DEPARTMENT OF TRANSPORTATION

OFFICE OF THE SECRETARY

14 CFR Part 255

(Dockets Nos. OST-97-2881, OST-97-3014, OST-98-4775, and OST-99-5888)

RIN 2105-AC65

Computer Reservations System (CRS) Regulations

AGENCY: Office of the Secretary, Department of Transportation

ACTION: Final rule.

SUMMARY: The Department is amending its rules governing airline computer reservations systems (“CRSs” or “systems”) to eliminate most of the rules now and to terminate additional rules as of July 31, 2004. The Department is readopting the rules prohibiting display bias and adopting rules that prohibit systems from imposing certain types of contract clauses on participating airlines that would unreasonably restrict their ability to choose how to distribute their services. These rules will be effective during a six-month transition period.

DATES: This rule is effective on January 31, 2004.

FOR FURTHER INFORMATION CONTACT: Thomas Ray, Office of the General Counsel, 400 Seventh St. S.W., Washington, D.C. 20590, (202) 366-4731.

SUPPLEMENTARY INFORMATION:

ELECTRONIC ACCESS:

You can view and download this document by going to the website of the Department’s Docket Management System (http://dms.dot.gov).  On that page, click on “simple search.” On the next page, type in the last four digits of the docket number shown on the first page of this document, 2881. Then click on “search.” An electronic copy of this document also may be downloaded from http://regulations.gov and from the Government Printing Office’s Electronic Bulletin Board Service at (202) 512-1661. Internet users may reach the Office of the Federal Register’s home page at: http://www.archives.gov/federal_register/index.html and the Government Printing Office’s database at: http://www.gpoaccess.gov/fr/index.html.  

Table of Contents

A. Summary of Final Rule

B. Background

    1. The CRS Business

    2. The Travel Agency Distribution System and the Business Relationships between Travel Agencies and the Systems

    3. Regulatory Background

C. Development of the Record in this Rulemaking

D. Procedural Issues

E. The Need for Limited CRS Regulation

    1. Introduction

    2. Final Rule

    3. Market Definition

    4. The Systems’ Market Power over Airlines

    5. The Potential for System Conduct Undermining Airline Competition

    6. System Practices that Preserve Market Power

    7. The Systems’ Ability to Engage in Display Bias

F. The Department’s Statutory Authority To Regulate CRS Practices
  
    1. Whether Non-Airline Systems Are Ticket Agents Subject to Section 411

    2. Antitrust Principles Relevant to System Practices

    3. First Amendment and International Law Issues

G. The Specific Rule Proposals

    1. The Scope of the Rules

    2. Exclusion of Internet-Based Systems

    3. Definitions

    4. Rules Barring Display Bias

    5. Contract Clauses Restricting Airline Choices on System Usage

    6. Equal Functionality

    7. The Mandatory Participation Rule

    8. Booking Fees

    9. Booking Fee Bills

    10. Other Participating Carrier Contract Rules

    11. Marketing and Booking Data

    12. Third-Party Hardware and Software

    13. Travel Agency Contracts

    14. The Tying of Commissions and Marketing Benefits with a Subscriber’s Choice of a System

    15. Regulation of the Internet’s Use in Airline Distribution

    16. Tying of Internet Participation

    17. International Issues

    18. Retaliation against Discrimination by Foreign Airlines and Systems

19. Sunset Date for the Rules

20. Effective Date of the Rules

21. Divestiture

REGULATORY PROCESS MATTERS

Regulatory Assessment and Unfunded Mandates Reform Act Assessment

Regulatory Flexibility Analysis

Paperwork Reduction Act

Federalism Implications

Taking of Private Property

Civil Justice Reform

Protection of Children

Consultation and Coordination with Tribal Governments

Energy Effects

Environment

Glossary
 
ASTA American Society of Travel Agents
 
Board The Civil Aeronautics Board
 
Booking fees Fees paid by airlines and other travel suppliers when a travel agent makes or changes a booking in a system
 
CRS Computer reservations system
 
Mandatory participation rule The rule requiring each airline that has a significant ownership interest in a system to participate in competing systems at as high a level of functionality as it does in its own system, if the terms are commercially reasonable
 
Network airlines The airlines that operate hub-and-spoke route systems, especially the five largest airlines (American, Continental, Delta, Northwest, and United)
 
Non-airline system A system that is neither owned nor controlled by any airline or airline affiliate
 
OMB Office of Management and Budget
 
Participate To make the services of an airline or other travel supplier available for sale through a system under a contract with that system
 
Parity clauses Clauses in participating airline contracts that require a participating airline to buy at least as high a level of service from the system as it does from any other system
 
Productivity pricing Pricing formula used in subscriber contracts that enables the travel agency to obtain lower CRS fees from a system if the travel agency meets minimum booking quotas established by the contract
 
Section 411 49 U.S.C. 41712, recodifying section 411 of the Federal Aviation Act
 
Subscriber A travel agency that obtains CRS services under a contract with the system
 
System Computer reservations system

Webfares Discount fares offered by an airline through its own website and often through selected distribution channels
 
A. Summary of Final Rule
 
In this proceeding we have reexamined whether our existing rules on computer reservations systems (“CRSs” or “systems”), 14 CFR Part 255, remain necessary and, if so, whether we should readopt them, with or without modifications. If we do not readopt the rules, they will expire on their sunset date, currently January 31, 2004. Our notice of proposed rulemaking asked for comment on these issues and proposed that most of the rules should be readopted. 67 FR 69366 (November 15, 2002). After reviewing the comments and the on-going changes in the airline distribution and CRS businesses reflected in those comments, we have concluded that most of the rules should be allowed to sunset on January 31, 2004. We believe, however, that we should adopt the rules prohibiting display bias and certain rules barring unreasonably restrictive requirements in the contracts between systems and their airline customers for a six-month transition period to provide an opportunity for the affected parties to prepare for complete deregulation of computer reservation systems. We intend to monitor developments in the industry during this period and beyond. We, of course, retain our authority to pursue future regulatory or enforcement actions against airlines or systems that engage in anti-competitive practices.
 
The systems’ operations have been subject to rules for twenty years. Although the systems now are commonly called global distribution systems, or GDSs, we will continue to refer to them here as CRSs. The Civil Aeronautics Board (“the Board”), the agency that had been responsible for the economic regulation of the airline industry, originally adopted those rules in 1984. 49 FR 32540 (August 15, 1984), aff’d, United Air Lines v. CAB, 766 F.2d 1107 (7th Cir. 1985). After reexamining whether those rules were necessary and effective, we readopted them with some changes in 1992. 14 CFR Part 255, adopted at 57 FR 43780 (September 22, 1992).
 
When these rulemakings were held, one or more airlines or airline affiliates owned or controlled each system, airlines depended heavily on travel agencies for distribution, travel agents used a system to research airline service options and to make bookings, and each travel agency predominantly relied on one system to perform these tasks. Systems therefore did not need to compete for airline participants (a “participant” is an airline that agrees to make its services saleable through a system). The airlines that controlled the systems had the incentive and ability to use them to prejudice the competitive position of non-owner airlines and to provide information on airline services through the systems to travel agents that gave an undue preference to the services operated by the owner airlines. Competitive market forces did not discipline the prices and terms for services offered by systems to participating airlines.
 
Our goal in CRS rulemakings has been to prevent practices that were likely to harm consumers by substantially reducing airline competition or by giving travel agents and their customers inaccurate or misleading information on airline services. The rules block system practices that would cause consumers and their travel agents to receive misleading information and would distort airline competition. We adopted most of the rules under our authority to prevent unfair methods of competition in the sale of airline transportation, an authority that empowers us to prohibit practices that violate the antitrust laws or antitrust principles, but, in adopting the rules prohibiting display bias, we additionally relied on our authority to prevent unfair and deceptive practices in the marketing of air transportation.
 
We should adopt rules regulating industry practices only if they are reasonably necessary to prevent anti-competitive or deceptive practices that are likely to occur, and would cause significant consumer harm if they did occur, and that market forces are unlikely to remedy. Any rule must be effective and enforceable. Rules intended to address a serious competitive concern may have unintended consequences that may reduce efficiency and consumer choice. As we explained in our notice of proposed rulemaking, we will not adopt rules that address all potential problems, for such detailed regulations would necessarily impose significant burdens on the systems and interfere with legitimate business practices. 67 FR 69389. Our approach for determining whether rules are necessary is essentially the same as that recommended by the Justice Department. The Department of Justice states that regulation is appropriate “only when (1) market participants have substantial and durable market power that will likely harm consumers directly, or will be exercised in ways that exclude or limit competition in contiguous markets, and (2) the regulation will likely be effective and enforceable without imposing significant costs of its own.” Justice Department Reply Comments at 18.
 
Our rules included a sunset date, currently January 31, 2004, to ensure that we would review whether the rules remained necessary in light of on-going developments in the CRS and airline distribution businesses. 57 FR 43829-43830; 68 FR 15350 (March 31, 2003). This proceeding carries out that reassessment. The major changes that have occurred since our last major rulemaking underscore the need for such a reassessment. All of the U.S. airlines that had controlled a system have divested their CRS ownership interests. As a result, none of the four systems now operating in the United States is owned or controlled by any U.S. airline or airline affiliate. Furthermore, airlines are selling an increasingly large share of their tickets through their Internet websites and a diminishing share through travel agencies using a system. The airlines’ control over access to their webfares, the discounted fares originally offered only through individual airline websites, has enabled them to obtain lower fees from two of the systems. And travel agencies are increasingly demanding -- and winning -- contracts from the systems that give them more freedom to use alternative booking channels and to switch systems periodically.
 
Our examination of these developments has persuaded us that we should allow most of the existing rules to sunset upon their expiration. The major predicate for the rules has always been the systems’ control by airlines. The U.S. airlines’ divestiture of their ownership interests has eliminated that basis for the rules. While each system still has market power over most airlines, that power is diminishing. Moreover, the record does not show a likelihood that the systems would use that power to distort airline competition except potentially through the sale of bias.
 
On the other hand, we have determined that we should readopt, for a six-month transition period, the rules prohibiting display bias and rules prohibiting certain types of contract clauses in the systems’ contracts with airlines. We are readopting the rules against display bias because we believe that, were the rules terminated immediately, systems might well be expected to bias their displays in ways that could mislead travel agents and their customers and prejudice airline competition. For that reason, we believe it is important to provide a measure of notice to the industry prior to the rules’ termination and a concomitant opportunity to prepare for the absence of regulation.
 
Similarly, we are adopting for the same short transition period two rules governing the contracts between the systems and airlines: rules prohibiting parity clauses (a parity clause would require an airline to participate in that system at at least as high a level as it participates in any other system) and clauses requiring airlines to provide access to all webfares as a condition to any participation in a system. However, an airline is free to agree to such clauses. We believe that, were these prohibitions terminated immediately, the systems would have sufficient market power to impose contract terms on airlines that would unreasonably restrict the airlines’ ability to bargain for better terms for participation. The transition period during which these prohibitions will be maintained will furnish the industry with reasonable notice of the forthcoming change with an opportunity to prepare for it. Our final decision is consistent with the recommendations made by the Justice Department.
 
The two rules on contract clauses and the rule prohibiting display bias therefore will sunset on July 31, 2004. We will actively monitor developments during the transition period and beyond and take appropriate investigative, enforcement, or regulatory action if we see evidence that systems or airlines are engaging in anti-competitive conduct in connection with airline distribution through the systems and other channels.
 
We will not readopt the other rules now in force, and we reaffirm our tentative decision not to adopt rules governing the use of the Internet in airline distribution. The rules that we are not readopting will automatically expire on January 31, 2004, their sunset date.
 
