Freight Railroad Regulation: Surface Transportation Board's	 
Oversight Could Benefit From Evidence Better Identifying How	 
Mergers Affect Rates (05-JUL-01, GAO-01-689).			 
								 
Railroads have been a primary mode of freight transportation for 
many years, especially for bulk commodities such as coal and	 
grain. During the last 25 years, the freight railroad industry	 
has undergone substantial consolidation largely to reduce costs  
and increase efficiency and competitiveness. This consolidation  
has raised concern from some companies that ship and receive	 
their goods by rail and others that mergers have led to a	 
reduction in railroad competition and consequently higher rail	 
rates or poorer service. This report reviews (1) the role the	 
Surface Transportation Board plays in reviewing proposed railroad
mergers and overseeing mergers that have been approved and how	 
post-merger oversight is conducted, (2) how the Board mitigates  
potential harm to competition, and (3) how the Union		 
Pacific/Southern Pacific merger affected rail rates in selected  
geographic areas. GAO found that the Board reviews railroad	 
merger proposals and approves those that are consistent with the 
public interest, ensures that any potential merger-related harm  
to competition is mitigated to preserve competition, and oversees
mergers that have been approved. The Board imposes conditions on 
mergers to mitigate potential harm to competition. The Board also
focuses on the overall direction and magnitude of rate changes	 
when analyzing rail rates as part of merger oversight. It does	 
not isolate the effects of mergers on rates from other effects.  
When GAO used this approach to analyze how the Union		 
Pacific/Southern Pacific merger affected rail rates, it found	 
that the merger reduced rates in four of six commodities studied.
However, for two of the commodities, the merger put upward	 
pressure on rates, even though other factors caused overall rates
to decrease. By focusing on overall rate decreases, the Board	 
will be unable to determine whether the decrease is due to the	 
merger or other factors.					 
-------------------------Indexing Terms------------------------- 
REPORTNUM:   GAO-01-689 					        
    ACCNO:   A01072						        
  TITLE:     Freight Railroad Regulation: Surface Transportation      
             Board's Oversight Could Benefit From Evidence Better Identifying 
             How Mergers Affect Rates                                         
     DATE:   07/05/2001 
  SUBJECT:   Competition					 
	     Corporate mergers					 
	     Freight transportation operations			 
	     Freight transportation rates			 
	     Railroad industry					 
	     Railroad transportation operations 		 
	     Reporting requirements				 
	     Colorado						 
	     Louisiana						 
	     Reno (NV)						 
	     Salt Lake City (UT)				 
	     Texas						 

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO Testimony.                                               **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
******************************************************************
GAO-01-689
     
Report to the Chairman, Subcommittee on Transportation and Related Agencies,
Committee on Appropriations, House of Representatives

United States General Accounting Office

GAO

July 2001 FREIGHT RAILROAD REGULATION Surface Transportation Board's
Oversight Could Benefit From Evidence Better Identifying How Mergers Affect
Rates

GAO- 01- 689

Page i GAO- 01- 689 Freight Railroad Regulation Letter 1

Results in Brief 2 Background 4 The Board Determines Whether Mergers Are in
the Public Interest

and Assesses the Implementation of Merger Conditions 5 The Board Has Acted
to Address Harm to Competition, but Some

Shippers and Carriers Are Concerned About Merger Oversight 12 The Board
Could Benefit From Evidence Better Identifying How

Mergers Affect Rail Rates 19 Conclusions 27 Recommendation for Executive
Action 28 Agency Comments and Our Evaluation 28

Appendix I Chronology of Class I Railroad Mergers- August 1995 Through June
2001 30

Appendix II Scope and Methodology 33

Appendix III Description and Discussion of Econometric Model Used to Conduct
Rail Rate Analysis 36

Tables

Table 1: Volume of Shipments, in Carloads, From the Reno and Salt Lake City
Areas, 1994- 99 21 Table 2: Changes in Postmerger Rail Rates Due to Merger
and

Other Factors, 1994- 99 23 Table 3: Changes in Postmerger Rail Rates for
Potential 2- to- 1

Shippers Compared With Rail Rates for Shippers Served Solely by UP 25 Table
4: Shipments to and From the Salt Lake City and Reno Areas,

in Carloads, 1994- 99 43 Table 5: Rates and Costs of Shipments to and From
the Salt Lake

City and Reno Areas, 1994- 99 44 Table 6: List of Variables Used in Our
Econometric Analysis of Rail

Rates 46 Contents

Page ii GAO- 01- 689 Freight Railroad Regulation

Table 7: Changes in Rail Rates, for Shipments to and From the Salt Lake City
and Reno Areas 48 Table 8: Econometric Results of Rail Rates, for Shipments
From

the Salt Lake City Area 51 Table 9: Econometric Results of Rail Rates, for
Shipments to the

Salt Lake City Area 52 Table 10: Econometric Results of Rail Rates, for
Shipments From

the Reno Area 53 Table 11: Econometric Results of Rail Rates, for Shipments
to the

Reno Area 54

Figure

Figure 1: Trackage Rights Granted to, and Lines Purchased by, BNSF in the
UP/ SP Merger 8

Abbreviations

BNSF Burlington Northern and Santa Fe Railway ICC Interstate Commerce
Commission SP Southern Pacific Transportation Company UP Union Pacific
Railroad

Page 1 GAO- 01- 689 Freight Railroad Regulation

July 5, 2001 The Honorable Harold Rogers Chairman, Subcommittee on

Transportation and Related Agencies Committee on Appropriations House of
Representatives

Dear Mr. Chairman: Railroads have been a primary mode of freight
transportation for many years, especially for bulk commodities such as coal
and grain. Over the last 25 years, the freight railroad industry has
undergone substantial consolidation. Since 1994, just prior to the recent
wave of mergers, the number of independent railroad systems with at least
one Class I railroad 1 has decreased from 12 to 7. 2 In 1999, the five
largest Class I railroads accounted for about 94 percent of the total Class
I operating revenue and about 95 percent of total Class I revenue ton-
miles. 3 Railroads have consolidated largely to reduce costs and increase
efficiency and competitiveness.

Industry consolidation has raised concerns from companies that ship and
receive their goods by rail (rail shippers) and others about the lack of
competition in the industry. In general, rail shippers are concerned that
mergers have led to a reduction in railroad competition and consequently
higher rail rates, poorer service, or both. In the context of railroad
mergers, the Surface Transportation Board (the Board), which reviews
railroad merger proposals, has defined ?competitive harm? as the extent to
which merging parties gain sufficient market power to profitably raise

1 Class I railroads are the nation?s largest freight railroads as measured
by revenue. In 1999 (the latest data available), Class I railroads were
those railroads whose operating revenues were $258. 5 million or more.

2 The seven independent railroad systems include eight U. S. Class I
railroads: Burlington Northern and Santa Fe Railway Co. (BNSF); CSX
Transportation, Inc.; Grand Trunk Western Railroad, Inc.; Illinois Central
Railroad Co.; Kansas City Southern Railway Co.; Norfolk Southern Railroad
Co.; Soo Line Railroad Co.; and Union Pacific Railroad Co. It should be
noted that Illinois Central and Grand Trunk Western are commonly controlled
by the Canadian National Railway Co. and the Soo Line Railroad is controlled
by the Canadian Pacific Railway Co.

3 A revenue ton- mile is 1 ton of revenue freight transported 1 mile.

United States General Accounting Office Washington, DC 20548

Page 2 GAO- 01- 689 Freight Railroad Regulation

rates, reduce service, or both. (See app. I for a chronology of Class I
railroad mergers since August 1995.)

This report responds to your request that we review the Board?s oversight of
railroad mergers. 4 In particular, this report discusses (1) the role the
Board plays in reviewing proposed railroad mergers and overseeing mergers
that have been approved and how postmerger oversight is conducted, (2) how
the Board mitigates potential harm to competition, and (3) how the Union
Pacific/ Southern Pacific merger affected rail rates in selected geographic
areas. This report primarily focuses on mergers of Class I railroads since
the Board was created in 1996. However, information on prior mergers is
included to show how merger oversight has changed over time. Moreover, this
report primarily focuses on issues pertaining to competition and not other
issues that might arise from a merger.

To accomplish our objectives, we reviewed applicable laws, regulations, and
decisions and met with Board officials, representatives of shippers? trade
associations, and railroad officials. We also developed an econometric model
to analyze selected merger- related rail rates using data from the Board?s
Carload Waybill Sample 5 for the period 1994 through 1999. We focused our
analysis on selected geographic areas associated with the 1996 merger of the
Union Pacific Railroad (UP) with the Southern Pacific Transportation Company
(SP) because of the significant competition issues pertaining to this
merger. (See app. II for a more detailed discussion of how we carried out
our work.)

The Board is the federal agency responsible for reviewing railroad merger
proposals and approving those that are consistent with the public interest.
The Board also ensures that any potential merger- related harm to
competition is mitigated. Mitigation efforts have focused on preserving
competition. The Board also oversees mergers that have been approved.
Oversight is not statutorily required, but when imposed, it has focused on
determining whether conditions (such as granting the authority for one

4 For the purposes of this report, the term ?merger? includes merger,
consolidation, and/ or acquisition transactions between Class I railroads. 5
The Carload Waybill Sample is a sample of railroad waybills (in general,
documents prepared from bills of lading that authorize railroads to move
shipments and collect freight charges); the sample contains information on
rail rates. Results in Brief

Page 3 GAO- 01- 689 Freight Railroad Regulation

railroad to operate over the tracks of another railroad) have been
implemented and have been effective in protecting against potential harm to
competition. As an adjudicatory agency, the Board relies on reports,
comments, and other information (called the merger record) submitted by
railroads, shippers, and others to conduct oversight. The merger record
serves as the basis for oversight decisions. In recent years, as the
complexity of mergers has increased and service disruptions associated with
the merger integration process and other problems have occurred, the Board?s
oversight activities and reporting requirements have increased as well.

The Board has found little competition- related harm during oversight of
recent mergers. The Board?s action to address competition- related harm
largely depends on the sufficiency of the evidence presented. In some cases,
the Board has not acted to address competition- related concerns during
oversight because it determined that the evidence of harm was not
sufficient. In other cases, during oversight the Board has modified
conditions that it originally imposed to mitigate potential harm to
competition when it believed such action was necessary to preserve
competition. Shipper association representatives and railroad officials with
whom we spoke generally agreed that the Board?s oversight process is a
valuable mechanism that allows them to participate in the oversight of
mergers. But some shipper association officials told us they were
dissatisfied with the Board?s oversight because they believe that the Board
is not responsive to their concerns and the process is too time- consuming.
Railroad officials told us that shippers try to use the process to address
non- merger- related issues.

Using an econometric approach that isolated the specific effects of the
Union Pacific/ Southern Pacific merger on rail rates for certain commodities
in two geographic areas- Reno, Nevada, and Salt Lake City, Utah- we found
that the merger reduced rates for four of the six commodities we studied.
However, for one of the commodities, the merger placed upward pressure on
rates, even though other factors caused the overall rates to decrease. For
the remaining commodity, rates were relatively unchanged by the merger. In
analyzing rail rates as part of merger oversight, the Board examines the
merger oversight record before it, which has generally focused on the
overall direction and magnitude of rate changes. According to Board
officials, in general, these records have not permitted the Board to
reliably and precisely isolate the effects of mergers on rates from effects
of other factors (such as the volume of shipments). Obtaining evidence that
quantitatively separates the effects of mergers on rates from the effects of
other factors, such as the volume of

Page 4 GAO- 01- 689 Freight Railroad Regulation

shipments, would help the Board identify whether competition- related
conditions imposed on mergers are meeting their objective. We are
recommending that, as part of merger oversight, the Board, when appropriate,
require the filing of information that identifies the effects of specific
factors, including mergers, on postmerger rail rates.

In 1995, the Congress passed the ICC Termination Act, which abolished the
Interstate Commerce Commission (ICC) and created the Board. The act
transferred many of ICC?s core rail functions to the Board, including the
responsibility to review and approve railroad mergers. The Board has
exclusive jurisdiction to review proposed rail mergers, and if approved by
the Board, such mergers are exempt from other laws (including federal
antitrust laws that would otherwise apply to the transaction) as necessary
to carry out the transaction. The Board also conducts oversight of mergers
that have been approved. However, there is no statutory requirement for
merger oversight. ICC had approximately 400 employees in 1995, its last year
of operation. For fiscal year 2001, the Board received an appropriation to
support 143 employees.

In October 2000, the Board proposed modifications to its regulations
governing major rail consolidations. According to the notice of proposed
rulemaking, the Board recognized that current merger regulations are
outdated and inappropriate for addressing future major rail mergers that, if
approved, would likely result in the creation of two North American
transcontinental railroads. In June 2001, the Board adopted final
regulations governing proposed major rail consolidations. The final
regulations recognize the Board?s concerns about what the appropriate rail
merger policy should be in light of a declining number of Class I railroads,
the elimination of excess capacity in the industry, and the serious service
problems that have accompanied recent rail mergers. The final rules
substantially increase the burden on applicants to demonstrate that a merger
is in the public interest, in part by providing for enhanced competition and
protecting service. The rules also establish a formal annual oversight
period of not less than 5 years following a merger?s approval. Background

Page 5 GAO- 01- 689 Freight Railroad Regulation

The Board is responsible for approving railroad mergers that it finds
consistent with the public interest. When necessary and feasible, conditions
are imposed by the Board to mitigate any potential harm to competition.
Oversight is designed to ensure that merger conditions have been implemented
and that they are meeting their intended purpose.

In determining, under the ICC Termination Act of 1995, whether proposed
mergers are consistent with the public interest, 6 the Board is required to
consider a number of factors that relate to competition. These include the
effect of a proposed transaction on the adequacy of transportation to the
public; the effect on the public interest of including, or failing to
include, other rail carriers in the area involved in the proposed
transaction; and the impact of the proposed transaction on competition among
rail carriers in the affected region or in the national rail system. 7 The
act also establishes a 15- month time limit for the Board to complete its
review of accepted applications for mergers between Class I railroads and
reach a final decision. 8 Since the Board was created, two applications for
merger between Class I railroads have been submitted- Conrail?s acquisition
by CSX and Norfolk Southern and Canadian National/ Illinois Central- both of
which were approved. The Board also approved the Union Pacific?s acquisition
of Southern Pacific, an application that had originally been submitted to
ICC.

