Aviation Competition: Effects on Consumers from Domestic Airline
Alliances Vary (Letter Report, 01/15/99, GAO/RCED-99-37).
Pursuant to a congressional request, GAO provided information on: (1)
the status of the proposed alliances to be formed by the six largest
U.S. airlines; (2) the potential beneficial and harmful effects on
consumers; and (3) the authority of the Department of Justice (DOJ) and
the Department of Transportation (DOT) to review these alliances and the
status of their reviews.
GAO noted that: (1) all six airlines have begun implementing various
aspects of their agreements; (2) Northwest Airlines completed its
acquisition of equity in Continental Airlines, and the two airlines
began implementing their reciprocal frequent flyer programs; (3) since
GAO testified in June 1998, however, Northwest and Continental have
revised their agreement; (4) under the terms of the revised agreement,
Northwest altered its equity investment in Continental, agreed to forgo
its right to place someone on Continental's Board of Directors, and
agreed to forgo code-sharing with Continental in certain domestic
markets; (5) even though Northwest and Continental have implemented
their agreement, it remains under review at both DOJ and DOT; (6) the
alliance between United Airlines and Delta Air Lines was originally to
include code-sharing, but it has been scaled back to an arrangement
involving reciprocal frequent flyer programs and access to airport
lounges; (7) this arrangement, which the airlines began implementing in
September 1998, is much the same as the one American Airlines and US
Airways proposed and began implementing in August 1998; (8) GAO analysis
of Northwest and Continental's proposed alliance showed that the
alliance could result in new, possibly improved, route options, and the
alliance's extended frequent flyer program may benefit members of each
airline's program; (9) GAO also found that this alliance will create
some new markets that are not already served by other airlines; (10)
however, GAO's analysis indicated fewer new markets than the alliance
partners estimated, and it showed that these new markets will serve
relatively few passengers; (11) GAO's analysis indicates that if
Northwest and Continental do not act independently, competition could
decline in 63 markets that served 2 million passengers in 1997, and the
two airlines could also increase by 5 percent in the number of markets
that they dominate; (12) DOJ and DOT are separately reviewing the three
alliances under different statutory authorities and have different
remedies available to them; (13) on October 23, 1998, DOJ filed a civil
antitrust action to prevent Northwest from acquiring or holding a
majority of Continental's voting stock; (14) DOJ said in its complaint
that Northwest's gaining control would lessen competition in interstate
trade and commerce and unreasonably restrain trade; and (15) Congress
recently authorized DOT to impose waiting periods before certain joint
venture arrangements involving major airlines.
--------------------------- Indexing Terms -----------------------------
REPORTNUM: RCED-99-37
TITLE: Aviation Competition: Effects on Consumers from Domestic
Airline Alliances Vary
DATE: 01/15/99
SUBJECT: Commercial aviation
Antitrust law
Jurisdictional authority
Corporate mergers
Competition
Competitive advantage
Trade regulation
Airline regulation
Airline industry
Transportation rates
******************************************************************
** This file contains an ASCII representation of the text of a **
** GAO report. Delineations within the text indicating chapter **
** titles, headings, and bullets are preserved. Major **
** divisions and subdivisions of the text, such as Chapters, **
** Sections, and Appendixes, are identified by double and **
** single lines. The numbers on the right end of these lines **
** indicate the position of each of the subsections in the **
** document outline. These numbers do NOT correspond with the **
** page numbers of the printed product. **
** **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced. Tables are included, but **
** may not resemble those in the printed version. **
** **
** Please see the PDF (Portable Document Format) file, when **
** available, for a complete electronic file of the printed **
** document's contents. **
** **
** A printed copy of this report may be obtained from the GAO **
** Document Distribution Center. For further details, please **
** send an e-mail message to: **
** **
** **
** **
** with the message 'info' in the body. **
******************************************************************
Cover
================================================================ COVER
Report to Congressional Requesters
January 1999
AVIATION COMPETITION - EFFECTS ON
CONSUMERS FROM DOMESTIC AIRLINE
ALLIANCES VARY
GAO/RCED-99-37
Domestic Airline Alliances
(348107)
Abbreviations
=============================================================== ABBREV
DOJ - Department of Justice
DOT - Department of Transportation
GAO - General Accounting Office
HHI - Herfindahl-Hirschmann Index
Letter
=============================================================== LETTER
B-280704
January 15, 1999
The Honorable John McCain
Chairman, Committee on Commerce,
Science, and Transportation
United States Senate
The Honorable Slade Gorton
Chairman, Subcommittee on Aviation
Committee on Commerce, Science,
and Transportation
United States Senate
Early in 1998, the six largest U.S. airlines, which account for
nearly 70 percent of domestic airline traffic, announced their
intentions to form three alliances, in which the partners--Northwest
and Continental, Delta and United, and American and US Airways--would
cooperate on some aspects of their business (see app. I for
information on the airlines' market shares). These alliances vary
from a limited marketing arrangement, such as reciprocal frequent
flyer programs, to more complex agreements, such as those involving
"code-sharing"\1 or one partner's ownership of an equity share in the
other partner's business. The airlines say that these alliances will
benefit consumers through expanded route networks and combined
frequent flyer programs. Others, however, say that the alliances
will decrease competition, ultimately reducing passengers' choices
and increasing fares. Concerned over the potential anticompetitive
effects of the alliances, the Department of Transportation is
reviewing them, and the Department of Justice filed suit in October
1998 to prevent Northwest from acquiring voting control of
Continental. Justice did not, however, request a temporary
injunction precluding the transfer of voting control.
At your request, we have been examining the implications of these
alliances. On June 4, 1998, we offered the preliminary results of
our analysis in testimony before the Subcommittee on Aviation, Senate
Committee on Commerce, Science, and Transportation.\2 This report
expands on that testimony and offers more information about the
implications of the alliances. As agreed with your offices, the
report (1) describes the status of each of the alliances, (2)
examines, for each alliance, the potential beneficial and harmful
effects on consumers, and (3) examines the authority of the
departments of Justice and Transportation to review these alliances
and the status of their reviews.
--------------------
\1 Code-sharing allows an airline to sell seats on its partner's
plane as if they were its own, enabling the airline to expand its
route network without adding any planes. For example, if Northwest
and Continental have a code-sharing agreement and Northwest flies
from Minneapolis to Duluth (and Continental does not), and
Continental flies from Amarillo through Houston to Minneapolis (and
Northwest does not), then both airlines could sell tickets from
Amarillo to Duluth as their own flights, and each computer
reservation system would indicate that both airlines provide seamless
("on-line") service to these cities. Thus, both Continental and
Northwest would increase their route networks without adding any new
flights. The computer reservation system could also show this flight
a third time as a connecting flight with segments served by both
airlines.
\2 Aviation Competition: Proposed Domestic Alliances Raise Serious
Issues (GAO/T-RCED-98-215, June 4, 1998).
RESULTS IN BRIEF
------------------------------------------------------------ Letter :1
All six airlines have begun implementing various aspects of their
agreements. Northwest completed its acquisition of equity in
Continental, and the two airlines began implementing their reciprocal
frequent flyer programs. Since we testified in June 1998, however,
Northwest and Continental have revised their agreement. Under the
terms of the revised agreement, Northwest altered its equity
investment in Continental, agreed to forgo its right to place someone
on Continental's Board of Directors, and agreed to forgo code-sharing
with Continental in certain domestic markets. Even though Northwest
and Continental have implemented their agreement, it remains under
review at both Justice and Transportation. The alliance between
United Airlines and Delta Air Lines was originally to include
code-sharing, but it has been scaled back to an arrangement involving
reciprocal frequent flyer programs and access to airport lounges.
This arrangement, which the airlines began implementing in September
1998, is much the same as the one American Airlines and US Airways
proposed and began implementing in August 1998.
The alliances may have both beneficial and harmful effects on
consumers. And because they differ in scope, their possible effects
vary. Officials from Northwest and Continental said that their
alliance will benefit consumers through expanded route networks, more
frequency options (that is, more flights on the same routes),
improved connections, and enhanced frequent flyer programs. Our
analysis showed that the alliance could result in new, possibly
improved, route options, and the alliance's extended frequent flyer
program may benefit members of each airline's program. We also found
that this alliance will create some "new" markets that are not
already served by other airlines. However, our analysis indicated
fewer new markets than the alliance partners estimated, and it showed
that these new markets will serve relatively few passengers. On the
other hand, consumers would be harmed if competition is reduced. But
it is difficult to determine whether the partners in the alliance
will continue to compete or whether the alliance will encourage them
to act in a manner that may reduce competition. Airline officials
have said that the partners will continue to compete. However,
industry experts have raised concerns that competition will likely
decline over time as firms recognize their interdependence and
maintain prices above the competitive level. Our analysis indicates
that if Northwest and Continental do not act independently,
competition could decline in 63 markets that served 2 million
passengers in 1997, and the two airlines could also increase by 5
percent the number of markets that they dominate. According to
industry experts, airlines that achieve dominant market positions can
drive out competitors with smaller shares and eventually raise fares.
In addition, we recently reported that certain airline marketing
practices, such as frequent flyer programs--a feature common to all
of the alliances--can make competitive entry more difficult for other
airlines, especially in markets where one airline has a substantial
share of the market, thus possibly limiting the benefits of
deregulation in the airline industry. However, we have not been able
to quantify the effects on competition that such practices would
exert.
The departments of Justice and Transportation are separately
reviewing the three alliances under different statutory authorities
and have different remedies available to them. On October 23, 1998,
Justice filed a civil antitrust action to prevent Northwest from
acquiring or holding a majority of Continental's voting stock.
Justice said in its complaint that Northwest's gaining control would
lessen competition in interstate trade and commerce and unreasonably
restrain trade. Justice believed that the alliance would
substantially diminish both airlines' incentives to compete against
each other and would cause consumers to pay higher prices and receive
lower-quality service in some markets. Justice will review other
aspects of the Northwest-Continental alliance, as well as the other
two alliances, using guidelines that are applied to traditional
mergers, but it is under no timetable for these reviews. The
Congress recently authorized Transportation to impose waiting periods
before certain joint venture arrangements involving major airlines,
such as frequent flyer programs, can be effective. Transportation
imposed a waiting period on the frequent flyer and code-sharing
aspects of the Northwest-Continental alliance under this new
authority but eventually agreed that the airlines could proceed with
their frequent flyer program. Transportation officials say that they
have received information about the United-Delta and American-US
Airways alliances' frequent flyer agreements, which are already in
effect. Transportation will request additional information if it
decides the information received is not sufficient and if the
airlines propose to extend their alliances to include code-sharing.
Transportation also has the authority to prohibit unfair methods of
competition in the airline industry, which it can use in reviewing
alliances after they have been implemented.
