should be treated as variable with respect to traffic levels) (STB 2010, 5). Engineering studies conducted over the past 50 years were consulted to help allocate certain costs more precisely, such as
20 There may be variable costs of activities that support both this train and other trains; allocating them between this and other trains would be inherently arbitrary. This complication is ignored, and the train’s total variable cost is assumed to be well defined.
21 The expense groups and methods used to allocate portions to traffic are described more fully by Wilson and Bitzan (2003).
the fuel used in switching a car (STB 2010, 4). URCS would even allow regulators for the first time to assign lower costs to shipments moved in unit trains and multicar allotments because of their added efficiencies. URCS would simply spread the cost savings, through a so-called “make-whole adjustment,” over all other traffic because it would be essential for all of the railroad’s costs that have been declared “variable” to be fully assigned.
URCS suffers from all of the methodological problems cited above. It is a cost allocation scheme that has no economic foundation, as amply illustrated by the “make-whole” contrivance to redistribute unallocated costs. Its refinements relative to earlier cost allocation schemes have done nothing to make the results any more reliable or less arbitrary because the large majority of cost items characterized by URCS as “variable” are clearly not variable (e.g., road property) with respect to priced units of traffic. Indeed, STB characterizes results from URCS as being “systemwide averages,” acknowledging that they do not reflect the actual cost of providing any specific service. The results from URCS cannot be represented as meeting the law’s requirement for economic accuracy, and they cannot be portrayed as having any relevance to the price charged for a given unit of traffic as implied by their use in the law’s R/VC formula. STB’s own Railroad–Shipper Transportation Advisory Council has referred to URCS as “an outdated and inadequate costing system.”22
Nevertheless, STB has stated that the results of URCS are sufficient for regulatory purposes: “Though imperfect, URCS has served as the agency’s costing tool for more than two decades and has produced costs sufficiently reliable for the Board to make regulatory determinations” (STB 2010, 1). The source of this confidence is unclear. In responding to a congressional inquiry about URCS, the agency pointed out that “there is no accounting process that can precisely attribute costs to particular movements” (STB 2010, 1) and that the “URCS system-wide average could be higher or lower than the actual cost of any particular movement” (STB 2010, 4). As STB further explained, “a railroad uses its physical assets (e.g., rail lines, locomotives, rail cars, yard equipment) to
transport hundreds of different commodities between many different locations. Thus, there are many common costs (akin to overhead) that the railroad will seek to recover from all of its customers. By necessity, the methodology must incorporate assumptions and generalizations about railroad operations, some of which may not reflect individual situations” (STB 2010, 1). Yet in conceding these deficiencies, STB overlooks their significance by acknowledging that “the role of URCS is to estimate that portion of the variable costs of providing rail service that can be attributed to any given [emphasis added] rail movement” (STB 2010, 1).
Evident Problems with Using URCS in the R/VC Formula
The arbitrary results produced by URCS are treated by regulators as if they were economically valid and are used for many regulatory purposes. In addition to being used in screening traffic for rate relief eligibility according to the R/VC formula, URCS is used in subsequent procedures to determine market dominance, to make assessments of whether a challenged rate is reasonable (by estimating the profitability of “crossover” traffic in SAC tests, as described later), and, if necessary, to prescribe the maximum tariff rate a railroad may charge. URCS is also used in measuring avoidable costs when a railroad applies to abandon a line and in calculating compensation fees for mandated access (STB 2010, 6–8).
The results of URCS are often used misguidedly by others. For example, in 2006 the U.S. Government Accountability Office (GAO) examined trends in shipments having rates with various R/VC percentages to determine whether railroads were obtaining and exercising more market power over time (GAO 2006). In finding that the share of traffic having R/VCs above 180 percent had dropped from 1985 to 2004, GAO surmised that the market power of railroads had been declining. Coincidental with these findings, however, GAO found that the amount of traffic having R/VCs exceeding 300 percent had increased from 4 to 6 percent, which caused the agency to question whether railroads were becoming more effective in exploiting market power when they possessed it (GAO 2006, 43). Fundamentally dependent on URCS-derived costs, such expanded uses of the R/VC formula have no basis in fact.
The reasons Congress had for introducing the R/VC formula or choosing 180 percent as the threshold below which rates would be immune from challenge are debatable.23 Such a formula is logically coherent only if the variable cost term has a connection to a railroad’s incremental cost of transporting a shipment and thus bears some resemblance to the rate that would be charged in a competitive market. A rate with a high R/VC percentage might be viewed as a sign that a railroad is exercising significant market power. However, estimates of variable cost that are derived in a highly arbitrary manner—in ways described above—cannot be expected to have a tenable or stable relationship to these incremental costs, and thus they can offer no meaningful insight into market power when they are compared with a shipment’s rate. Box 3-1 indicates how a number of cost allocation rules that might be viewed as reasonable can yield substantially different R/VC percentages, which would cause shipments to fall below or exceed the 180 percent threshold.
In its STB-sponsored study, Laurits R. Christensen Associates (2009) examined the relevance of the URCS-based R/VC formula in assessing market power and competition. On the basis of rate and variable cost data recorded in STB’s Carload Waybill Sample (CWS) from 2001 to 2008, Christensen Associates reported that the share of ton-miles exceeding the 180 percent threshold had been stable during the period, ranging from about 15 to 20 percent, while the share exceeding 300 percent fluctuated between 2 and 5 percent (Table 3-1). Most significantly, however, Christensen Associates (2009, 11-25) reported that one-fifth to one-third of all traffic had an R/VC below 100 percent. The comparisons were updated in this study for 2012. Table 3-1 shows that in 2012, 20 percent of all rail traffic was purported to have moved at rates below variable cost. A large share of traffic earning revenues below what is represented as an approximation of the incremental cost of a service means that large portions of rail traffic are being priced at an economic loss. This outcome is nonsensical or at least very difficult to reconcile with the railroad industry’s profit motive.
23 The Staggers Rail Act actually phased in the 180 percent standard over 5 years, beginning with a standard of 160 percent and rising by 5 percentage points annually until October 1984.
EXAMPLE OF THE ARBITRARINESS OF COST ALLOCATION RULES
STB compares a disputed rate with 180 percent of the URCS estimate of the average variable cost of the movement to determine whether the rate is presumptively excessive. Because there are substantial common costs involved in the production of rail services, URCS uses administrative rules to allocate them to derive the average variable cost estimate that is applied to a specific movement. The following examples show how choosing among various common cost allocation rules can result in different rates violating the 180 percent R/VC threshold and therefore becoming candidates for scrutiny by STB.
First, suppose a railroad can provide three types of service (1, 2, and 3) with the following demand curves: q1 = 100 – 5p1, q2 = 100 – p2, and q3 = 75 – p3, where each p is the price of the corresponding service and each q is the number of units of the service produced. Further, suppose the railroad’s variable cost function takes the form C(q1, q2, q3) = 500 + 5q1 + 10q2 + 3q3. This function supposes that common variable costs are $500 and the average incremental, or marginal, cost of shipping Products 1, 2, and 3 is $5, $10, and $3, respectively. Suppose the railroad sets p1 = $15, p2 = $13, and p3 = $10. On the basis of the above demand curves, these prices imply that q1 = 25, q2 = 87, and q3 = 65. This set of outputs implies a total variable cost equal to $1,690 = $500 + $5 * 25 + $10 * 87 + $3 * 65. Total revenue is equal to $2,156 = $15 * 25 + $13 * 87 + $10 * 65, so the railroad just breaks even if its fixed costs are ($2,156 – $1,690) = $466, the difference between its total revenue and total variable costs.