6 The Board is charged with determining whether proposed transactions are in
the public interest, not determining whether they comply with the antitrust
laws. The Board is empowered to disapprove transactions that would not
violate the antitrust laws and to approve transactions even if they
otherwise would violate the antitrust laws.

7 The Board is also required to consider the total debt (fixed charges) that
would result from the proposed transaction and the interest of rail carrier
employees affected by the proposed transaction. In addition, the Board must
consider whether there will be significant effects on the quality of the
human environment and the conservation of energy resources in its assessment
of proposed merger transactions.

8 If a merger application is approved, parties that would be affected by the
merger may ask the Board to reconsider its decision. They may also appeal
the decision directly to the federal courts. The Board

Determines Whether Mergers Are in the Public Interest and Assesses the
Implementation of Merger Conditions

Merger Approval Involves Assessing the Public Interest and Mitigating
Potential Harm to Competition

Page 6 GAO- 01- 689 Freight Railroad Regulation

During the merger review process, the Board considers comments and evidence
submitted by all interested parties, which, together with the application,
form the record upon which the Board bases its decision. The applicants as
well as interested parties may submit information on the potential public
benefits and potential harm of a proposed merger. Public benefits can
include such things as gains in a railroad?s efficiency, cost savings, and
enhanced opportunities for single- line service. 9 Potential harm can result
from, among other things, reductions in competition and harm to a competing
carrier?s ability to provide essential services- that is, services for which
there is a public need but for which adequate alternative transportation is
not available.

Whenever necessary and feasible, the Board imposes conditions on mergers
that it approves so as to mitigate potential harm associated with a merger,
including harm to competition. In determining whether to approve a merger
and to impose conditions on its approval, the Board?s concern has focused on
the preservation of competition and essential services- not on the survival
of particular carriers or enhancing competition. Board officials told us
that, while the Board?s efforts to preserve competition have primarily
focused on maintaining competitive options for those shippers that could
face a reduction in service from two railroads to service by only one
railroad, competition that is the result of having two

?nearby? railroads has also been preserved. 10 Conditions can include such
things as trackage rights, 11 switching arrangements, 12 access to another
railroad?s facilities or terminal areas, or divestiture of lines. For
example, in the UP/ SP merger, the Board granted about 4,000 miles of
trackage rights to the Burlington Northern and Santa Fe Railway (BNSF) to
address competition- related issues for those rail

9 Single- line service is the ability to transport products from an origin
to a final destination without having to transfer the shipment to another
railroad. 10 Board officials said this has been accomplished by such things
as conditioning approval of a merger on preservation of shippers? options to
(1) build (or have some other party build) a track connection to a competing
railroad (called the build- in/ build- out condition) or (2) locate new
facilities, including truck- to- rail or rail- to- truck ?transload?
facilities, on the lines of competing railroads.

11 Trackage rights are the authority of one railroad to use the tracks of
another railroad for a fee. 12 Under switching arrangements, a carrier
transports the railcars of a competing carrier at origin or destination for
a fee.

Page 7 GAO- 01- 689 Freight Railroad Regulation

corridors and shippers that could have potentially faced a reduction in
service from two railroads (UP and SP) to service by only one railroad (UP).
(See fig. 1.) The Board may also impose privately negotiated settlement
agreements as conditions to mergers. 13 The Board will normally impose
conditions only when a merger would produce effects harmful to the public
interest (such as a significant reduction in competition) and the condition
will ameliorate or eliminate these harmful effects. In addition, a condition
must be operationally feasible, produce net public benefits, and be tailored
to address the adverse effects of a transaction.

13 The Board will impose a privately negotiated agreement as a condition
only if the Board would have imposed a condition to address the identified
harm without the agreement or if the parties to the agreement request its
imposition. Board officials have stated that privately negotiated terms and
conditions are generally preferred over Board- crafted terms and conditions.

Page 8 GAO- 01- 689 Freight Railroad Regulation

Figure 1: Trackage Rights Granted to, and Lines Purchased by, BNSF in the
UP/ SP Merger

Source: GAO?s analysis of data from the Federal Railroad Administration and
Burlington Northern and Santa Fe Railway.

If a merger is approved, the Board has broad discretion to impose oversight
conditions, as well as flexibility in how it conducts oversight. Such
oversight conditions establish the Board?s intent to monitor a merger?s
implementation and to conduct annual oversight proceedings (called formal
oversight in this report). An oversight condition may also establish a time
period during which the Board will monitor the effects of The Board Conducts

Oversight of Mergers

Page 9 GAO- 01- 689 Freight Railroad Regulation

a merger. Although oversight conditions are not necessary for the Board to
retain jurisdiction over a merger- particularly with regard to carrying out
conditions the Board has imposed- oversight conditions ensure that the
Board?s retained jurisdiction will be meaningfully exercised and gives
parties an added opportunity to demonstrate any specific anticompetitive
effects of a merger. According to the Board, oversight also (1) permits the
Board to target potential problem areas for the subsequent imposition of
additional conditions if this proves warranted in light of experience, (2)
puts applicants on notice that they consummate the transaction subject to
reasonable future conditions to mitigate harm in limited areas, and (3)
helps to ensure cooperation by the merging carriers in addressing problems
and disputes that may arise following merger approval. As such, oversight
provides an additional check that Board- approved mergers are in the public
interest. When an oversight period ends, the Board has stated that it
continues to retain jurisdiction and can reopen a merger proceeding, if
necessary, to address concerns pertaining to competition and other problems
that might develop.

Board officials described postmerger oversight as a process consisting
mainly of an annual oversight proceeding. This proceeding is an examination
of the implementation of merger conditions and whether conditions have
effectively met their intended purpose. Oversight is generally conducted
each year for 5 years after a merger has been approved. 14 As part of the
oversight proceeding, public comments and supporting information are
formally submitted into the record by shippers, carriers, and other
interested parties. Periodic progress reports, which provide, among other
things, details on the implementation of conditions, are also submitted by
merging railroads as required. Board officials told us that reporting
requirements are frequently used as part of oversight and that such
reporting has served to replace the industry and merger monitoring once
conducted by ICC?s field staff. 15

As an adjudicatory body, the Board relies on parties affected by a merger to
identify whether a proposed transaction has harmed competition and, if

14 According to Board officials, this time period is based on an expected
length of 1 to 2 years for the merging carriers to prepare to integrate
computer systems and negotiate labor contracts, and another 2 to 3 years to
fully integrate the carriers operationally. In addition, Board staff told us
that the full range of anticipated merger benefits should begin to be
realized by about the third year following system integration.

15 The Board has no field staff.

Page 10 GAO- 01- 689 Freight Railroad Regulation

so, to what extent; the Board does not independently collect this type of
information. Board officials noted that it has been standard practice in
merger oversight to require relevant railroads, such as UP and BNSF in UP/
SP oversight, to make available under seal to interested parties the
railroads? confidential 100 percent traffic tapes- tapes that include
information such as shipments moved and freight revenue generated- so that
parties other than the merging carriers would also have the opportunity to
submit postmerger rate analyses to the Board. As part of the oversight
process, the Board may consider information obtained from monitoring
industry operations, such as service levels, as well as any studies
conducted, whether specific to that merger or industrywide. In conducting
formal oversight, the Board may modify existing conditions if they are not
achieving their intended purpose or may impose additional reporting
requirements if necessary. The Board also has the authority to initiate a
new proceeding to determine if additional conditions should be imposed to
address unforeseen merger- related issues.

Board officials noted that the agency engages in other activities associated
with oversight. Included are such things as informal monitoring of merging
railroads? operations and service performance 16 and responding to certain
filings, such as petitions to clarify or modify a merger condition based on
competition- related issues or other claims of merger harm.

Although the Board retains some form of oversight jurisdiction for all rail
mergers, the use of formal merger oversight has become standard only since
the mid- 1990s. Board officials told us that before 1995, formal
postapproval oversight of mergers was rare and was instituted only in
unusual situations when strong concerns about competition were present.
These officials pointed to only two cases when a period of formal oversight
was imposed prior to 1995: once in 1984 in a rail/ barge merger between CSX
Corporation and American Commercial Lines, Inc., and in 1992 as part of the
merger of Wisconsin Central Transportation Corporation and Fox Valley &
Western, Ltd. Neither case involved the merger of two or more Class I
railroads. In both cases, however, oversight conditions were imposed in
response to concerns raised about potential harm to competition.

16 Board officials told us that in informal monitoring, the Board?s Office
of Compliance and Enforcement requires merged railroads to report various
metrics, such as average train speed, in order to monitor service levels and
the operational performance of those carriers. Oversight Has Changed

Over Time

Page 11 GAO- 01- 689 Freight Railroad Regulation

In recent years, in light of the complexity of transactions and the service
and competitive issues that have arisen, the Board has expanded its use of
formal oversight of railroad mergers. ICC did not impose specific oversight
conditions on its approval of the 1995 Burlington Northern and Santa Fe
Railway merger because, according to Board officials, there were few
concerns raised in that merger about service issues or potential harm to
competition. 17 Since August 1995, when the BNSF merger was approved, the
Board has imposed oversight on all three Class I railroad mergers that it
has approved: the 1996 UP/ SP merger, the 1998 Conrail acquisition by CSX
and Norfolk Southern, and the 1999 Canadian National/ Illinois Central
merger. For two of the three transactions (UP/ SP and Conrail), the
oversight period was set for 5 years. In the third merger- Canadian National
and Illinois Central- a 5- year oversight period was established with
continuation to be reviewed annually. All three oversight periods are
ongoing.

The Board has significant discretion and flexibility to adapt its oversight
as circumstances warrant. For example, in conducting oversight in recent
years, the Board has, when necessary, incorporated additional monitoring
elements to supplement its oversight activities. For example, it has added
more reporting requirements. The UP/ SP merger provides a good illustration
of service monitoring. As the result of a service crisis 18 that developed
during the implementation of this merger, the Board required both UP/ SP and
BNSF to provide weekly and monthly reports to its Office of Compliance and
Enforcement- information which, according to Board officials, had never been
available before. These reports included statistics on such things as
average train speed, cars on line, and terminal dwell time- the time loaded
railcars spend in a terminal awaiting continued

17 The BNSF merger was largely an end- to- end merger, meaning there was
little overlap in routes and few locations where shippers could go from
service by two railroads to one. 18 The UP/ SP system started experiencing
service problems in July 1997 during integration of the two railroads. As
the result of aging rail infrastructure in the Houston area that was
inadequate to cope with a surge in demand, congestion on this system began
affecting rail service throughout the western United States. Rail service
disruptions and lengthy shipment delays continued throughout the rest of
1997 and into 1998. The Board issued a series of decisions, many focused on
the Houston/ Gulf Coast area, including an emergency service order, to
address the service crisis. The Board also initiated a separate postmerger
proceeding to consider requests from various parties for additional merger
conditions to modify the way in which rail service was being provided in the
Houston area. In that proceeding, the Board added one condition and modified
another to ease congestion in the Houston/ Gulf Coast area. The Board also
concluded that the service crisis did not stem from any competitive failure
resulting from the UP/ SP merger.

Page 12 GAO- 01- 689 Freight Railroad Regulation

movement. 19 This information allowed the Board to monitor the operations
and service levels of both railroads. Similar reporting requirements were
imposed on both CSX and Norfolk Southern in the Conrail merger. In this
instance, the Board, anticipating possible transitional service problems
during the integration process, required the weekly and monthly reports both
to monitor the merger?s implementation and to identify potential service
problems.

Board officials told us that as a result of the lessons learned in the UP/
SP merger, oversight has expanded to incorporate monitoring of operational
and service issues- in part to serve as an early warning of problems that
might occur during the merger integration process. Future mergers will also
be subject to operational monitoring. The merger rules adopted by the Board
in June 2001 state that the Board will continue to conduct significant
postapproval operational monitoring of mergers to insure that service levels
after a merger are reasonable and adequate.

In general, the Board has found few competition- related problems when
conducting oversight of recent mergers but has acted to modify some
conditions designed to address such problems when it felt such action was
necessary. Even though many of the shipper and railroad trade associations
told us that the oversight process is valuable, some shippers and small
railroads are dissatisfied with aspects of the Board?s oversight. In
addition, some larger carriers are concerned that shippers are using the
oversight process to address issues not related to mergers. The Board?s
recently adopted merger rules could affect oversight by changing the focus
of merger approval toward enhancing rather than preserving competition.

19 A Board official said the latter does not include railcars waiting for
repairs, maintenanceof- way equipment, or railcars arriving and departing on
the same train. The Board Has Acted

to Address Harm to Competition, but Some Shippers and Carriers Are Concerned
About Merger Oversight

Page 13 GAO- 01- 689 Freight Railroad Regulation

A review of oversight decisions in recent merger cases shows that the Board
has found few problems related to competition. Board officials also told us
they believe that, to date, the conditions originally imposed on mergers
have met their intended purpose and have mitigated any potential harm to
competition. In determining whether to modify a condition, 20 the Board
reviews the evidence presented, considers the nature and extent of the
alleged harm, and assesses what action may be warranted. In general, the
Board has not found it necessary to modify or add conditions during
oversight of recent mergers. However, the Board has found such action to be
appropriate in some cases. For example, in December 1998, the Board added a
condition and modified a condition in the UP/ SP merger. The added condition
addressed traffic congestion in the Houston/ Gulf Coast area; the modified
condition changed the location where BNSF railcars are transferred to
another railroad. Similarly, in 1998 and 1999, the Board modified four
conditions in the Conrail transaction. These modifications were designed to
preserve competition by, among other things, introducing a second carrier
and requiring carriers to negotiate an acceptable transfer point to
interchange railcars bound for an Indiana power plant.