BACKGROUND
------------------------------------------------------------ Letter :2
Two or more airlines may enter into an alliance to increase their
revenues and the number of passengers they carry. Code-sharing is
one type of alliance. It can be an important marketing tool for
airlines because it allows the code-sharing partners to replicate the
"seamless travel" that can be provided by a single airline, known as
"on-line" service. Airline passengers prefer this type of service
because it allows the convenience of single ticketing and check-in,
among other things.\3 For a code-sharing flight, each partner sets
its own fare for the entire trip, covering both the segment it flies
and the one its partner flies.
In recent years, alliances among airlines have become very common in
international aviation because they allow airlines to enter markets
that would be (1) too expensive to serve with their own aircraft or
(2) restricted under a bilateral aviation agreement with another
nation.\4 The Department of Transportation (DOT) reported that there
were 74 active alliances between U.S. and foreign carriers as of
June 1998. In the simplest case, an international code-sharing
alliance links the route network of one airline with the route
network of another, forming an end-to-end alliance with little
overlap. Figure 1 shows how such an alliance links two airlines, one
with an extensive route network in the United States and the other
with an extensive route network in Europe and Africa.
Figure 1: Illustration of a
Hypothetical End-to-End
International Alliance
(See figure in printed
edition.)
In our previous work, we found that alliances between U.S. and
foreign airlines that involved code-sharing in a large number of
markets did benefit the alliance partners. The alliance generated
large gains for the partners in terms of passengers and revenues,
mainly at the expense of other airlines.\5 We also found that
although consumers benefited from conveniences such as shorter
layovers, the data were insufficient to determine the effects of the
alliances on fares in the short term and on competition and fares in
the long term.
Domestically, code-sharing alliances have generally occurred between
major U.S. airlines and regional commuter airlines that transport
passengers, usually from smaller communities to the cities served by
the major carriers. Similar to international alliances, these
alliances generally link end-to-end networks without creating a
significant overlap in service. For example, United Airlines has a
code-sharing agreement with Atlantic Coast Airlines to bring
passengers into its hub at Washington Dulles Airport. Additionally,
several major U.S. airlines have used code-sharing in a limited
number of markets. For example, in 1994, Continental and America
West airlines entered into a limited code-sharing agreement, and in
1996, Northwest and Alaska Airlines also entered into a limited
code-sharing agreement.
The federal government has limited authority over proposed airline
alliances. In the international sector, the routes that airlines can
fly, the frequency of their flights, and the fares they can charge
are governed by 72 bilateral agreements between the United States and
other countries. Many of these agreements are very restrictive.
Since the late 1970s, U.S. policy has been to negotiate agreements
that substantially reduce or eliminate bilateral restrictions ("open
skies" agreements). In the domestic sector, however, the Airline
Deregulation Act of 1978 generally eliminated the government's
authority over airline routes, frequencies, and pricing. Under new
legislation passed in October 1998, DOT has the authority to impose
an initial 30-day waiting period, which it may extend another 150
days for joint ventures involving code-sharing between major
airlines.\6 This authority does not limit the Department of Justice's
(DOJ) authority to enforce the antitrust laws. DOT may also
investigate whether an alliance is an unfair method of competition.
--------------------
\3 Similar conveniences can be obtained between two airlines that
have certain agreements, commonly referred to as "interline
agreements." Interline agreements provide for the mutual acceptance
by the participating airlines of passenger tickets, baggage checks,
and cargo waybills, as well as establish uniform procedures in these
areas. These agreements are common, but not universal, among the
major U.S. airlines. All major U.S. airlines except for Southwest
have interline agreements. According to the Department of
Transportation, there are important differences between code-sharing
and interline agreements. For example, interline agreements
typically do not include reciprocal frequent flyer and airport lounge
rights, and airlines will generally not hold outgoing connecting
flights to wait for delayed incoming flights. Most importantly,
however, fares for code-sharing flights are generally much cheaper
than for interline flights. See pp. 19-20 for additional
information on the relative prices of interline and code-sharing
flights.
\4 According to United, its decision to enter a particular
international market tends to be revenue-based and considers such
factors as the airline's overall network size and presence at that
city. Certain markets may simply be too small to support
head-to-head alliance competition.
\5 International Aviation: Airline Alliances Produce Benefits, but
Effect on Competition Is Uncertain (GAO/RCED-95-99, Apr. 6, 1995).
\6 P. L. 105-277, sec. 110(f). Generally, the joint venture
agreements subject to this waiting period include only those that
were entered into by a major airline after Jan. 1, 1998, and involve
code-sharing, certain leasing arrangements, frequent flyer programs,
and other cooperative working arrangements designated by DOT.
STATUS OF THE ALLIANCES: ALL
THREE ALLIANCES HAVE BEEN
INITIATED
------------------------------------------------------------ Letter :3
Since we testified in June 1998, each of the three domestic alliances
has begun to implement at least part of its agreement. On November
20, 1998, Northwest and Continental announced that Northwest had
completed its acquisition of the equity position in Continental. The
two airlines also announced that they had revised their agreement and
would implement the marketing aspects of their alliance in December
1998. United and Delta and American and US Airways have begun to
implement their alliances, although the United-Delta proposal no
longer includes code-sharing. Like the American-US Airways proposal,
it is now limited to offering consumers reciprocal frequent flyer and
airport lounge benefits. The alliance between Northwest and
Continental airlines remains under review at both DOJ and DOT, which
has requested additional information from each of the six airlines on
the frequent flyer arrangements with its respective alliance partner.
NORTHWEST AND CONTINENTAL
HAVE BEGUN TO IMPLEMENT
THEIR ALLIANCE AGREEMENT
---------------------------------------------------------- Letter :3.1
On November 20, 1998, Northwest and Continental announced that
Northwest had completed the acquisition of 8.7 million shares of
Continental's stock, which it then deposited in a voting trust.\7 The
agreement announced on that date is somewhat different from that
originally announced by the airlines in January 1998. According to
the airlines, under the new terms of the agreement between them,
Northwest will acquire less than a majority of the voting control of
Continental,\8 forgo its right to place someone on Continental's
board of directors, and forgo code-sharing with Continental in
certain hub-to-hub domestic markets. Continental said that to
protect its stockholders, its board has adopted a shareholder rights
plan that will become effective if 15 percent or more of its voting
stock is acquired by a single investor.
Nevertheless, the closing of the stock acquisition represented a
major step toward implementing the "strategic global alliance" that
the airlines announced in January 1998 to connect their route
systems. The two airlines announced that they would accept
code-sharing bookings starting December 12, 1998, with code-sharing
flights to Japan beginning December 29, 1998, and to other
destinations beginning January 7, 1999. The code-sharing plan
ultimately will include the airlines' international code-sharing
partners, such as Air China\9 and KLM Royal Dutch Airlines.\10 Each
airline has a separate code-sharing agreement with America West\11
and may independently pursue other domestic alliances.\12
The alliance will also include the airlines' regional partners,
Northwest Airlink,\13 in which Northwest holds some equity, and
Continental Express, which is wholly owned by Continental. As part
of the agreement, the airlines will also undertake other cooperative
activities, including marketing and coordinating their flight
schedules to improve connection times. Airline executives stated
that they will not coordinate pricing or capacity, and they submitted
their alliance proposal to DOJ under the Hart-Scott-Rodino Antitrust
Improvements Act of 1976\14 (Hart-Scott-Rodino Act) as if it were a
full merger. Northwest and Continental invited DOJ to review the
transaction under its stringent merger guidelines.
In addition, the airlines established reciprocal frequent flyer
programs, allowing members to earn and redeem their miles on either
carrier. Beginning December 6, 1998, members of each airline's
frequent flyer programs can earn miles for travel on the other
airline's flights and can begin redeeming miles on February 1, 1999,
for travel beginning on March 1, 1999. Both airlines announced
several changes to their frequent flyer programs, as well. For
example, members will be able to redeem mileage at reduced levels and
travel during off-peak times. At other award levels, the airlines
will have fewer blackout dates (when frequent flyer awards may not be
used). On November 21, 1998, each airline opened its club facilities
to members of the other airline's clubs, thereby expanding the number
of airports where members can use clubs.
Northwest's principal service areas are the Midwest and Midsouth,
while Continental's are the Northeast and Southwest. Figure 2 shows
the percentage share of departing passengers that the
Northwest-Continental alliance, had it been in effect in 1997, would
have carried. It indicates that the Northwest-Continental alliance
would carry large percentages of passengers (i.e., would have
relatively large market strength) in the upper Midwest, South, and
Northeast. According to the airlines, these were essentially
preexisting market shares of Northwest and Continental, respectively,
and the market shares in these regions did not increase materially as
a result of the alliance.
Figure 2: Percentage of Each
State's Departing Passengers
That the Northwest-Continental
Alliance Would Have Carried
(See figure in printed
edition.)
Note: This figure is based on
the alliance's share of the
number of departing passengers
carried by all major airlines,
by state, during 1997. In
1997, the major airlines
carried 85.8 percent of all
passengers enplaned in the
United States.
(See figure in printed
edition.)
For the purposes of this and
subsequent figures, we
classified passengers
originating from the Greater
Cincinnati International
Airport as originating from
Ohio, rather than Kentucky
(where the airport is located),
and those originating from
Washington Reagan National
Airport as originating from
Virginia, rather than the
District of Columbia or
Maryland. We did so because
information from the airports
indicated that more of their
passengers came from these
states than from the other
jurisdictions.
(See figure in printed
edition.)
Source: GAO's analysis of
information provided by Data
Base Products, Inc.
(See figure in printed
edition.)
Under the alliance, the two airlines combined carry approximately 15
percent of the total domestic passenger market. Such a combination
effectively creates the second largest domestic market share (in
terms of the number of passengers carried), behind Delta's 18 percent
and ahead of United's 13 percent. (See app. I for more detailed
information on the domestic market shares of U.S. airlines.)
--------------------
\7 According to Northwest and Continental officials, the voting trust
means that Northwest's shares will generally be voted in proportion
to the votes of the non-Northwest shareholders. According to airline
officials, except in exceptional circumstances, the outcome of a vote
will not be affected. The voting trust ends after 6 years, and
Northwest has agreed on significant voting restrictions for 4 years
thereafter. Northwest has also agreed to vote for a majority of
independent directors on Continental's board. After 10 years,
Northwest can exercise the full power of its ownership.
\8 Northwest's equity purchase equates to approximately 46 percent of
the fully diluted voting rights rather than 51 percent as originally
proposed. (Fully diluted voting rights are those adjusted for the
conversion into common stock of all convertible securities.) Under
the terms of the November agreement, certain partners of the original
holders of that stock retained approximately 5 percent of the total
stock but granted Northwest a "limited proxy" to vote these shares.