There is no unambiguous way to allocate the common variable costs of $500 to the individual products to obtain product-specific variable costs. Consider four possible rules for allocating the common costs to the three products. The first rule allocates on the basis of quantity: the $500 of common costs is allocated to each product according to its share of total output. In this example, Product 1 is allocated 25/(25 + 87 + 65) of the common costs, Product 2 is allocated 87/(25 + 87 + 65) of the common costs, and Product 3 is allocated 65/(25 + 87 + 65) of the common costs. To compute the average variable cost of Product 1 requires adding the marginal cost of Product 1 to Product 1’s share of the $500 of common costs divided by the number of units sold of Product 1. This yields $7.82.
EXAMPLE OF THE ARBITRARINESS OF COST ALLOCATION RULES
Multiplying this fully allocated average variable cost of Product 1 by 1.8 yields $14.08, which is less than the price set for Product 1 of $15. The first row of the embedded table contains the 180 percent of the average variable cost thresholds for each product on the basis of the quantity-based common cost allocation rule. For this rule, only the price charged for Product 1 violates the 180 percent threshold.
The second rule uses revenue shares to allocate the $500 of common costs to each product. Product 1’s total revenue is $15 * 25 = $375, Product 2’s is $13 * 87 = $1,131, and Product 3’s is $10 * 65 = $650. To compute Product 1’s average fully allocated cost requires taking the marginal cost of Product 1 and adding its revenue share of the $500 common cost divided by the amount sold of Product 1. This yields $8.48. Multiplying this fully allocated average variable cost figure for Product 1 by 1.8 yields $15.26, which is greater than the price charged for Product 1. Consequently, this cost allocation rule would not result in the price of Product 1 being deemed presumptively excessive. The second row contains the 180 percent of the average fully allocated cost thresholds for each product for the revenue-based allocation rule. For this allocation rule, only Product 3 violates the 180 percent threshold.
The third cost allocation rule uses the incremental cost shares of each product to allocate the $500 of common costs. Product 1’s incremental cost is $5 * 25 = $125, Product 2’s is $10 * 87 = $870, and Product 3’s is $3 * 65 = 195. Use of the same procedure as described above for Product 1’s marginal cost share to allocate the common variable costs to each product yields an average fully allocated variable cost for Product 1 of $7.10. Multiplying this result by 1.8 yields $12.78. The third row contains the 180 percent of the average fully allocated cost thresholds for each product under this common cost allocation rule. Both Product 1 and Product 3 violate the 180 percent threshold.
The fourth cost allocation rule simply assigns one-third of the $500 of common variable costs to each product. For Product 1, this yields an average fully allocated cost of $11.67, which implies a 180 percent threshold of $21.00. From the table below, only the price of Product 2 violates the 180 percent threshold under this cost allocation rule.
EXAMPLE OF THE ARBITRARINESS OF COST ALLOCATION RULES
This numerical example demonstrates that which of the prices of the three products would violate the 180 percent threshold and be deemed presumptively excessive depends on the ad hoc cost allocation rule used. The example points out the arbitrary nature of using a fully allocated cost approach to determine whether a price charged should be deemed excessive and therefore be a candidate for regulation.
Examples of Alternative Rules for Computing a Rate’s Fully Allocated Cost Threshold: 180 Percent R/VC Value
|Method Used to Compute Fully Allocated Cost||Product 1||Product 2||Product 3|
|Quantity-based cost allocation||$14.08||$23.08||$10.48|
|Revenue-based cost allocation||$15.26||$23.43||$9.57|
|Variable cost–based cost allocation||$12.78||$25.56||$7.67|
|Equal cost allocation by product||$21.00||$12.45||$13.62|
|Actual rates for shipments||$15.00||$13.00||$10.00|
Christensen Associates concluded that URCS was flawed by failing to take into account “latent cost-causing factors or other shipment features.”24 STB has defended URCS from such findings by stating that ratios below 100 percent are possible for some traffic
24 Laurits R. Christensen Associates 2009, 11-24. Christensen Associates findings about URCS were acknowledged by STB in its 2010 URCS critique (STB 2010).
|R/VC < 100%||100% ≤ R/VC < 180%||180% ≤ R/VC < 300%||R/VC ≥ 300%|
|Study Committee Review, 2012|
|Christensen Associates, 2001–2008|
SOURCE: 2012 CWS; Laurits R. Christensen Associates 2009, 11-25; M. Meitzen and K. Eakin, Christensen Associates, presentation to the committee, January 10, 2014 (http://www.trb.org/PolicyStudies/RailTransReg.aspx).
for short periods. The reason given is that URCS is not a measure of short-run variable costs but rather a measure of “intermediate” variable costs made on a system average basis that includes cost items such as rails and ties “that are fixed in the short term.”25 That defense illustrates the inherent problem with cost allocation (i.e., how does URCS decide where to allocate rail and tie costs?), but it is also an implausible defense when 20 to 30 percent of traffic is assumed to be operating at below cost for many years and railroads remain financially solvent.
25 STB Ex Parte No. NOR-42088, Western Fuels Association, Inc., and Basin Electric Power Cooperative v. BNSF Railway Company, June 15, 2012, p. 7.
FIGURE 3-1 Share of nonexempt traffic (ton-miles) by selected R/VC ratios, 2012. (Source: 2012 CWS.)
Even though URCS allocation rules and procedures are arbitrary, analyses of R/VCs suggest that they do not produce random results. They can produce systematic biases in the traffic identified as falling below or above the 180 percent threshold. For example, analyses of R/VCs for nonexempt (tariff and contract) traffic grouped by market distance show that shorter-haul shipments have much higher shares of ton-miles exceeding the 180 percent threshold than do longer-haul shipments (Figure 3-1). As discussed in the previous chapters, railroads have expressed concern that STB does not properly cost hazardous materials shipments because URCS omits a large number of costs associated with their transportation risk and its mitigation.26 They maintain that because of these costs, the rates charged for hazardous materials shipments may appear excessive when they are not. Indeed, a review of the R/VCs for hazardous materials shipments indicates that a large percentage of this traffic moves at rates above 180 percent, suggesting the potential for systematic bias (Figure 3-2).
26 See comments by the Association of American Railroads to STB Ex Parte No. 677-1, July 10, 2008, p. 26.
FIGURE 3-2 Average R/VC for hazardous materials shipments by market distance, 2012. (Source: 2012 CWS.)
Criticisms of URCS because of its omission of relevant costs such as those directly attributable to the movement of hazardous materials are valid. However, the addition of those costs to URCS allocations, if that is possible, would not make the results any more economically meaningful. Replacing or reforming URCS with more “refined” methods of cost allocation would be substituting one contrivance for another. The more appropriate solution is to replace reliance on URCS in rate regulation with a system for identifying unusually high rates that is economically sound and that does not apply arbitrary cost allocation rules. At the conclusion of this chapter, such a methodology is demonstrated.
When a tariff rate exceeds the 180 percent R/VC threshold and a shipper paying that rate complains, the law requires a direct review of the competitive structure of the market in question, often referred to as a qualitative assessment of market dominance. Market dominance is defined in the law as the “absence of effective competition from other
rail carriers or modes of transportation for the transportation to which the rate applies.”27
These assessments are made on a case-by-case basis, and the complainant shipper and the railroad present evidence. In some cases, the railroad concedes a shipper’s characterization of dominance. In others, the railroad may submit evidence to counter the shipper’s evidence, and STB must decide. Because of the limited suitability of long-haul trucking for the bulk commodities normally moved under common carriage, market dominance inquiries tend to focus on the shipper’s proximity to other bulk transportation modes, including other railroads, barges, pipelines, and trans-loading operations (short truck hauls between rail stations).