The elimination of most of the rules will ensure that government regulation does not interfere with market forces and innovation in the CRS and airline distribution businesses. The record indicates that market forces are beginning to discipline business practices in the CRS industry. Ending the broad regulation of CRS practices will enable each system and each airline to bargain over the terms on which CRS services should be provided, just as airlines obtain products and services from other suppliers under agreements negotiated by the parties. The systems will have the same ability to bargain with their other customers, the travel agencies. The resulting terms under which airlines and travel agencies obtain system services will likely reflect the interests of both sides better than if we maintained broad regulations restricting the parties’ behavior. While we cannot predict exactly what will happen, we believe that ending most of the rules will produce the best results for consumers over time. We base this judgment on our experience with airline deregulation. Airline deregulation has provided lower fares and better service for consumers, in part by enabling new firms to enter the airline business. Several of the new airlines have followed new business plans that have provided great benefits for airline travelers. Airline deregulation has produced these benefits even though the deregulated airline industry has not operated in the manner expected by industry experts on the eve of deregulation. The deregulation of the CRS business should also benefit consumers, even though we cannot forecast how it will play out.
 
Our final rule also conforms to the limits imposed by Congress on our authority to regulate the airline and airline distribution businesses. Congress has given us the authority to prevent practices that violate the antitrust laws or antitrust principles and practices that are deceptive, but no comprehensive oversight authority over airline distribution. We are adopting only those rules that are necessary to prevent practices in the CRS business that would constitute unfair or deceptive practices, or unfair methods of competition.
 
We are aware that some participants in the airline distribution and CRS businesses may seek to engage in anti-competitive conduct that would reduce competition in the airline and airline distribution businesses and thereby harm consumers. A system, for example, might develop vertical ties with an airline that would cause the system to operate in a way that could prejudice airline competition. Some systems may seek to pursue practices that would reduce competition in the CRS business and preserve their market power over airlines. Even without specific regulations, any such practices could be unfair methods of competition and thus unlawful. We retain the authority to bring enforcement cases against firms that violate the statutory prohibition against unfair methods of competition, and we will take appropriate action if we have evidence of unlawful conduct. As Congress stated when it deregulated the airline industry, S. Rep. No. 95-631, 95th Cong., 2d Sess. (1978) at 52:
 
Vigorous enforcement of antitrust policy is the discipline by which competition can remain free and markets can operate in a healthy fashion. Predatory behavior, market concentration, and other economic evils should be avoided and remedied by the Board when they exist.

See also H. R. Rep. No. 98-793, 98th Cong., 2d Sess. (1984) at 5: “Although the airline industry has been deregulated, this does not mean that there are no limits to competitive practices. As is the case with all industry, carriers must not engage in practices which would destroy the framework under which fair competition operates.”
 
We will also actively monitor the systems’ reactions to the substantial deregulation of their business, and we, of course, retain the power to reexamine our decision that all rules should terminate by July 31, 2004, if the systems’ conduct or other developments makes such a reexamination necessary.    
 
Our final rule departs from the proposals made by our notice of proposed rulemaking. Our notice proposed to eliminate two of the major rules, the rule barring discriminatory booking fees and the rule requiring airlines with a significant ownership interest in one system to participate in competing systems at an equivalent level if the terms for doing so were commercially reasonable, but to readopt most of the remaining rules. Our review of the rulemaking record up to that point suggested that rules were still necessary, notwithstanding the changes in the systems’ ownership and the growing role of the Internet. 67 FR 69375-69384. The notice, however, did request comment on whether we should sunset more of the rules now, and we predicted that the rules would become unnecessary in a few years. 67 FR 69368, 69376, 69388-69389.
 
The comments and the continuing developments in airline distribution and the CRS business have convinced us that most of the rules are no longer appropriate. In particular, one of the systems, Worldspan, was owned by three U.S. airlines when we issued our notice of proposed rulemaking but was sold several months ago to two private venture capital firms. The airline distribution business has continued to evolve since we issued the notice. Airlines are selling more tickets through the Internet. Moreover, as we predicted, the airlines’ control over access to their webfares has led some of the systems to offer airlines discounted booking fees in return for the ability to sell those fares. 67 FR 69381; Galileo Supp. Comments at 5-8. And the comments have shown that the systems’ contracts with travel agencies are significantly less restrictive than they were even a few years ago. See, e.g., ASTA Comments at 14-16.
 
That our final rule does not duplicate our proposal is consistent with the purpose of rulemaking procedures. The notice of proposed rulemaking was designed to obtain comments from interested persons on our tentative findings and our economic and policy analysis and to enable them to submit current information. We held a public hearing to give interested persons an additional opportunity to present their views and respond to our questions. The comments submitted in this proceeding, together with the on-going developments in the airline distribution and CRS businesses, have persuaded us that our proposals should not be made final. Those proposals, while reasonable in light of industry conditions two or three years ago, to a large extent no longer reflect current conditions.
 
We will begin our explanation of our final rule by updating our description of the CRS and travel agency businesses, and we address several procedural issues. We then discuss our conclusions on the need for adopting some CRS rules, including our findings that the systems continue to have market power over airlines, and discuss the question of our legal authority to readopt the rules and to apply them to systems that are not owned by airlines. We thereafter present the rationale for our decisions on each of the rule proposals.
 
Our notice of proposed rulemaking included a request for comments on whether we should clarify our policy on fare disclosures as regards the disclosure of travel agency service fees. We have decided to address that question in a separate rule.
 
We will refer to commenters by their common names (for example, “Alaska,” not “Alaska Airlines”). References to comments and reply comments are to the pleadings filed in response to the notice of proposed rulemaking, not the pleadings filed in response to the advance notices of proposed rulemaking, which were discussed in the notice of proposed rulemaking. We will refer to the statutory provision that is the principal basis for our adoption of CRS rules, 49 U.S.C. 41712, by its traditional name, section 411, as we did in the notice of proposed rulemaking. The glossary at the beginning of this document gives the meaning of the abbreviations and technical terms used in this rule.
 
B. Background
 
Our notice of proposed rulemaking described in some detail the nature of the airline distribution and CRS businesses, including the travel agency business. 67 FR 69369-69375. Here we will update our factual description on the basis of the information provided by the comments and set forth the factual findings underlying our final decision.
 
1. The CRS Business
 
Airlines use several distribution methods: direct sales through their reservations agents, sales through “brick-and-mortar” travel agencies, sales through individual airline websites, and sales through on-line travel agencies. In the past, the “brick-and-mortar” travel agency channel produced the great majority of airline revenues for almost all airlines. In 1999 travel agencies sold almost three-quarters of airline tickets, almost all through off-line travel agencies. 67 FR 69369, citing Bear, Stearns & Co., “Point, Click, Trip: An Introduction to the On-Line Travel Agency” (April 2000) at 17. Since then the Internet has become an increasingly important distribution channel. Galileo states that the different channels’ shares of total airline tickets in 2002 were as follows, Galileo Comments, Guerin-Calvert, Jernigan, & Hurdle Declaration at 24:
 
off-line sales by airlines 17 percent
on-line sales by airlines 10 percent
off-line sales by travel agencies 58 percent
on-line sales by travel agencies 15 percent.
 
Until recently the great majority of all travel agency airline ticket sales, whether off-line or on-line, have been made through one of the systems.
 
Four systems operate in the United States: Sabre, Galileo, Worldspan, and Amadeus. Each of them was originally developed by one or more U.S. airlines (Amadeus entered the U.S. market by acquiring a U.S. system). Two of the systems -- Sabre and Galileo -- were no longer owned or controlled by any U.S. airlines when we issued the notice of proposed rulemaking. At that time, three U.S. airlines -- American, Delta, and Northwest -- owned Worldspan. Amadeus was then owned by three European airlines -- Air France, Iberia, and Lufthansa -- as well as by public shareholders (and has the same ownership today). Worldspan’s airline owners sold that system to two private venture capital firms on June 30, 2003, after the issuance of our notice of proposed rulemaking. As part of that sale, the airline owners agreed to certain parity clauses and marketing commitments. Galileo Comments, Guerin-Calvert, Jernigan, & Hurdle Declaration at 20; Amadeus Comments at 32-33; August 1, 2003, Letter from Charles Simpson, Jr.; Sabre Supp. Reply at 4. Amadeus is now the only system with any airline ownership.
 
The systems that have no airline owners have marketing ties with their former owners. United markets Galileo, American markets Sabre, and Delta and Northwest have agreed to market Worldspan for several years following the closing of the system’s sale. Amadeus Comments at 25, n. 24; Galileo Supp. Comments at 1-4. Southwest also markets Sabre, although Southwest never had an ownership interest in the system.
 
Each system’s share of CRS airline bookings in the United States in 2002 was as follows, Galileo Comments, Guerin-Calvert, Jernigan, & Hurdle Declaration at 18:
 
Sabre 44.7 percent
Worldspan 26.5 percent
Galileo 19.7 percent
Amadeus 9.2 percent
 
Since 1999 the shares of Galileo and Amadeus have been declining, while Worldspan’s share has risen sharply, from 19.3 percent to 26.5 percent. The growth in Worldspan’s share in large part reflects its status as the booking engine for two of the three largest on-line travel agencies, Expedia and Orbitz.
 
Each system provides information and booking capabilities on the airlines that “participate” in it, that is, agree to make their services saleable through the system. The system obtains its availability information from the airlines’ internal reservations systems, and it makes bookings in those systems, which are used by the airlines’ own reservations agents and other staff members. The systems also provide information and booking capabilities for rental cars, hotels, and other travel services. Airline transportation is the most important travel service sold through the systems, and airlines obtain a larger share of their revenues from CRS bookings (sales made through the systems) than do other travel suppliers. 67 FR 69370.
 
An airline (or other travel supplier) participating in a system must pay fees for each booking transaction (the fees paid by participating airlines are usually called “booking fees”). Airlines can participate at different levels. At higher levels the information provided travel agencies will be more timely and so more reliable, and travel agents can carry out tasks like reserving specific seats for their customers. An airline that chooses a higher level of participation must pay a higher booking fee. 67 FR 69370. Booking fees paid by airlines provide well over half of the systems’ total revenues. 67 FR 69380.
 
The average airline booking fee per segment is $4.25. Because the average ticket includes more than one segment, the average booking fee per ticket is $11. United Reply Comments at 28; “Upheaval in Travel Distribution: Impact on Consumers and Travel Agents,” National Commission to Ensure Consumer Information and Choice in the Airline Industry” (November 13, 2002), at 16. United alleges that its average booking fee per segment equals 3.3 percent of its average revenue per segment. United Reply Comments at 29. Sabre has stated that the effective booking fee per segment for its highest level of participation was $4.38 in 2002, about 2.4 percent of the average airline ticket price for tickets sold through Sabre. Sabre charges $2.12 per segment for airlines participating at its low level, Basic Booking Service. Sabre Comments at 14; Sabre Comments, Wilson Declaration at 6.
 
Sabre and Galileo have created programs that give participating airlines lower booking fees in return for a commitment to provide the system with all of their webfares. Under Sabre’s Direct Connect Availability program (“DCA program”), an airline can obtain a 10 percent reduction in its booking fees, guaranteed for three years, in exchange for a commitment to provide the system with all of the airline’s published fares, including its webfares. American, Continental, Delta, Northwest, United, US Airways, and a number of smaller airlines now participate in this program. Sabre Supplemental Reply at 1.
 