Providing specific evidence of harm to competition is critical in obtaining
additional Board relief. According to the Board?s decisions, shippers and
others have sometimes alleged harm to competition during oversight without
presenting specific evidence of such harm. For example, as part of the UP/
SP merger, the Board granted over 2,100 miles of trackage rights to BNSF on
the Central Corridor 21 to preserve competition for those shippers that
could have been reduced from service by two carriers (UP and SP) to service
by only one (the merged UP/ SP) and for those exclusively served shippers
who benefited from having another railroad nearby. Some organizations have
asserted that, despite the trackage rights, postmerger competition has not
been adequate on this corridor. However, in its UP/ SP oversight decisions,
the Board has concluded that postmerger competition on this corridor has
been adequate, in part because no shippers came forward with specific
evidence of harm. In another instance, in the Conrail

20 In addition to considering whether to modify or add conditions, Board
officials said the merger oversight process has also been used to clarify
conditions imposed in the original merger approval.

21 The Central Corridor is generally defined as an area stretching from St.
Louis, Missouri, to Oakland, California, by way of Denver, Colorado; Salt
Lake City, Utah; and Reno, Nevada. BNSF?s trackage rights were over the
Denver- to- Oakland segment of the corridor. The Board Has Found Few

Postmerger Problems Regarding Competition in Recent Merger Oversight
Proceedings and Has Acted to Address Them

Page 14 GAO- 01- 689 Freight Railroad Regulation

merger, the Board granted trackage rights to Norfolk Southern to access a
power plant in Indiana. In order to use the trackage rights, Norfolk
Southern negotiated a fee with CSX. The power plant owner believed that the
negotiated fee was too high to allow adequate competition between the
railroads and requested a lower fee so that Norfolk Southern could compete
for its business. In denying this request, the Board stated that the
evidence of harm presented was not sufficient, in part because both CSX and
Norfolk Southern demonstrated that the negotiated fee would amount to only a
minimal cost increase ($ 0.004 per ton) over the amount the Board had
previously found to be reasonable.

A review of merger oversight documents shows the Board has acted to address
competition- related postmerger issues when it believed such action was
necessary. For example, during oversight of the Conrail acquisition, the
Board reduced fees for trackage rights and switching charged to Canadian
Pacific to permit competition between CSX and Canadian Pacific Railway in
the Albany, New York, to New York City corridor. Although the Board had
initially set these fees in a postmerger decision, the Board later
determined that the fees were too high to allow Canadian Pacific to use CSX
tracks to provide meaningful competition between the carriers. Consequently,
the Board acted to reduce the fees to promote competition. The Board also
acted during the Conrail oversight period to void provisions in two
contracts between CSX Intermodal, Inc., a rail shipper, and Norfolk Southern
that required Norfolk Southern to be the primary carrier of CSX Intermodal
goods between northern New Jersey and Chicago during the contract period.
Voiding these provisions allowed CSX immediately to compete with Norfolk
Southern for these shipments.

Shipper and railroad trade associations and railroad companies with whom we
spoke believe postmerger oversight is a valuable process. Officials from the
National Grain and Feed Association and the National Industrial
Transportation League told us that the Board has always been willing to
listen to their concerns. Officials from Norfolk Southern and BNSF said the
merger oversight process provides shippers and railroads with an opportunity
to submit merger- related questions, problems, and concerns. Railroad and
railroad association officials stated that the Board acts to protect the
interests of the public and the shipping community by allowing railroads and
shippers to work together during oversight to resolve actual and potential
merger- related problems. Officials from one trade association said that
without an oversight process, their members might be faced with a less
desirable alternative. For example, officials All Parties Believe

Oversight Is Valuable, but Some Shipper and Railroad Associations Are
Dissatisfied

Page 15 GAO- 01- 689 Freight Railroad Regulation

from the American Chemistry Council told us that the only other option for
shippers would be to use the Board?s time- consuming and expensive complaint
process. Officials from the American Chemistry Council, as well as officials
from UP and BNSF, said a 5- year oversight period has been a benefit to both
railroads and shippers. However, an American Chemistry Council official said
some mergers may need oversight for a longer or shorter period than 5 years
and that it is unclear what type of oversight will occur after the 5- year
oversight period for the UP/ SP merger expires in 2002.

Despite seeing oversight as a valuable process, some shipper and small
railroad associations are dissatisfied with aspects of the Board?s oversight
procedures. A number of reasons were cited. The Board has been viewed as
unresponsive to concerns of shippers and small railroads. For example, an
official representing the Edison Electric Institute told us that it had
expressed concern to the Board in 2000 about the degree of competition for
the transport of Utah and Colorado coal in the Central Corridor, but that
the Board declined to answer questions about this issue. 22 An official from
the American Chemistry Council expressed similar frustration that the Board
did not adopt any part of a plan developed by shippers and others to address
the Houston/ Gulf Coast service crisis that occurred during the
implementation of the UP/ SP merger. This plan had broad support from both
private sector and state government officials. 23 Dissatisfaction was also
expressed about the time and resources required for preparing and submitting
comments during the postmerger oversight period, especially for small
shippers. For example, officials from the Edison Electric Institute and the
American Chemistry Council told us that small shippers might not have the
time or the money to invest in the formal oversight process. Finally,
officials from several shipper associations and the American Short Line and
Regional Railroad Association (an association representing smaller
railroads) said their

22 Board officials told us that there was very little coal traffic affected
by the merger (primarily in Utah) involving shippers that could have been
reduced in service from two railroads to only one railroad. Board officials
said that BNSF has competed for, and obtained, some of this traffic.

23 Board officials noted that the Board issued a series of decisions, many
focused on Houston, including an emergency service order, to address the
service crisis. Board officials said that the plan developed by the shippers
and others would have taken property from UP and injected additional
operators without improving service.

Page 16 GAO- 01- 689 Freight Railroad Regulation

members are discouraged from participating in the oversight process, in part
because of the reasons cited above. 24

Although generally satisfied with the Board?s oversight process, officials
at some Class I railroads have cited certain drawbacks to it. For example,
officials at Norfolk Southern, CSX Transportation, and UP said some shippers
use the formal oversight process as a mechanism to raise nonmerger- related
issues, which they claim have protracted the oversight process. Railroad
officials told us that inviting comments by interested parties allows them
to reintroduce issues that were initially denied during the merger approval
process. They noted that, as a result, they must invest their time to
address non- merger- related issues. Officials with Norfolk Southern said
that if the Board allows parties to reintroduce issues already decided, this
could delay implementation of a merger.

Board officials told us that oversight is an open process and anyone can
submit comments. The basis for making decisions is the merger and postmerger
oversight record and Board officials said they encourage parties such as
shippers, railroads, and others to submit information into the record so
that the Board can act with as much information as possible. However, Board
officials acknowledged that parties sometimes reargue issues during
oversight that were not decided in their favor in the merger decision. For
example, in its November 2000 oversight decision in the Canadian National/
Illinois Central merger, the Board refused to require that Canadian National
sell its share of the Detroit River Tunnel as requested by various parties.
The parties were concerned that Canadian National would competitively
disadvantage the Detroit River Tunnel by not allowing needed capital
investments to be made and favoring another nearby tunnel it owned. The
Board found that this issue was not directly related to the merger and was a
matter being privately negotiated between the parties. Finally, Board
officials have said the oversight process has evolved over time and the
Board has incorporated additional reporting and other requirements to
provide more information on actual and potential problems experienced during
merger implementation. Moreover, the Board has focused on preserving, not
enhancing, competition and does not seek to restructure the competitive
balance of the railroad industry during postmerger oversight.

24 Board officials noted, however, that numerous regional and shortline
railroads, as well as smaller shipper interests and communities,
participated in recent merger and merger oversight proceedings.

Page 17 GAO- 01- 689 Freight Railroad Regulation

Both shipper association and railroad officials with whom we spoke
recognized that the Board has a limited number of staff to conduct formal
oversight. According to officials from the American Short Line and Regional
Railroad Association, the Board?s perceived slowness in handling oversight
issues may be attributable to the significant amount of information that
needs to be processed during the annual oversight proceeding- information
that is generally handled by a core team of 15 employees (who, Board
officials noted, also work on agency matters other than mergers). Board
officials acknowledged that their resources are limited. However, they said
oversight offers an open, no- fee process in which any interested party may
participate. They also said the Board has issued in a timely manner its
decisions in the annual oversight proceeedings, as well as in matters
involving specific material issues during oversight.

The rail consolidation rules issued in June 2001 could change how the Board
conducts oversight by providing for merger applications to include plans to
enhance competition and to ensure reasonable service and by holding
applicants accountable if they do not act reasonably to achieve promised
merger benefits. Shifting the focus of merger review towards enhancing
competition and ensuring reasonable service, as well as including some
degree of accountability for postmerger benefits, could require the Board to
expend additional time and resources reviewing these issues. For example,
the final rules would call upon merger applicants to enhance competition so
as to offset any negative effects resulting from a merger, such as potential
harm to competition and disruptions of service. This could affect the way
the Board uses and oversees conditions during the merger approval and
oversight processes. Similarly, to require railroads to calculate the net
public benefits to be gained through a proposed merger and to hold them
accountable for acting reasonably to achieve these benefits, such as
improved service, the Board will monitor as part of the general oversight
proceeding the realization of merger benefits claimed. These activities
would enlarge the current focus of assessing whether conditions are working
as intended. In the event that public benefits fail to materialize after a
merger is approved, the Board said it would consider the applicant?s
proposals for additional measures. 25

25 The Board did not specify what these additional measures might be in the
final rules. New Merger Rules Could

Change Oversight but May Not Address All Concerns of Shippers and Carriers

Page 18 GAO- 01- 689 Freight Railroad Regulation

It is not likely that the final merger rules will resolve all concerns
expressed by shipper and railroad organizations about oversight. The final
rules will not change the basic process established for oversight. While the
final rules may address concerns of shippers and railroads about service
levels by requiring merger applicants to develop service assurance plans,
they will not address more general concerns that the Board is not responsive
to their issues. 26 Furthermore, the final rules will not likely address
concerns about the time and resources necessary to participate in postmerger
oversight. Rather, the amount of time and resources required could increase,
given that during oversight the Board will assess enhancement of
competition, service issues, and accountability for proposed merger benefits
as well as whether conditions are working as intended. In addition, issues
may continue to be introduced that are not directly related to the merger
under review. Board officials said they do not consider participation in
oversight to be an expensive or burdensome process. However, they
acknowledged that the new merger rules would require applicants to provide
more detailed information on competition, service, and benefits as part of
the merger application and that the amount of time and resources required
during oversight could increase.

Finally, the final rules may also not address all of the shippers? concerns
about the extent of competition in the rail industry resulting from mergers.
While provisions regarding the enhancement of competition may address some
competition- related issues, it is not clear how these provisions will be
implemented. Both shipper and railroad officials told us that enhanced
competition had not been defined in the proposed rules and, therefore, they
were not clear how the provisions might affect specific situations involving
competition. The final rules acknowledge that the Board cannot predict in
advance the type and quantity of competitive enhancements that would be
appropriate in a particular merger proposal. Lastly, the new merger rules
make clear that the Board will not use its authority to impose conditions
during merger approval to provide a broad program of open access. 27

26 However, the final rules will require merging railroads to address
regional and shortline railroad issues as part of oversight in order to
monitor potential adverse effects of a merger on these railroads.

27 A system of open access would allow shippers, wherever possible, to be
served by more than one railroad even if, to produce such a system,
railroads that own the rail infrastructure used would be required to share
their property with others that do not own this property.

Page 19 GAO- 01- 689 Freight Railroad Regulation

We analyzed the effects of the 1996 UP/ SP merger on rail rates 28 in two
selected geographic markets that have high concentrations of shippers that
faced a reduction in service by two railroads to service by only one
railroad (called 2- to- 1 shippers). We found that the merger reduced rail
rates for four of the six commodities we reviewed. However, in one instance,
the merger placed upward pressure on rates, even though other factors caused
overall rate decreases. For the remaining commodity, rates were relatively
unchanged. Our analysis illustrates that the Board could make more informed
decisions during oversight about whether merger conditions are protecting
against harm to competition, as measured by the merger?s effect on rates, if
it had information that separated rate changes specifically resulting from a
merger from rate changes caused by other factors.

A merger reduces the number of rail carriers and can potentially enhance the
market power of remaining carriers. This enhanced market power could be used
to profitably increase rail rates if no action were taken to preserve
competition. Board officials told us that rate trends are a good indicator
of postmerger competition. In 1996, UP acquired SP in a transaction that
raised significant competition- related issues. This merger encompassed a
number of geographic areas where the loss of competition from SP could have
reduced the number of carriers from 2 to 1. Most of these areas were in
Texas and Louisiana, but some were in the Central Corridor between
California and Colorado. (See fig. 1.) In granting trackage rights to BNSF
in this merger, the Board sought to replace the competition for potential 2-
to- 1 shippers in these geographic areas. To understand how the UP/ SP
merger affected rail rates, we looked at rail rates in two geographic areas-
Reno, Nevada, and Salt Lake City, Utah- both in the Central Corridor. We
selected these areas because they had high concentrations of potential 2-
to- 1 shippers and, according to BNSF and UP/ SP officials, were less
affected by the service crisis that developed during implementation of the
UP/ SP merger. 29 They also provided relatively clear examples of where BNSF
service substituted for SP service.

The primary commodities shipped to and from Reno and Salt Lake City were
nonmetallic minerals (such as barites) and chemicals (such as

28 Rail rates in this section are inflation- adjusted gross revenue per ton-
mile of freight originated, in 1996 dollars. 29 A more complete discussion
of our rate analysis is presented in app. III. The Board Could

Benefit From Evidence Better Identifying How Mergers Affect Rail Rates

UP/ SP Merger Expected to Affect Competition in Selected Geographic Areas

Page 20 GAO- 01- 689 Freight Railroad Regulation

sulfuric acid or sodium). (See table 1.) Farm products (such as corn and
wheat) accounted for about 13 percent of the traffic shipped to Salt Lake
City. We also included coal in our analysis of Salt Lake City rail rates,
since it accounted for the highest percentage of carloads shipped to and
from that area. However, BNSF officials told us that, in general, they have
not yet used the trackage rights they were granted to transport coal to or
from the Salt Lake City area. In its decision approving the UP/ SP merger,
the Board noted that BNSF was granted access to only a small portion of coal
traffic on the Central Corridor, mostly in the northwestern section of Utah.
As the table shows, the potential 2- to- 1 shippers served by BNSF, as a
percentage of total shippers in these geographic areas, ranged from 10 to 22
percent. This is consistent with comments made by Board officials that BNSF
received trackage rights to serve about 20 percent of the postmerger UP/ SP
traffic on the Central Corridor.