In inquiries during ICC’s tenure, railroads had additional latitude in characterizing a shipper’s rail substitution possibilities, including nontransportation options. A railroad could, for example, show that a shipper or its customers could readily substitute another product for the one transported by rail (i.e., product competition) or ship to and from alternative locations (geographic competition).28 However, since 1998 STB has prohibited such showings on the basis that they “significantly impede the efficient processing” of the proceedings and present “undue burdens and obstacles” for shippers challenging rates.29 This decision was in direct response to the demand by Congress for the timely handling of rate challenges, “avoiding delay in the discovery and evidentiary phases” of proceedings.30
STB reported that rate cases proceeded more quickly after the exclusion of geographic and product competition from market dominance inquiries. However, a number of cases in the past several years have caused STB to express concern that assessments of market dominance will once again slow down and potentially deter rate challenges. It has stated that “new cases involving challenges
27 49 USC §10707.
28 For example, an electric utility that burns coal may be able to convert to natural gas supplied by a pipeline or obtain power on the wholesale market (creating product competition), or a water treatment plant may obtain chemicals from suppliers in regions served by other carriers (creating geographic competition).
29 STB Ex Parte No. 627, December 10, 1998.
30 49 USC §10704(d), §10701(d)(3).
to dozens, if not hundreds, of transportation rates raise complex market dominance issues. Without some more objective means of resolving these issues quickly, the market dominance inquiry will soon dwarf the rate reasonableness inquiry.”31
In view of the requirement to expedite market dominance inquiries, STB has increasingly turned to URCS. It has reasoned that its use is valid because “Congress regarded R/VC ratios as an appropriate measure for allocating joint and common costs among rail shippers, as reflected in the 180 percent R/VC jurisdictional floor for rate relief.”32 A case in 2012, for example, was brought by a producer of plastic pellets contending that the defendant railroads had market dominance affecting rates in 42 markets.33 The central issue was whether trucks, which are sometimes used to transport the pellets, function as a practical constraint on railroad pricing. To assess this potential, STB has resorted to using URCS and its R/VC values to estimate the highest price that the railroad could charge the pellet shipper without causing substantial traffic to be diverted to trucks.34
In 2013, the Association of American Railroads petitioned STB to restore product and geographic competition. It cited the changing nature of rate relief complaints and the growing complexities of making market dominance decisions. STB ruled against the restoration. It found that railroads did not offer a practical framework that could be used in proceedings to establish the existence and practical effect of these nontransportation forms of competition.35
31 STB NOR No. 42123, M&G Polymers USA, LLC, v. CSX Transportation, Inc.
32 STB Ex Parte No. 657-1, Major Issues in Rail Rates Cases.
33 STB NOR No. 42123, M&G Polymers USA, LLC, v. CSX Transportation, Inc.
34 STB reasoned that truck prices close to 180 percent of R/VC would be a clear indicator of the competitive viability of trucks; however, the agency needed to establish how high the R/VC percentage could go before attracting truck competition. To do this, STB consulted URCS to calculate an R/VC that the railroad would need to average for all of its potentially high (>180 percent) R/VC traffic in order to earn a return on investment equal to the cost of capital. (The idea was that the traffic above 180 percent R/VC is primarily responsible for the railroad’s revenue adequacy.) STB then compared this average R/VC with the R/VC that would trigger truck competition. STB ruled that the latter exceeding the former would indicate that trucks do not provide an effective means of competition. Because the URCS variable cost numbers are essentially arbitrary, so is this procedure.
35 STB Ex Parte No. 717.
Antitrust agencies routinely consider product and geographic competition when they define the relevant market in merger reviews. STB itself has continued to permit evidence of product and geographic competition in deciding rate cases involving pipelines.36 Such forms of competition can affect the willingness to pay for rail transportation on a route, perhaps significantly (USDOJ and FTC 2010, Chapter 4). Many other factors can affect willingness to pay, including wages, the price of other inputs, and the productive capacity at a plant. Accordingly, thorough assessments of the competitive and demand conditions in a market will be inherently site specific and fact intensive. Under such circumstances, market dominance proceedings that are undisciplined could be prone to lengthy delays that deter rate complaints.
An alternative to the categorical prohibition of certain kinds of evidence as a way of preventing delay is to discipline the process itself by using deadlines to compel parties to prioritize their arguments and evidentiary presentations. When they conduct merger reviews, the antitrust agencies follow a process with legislated timelines that allow 30 days for initial assessments about the relevant market and other substantive issues.37 In railroad market dominance inquiries, statutory deadlines might bring about faster and more efficient competition evaluations without the need for excluding types of evidence. Their introduction is considered later in this report.
If a shipper is charged a tariff rate that exceeds 180 percent of variable cost and can prove it ships in a dominated market, it is eligible to challenge the rate by using one of three main methods for judging rate reasonableness. The traditional method for large claims is a SAC proceeding. In response to congressional demands for faster handling of
36 See STB Docket No. 41685, CF Industries, Inc., v. Koch Pipeline Company, LP, May 3, 2000.
37 The U.S. Department of Justice (USDOJ) and the Federal Trade Commission (FTC) review most mergers under the timelines set by the Hart–Scott–Rodino Act of 1976. Once information compliance is met, USDOJ and FTC can request a second 30-day period to seek additional information if concerns arise.
rate cases,38 STB has instituted two additional streamlined methods to be used mostly for smaller claims.
One stated priority of ICC in first implementing the maximum rate provisions in the Staggers Rail Act was to develop a method for assessing rate reasonableness that could be used in cases brought by coal shippers. The concern was that they were “caught in the transition to a less regulated environment … and may be subject to monopoly abuse because they had made investments or locational decisions, or entered into long-term supply contracts, during the pre–Staggers Act period of greater rate scrutiny” (ICC 1985, 3). The agency’s other stated priority was to ensure that the assessment method introduced did not conflict with the law’s requirement that railroads be allowed to earn adequate revenues.39
The 1985 Coal Rate Guidelines introduced the regulatory concept of “constrained market pricing.” The guidelines state that a railroad’s rates for the transportation of market-dominated traffic would be subject to several constraining factors. The primary constraint is revenue adequacy. The guidelines declare that “captive shippers could not be required to continue to pay differentially higher rates than other shippers when some or all of that differential is no longer necessary to ensure a financially sound carrier capable of meeting its current and future service needs” (ICC 1985, 11). However, ICC did not explain how it would implement the constraint and whether the application of a firmwide revenue adequacy constraint implied an intention to scrutinize, or even cap, a railroad’s overall profitability.
Absent further guidance on the application of a firmwide revenue adequacy constraint—or any near-term prospects for its use given the still tenuous financial condition of railroads in 1985—the Coal Rate Guidelines’ two other constraints for judging rate reasonableness took precedence. First, the shipper would not be required to bear the cost
38 49 USC §10704(d), §10701(d)(3).
39 H.R. 1035, 96th Congress, Second Session, 54 (1980), as cited by ICC 1985, 10–11.
of any demonstrated management inefficiencies; that is, ICC would not accept management inefficiencies as a defense for a high rate. Second, the shipper would not be required to bear the cost of facilities or services from which it derives no benefits. ICC declared that such “[c]ross-subsidization of other shippers is effectively precluded” (ICC 1985, 4).