According to a Continental official, this means that these shares may
be voted generally in accordance with Northwest's wishes only in
extraordinary circumstances (e.g., matters that would materially
affect Northwest's ownership position, such as potential mergers,
recapitalizations, or liquidations) or in circumstances where
Northwest votes for directors in accordance with the recommendation
of Continental's board, in which limited cases they would be voted as
directed by Northwest. DOJ filed an amended complaint on Dec. 18,
1998, to reflect Northwest's consummation of its stock purchase
agreement and changes made to various agreements by the parties
involved in the transaction. The amended complaint alleges that
Northwest would still own more than 50 percent of the fully diluted
voting power over Continental.
\9 Northwest and Air China, the largest airline in China, announced
their code-sharing arrangement in May 1998, including Alaska
Airlines, Continental, and America West. The arrangement gives
Alaska and America West their first Asian code-sharing partner,
Continental its first access to mainland China, and Northwest
improved access to China.
\10 In Jan. 1993, DOT granted antitrust immunity to the
Northwest-KLM alliance in conjunction with the U.S.-Netherlands open
skies accord.
\11 America West and Continental have had a relatively limited
code-sharing agreement since 1994. The agreement has grown from 45
to 72 route segments, allowing Continental to place its code on
America West's flights west of Phoenix and Las Vegas and allowing
America West to place its codes on Continental's flights east of
Houston, Cleveland, and Newark.
\12 Alaska Air Group, Inc., owners of Alaska Airlines and Horizon
Airlines, has a code-sharing agreement with Northwest and recently
agreed to one with KLM. The Alaska--Northwest code-sharing agreement
covers all of Northwest's markets and Alaska's feeder markets (routes
into Seattle and Los Angeles through Horizon Airlines, a wholly owned
subsidiary of Alaska Air Group, Inc., which also owns Alaska
Airlines). These routes serve to funnel Alaska's passenger traffic
on to Northwest's trans-Pacific, transcontinental, and Midwest
flights.
\13 The Northwest Airlink carriers are Mesaba Airlines and Express
Airlines I. Code-sharing with other regional partners will have to
be negotiated separately.
\14 15 U.S.C. � 18a.
UNITED AND DELTA HAVE BEGUN
IMPLEMENTING A MORE LIMITED
ALLIANCE THAN ORIGINALLY
PROPOSED
---------------------------------------------------------- Letter :3.2
In April 1998, 3 months after Northwest and Continental proposed
their alliance and investment agreements, United and Delta announced
their plan to form a global alliance. This alliance would have
linked the two largest domestic airlines through code-sharing and
reciprocal frequent flyer programs. Moreover, as envisioned, it
would have produced a larger combined market share than either of the
other alliances--almost 31 percent of domestic passengers. It would
have joined United's extensive route networks in the West and Midwest
with Delta's similarly extensive networks in the East, Southeast, and
Southwest. Figure 3 shows the percentage share of departing
passengers that the United-Delta alliance, had it been in effect in
1997, would have carried. It indicates that the alliance, as
originally proposed, would have had market strength virtually
nationwide.
Figure 3: Percentage of Each
State's Departing Passengers
That the Originally Proposed
United-Delta Alliance Would
Have Carried
(See figure in printed
edition.)
Note: Figure 3 is based on the
alliance's share of the number
of departing passengers carried
by all major airlines, by
state, during 1997. In 1997,
the major airlines carried 85.8
percent of all passengers
enplaned in the United States.
(See figure in printed
edition.)
Source: GAO's analysis of
information provided by Data
Base Products, Inc.
(See figure in printed
edition.)
On September 1, 1998, however, Delta and United announced that they
had discontinued discussions concerning code-sharing arrangements.\15
The two airlines implemented their reciprocal frequent flyer programs
on September 1, 1998, allowing passengers who fly on either airline
to choose the program where they accrue and redeem their miles.
--------------------
\15 On Aug. 17, 1998, Delta's board said that it would not grant an
Air Line Pilots Association proposal to convert its existing
nonvoting seat on the board to full voting status.
AMERICAN-US AIRWAYS BEGAN
IMPLEMENTING THEIR ALLIANCE
IN AUGUST
---------------------------------------------------------- Letter :3.3
Also in April 1998, 3 months after Northwest and Continental
announced their alliance, American Airlines and US Airways announced
that they had agreed on a limited marketing relationship involving
their frequent flyer programs and club facilities. The airlines
began implementing the frequent flyer agreement in August 1998. The
airlines further announced a special methodology for allocating costs
between the two airlines on interline flights. According to US
Airways, the methodology allows each airline to recover its costs for
the segment it flies and to divide revenues in the same proportion.
Airline officials have said that they may consider some very limited
code-sharing with their regional partners, American Eagle and US
Airways Express.\16
American and US Airways offer greater market presence in different
parts of the country. US Airways has a strong presence in the
Northeast, Mid-Atlantic, and Southeast, while American's network
extends across much of the rest of the United States.\17 Should this
arrangement move beyond reciprocal frequent flyer programs to
code-sharing, the alliance's market share would be about 22 percent
of total domestic passengers. Figure 4 shows the percentage share of
departing passengers that the American-US Airways alliance, had it
been in effect in 1997 and extended to include code-sharing, would
have carried. It indicates that such an alliance would have had
market strength mostly on the East Coast.
Figure 4: Percentage of Each
State's Departing Passengers
That a Code-Sharing Alliance
Between American and US Airways
Would Have Carried
(See figure in printed
edition.)
Note: Figure 4 is based on the
alliance's share of the number
of departing passengers carried
by all major airlines, by
state, during 1997. In 1997,
the major airlines carried 85.8
percent of all passengers
enplaned in the United States.
This figure also includes Reno
Air's share of departures in
the markets served by Reno in
1997.
(See figure in printed
edition.)
Source: GAO's analysis of
information provided by Data
Base Products, Inc.
(See figure in printed
edition.)
--------------------
\16 According to the airlines, there has been some discussion of such
code-sharing, but certain labor and commercial issues remain to be
resolved. Senior management of US Airways has stated that it does
not favor domestic code-sharing between the mainline divisions of US
Airways and American but would consider such code-sharing as a
competitive response to actions of other airlines and alliances. As
of the end of Dec. 1998, the two airlines had not discussed mainline
code-sharing.
\17 On Nov. 19, 1998, American Airlines announced that it had agreed
to acquire Reno Air for $124 million. Reno's 16-city route system
extends from Oklahoma City to Anchorage and includes service to
Tucson, San Diego, Los Angeles, San Francisco, Portland, and Seattle.
It also serves San Jose, subleasing the gates formerly operated by
American. This purchase will strengthen American's north-south route
system on the West Coast. On Dec. 9, 1998, Alaska Air Group, Inc.,
announced that it had signed a letter of intent for the creation of a
marketing partnership between its subsidiaries, Alaska Airlines and
Horizon Air, and American Airlines--American Eagle. Alaska and
Horizon intend to implement fully reciprocal frequent flyer
relationships with American and American Eagle, allowing customers to
earn and use mileage awards across each other's networks.
Code-sharing by the airlines has also been discussed but is subject
to labor contract provisions.
ALLIANCES COULD HAVE BENEFICIAL
AND HARMFUL EFFECTS ON
CONSUMERS
------------------------------------------------------------ Letter :4
The alliances may have beneficial and harmful effects on consumers.
Because the alliances currently differ in scope, their potential
benefits range from increased on-line service and more efficient
routings to enhanced frequent flyer programs and access to more club
facilities. Harmful effects will occur if competition is reduced.
Alliance airline officials have stated that they will act
independently. On the other hand, industry experts have said that
they think competition between alliance partners will decline over
time. It is difficult to determine what will happen in the future,
but some experts' concerns are consistent with widely held economic
principles and past experience in the airline industry. With
code-sharing, the opportunity for harm increases with the number of
overlapping markets and the number of passengers flying in these
markets. Thus, an end-to-end alliance with few overlapping routes
that serves fairly small markets is less likely to threaten
competition than an alliance between partners that share a number of
heavily traveled markets. In addition, under enhanced frequent flyer
programs, it may be more difficult for would-be competitors to enter
markets where the alliance partners already have a substantial market
share, thus further limiting competition and the potential benefits
of deregulation in the airline industry.
This section examines the potential beneficial and harmful effects on
consumers of the arrangement between Northwest and Continental--the
only one with plans for equity acquisition and code-sharing as of
December 1998 (see app. II for a detailed description of our scope
and methodology). However, the other alliances may introduce
code-sharing if the Northwest-Continental proposal is implemented.
Consequently, we analyzed the potential effects of the United-Delta
alliance (see app. III) and of the American-US Airways alliance (see
app. IV) using the same approach and framework that we applied to
the Northwest-Continental alliance. We also analyzed the potential
effects, on consumers and the industry as a whole, of all three
alliances' implementing code-sharing (see app. V).
NORTHWEST-CONTINENTAL
ALLIANCE COULD HAVE BOTH
BENEFICIAL AND HARMFUL
EFFECTS ON CONSUMERS
---------------------------------------------------------- Letter :4.1
Because the Northwest-Continental alliance involves code-sharing, it
could benefit consumers in several ways. First, it will create new
on-line destinations, even though no aircraft or flights are added.
In addition, it will increase the number of flights in a market
attributable to a single airline (flight frequencies) and provide
better connections so that consumers can reach their destinations
sooner. Finally, the alliance will give consumers more frequent
flyer award destinations and access to more club facilities. At the
same time, however, the alliance creates the possibility of harm for
air travelers, especially over the longer term. If competition is
reduced, consumers would be harmed. Specifically, should the
airlines compete less vigorously or act as one entity if Northwest
acquires a majority interest in Continental, then consumers could be
harmed through the decrease in competition, especially on routes
where both airlines previously competed. Our analysis of the
potential beneficial and harmful effects of the alliance between
Northwest and Continental follows.
ALLIANCE COULD BENEFIT
CONSUMERS BY PROVIDING
MORE ON-LINE DESTINATIONS
-------------------------------------------------------- Letter :4.1.1
According to Northwest and Continental officials, their alliance will
create new on-line service to 2,007 new domestic and international
city pairs, thereby benefiting the approximately 3.4 million
passengers who flew on these routes in 1997.\18 One airline official
believes that the new on-line service will divert passengers from
other airlines that provide poorer connections and will stimulate new
flying by passengers who previously chose not to fly. Because the
alliance will convert passenger routings from interline connections
to on-line connections, the official also expects the alliance
partners to offer lower fares in some markets that previously could
be served only on an interline basis.\19 In turn, such lower fares
will also stimulate travel on these routes.