Galileo first established its Momentum program, which gave airlines a 20 percent reduction in booking fees for tickets sold through participating travel agencies, if the airlines agreed to give Galileo access to all of their publicly-available fares. Travel agencies could participate in the program if they agreed to a reduction in their incentive payments from Galileo. United and US Airways were the first airlines that joined this program. One of the travel agencies that joined the program was Rosenbluth International, the fourth largest U.S. corporate travel agency. Due to complaints from America West and other airlines, Galileo dropped the initial requirement that any airline participating in the Momentum program must upgrade its participation level to the highest level. More recently Galileo introduced Preferred Fares Select, which will enable airlines to obtain lower booking fees on all of their bookings if they agree to make all of their publicly-available fares saleable through Galileo. Galileo Comments, Guerin-Calvert, Jernigan, & Hurdle Declaration at 52-56; Galileo Reply Comments at 33-34; Galileo Supplemental Comments at 5-8; Sabre Comments, Fahy Declaration at 10-11.
 
The record does not indicate that Amadeus or Worldspan has introduced comparable programs.
 
Travel agencies often obtain CRS services at no cost or receive bonus payments in exchange for agreeing to use a system. ASTA states that in 2002 fewer than half of all travel agencies paid monthly fees for system services and that 60 percent of them received a signing bonus of some kind from the system that they were using. ASTA Comments at 17. The systems pay on average $1 to $1.50 per booking to travel agencies for using a system. Sabre Comments at 7.
 
As we stated in the notice of proposed rulemaking, travel agents have depended heavily on the systems to determine what airline services are available and to make bookings. There we cited statistics showing that travel agencies in 1999 sold almost three-quarters of all airline tickets and made 93 percent of their domestic airline bookings and 81 percent of their international airline bookings through a system. 67 FR 69369-69370. The record shows that since then the share of airline revenues produced by travel agents using a system has been declining. The Justice Department states that the share of revenues produced by “brick-and-mortar” travel agencies for the five airlines that own Orbitz has fallen from 76 percent in May 2000 to 67 percent in March 2002, primarily due to the growth in Internet sales. Justice Department Reply Comments at 14-15.
 
In the past, almost all U.S. airlines participated in every system. Southwest, which has participated only in Sabre and at a low level, was the major exception. JetBlue, which began operations in 2000, also participates only in Sabre and at the same level as Southwest. Sabre Comments at 38. Airlines that can avoid participation in every system focus their marketing efforts instead on direct sales to consumers, made through either the airline’s website or its reservations agents. Airlines that have been participating in all of the systems, such as Alaska, have been shifting many of their bookings away from the travel agency channel, which required them to pay the systems’ booking fees. See, e.g., Alaska Comments at 5. The large network airlines nonetheless still obtain at least 60 percent of their revenues from bookings made by travel agents using a system, as discussed below. American, for example, states that over 70 percent of its bookings are made through the systems. American Reply Comments at 19. The share of total industry bookings made through the systems has been declining in part due to the growth of airlines like Southwest that do not depend on travel agencies for the major share of their revenues. American Reply Comments at 19.
 
The systems have played a major role in airline distribution because travel agents -- the airlines’ primary distribution channel -- have relied so much on the systems for investigating airline service options and booking tickets, because the systems are so efficient. They electronically provide comprehensive information and booking capabilities on airlines and other travel suppliers. Each system presents displays that integrate almost all services offered in a market. Each system shows the schedules and fares offered by airlines in each market that are available for sale through travel agents using that system and whether seats are available on specific flights at specific fares (some fares are often not available through the systems, notably corporate discount fares and webfares). The system thus allows the travel agent to compare the schedules and fares offered by different airlines and determine which would best meet a customer’s needs. The agent using a system can reserve a seat and issue a paper ticket or print an E-ticket.
 
On-line agencies also use systems -- Travelocity uses Sabre, while Expedia and Orbitz use Worldspan, for example. 67 FR 69370. Orbitz and Expedia have been developing direct connection technologies which enable bookings to be made directly with an airline’s internal reservations system, bypassing Worldspan. Sabre Comments, Fahy Declaration at 8-9.
 
Since the Board first adopted CRS rules, no firm has entered the CRS business. Until recently, entry into the CRS business would have been prohibitively costly and time-consuming. 67 FR 69381.  This may no longer be true. Sabre Comments, Fahy Declaration at 8. New direct-connection technologies can enable firms to provide airline information and booking services that replicate at least some of the services provided by the systems. Galileo Comments at 42, n. 38. Orbitz, which now operates as an on-line travel agency, plans to make its services available to travel agencies through software being developed by Aqua. Orbitz continues to rely on Worldspan for some functions involved in the search and booking process. 67 FR 69373, 69374. Another commenter in this proceeding, AgentWare, is also offering travel agencies fare and schedule information and links to booking sites. Galileo Comments at 66-67.
 
The development of sources of airline information and booking capabilities on the Internet has created additional resources that travel agents can use. Travel agents are increasingly checking the fares and services offered on websites because some airline discount fares have not been sold through the systems. Travel agents, however, continue to make most of their airline bookings through a system. Using alternative booking channels is less efficient for travel agents, as discussed below. Nevertheless, the development of alternative sources of information and booking capabilities on the Internet, and the airlines’ control over access to their webfares, have begun to make the systems responsive to market force discipline.
 
Corporate travel departments as well as travel agencies use the systems.  A corporate travel department can book travel for its company’s employees by accessing a system through the Internet or by Intranet (an internal corporate communications network based on Internet technology). 67 FR 69370.
 
Systems operate throughout the world. U.S. systems like Sabre and Worldspan market their services to travel agencies in foreign countries, and Amadeus is a major system in the Eastern Hemisphere. The systems had the following shares of worldwide CRS airline bookings in 2002, Galileo Comments, Guerin-Calvert, Jernigan, & Hurdle Declaration at 18:
 
Sabre 30.8 percent
Worldspan 15.1 percent
Galileo 26.4 percent
Amadeus 27.7 percent
 
The European Union, Canada, and other governments have regulations governing CRS operations. The United States has entered into a number of international air services agreements that require each party to ensure that the systems operating in its country and their owners do not subject airlines and systems from the other country to discriminatory treatment. 67 FR 69371-69372.
 
2. The Travel Agency Distribution System and the Business Relationships between
Travel Agencies and the Systems
 
The systems’ practices have affected airline competition because of the importance of travel agents in airline distribution. The travel agency system has provided airlines with an efficient means of distribution. Travel agencies have acted as agents for virtually all airlines and generally hold themselves out to the public as sources of impartial advice on airline services and other travel services. 67 FR 69371.
 
In 2001, there were 18,425 travel agencies. The travel agency business is dominated by the largest travel agencies. In 2001, the 117 travel agencies with revenues of more than $50 million (as measured by sales of air transportation) accounted for 57.2 percent of all travel agency sales. The 1,015 travel agencies with revenues of $5 million to $50 million accounted for another 20.1 percent of all travel agency sales. “Upheaval in Travel Distribution: Impact on Consumers and Travel Agents,” National Commission to Ensure Consumer Information and Choice in the Airline Industry” (November 13, 2002), at 113. See also Sabre Comments, Salop & Woodbury Declaration at Table 3 (Sabre’s top five subscribers produced 25.7 percent of its total bookings, excluding Travelocity, and the top 100 produced 49.6 percent of its total bookings, excluding Travelocity).
 
As noted above, in 2002 the airlines obtained 58 percent of their bookings from “brick-and-mortar” travel agencies and 15 percent from on-line travel agencies. Galileo Comments, Guerin-Calvert, Jernigan, & Hurdle Declaration at 24. The three largest on-line travel agencies had the following shares of all on-line travel agency bookings in 2002: Travelocity, 28.5 percent; Expedia, 28.7 percent; and Orbitz, 21.3 percent. Sabre Comments, Salop & Woodbury Declaration at Table 2. Travelocity is a Sabre subsidiary, while Orbitz is owned by the five largest U.S. airlines -- American, Continental, Delta, Northwest, and United. Travelocity has been using Sabre as its source of airline information and booking capabilities, while Expedia and Orbitz have been using Worldspan for these functions. Orbitz and Expedia have been developing direct connections with airlines that bypass Worldspan. Airlines that agree to be “charter associates” in Orbitz, which includes a commitment to make all publicly-available fares available for sale through Orbitz, receive a rebate on their booking fees. 67 FR 69374.
 
The larger airlines still obtain most of their revenues from bookings made by travel agents. However, despite the continuing importance of travel agencies in airline distribution, the travel agency business has faced severe business problems in recent years, due to developments such as the airlines’ elimination of base commissions (but not incentive commissions), the growing use of the Internet by many travelers, particularly leisure travelers, and the overall decline in airline traffic. See “Upheaval in Travel Distribution: Impact on Consumers and Travel Agents,” National Commission to Ensure Consumer Information and Choice in the Airline Industry” (November 13, 2002). From 1994 to 2002, the number of travel agencies fell by 31 percent and the number of travel agency locations by 21 percent. “Upheaval in Travel Distribution” at 21. The number of travel agencies declined by 12 percent in the year ended September 2002 and by another 7 percent through April 2003. ASTA Reply Comments at 15-16.
 
The nature of the travel agencies’ operations is important to this proceeding, because we must consider the impact of our decisions on the travel agencies’ business and because the rules have covered some features of the relationships between the systems and travel agencies. However, providing support for travel agencies that would offset other economic developments is not within our statutory authority and therefore not a proper goal of this proceeding. This proceeding must be, and is, limited to preventing system practices and related airline practices that would harm consumers by significantly reducing airline competition.
 
A critical factor in our decision-making is that travel agencies, unlike most airlines, can choose which system to use. Most travel agencies need to use only one system, and for most travel agencies no system has features and information that are indispensable, as discussed below. Because most travel agencies are free to decide to use one system rather than its competitors, the systems compete vigorously for travel agency customers. As noted above, systems usually pay travel agencies for choosing one system rather than another. See, e.g., 67 FR 69371; Sabre Comments at 7.
 
In past rulemaking proceedings, and in our notice of proposed rulemaking in this proceeding, we cited evidence that the systems’ contracts with travel agencies often contained provisions that unreasonably restricted the travel agencies’ ability to use more than one system or to use alternative electronic sources of airline information and booking channels. 67 FR 69405; 57 FR 43822. For example, each system formerly kept travel agencies from buying their own equipment and made them use equipment provided by the system for accessing its services. 57 FR 43796. The record further suggested that the systems’ contracts with travel agencies typically included “productivity pricing” programs that imposed financial penalties on an agency that began using another system or other booking channel for making a substantial number of bookings, or that gave the agency incentive payments if it made most of its bookings through that system. 67 FR 69408. These types of restrictive contract provisions concerned us because they tended to preserve the systems’ market power and denied airlines an opportunity to encourage travel agencies to use alternative electronic means for obtaining information on airline services and making bookings, such as direct links between a travel agency and an airline’s own internal reservations system. Our notice observed, however, that the systems were giving at least some travel agencies more flexible terms. 67 FR 69405.
 
The proposals made by our notice fairly reflected industry conditions when the comments on our advance notices of proposed rulemaking were filed. Large Agency Coalition Comments at 7. However, the comments submitted in response to our notice of proposed rulemaking show that travel agencies since then have been successfully demanding more flexible contracts and winning the ability to use alternative booking channels. ASTA’s October 2002 travel agency survey made the following finding (quoted in Sabre Comments at 151):
 
[CRS] vendors are introducing a new crop of more flexible contracts with less rigid productivity requirements and more pricing options. [C]ontract terms have gotten more favorable towards agencies with shorter overall length, lower required segments and a higher percentage of agencies receiving booking incentives.
 
See also Large Agency Coalition Comments at 7-14.
 
For example, subscriber contracts typically have a term that is substantially shorter than the maximum permitted by our rules. Our rules prohibit contracts with a term of more than five years and require a system to offer a three-year contract to any travel agency offered a five-year contract. 57 FR 43825. For some time after we adopted that rule, few travel agencies had contracts with a term of less than five years. 67 FR 69405. Now, however, many travel agencies have contracts that are no more than three years in length. The percentage of travel agencies with five-year contracts has declined from 85 percent in 1998 to 47 percent in 2002, while the percentage with three-year contracts has risen from 9 percent in 1998 to 39 percent in 2002. Almost 60 percent of Worldspan subscribers had five-year contracts in 2002, while only 35 percent of Sabre’s subscribers had such contracts. Sabre Comments at 17-18; Sabre Comments, Fahy Declaration at 14-15.
 