Page 21 GAO- 01- 689 Freight Railroad Regulation

Table 1: Volume of Shipments, in Carloads, From the Reno and Salt Lake City
Areas, 1994- 99

Location of shipment a Commodity b

Commodity?s carload share of

area traffic ( in percent)

BNSF 2- to- 1 shippers to all shippers

(in percent)

From the Salt Lake City area Chemicals 44 c 15 From the Salt Lake City area
Coal 69 d To the Salt Lake City area Farm products 13 e 22 To the Salt Lake
City area Coal 52 d From the Reno area Nonmetallic

minerals 79 10 To the Reno area Chemicals 32 f 15 a The areas are business
economic areas. An economic area is a collection of counties used by the
Bureau of Economic Analysis, within the U. S. Department of Commerce, for
statistical reporting of regional economic activity. b The commodities
reported are generally those with the highest share of area carloads, from
1994

through 1999, excluding 1996 and intermodal traffic. c Excludes coal
shipments.

d Not applicable. e Excluding coal shipments, the commodity with the largest
market share of traffic to the Salt Lake City

area (in carloads) was waste and scraps (28 percent). However, the BNSF
share of this traffic was only 1 percent. Consequently, we used the
commodity with the next highest percentage of joint traffic- farm products.
f The commodity with the largest market share of traffic to the Reno area
(in carloads) was coal (50

percent). However, BNSF did not ship coal to the Reno area in 1999, the most
recent year of our study. Consequently, we used the commodity with the next
highest percentage of joint traffic- chemicals.

Source: GAO?s analysis of Surface Transportation Board data.

Our analysis found that by itself the merger would have served to reduce
rates for four of the six commodities shipped to or from the geographic
areas we chose. (See table 2.) Specifically, the merger would have reduced
rates for coal shipments to and from the Salt Lake City area (by 8 percent
and 10 percent, respectively), chemical shipments from the Salt Lake City
area (by 6 percent), and farm products to the Salt Lake City area (by 5
percent). However, the rates for shipments of chemicals to the Reno area UP/
SP Merger Generally

Decreased Rail Rates, but Not for All Commodities and Not for All Shippers

Page 22 GAO- 01- 689 Freight Railroad Regulation

would have increased by 21 percent because of the merger, 30 while rates for
shipments of nonmetallic minerals originating in the Reno area would have
been relatively unchanged by the merger (i. e., the merger- related change
was not statistically significant). 31 The effect of a merger on rail rates
depends on the cost savings the merger might generate relative to the
exercise of any enhanced market power by the railroad carriers. Since the
Board acted to preserve the level of competition by granting trackage rights
to BNSF to serve potential 2- to- 1 shippers in these geographic areas, the
rate decreases from the merger likely reflect cost savings from the
consolidation. Another way in which the merger could result in lower rates
is if BNSF provided more effective competition to UP in the postmerger
period than SP did in the premerger period.

30 This result may be attributable to changes in competition between the
premerger and postmerger periods. Compared with the Salt Lake City area, the
Reno area has a very small volume of chemicals shipments- based on carloads,
the Salt Lake City area has more than three times the volume of the Reno
area. Given the high fixed costs in the rail industry, large volumes of
shipments are generally necessary to attract competition. This result may
also reflect cost differences between SP and BNSF. According to BNSF
officials, SP, because it had access to both 2- to- 1 and non- 2- to- 1
shippers in the Reno area, had substantially more customers in the premerger
period than BNSF (which has access only to 2- to- 1 shippers) has in the
postmerger period. As a result, SP could have spread its costs among more
customers and therefore offered more competitive rates to UP than rates
currently offered by BNSF. Without commenting on Reno or Salt Lake City,
Board officials also noted that, overall, SP?s premerger rates were not
covering all its costs and were thus not sustainable. Board officials
further noted that BNSF has been able to successfully compete for the
business of a facility near Reno after the Board clarified that this
facility qualified for BNSF service under the Board?s new facility
condition. Finally, Board officials said that business from new facilities
such as this would add to BNSF?s volumes and enable it to compete more
effectively for other traffic.

31 For the nonmetallic minerals, the changes were not statistically
significant, meaning there was no meaningful difference between premerger
and postmerger rates.

Page 23 GAO- 01- 689 Freight Railroad Regulation

Table 2: Changes in Postmerger Rail Rates Due to Merger and Other Factors,
1994- 99

Changes in percent Shipments Commodity

Shipper/ postmerger railroad

Rate changes

due to merger

Rate changes

due to other factors

Overall changes

in rates

From the Salt Lake City area Chemicals All shippers/

BNSF & UP -6 16 10

From the Salt Lake City area Coal All shippers/

UP -10 20 10

To the Salt Lake City area Farm

products All shippers/ BNSF & UP -5 6 1

To the Salt Lake City area Coal All shippers/

UP -8 23 15

From the Reno area Nonmetallic minerals All shippers/

BNSF & UP 4 -26 -22

To the Reno area Chemicals All shippers/ BNSF & UP 21 -27 -6

Notes: The overall rate changes and the rate changes reflecting effects of
the merger are determined by two different methods. Overall rate changes,
which are unweighted, are based on a meandifference analysis that subtracts
the premerger rates from the postmerger rates. Results for the rate changes
due to the merger are based on an econometric analysis. For the overall
changes in rates and the merger effects, all values in bold/ italics are
statistically significant at the 5- percent level. The effects due to
changes in other factors are calculated as the overall changes in rates less
the merger effects.

See also notes to table 1. Source: GAO?s analysis of Surface Transportation
Board data.

While the effects of a merger can put downward (or upward) pressure on
rates, an analysis focused on overall rate changes alone could lead to an
inaccurate conclusion about whether conditions imposed on a merger to
mitigate potential harm to competition have been effective. The results of
our analysis indicate that, in addition to merger effects, other factors,
such as the volume of shipments, had an equal or greater influence on
overall rate changes for the specific movements we examined. In some cases,
the effects of these other factors were strong enough to offset or even
reverse the downward pressure of the merger on rates. (See table 2.) For
example, for shipments of chemicals from the Salt Lake City area and for
shipments of coal to and from the Salt Lake City area, while the merger
alone would have decreased rates, the rates nevertheless increased overall.
On the

Page 24 GAO- 01- 689 Freight Railroad Regulation

other hand, while rates decreased overall for chemicals shipments to the
Reno area, the merger by itself put an upward pressure on rates. 32

Finally, we found that postmerger rates for potential 2- to- 1 shippers
(served by BNSF) in the Reno and Salt Lake City areas decreased for one of
the commodities we looked at but were essentially unchanged in three other
instances. 33 (See table 3.) The rate changes for potential 2- to- 1
shippers (served by BNSF) shipping chemicals from the Salt Lake City area
were about 16 percentage points less than similar rates for shippers
shipping similar products but served solely by UP. 34 However, rail rate
changes for potential 2- to- 1 shippers (served by BNSF) who shipped farm
products to the Salt Lake City area, nonmetallic minerals from the Reno
area, and chemicals to the Reno area were all higher than for shippers
served exclusively by UP, but this difference was not statistically
significant, meaning that the rates were essentially unchanged. These
results are not wholly unexpected, since the levels of rail competition for
the two kinds of shippers- potential 2- to- 1 and non- 2- to- 1- differ and
rail rates are set using differential pricing. 35 Under differential
pricing, shippers

32 As table 2 shows, the overall changes in rates are different between the
Salt Lake City and Reno areas. This may be attributable to several reasons.
First, the volume of shipments from the Reno area is much smaller compared
with the Salt Lake City area. This difference in volume can magnify the
impact of slight changes in tonnage on our measured rail rates (revenue per
ton- mile). Second, costs between the two areas were different. For the
commodities that we examined, postmerger costs were slightly lower in the
Reno area compared with the Salt Lake City area (see app. III). Finally,
there was a change in demand for some products shipped from the Reno area.
According to UP officials, the decline in the prices of oil, gold, and
copper in the latter part of the 1990s reduced the demand for the mining-
related products shipped from the Reno area. Each of these nonmerger factors
may have contributed to decreased overall rates in the Reno area.

33 Although the UP traffic used in our analysis consists of both 2- to- 1
shippers and non- 2- to1 shippers, according to UP officials, most of the
postmerger traffic in Reno and Salt Lake City retained by UP after the
merger was solely served by UP prior to the merger.

34 Similar statistical analyses could not be performed for the overall rate
changes because the data were unbalanced- that is, the sample sizes were not
the same. See app. III for more information.

35 These results were similar to those found in other recent studies of rail
rates (see Curtis Grimm and Clifford Winston, ?Competition in the
Deregulated Railroad Industry: Sources, Effects, and Policy Issues? in
Deregulation of Network Industries: What?s Next?, Peltzman and Winston
(ed.), AEI- Brookings Joint Center for Regulatory Studies: Washington, DC:
2000. This study, which is based on a survey of shippers who are members of
the National Industrial Transportation League, Edison Electric Institute,
and the Alliance for Rail Competition, and which uses a more restrictive
definition of shippers served solely by one railroad than the definition we
applied here, estimated that shippers served solely by one railroad paid
freight charges that were about 21 percent higher than those paid by other
shippers.

Page 25 GAO- 01- 689 Freight Railroad Regulation

with less effective transportation alternatives generally pay a
proportionately greater share of a railroad?s fixed costs than shippers with
more effective transportation alternatives. 36

Table 3: Changes in Postmerger Rail Rates for Potential 2- to- 1 Shippers
Compared With Rail Rates for Shippers Served Solely by UP

Shipments Commodity a Shipper category Rate changes

due to merger b (in percentage points)

From the Salt Lake City area Chemicals 2- to- 1 shippers compared

with shippers served solely by UP

-16

To the Salt Lake City area Farm products 2- to- 1 shippers compared

with shippers served solely by UP

0.3 From the Reno area Nonmetallic

minerals 2- to- 1 shippers compared with shippers served solely

by UP 6

To the Reno area Chemicals 2- to- 1 shippers compared with shippers served
solely by UP

4 a Coal is excluded from this table because there were no potential 2- to-
1 shippers for this commodity in either the Salt Lake City or the Reno
areas. b This column shows the rate changes for potential 2- to- 1 shippers
(served by BNSF) less the rate changes for shippers served solely by UP/ SP.
A negative value indicates that rates for potential 2- to 1 shippers
decreased more than the rates for shippers served solely by UP. The numbers
are shown as percentage point differences. This is the difference between
two percentage values. All values in bold /italics are statistically
significant at the 5- percent level.

See also notes to table 1. Source: GAO?s analysis of Surface Transportation
Board data.

There are limitations in the analysis and data we used. The results
presented are only for the two geographic markets we reviewed and cannot be
generalized to other geographic locations or for rate changes from the UP/
SP merger as a whole. In addition, although econometric models of the
factors that determine rail rates have been used to analyze a variety of
policy- related issues in rail transportation 37 and have been useful, such
a model can be sensitive to how it is specified. We tested the

36 For more information on differential pricing, see Railroad Regulation:
Changes in Railroad Rates and Service Quality Since 1990 (GAO/ RCED- 99- 93,
Apr. 16, 1999). 37 See, for example, studies of the 1980 Staggers Rail Act
on rail rates by Wesley Wilson,

?Market- Specific Effects of Rail Deregulation,? Journal of Industrial
Economics, Vol. XLII (1994), pp. 1- 22.

Page 26 GAO- 01- 689 Freight Railroad Regulation

model?s key results to ensure that our findings were reliable and are
confident that the results are reasonable for the commodities in the
geographic areas we examined. Finally, the Carload Waybill Sample data used
in our model also have limitations. For example, these data do not
necessarily reflect discounts or other rate adjustments that might be made
retroactively by carriers to shippers exceeding certain volume requirements.
38

Our analysis provides an example of how rates subject to merger conditions
could be analyzed. Although the results in this study are not directly
comparable to those in other studies of rates that are based on broader
geographic areas, our analysis suggests that overall rate changes do not
identify the specific impact of mergers on rates. In general, the Board has
been presented with rate studies that have focused on overall rate changes,
not on the portion of changes caused by a merger. For example, rate studies
39 prepared by UP during merger oversight indicate that, overall, rates
decreased immediately after the merger and have continued to decrease at 2-
to- 1 points and for traffic moving in the Houston- Memphis and Houston- New
Orleans corridors. Similarly, both CSX and Norfolk Southern have conducted
studies of rail rates in the Buffalo, New York, area since their acquisition
of Conrail in 1999. Again, these studies have focused on the overall
direction of rate changes and have shown that rail rates in the Buffalo area
have generally decreased. Neither the UP nor the CSX/ Norfolk Southern rate
studies identified the specific effects of mergers on rates- effects that
could have potentially been different from the overall rate trends.

According to Board officials, in general, the parties in merger oversight
proceedings have focused on determining the overall magnitude and direction
of rate changes without trying to relate such changes to specific causes,
and the Board?s own December 2000 staff study of nationwide

38 For additional information on the data and model used in this analysis as
well as limitations, see apps. I and III. 39 UP has conducted three studies
of rail rates associated with its merger with SP. They compare the periods
October 1995- March 1996 to October 1996- March 1997, October 1996March 1997
to October 1997- March 1998, and October 1997- March 1998 to October
1998March 1999. These studies have been used during oversight of the UP/ SP
merger. The Board?s Oversight

Could Benefit From Evidence Better Identifying Merger- Related Rail Rate
Changes

Page 27 GAO- 01- 689 Freight Railroad Regulation

changes in rail rates took this approach. 40 Board officials said they have
attempted to take into account, in the context of postmerger oversight, such
non- merger- related factors as the recent significant rise in diesel fuel
prices but have not been presented with an econometric approach to analyze
rail rates in the context of merger oversight. They said that they had
questions and concerns about the precision and reliability of the analysis
we conducted. However, the Board is amenable to seeing this general approach
developed in the context of a public merger oversight record where it would
be subject to scrutiny and refinement by relevant parties. Board officials
noted that presenting and rebutting econometric studies, because of their
sophisticated nature, could increase the burden of participating in the
merger oversight process. It is important to note that the Board, in
approving the UP/ SP merger, was provided with various empirical rate
studies by the applicants and interested parties that included econometric
analyses. 41 In addition, econometric evidence has played an important role
in merger- related cases that have been reviewed by courts and other
government agencies. 42

As an adjudicatory agency, the Board relies on affected parties to identify
alleged harm when it exercises oversight to ensure that conditions

40 See Surface Transportation Board, Rail Rates Continue Multi- Year
Decline, Office of Economics, Environmental Analysis, and Administration
(Dec. 2000). See also GAO/ RCED- 99- 93, which reviewed an earlier similar
Board rate study. According to Board officials, the Board?s 2000 staff study
showed that rail rates in the West were stable from 1992 to 1994, but
resumed their long- term decline once the restructuring of the western rail
network had begun, and fell 9 percent, or about 3.1 percent per year, during
the 3- year period following the UP/ SP merger. Rail rates on coal movements
in the West declined even faster during this 3- year period- 14. 2 percent,
or about 5 percent per year. The Board stated that rate decreases of this
magnitude could not have been realized if the UP/ SP and BNSF mergers had
substantially decreased rail competition in the region.