Because this study is limited to domestic airline alliances, we did
not analyze the impact of the alliance on Northwest's and
Continental's international markets. We did, however, analyze its
impact on both airlines' domestic markets and included in our review
new markets that would be created through routes with one or two
connections. Although the airlines also included new city pairs that
would be created through routings that require three connections, we
did not include these markets because we believe that the number of
passengers who would book flights with so many connections is
relatively small.\20
Our analysis showed that a code-sharing alliance between Northwest
and Continental would serve considerably fewer new airport pairs than
the airlines estimated. We found, first, that such an alliance could
create on-line single-connection service in 74 markets that served
about 400,000 passengers in 1997 (or about 15 per day per route). In
22 of these markets, the Northwest-Continental alliance would produce
superior service to that already offered by a competitor. In the
other 52 markets, competing airlines' existing service would be
superior or at least equal to that anticipated under the alliance.
At the same time, however, we found that the alliance would create no
new service in any of these single-connection markets because all 74
were already served either by one of the alliance partners or by a
competitor. We also found that a Northwest-Continental alliance
could create 71 new double-connection markets. The
Northwest-Continental alliance would provide superior service to that
offered by competitors in two markets, but other competing airlines
already provided superior or equal service in the other 69 markets.
Table 1 provides our detailed analysis of the new on-line benefits
that consumers could expect from this alliance.
Table 1
GAO's Analysis of Airport Pairs That
Would Receive New On-Line Service
Through Single or Double Connections
Under a Northwest-Continental Alliance
Total markets and average
number of passengers per
day who would receive new
Northwest-Continental
service
--------------------------
Number of Number of Number of
markets in markets in markets in
Average which which which
number of competing competing competing
passengers airlines airlines airlines
per day per provide provide provide
Type of new on- Number of market in superior\ equal\ inferior\
line service markets 1997 service\a service\b service\c
------------------- ------------ ------------ ------------ ------------ ------------
Single-connection 74 15 10 42 22
Double-connection 71 6 54 15 2
-----------------------------------------------------------------------------------------
\a Superior service includes current direct and nonstop service by
competitors between airports that Northwest-Continental could serve
at best with single-connection service; and direct, nonstop, or
single-connection service by competitors between airports that
Northwest-Continental could serve at best with double-connection
service. Direct service differs from nonstop service in that a
"direct" flight makes a scheduled stop between an origin and a
destination, but passengers flying between the origin and destination
are not required to change planes, while a nonstop flight between an
origin and a destination makes no scheduled stops en route.
\b Equal service means that competing airlines currently offer
service that matches the Northwest-Continental alliance's potential
single- or double-connection service.
\c Inferior service means that competing airlines currently offer (1)
no service or double-connection service between airports where the
Northwest-Continental alliance could offer single-connection service
or (2) no service between airports where the alliance could provide
double-connection service.
Source: GAO's analysis of information provided by BACK Associates
and Data Base Products, Inc.
--------------------
\18 According to an economic consultant for Northwest, these cities
are mainly, but not exclusively, located in the United States.
Included are cities served by Northwest and Continental in North and
Central America, but not those in Europe, Asia, or South America.
\19 According to airline officials, interline fares are most
frequently the sum of the fares for each airline's segment of the
itinerary. Even when tickets are purchased well in advance to take
advantage of possible airline discounting, such fares tend to be
considerably higher than on-line fares from the same origin to the
same destination. Newly created on-line fares should be less
expensive than these interline fares because each airline in an
alliance now has an incentive to compete on fares to attract
passengers to its own airline.
\20 We also limited our analysis of this and the other alliances in
other ways. For example, using the airlines' published schedule for
May 1998, we required arriving and departing flights to be scheduled
within 30 and 150 minutes to qualify as potentially valid
connections. For double connections, we limited our analysis to
flights that connected through the partners' hubs. For additional
information on our methodology, see app. II.
ALLIANCE COULD BENEFIT
CONSUMERS BY PROVIDING
MORE FREQUENCIES AND
IMPROVED ROUTINGS
-------------------------------------------------------- Letter :4.1.2
According to Northwest and Continental officials, the alliance will
provide 17,446 new on-line flight opportunities in 10,459 markets
that, they say, will create additional routings in existing markets,
increasing convenience and choice for their passengers and reducing
travel times for an estimated 250,000 passengers. For example,
Northwest and Continental say that by routing passengers from
Milwaukee to Honolulu through San Francisco instead of Seattle, the
alliance will shorten these passengers' connection times by about 1.5
hours.
Our limited analysis of flight frequencies suggests that the alliance
could result in new, possibly improved, route options. For various
technical reasons, we were unable to duplicate the model that
Northwest and Continental used to calculate the number of new routing
possibilities. We note, however, that their earlier modeling of
benefits included destinations other than those in the United States,
causing them to overestimate the new domestic routing possibilities.
ALLIANCE COULD BENEFIT
MEMBERS OF THE AIRLINES'
FREQUENT FLYER PROGRAMS
AND CLUBS
-------------------------------------------------------- Letter :4.1.3
Although there is some uncertainty about the extent of the service
and routing benefits under the Northwest-Continental alliance, the
agreement to offer reciprocal access to the airlines' frequent flyer
programs and club facilities could provide direct benefits to
customers of both airlines. Passengers may choose to participate in
one or both of the frequent flyer programs (either Northwest's
"WorldPerks" program or Continental's "OnePass" program), but miles
earned on any given alliance flight can be awarded to only one
program. Travelers will not be able to combine miles from both
programs to achieve an award. This reciprocal relationship should
increase the value of these programs to frequent flyers because more
destinations and frequencies will be available. Northwest and
Continental do not expect to significantly change how they distribute
frequent flyer seats, which may be different on each flight because
of various market forces. According to the airlines, club members
will have access to facilities at all the airports where the alliance
partners have club facilities starting November 21, 1998.
Members of each airline's frequent flyer program could benefit from
the alliance's ability to offer new destinations. When announcing
the consummation of their alliance, both airlines announced
enhancements to their programs. Northwest noted that its WorldPerks
members will still be able to redeem free travel awards for as few as
20,000 miles within North America and that these awards will be
available for travel during 9 months of the year instead of just 10
weeks. Northwest also announced a number of changes to its frequent
flyer program, including fewer blackout dates (when frequent flyer
awards cannot be used) for WorldPerks travel within North America.
Continental announced that travel during off-peak times will be
available at reduced mileage levels starting March 1, 1999.
POTENTIAL HARMFUL EFFECTS
OF ALLIANCE DEPEND ON
WHETHER PARTNERS COMPETE
LESS VIGOROUSLY
-------------------------------------------------------- Letter :4.1.4
According to Northwest and Continental, their alliance is unlikely to
harm consumers because it is an end-to-end alliance with relatively
few overlapping markets. Nevertheless, the alliance could harm
consumers if it decreases competition in markets that are already
served by both partners,\21 especially in those where the alliance
gives the partners a "dominant" market position (i.e., control of
more than 50 percent of the market). Such an increase in market
dominance is significant, according to industry analysts, who
predicted that the partners might not be able to gain much market
share overall but might be able to increase revenues--presumably by
raising fares--in individual markets where they held a dominant
position.\22 The Northwest-Continental alliance could also increase
barriers for other airlines that might want to enter particular
markets. To examine the potential harm from the alliance, we
determined the extent to which each airline's routes overlapped with
those of its alliance partner by analyzing 1997 data on the 5,000
busiest domestic airport-pair origin and destination markets or
markets for air travel between two airports. These 5,000 markets
accounted for over 90 percent of the total U.S. domestic air traffic
in 1997.\23
--------------------
\21 DOJ's lawsuit against the investment agreement between Northwest
and Continental states that the acquisition would substantially
diminish the incentives for the two airlines to compete against each
other.
\22 We have also reported in the past that airfares in dominated
markets tend to be higher than in other markets. See, for example,
Airline Competition: Effects of Airline Concentration and Barriers
to Entry on Airfares (GAO/RCED-91-101, Apr. 26, 1991).
\23 The following example indicates how large these markets are: The
smallest of the top 5,000 were between Jacksonville, Florida, and
Newport News/Williamsburg, Virginia, and between Burlington, Vermont,
and Cleveland, Ohio. In each of these markets, 7,964 passengers (an
average of 22 each day) flew in 1997.
EXPERTS SUGGEST PARTNERS
MAY NOT COMPETE
VIGOROUSLY, BUT AIRLINES
SAY THEY WILL
-------------------------------------------------------- Letter :4.1.5
The potential for the Northwest-Continental alliance to harm
consumers depends on the features of the agreement and on whether it
creates an environment that discourages competition between the two
partners. Several factors could influence whether the partners
cooperate rather than compete as independent entities. These factors
include how much ownership one partner has in the other, how revenue
from jointly operated flights is shared, whether the airlines'
pricing practices change, and whether the airlines coordinate their
schedules, especially if such coordination reduces the total number
of flights or available seats in a market. Some of this information
is contained in proprietary documents provided by Northwest and
Continental to DOJ and DOT. Northwest and Continental declined to
let us review these confidential documents, but we discussed various
elements of the agreement with airline officials and industry
experts.\24
All but one of the industry experts whom we interviewed agreed that
an investment such as Northwest's in Continental would be likely to
affect the behavior of the airlines over time, encouraging them to
cooperate in order to maximize the value of the investment to their
stockholders. Some of these industry experts believed that the
alliance partners would not vigorously compete with each other,
particularly over the long term. If they are correct, then the
alliance could have harmful effects on consumers. In contrast,
officials from Northwest and Continental insist that while their
alliance arrangement is in many ways like a merger, the airlines will
continue to operate independently. They said that although Northwest
will own the largest percentage of the voting stock in Continental,
it will not own a majority of the voting stock and its voting trust
agreement with Continental is such that it effectively has no ability
to affect or influence the control or management of Continental for
the next 10 years.\25
Industry experts expressed concern that this alliance could create an
environment in which the airlines would compete less vigorously even
if there were no change in equity between the alliance partners. One
industry expert we spoke with said that in the airline industry,
where there are only a few firms and where rivals' fare and schedule
information is widely available, tacit coordination is relatively
easy. Some industry experts pointed out, for example, that the two
airlines could cooperate on some aspects of their business. As a
result, according to this expert, Northwest and Continental might not
compete aggressively on fares or schedules, fares could rise, and
service levels could eventually decline. In contrast, Continental
and Northwest officials said that they will continue to set fares and
schedules independently and that competition will not decline because
of the alliance. For example, on nonstop routes that both airlines
serve, Continental will receive a booking fee that is similar to a
travel agent's fee (an amount equal to Continental's costs) for
ticketing a passenger on a Northwest flight, but none of the revenue,
and vice versa. As a result, airline officials say, each airline
will have an incentive to book passengers on its own airplanes. When
a passenger's travel involves segments served by both airlines,
revenues will be split between the two airlines on the basis of a
prorated schedule that is standard in the industry. When announcing
the stock acquisition, Northwest and Continental also said that they
would not implement code-sharing on the seven hub-to-hub routes that
they both serve. For example, for flights from New York City to
Detroit, the two would continue to compete directly.