Travel agencies, moreover, have a substantial ability to switch systems when their existing contract expires. Half of the responding agencies in the ASTA survey stated they intended to obtain competitive bids at the end of their current contract, while another third stated that they might seek competitive bids and only one sixth stated they definitely intended to continue using the same system. Sabre Comments at 153. Nonetheless, switching systems can impose significant costs on travel agencies, at least for smaller travel agencies. Galileo Comments, Guerin-Calvert, Jernigan, & Hurdle Declaration at 81.
 
When we last readopted the rules, we added a provision giving travel agencies the right to use their own equipment to access a system and to use third-party software. Before then, each system typically demanded that its subscribers use equipment provided by the system and barred subscribers from accessing other systems and databases from that equipment. 57 FR 43796-43797. Travel agencies are increasingly using their own equipment. Only 70 percent of travel agencies leased equipment from a system in 2002, while 85 percent did so in 2000. ASTA Comments at 14. Sabre alleges that it seeks to exit the equipment-leasing business, that 73 percent of the equipment used by Sabre subscribers will be provided by third parties by the end of 2003, and that 62.5 percent of their equipment was being provided by third parties as of November 2002. Sabre Comments at 131. Amadeus states that only one fourth of its subscribers rely entirely on equipment provided by Amadeus. Amadeus Comments at 45. Subscribers to other systems are more likely to use equipment provided by the system. ASTA represents that systems do not resist subscriber efforts to use their own equipment instead of equipment provided by the system. ASTA Comments at 15. Sabre represents that it does not enforce the provisions in its older subscriber contracts that barred the travel agencies from using Sabre equipment to access other systems. Its subscribers are free to use multiple systems. Sabre Comments at 17, n. 17, and 71. Amadeus has made a similar representation. Amadeus Comments at 45.
 
Sabre further represents that the larger travel agencies often have complete flexibility in using the systems. Sixteen of Sabre’s 20 largest “brick-and-mortar” travel agency customers use multiple systems, and many use their own software to direct bookings to a specific system, often in order to maximize their incentive payments. Those 16 agencies produce 35 percent of Sabre’s total volume from “brick-and-mortar” travel agencies. Sabre Comments at 71. However, as discussed below in our market definition analysis, each location of a travel agency that subscribes to more than one system tends to predominantly rely on one system rather than make substantial use of every system whose services are being purchased by the parent firm.
 
Using alternate booking channels and sources of information has become easier for travel agents in recent years. New software, for example, allows travel agents to conduct fare searches simultaneously through a system and airline websites. Galileo Comments, Guerin-Calvert, Jernigan, & Hurdle Declaration at 29. The systems allegedly do not seek to block their subscribers from using alternative booking channels and sources of information, and they help develop tools enabling travel agents to use alternative sources of information. Galileo Comments at 64, 66-67. In 2002, 98 percent of all travel agencies had Internet access, according to an ASTA survey. Galileo Comments, Guerin-Calvert, Jernigan, & Hurdle Declaration at 81.
 
However, despite the widespread use of the Internet by travel agents, they make relatively few bookings through the Internet. According to the ASTA survey, travel agents made only 10 percent of their bookings through websites, and most of those bookings were for tours booked through tour operator sites. ASTA Comments at 12. The inefficiency of using the Internet for airline bookings is probably the most important deterrent to a greater use of the Internet. See “Upheaval in Travel Distribution: Impact on Consumers and Travel Agents,” National Commission to Ensure Consumer Information and Choice in the Airline Industry” (November 13, 2002), at 47-50.
 
Our notice further identified the systems’ pricing practices as a factor that seemingly kept travel agencies from using alternative systems and booking channels. Each system’s productivity pricing program generally gave travel agencies incentive payments if a subscriber used the system for a large majority of its bookings (or imposed financial penalties if it did not). We believed that such productivity pricing programs effectively deterred travel agencies from making significant use of alternative booking channels, such as airline websites. While we noted that the percentage of subscriber contracts with productivity pricing had been declining, most subscriber contracts still included productivity pricing. 67 FR 69408-69409.
 
The comments show that the systems’ productivity pricing provisions have become significantly less widespread and less restrictive in the last few years. In 1998 91 percent of subscriber contracts had productivity pricing, but only 56 percent did in 2002. The average number of bookings required before a travel agency can obtain incentive payments has fallen from 252 in 1998 to 194 in 2002. ASTA Comments at 15; Sabre Comments at 69, 162. The Large Agency Coalition represents that the systems’ incentive payment programs typically allow the travel agency to make up to thirty percent of its bookings outside the system before it suffers a financial penalty. Transcript at 231. Despite these changes, however, Sabre states that it has contracts with some small travel agencies that require the subscriber to use no system other than Sabre. Sabre argues that this requirement is reasonable under the circumstances because Sabre is providing support for the agency’s operations that would otherwise not be economical. Sabre Comments, Salop & Woodbury Declaration at 20. Nonetheless, despite the greater flexibility allowed travel agencies by recent productivity pricing arrangements, the record suggests that the systems’ current contractual arrangements may still deter travel agencies from making many bookings through the Internet. Orbitz Comments at 23, n. 10; ASTA Comments at 26, n. 44, and 34-35; Travel Management Alliance Comments.
 
The increasing flexibility of the contracts obtained by travel agencies is the result of changes in the travel agency business. ASTA states that travel agencies must have a greater ability to respond to changing technology, especially the growth of the Internet. The increasing uncertainties of the travel agency business itself, moreover, are likely to encourage many travel agencies to avoid long-term commitments if possible. ASTA Comments at 14. The large travel agencies created in recent years have more bargaining leverage with the systems.
 
In the past, we have endeavored to prevent system practices that would deter travel agencies from using multiple systems. We reasoned that the systems’ market power over airlines would be reduced if travel agencies had the ability to use alternative sources of airline information and booking capabilities. 57 FR 43797. Travel agency parties had encouraged those efforts. 67 FR 69391; 57 FR 43796.
 
The travel agency commenters in this proceeding assert, however, that rules designed to encourage travel agencies to use multiple systems will be futile. They contend that almost all travel agencies predominantly or entirely use one system. ASTA thus alleges, ASTA Comments at 3-4:
 
Use of a single CRS is a function of the market reality that multiple CRS’s are highly inefficient for travel agencies, who therefore do not employ them. No amount of realistically foreseeable inducement from competing CRS’s or regulatory pressure from DOT is going to overcome the inefficiencies for most agencies of operating multiple CRS’s in today’s environment.
 
See also Transcript at 213.
 
Using more than one system is generally inefficient for travel agencies, because, among other things, it requires training staff members to work with different systems and will cause the booking records of different customers to be in different places. Cardinal Travel Service Comments; Galileo Comments at 64-65; Galileo Comments, Guerin-Calvert, Jernigan, & Hurdle Declaration at 79; ASTA Comments at 23-24; Large Agency Coalition Comments at 20. At travel agencies that have multiple offices, each office tends to use one system even though the firm subscribes to several systems. Carlson Wagonlit Comments at 11.
 
Travel agencies, moreover, assertedly have no need to use multiple systems. Large Agency Coalition Comments at 20; Transcript at 236-237. While some travel agencies use multiple systems, they appear to make relatively little use of the secondary system. Galileo Comments, Guerin-Calvert, Jernigan, & Hurdle Declaration at 79-80. The Large Agency Coalition is a group of 22 large, corporate-oriented travel agencies, all but one of which was included in a recent listing of 84 top corporate travel agencies. Although many of the 22 use two or three systems, they typically do so because (i) the dominant airline in a city other than the agency’s headquarters city insisted that the agency use the system affiliated with the airline, (ii) a newly-won corporate client wished to keep its existing system at an on-site location rather than switch to the agency’s primary system, or (iii) the agency acquired another agency which had a contract obligating it to continue using another system. Large Agency Coalition Comments at 1-3. See also Transcript at 212.
 
3. Regulatory Background
 
The Board’s rules, adopted in 1984, included an expiration date to ensure that we would reexamine the rules after they had been in force for several years. We therefore reexamined those rules through our rulemaking completed in 1992. 57 FR 43780 (September 22, 1992). We readopted the rules, because we found that CRS rules remained necessary then to protect airline competition and to help ensure that consumers did not receive inaccurate or misleading information on airline services. We based our decision on the systems’ control by airlines and airline affiliates, which could still use their control of the systems to prejudice airline competition if there were no rules. Airlines then relied on travel agencies for distribution and had no practical ability to induce travel agencies to use systems charging lower fees, and travel agencies did not choose systems on the basis of their treatment of airlines. See 67 FR 69367, 69372.
 
The rules adopted by us regulate the operations of systems owned or marketed by an airline or airline affiliate insofar as the system was providing services to travel agencies.
 
The current rules (i) bar each system from using carrier identity as a factor for editing and ranking services, (ii) prohibit systems from charging airlines discriminatory booking fees, (iii) require each system to make available to any participating airline the booking and marketing data generated by the system from bookings for domestic travel made through the system, and (iv) prohibit certain types of restrictive contract provisions that unreasonably limit the travel agencies’ ability to switch systems or use more than one system. The rules also require each system to provide non-owner airlines with information and booking capabilities as accurate and reliable as those provided the owner airline, and they give each travel agency the right to use its own equipment in conjunction with a system and to access other systems and databases from the same terminals used to access its primary system, unless the agency uses equipment provided by that system. The rules additionally require each airline with a significant CRS ownership interest to participate in other systems at as high a level of functionality as it does in its own system, if the terms for participation are commercially reasonable (this is the mandatory participation rule).
 
Five years after our last overall reexamination of the rules, we revised the rules in two respects. First, we prohibited systems from enforcing “parity clauses” against airlines that did not own or market a competing system. 62 FR 59784 (November 5, 1997). The parity clauses required each airline to buy at least as high a level of service from the system as it did from any other system. The parity clauses made it unnecessary for systems to compete for airline participation at higher levels of service. Secondly, we strengthened the prohibition against display bias by requiring each system (i) to offer at least one display that does not give on-line connections a preference over interline connections and (ii) to either list one-stop and other direct flights before connecting services or use elapsed time as a significant factor in selecting flight options from the database. 62 FR 63837 (December 3, 1997). We strengthened the rule in large part because of evidence that United had caused Galileo to create displays that prejudiced United’s competitors. 62 FR 63840-63841.
 
C. Development of the Record in This Rulemaking

To ensure that the record in this proceeding would be as complete as possible and that all interested persons would have the opportunity to present their views and to respond to points made by other commenters, we have used procedures in addition to those required by the Administrative Procedure Act for informal rulemakings. We began this proceeding by issuing an advance notice of proposed rulemaking, 62 FR 47606 (September 10, 1997). We issued a supplemental advance notice of proposed rulemaking that asked interested persons to update the record and to comment on the implications of two developments, the Internet’s growing role in airline distribution and the systems’ shrinking airline ownership. 65 FR 45551 (July 24, 2000).
 
After reviewing the comments submitted in response to those notices, we issued our notice of proposed rulemaking on November 15, 2002. That notice, as stated above, proposed to readopt most of the existing rules but also asked for comments on whether the rules had become unnecessary. We additionally proposed to eliminate the mandatory participation rule and the prohibition against discriminatory booking fees. We tentatively concluded that we should not extend the rules to cover the distribution of airline tickets through the Internet. We asked for comment on whether we should change our policy statement requiring travel agents to disclose the full amount of airline fares to consumers so that travel agents would be obligated to state separately the amount of any travel agency service fee, as long as the fee did not exceed certain levels. We took into account the changes in the systems’ airline ownership, although only Galileo and Sabre then had no airline owners. We tentatively believed that the systems might engage in practices that would undermine airline competition due to the marketing relationships and other ties that continued to exist between the systems and their former airline owners.
 