To implement these two constraints, ICC declared that a shipper could not be charged more than the “stand-alone” cost of providing service, defined as a cost that “approximates the full economic costs, including a normal profit, that need to be met for an efficient producer to provide service to the shipper(s) identified” (ICC 1985, 7). Because a railroad’s network and other production facilities are used to provide a range of services to many shippers, the SAC is the theoretical cost the railroad would incur if it only provided the single service in question, without supplying the additional services sharing the production facilities. Of course, this cost cannot be directly observed. Hence, ICC introduced the SAC test, the stated purpose of which was to estimate a competitive rate level “to determine the least cost at which an efficient competitor could provide the [stand-alone] service” (ICC 1985, 15). That estimated rate level would thus “represent the theoretical maximum rate that a railroad could levy on shippers without substantial diversion of traffic to a hypothetical competing service” (ICC 1985, 6).
SAC’s Questionable Applicability to Railroad Regulation
The SAC test was not invented by railroad regulators but imported from the economics literature on utility regulation during the 1970s, particularly as applied to the regulated telecommunications sector of that time.40 As Pittman (2010) points out, the idea of calculating the SAC of a service was conceived to aid regulators in setting rates charged by firms whose economies of scale and declining average costs made setting regulated rates equal to marginal cost problematic for cost recovery. When they were regulated as monopolies, these firms were generally required to provide service at long-run break even; that is, they were restricted to earning a rate of return that is just competi-
40 Credit for formalizing the concept of subsidy-free pricing to define the concept of cross-subsidization is generally given to Faulhaber (1975).
tive (i.e., not supracompetitive or earning residual “economic profits”). The SAC test was conceived to provide regulators with an approximation of the rates that groups of customers (whose services are supplied through use of the same production facilities) should be expected to pay to cover fully the stand-alone cost of providing their service without subsidizing the services provided to other customers.
Consider a regulated monopolist that is subject to a constraint of zero economic profits (i.e., break even). Any rate that it charges to one group of customers that is above the SAC of supplying service to them necessarily means that the other customers are paying less than the incremental cost of providing them service (i.e., the difference between the firm’s total cost and the SAC of providing service to the first group). In this sense, the second group is being subsidized.41 If the monopolist has economies of scale and scope, it also follows that the second group is paying less than the SAC of serving them. A ratemaking structure that is designed to avoid such cross-subsidies may be desirable to regulators to avoid imposing an unfair burden on one set of customers. It may also be desirable on efficiency grounds to keep the rate structure of the regulated firm from inadvertently inviting entry by less efficient, higher-cost suppliers seeking to attract the business of the customers paying the subsidy. Of course, any concern over inefficient entry is inapplicable to railroads, inasmuch as the prospect of high rates inviting railroad entry, with its large fixed and sunk costs, is negligible.42
As noted above, the Coal Rate Guidelines are explicit in referring to the subsidy-free goal as a rationale for the SAC test, which implies an interest in the fairness aspect of cross-subsidization—that is, to ensure that shippers are not forced to pay higher rates that benefit other shippers. However, as Pittman (2010) explains, railroads are not utilities with rates that are fully regulated or that are precluded from earning positive economic profits. Railroads are free to set their own rates and to earn profits. A private railroad not facing a risk of competitive entry
41 The profit constraint implies that the firm is not allowed to “pocket” the difference between the SAC and the revenues earned from the higher rate but must pass it along to other customers through lower rates.
42 Of course, the complainant shipper should welcome entry and would obviously not dispute a rate on grounds that it would spur inefficient entry.
Thus, the absence of any legitimate risk of shipper cross-subsidies or of concern over efficiency or fairness effects arising from such cross-subsidies makes the conceptual basis for applying a SAC test for railroad rate regulation highly questionable. The test might still be useful in ensuring that a railroad does not try to defend a disputed rate by pointing to costs that actually arise from management inefficiencies. The simulated, stand-alone railroad should be designed to be efficient and thus to reveal any such inefficiencies to preclude such a defense. While the Coal Rate Guidelines make clear that part of SAC’s purpose is to identify such management inefficiencies,43 that purpose appears highly questionable today in light of the railroads’ own profit incentives, which should motivate management efficiency.
Thus, as Pittman explains in his critique, the decision by regulators to apply the SAC test must hinge on purposes other than those arising from cross-subsidies or an intention to protect efficiency. One obvious advantage of SAC is that it links directly, even if only ostensibly, to the law’s interest in ensuring railroad revenue adequacy. Viewed in this way, SAC is potentially defensible from a legal and administrative standpoint if it can be applied in a consistent way. However, its disadvantages, as explained next, are substantial. SAC proceedings are costly to bring, and shippers with characteristics fundamentally different from those of the coal shippers for whom the process was designed find the test especially difficult to apply.
Designing an efficient stand-alone railroad and estimating its costs for adjudicatory purposes are complicated endeavors. The SAC procedure requires the complainant shippers and the railroad to design a hypothetical railroad that offers stand-alone service. As detailed in Box 3-2, this requires extensive documentation with regard to its configuration and investment and operating expense items such as locomotives, car
43 Indeed, the second constraint of the Coal Rate Guidelines specifies that “a captive coal shipper would not be required to bear the cost of demonstrated management inefficiencies in the carrier’s operations and pricing structure” (ICC 1985, 2).
STEPS IN A SAC PRESENTATION
To make a SAC presentation, a shipper designs a stand-alone railroad (SARR) tailored to serve an identified traffic group. It is based on the optimum physical plant or rail system needed for that traffic. By using information on the types and amounts of traffic moving over the railroad’s system, the complainant selects a subset of that traffic (including its own traffic, to which the challenged rate applies) that the SARR would serve.
On the basis of the traffic group to be served, the level of services to be provided, and the terrain to be traversed, a detailed operating plan must be developed for the SARR. Once an operating plan is developed that would accommodate the traffic group selected by the complainant, the SARR’s investment and operating expense requirements (including such expenses as locomotive and car leasing, personnel, materials and supplies, and administrative and overhead costs) must be estimated. The parties must provide documentation to support their estimates.
It is assumed that investments normally would be made before the start of service, that the SARR would continue to operate into the indefinite future, and that recovery of the investment costs would occur over the economic life of the assets. However, STB’s SAC analysis only examines a set period of time, commonly 10 years. In that analysis, the revenue requirements for the SARR are estimated on the basis of the operating expenses that would be incurred and the portion of capital costs that would need to be recovered during that period. A computerized discounted cash flow model simulates how the SARR would likely recover its capital investments. It takes into account inflation, federal and state tax liabilities, and the need for a reasonable rate of return. The annual revenues required to recover the SARR’s capital costs (and taxes) are combined with the annual operating costs to calculate the SARR’s total annual revenue requirements.
The revenue requirements of the SARR are then compared with the revenues that the railroad is expected to earn from the traffic group. There is a presumption that the revenue contributions from non–issue traffic (that is, the traffic of noncomplaining shippers) should be based on the revenues produced by the current rates. Traffic and rate level trends for the traffic group are forecast to determine the future revenue contributions from that traffic.
STEPS IN A SAC PRESENTATION
STB then compares the revenue requirements of the SARR with the total revenues to be generated by the traffic group over the SAC analysis period. URCS is used to calculate the variable costs to allocate revenues from shared, or crossover, traffic. Because the analysis period covers multiple years, a present value analysis is used that takes into account the time value of money; the annual over- and underrecovery are netted as of a single point in time. If the present value of the revenues that would be generated by the traffic group is less than the present value of the SARR’s revenue requirements, the complainant has failed to demonstrate that the challenged rate levels violate the SAC constraint. On the other hand, if the present value of the revenues from the traffic group exceeds the present value of the revenue requirements of the SARR, STB must decide what relief to provide to the complainant by allocating the revenue requirements of the SARR among the traffic groups and over time.