It is difficult to determine precisely how the alliance will affect
competition, but the industry experts' concerns and the airlines'
past records establish cause for concern. As discussed, there is
widespread agreement among these experts that competition will likely
decline over time as firms recognize their interdependence and
maintain prices above the competitive level. This understanding is
consistent with widely held economic principles showing that
competition lessens when there are fewer competitors. It is also
consistent with a recent DOT paper containing evidence that, when
given the chance, airlines do not compete vigorously.\26 According to
DOT, prices on routes of less than 750 miles were generally not
competitive unless a low-fare carrier, such as Southwest, served the
route and kept prices closer to a competitive level. On routes not
served by a low-cost carrier, major airlines have recognized their
interdependence and kept prices above the competitive level. If the
ties between major airlines are strengthened, the airlines will have
even more opportunities to recognize their interdependence. If this
occurs, competition may suffer and prices may rise.
--------------------
\24 These industry experts included several financial analysts in New
York investment firms and brokerage services, nationally recognized
academicians, and other individuals with expertise in specific
aspects of the industry, such as computer reservation systems.
\25 According to Continental officials, the governance agreement
between the two airlines deprives Northwest of substantially all
voting power over the Continental securities it acquired, severely
limits Northwest's ability to acquire additional voting securities,
and prohibits Northwest from seeking to affect or influence
Continental's business. As a result, they say, Northwest effectively
has no ability to affect or influence the control or management of
Continental. In its antitrust complaint against Northwest and
Continental, DOJ disagreed with this conclusion. DOJ said that ". .
. the governance agreement does not prevent the harm likely to
result from Northwest's acquisition of voting control of Continental.
No private agreement can alter the fact that Northwest still owns
Continental, and Continental will not compete vigorously with its
owner during the term of the governance agreement." After DOJ filed
its complaint, Northwest and Continental announced a revision to the
acquisition agreement that provides Northwest with less than half--46
percent--of the voting control of Continental. Certain partners of
the original seller of the majority control of Continental, Air
Partners, L. P., retained ownership of 853,644 shares of
Continental's stock, or about 5 percent, and granted Northwest a
limited proxy for the voting of these shares, thereby limiting the
ability of Northwest to exercise strict voting control over
Continental. On Dec. 18, 1998, DOJ filed an amended complaint
reflecting the consummation of Northwest's stock purchase agreement
and changes made to various agreements by the parties involved in the
transaction. The amended complaint alleges that Northwest would
still own more than 50 percent of the fully diluted voting power over
Continental. See pp. 37 to 39 for additional information.
\26 See Competition in the U.S. Domestic Airline Industry: The Need
for a Policy to Prevent Unfair Practices, DOT (July 1998).
NORTHWEST-CONTINENTAL
ALLIANCE MAY DECREASE
COMPETITION IN CERTAIN
MARKETS AND COULD
POTENTIALLY HARM
CONSUMERS
-------------------------------------------------------- Letter :4.1.6
If the partners in the Northwest-Continental alliance do not
vigorously compete over time, the harm to consumers will depend on
the number of markets where the partners provide overlapping service
(including the number of passengers served), the alliance's share\27
of these markets, and the effects of the alliance on barriers to
entry.
The Alliance Could Reduce Competition in Shared Markets. We found
that a Northwest-Continental alliance could reduce or eliminate
competition in 63 of the top 5,000 markets. Nearly 2 million
passengers traveled in these markets in 1997. The alliance could
effectively eliminate competition in five markets, with about 36
percent of these passengers (723,186 in 1997), by reducing the number
of competing airlines from two to one. The largest of these markets
is Detroit to Newark, where the alliance partners served 97 percent
of the market in 1997--428,574 passengers. In 24 markets, the
alliance could reduce the number of competitors from three to two,
and it could become the largest carrier in nearly two-thirds of these
markets. The markets where the alliance could decrease the number of
competitors to three or fewer served approximately 1.6 million
passengers in 1997. These are the markets that could suffer the
greatest potential harm from the alliance.\28 Figure 5 illustrates
the number of markets and passengers with the potential to be harmed
under a Northwest-Continental alliance if the partners do not price
their fares independently.
Figure 5: Number of Markets
and Passengers Subject to
Potentially Decreased
Competition Under the
Northwest-Continental Alliance
(See figure in printed
edition.)
Note: So that readers can more easily compare the number of markets
and passengers potentially affected by the Northwest-Continental
alliance with the number of those potentially affected by the two
other alliances and by all three alliances, should they eventually
move to code-sharing (see figs. III.1, IV.1, and V.1), we are using
the same scales for all figures.
Source: GAO's analysis of data provided by Data Base Products, Inc.,
on the top 5,000 origin and destination markets in 1997.
Northwest and Continental have attempted to address DOJ's concerns
about competition. According to Northwest and Continental officials,
the airlines have agreed not to implement code-sharing in the local
hub-to-hub markets, such as Detroit to Newark, where DOJ identified
potentially anticompetitive effects.\29 While refraining from
code-sharing is a positive step on the part of the airlines to
address these concerns, some industry experts with whom we spoke
expressed doubt about how much it would reduce the potential harm to
consumers. They explained that because the alliance partners would
continue to exchange information on other aspects of their
operations, such as scheduling, they may not be able to act as
independent competing airlines.
To a certain extent, the harmful effects of reducing competition in
some markets, assuming the alliance partners will not compete, could
be mitigated by increasing competition in other markets. In 286 of
the top 5,000 markets, each of the two carriers has a limited market
share that, if combined, would create a larger, active competitor.\30
These markets served a total of 15.1 million passengers in 1997.\31
However, on average, the effect of creating this potentially stronger
competitor is relatively small because some of these markets continue
to be dominated by other airlines.\32 For example, in 36 of the 286
markets where the Northwest-Continental alliance's market share will
equal or exceed 10 percent, another single competitor will remain
dominant. These 36 markets served 1.2 million passengers in 1997.
Thus, although the alliance will create a greater presence in many
markets, offering more flights and choices to travelers, the overall
improvement in competition will be limited by other dominant
competitors.
The Alliance Could Create Some Additional Dominant Markets. Market
share is particularly important in determining the extent to which an
alliance may be harmful to consumers. Industry experts generally
agree that if an airline can capture significant market share, its
revenue share will increase faster than its market share; in other
words, an airline can earn a fare premium because, among other
reasons, it offers the high level of flight frequencies preferred
especially by business travelers. Airlines that achieve a dominant
market position can drive out airlines with smaller market shares by
making it unprofitable for them to compete. Additionally, a number
of studies have shown that markets with fewer competitors, especially
those dominated by a single carrier, have higher fares.\33 We
reported in 1993, for example, that fares at concentrated airports
were about 22 percent higher than fares at less concentrated
airports.\34
Table 2 shows how many of the top 5,000 busiest airport-pair markets
in 1997 were already dominated by either Northwest or Continental, as
well as how many will be dominated through the alliance's creation.
Northwest dominated 325 markets and Continental dominated 142
markets. By combining the partners' market shares, the alliance will
gain a majority share in 25 additional markets (an increase of 5
percent). One market, Atlanta-Newark, was relatively large, having
served 1.2 million passengers in 1997 (3,205 passengers per day, on
average). However, none of the other markets were relatively large;
they served an average of 135 passengers per day in 1997. Markets
that will be dominated by the new alliance include Atlanta-Newark and
Cleveland-Phoenix.
Table 2
Effect of the Northwest-Continental
Code-Sharing Alliance on Market
Dominance
Number of markets
with more than 50-
percent market Number of
share passengers, 1997
------------------------------ ------------------ ------------------
Northwest 325 20,790,192
Continental 142 17,548,616
Alliance (new) 25 2,355,163
======================================================================
Total 492 40,693,971
----------------------------------------------------------------------
Notes: Market share is defined as the percentage of domestic traffic
in 1997. Dominant markets are defined as those with more than
50-percent market share.
Source: GAO's analysis of data provided by Data Base Products, Inc.,
on the top 5,000 origin and destination markets in 1997.
The Alliance Could Increase Operating and Marketing Barriers to
Entry. To the extent that an alliance creates barriers to market
entry, it can limit competition. If entry is easy, a single airline
cannot charge monopoly prices over the long term because other
airlines will enter and provide competition. However, if entry is
difficult, then the airline may offer fares above competitive levels.
As we have reported in the past, operating barriers such as slot
controls and gate constraints (long-term leases for one airline's
exclusive use of a large number of gates) have contributed to higher
fares on routes to and from some airports.\35 If the formation of an
alliance further concentrated control in one entity at any of these
airports, then fares might rise and the prospects of new competition
at these airports might be further diminished. Similarly, marketing
barriers such as frequent flyer programs and travel agent commission
overrides\36 can also inhibit competition, particularly in markets
already dominated by a given airline.\37
The Northwest-Continental alliance will not appreciably change the
level of concentration at slot-controlled and gate-constrained
airports. Northwest's hubs at Minneapolis and Detroit and
Continental's hub at Newark are gate-constrained airports because
Northwest and Continental control all or a majority of the leases
there. However, these three airports are already heavily
concentrated, and the combination of these two carriers will not
change that. For example, in 1997, Northwest held 77.8 percent of
the market (defined as the total number of enplanements for the year)
at Detroit, and Continental held only 1.6 percent.
We did not evaluate the potential effect on other carriers or
consumers of combining the two airlines' frequent flyer benefits.
Earlier this year, however, we reported on how certain marketing
practices, such as frequent flyer programs, can act as barriers to
entry.\38
The combination of frequent flyer programs under an alliance could
present a more formidable barrier to entry for new-entrant airlines
and point-to-point carriers lacking extensive networks. Northwest
has approximately 20 million members in its WorldPerks frequent flyer
program, and Continental has approximately 16 million in its OnePass
program.
Consumer advocates we interviewed expressed some concern that the
airlines might increase their frequent flyer award requirements in
the future, thereby discounting the value of the frequent flyer
benefits. They also noted that because members of both airlines'
frequent flyer programs may redeem miles on both carriers, finding
available mileage awards could become far more difficult.
In addition, a nonaligned airline, consumer groups, and an industry
expert stressed the adverse effect on competition that alliances like
the one between Northwest and Continental could have through computer
reservation systems. The alliance could gain a competitive advantage
through multiple listings of the same code-sharing flight on the
reservation screen, increasing the likelihood that the alliance's
flights would be the first offered to the consumer.
--------------------
\27 Market share may be determined by any number of measures, such as
the total number of passengers that an airline carries, the number of
available seats flown by the airline, or the total number of
departures in a given market. For this report, we measured market
share by the number of passengers carried in a given airport-pair
market.