To make certain that interested persons had ample opportunity to present their evidence and positions on the issues, we established a lengthy comment period and asked for reply comments. 67 FR 69366. We later extended the comment period and reply comment period by two months and one month, respectively. 67 FR 72869 (December 9, 2002). To provide an additional opportunity for public participation, we also held a public hearing on May 22, where interested persons could present their views to a Department official, Michael W. Reynolds, the Deputy Assistant Secretary for Aviation and International Affairs, and answer his questions. 68 FR 25844 (May 14, 2003); 68 FR 27948 (May 22, 2003).
 
We received about 95 comments and 35 reply comments. The commenters included members of Congress, other Federal agencies, the systems, many U.S. and foreign airlines, many travel agencies and travel agents, firms that process the marketing and booking data sold by the systems, and several public interest groups. Because of the complexity of the issues and the varying effects of the rule proposals, the commenters do not share common views.
 
The Justice Department argues that we should readopt the rules prohibiting display bias and should not adopt any other rules except possibly transitional rules barring the systems from demanding most-favored-nation clauses in their contracts with participating airlines. Sabre, Worldspan, United, Expedia, and Travelocity contend that we should terminate all of the CRS rules. Amadeus, Galileo, Alaska, America West, Midwest, and US Airways generally assert that most of the rules should be readopted. Orbitz, American, Continental, Delta, and Northwest argue that we should maintain some rules only for a transition period to ensure that the CRS industry’s deregulation will succeed. The travel agency commenters largely support the continuation of rules governing the systems’ contracts with their travel agency customers but object to any significant restrictions on the systems’ incentive pricing programs. The public interest groups generally oppose continued regulation, but some argue that we should take action to prevent Orbitz’ operations from reducing competition.
 
As stated above, we have determined not to make final our tentative proposals to readopt most of the rules. The comments on our notice of proposed rulemaking have shown that market forces in the CRS business are more effective than was shown by the comments submitted before we issued that notice: the airlines’ control over access to their webfares has enabled them to obtain better terms for participation in some systems, the systems’ subscriber contracts are giving travel agencies increasing flexibility to use alternative booking channels, and the airlines’ share of revenues from travel agents has continued to decline. Furthermore, as a result of the Worldspan sale, no system is now controlled by U.S. airlines.
 
Before turning to the detailed discussion of the substantive issues, we will address the procedural questions raised by commenters.
 
D. Procedural Issues
 
For this proceeding we have followed the notice-and-comment procedures established by the Administrative Procedure Act for informal rulemakings, as we have done in all past CRS rulemakings. 67 FR 69369. We also held a public hearing and invited interested persons to submit reply comments as well as comments. These informal rulemaking procedures have given commenters a fair opportunity to present their evidence and policy and legal arguments and have enabled us to resolve the issues rationally and efficiently.
 
Some parties filed comments or reply comments after the due date for those documents. We have accepted all such documents, and we have considered them to the extent practicable.
 
Sabre’s comments included several exhibits for which Sabre requested confidential treatment. Sabre thereafter concluded that some of these exhibits did not require confidential treatment, because their information was equivalent to that provided by other commenters without any request for confidential treatment. We were unable to work out an arrangement with Sabre on the remaining documents that would meet Sabre’s interests in protecting the confidentiality of the information while satisfying our need to give all interested persons an adequate opportunity to review the information while preparing their comments. We are therefore returning those documents to Sabre, and we have not considered them at all in this rulemaking.
 
Some commenters requested a more formal hearing where they could cross-examine members of our staff and representatives for other commenters. We found such additional procedures would be unnecessary for the development of an adequate record in this proceeding. 68 FR 12883 (March 18, 2003).
 
Several commenters assert that the record is stale or incomplete. See, e.g., Galileo Reply Comments at 9-13; ASTA Reply Comments at 4-8. We disagree. While our notice of proposed rulemaking cited some factual material that may not have reflected current conditions, the notice set forth our tentative factual findings, our reasoning on the economic and policy issues, and, most importantly, gave all interested persons ample opportunity to submit their own factual information. Any commenter who considered the factual record outdated or incomplete could have corrected any inadequacies by submitting current information. We believe that the record is more than adequate for our decision.
 
We also disagree with those commenters who contend that we cannot reach a rational decision on the issues without learning the details of the marketing and other on-going relationships between Worldspan and its former airline owners. See, e.g., Galileo Reply at 10. In this proceeding we are considering what general rules, if any, should be adopted that will regulate each system’s operations, not whether specific features of the arrangements between Worldspan and its former owners may be unlawful as unfair methods of competition. The record is entirely adequate for us to determine what general rules should be adopted. If it becomes apparent that specific features of the relationships between Worldspan and its former owners present questions about possible violations of section 411, we can address those issues through our investigatory and enforcement powers. In addition, the record does not include information on the details of the relationships between Galileo and United, or between Sabre and American or Southwest. Some commenters, however, have submitted evidence on their experience with those relationships, and other commenters could have done so as well. That evidence indicates neither that we must obtain additional information nor that the existing relationships create a likelihood of anti-competitive behavior that would injure airline competition and that requires regulations.
 
Our notice of proposed rulemaking included an initial regulatory flexibility analysis as required by the Regulatory Flexibility Act of 1980, 5 U.S.C. 601 et seq. That analysis discussed the potential impact of our rule proposals on small entities and invited comments on that analysis. 67 FR 69423-69424. Travel agencies, several members of Congress, the Small Business Administration’s Office of Advocacy, and some other commenters contend that we failed to comply with the Regulatory Flexibility Act, because our initial regulatory flexibility analysis allegedly failed to provide adequate analysis and an opportunity for comment on several rule proposals affecting travel agencies, particularly our proposal to restrict the systems’ incentive payment programs. See, e.g., June 9, 2003, Letter from Senators Snowe and Kerry; March 19, 2003, Letter from the Democratic Members of the House Committee on Small Business; Comments of the Small Business Administration Office of Advocacy; ASTA Comments at 51-54. We recognize the importance of the goal of ensuring that our rules do not unreasonably or unnecessarily affect small businesses and the importance of compliance with the Regulatory Flexibility Act. We believe that we have fulfilled our obligations under that statute. However, the issue is moot for the most part because we are not adopting the rule proposals that generated most of the complaints. In addition, certain other proposals sought by travel agency groups, such as a requirement that every airline make all publicly-available fares saleable through every distribution channel, are not alternatives that we have the statutory authority to adopt on the basis of the record in this proceeding. Our final regulatory flexibility analysis is set forth later in this rule.
 
We also conducted a review under 5 U.S.C. 610 of the CRS rules, Part 255, in this proceeding. As discussed below, we concluded that changes were necessary to relieve regulatory burdens and respond to changed circumstances.
 
E. The Need for Limited CRS Regulation
 
1. Introduction
 
We adopted the current rules because we found that regulations were necessary to prevent the systems from engaging in anti-competitive conduct that was likely to prejudice competition in the airline industry (for example, display bias and unjustly discriminatory booking fees). We additionally concluded that some practices followed by the systems represented efforts to preserve their market power over airlines (for example, subscriber contract provisions that kept travel agents from using alternative booking channels). We further determined that, if there were no rules, the systems would probably bias their displays, thereby denying travel agents and their customers impartial and information on airline services. 57 FR 43781-43787. In addition, as the Justice Department observes, the system owned by an airline that dominated a region had a substantially greater ability to obtain subscribers than did other systems. If that system operated in ways designed to prejudice the competitive position of rival airlines, it would reinforce its owner’s dominant position in the airline market. Justice Department Reply Comments at 9.
 
We based these conclusions on our findings that airlines relied heavily on travel agencies for distribution, that travel agents generally used a system to determine what airline services were available and to make bookings, that each travel agency predominantly or entirely used one system for these tasks, and that the resulting need of almost all airlines to participate in each system meant that market forces did not discipline the prices and terms offered by the systems for airline participation. We further relied on the fact that each system was then owned and controlled by one or more airlines or airline affiliates. 57 FR 43781, 43790, 43794.
 
Recent developments, such as the systems’ ownership changes and the growth of on-line bookings, have seriously eroded the basis for the findings on which the current rules were based. We must thus examine whether the regulation of system operations remains necessary. When we issued our notice, one system was still controlled by three U.S. airlines, and we tentatively found that the rules remained necessary because the systems still had market power over airlines and because the continuing ties between the systems and their former owners created a likelihood that systems would engage in conduct that would prejudice airline competition. 67 FR 69377-69384. We nonetheless invited comments on whether we should allow all of the rules to sunset, 67 FR 69368, and we stated that we anticipated that the on-going changes in the marketing of airline tickets could in time make the rules unnecessary. 67 FR 69376.
 
The commenters disagree on whether rules are still necessary. The Justice Department recommends that we maintain only the rules prohibiting display bias and possibly short-term rules barring certain types of most-favored-nation clauses in the systems’ contracts with participating airlines. Some commenters, such as Expedia and United, contend that the rules should be terminated now. Sabre argues that no rules are necessary unless a system is still controlled by U.S. airlines. Other commenters, like Orbitz, American, Continental, and Northwest, contend that we should adopt regulations for a transition period to ensure that the ultimate deregulation of the CRS business will be effective. And still others, like Midwest, argue that the regulations are likely to remain essential for a number of years. Some commenters, like United, argue that we may not regulate non-airline systems at all and that we should not regulate systems owned or controlled by airlines.
 
2. Final Rule
 
We have concluded that market forces are beginning to discipline the systems’ prices and terms for airline participation, and the systems’ competition for subscribers is in large part eliminating contract provisions that substantially restrict travel agents from using alternative electronic sources of airline information and booking capabilities. Furthermore, the record does not contain evidence showing a likelihood that a system will engage in conduct designed to distort competition in the airline industry, except for display bias. Readopting most of the existing regulations would not be justified without such evidence. For these reasons, we have determined to permit most of the rules to sunset upon their expiration on January 31, 2004.
 
The only exceptions are the rules that prohibit display bias and foreclose certain contract clauses with airlines that would maintain the systems’ market power. We find that the systems continue to have market power over airlines, as argued by the Justice Department; that there is some potential for conduct by the systems that could  prejudice airline competition (most notably the sale of display bias); and that systems could engage in practices that could  unreasonably preserve their market power. For these reasons, we will adopt these rules for a six-month period in order to facilitate an orderly transition to a completely deregulated distribution marketplace. We retain the power to reexamine this decision if unexpected developments show that continuing regulation may be necessary. We are also prepared to take enforcement action if a system engages in conduct that appears to violate section 411.
 
We explain in this section why we have concluded that most of the current rules are no longer needed, and that the remaining rules will be maintained only for a short transition period. The several types of system conduct that create concern require separate discussion, because they involve different groups of system users -- airlines, travel agencies, and travel agents and their customers -- and the degree and effectiveness of market forces for each group is different. For airlines, the question is whether competition disciplines the prices and terms for CRS services offered airlines. For travel agencies, the question is whether the systems can engage in conduct that tends to preserve any market power they may have over airlines by unreasonably restricting a travel agency’s use of alternative information sources and booking channels. For travel agents and their customers, the question is whether the systems could engage in display bias and similar practices that would lead to consumer deception and undermine airline competition. As a separate matter, we must determine whether, assuming that the systems do have market power over airlines, they are likely to pursue practices that would distort airline competition, even though no U.S. airlines now control any system.
 