SOURCE: STB Ex Parte No. 646-1, September 4, 2007.
leasing, personnel, materials, and administration. Complex computer programs are required to model the hypothetical railroad and to test its operating plan and configuration against the forecast demand of the traffic groups it is supposed to serve. According to the Coal Rate Guidelines, the simulations and their assumptions must be able to “show that the alternative [railroad] is feasible and could satisfy the shipper’s needs. All of its data on construction and operating costs must be verifiable” (ICC 1985, 15).
A SAC presentation is inevitably complex. The minimal evidence that litigants must provide is substantial, and litigation costs are constrained only by the amount of revenues at stake for the shippers and railroads involved in the dispute. As Pittman (2010) and Johnstone (2009) have documented, the higher the monetary sums at stake, the more elaborate, and in some respects fanciful, the SAC scenarios tend
to become. A review of past cases is punctuated with estimates of landscaping costs, minutiae about station dwell times, debates over operating plan details, and speculation about maintenance needs as the hypothetical railroad ages. They contain details and determinations with regard to matters whose relevance cannot in practice be evaluated by outsiders and presumably can be evaluated by STB only with much dedicated expertise and effort.44 Whether the complex hypothetical scenarios that emerge have any connection to the genuine revenue needs of the defendant railroad, which operates a broader network shared by many shippers with many fixed and sunk costs, cannot be readily discerned.
The SAC evidentiary rules were originally intended for use by coal shippers, who ship large volumes on a regular basis over fixed traffic corridors. They have been modified to allow their use by other shippers whose flows do not dominate a corridor. Shippers can propose the inclusion of traffic that crosses over the corridor (or set of corridors) and contributes net revenues (i.e., profits) that would effectively lower the amount of revenue that the stand-alone railroad would need to earn from the complainant shipper to maintain the service. Railroads can argue that some or all of this proposed crossover traffic should be excluded. For shippers of relatively small quantities, the significance of the railroad’s profits earned from the crossover traffic is crucial. The profit contribution from crossover traffic is estimated by using the R/VC markups derived from URCS.
If STB finds that the revenue earned by the defendant railroad from the complainant shipper exceeds the revenue needed by the standalone railroad to serve this traffic, after profits from any crossover traffic are factored in, it will find the rate to be unreasonable. The revenue-adequate rate that the stand-alone railroad would need to charge the shipper would become the maximum rate judged reasonable by STB. If the revenue-adequate rate, as determined by SAC, is lower than the defendant railroad’s rate, it becomes the basis for the assessment of overcharge penalties and a prescribed rate for future
44 After reviewing the evidence submitted in many cases, Pittman (2010, 4) points out that the “process is plagued with both problems of asymmetric information and the resulting incentives and ability to pick and choose among such information in order to further one’s own agenda.”
a One SAC case was originally ruled as rate reasonable, but the case was readjudicated after a court remand and subsequently settled.
SOURCE: STB (http://www.stb.dot.gov/stb/industry/Rate_Cases.htm).
traffic. The prescribed rate can be no lower than 180 percent of the traffic’s R/VC ratio as derived from URCS.
SAC’s High Cost and Inappropriateness
STB estimated in 2013 that a SAC case costs about $5.8 million for a shipper to litigate.45 The portion of these costs pertaining to the minimal evidentiary requirements of a SAC case is difficult to determine because of the parties’ incentive to keep adding evidence and details in proportion to the size of the claim. As noted, SAC was introduced by ICC with coal shippers in mind. Thirty-seven of the 44 SAC cases brought before STB through 2014 involved coal shippers, who won seven and settled in 21 others (Table 3-2). Coal shippers maintain that the SAC process is burdensome and leads to rates being judged reasonable at conservatively high levels out of deference to revenue adequacy, when the true revenue needs of the railroad are lower because of network economies. However, even with these concerns, coal shippers have demonstrated the ability to bring SAC cases and prevail in them.
45 STB Ex Parte No. 715, Rate Regulation Reforms, July 8, 2013, pp. 10–11.
For coal shippers who transport large volumes over fixed routes, the multimillion-dollar litigation costs involved in bringing a SAC case may be recouped by the large reparations and lower rates prescribed on winning a case. The demonstrated ability of coal shippers to bring and win SAC cases may have played a role in the large shift of coal traffic from common to contract carriage during the past decade, as documented in Chapter 2. Accordingly, the law’s rate relief provisions may give coal shippers leverage in negotiating more favorable contract rates and service. Whether SAC has contributed to “fairer” rate outcomes for coal shippers cannot be judged in the absence of legislated evaluation criteria on what constitutes a fair rate.
In comparison, shippers who transport smaller volumes of traffic over more varied routes have not shown an ability to use SAC, much less to win a case. Since SAC’s inception 30 years ago, only one case has been filed by shippers of grain or other agricultural products. McCarty Farms lost the case in 1997.46 Grain shippers in particular have argued that the design of a stand-alone railroad entails large litigation expenses that cannot be justified when a grievance involves a relatively small claim. In addition, these shippers, whose traffic may not be the dominant flow (or remotely close to it) in a corridor or set of corridors, must depend heavily on the profits generated by any crossover traffic to cover common costs and lower the revenue-adequate rate as determined by SAC. These profit contributions in turn are computed from R/VC markups derived from the unreliable URCS. Complainant shippers can therefore face substantial uncertainty about a fundamental aspect of their SAC cases. Such uncertainties, coupled with the high litigation costs of SAC, have discouraged most shippers of commodities other than coal from making use of the process.
Within a short period after adopting the Coal Rate Guidelines, ICC recognized that shippers of lower volumes over more varied routes would not be strong candidates for disputing a rate with the SAC test. However, ICC took many years to develop a set of alternative procedures.
46 See Johnstone (2009) for a history of the McCarty Farms rate relief case. Johnstone points out that McCarty Farms itself, which was one of the largest grain shippers in the case named for it, shipped only 3,000 tons of grain per year, which is less than the total tonnage in a single unit train carrying coal.
When Congress created STB in 1995, it required the introduction of alternatives to SAC.47
Congress recognized that the SAC process was not being used by many shippers. When it created STB in 1995, it instructed the agency to “establish a simplified and expedited method for determining the reasonableness of challenged rail rates in those cases in which a full standalone cost presentation is too costly, given the value of the case.” This directive led STB to create several alternative procedures intended to be less expensive and faster to litigate and administer but accompanied in some cases by limits on the potential award to shippers. The two expedited options, which were introduced in 1997 and revised in 2007, are the simplified SAC and the three-benchmark methodologies.48 Both procedures restrict the evidence parties can submit and set a time limit on decisions.49 STB is required to decide on the case within 120 days of the close of arguments, and parties are required to participate in a 20-day nonbinding mediation process at the outset of the case.