41 See Surface Transportation Board, Finance Docket No. 32760, Decision 44
(Aug. 6, 1996), pp. 119- 121 and 267- 273. According to Board officials,
these studies were considered but were found to contain certain flaws that
limited their applicability in that proceeding. Board officials also noted
that no party has attempted to submit such an analysis in an oversight
proceeding to date.

42 For instance, in FTC v. Staples, Inc., 970 F Supp. 1066 (D. D. C. 1997)
the court relied heavily on the results of econometric analysis. According
to a Federal Trade Commission official, one of the first examples of the use
of econometrics in a regulatory proceeding was in connection with the
hearings on concentration in American industry conducted by the Temporary
National Economic Committee on the iron and steel industry in 1940. (See
prepared remarks of Jonathan B. Baker, Director, Bureau of Economics,
Federal Trade Commission, Before the George Mason University Law Review
Symposium, Oct. 11, 1996.) Conclusions

Page 28 GAO- 01- 689 Freight Railroad Regulation

imposed in railroad mergers are working and that competition has not been
harmed. Therefore, it is necessary for shippers, railroads, or others not
only to identify instances when they have been, or might be, harmed, but
also to present evidence to the Board demonstrating this harm. For the Board
to make sound decisions about the extent to which mergers affect rate
changes, the Board should have information that separately identifies the
factors that affect rates and the specific impact of these factors. Without
such information, the Board?s ability to evaluate whether merger conditions
have been effective in protecting against potential harm to competition may
be limited.

To better assist the Board in the oversight of railroad mergers and in
ensuring that conditions imposed in such mergers protect against potential
harm to competition, we recommend that the Board, when appropriate, require
railroads and others to provide information to the Board that separately
identifies the factors affecting postmerger changes in rail rates and the
specific impact of these factors on rate changes. In particular, the Board,
when appropriate, should require railroads and others to provide information
that identifies the effects of mergers on changes to rail rates,
particularly in those geographic areas subject to potential reductions in
competition. This information should be considered in deliberations on the
need to modify conditions, add reporting requirements, or initiate
proceedings to determine if additional conditions are required to address
competition- related issues.

We provided a draft of this report to the Surface Transportation Board and
the Department of Transportation for their review and comment. The Board did
not express an overall opinion on the draft report, but rather supplied
suggested revisions to it. Most importantly, while the Board is amenable to
seeing an econometric approach developed in the context of a public
oversight record, it commented that such an approach could increase the
burden of the parties participating in the merger oversight process. This
increased burden might occur because of the effort entailed to develop,
present, and rebut econometric studies. We agree that an increased burden
might occur and incorporated this view into our report. Allowing parties to
critique the usefulness of our recommendation and the effort involved in
implementing it should provide the Board with the information it needs on
implementation. The Board offered extensive clarifying, presentational, and
technical comments which, with few exceptions, we incorporated into our
report. Recommendation for

Executive Action Agency Comments and Our Evaluation

Page 29 GAO- 01- 689 Freight Railroad Regulation

The Department of Transportation did not express an overall opinion on the
draft report. Its comments were limited to noting that several Class I
railroads were under common control. We incorporated this change into our
report.

As agreed with your office, unless you publicly announce its contents
earlier, we plan no further distribution of this report until 15 days after
the date of this letter. At that time, we will send copies of the report to
congressional committees with responsibilities for transportation issues;
the Secretary of Transportation; the Acting Administrator of the Federal
Railroad Administration; the Chairman of the Surface Transportation Board;
and the Director, Office of Management and Budget. We will also make copies
available to others upon request. This report will also be available on our
home page at http:// www. gao. gov.

If you or your staff have any questions about this report, please contact me
at (202) 512- 2834. Key contributors to this report were Stephen Brown,
Helen Desaulniers, Leonard Ellis, John Karikari, Tina Kinney, Richard
Jorgenson, Mehrzad Nadji, Melissa Pickworth, James Ratzenberger, and Phyllis
Scheinberg.

Sincerely yours, John H. Anderson, Jr. Managing Director, Physical

Infrastructure Issues

Appendix I: Chronology of Class I Railroad Mergers- August 1995 Through June
2001

Page 30 GAO- 01- 689 Freight Railroad Regulation

1995

Burlington Northern Inc. and Burlington Northern Railroad Company -

Control and Merger - Santa Fe Pacific Corporation and the Atchison, Topeka
and Santa Fe Railway Company

Merger approval date: August 16, 1995 Total route mileage: 35,400 Service
area: Western United States and Canada Acquisition cost: $1.3 billion, plus
assumed liabilities Type of merger: Largely end- to- end. However, in
approving

this merger, ICC found that of the approximately 29 locations that were
served by both railroads, only a few would have potentially sustained harm
from reduced competition given the presence of other railroads and of
extensive truck competition at many of the locations. Conditions were
attached to preserve competition where necessary. 1996

Union Pacific Corporation, Union Pacific Railroad Company, and Missouri
Pacific Railroad Company - Control and Merger - Southern Pacific Rail
Corporation, Southern Pacific Transportation Company, St. Louis Southwestern
Railway Company, SPCSL Corp., and The Denver and Rio Grande Western Railroad
Company

Merger approval date: August 6, 1996 Total route mileage: 38,654 Service
area: Western United States Acquisition cost: $3.3 billion in cash and
stock, plus assumed

liabilities Type of merger: Significant parallel components. In

approving this merger, the Board granted about 4,000 miles of trackage
rights to BNSF and other railroads to protect potential 2- to- 1 shippers
and others from loss of competition. Appendix I: Chronology of Class I
Railroad

Mergers- August 1995 Through June 2001

Appendix I: Chronology of Class I Railroad Mergers- August 1995 Through June
2001

Page 31 GAO- 01- 689 Freight Railroad Regulation

1997 No Class I merger transactions. 1998

CSX Corporation and CSX Transportation, Inc., Norfolk Southern Corporation
and Norfolk Southern Railway Company - Control and Operating Leases/
Agreements - Conrail, Inc. and Consolidated Rail Corporation

Merger approval date: July 20, 1998 Route mileage (CSX and CSX portion of
Conrail): About 23,000 Route mileage (Norfolk Southern and Norfolk Southern
portion of Conrail): About 21,800 Service area: Eastern United States and
Canada Acquisition cost: $9.9 billion, plus assumed liabilities

and fees Type of merger: Largely end- to- end.

Additional information: Although CSX Corporation and Norfolk Southern
Corporation jointly acquired Conrail and then divided most of the assets
between them, Conrail continues to operate certain shared assets areas for
the joint benefit of CSX and Norfolk Southern. These shared assets areas are
located in North Jersey (generally from northern New Jersey to Trenton, New
Jersey), South Jersey/ Philadelphia (generally from Trenton, New Jersey, to
Philadelphia and southern New Jersey), and Detroit. Both CSX and Norfolk
Southern have the right to operate their own trains, with their own crews
and equipment and at their own expense, over any track included in the
shared assets areas. Various other areas formerly operated by Conrail are
subject to special arrangements that provide for a sharing of routes or
facilities to a certain extent. For example, the Monongahela Area in
Pennsylvania and West Virginia, although conveyed to Norfolk Southern, is
available to CSX on an equal- access basis for 25 years, subject to renewal.

Appendix I: Chronology of Class I Railroad Mergers- August 1995 Through June
2001

Page 32 GAO- 01- 689 Freight Railroad Regulation

1999

Canadian National Railway Company, Grand Trunk Corporation, and Grand Trunk
Western Railroad Incorporated - Control - Illinois Central Corporation,
Illinois Central Railroad Company, Chicago, Central and Pacific Railroad
Company and Cedar River Railroad Company

Merger approval date: May 21, 1999 Total route mileage: 18,670 Service area:
Midwestern United States and Canada Acquisition cost: $1.8 billion, plus the
value of 10.1 million

common shares of Canadian National stock Type of merger: End- to- end.

2000 No Class I merger transactions. 2001 No Class I merger transactions
proposed through June 2001.

Appendix II: Scope and Methodology Page 33 GAO- 01- 689 Freight Railroad
Regulation

Our review focused primarily on the Board?s oversight of Class I railroad
mergers that occurred since its creation in January 1996. These mergers
included (1) the Union Pacific Railroad Company (UP) with the Southern
Pacific Transportation Company (SP), (2) the Canadian National Railway
Company with the Illinois Central Railroad and (3) the acquisition of the
Consolidated Rail Corporation (Conrail) by CSX Transportation, Inc., and the
Norfolk Southern Corporation. However, to aid in showing how merger
oversight has changed over time, we also included information on the
Burlington Northern Railroad Company merger with the Atchison Topeka and
Santa Fe Railway Company, which was approved by ICC in August 1995.

To address the role of the Board in approving and overseeing railroad
mergers and to determine how merger oversight is conducted, we reviewed
relevant laws and regulations and analyzed documents prepared by the Board
addressing its merger authority and functions. We also discussed with the
Board?s staff how merger oversight is conducted and how such oversight has
changed over time. In addition, we discussed with the Board?s staff the
activities conducted as part of formal oversight- that is, activities
included in an annual general oversight proceeding- as well as informal
oversight activities (such as monitoring of railroad performance data)
associated with mergers.

To address how the Board acts to mitigate potential merger- related harm to
competition, we reviewed documents contained in its merger dockets,
including merger approval and oversight decisions and progress reports filed
by merged railroads. We discussed with Board officials how oversight of
conditions is conducted and the factors considered by the Board in
determining if conditions imposed have been effective in mitigating
potential harm to competition. We also discussed oversight issues with
various trade associations representing shipper and railroad interests as
well as with officials from Class I railroads. (The organizations we
contacted are listed at the end of this app.) The shipper trade associations
represented major commodities shipped by rail. Finally, to identify how
merger oversight might change in the future, we reviewed the Board?s notice
of proposed rulemaking on major rail consolidations published in October
2000 and the final regulations issued in June 2001. We discussed with the
Board how the final merger rules differed from the proposed rules.

To address how the UP/ SP merger affected rail rates in selected geographic
areas, we obtained data from the Board?s Carload Waybill Sample for the
years 1994 through 1999. The Carload Waybill Sample is a Appendix II: Scope
and Methodology

Appendix II: Scope and Methodology Page 34 GAO- 01- 689 Freight Railroad
Regulation

sample of railroad waybills (in general, documents prepared from bills of
lading authorizing railroads to move shipments and collect freight charges)
submitted by railroads annually. We used these data to obtain information on
rail rates charged by different railroads for specific commodities in
specific markets subject to potential reduction in competition in the UP/ SP
merger. We focused on this merger because it was identified by the Board as
having significant competition- related issues, especially in the number of
shippers potentially going from service by two railroads to service by only
one railroad (called 2- to- 1 shippers).

Using documents submitted by the Union Pacific Railroad, as well as
discussions with officials from both the Union Pacific Railroad and the
Burlington Northern and Santa Fe Railway, we identified those locations and
corridors containing the majority of potential 2- to- 1 shippers. Using
economic areas defined by the Department of Commerce?s Bureau of Economic
Analysis, our analysis focused on those economic areas containing the
majority of these potential 2- to- 1 shippers. We used the

Carload Waybill Sample instead of more specific data on rates for individual
shippers because of the lack of sufficient premerger rate data from SP?s
operations. Although it is possible to get rates for 2- to- 1 shippers from
the Carload Waybill Sample, the sample is not designed for use in analyzing
rates for specific shippers. However, the sample can be used to analyze rail
rates within and between geographic areas. For these reasons, we used
economic areas containing a majority of potential 2- to- 1 points in
conjunction with the Carload Waybill Sample to conduct our analysis. The
rate data obtained from the Carload Waybill Sample were then used in an
econometric model that analyzed the effects of the UP/ SP merger on changes
to rail rates for various commodity shipments to and from the economic areas
with the majority of potential 2- to- 1 shippers. A detailed description and
discussion of this model can be found in appendix III.

Some railroad movements contained in the Carload Waybill Sample are governed
by contracts between shippers and railroads. To avoid disclosure of
confidential business information, the Board provides for railroads to mask
the revenues associated with these movements prior to making this
information available to the public. We obtained a version of the Carload
Waybill Sample that did not mask revenues associated with railroad movements
made under contract. Therefore, the rate analysis presented in this report
presents a truer picture of rail rates than analyses that are based solely
on publicly available information. There are also limitations associated
with data from the Carload Waybill Sample. For example, according to Board
officials, revenues derived from this sample are not adjusted for such
things as year- end discounts and refunds that

Appendix II: Scope and Methodology Page 35 GAO- 01- 689 Freight Railroad
Regulation

may be provided by railroads to shippers that exceed certain volume
requirements. However, both Board and railroad officials agreed that, given
the lack of sufficient premerger SP data, the Carload Waybill Sample was the
best data source available for conducting our analysis.

We performed our work from July 2000 through June 2001 in accordance with
generally accepted government auditing standards.