Most commenters supporting continuing regulation assume that any rules should apply equally to all systems, whether or not owned and controlled by airlines. None of the commenters argues that Amadeus’ ownership by three European airlines provides a basis for regulating that system if the others are unregulated. We agree. We doubt that the alliance relationships between each Amadeus owner and one or more U.S. airlines will substantially increase the potential for anti-competitive behavior affecting the U.S. airline market, especially since the Amadeus owners belong to different alliances. In addition, Amadeus has substantial public ownership, and its obligations to its public shareholders should lessen any potential for action by Amadeus designed only to distort airline competition in the United States. Amadeus also has the smallest market share in the United States. Amadeus Comments at 32-33; Sabre Comments at 4, n.6.
 
The primary basis for our rule proposals was our belief that the proposals appeared necessary to prevent system practices that would constitute unfair methods of competition and that market forces would not prevent those practices. We will begin our explanation of the need for maintaining some short-term, residual regulation with our analysis of the systems’ market power over most airlines, an analysis that begins with our conclusions on market definition. We then discuss whether systems are likely to engage in conduct that would prejudice airline competition, preserve their existing market power, or give consumers and their travel agents misleading information on airline services. Despite our conclusion that the systems have market power over airlines, we are allowing most of the existing rules to expire because we find that the systems are not likely to engage in practices that would prejudice airline competition or tend to maintain their existing market power, except for display bias and the potential imposition of some contract clauses on participating airlines that would reduce the airlines’ bargaining power. Because we conclude that the systems would probably sell display bias if our prohibition against doing so were immediately terminated, thereby misleading travelers, we have decided to retain that prohibition for a six-month transitional period to furnish the industry notice of the change.
 
Where we find short-term, transitional regulation necessary, our analysis is substantially the same for both airline and non-airline systems. Elsewhere, as discussed below, our conclusions that rules are not necessary stems in large part from the lack of any U.S. airline control of the systems now operating in the United States. If Orbitz enters the CRS business, there would again be a system controlled by U.S. airlines. However, we are unwilling at this time to adopt general regulations based upon Orbitz’ potential entry.
 
3. Market Definition
 
In judging whether any regulation is necessary, the fundamental question is whether market forces would discipline system practices. If competition would do so, no rules should be necessary. Cf. Justice Department Reply Comments at 18.
 
When we adopted the current rules, we found that they were necessary because each system had market power over almost all airlines and market forces would not discipline the systems’ anti-competitive practices. We also adopted rules governing subscriber contracts, even though we did not find that systems generally had market power over travel agencies, because the systems’ contracts with travel agencies contained clauses that would maintain the systems’ market power over airlines. 67 FR 69405. In the current rulemaking, we again made a tentative determination that the systems had market power over airlines.
 
Determining whether the systems have market power over airlines requires us to define the relevant market. The relevant market must contain all products or services that consumers -- here the airlines -- are likely to consider using for the same purpose. The relevant market includes all reasonably interchangeable products and services, because “the ability of consumers to turn to other suppliers restrains a firm from raising prices above the competitive level.” United States v. Microsoft Corp., 253 F.3d 34, 51-52 (D.C. Cir. 2001), quoting Rothery Storage & Van Co. v. Atlas Van Lines, Inc., 792 F.2d 210, 218 (D.C. Cir. 1986).
 
In our notice of proposed rulemaking, we tentatively found that, for airlines, each system is a relevant market. Most airlines still obtain the great majority of their revenues from travel agents, each travel agency office normally uses only one system, and travel agents rarely make airline bookings outside a system. If travel agents routinely used several electronic sources of airline information and booking capabilities when making reservations for their customers, an airline could then afford to withdraw from one or more systems, because the travel agents’ use of alternative systems would still enable the airline to obtain bookings. Travel agencies, however, typically rely entirely or predominantly on one system for investigating airline service options and making bookings. 67 FR 69375-69376, 69377-69381.
 
As a result, an airline that wants its services to be readily saleable by travel agencies must participate in each system, because otherwise it will lose a significant amount of revenue. As the Justice Department had stated in an earlier rulemaking, quoted at 67 FR 69376:
 
Each CRS provides access to a large, discrete group of travel agents, and unless a carrier is willing to forego access to those travel agents, it must participate in every CRS. Thus, from an airline's perspective, each CRS constitutes a separate market and each system possesses market power over any carrier that wants travel agents subscribing to that CRS to sell its airline tickets.
 
We further noted that, due to the economics of the airline industry, the addition or loss of a few passengers on an airline flight will determine whether the flight is profitable. The importance of marginal revenues in the airline business meant that airlines cannot afford to lose access to any significant distribution channel. In that regard, we quoted the statement of one industry economist, Daniel Kasper, 67 FR 69375:
 
Airlines utilize many different distribution channels for the simple reason that they must do so in order to ensure that their products are easily accessible to the broadest possible array of prospective
travelers. . . . Because attracting incremental passengers is critically important to an airline’s profitability, each airline strives to match or surpass the visibility to purchasers enjoyed by its rivals. That is, airlines must compete for “shelf space” in any channel where consumers prefer to shop.
 
The comments support our tentative factual findings on market definition. First, most airlines still obtain the majority of their revenues from bookings made by travel agencies through a system. The Justice Department states that the five airlines that own Orbitz derived 65 percent of their total revenues in March 2002 from “brick-and-mortar” travel agency bookings. Justice Department Reply Comments at 14. America West states that 67 percent of its revenues in 2002 came from bookings made through the systems. America West Comments at 7. Alaska similarly states that it obtains 56 percent of its revenues from travel agencies. Alaska Comments at 5. Delta states that 55 percent of its revenues are produced by “brick-and-mortar” travel agencies and that another 10 percent are produced by on-line travel agencies through a system. Delta Reply Comments at 39. Sabre by itself produces about one-third of a typical airline’s revenues. Orbitz Comments at 10. While the Justice Department suggests that the systems’ use by on-line travel agencies (as opposed to “brick-and-mortar” travel agencies) adds little to their market power over airlines, because most consumers check two or more websites before making a booking on-line, the Justice Department agrees that the systems have market power due to their usage by “brick-and-mortar” travel agencies. Justice Department Reply Comments at 15. About 80 percent of CRS bookings made by travel agencies are made by “brick-and-mortar” agencies. Galileo Comments, Guerin-Calvert, Jernigan, & Hurdle Declaration at 24.
 
In arguing that the systems do not have market power, Sabre cites figures showing that less than half of all tickets will be sold this year by travel agencies using a system. See, e.g., Sabre Comments, McAfee and Hendricks Declaration at 2; Transcript at 8. We believe that market shares based on revenues, not individual tickets, should be determinative. A firm’s profitability directly depends on its total revenues, not on the number of units sold. The travelers who make bookings on-line tend to buy tickets that are sold at greater discounts. The travelers using “brick-and-mortar” travel agencies are more important to the airlines because they tend to buy the more expensive tickets. Justice Department Reply Comments at 16.
 
We agree with Sabre that the travel agencies’ share of total bookings has been declining and will likely continue to decline. See, e.g., Justice Department Reply Comments at 14. However, as noted, the large network airlines still obtain the large majority of their revenues from travel agencies using a system, a situation likely to persist for some time to come.
 
Business travelers -- the travelers that produce a disproportionate share of the network airlines’ revenues -- have been reluctant to make bookings on-line or otherwise outside the travel agency channel. Justice Department Reply Comments at 16; NBTA Comments at 11-14. Consumers make about five times as many on-line bookings as do corporate travelers. Galileo Comments, Guerin-Calvert, Jernigan, & Hurdle Declaration at 26, n. 40. We recognize that a growing number of business travelers are booking on-line, but they appear to be doing so through websites offered by travel agencies using a system, or through one of the corporate booking firms acquired by systems like Sabre. Sabre Reply Comments at 34-35; American Reply Comments at 25.
 
It may well be that within several years even a large proportion of business travelers will book their air travel outside of travel agencies using a system, but they do not do so now. Most airlines, including the major network airlines, derive the large majority of their revenues from bookings made through a system. See also Galileo Comments, Guerin-Calvert, Jernigan, & Hurdle Declaration at 29.
 
Secondly, travel agents continue to rely on systems for booking airline tickets. ASTA states that, on average, 87 percent of travel agency airline bookings are made through a system. ASTA Comments at 23. Galileo estimates that an even higher percentage of travel agency bookings are made through a system. Galileo Comments, Guerin-Calvert, Jernigan, & Hurdle Declaration at 25, n. 37. Travel agents generally have access to the Internet and use it, primarily for research on travel options, but they have not made much use of the Internet for airline bookings, as noted above, because using the Internet is significantly less efficient than using a system.  ASTA Comments at 12-13.
 
Thirdly, to operate more efficiently, most travel agencies use only one system, as discussed above. While the largest travel agencies tend to have two or more systems, they do not seem to make substantial use of all of them. Those agencies typically rely predominantly on one system. The Large Agency Coalition states that its members -- all large corporate travel agencies -- do not subscribe to multiple systems in order to improve their ability to book airline travel, but because of continuing business relationships between the agency and the dominant airline in local markets, between some of their corporate customers and airlines, or between an acquired agency and its system. Large Agency Coalition Comments at 1-3. Carlson Wagonlit alleges that each of its branch offices relies predominantly on one system even though the travel agency firm subscribes to all of the systems: “Using multiple CRSs at one location creates numerous operational difficulties related to training agents on multiple CRSs and because client information is maintained within the CRS.” Carlson Wagonlit Comments at 11.
 
Fourthly, the airlines’ dependence on marginal revenues requires them to participate in every significant distribution channel. No commenter denies that marginal revenues are critical in the airline industry. Sabre’s experts agreed with our finding: “Air transportation involves high fixed costs and low marginal costs. Thus a few incremental bookings can spell the difference between profit and loss.” Sabre Comments, Salop & Woodbury Declaration at 29.
 
We are unconvinced by the claims of several commenters that airlines can nonetheless find substitutes for the travel agency channel and that travel agents can use substitutes for the systems. We recognize that Southwest, JetBlue, and some other low-fare airlines operate successfully without obtaining many bookings from travel agents. Southwest and JetBlue reportedly obtain only 20 percent and 10 percent of their revenues, respectively, from travel agencies. Justice Department Reply Comments at 15, n.14. Other airlines, particularly the large network airlines, cannot now practicably end their reliance on the travel agency channel. The low-fare airlines have traditionally focused on attracting leisure travelers. As shown, leisure travelers are much more likely to book flights through the Internet without using a “brick-and-mortar” travel agency (or an on-line agency). Insofar as other airlines follow a business strategy that involves attracting business customers -- the travelers most likely to use travel agencies -- those airlines continue to be dependent on travel agencies for the largest share of their revenues and may have limited bargaining leverage against the systems, at least in the near future. The network airlines, moreover, tend to operate more complex hub-and-spoke route systems than the low-fare airlines, and that complexity limits their ability to obtain direct sales, unlike airlines such as Southwest that primarily operate point-to-point services. It may be that the network airlines would be more successful if they adopted the same business strategy as the low-fare airlines. They have not done so, however, and presumably could not do so without significant expense. American Comments at 17-21; 67 FR 69379. As a result, these airlines rely on travel agencies for the majority of their revenues. Our determination of the relevant market must rely on the choices actually made by airlines and consumers, not on the choices that some think they should make. Cf. U.S.-U.K. Alliance Case, Order 2002-1-12 (January 25, 2002) at 42-43.
 