\28 For all markets that experienced decreases in the number of
active competitors, we calculated the Herfindahl-Hirschmann Index
(HHI), a measure of concentration used by DOJ, in which higher scores
reflect greater increases in market dominance. The HHI is calculated
by summing the squares of the individual market shares of all
participants. An HHI below 1000 is considered unconcentrated,
between 1000 and 1800 moderately concentrated, and above 1800 highly
concentrated. The average change in the HHI for the
Northwest-Continental alliance was 994, with a range of 219 to 4,588.
This suggests that the Northwest-Continental alliance would create
much greater concentration in markets where the alliance might reduce
the number of active competitors from three to two and from two to
one than in markets where the alliance might reduce the number of
active competitors from five to four.
\29 The DOJ complaint also cited the following six city-pair market
as potentially troublesome: Detroit-Cleveland, Detroit-Houston,
Cleveland-Minneapolis, Minneapolis-New York City,
Houston-Minneapolis, and Houston-Memphis. By forgoing code-sharing
in these markets, the airlines have adopted an approach similar to
that often used by international airline alliances seeking antitrust
immunity. For example, in reviewing the alliance between United and
Lufthansa German Airlines, DOJ expressed concern that the two
airlines could dominate particular city- or airport-pair markets.
DOT ordered that antitrust immunity would not extend to airline
activities relating to pricing, inventory or yield management
coordination, or pooling of revenues, with respect to certain
passengers flying nonstop between Chicago and Frankfurt, and
Washington and Frankfurt.
\30 Our prior testimony, Aviation Competition: Proposed Domestic
Airline Alliances Raise Serious Issues (GAO/T-RCED-98-215, June 4,
1998), which presented our preliminary work, used 5 percent as a
minimum market share to determine the presence of a competitor, a
standard consistent with the work of some other researchers. DOT
uses 10 percent to define a competitor, and subsequent discussions
with DOT officials, some airline officials, and industry experts
convinced us that 10 percent better reflected a competitor in the
market. Accordingly, for the purposes of this report, we are
defining any airline or alliance as an "active competitor" if it
carries 10 percent or more of any given market. Conversely, if an
airline or an alliance has less than 10 percent of any given market,
we are defining its market share as "limited." The effect of raising
the market threshold should help eliminate some potential problems in
the quality of the data reported to DOT and reduce the counts of
markets (and thus of passengers) affected positively or negatively by
the formation of the alliances.
\31 Northwest-Continental officials used a higher threshold in
calculating when the alliance would increase service. They stated
that by creating a market presence of at least 15 percent, their
alliance would create broader service offerings in 32 other major
cities, affecting 33.4 million passengers. For example, cities like
Orlando and Indianapolis would have two significant carriers instead
of one; and Baltimore, Tampa, and Columbus would have three
significant carriers instead of two. We could not verify the number
of passengers affected.
\32 The average increase in the HHI attributable to the creation of
the Northwest-Continental alliance for these 286 markets is 71, with
a range of 2 to 191. This indicates that the addition of the
alliance as an active competitor would make relatively little
difference in the competitive structure of these markets.
\33 See, for example, Steven A. Morrison, "New Entrants, Dominated
Hubs, and Predatory Behavior," Statement before the Subcommittee on
Antitrust, Business Rights, and Competition, Committee on the
Judiciary, United States Senate (Apr. 1, 1998).
\34 Airline Competition: Higher Fares and Less Competition Continue
at Concentrated Airports (GAO/RCED-93-171, July 15, 1993).
\35 See Airline Deregulation: Barriers to Entry Continue to Limit
Competition in Several Key Domestic Markets (GAO/RCED-97-4, Oct. 18,
1996).
\36 A travel agent commission override is a special bonus paid by an
airline to travel agents or agencies as a reward for booking a
targeted proportion of passengers on the airline.
\37 See Aviation Competition: International Aviation Alliances and
the Influence of Airline Marketing Practices (GAO/T-RCED-98-131, Mar.
19, 1998).
\38 See footnote 38.
ALLIANCES BETWEEN
AMERICAN-US AIRWAYS AND
UNITED-DELTA MAY INCREASE
OPERATING AND MARKETING
ENTRY BARRIERS FOR CERTAIN
AIRLINES
---------------------------------------------------------- Letter :4.2
As of October 1998, the arrangements between American and US Airways
and between Delta and United were generally limited to ties between
their respective frequent flyer programs, including reciprocal access
to airport lounges. Implementation of the frequent flyer agreement
between American and US Airways began in August 1998, and Delta and
United began participating in each other's frequent flyer programs on
September 1, 1998.\39
Reciprocal access to both airlines' airport clubs will follow. Both
alliances will provide benefits to members of each airline's frequent
flyer program by offering new destinations to which members can fly.
At the same time, however, by solidifying their customer base through
such marketing efforts, the alliances may raise further barriers to
entry for new airlines in certain markets.
Passengers who belong to both American's and US Airways' frequent
flyer programs will be able to combine miles from both airlines to
redeem an award for travel on either airline. Because the partners
offer greater market presence in different parts of the country (as
well as complementary international destinations), according to
airline officials, American's frequent flyers will have access to 105
new award destinations and US Airways' frequent flyers will be
eligible for award travel to 120 new destinations. In addition,
American's frequent flyers will be able to use their miles from
either airline to claim awards on certain US Airways Shuttle flights
between Washington, D.C.; New York; and Boston.\40 The two airlines
have also agreed to allow reciprocal access to all domestic and
international club facilities. American's club members will gain
access to US Clubs at 12 additional airports, and US Club members
will gain access to 35 American clubs. The agreement between Delta
and United is somewhat more limited. Passengers may choose to
participate in one or both frequent flyer programs, but miles earned
on any given alliance flight may be awarded to only one program.
Passengers will not be able to combine miles earned from both
programs to obtain an award from one of the alliance partners.
Because of the number of locations served uniquely by United and
Delta, according to airline officials, United's frequent flyers will
have access to 75 new domestic award destinations and Delta's
frequent flyers will be eligible for award travel to 108 new domestic
destinations (assuming the alliances' frequent flyer program award
destinations extend to the airlines' commuter partners).
As noted in discussing the Northwest-Continental alliance, we did not
evaluate the effect on other carriers or consumers of combining
frequent flyer benefits through an alliance, but we have reported
that the existence of frequent flyer programs can act as a barrier to
entry for new-entrant airlines and point-to-point carriers lacking
extensive networks.
--------------------
\39 According to data from InsideFlyer magazine, as of Aug. 1998,
American had 32.0 million members in its Aadvantage program (the
largest membership of any frequent flyer program in the world), US
Airways had 4.0 million Dividend Miles members, United had 23.0
million MileagePlus members, and Delta had 17.5 million SkyMiles
members.
\40 Effective Feb. 1, 1999, American will raise the number of miles
required for first class and business class tickets in most
international markets. In addition, an upgrade from a discounted
domestic ticket to a first class ticket will require 30,000 miles
instead of 20,000 miles. Coach class awards will remain the same.
American also eased some mileage requirements, such as the number of
miles needed to claim an upgrade for passengers traveling on a
full-fare ticket.
FEDERAL REVIEWS OF ALLIANCES
ARE ONGOING
------------------------------------------------------------ Letter :5
DOJ and DOT have both been examining the alliances. Although
cooperating, they are conducting separate reviews because they have
different statutory authorities, responsibilities, and remedies. DOJ
filed a civil antitrust action in October to block the equity
investment agreement between Northwest and Continental and amended
its complaint against the revised agreement in December. DOT has the
authority to prevent unfair, deceptive, or anticompetitive practices,
and its statute lists the avoidance of unreasonable concentration in
the airline industry as a public interest factor to be considered in
its decision-making. Under authority provided by a new law, DOT has
imposed waiting periods for certain aspects of the
Northwest-Continental alliance.\41
Although it does not have the authority to preapprove an alliance,
DOT can seek to stop specific anticompetitive practices.
--------------------
\41 P. L. 105-277, division C, title I, sec. 110(f) (1998).
DOJ HAS FILED LEGAL ACTION
TO BLOCK NORTHWEST FROM
OBTAINING VOTING CONTROL
OVER CONTINENTAL
---------------------------------------------------------- Letter :5.1
On October 23, 1998, DOJ filed a civil antitrust complaint to prevent
Northwest from acquiring or holding a majority of Continental's
voting stock. DOJ said in its complaint that Northwest's gaining
voting control would lessen competition in interstate trade and
commerce and unreasonably restrain trade and that the transaction
would also likely create "interlocking directors" on the boards of
directors of both airlines, with certain individuals sitting on both
boards. DOJ believed that the alliance would substantially diminish
both airlines' incentives to compete against each other and would
cause consumers to pay higher prices and receive lower-quality
service in some markets.
DOJ has specific authority to review mergers or stock acquisitions
before they take place under the Hart-Scott-Rodino Act to see whether
they violate antitrust laws, and it has general authority to review
alliances under the Sherman Antitrust Act and the Clayton Act. Under
the Hart-Scott-Rodino Act, an acquisition of voting securities above
a set monetary amount must be reported to DOJ for prior review. DOJ
has the authority to institute civil or criminal proceedings under
the Sherman Act if a merger or acquisition may restrain competition
or is an attempt to monopolize a particular market.\42 In addition,
DOJ may bring a civil action under the Clayton Act if a merger or
acquisition may substantially lessen competition in a relevant market
or tend to create a monopoly.\43
If DOJ believes any agreement is anticompetitive in whole or in part,
it may seek to block the agreement in federal court.
DOJ has been reviewing the Northwest-Continental alliance since it
was first announced in January 1998, under the Hart-Scott-Rodino
process. Under the Hart-Scott-Rodino Act, an acquisition of voting
securities above a set monetary amount must be reported to DOJ\44 for
review prior to the merger. On October 23, 1998, DOJ filed a civil
antitrust action to prevent Northwest from acquiring or holding a
majority of the Continental's voting stock. DOJ alleged that the
effects of the alliance might be to substantially lessen competition
in interstate trade and commerce in violation of the Clayton Act and
to unreasonably restrain trade in violation of the Sherman Act.
According to the complaint filed by DOJ, Northwest's acquisition of
an equity stake and controlling interest in Continental would reduce
Continental's incentive to compete aggressively against Northwest.
As a result, consumers would pay higher prices and receive
lower-quality service in markets dominated by Northwest and
Continental. In addition, according to the complaint, consumers
would lose the benefits of new, competitive entry by Continental
against Northwest in the future and of potential competition in other
markets.