Department of Transportation Federal Railroad Administration Surface
Transportation Board

American Chemistry Council Edison Electric Institute National Grain and Feed
Association National Industrial Transportation League National Mining
Association Society of the Plastics Industry

American Short Line and Regional Railroad Association Association of
American Railroads

Burlington Northern and Santa Fe Railway Co. CSX Transportation, Inc.
Norfolk Southern Corporation Union Pacific Railroad Co.

Covington and Burling Gollatz, Griffin & Ewing, P. C. LeBoeuf, Lamb, Greene,
MacRae, LLP Thompson, Hine, and Flory Organizations

Contacted Federal Agencies

Shipper Associations Railroad Associations Railroads

Law Firms Representing Railroads or Shipper Associations

Appendix III: Description and Discussion of Econometric Model Used to
Conduct Rail Rate Analysis

Page 36 GAO- 01- 689 Freight Railroad Regulation

This appendix describes and discusses our analysis of the effects of the
1996 UP/ SP merger on rail rates in selected geographic areas where the
merger had the potential for harm to competition because 2- to- 1 shippers
could have lost one of the two railroad carriers upon which they had relied.
In particular, we discuss (1) the econometric model we developed to analyze
separately the effects of the merger and of other factors on rail rates, (2)
the construction of the data used for the analysis, and (3) our analysis,
including a comparison of overall changes in rates, based on mean-
difference analysis, with the results of the econometric model.

We developed an econometric model to examine both the specific impact of the
1996 UP/ SP merger and the impact of other factors on rates in selected
geographic areas where competition could have been potentially reduced. In
developing the model, we focused on the trackage rights granted to BNSF by
the Board, and applied existing empirical literature on how rail rates are
determined.

The UP/ SP merger covered areas where the services provided by UP overlapped
those provided by SP. As a result, some rail shippers could have been
reduced from being directly served by both SP and UP to being directly
served by UP only. In order to preserve competition in those potential 2-
to- 1 situations and for those shippers exclusively served by UP or SP who
benefited from having another independent railroad nearby, the Board granted
trackage rights to BNSF in order to replace the competition that would be
lost when SP was absorbed by UP. 1

As done in previous studies, we use an econometric model to identify the
factors affecting rail rates following the UP/ SP merger- rail rates being
the dependent variable used in the model. 2

1 For the most part, the BNSF trackage rights condition imposed by the Board
does not provide for direct access by BNSF to 3- to- 2 shippers (shippers
who could obtain service from UP, SP, and one other rail carrier before the
merger, but would have only two carriers available to them after the UP/ SP
merger).

2 Some of the previous models examined the effects of the Staggers Rail Act
of 1980 on rates. See, for example, Mark Burton, ?Railroad Deregulation,
Carrier Behavior, and Shipper Response: A Disaggregated Analysis,? Journal
of Regulatory Economics, Vol. 5 (1993), pp. 417- 434; and Wilson (1994).
Appendix III: Description and Discussion of

Econometric Model Used to Conduct Rail Rate Analysis

An Econometric Model of the Impact of the UP/ SP Merger on Rail Rates

The Board Granted Trackage Rights to Preserve Competition in the UP/ SP
Merger

Factors Affecting Rail Rates in the UP/ SP Merger

Appendix III: Description and Discussion of Econometric Model Used to
Conduct Rail Rate Analysis

Page 37 GAO- 01- 689 Freight Railroad Regulation Rail Rates: We measured
rail rates- the freight rate charged by a railroad

to haul a commodity from an origin to a destination- by revenue per tonmile,
adjusted for inflation. 3 We used data from 1994 and 1995 for the premerger
period, and data from 1997 through 1999 for the postmerger period. We
excluded 1996 data, since the UP/ SP merger was approved in August 1996. We
also excluded shipments with rail transportation charges less than $20,000
(in 1996 dollars) in order to focus on the major movements. The level of
each observation was shipments at the 7- digit Standard Transportation
Commodity Code- a classification system used to group similar types of
commodities such as grains- between an origin and a destination. The factors
that explained the rail rates were generally those related to market
structure and regulatory conditions, as well as cost and demand factors.

Market Structure and Regulatory Conditions: We included the variable MERGER
to capture the effect of the merger on rates. The extent of rail competition
is expected to affect rail rates. 4 We used a variable that would reflect
the difference in rates charged to shippers with competitive options- SP and
UP before the merger, and BNSF and UP afterwards- and shippers served solely
by one railroad both before and after the

3 Although revenue per ton- mile is not the actual rail rate that is paid to
transport freight, it is the most widely accepted measure of rates in the
rail industry. (See, for example, Rail Rates Continue Multi- Year Decline,
Surface Transportation Board, Office of Economics, Environmental Analysis,
and Administration (2000), and the academic studies cited in this report.)

4 A full assessment of the effects of mergers on the extent of competition
and rates should consider their impact on origin- to- destination rivalry
among existing railroads (intramodal competition) as well as competition
from nonrail carriers (intermodal competition). We could not include
intermodal competition in the analysis because of data limitations. Other
sources of competition in the rail industry include geographic or source
competition- that is, the ability of customers to use an alternative carrier
to obtain similar products from another source (origin or destination)- and
product competition- the ability of customers to use an alternative carrier
to obtain a substitute product. Since 1999, the Board has excluded
geographic and product competition from consideration in determining whether
a rail rate can be subjected to regulatory review for practical reasons, and
the Board has recently reaffirmed that policy. (See Surface Transportation
Board, Market Dominance Determinations-- Product and Geographic Competition,
STB Ex Parte No. 627, Apr. 6, 2001. Note: This decision has been appealed to
the United States Court of Appeals for the District of Columbia Circuit.)
Board officials said they continue to assess the need to preserve product
and geographic competition in the context of rail merger applications.

Appendix III: Description and Discussion of Econometric Model Used to
Conduct Rail Rate Analysis

Page 38 GAO- 01- 689 Freight Railroad Regulation

merger to capture the influence of this fact on rates. The variable is
RAILROAD- BNSF. 5

Cost and Demand Factors: These factors are generally captured by the
shipment and shipper characteristics of the traffic. 6 As in previous
studies, we use the following variables to measure the influence of cost and
demand factors: variable cost per ton- mile (COST), the weight of shipments
(TON), the length of haul (DISTANCE), the annual tonnage shipped between an
origin- destination pair (DENSITY), and OWNERSHIP of railcars. 7

In addition to the explanatory factors mentioned above, we included the
following factors: First, we introduced a variable for contract rates
(CONTRACT) to account for possible differences between contract rates and
noncontract rates. Second, we included a variable to account for the
possible effects of the service crisis that arose after the merger and
lasted through 1998 (CRISIS). Third, following previous studies, we included
the squared terms for the variables TON (TON_ SQ) and DISTANCE (DISTANCE_
SQ), to account for possible nonlinear relationships between these variables
and rates. 8 We also included dummy variables for the major commodity groups
(COMMODITY) where appropriate.

We selected geographic markets that had high concentrations of potential 2-
to- 1 shippers because of the possibility for harm to competition in those
areas. Using the Carload Waybill Sample, we performed several dataprocessing
tasks that included matching similar sets of traffic before and after the
merger, and selecting the primary commodities that were shipped, based on
carloads, for analysis.

5 The procedure for constructing the traffic for only 2- to- 1 shippers and
other shippers is detailed in the data section below. 6 See, for example,
Curtis Grimm and Clifford Winston (2000), and Stephen Schmidt,

?Market Structure and Market Outcomes in Deregulated Rail Freight Markets,?

International Journal of Industrial Organization, Vol. 19 (2001), pp. 99-
131. 7 Based on previous studies, COST is expected to be positively related
to rates, while TON, DISTANCE, and DENSITY are negatively related to rates.
The impact of OWNERSHIP is inconclusive. (See, for example, Wilson (1994),
and Grimm and Winston (2000)). However, Board officials told us that the use
of railroad- owned cars invariably is reflected in higher rates than if
shipper- owned cars had been used.

8 See, for example, Grimm and Winston (2000). Data Sources,

Selection, and Processing

Appendix III: Description and Discussion of Econometric Model Used to
Conduct Rail Rate Analysis

Page 39 GAO- 01- 689 Freight Railroad Regulation

All the data used for the study were constructed from the Carload Waybill
Sample, which is a sample of railroad waybills (in general, documents
prepared from bills of lading that authorize railroads to move shipments and
collect freight charges) that are submitted annually by the railroads. 9
However, there are limitations in using the Carload Waybill Sample for rate
analysis. Among these limitations is that no specific information is
provided about the identity of the shippers. This makes it difficult to
identify potential 2- to- 1 traffic by shipper name. Also, data for rates
for shipments moved under contract between railroads and shippers (called
contract rates), which are masked or disguised in the Carload Waybill
Sample, may be incomplete. 10

We selected the Reno, Nevada, and Salt Lake City, Utah, business economic
areas, which are in the Central Corridor and which had high concentrations
of potential 2- to- 1 shippers. 11 Both SP and UP served these two areas
prior to the merger; BNSF service was not available in the area at that
time. 12 Also, according to BNSF officials, the Central Corridor was
relatively less affected by the service crisis that emerged after the UP/ SP
merger. In addition, UP fully integrated its computer and information
systems with SP in the Central Corridor much earlier than in the other
regions, making rate and other data there more reliable. However, there

9 An alternative data source would be a survey of shippers, as was done by
Grimm and Winston (2000). However, this approach has the potential problem
of shipper bias- that is, shippers could provide biased responses or the
self- selected nature of those choosing to respond could result in a sample
that is not representative of the group. In addition, there is the potential
problem of allocating revenues to multiple origin- destination pairs of
traffic. Furthermore, this approach typically yields data for a single year,
which means we could have been limited to data for 1999 or 2000 only.
Finally, we could have obtained data on all car movements directly from the
railroads. Unfortunately, data of sufficient quality on individual potential
2- to- 1 shippers were not available for the premerger period.

10 About 70 percent of the tonnage in 1997 moved under contract. Contracts
generally offer reduced rates in return for guaranteed volumes. However,
even unmasked Carload Waybill Sample revenues may not reflect the actual
rates paid. This is because negotiated contract volumes may not always
materialize and subsequent upward adjustments are thus made to the rates,
or, more typically, rebates are offered late in the year or early in the
next year when minimum volume commitments have been met. However, according
to BNSF and UP officials, the margin of error in using Carload Waybill
Sample revenues as a surrogate for contract rates is likely to be very small
and within a few percentage points of actual rates.

11 See figure 1. 12 BNSF had provided some services in the Reno economic
area, in Lassen County, California, hauling primarily lumber/ wood in the
premerger period. However, this did not affect our analysis, since this
traffic was not utilized in our analysis. Data Sources

Selection of Geographic Markets

Appendix III: Description and Discussion of Econometric Model Used to
Conduct Rail Rate Analysis

Page 40 GAO- 01- 689 Freight Railroad Regulation

are limitations in using the Central Corridor to illustrate the possible
effects of the UP/ SP merger on rates. According to the Board, BNSF
generally had problems ramping- up its trackage- rights service in the
Central Corridor. Also, the Reno and Salt Lake City areas are not typical
rail hubs, because the traffic to and from these areas is not high volume,
compared with other areas, such as the Houston- Gulf Coast area. Despite
these limitations, the two selected areas provide an opportunity to
illustrate the impact of the UP/ SP merger on rates in predominantly
potential 2- to- 1 situations.

We performed several tasks to organize the Carload Waybill Sample for our
analysis. 13 We identified traffic by origin and destination, and at the
7digit Standard Transportation Commodity Code level separately for periods
before the merger and periods after the merger.

We then matched similar sets of railroad traffic existing before and after
the merger. The matching involved shipments that we could determine, on a
commodity and origin- and- destination basis, that were made in both
periods. To help identify traffic associated with BNSF?s trackage rights, we
also identified the railroad carrier( s) associated with the shipments that
we matched for both periods. There were two Class I railroads serving the
two geographic areas before the merger (SP and UP). After the UP/ SP merger,
all the traffic belonging to SP and UP came under the merged UP?s sole
control, except for potential 2- to- 1 shippers and shippers that could take
advantage of such provisions as build- in/ build- out and new facilities
conditions. As a result of the trackage rights imposed by the Board as part
of the merger conditions, BNSF obtained access to the potential 2- to- 1
traffic, regardless of whether the traffic had been carried by SP or UP
prior to the merger. Our matching process was intended to identify this
potential 2- to- 1 traffic. The following matching was done in the following
sequence:

1. SP premerger traffic was matched to BNSF postmerger traffic- this is BNSF
trackage rights over SP (BNSF- SP).

2. UP premerger traffic was matched to BNSF postmerger traffic that is still
unmatched- this is BNSF trackage rights over UP (BNSF- UP).

13 Using data provided by the Board, we converted the economic areas used in
1994 and 1995 to the new economic areas issued by the Bureau of Economic
Analysis in 1996. Data Processing

Appendix III: Description and Discussion of Econometric Model Used to
Conduct Rail Rate Analysis

Page 41 GAO- 01- 689 Freight Railroad Regulation

3. SP premerger traffic that was still unmatched was matched to UP
postmerger traffic- this is UP traffic over SP (UP- SP).

4. UP premerger traffic that was still unmatched was matched to UP
postmerger traffic that is still unmatched- this is UP traffic over UP (UP-
UP).

The BNSF- SP and BNSF- UP traffic (henceforth BNSF) consists of only
potential 2- to- 1 traffic that was served by SP or UP before the merger but
served by BNSF in the postmerger period. The UP- SP and UP- UP traffic
(henceforth UP) includes potential 2- to- 1 traffic as well as non- 2- to- 1
traffic. 14 However, according to UP officials, the latter traffic
substantially comprises shippers that are served solely by one railroad
because they could be served in the premerger period only by UP or SP, but
not both, and in the postmerger period, only by UP. The two broad types of
shippers identified reflect different levels of rail competition. The
potential 2- to- 1 traffic (served by BNSF) is considered more competitive
than the traffic served solely by UP because direct rail competition was
preserved or maintained for the potential 2- to- 1 shippers, while the
traffic solely- served by UP had only indirect competition, which was
preserved through buildin/ build- out and new facilities conditions.