We recognize that airlines have been shifting some bookings away from the travel agency channel to their own websites. This shift has been much stronger for low-fare airlines than for the large network airlines. Despite these efforts, some believe that the Internet is unlikely to produce more than 40 percent of airline revenues by 2005. Galileo Comments, Guerin-Calvert, Jernigan, & Hurdle Declaration at 23-24. Airlines have also taken steps to encourage travel agencies to bypass the systems. For example, American has an arrangement with American Express that enables that travel agency to make bookings directly with American. Amadeus Comments at 12-13. The record does not indicate that direct booking arrangements will substantially reduce the agencies’ use of the systems for airline bookings any time in the near future. As shown, the larger airlines still obtain the large majority of their revenues from bookings made through the systems.
 
Several commenters contend that travelers can use alternative distribution channels and are not locked into the travel agency channel, or, alternatively, can switch between travel agencies if one agency uses a system that provides inferior service. See, e.g., Sabre Comments at 59-65. We agree that consumers can choose where to book and need not book through a travel agency if they do not wish to, and that many consumers can easily switch between travel agencies. At least for corporate customers, however, changing agencies will impose some switching costs. Justice Department Reply Comments at 16, n.19. Airlines do not enjoy such choices. If a substantial number of travelers choose to use travel agencies, as they do, and if those travel agencies, with few exceptions, use only one system and do not readily make bookings outside the system, as is true, then each airline must participate in each system used by a significant number of travel agencies in order to avoid losing bookings from those agencies. As we stated in the notice, 67 FR 69378:
 
The existence of one distribution channel that is attractive to a significant and growing number of travelers does not make that channel competitive with another channel that a larger if shrinking share of travelers finds preferable. With a very few exceptions, any airline that uses only one channel will not obtain the business of those travelers that prefer the other channel.
 
See also American Comments at 16-17 and Dorman Declaration at 5. While the airlines’ customers have alternatives, that does not make irrelevant the question of whether systems have market power over airlines. Cf. United States v. Visa U.S.A., Inc., 344 F.3d 229, 239 (2d Cir., 2003); In Re Visa Check/Mastermoney Antitrust Litigation, E.D.N.Y. No. 96-CV-5238, April 1, 2003, Memorandum and Order at 5.
 
Some arguments made by the commenters opposing our preliminary analysis mischaracterize our reasoning. Sabre wrongly alleges that we concluded that systems have market power over travel agencies. Sabre Comments at 59, 71, 84. Nothing could be further from the truth. We expressly found that systems compete vigorously for travel agency subscribers, 67 FR 69371, 69405, and nowhere did we state that systems have market power over travel agencies. Sabre additionally misstates our analysis by asserting that we found that travel agencies control their customers. Sabre Comments at 59, 63.
 
Sabre has failed to show that the relevant market is not each system, but the broader market of providing travel information to consumers, or airline ticket distribution, a market in which each system’s share would be relatively small. Sabre Comments at 57-59, 79. As a practical matter, airlines wishing to electronically provide information and booking capabilities to travel agencies currently have no effective substitute for participation in each system. Similarly, because travel agencies do not use multiple systems, Sabre’s observation that no system has even a 50 percent share of the CRS business, Sabre Comments at 81, is irrelevant. Each system is a separate market insofar as airlines are concerned. Furthermore, each system has a dominant share of the CRS business at cities where its former airline owners were the dominant airlines. Justice Department Reply Comments at 22.
 
 
4. The Systems’ Market Power over Airlines
 
Because readopting CRS rules to block anti-competitive behavior will require a finding that the systems have market power over most airlines, we must determine whether they do have such power. If systems have market power over airlines, they will be able to charge them prices that exceed competitive levels, and the resulting costs will be passed on to consumers, even if many or most consumers can choose between different distribution channels when buying airline tickets.
 
We are following the definition of market power applied by the Supreme Court in antitrust cases. In Eastman Kodak Co. v. Image Technical Services, 504 U.S. 451 (1992), the Court stated that market power is the power "to force a purchaser to do something that he would not do in a competitive market," 504 U.S. at 464, quoting Jefferson Parish Hospital v. Hyde, 466 U.S. 2, 14 (1984), and "the ability of a single seller to raise price and restrict output," 504 U.S. at 464, quoting Fortner Enterprises, Inc. v. United States Steel Corp., 394 U.S. 495, 503 (1969). The courts have similarly stated that a firm is a monopolist “if it can profitably raise prices substantially above the competitive level.” United States v. Microsoft Corp., 253 F.3d at 51.
 
Our notice of proposed rulemaking stated our belief that each system still has market power over most airlines. We noted in that regard that some airlines that had otherwise supported the elimination of most or all of the rules still conceded that the systems have market power. Northwest had thus stated, as quoted by us at 67 FR 69378:
 
Sales to consumers made over the Internet, via both airline websites and online agents, have provided significant new competition to CRSs, but each CRS typically remains the only means by which to reach the travel agents who use that system. Each CRS therefore continues to have significant market power based on the travel agents to which it has exclusive access.
 
 
First, until now an airline or other firm could not practicably create competitive alternatives for the systems. Among other things, building a new system would be costly and time-consuming, and the great majority of travel agencies already had contracts to use an existing system. 67 FR 69381. Entry into the business has become easier, as argued by Sabre. Sabre Comments at 52-85. However, because travel agencies generally rely entirely or predominantly on one system for information and bookings on airline services, new entry is unlikely in the near term to eliminate the systems’ existing market power.
 
Secondly, airlines have generally been unable to persuade travel agencies to use one system rather than another. If they could, they would have some bargaining leverage against the systems. Airlines could then shift business to systems offering better terms for airline participants and away from systems offering poorer terms. Because travel agencies do not pay booking fees, they have no direct incentive to use the system charging the lowest fees. The record suggests, in fact, that the incentive payment programs used by the systems encourage travel agencies to choose the system that is the most expensive for participating airlines.  The systems then obtain subscribers typically by offering to give them bonus payments. The revenues used for those incentive payments come from the fees paid by participating airlines (and to a smaller extent by other travel suppliers). See, e.g., American Reply Comments, Dorman Declaration at 2-4.
 
Airlines have had no effective incentives that they can offer travel agencies to encourage the use of one system rather than another, except in local markets where a dominant airline can influence travel agency choices by denying access to its corporate discount fares and marketing benefits to travel agencies that do not use its preferred system. As discussed in our notice of proposed rulemaking, airlines that dominate an area’s airline markets, like Delta at Atlanta and American in southern Florida, can influence local travel agencies to use the airline’s preferred system, because those travel agencies cannot easily succeed without the ability to sell the corporate discount fares offered by the area’s major airline. 67 FR 69381.
 
Airlines have developed programs to encourage travel agents to agree to terms that offset some CRS costs, or to bypass the systems, but those programs do not yet seem to have had great success. American’s “Everyfare” program gave travel agencies access to American’s webfares if they agreed to assume the airline’s booking fee liability. Amadeus Comments at 10-13. Northwest and other airlines have created websites designed for travel agent bookings. Sabre Supp. Reply at 2.
 
We recognize that airlines have been gaining bargaining leverage against the systems, a factor that caused us to propose the elimination of the mandatory participation rule and the rule barring discriminatory booking fees. Nonetheless, the systems currently have significantly greater leverage. An airline’s greatest leverage for obtaining lower fees or better terms for participation will be a threat to withdraw from the system. If an airline withdraws, however, it will immediately begin losing bookings from that system, and those losses will not be entirely offset by increased bookings through the Internet. Any saving in CRS participation expenses will arrive later, and will not quickly offset the revenues lost from the reduction in bookings. Booking fees, after all, equal about two percent of the revenues obtained by an airline from sales made through a system. Orbitz Comments at 10, n.4. Cf. Amadeus Comments at 18-19.
 
It is true that an airline’s withdrawal from a system will make that system less attractive to travel agencies, and over time the system will lose subscribers. Because the average travel agency contract has a term of three years, however, only a relatively small portion of the system’s subscribers will have the ability to switch to another system in the short term.
 
Thus the airline’s revenue losses from withdrawal will be substantial and begin occurring immediately, while the system’s losses in subscribers will be gradual and occur only over a period of some months. In these circumstances, the system should have the upper hand in bargaining. See, e.g., Orbitz Comments at 10.
 
An airline could also put pressure on the system by attempting to reduce the number of tickets sold through the system without withdrawing completely. One possibility would be to increase their efforts to encourage travelers to book directly with the airline. These lost sales would lower the systems’ revenues, but may also increase the airline’s distribution costs.
 
An airline could put pressure on the system by lowering its participation level, because doing so would make the system less attractive to travel agencies that frequently book the airline without drastically reducing the airline’s bookings from that system’s subscribers. The lower level of participation would make it somewhat harder for travel agents to obtain information and reliably make bookings, and could block travel agents from conducting functions that are important to their customers. These functionality differences would not lead to a loss of as many bookings as would withdrawal but presumably would still result in lower revenues from the travel agents using that system. On the other hand, the lower level of participation would have less impact on the system’s ability to market itself to travel agencies in the future. We expect that airline changes in participation levels will give airlines bargaining leverage.
 
Our notice of proposed rulemaking predicted that the airlines’ control over access to their webfares could enable them to obtain better terms for system participation. 67 FR 69381. As discussed above, Sabre and Galileo have begun programs that give airlines a discount from the standard booking fee levels in exchange for a commitment to provide all publicly-available fares, including webfares. The commenters disagree over the implications of these programs. Some commenters assert that airlines have gotten little in exchange for the commitments required of them. See, e.g., American Reply Comments at 21-23. America West states that Orbitz has offered substantially larger fee reductions for airlines that agree to its most-favored-nation clause. America West Reply to Supp. Comments at 2-3. Other commenters contend that the programs demonstrate that airlines have bargaining power and that the systems do not have market power. See, e.g., Sabre Reply Comments, Salop & Woodbury Declaration at 15-16.
 
We believe that the airlines’ ability to change their participation levels and their control over access to webfares is reducing the systems’ market power. Overall, however, we find that the systems currently still have market power over most airlines, although the continuing changes in airline distribution, particularly the growing importance of the Internet for airlines, travel agents, and travelers, should continue to erode the systems’ market power. Our finding that the systems have market power is consistent with the Justice Department’s conclusions. Justice Department Reply Comments at 2, 16-17.
 
We disagree with Sabre’s contention, first made in its reply comments, that the airlines’ contracts with corporate customers keep systems from having market power. Sabre asserts that system practices cannot significantly affect airlines, because ”much business travel” involves fares directly negotiated with specific airlines, often booked through direct links. Sabre Reply Comments at 36; Sabre Reply Comments, Salop & Woodbury Declaration at 7-9. Airlines obtain substantial amount of business from corporate customers that do not have such contracts, and the contracts do not normally bar employees from traveling on alternative airlines.
 
We have based our finding of market power on the industry’s structural characteristics, not on an analysis of whether the systems’ fees are at supracompetitive levels. The best evidence of a firm’s monopoly power would be a showing that it has been able to profitably charge prices that significantly exceed competitive levels. Because direct evidence of this ability is usually not available in Sherman Act monopolization cases, the courts usually rely on market structure evidence to determine whether a firm has monopoly power. United States v. Microsoft Corp., 253 F.3d at 51. We have taken the same approach here.
 
When we last compared the systems’ prices with their costs, we concluded that the larger systems at least were charging supracompetitive prices. See 56 FR 12586, 12595 (March 26, 1991). We have not done such an analysis since then, as we noted in our notice, but stated our belief that the systems’ booking fees were probably above competitive levels, because they were not disciplined by market forces. 67 FR 69382. t with our findings that the systems must compete for travel agency subscribers but do not compete for airline participants.
 
The airline commenters generally support our finding that booking fees are not disciplined by competition and contend that the fees substantially exceed competitive levels. They point out, for example, that the network airlines’ financial crisis since 2001 has enabled them to drive down costs from other suppliers while the systems have been raising their fees and reporting large profits. See, e.g., America West Comments at 7-9.
 