DOJ did not seek a restraining order to stop all aspects of the
alliance from moving forward. While it filed a complaint against the
alliance's investment or stock acquisition agreement, it did not
address the code-sharing alliance and frequent flyer agreements. In
December 1998, Northwest and Continental proceeded with their
alliance under a modified arrangement. For example, they said that
they would not introduce code-sharing on flights between each other's
hubs, where both airlines currently compete, and announced that
Northwest would acquire less than half of the voting control of
Continental's stock. DOJ responded to the revised agreement between
Northwest and Continental by filing an amended complaint on December
18, 1998. DOJ alleged that despite the amended agreement, Northwest
could still own more than 50 percent of the fully diluted voting
power over Continental.\45 As a result, DOJ alleged that consumers
would likely still be harmed in the markets dominated by Northwest
and Continental. DOJ has also indicated that it has competitive
concerns about certain specific aspects of the alliance between the
airlines and that its investigation of these aspects of the alliance
continues. According to DOJ officials, Northwest and Continental
have 60 days from the date of DOJ's filing to respond. During that
period, however, the parties may also file legal motions on various
subjects, such as the procedural schedule.
DOJ has also indicated that it is looking at proposals for the other
two alliances, as well as the potential impact of the alliances on
the entire airline industry. Without detailing its ongoing reviews,
it stated that its analyses of the three alliances under the Sherman
and Clayton acts' authorities will follow an approach similar to that
found in the Merger Guidelines, which are applied to a traditional
merger. Under these guidelines, DOJ uses a five-part analytical
process. First, DOJ defines the markets in which the partners
operate and determines whether they are actual or potential
competitors. DOJ testified before the House Committee on the
Judiciary on May 19, 1998, that the greatest threat to competition
comes when two of very few airlines that compete in a market enter
into a code-sharing agreement in that market. DOJ stated that it is
concerned about the effect on competition any time two of very few
airlines in a market act jointly. Second, DOJ examines aspects of
the agreement that may affect competition--for example, whether the
partners' capacity, scheduling, and pricing decisions will remain
independent. Third, DOJ considers the extent to which new
competitors are likely to enter the market in response to
anticompetitive behavior by the alliance partners. Fourth, DOJ
examines the operational efficiencies or other benefits that may be
generated by the agreement. Finally, DOJ considers whether one of
the partners is likely to exit a market because of financial
considerations if the alliance does not occur. After completing
these parts, DOJ attempts to balance all of the factors in deciding
whether the alliance raises any antitrust concerns. DOJ officials
told us that they are under no timetable for their antitrust review
of the United-Delta and American-US Airways alliances.
If DOJ has concerns, it usually attempts, before bringing a suit, to
negotiate a consent agreement that will restructure the transaction
to remedy the competitive harm. DOJ pointed to its recommendations
to DOT for the international code-sharing arrangements between
American and British Airways and American and the TACA group as an
indication of the types of solutions that could be used domestically.
In its recommendations to DOT on these international alliances, DOJ
outlined several possible solutions, including slot and/or
market-specific divestitures to allow greater opportunity for new
entry and carve-outs to eliminate certain city-pairs from an alliance
where two airlines serve overlapping markets.
--------------------
\42 15 U.S.C. �� 1-7.
\43 15 U.S.C. �� 12-27.
\44 DOJ and the Federal Trade Commission have concurrent jurisdiction
over Hart-Scott-Rodino matters and the Clayton Act. DOJ has
jurisdiction over airline alliances and the airline industry in
general because of an exemption in the Federal Trade Commission Act.
\45 According to DOJ's amended complaint, notwithstanding the Nov.
amendment to the Investment Agreement between Northwest and Air
Partners, L.P., Northwest separately entered into another agreement
on Mar. 2, 1998, with Barlow Investors III, LLC, to purchase
approximately 5 percent of the voting power over Continental to
ensure that Northwest would own over 50 percent of the fully diluted
voting power over Continental.
DOT CAN DELAY ALLIANCES BUT
HAS NO PREAPPROVAL AUTHORITY
UNLESS A CHANGE IN OWNERSHIP
OCCURS
---------------------------------------------------------- Letter :5.2
DOT is authorized to impose waiting periods on proposed alliances
involving major U.S. airlines but has no authority to preapprove
domestic airline alliances, except to conduct a fitness review when a
change in ownership occurs. However, once an alliance is in place,
DOT has the authority to prohibit unfair methods of competition in
the airline industry, which it can use in reviewing the alliances.
DOT derives its authority to review proposed alliances from several
statutory sources. DOT can review a proposed alliance under its
certification and fitness procedures. However, once an alliance is
in place, DOT can challenge it under its authority to prohibit unfair
methods of competition and unfair and deceptive practices.\46 In
addition, DOT's authorizing statute specifies that DOT should
consider such policy matters as preventing unfair, deceptive, and
predatory practices and avoiding unreasonable industry concentration
and excessive market domination in carrying out its duties.\47
Furthermore, under new legislation passed in October 1998, DOT has
the authority to impose an initial 30-day waiting period on certain
joint ventures between major air carriers, which it may extend
another 150 days for such joint venture agreements involving
code-sharing.\48 This authority of DOT's does not limit DOJ's
authority to enforce the antitrust laws.
DOT officials said they have focused their efforts to date on the
Northwest-Continental alliance and on whether it meets fitness
requirements.\49 DOT's fitness procedures include looking at U.S.
citizenship requirements for the airlines involved, as well as their
financial statements and safety records. Only the
Northwest-Continental alliance is subject to these requirements
because it alone is proposing a change in ownership that requires
that a new owner's fitness be determined.\50 As of December 31, 1998,
DOT's fitness determination was ongoing. DOT also imposed waiting
periods on the implementation of the Northwest-Continental alliance's
frequent flyer and code-sharing agreements but said on December 4,
1998, that the airlines could proceed with their reciprocal frequent
flyer program, which had been modified somewhat at DOT's request.
Although DOT does not have the authority to preapprove an alliance,
it could institute an administrative enforcement proceeding if it
determined that some aspect of an alliance or the alliance itself
amounted to an unfair method of competition.\51 DOT did not object to
previous domestic alliances--Continental-America West and
Northwest-Alaska--because it did not find them to be anticompetitive.
As for its review of the United-Delta and American-US Airways
alliances, a DOT official said the Department has received some
information from the airlines on their frequent flyer agreements and
certain other matters. A DOT official said that the Department will
request additional information if it decides the information received
is not sufficient and if the airlines propose to extend their
alliances to include code-sharing. DOT will coordinate with DOJ in
reviewing these alliances.
--------------------
\46 49 U.S.C. � 41712.
\47 See 49 U.S.C. �� 40101(a)(9), (10), and (12).
\48 P.L. 105-277, � 110(f). The joint ventures subject to this
waiting period include only those entered into by a major airline
after Jan. 1, 1998, that involve code-sharing, certain leasing
arrangements, frequent flyer programs, and certain other cooperative
working arrangements.
\49 49 U.S.C. � 41102.
\50 14 C.F.R. � 204.5.
\51 49 U.S.C. � 41712.
AGENCY COMMENTS
------------------------------------------------------------ Letter :6
We provided copies of a draft of this report to DOT and DOJ for their
review and comment. We met with DOT officials from the Office of the
Secretary, including the Deputy Assistant Secretary for Aviation and
International Affairs and the Special Counsel. These officials
indicated that the report provides a useful and constructive
discussion of issues presented by alliances among airlines. They
indicated that because DOT is reviewing each of these alliances, it
is not at liberty, at this time, to provide specific comments on the
potential effects of the alliances discussed in the draft report.
Nonetheless, they indicated that DOT's authority over these three
alliances is not as broad as the draft report might have led some
readers to believe. DOT noted that it could challenge an alliance
after it is in place, citing the Department's authority to prohibit
unfair methods of competition and unfair and deceptive practices.
DOT provided a number of specific and technical comments that
addressed this and other issues; we incorporated these comments as
appropriate. We also held discussions with DOJ officials, including
the Chief of the Transportation, Energy, and Agriculture Section,
within DOJ's Antitrust Division. In general, DOJ found the report to
be a useful and constructive discussion of the issues presented by
the airline alliances, and it provided technical corrections, which
we incorporated into the report as appropriate. DOJ asked that we
more clearly distinguish our use of the term "alliance," (which DOJ
applies only to code-sharing agreements, frequent flyer agreements,
and similar arrangements between separate airlines) from arrangements
in which an equity share is involved. We did so by specifically
noting when the equity investment agreement was at issue. We also
provided each participating airline with a copy of the section of the
report describing its alliance agreement or proposal. The airlines
generally agreed with our characterization of their arrangements and
offered technical corrections, which we incorporated into the report
as appropriate. Some airlines suggested other methods we might use
to determine markets and market shares. In discussing our scope and
methodology (see app. II), we point out that there may be several
ways to define markets in the airline industry, and we carefully
describe what we did and why. We did not share our analysis of the
beneficial or harmful effects of the alliances with the airlines.
---------------------------------------------------------- Letter :6.1
As arranged with your offices, unless you publicly announce its
contents earlier, we plan no further distribution of this report
until 10 days after the date of this letter. At that time, we will
send copies to the Secretary of Transportation; the Attorney General;
the Director, Office of Management and Budget; and other interested
parties. We will send copies to others upon request.
If you have any questions, please call me at (202) 512-2834. Major
contributors to this report are listed in appendix VI.
John H. Anderson, Jr.
Director, Transportation Issues
AIRLINES' MARKET SHARES OF
DOMESTIC PASSENGER TRAFFIC--1997
=========================================================== Appendix I
1997 domestic
traffic (total
number of 1997 market share
passengers (percentage of
enplaned, in total
Airline or alliance millions)\a passengers)\c
------------------------------ ------------------ ------------------
Northwest 47.1 8.5
Continental 34.2 6.2
======================================================================
Northwest-Continental 81.3 14.7
Delta 97.3 17.6
United 72.9 13.2
======================================================================
Delta-United 170.2 30.8
American 66.1 12.0
US Airways 57.4 10.4
======================================================================
American-US Airways 123.5 22.3
======================================================================
Alliance subtotal 375.0 67.8
America West 17.9 3.2
Alaska 11.5 2.1
Southwest 56.1 10.1
Trans World 22.2 4.0
======================================================================
Other major airlines subtotal 107.7 19.5
Other large airlines\b 70.2 12.7
======================================================================
Total\c 552.8 100.0
----------------------------------------------------------------------
\a "Passenger enplanements" represent the total number of passengers
boarding an aircraft. Thus, for example, a passenger that must make
a single connection between his or her origin and destination counts
as two enplaned passengers because he or she boarded two separate
flights. Other measures, such as the number of domestic origin and
destination passengers or revenue passenger miles flown during a
year, are sometimes used to illustrate the relative sizes of
airlines. We believe that using these measures would produce few
differences from the results shown in this table.
\b This category includes such airlines as Reno, Midwest Express, and
AirTran, along with the larger "regional" commuter airlines such as
Atlantic Southeast, Continental Express, Horizon Air, Mesa, and
Simmons. We are excluding other smaller commuter airlines because
they tend not to compete for the same passengers as the larger
airlines and carried only 1.5 percent of the total number of
passengers that flew within the United States in 1997.