The simplified SAC procedure is conceptually the same as a full SAC procedure. It is streamlined by replacing the design of a hypothetical stand-alone railroad and postulation of its customer base with the apparently less demanding requirement of estimating the SAC of providing the current service with its current traffic on the actual railroad involved. The core analysis is thus an assessment of the existing railroad facility rather than the design of an efficient railroad optimized for the traffic at issue. The litigants are required to estimate the return on investment that would be needed to replicate the existing facility by using estimates of replacement costs, while URCS is used to estimate the railroad’s operating costs. In 2012, STB removed restrictions on the
47 See STB Ex Parte No. 347-2, Rate Guidelines—Non-Coal Proceedings, December 27, 1996.
48 STB Ex Parte No. 646-1, September 5, 2007.
49 STB Ex Parte No. 646-1, September 5, 2007.
total rate relief that could be awarded. The removal was justified on the grounds that any award from a simplified SAC (i.e., lower prescribed rate in conformance with the stand-alone revenue needs) would likely be smaller than that from a full SAC case because of the assumption that the rail service replicated in the simplified SAC is already efficient. This assumption could reduce any differential between the disputed rate and simulated stand-alone rate and thereby lower the potential reward to the shipper.50
In 2013, STB estimated that a simplified SAC would cost a shipper about $4 million to litigate.51 From 1999 through 2014, five rate cases have been adjudicated before STB on the basis of the simplified SAC procedure. All involved chemical shippers and all led to a settlement.52
The three-benchmark process is even more streamlined than the simplified SAC. The procedure’s simplicity derives largely from its heavy reliance on URCS. STB compares the R/VC of the disputed rate with the average R/VC of the portion of the defendant railroad’s other “potentially captive” traffic that has an R/VC higher than 180 percent; this is one benchmark. For a second benchmark, STB compares the disputed rate’s R/VC with the average R/VC of the railroad’s traffic that most resembles that of the traffic at issue with regard to such characteristics as commodity type, carload size, and travel distance. Finally, STB compares the disputed rate’s R/VC with calculations of the average markup that the railroad is presumed to need on all of its potentially captive traffic to make the railroad revenue adequate. The purpose of the three benchmarks is “to ensure that the complaining shipper’s traffic is not bearing a disproportionate share of the carrier’s revenue requirements vis-à-vis other relatively demand-inelastic traffic without good cause.”53 STB applies the three benchmarks to a formula for ascertaining when a disputed rate is reasonable.54
50 STB Ex Parte No. 715, Rate Regulation Reforms, July 25, 2012.
51 STB Ex Parte No. 715, Rate Regulation Reforms, July 8, 2013, p. 22.
53 See STB Ex Parte No. 347-2, Rate Guidelines—Non-Coal Proceedings, December 27, 1996.
54 For a description of the calculations, see STB Ex Parte No. 646-1, September 5, 2007.
The three-benchmark method was estimated by STB in 2007 to cost a shipper about $250,000 to litigate. It is now subject to a $4 million cap on rate relief over 5 years.55 Through 2014, five cases have been brought, all by chemical shippers. One led to a ruling of a rate being unreasonable, and four led to settlements. In 2014 comments to STB, a number of shipper groups, including the National Grain and Feed Association, complained that the caps on awards were too low to justify the expense of bringing a case and that the evidentiary burdens, decision-making uncertainties, and procedural delays remained too great.56
STB’s rate relief procedures involve three steps that appear to be distinct and independent but that are integrated in an important sense because of the need of regulators to ensure that the amount of rate relief resulting from them does not impair railroad revenue adequacy. Perhaps most significantly, the identification of rates that are candidates for further scrutiny with respect to market dominance and reasonableness is determined by a statutorily defined R/VC screen. The screen itself is arbitrary and cannot be implemented in a reliable and economically valid way because of its reliance on cost allocations as implemented by URCS. This unreliability appears to have prompted regulators to institute exacting and often costly procedures for assessing market dominance and rate reasonableness. The cost and complexity of the process have contributed to shippers who have relatively small claims not taking advantage of the law’s maximum rate protections.
In stipulating that the variable cost of a priced unit of traffic be calculated as a means of screening it for eligibility for rate relief, the Staggers Rail Act created an insoluble problem for regulators. Most
55 STB Ex Parte No. 715, July 18, 2013.
56 See shipper comments to STB Ex Parte No. 665-1, Rail Transportation of Grain, Rate Regulation Review. In particular, see comments by the National Grain and Feed Association, June 26, 2014, p. 14. In revising the simplified procedures in 2007, STB concluded that shippers were reluctant to use the simplified procedures because of vagueness in the requirements and uncertainty about eligibility.
railroad costs are shared by traffic and cannot be unambiguously allocated to individually priced units of traffic. URCS, like all similar schemes for allocating variable costs, is incapable of producing results with any stable relationship to an individual movement’s price. Previous studies have not found any connection between the R/VC relationships that emerge from URCS and a railroad’s market power. However, those studies, as well as this study, have revealed that URCS produces inexplicable results. Among them are the pricing of about 25 percent of all railroad shipments below their presumed variable cost and most short-haul and hazardous materials traffic exceeding the 180 percent R/VC threshold.
Perhaps to compensate for an unreliable URCS-based rate screening process, STB has instituted exacting and complex procedures for assessing market dominance and rate reasonableness. Railroad revenue adequacy is safeguarded as a practical matter because of these imposing procedures, but shipper access to rate relief is restricted in a biased manner. The once-burdensome market dominance assessment, which was designed to ensure that any rate that passes the unreliable URCS-based R/VC screen is from a market lacking competition, was streamlined by STB by limiting the evidence that could be introduced by railroads rather than simply by adopting review timelines. The processing speed gained from this action has only marginally expanded access to rate relief, in part because the main evidentiary method used for ruling on the reasonableness of a rate requires a complex SAC presentation. Such presentations entail large minimum litigation expenses that cannot be justified by shippers with relatively small claims, and they are inapplicable to shippers whose traffic is not close to being the dominant flow along the corridor it uses. The high fixed cost and inappropriateness of this standard have led to few shippers of commodities other than coal bringing a SAC case.
After more than a decade of complaints about SAC from shippers of grain, chemicals, and other commodities that regularly use common carriage, STB introduced two expedited evidentiary methods for assessing rate reasonableness. One is a somewhat simplified version of SAC, and one assesses the profitability of traffic as determined by R/VC ratios derived from URCS. With each effort to streamline the rate relief process, STB has increased its dependence on the arbitrary cost
allocations derived from URCS. STB is thus moving toward replacing the inappropriate SAC test with URCS-based procedures that offer even less predictable decision criteria and lack even that test’s weak conceptual basis. Meanwhile, shippers of some commodities that are heavy users of common carriage, including grain, have not used the expedited methods after more than 15 years. They contend that the procedures continue to be irrelevant and to impose a substantial cost burden and uncertainties about decision criteria.
To simplify and expand access to rate relief, STB has periodically considered the use of arbitration as a method for rate dispute resolution, as have members of Congress in various legislative proposals.57 Arbitration is normally viewed as an alternative method for resolving disputes. It involves proceedings that are less formal and therefore faster and more economical than formal discovery and adjudication proceedings. The arbitrator reviews the evidence, listens to the parties, and then makes a decision. STB now uses arbitration to resolve certain nonrate disputes between shippers and railroads, and a voluntary arbitration and mediation program is used by BNSF Railway and grain shippers to settle rate and service disputes in Montana.58 The National Grain and Feed Association has operated an arbitration system for railroad–shipper disputes since the early 20th century, which it expanded in 1998 to include involvement by major railroads.59
In conducting hearings during 2001 on the potential to use arbitration for small rate cases, STB also examined the potential for using the binding, final-offer form that is practiced in Canada.60 The agency
57 A recent legislative proposal for rate arbitration was introduced on September 8, 2014, in S. 777, 113th Congress, second session.
58 BNSF, the Montana Farm Bureau, and the Montana Grain Growers have informally mediated grain farmers’ rail issues since 2004, but they agreed to establish a formal process in 2008.
59 See National Grain and Feed Association Trade Rules, Arbitration Rules, Rail Arbitration Rules and Rail Mediation Rules, 2014 (http://www.ngfa.org/wp-content/uploads/2014_TradeRules.pdf).