In response, the systems have denied that their fees are not disciplined by competition, and they argue that the fees are reasonable. They contend that their costs have been rising due to increased functionality provided airlines and the growing number of messages carried by their communications links. See, e.g., Galileo Comments at 38-39. While the systems thus contend that several important cost factors have increased significantly in recent years, they have not submitted a detailed cost analysis that would show that their booking fees do not significantly exceed their costs, nor have they attempted to demonstrate that the booking fees charged before the beginning of the cited cost increases did not significantly exceed their costs.
 
We continue to believe that the systems’ fees exceed competitive levels for the reasons set forth in the notice of proposed rulemaking. We have not seen evidence that the systems’ fees generally respond to market forces, although two of the four systems have made modest concessions in exchange for access to airline webfares. However, we have not done an analysis of the systems’ costs and revenues that would demonstrate that their fees exceed competitive levels. As explained above, a finding that the fees are at supracompetitive levels is not necessary for our determination that the systems have market power over airlines.
 
We also cannot accept Sabre’s claim that bookings made through a system are relatively inexpensive for airlines while bookings made through airline websites are not (and that bookings made through airline websites are more expensive than those made by an airline’s reservations agents). Sabre Comments, Wilson Declaration at 22. Sabre’s analysis is belied by the efforts of virtually every airline to shift bookings to its own website. Several low-fare airlines have claimed that their ability to obtain most of their revenues from direct sales gives them a great cost advantage over other airlines. See American Reply Comments at 32. See also 67 FR 69373, 69374. Sabre in any event has failed to demonstrate that its calculation is valid. American Reply Comments, Dorman Declaration at 8-9; United Reply Comments at 35, n.96; America West Reply Comments at 27. See also Northwest Reply Comments at 19-20.
 
5. The Potential for System Conduct Undermining Airline Competition

Our finding that each system has market power over airlines is not sufficient by itself to justify the adoption of rules. To adopt rules regulating the systems in order to prevent potential unfair methods of competition, we should have evidence that, if there were no regulations, systems would likely engage either in anti-competitive conduct designed to preserve their market power, a subject discussed below, or in conduct intended to distort airline competition. Any such conduct would harm consumers, either by causing airlines to pay supracompetitive prices for CRS services or by denying consumers the benefits of lower fares and better service created by competition between airlines.
 
When each system was owned and controlled by one or more airlines or airline affiliates, experience demonstrated that systems were likely to engage in conduct designed to prejudice the competitive position of rival airlines, for example, by biasing displays against the owner airlines’ competitors and charging competing airlines discriminatorily high booking fees. See 56 FR 12589. None of the systems now operating in the United States, however, is owned by a U.S. airline. Obviously a system that is not owned or controlled by a U.S. airline will not have the same incentives to prejudice the competitive position of rival airlines. Justice Department Reply Comments at 13-14; Sabre Comments, Salop & Woodbury Declaration at 26-30 and McAfee & Hendricks Declaration at 53-59. We must therefore determine whether a non-airline system (a system not owned or controlled by an airline or airline affiliate) is likely to engage in unfair methods of competition.
 
We have found, as shown, that the systems have market power over airlines. To the extent that they do, their booking fees may exceed the fee levels that would exist in a competitive market, and the service offered airlines by the systems may be below the level of service that would exist in a competitive environment. The systems’ possession of market power, however, by itself would not justify rules regulating their practices. The antitrust laws permit firms with monopoly power to use that power as long as they do not engage in conduct that is designed to maintain or extend that power. “[M]erely possessing monopoly power is not itself an antitrust violation.” United States v. Microsoft Corp., 253 F.3d at 51. As explained below in our analysis of our authority under section 411, we may prohibit unfair methods of competition, which are practices that violate the antitrust laws or antitrust principles.
 
Our notice of proposed rulemaking stated our belief that there was a risk that non-airline systems would engage in anti-competitive conduct in order to prejudice airline competition. Each of the non-airline systems still had ties with its former U.S. airline owners, and each of the non-airline systems was being marketed by one or more of its former owners. The record suggested, moreover, that marketing airlines took actions favoring a system even when doing so appeared to be contrary to their interests in selling their own tickets. We therefore proposed to apply the rules, to the extent they were readopted, to non-airline systems. 67 FR 69383.
 
The systems continue to have marketing relationships and other relationships with their former owner airlines. See, e.g., Amadeus Comments at 25, n.24; Galileo Supp. Comments at 3. The lack of control by any U.S. airline will not eliminate the possibility that a system would agree with an airline to engage in conduct that would undermine the competitive position of the airline’s rivals. Each system, after all, continues to have market power over most airlines, and each of the larger airlines dominates some local markets, primarily at its hubs. A system and such an airline might agree that the system would change its operations so as to benefit the airline while the airline would use its local dominance to strengthen the system’s marketing efforts. Justice Department Reply Comments at 19.
 
The record suggests that the systems are willing to sell preferential treatment to airlines at least insofar as display bias is concerned. Their willingness to do so is apparent from their own comments, which argue that we should allow systems to sell bias. Amadeus Comments at 53-54; Sabre Comments at 141-142. The Justice Department believes that the systems are likely to engage in display bias. Justice Department Reply Comments at 19-21. See also American Antitrust Institute Comments at 8. Our notice cited evidence that display bias is sold to suppliers in other travel industries. 67 FR 69383. Although Amadeus has denied that it biases its displays for hotels and rental cars, Amadeus Reply Comments at 12, n.16, the other systems’ comments do not address this issue.

Apart from bias, however, the record does not indicate that systems are likely to seek to operate in ways designed to prejudice airline competition. Our notice of proposed rulemaking expressly invited commenters to submit evidence on whether systems had sought to distort competition in other travel industries. 67 FR 69383. One speaker at our public hearing stated that he did not know of any system practices that distorted competition in other industries, Transcript at 85, and one commenter asserted that there is no evidence of competitive harm resulting from the systems’ treatment of firms in other travel industries. Worldspan Reply at 17. See also Transcript at 116-117, 151-154. The record further suggests that the marketing relationships between systems and airlines currently give the marketing airline little incentive to help the system and that marketing airlines, in fact, do little to help the system being marketed. American Comments at 30; Large Agency Coalition Comments at 14-15; Large Agency Coalition Reply Comments at 16-17. This suggests that the ties between airlines and systems may have weakened enough so that systems would have little interest in taking action that undermined airline competition in order to favor one airline. The Justice Department additionally believes that contractual arrangements between airlines and systems do not pose a sufficient threat to competition to justify the adoption of general rules at this time. Justice Department Reply Comments at 1-2. See also Expedia Reply Comments at 3, n.1. We note, nonetheless, Amadeus’ complaint that American, Delta, and Northwest have recently tied a travel agency’s ability to sell corporate discount fares with the use of the system affiliated with the airline. Amadeus Comments at 91-92. However, this tying affects competition between the systems and does not necessarily show that systems will engage in conduct designed to distort airline competition.

Furthermore, we cannot predict at this point what kinds of relationships may arise as a result of the CRS industry’s deregulation. We do not wish to adopt rules now when we do not know what types of potential anti-competitive practices, if any, may occur. We therefore do not agree with the arguments of some commenters that rules should be maintained on the ground that systems have continuing marketing and other special arrangements with selected airlines. See, e.g., Galileo Comments at 7-11.
 
We fully agree with the Justice Department, however, that there is a potential for contractual relationships between systems and airlines that would be designed to reduce competition in either or both the CRS and airline industries. The Justice Department has stated its intent to take action against any such agreements that violate the antitrust laws, and we also have statutory authority to take appropriate action if such contractual relationships appear to be unfair methods of competition that violate section 411. Under 49 U.S.C. 41708, formerly section 407 of the Federal Aviation Act, we can obtain copies of any agreements between airlines and systems if we see a need to investigate contractual relationships between systems and participating airlines.

6. System Practices that Preserve Market Power

While we have determined that most of the rules should not be readopted, even though each system continues to have substantial market power over airlines, we are readopting for a short transition period the rule prohibiting parity clauses and adopting an analogous rule prohibiting most-favored-nation clauses demanded as a condition for any participation in a system. These types of contract clauses would tend to maintain the systems’ market power and reduce the bargaining leverage of participating airlines. Because we are essentially deregulating the CRS business notwithstanding the systems’ market power, we decided to adopt the parity and most-favored-nation clause prohibitions for a period long enough allow affected parties to respond to the transition to complete deregulation.
 
We originally adopted the rule prohibiting systems from enforcing parity clauses (except as to airlines that owned or marketed a competing system) because three of the systems had imposed parity clauses on airline participants. These clauses required each airline to participate in the system at at least as high a level as it participated in any other system. Thus, for example, Sabre’s parity clause required Alaska to participate in Sabre at the full availability level as long as Alaska participated in any other system at that level, even if Alaska considered Sabre’s service at that level too costly or not as attractive as the comparable service offered by other systems. 62 FR 59786-59787, 59791-59792. Because these parity clauses eliminated some possibility of system competition for airline participants, and required each airline to buy a level of service that an airline might not wish to buy, we adopted a rule prohibiting the systems from enforcing airline parity clauses except as to airline participants that owned or marketed a competing system. 62 FR 59784.
 
We have concluded that this rule should be readopted for another six months. We are also adopting for the same period an analogous rule that will prohibit each system from requiring airlines as a condition to any participation in the system to make all publicly-available fares saleable through the system. If we did not provide for an orderly transition, a contract clause requiring a participating airline to provide all webfares as a condition to participation, sometimes referred to as a most-favored-nation clause, would deny the airline the ability to use its control over access to its webfares as bargaining leverage to obtain better terms and prices for system participation. Such a clause would additionally tend to prevent the development of alternative sources of information and booking channels, for a travel agency would have less incentive to use alternatives if the system used by the agency already provided complete information on webfares. It is our expectation that the six-month period during which our prohibition on such clauses will remain in place will enable airlines to prepare more effectively for the termination of these rules.
 
On the other hand, we have decided not to readopt rules designed to prohibit system contract practices that would unreasonably restrict travel agency subscribers from switching systems or using alternative systems or booking channels. In the past, the systems engaged in subscriber contract practices that appeared to be designed to preserve their market power. Travel agencies accepted such contract clauses even though most travel agencies could choose between systems. 67 FR 69405. We therefore adopted rules barring subscriber contracts from having a term that exceeded five years and giving travel agencies the right to use their own third-party equipment and software in conjunction with a system.
 
As discussed above, the record shows that travel agencies in recent years have been obtaining more flexible contracts from the systems. The term of the average subscriber contract, for example, is well under five years. While most subscriber contracts still have productivity pricing clauses, the productivity pricing clauses in the contracts currently offered travel agencies do not seem to effectively block travel agents from using alternative booking channels. And travel agencies appear to have a substantial ability to switch systems at the end of their contract term. While systems may have some contracts that may be unreasonably restrictive, their contracts in general do not seem to block travel agents from obtaining information and making bookings outside the system. Moreover, the market is moving in a more competitive direction -- travel agencies are obtaining more flexibility, not less, in their newest contracts.
 
As a result, the current record shows that rules regulating travel agency contracts are no longer necessary. Several airline commenters and Orbitz have argued that we should continue to regulate the systems’ subscriber contract practices, because the existing contracts are alleged to unreasonably lock travel agencies into using their existing system. See, e.g., Orbitz Comments at 46-49; America West Comments at 26-29; American Comments at 33-35; Continental Comments at 17-20; Delta Comments at 41-42. For the reasons discussed below in connection with the specific subscriber contract issues, the systems’ current contracts do not appear to unreasonably keep travel agencies from using alternative booking channels.
 
 
7. The Systems’ Ability to Engage in Display Bias
 
Display bias has been a concern since the systems were first developed. Experience has demonstrated that travel agents are likely to book one of the first services d