Finally, because our study focuses on potential 2- to- 1 shippers, we
included only the commodity groups for which BNSF had presence. Although
BNSF officials told us they had not aggressively exercised their trackage
rights for coal shipments in the Salt Lake City area, we included these
shipments because coal is a major commodity shipped to and from the Salt
Lake City area. Summary statistics of the commodities shipped to and from
the Salt Lake City and Reno economic areas are provided in tables 4 and 5.
The commodities include coal, chemicals, primary metals, farm products (such
as corn and wheat), petroleum/ coal, food, nonmetallic minerals, lumber/
wood, and stone/ clay/ glass/ concrete. Each of these commodities accounted
for at least 10 percent of the traffic to or from an area. 15 The share of
BNSF?s potential 2- to- 1 shippers to all shippers was mostly between 10 and
25 percent. (See table 4.) Also, the rail rates

14 Board officials indicated that both BNSF and UP used a mix of former SP
and former UP routes to reach potential 2- to- 1 shippers. This activity is
not expected to affect the matching since the matching is not based on the
ownership of the railroad routes that were used.

15 The matched traffic, compared with all the traffic in an area, excluded
traffic no longer transported by rail, new traffic, and traffic from new
facilities.

Appendix III: Description and Discussion of Econometric Model Used to
Conduct Rail Rate Analysis

Page 42 GAO- 01- 689 Freight Railroad Regulation

and the direct costs for the total traffic were very similar to the rates
for the matched traffic. (See table 5.) 16

16 The rail rates we calculated had some extreme values. This could be due
to contract rates for which the guaranteed volume may not have been realized
or to local shipments (shipments over very short distances). After examining
the distributions of the calculated rates, we deleted the top 1 percent and
the bottom 1 percent of the data. The Board had mentioned that outliers
represent about one- quarter to one- half of 1 percent of all Carload
Waybill Sample records.

Appendix III: Description and Discussion of Econometric Model Used to
Conduct Rail Rate Analysis

Page 43 GAO- 01- 689 Freight Railroad Regulation

Table 4: Shipments to and From the Salt Lake City and Reno Areas, in
Carloads, 1994- 99

Area/ Commodity a Commodity share of area

traffic, by carloads (in percent) b

BNSF 2- to- 1 shippers to all shippers, by carload

(in percent) From the Salt Lake City area c

Coal 69 d Chemicals 44 15 Primary metals 30 3

To the Salt Lake City area e

Coal 52 c Farm products 13 22 Chemicals 12 25 Petroleum/ Coal 11 25 Food 10
20

From the Reno area f

Nonmetallic minerals 79 10 Lumber/ Wood 21 21

To the Reno area f

Coal 50 10 Chemicals 32 15 Stone/ Clay/ Glass/ Concrete 10 10 a The values
reported are averages for 1994 through 1999, excluding 1996.

b The share of commodity in area traffic, excluding coal. c For shipments
from the Salt Lake City area, the following commodities were excluded: food,
petroleum/ coal, farm products, metallic ores, miscellaneous manufacturing,
and mail/ express/ other contract traffic. d Not applicable.

e For shipments to the Salt Lake City area, the following commodities were
excluded: stone/ clay/ glass/ concrete, primary metals, lumber/ wood, pulp/
paper, mail/ other contract traffic, furniture/ fixtures, and electrical
machinery. Waste/ scrap was excluded from the analysis because BNSF?s share
of this traffic was only 1 percent. f For shipments to the Reno area, the
following commodities were excluded: food, petroleum/ coal,

farm products, metallic ores, miscellaneous manufacturing, and mail/ other
contract traffic. BNSF did not ship coal to the area in 1999, the most
recent year of our study.

Source: GAO?s analysis of STB?s Carload Waybill Sample.

Appendix III: Description and Discussion of Econometric Model Used to
Conduct Rail Rate Analysis

Page 44 GAO- 01- 689 Freight Railroad Regulation

Table 5: Rates and Costs of Shipments to and From the Salt Lake City and
Reno Areas, 1994- 99 Total traffic (in cents) Matched traffic (in cents) a
Premerger Postmerger Premerger Postmerger Area/ Commodity b Rate c Cost c
Rate c Cost c Rate c Cost c Rate c Cost c From the Salt Lake City area

Coal 2.8 1. 3 2.5 1. 1 2.6 1. 2 2.8 1. 3 Chemicals 2. 4 2.4 2. 4 2.3 2. 3
2.4 2. 5 2.5 Primary metals 2.2 2. 0 2.1 2. 3 2.2 2. 1 2.1 2. 3

To the Salt Lake City area

Coal 4.3 1. 7 4.1 1. 6 3.6 1. 5 4.1 1. 6 Farm products 2. 6 2.5 2. 6 2.3 2.
5 2.2 2. 6 2.3 Chemicals 3. 8 2.7 3. 6 2.6 3. 7 2.7 4. 1 2.8 Petroleum/ Coal
4.2 3. 1 3.1 2. 4 3.8 2. 8 3.7 2. 7 Food 3.3 3. 2 3.0 2. 7 3.0 3. 0 3.0 2. 8

From the Reno area

Nonmetallic minerals 2. 5 2.6 2. 1 2.2 2. 8 2.6 2. 1 2.2 Lumber/ Wood 3. 6
2.3 2. 9 2.3 3. 1 2.0 2. 8 2.2

To the Reno area

Coal 1.7 0. 9 1.8 0. 9 1.8 1. 0 1.8 0. 9 Chemicals 3. 2 2.4 3. 1 2.2 3. 1
2.4 3. 0 2.1 Stone/ Clay/ Glass/ Concrete 2.5 2. 2 2.9 3. 3 2.5 2. 1 3.1 2.
7

Note: See also notes to table 4. a Matched traffic is similar sets of
traffic before and after the merger, excluding intermodal.

b The values reported are averages for 1994 through 1999, excluding 1996. c
Rate is weighted average of revenue per ton- mile; weights are based on the
Carload Waybill Sample

sampling rates. Cost is weighted average of variable costs per ton- mile;
weights are based on the

Carload Waybill Sample sampling rates. Source: GAO?s analysis of STB?s
Carload Waybill Sample.

The econometric model that we developed was estimated using an appropriate
estimation technique. We also discuss the results of our study in terms of
the effects on rail rates attributable to the merger and the effects of
other factors.

We used a reduced- form rate model of shipping a commodity between an origin
and a destination because such a model is useful for analyzing the
Methodology for

Estimation and Its Results

Estimation of the Model

Appendix III: Description and Discussion of Econometric Model Used to
Conduct Rail Rate Analysis

Page 45 GAO- 01- 689 Freight Railroad Regulation

impact of a regulatory policy, such as a merger, on rates. 17 The service
crisis of 1997 and 1998 could potentially make the estimation results less
reliable because the rates may not be at the market- clearing level.
However, we included a CRISIS variable to account for this possible
structural shift. 18 The reduced- form model we used was as follows:

The term ?ln? is a natural logarithm, and ?i? is representative of a
commodity group. 19 The ?s are parameters to be estimated, and is the
random- error term. A complete list of the variables used to estimate the
regression model is presented in table 6. We could not directly incorporate
certain factors into the model primarily because of data limitations. 20

17 Furthermore, reduced- form estimates are preferred when determining the
net effect of the merger on rates after all other endogenous variables have
been adjusted. Also, a reduced- form specification may provide more robust
and reliable estimates. (See, for example, Stephen Schmidt (2001).)

18 A similar approach was used in the Grimm and Winston study (see Grimm and
Winston 2000). 19 All the variables are expressed in natural logarithms,
except for the dummy variables. This is done, following previous studies, to
obtain a better fit for the estimates and to help deal with potential
problems of heteroscedasticity. (See, for example, Schmidt (2001), and
Wilson (1994).)

20 For instance, railroad carriers may react to changes in their economic
environment by changing rates and/ or the quality of services, meaning that
service quality information would be useful for explaining rates. Also,
because shippers might incur additional investment costs (for such
activities as extending track and adding storage capacity), rates could be
affected by such behavior. (See, for example, Traffic World, July 17, 2000,
p. 11).

Appendix III: Description and Discussion of Econometric Model Used to
Conduct Rail Rate Analysis

Page 46 GAO- 01- 689 Freight Railroad Regulation

Table 6: List of Variables Used in Our Econometric Analysis of Rail Rates
Variable Definition

RATE Real revenue per ton- mile (adjusted by the Gross Domestic Product
Deflator), in 1996 dollars MERGER A dummy variable, equals 1 if postmerger
period (1997- 99), 0

otherwise COST Variable costs per ton- mile (in 1996 dollars), as defined by
the Board?s

Uniform Rail Costing System (URCS) TON Billed weight of shipments (in tons)
TON_ SQ Squared value of TON DISTANCE Length of haul (in miles) DISTANCE_ SQ
Squared value of DISTANCE DENSITY Total tonnage shipped from origin to
destination (in tons) OWNERSHIP A dummy variable, equals 1 if railcars are
railroad- owned, 0 otherwise RAILROADBNSF A dummy variable, equals 1 if
traffic is potential 2- to- 1 and postmerger

railroad carrier is BNSF, 0 otherwise CRISIS A dummy variable, equals 1 if
crisis period (1997- 98), 0 otherwise CONTRACT A dummy variable, equals 1 if
rate is based on contract, 0 otherwise BNSF* MERGER Interaction term for
RAILROAD- BNSF and MERGER- equals 1 if

traffic is potential 2- to- 1, postmerger railroad carrier is BNSF, and the
period is postmerger, 0 otherwise COMMODITYCOAL a A commodity dummy
variable, equals 1 if major commodity group is

coal, 0 otherwise a Similar dummy variables were created for the other major
commodity groups.

We estimated the regression model using the SAS SURVEYREG procedure, since
the data are from stratified samples. This procedure is appropriate for
dealing with a stratified sample because it adjusts both the coefficients
and the standard errors of the estimates to account for the sampling design.
21 The econometric model was run for different samples- shipments of the
primary commodities to or from an economic area, and for subsamples of
individual commodities and shippers.

We tried different specifications of our basic model to check the robustness
of our key model results. We found that the results were not

21 We preferred the SURVEYREG procedure to the Weighted Least Squares
method, which adjusts only the coefficients and assumes a simple random
sampling design. However, the econometric results were generally consistent
with either method, probably because most of the observations were
concentrated in only one or two sampling strata. With the SURVEYREG
procedure, we could not check for possible problems of heteroscedasticity or
serial correlation because these tests are not available for this procedure.
We believe that since the data are from a stratified sample, it is more
appropriate to use SURVEYREG, which is consistent with the data design.

Appendix III: Description and Discussion of Econometric Model Used to
Conduct Rail Rate Analysis

Page 47 GAO- 01- 689 Freight Railroad Regulation

highly sensitive to model specification. While we used a reduced- form
specification, it is still possible that some of the explanatory variables
on the right- hand side of the equation may be endogenous. Since there are
no available instruments in a reduced- form model, we could not perform the
usual test. 22 Rather, we checked the robustness of our results by excluding
possible endogenous variables. 23 In particular, when DENSITY was excluded
from the model, our findings regarding the effects of mergers on rates and
the effects of the other factors on rates were essentially unchanged. It is
also likely that COST is related to the variables TON, DISTANCE, and
OWNERSHIP, which could produce unreliable results. In other specifications
of the model, we eliminated the COST variable, but our key findings were
robust to such specifications.

Summaries of the effects of the merger on rates, based on the econometric
results, are presented in table 7. 24 The rates for shipments to and from
the Reno and Salt Lake City areas generally would have declined for all the
shippers as a result of the merger, especially in the Salt Lake City area.
Although the effects of the merger on rates depend on both the potential
cost savings from the merger and the exercise of any enhanced market power
by the railroads, the UP/ SP merger is generally expected to lower rates in
those areas where the Board imposed trackage rights.

22 This is a Hausman test. See J. Hausman, ?Specification Tests in
Econometrics,?

Econometrica, Vol. 46 (1978), pp. 1251- 1271. 23 See, for example, Wilson
(1994) for a similar approach.

24 The complete econometric results are presented in tables 8- 11. The
overall econometric results are very significant, based on the significance
levels of the prob- F values. The econometric results are presented for only
the commodities that are discussed in table 1 in the text above. Econometric
Results of

Effects of Merger and Other Factors on Rates

Appendix III: Description and Discussion of Econometric Model Used to
Conduct Rail Rate Analysis

Page 48 GAO- 01- 689 Freight Railroad Regulation

Table 7: Changes in Rail Rates, for Shipments to and From the Salt Lake City
and Reno Areas

In percent Shipments Commodity

Shipper/ post merger railroad a

Rate changes

due to merger b

Rate changes due to other

factors c Overall

changes in rates d

From the Salt Lake City area

Coal All shippers/ UP

-10 20 10

Chemicals All shippers/ BNSF & UP

-6 16 10

Chemicals BNSF v UP -16 e e To the Salt Lake City area

Coal All shippers/ UP

-8 23 15

Farm products All shippers/

BNSF & UP

-5 6 1 Farm products BNSF v UP 0.3 e e From the Reno area Nonmetallic

minerals All shippers/ BNSF & UP

4 -26- 22

Nonmetallic minerals BNSF v UP 6 e e To the Reno area Chemicals All
shippers/

BNSF & UP

21 -27 -6

Chemicals BNSF v UP 4 e e a BNSF & UP is for all the shippers- potential 2-
to- 1 shippers served by BNSF and shippers served solely by UP. On the other
hand, BNSF v UP is the changes in BNSF rates less changes in UP rates. The
results for the rate changes are based on the estimated coefficients for
BNSF* MERGER from the regression equations in tables 8- 11. Since the
dependent variable is in logs, the percentage change in rates between the
BNSF shippers and the UP shippers as a result of the merger is obtained as:
[exp ( ) - 1] x 100, where ?exp? is an exponential, and is the estimated
coefficient for BNSF* MERGER. b The results for the rate changes due to the
merger are based on econometric results, using the

estimated coefficients for MERGER from the regression equations in tables 8-
11. See also note ?a?

above. c The effects of changes in rates due to other factors are calculated
as the overall changes in rates

less the merger effects on rates. d The overall changes in rates, which are
unweighted, are based on a mean- difference test that subtracts the
premerger rates from the postmerger rates. The BNSF v UP values are not
available for the mean- difference tests because the samples were unbalanced
(unequal sample sizes) for the postmerger and premerger periods. For the
overall changes in rates and the rate changes due to the merger, values in
bold/ italics are significant at the 5- percent level or better. e Not
applicable.