60 STB Ex Parte No. 586, Arbitration—Various Matters Relating to Its Use as an Effective Means of Resolving Disputes That Are Subject to the Board’s Jurisdiction, September 18, 2001.
concluded that because the law states that STB itself must adjudicate all challenged rail rates, such an alternative process would need to be legislatively authorized.61
Canada’s arbitration process merits consideration because of the similarities and interconnectivity of the U.S. and Canadian railroad industries. As in the United States, the Canadian freight railroad industry has been substantially deregulated for about 30 years, and the easing of regulation has been accompanied by improved productivity and lower freight rates.62 Canada requires railroads to offer common carrier service and exempts traffic moved under confidential contract from regulations governing rate and service offerings, as does the United States. The two major Canadian railroads are closely integrated with the U.S. rail system and share the characteristics of vertical integration and private ownership (Cairns 2013). Both the Canadian Pacific (CP) and Canadian National (CN) railroads are part of larger companies that include two U.S. Class I railroads, the Soo Line (CP) and Grand Trunk (CN). About 30 percent of the revenues of CN and CP are earned from cross-border movements (Cairns 2014). Although each concentrates its operations in different parts of Canada—CP in the West and CN in the East—the two railroads are competitors for substantial amounts of traffic, since both operate transcontinental systems.63
Despite these similarities and the integration of the two countries’ railroad systems, Canada’s regulatory regime differs from that of the United States, particularly with respect to provisions granting rate relief. Canadian law requires that rate disputes be decided by arbitration according to the final-offer decision-making rule. Any shipper dissatisfied with a tariff rate or service offered by CP or CN may apply to the Canadian Transportation Agency (CTA) for arbitration to
61 STB Ex Parte No. 586, September 20, 2001, Footnote 7.
62 In comparison, freight rail systems in Europe and Australia are vertically separated or hybrids of vertical separation and integration. In the United Kingdom and Sweden, for example, independent train operators compete over a single rail network owned by a separate entity, whereas in Germany and Australia a single vertically integrated railroad is required to provide access to independent train operators (Cairns 2013).
63 According to Cairns (2013, 33), the two railroads compete directly for about 40 percent of Canadian rail freight traffic, and another 20 percent could be transloaded (i.e., short-haul movements could be made by truck to gain access to the second railroad).
CTA manages the application, but an independent arbitrator is selected to decide on the dispute. Both parties to the arbitration are required to present their arguments, evidence, and offers for resolution of the dispute in their final form (that is, with no additional opportunity for amendments). The arbitrator is bound to choose one of the two offers without modification. The law limits the decision’s applicability to a duration of 1 year. It also limits the duration of the arbitration process itself: 60 days is allowed for most disputes, and 30 days is allowed for disputes involving monetary sums of less than $750,000 (Cairns 2013). The details of the arbitrator’s decision, as well as the specific offers proposed by the parties, are kept confidential.
The requirement of confidentiality is intended to reduce the formality of the proceedings and to avoid introducing precedents that could discourage parties from negotiating a settlement or from making other concessions that help resolve the dispute more quickly. The imposition of time limits is intended to bring economy to the process and to ensure that shippers are not precluded from access to rate relief as a consequence of slow processing and high litigation costs. In addition, the time limit in conjunction with the final-offer rule injects uncertainty into the process, which limits the likelihood that any one party will take an extreme position and encourages the settlement of disputes. Box 3-3 provides an overview of the key features of final-offer arbitration and the procedures used in Canada.
Canada does not mandate that arbitrators have any special knowledge or training in the railroad industry, and arbitration decisions are confidential and do not create precedent. For these reasons, the process can be even more unpredictable and compel compromise, and only on rare occasions do parties avail themselves of the process in successive years (Cairns 2014). Because the decisions of arbitrators are kept confidential, the number of decisions favoring shippers versus railroads is unknown. Cairns estimates that about 30 decisions in total have been rendered since the process was instituted in 1988. This estimate is consistent with that in a 2001 report by a
FINAL-OFFER RATE ARBITRATION IN CANADA
Arbitration is intended to be a faster, more flexible, and more economical means of dispute resolution than litigation. In conventional arbitration, the arbitrator is free to impose any award that he or she believes is appropriate. The tendency of arbitrators to split the difference between each side’s offer can reduce the prospects for a negotiated settlement, since both sides are motivated to make extreme offers under the assumption that the arbitrator will select a compromise position (Kochan and Katz 1988, 279–284).
Final-offer arbitration (FOA), with its “either-or” format, is designed to have the opposite effect by prompting a convergence of offers, since each party believes that making a more reasonable offer will increase the odds of its being selected. By introducing uncertainty into the process, FOA is designed to prompt the parties toward their most reasonable offer and therefore to induce settlements. Kochan and Katz (1988, 284) report that only 10 to 15 percent of FOA cases lead to an arbitrator decision, compared with 25 to 30 percent of conventional arbitrations. FOAs are often preceded by mediation to start the settlement discussion.
In Canada, individual shippers or groups of shippers can apply for arbitration to challenge an existing tariff rate or a tariff rate being offered by a railroad in anticipation of an expiring contract rate.a The railroad must receive written notice of the shipper’s intention to pursue arbitration at least 5 days before the shipper applies with CTA. CTA appoints the arbitrator, and the shipper and railroad share the costs of the process. Within 10 days of the application being filed, the shipper and railroad must submit their final offers to CTA, including the proposed rates. Within 5 days of receiving the final offers, CTA must refer the matter to the arbitrator.
The arbitrator must consider whether the shipper could use any other competitive means of transportation. This requirement is intended to
a For a detailed review of the process, see http://www.pulsecanada.com/transportation/resources/final-offer-arbitration-general-information.
FINAL-OFFER RATE ARBITRATION IN CANADA
make it more difficult for shippers who have access to viable transportation alternatives to succeed and thus to make them less inclined to bring a case. Otherwise, CTA does not specify the criteria that an arbitrator must consider in reaching a decision or the kinds of evidence that parties must submit. Because the process and its results are kept confidential, the decisions do not establish precedents concerning the evidence submitted or how it is evaluated and weighed.
The arbitrator in a Canadian rate case is required to render a decision within 60 days of the shipper’s original submission to CTA, and the decision is made retroactive to the date on which the shipper filed its application. The duration of the decision is limited to 1 year.
panel appointed by the Canadian government to review the regulatory process. The panel reported 23 decisions during the process’s first 13 years and estimated that half of the arbitration cases were settled before a decision (Minister of Public Works and Government Services Canada 2001, 35).
Canadian disputes can involve service issues, but Cairns (2014) estimates that most involve rate grievances. Because there is no eligibility requirement for a shipper to dispute a rate, any shipper can file a dispute or, as Cairns points out, threaten to do so to gain leverage in negotiations with railroads. According to the 2001 panel report, railroads have expressed concern about the ability of shippers to seek rate relief without good cause. However, the panel pointed out that arbitrators are instructed by CTA to assess whether the shipper has competitive means of transporting goods. The panel concluded that as a consequence of this requirement, a shipper in a competitive market would be unlikely to endure the complexity and expense of a case in view of the low likelihood of prevailing once its competitive status was exposed (Minister of Public Works and Government Services Canada 2001, 71).
When the R/VC formula was introduced in the Staggers Rail Act more than 30 years ago, rates had long been set by cartels under regulatory oversight. Hence, there was little practical value in trying to compare rates in dominated versus more competitive markets. The regulatory practice of assigning portions of a railroad’s total costs to individual units of traffic was therefore retained as a way of identifying potentially excessive rates. For reasons that have been explained in this chapter, STB’s URCS-derived R/VC formula does not indicate whether a rate is unusually high relative to rates of comparable shipments in competitive markets. In short, comparing the arbitrarily defined URCS variable cost with an actual rate is not a sound basis for screening shippers for eligibility for relief. It cannot be justified on economic grounds and has led to the development of a rate relief system that is characterized by large inequalities in shipper access to relief.