Source: GAO?s analysis of STB?s Carload Waybill Sample.

Appendix III: Description and Discussion of Econometric Model Used to
Conduct Rail Rate Analysis

Page 49 GAO- 01- 689 Freight Railroad Regulation

We also compared the effects of the merger on rates charged to potential 2-
to- 1 shippers served by BNSF to rates charged to shippers served solely by
UP in the same general locations. In particular, the results show that the
rates charged to the potential 2- to- 1 shippers served by BNSF were lower
than the rates charged to the shippers served solely by UP for shipments of
chemicals from the Salt Lake City area. The rate differentials for the Reno
area were positive, but none was statistically significant. The result that
rates for the potential 2- to- 1 shippers served by BNSF were generally
lower than rates charged to shippers served solely by UP is consistent with
demand- based differential pricing, which reflects the differing
transportation alternatives available to shippers.

We found that the effects of other factors on rail rates during the period
are generally consistent with what has been found in previous studies. (See
results in tables 8 through 11 for all commodities.) We used the econometric
results for all the commodities because most of these effects are not
commodity- specific and can be better captured across commodities. The
impact of COST on rates was positive and significant for traffic in each of
the selected areas, meaning that rates were lower (or higher) as costs
decreased (or increased). 25 TON had mixed results, meaning that larger
shipment volumes sometimes resulted in higher or lower rates. DISTANCE
generally decreased rates. 26 DENSITY, which captures the volume of traffic
on the route used for a particular shipment, unambiguously decreased rates.
This effect is consistent with decreasing costs in railroad operations,
since increased shipment levels over a rail route spread fixed costs over
larger volumes and reduce rates. 27 OWNERSHIP had mixed results. CONTRACT
rates were generally lower. Finally, the impact of CRISIS on rates was
generally inconclusive. This is not unexpected, since most shipments are
under contract and the crisis affected primarily the services that were
provided rather than the rates.

25 The estimated coefficients were also reasonable, between zero and one,
except for shipments from the Reno area. 26 The squared terms for TON (TON_
SQ) and DISTANCE (DISTANCE_ SQ) were generally significant, implying that
these variables had nonlinear relationships with rates. 27 Other previous
studies have found evidence of decreasing costs. See, for example, Burton
(1993).

Appendix III: Description and Discussion of Econometric Model Used to
Conduct Rail Rate Analysis

Page 50 GAO- 01- 689 Freight Railroad Regulation

To compare the changes in rates due to the merger that we obtained from the
econometric analysis to the overall changes in rates, we separated the
overall changes in rates into changes due to the merger and changes due to
other factors, such as costs and volume of shipments. The overall changes in
rates were estimated using a difference in means analysis that compares the
rates in the postmerger period with rates in the premerger period. 28

We found that the overall changes in rates could be in the opposite
direction from the rate changes due to the merger. For instance, for coal
shipments from the Salt Lake City area, the overall changes in rates were
about 10 percent higher, while the rate changes due to the merger alone
would have been about 10 percent lower. On the other hand, for shipments of
chemicals to the Reno area, the overall changes in rates were about 6
percent lower, while the rate changes due to the merger alone would have
been about 21 percent higher. These illustrations indicate that a complete
analysis of merger- related rate changes could benefit from the application
of an analytical approach that identifies and determines the separate
effects of the various factors, including those associated with a merger,
affecting rail rates.

28 The mean- difference test uses the TTEST procedure in SAS. Direction of
Rate Changes

Due to a Merger Could Differ From Overall Changes in Rates

Appendix III: Description and Discussion of Econometric Model Used to
Conduct Rail Rate Analysis

Page 51 GAO- 01- 689 Freight Railroad Regulation

Table 8: Econometric Results of Rail Rates, for Shipments From the Salt Lake
City Area

Variable All commodities a Coal

UP Chemicals

BNSF & UP Chemicals

BNSF v UP

MERGER -0.0658 [0.0001]

-0.1057 [0.0001]

-0.0640 [0.0018]

-0.0312 [0.1728] CRISIS 0. 0170

[0.1164] 0.0049

[0.6897] 0.0535

[0.0004] 0.0524

[0.0005] CONTRACT -0. 0773

[0.0001] -0.1438

[0.0001] 0.0379

[0.1076] 0.0495

[0.0382] COST 0.3286

[0.0001] 0.1145

[0.0112] 0.1501

[0.0001] 0.1141

[0.0012] TON 0. 0697

[0.2325] 4.2914

[0.0001] -0.8073

[0.0001] -0.8560

[0.0001] TON_ SQ -0.0080

[0.0774] -0.2786

[0.0001] 0.0800

[0.0001] 0.0836

[0.0001] DISTANCE -0. 7108

[0.0001] 0.5624

[0.0001] -1.9305

[0.0001] -1.9862

[0.0001] DISTANCE_ SQ 0. 0190

[0.0010] -0.0855

[0.0001] 0.1060

[0.0001] 0.1094

[0.0001] DENSITY -0.0268

[0.0001] 0.0270

[0.0001] -0.0268

[0.0001] -0.0259

[0.0001] OWNERSHIP -0. 0089

[0.5234] -0.0809

[0.0001] 0.0612

[0.0010] 0.0690

[0.0002] RAILROAD- BNSF -0. 0134

[0.4503] b 0.0572

[0.0098] 0.1585

[0.0001] BNSF* MERGER b b b

-0.1703 [0.0001] Chemicals -0.3939

[0.0001] b bb Primary metals -0.1221

[0.0005] b bb INTERCEPT 1.7921

[0.0001] -19.6003

[0.0001] 7.0212

[0.0001] 7.2153

[0.0001] Prob- F 0. 0001 0. 0001 0. 0001 0. 0001 R 2 0.64 0.84 0.71 0.71
Sample size 6359 2227 2323 2323

Note: P- values are in brackets. a Includes all chemicals, primary metals,
and coal shipments. Coal was the excluded commodity

dummy. b Not applicable.

Source: GAO?s analysis of STB?s Carload Waybill Sample.

Appendix III: Description and Discussion of Econometric Model Used to
Conduct Rail Rate Analysis

Page 52 GAO- 01- 689 Freight Railroad Regulation

Table 9: Econometric Results of Rail Rates, for Shipments to the Salt Lake
City Area

Variable All commodities a Coal

UP Farm products

BNSF & UP Farm products

BNSF v UP

MERGER 0.0628 [0.0023]

-0.0838 [0.0001]

-0.0462 [0.0360]

-0.0468 [0.0203] CRISIS -0.0681

[0.0006] -0.0084

[0.5394] -0.0280

[0.2466] -0.0281

[0.2517] CONTRACT -0. 0812

[0.0001] -0.1454

[0.0001] 0.1727

[0.0001] 0.1729

[0.0001] COST 0.5191

[0.0001] 0.0979

[0.0077] 0.5313

[0.0001] 0.5315

[0.0001] TON -0.4302

[0.0001] 0.7310

[0.0037] -0.6961

[0.0002] -0.6960

[0.0002] TON_ SQ 0.0387

[0.0001] -0.0581

[0.0013] 0.0614

[0.0001] 0.0614

[0.0001] DISTANCE -0. 6732

[0.0001] 0.4639

[0.0001] 2.5095

[0.0001] 2.5124

[0.0001] DISTANCE_ SQ 0. 0263

[0.0024] -0.0733

[0.0001] -0.2309

[0.0001] -0.2312

[0.0001] DENSITY -0.0442

[0.0001] 0.1095

[0.0001] -0.0392

[0.0001] -0.0393

[0.0001] OWNERSHIP -0. 0430

[0.0602] -0.2059

[0.0001] -0.0621

[0.0012] -0.0622

[0.0017] RAILROAD- BNSF -0. 1216

[0.0001] b -0.1367

[0.0001] -0.1379

[0.0001] BNSF* MERGER b b b

0.0025 [0.9534] Farm products 0. 3243

[0.0018] b b b Chemicals 0. 5001

[0.0001] b b b Petroleum/ Coal 0.3969

[0.0002] b b b Food 0.3514

[0.0011] b b b INTERCEPT 2.9016

[0.0001] -6.6043

[0.0001] -5.6704

[0.0001] -5.6781

[0.0001] Prob- F 0. 0001 0. 0001 0. 0001 0. 0001 R 2 0.63 0.88 0.84 0.84
Sample size 3079 954 492 492

Note: P- values are in brackets. a Includes farm, chemicals, petroleum/
coal, food, and coal shipments. Coal was the excluded commodity dummy. b Not
applicable.

Source: GAO?s analysis of STB?s Carload Waybill Sample.

Appendix III: Description and Discussion of Econometric Model Used to
Conduct Rail Rate Analysis

Page 53 GAO- 01- 689 Freight Railroad Regulation

Table 10: Econometric Results of Rail Rates, for Shipments From the Reno
Area Variable All commodities a

Nonmetallic minerals BNSF & UP

Nonmetallic minerals BNSF v UP

MERGER -0.0230 [0.4164]

0.0361 [0.3093]

-0.1209 [0.0043] CRISIS -0.0235

[0.2494] -0.0035

[0.8912] 0.0324

[0.2817] CONTRACT -0. 0162

[0.5981] -0.0212

[0.5720] -0.1392

[0.0072] COST 1.1828

[0.0001] 1.2656

[0.0001] b TON 0. 6393

[0.0001] 0.7497

[0.0001] -0.4051

[0.0106] TON_ SQ -0.0525

[0.0001] -0.0599

[0.0001] 0.0303

[0.0566] DISTANCE -1. 1022

[0.0527] -1.8058

[0.0185] -4.0021

[0.0001] DISTANCE_ SQ 0. 0749

[0.0685] 0.1296

[0.0204] 0.2693

[0.0001] DENSITY -0.0436

[0.0001] -0.0467

[0.0001] -0.0598

[0.0001] OWNERSHIP -0. 1080

[0.0001] -0.1423

[0.0001] 0.0146

[0.7428] RAILROAD- BNSF 0. 0424

[0.2572] 0.0428

[0.4405] -0.1076

[0.2325] BNSF* MERGER c c

0.0627 [0.5488] Lumber/ Wood 0. 2609

[0.0001] c c INTERCEPT 3.4396

[0.0777] 5.6220

[0.0292] 12.6568

[0.0001] Prob- F 0. 0001 0. 0001 0. 0001 R 2 0.60 0.55 0.24 Sample size 933
708 708

Note: P- values are in brackets. a Includes nonmetallic minerals and lumber/
wood. Nonmetallic minerals was the excluded commodity

dummy. b The COST variable was deleted from the regression because it was
very significantly correlated with BNSF* MERGER. See also footnote 25. c Not
applicable.

Source: GAO?s analysis of STB?s Carload Waybill Sample.

Appendix III: Description and Discussion of Econometric Model Used to
Conduct Rail Rate Analysis

Page 54 GAO- 01- 689 Freight Railroad Regulation

Table 11: Econometric Results of Rail Rates, for Shipments to the Reno Area
Variable All commodities a

Chemicals BNSF & UP

Chemicals BNSF v UP

MERGER 0.1103 [0.0001]

0.1914 [0.0001]

0.1902 [0.0001] CRISIS -0.0027

[0.8921] -0.0231

[0.3418] -0.0235

[0.3361] CONTRACT 0. 0633

[0.0146] 0.1291

[0.0001] 0.1278

[0.0001] COST 0.1691

[0.0094] 0.2891

[0.0047] 0.2996

[0.0047] TON -1.0161

[0.0001] -0.9636

[0.0003] -0.9458

[0.0004] TON_ SQ 0.0954

[0.0001] 0.0994

[0.0001] -0.0978

[0.0002] DISTANCE -1. 3817

[0.0001] -1.9342

[0.0001] -1.9242

[0.0001] DISTANCE_ SQ 0. 0836

[0.0004] 0.1236

[0.0001] 0.1231

[0.0001] DENSITY -0.0845

[0.0001] -0.1070

[0.0001] -0.1072

[0.0001] OWNERSHIP 0. 0199

[0.5562] -0.0888

[0.2386] -0.0889

[0.2353] RAILROAD- BNSF -0. 2071

[0.0001] -0.1760

[0.0001] -0.2039

[0.0005] BNSF* MERGER b b

0.0352 [0.5691] Chemicals 0. 0600

[0.1023] b b INTERCEPT 5.9309

[0.0001] 8.1804

[0.0001] 8.1334

[0.0001] Prob- F 0. 0001 0. 0001 0. 0001 R 2 0.42 0.41 0.41 Sample size 1116
866 866

Note: P- values are in brackets. a Includes chemicals and stone/ clay/
concrete/ glass. Stone/ clay/ concrete/ glass shipments was the

excluded commodity dummy. b Not applicable.

Source: GAO?s analysis of STB?s Carload Waybill Sample.

(348244)

The first copy of each GAO report is free. Additional copies of reports are
$2 each. A check or money order should be made out to the Superintendent of
Documents. VISA and MasterCard credit cards are also accepted.

Orders for 100 or more copies to be mailed to a single address are
discounted 25 percent.

Orders by mail:

U. S. General Accounting Office P. O. Box 37050 Washington, DC 20013

Orders by visiting:

Room 1100 700 4 th St., NW (corner of 4 th and G Sts. NW) Washington, DC
20013

Orders by phone:

(202) 512- 6000 fax: (202) 512- 6061 TDD (202) 512- 2537

Each day, GAO issues a list of newly available reports and testimony. To
receive facsimile copies of the daily list or any list from the past 30
days, please call (202) 512- 6000 using a touchtone phone. A recorded menu
will provide information on how to obtain these lists.

Orders by Internet

For information on how to access GAO reports on the Internet, send an email
message with ?info? in the body to:

Info@ www. gao. gov or visit GAO?s World Wide Web home page at: http:// www.
gao. gov

Contact one:

 Web site: http:// www. gao. gov/ fraudnet/ fraudnet. htm

 E- mail: fraudnet@ gao. gov

 1- 800- 424- 5454 (automated answering system) Ordering Information

To Report Fraud, Waste, and Abuse in Federal Programs
*** End of document. ***