Nevertheless, STB must identify traffic eligible to pursue rate relief. Its method of doing so must ensure that shippers in markets lacking effective competition do not pay unreasonably high rates and that railroads are not denied the ability to attain revenue adequacy. The challenge for regulators is to replace an unreliable and arbitrary cost allocation scheme with an economically sound approach for identifying rates that may be unreasonable. Such an approach will, in turn, allow for the development of procedures for assessing the reasonableness of rates that are not so burdensome and costly that they effectively deny eligibility for relief to shippers for reasons unrelated to the legitimacy of their claims.
A wealth of information on unregulated, market-based rail prices now exists in STB’s annual CWS, along with detailed information on important characteristics of individual shipments. These data, supplemented with information on the competitive structure of individual markets, can be used to develop models of rates determined under effectively competitive conditions. A shipper in a market lacking effective competition can compare its tariff rate with the rate predicted by the model if the shipment were made in a competitive market. The predicted rate, or some designated percentage above it, would become
An illustrative methodology that uses statistical models and readily available data to implement this benchmarking approach is demonstrated in Appendix B. Data on shipment characteristics and rates in effectively competitive markets are used to construct a predicted, or benchmark, rate for any given rail shipment in a presumptively noncompetitive market on the basis of key observable characteristics of the shipment. The benchmark rate is computed by estimating the distribution of average rates (i.e., revenue per ton-mile) for the shipment conditional on observable characteristics of the shipment mainly derived from CWS data. The characteristics include the distance traveled, the number of carloads in the shipment, the number of railroads involved, and competitive circumstances at the origin and destination (i.e., number of competing railroads and proximity to other modes), as well as controls such as calendar year and railroad. The demonstration shows how models can be constructed for broad commodity groups such as farm products, coal, and chemicals. Models could similarly be developed for narrower product groups (e.g., grain, hazardous materials) as long as there are enough observations for reasonably precise estimation of the conditional distribution.
Once a model of this sort is developed and made public, a shipper could enter a defined set of shipment characteristics, such as travel distance, commodity, car type, and railroad (or railroads) used, into a website or spreadsheet that would predict the competitive benchmark rate. Most of the characteristics to be entered into the model would be readily identifiable by the shipper or could be integrated into the program. For example, the number of railroads serving the market and distance to a waterway can be preprogrammed. The demonstration in Appendix B suggests that such models would be no more complicated to construct and run, and would probably be less so, than the annual derivation of variable costs from URCS. The complexity of the latter (and the potential to upset existing patterns of results) has prevented its basic structure from being changed for decades despite fundamental methodological flaws. Once the statistical models of the type demonstrated in Appendix B are constructed, they could be used from year to year but updated regularly with new data on shipment rates and characteristics as obtained from the annual CWS.
Designation of the percentage above the benchmark rate that would qualify a shipper’s tested rate for further scrutiny for reasonableness would be key to the implementation of the models. A shipper’s tested rate is unlikely to match the predicted rate perfectly because of the impracticality of having a complete set of information on all of a shipment’s economically meaningful characteristics for precise matching with characteristics of shipments in the benchmark group. However, the more the tested rate deviates upwardly from its predicted value, the less likely is its high level due only to the omission of economically meaningful characteristics. Regulators would need to decide the percentage by which a tested rate must exceed its predicted benchmark competitive level before a shipper would be entitled to have its rate scrutinized as potentially unreasonable. Examinations of each model’s predictive capability may be undertaken to inform such decisions. Such examinations might cause regulators to apply different qualifying thresholds for different commodities.
Regulators would need to consider an obvious trade-off in making such decisions. On the one hand, a stricter screen (i.e., large percentage threshold) will provide less of a threat to railroad revenue adequacy but would make fewer shippers with legitimate rate grievances eligible for relief. On the other hand, a less strict screen would offer greater opportunity for aggrieved shippers to challenge their rates but would pose a greater threat to railroad revenue adequacy. Although decisions about the appropriate threshold could be controversial, they would be transparent. That is preferable to the current system, which relies on arbitrary cost allocation rules that are used in implementing an arbitrary R/VC threshold.
|FTC||Federal Trade Commission|
|GAO||Government Accountability Office|
|ICC||Interstate Commerce Commission|
|STB||Surface Transportation Board|
|USDOJ||U.S. Department of Justice|
Cairns, M. 2013. Expansion of Regulated Access to Railway Infrastructure in North America: Implications from Practices and Recent Experiences with Regulated Access Overseas. Research in Transportation Business and Management, Vol. 6, pp. 31–44.
Cairns, M. 2014. Regulation of Freight Rail in Canada. Presentation to the study committee, May 29. http://onlinepubs.trb.org/onlinepubs/sp/Regulation_of_Freight_Rail_in_Canada_Cairns.pdf.
Faulhaber, G. 1975. Cross-Subsidization: Pricing in Public Enterprises. American Economic Review, Vol. 65, pp. 966–977.
GAO. 2006. Freight Railroads: Industry Health Has Improved but Concerns About Competition and Capacity Should Be Addressed. Washington, D.C.
Grimm, C., and C. Winston. 2000. Competition in the Deregulated Railroad Industry: Sources, Effects, and Policy Issues. In Deregulation in Network Industries: What’s Next? (S. Peltzman and C. Winston, eds.), American Enterprise Institute–Brookings Joint Center for Regulatory Studies, Washington, D.C., pp. 41–71.
ICC. 1985. Coal Rate Guidelines, Nationwide.
Johnstone, A. 2009. Captive Regulators, Captive Shippers: The Legacy of McCarty Farms. Montana Law Review, Vol. 70, pp. 239–272.
Kochan, T., and H. Katz. 1988. Collective Bargaining and Industrial Relations: From Theory to Policy to Practice, 2nd ed. Irwin, Homewood, Ill.
Laurits R. Christensen Associates. 2009. A Study of Competition in the U.S. Freight Railroad Industry and Analysis of Proposals That Might Enhance Competition. Madison, Wis., Nov.
Meyer, J. R. 1958. Some Methodological Aspects of Statistical Costing as Illustrated by the Determination of Rail Passenger Costs. American Economic Review, Vol. 48, No. 2, pp. 209–222.
Minister of Public Works and Government Services Canada. 2001. Vision and Balance: Report of the Canada Transportation Act Review Panel. http://publications.gc.ca/collections/Collection/T22-107-2001E.pdf.
Pittman, R. 2010. Against the Stand-Alone-Cost Test in U.S. Freight Rail Regulation. Journal of Regulatory Economics, Vol. 38, pp. 313–326.
Scharfman, I. L. 1915. Railroad Regulation: An Analysis of the Underlying Problems in Railroad Economics from the Standpoint of Government Regulation. LaSalle Extension University, Chicago, Ill.
Schmalensee, R. L. 1979. The Control of Natural Monopolies. Lexington Books, Lexington, Mass.
STB. 2010. Report to Congress Regarding the Uniform Rail Costing System, Submitted Pursuant to Transportation and Housing and Urban Development, and Related Agencies Appropriations. May 27.
USDOJ and FTC. 2010. Horizontal Merger Guidelines.
Wilson, W. W., and J. D. Bitzan. 2003. Costing Individual Railroad Movements. Federal Railroad Administration, Sept.