\c Percentages may not sum to subtotals because of rounding.
Source: GAO's analysis of DOT's data.
SCOPE AND METHODOLOGY
========================================================== Appendix II
At the request of the Chairman of the Senate Committee on Commerce,
Science, and Transportation and the Chairman of that committee's
Subcommittee on Aviation, we addressed three issues relating to the
alliances among six of the largest U.S. airlines. Specifically, our
objectives were to (1) determine the status of each of the alliances;
(2) examine, for each alliance individually, the potential beneficial
and harmful effects on consumers; and (3) examine the authority of
the departments of Justice (DOJ) and Transportation (DOT) to review
these alliances and the status of these reviews.
To determine the status of each of the alliances, we interviewed
officials from each of the participating airlines, along with
officials from DOJ and DOT. Because each of the alliances changed
somewhat during the assignment, we maintained contact with officials
from the airlines involved. We reviewed publicly available
documentation that described the structure of each alliance and asked
each airline about the involvement of its regional "feeder" commuter
partners, international code-sharing partners, and other domestic
code-sharing airlines (in the case of Continental and Northwest).\52
To ensure that we understood the structure of each alliance, we also
sought explanations from the partners on how they would share revenue
in a variety of situations (for example, when one airline would sell
a ticket to a passenger for a trip that was to involve both the main
alliance airline and one of its regional partners.)
To determine the extent of the service benefits associated with the
alliances, we examined the airlines' statements on their routing and
connection benefits. To evaluate these statements, we analyzed
airline schedule information for May 1998. To obtain these data for
this report, we contracted with BACK Associates, Inc., an aviation
consulting firm. BACK Associates, Inc., used information on flight
schedules submitted by all U.S. airlines to the Official Airline
Guide Worldwide and prepared computer programs to produce tables to
our specifications. We based these specifications on information
submitted by the airlines on the scope of their alliances. We also
imposed some limiting assumptions that our discussions with industry
experts led us to believe were reasonable. We adopted these to
prevent our consulting firm's computer programming from identifying
what would seem to be obviously unreasonable flight patterns (e.g.,
flights between New York and Chicago that would connect through Los
Angeles). These included (1) requiring potential connections between
flights to fall between 30 and 150 minutes of one flight's scheduled
arrival at the connecting airport and the next flight's scheduled
departure\53 and (2) limiting the distance that possible flight
segments might cover ("circuity") to 125 percent of the distance
between a flight's origin and destination, unless the flight
connected through one of the airline's hubs, in which case we allowed
up to 150 percent of the distance. Furthermore, with possible new
double-connection markets, we limited the number of potential
connection points by requiring connections to be made through the
airlines' hubs. Because we were principally interested in the impact
of the alliances on the U.S. domestic market, we excluded
international destinations (including U.S. territories). We did not
review BACK Associates, Inc.'s programming but verified the logic of
that programming and discussed with company officials the approach
that they used. This analysis provided information on the extent to
which the alliances may produce actual new "on-line" connections or
service opportunities for passengers. Finally, we examined whether
competitors provided equivalent, superior, or inferior service to
these destinations.\54 To estimate the number of passengers who might
benefit from the new on-line service, we then matched the new on-line
origins and destinations against the 1997 passenger traffic in these
markets. We did not independently assess the airlines' basis for
stating that their improved service options would generate additional
traffic. We also did not attempt to quantify the benefits of the
reciprocal relationships among frequent flyer programs and clubs that
would be established under the alliances.
To examine the potential harm from the alliances, we determined the
extent to which each airline's routes overlap with those of its
alliance partner by analyzing 1997 data on the 5,000 busiest domestic
airport-pair origin and destination markets or markets for air travel
between two airports. These 5,000 markets accounted for over 90
percent of the total 396 million U.S. domestic passengers in 1997.
To obtain these data for this report, we contracted with Data Base
Products, Inc., which used information submitted by all U.S.
airlines to DOT for 1997 and produced various tables to our
specifications. Data Base Products, Inc., used three different data
sources from DOT: the Origin and Destination Survey (O&D) based on a
10-percent sample of tickets containing itinerary and pricing
information; T-100 on-flight data;\55 and 298C T-1 data, which
supplement the T-100 data with data on commuter and small certified
air carriers. Data Base Products, Inc., made certain adjustments to
these data, such as correcting recognized deficiencies in the air
carriers' O&D data submissions, which have not met DOT's standard of
95-percent accuracy. For example, Data Base Products, Inc., used the
T-100 and the 298C T-1 data to obtain more accurate passenger counts.
We did not review the company's programming but did discuss with
company officials the adjustments that they made.
We examined whether the formation of alliances might reduce
competition within a given airport-pair market under various
assumptions about how large a market share an alliance would need to
be considered an "active competitor." In our testimony on June 4,
1998, which reported our preliminary results on the effects of the
proposed alliances, we defined a competitor as one that carried at
least 5 percent of the enplaned passengers in a particular market.\56
DOT uses 10 percent as a minimum market share. Our conversations
with DOT officials, some airline officials, and industry experts
convinced us that 10 percent better represented a competitor in the
market and also eliminated some potential problems in the quality of
the data that are reported to DOT. For all these reasons, we decided
to use DOT's minimum. To provide additional insight into markets
where an alliance may exert additional or disproportionate market
influence, we then further refined our analysis to focus on those
markets where an alliance would have a dominant share--more than 50
percent. Various studies support the finding that airlines holding
dominant shares of a market reap disproportionate amounts of the
revenue available in that market, because they are able to provide
far more frequent service, which is important for time-sensitive,
high-yield business travelers. To a limited extent, the harmful
effects of having fewer competitors in some markets, assuming the
alliance partners would not compete, could be mitigated by an
increase in competition in other markets.
We defined a market as one involving airport pairs, rather than city
pairs, because certain groups of passengers--particularly business
travelers--who may need to make connections to reach their final
destination do not regard various airports as substitutes. For
example, many, if not most, business travelers going to and from
Chicago, would not regard Midway Airport as an adequate substitute
for O'Hare. This market definition is in line with the analysis of
international airline alliances that appeared in DOJ's comments on
the proposed code-sharing alliance between American Airlines and
British Airways.\57
In our analysis of markets that lost or gained a competitor with the
formation of an alliance, we determined the significance of this loss
or gain by looking at changes in market concentration. To do this,
we used an index used by DOJ, the Herfindahl-Hirschmann Index (HHI),
in which higher scores reflect greater increases in market dominance.
The HHI is calculated by summing the squares of the individual market
shares of all participants. We calculated changes in the HHI in the
markets where a competitor was lost and where a competitor was added,
as well as the changes in the range of HHIs for these markets. We
recognize that the HHI may provide somewhat misleading results when
applied to the airline industry but believe that it is nonetheless
useful in suggesting the change in the amount of concentration over
time.
To determine the cumulative effect that all three alliances might
exert on the U.S. traveling public, we analyzed the routings and
traffic simultaneously. This analysis accounted for various
offsetting effects that the alliances might produce and explains why
the total number of passengers potentially benefiting from or harmed
by the three alliances is not simply the sum of those affected by the
three individual alliances. For example, the analysis of a single
alliance might suggest that if two partners did not continue to
charge competitive prices, then a particular airport-pair market
might lose a competitor, and that loss, according to 1997 origin and
destination information, would affect some number of passengers.
However, that same airport-pair market might conceivably benefit from
another alliance, whose formation would create a new active
competitor. Consequently, the net effect of both alliances might be
to produce the same number of competitors in the market, and the
total number of passengers affected would be the difference between
the two. Because of technical and computing limitations, we were
unable to analyze the cumulative benefits that implementing all three
alliances' would have had on flight connections, flight frequencies,
and the number of new on-line destinations.
We interviewed officials from DOT, DOJ, consumer groups, a flight
attendants' union, and each of the six major airlines contemplating
domestic alliances. For the remaining nonaligned major U.S.
passenger airlines, we either interviewed officials or obtained their
views through published speeches or news releases.
We also conducted interviews with recognized industry experts. These
included academic experts recognized nationally for their expertise
in airline competition work (including the effects of various
cooperative ventures, such as frequent flyer programs); individuals
with expertise in particular aspects of the industry, such as
ticketing and computer reservation system issues; and several Wall
Street airline financial analysts. We asked for their views on how
the alliances might affect domestic airline competition,
participating airlines, and consumers. We selected these individuals
on the basis of their published work on the industry.
To determine the authority of DOT and DOJ to review the domestic
alliances, we reviewed the statutory basis for each department's
work, including DOJ's Merger Guidelines, and we interviewed officials
at both DOT and DOJ. We also reviewed the complaint filed by DOJ
against Northwest and Continental, as well as the public responses
provided by these airlines.
We conducted our work from May 1998 through December 1998 in
Washington, D.C., and Seattle, Washington, in accordance with
generally accepted government auditing standards.
--------------------
\52 Some of the alliance airlines--notably Continental and
Northwest--said that they intended to extend code-sharing to their
international partners. United and Delta officials said that if
their alliance had proceeded, they would have liked to extend it to
their international partners as well, but would wait for the
resolution of a number of issues related to their existing alliances
within Europe. (The European Union is reviewing all international
code-sharing agreements involving European and U.S. airlines and
proposing various changes in these business arrangements.) Because of
differences in the extent to which the alliances have announced how
their international partners would be integrated, we limited our
analysis to the effects on U.S. domestic travel only.
\53 On the basis of information provided by the alliance airlines, we
assumed that the alliance partners would make minimal changes to
their schedules. Airline officials told us that larger changes would
disrupt their entire systems.
\54 Equal service means that competing airlines currently offer
service that matches the alliance's potential single- or
double-connection service. Superior service includes current direct
and nonstop service by competitors between airports that the alliance
could serve at best on a single-connection basis; and direct,
nonstop, or single-connection service by competitors between airports
that the alliance could serve at best on a double-connection basis.
Inferior service means that competing airlines currently offer no
service or double-connection service between airports where the
alliance could offer single-connection service, or no service where
the alliance would provide double-connection service.
\55 14 C.F.R. 241 prescribes the collection of scheduled and
nonscheduled service traffic data from the domestic and international
operations of U.S. air carriers. The schedules submitted by the air
carriers to DOT under this requirement collect nonstop segment data
and on-flight market information by equipment type and by service
class. This report is known as the "T-100" report.
\56 See Aviation Competition: Proposed Domestic Airline Alliances
Raise Serious Issues, (GAO/T-RCED-98-215, June 4, 1998). Our use of
5 percent as a minimum market share to determine the presence of a
competitor is supported by some researchers, notably Belobaba and Van
Acker in "Airline Market Concentration: An Analysis of U.S.
Origin--Destination Markets," Journal of Air Transport Management
(1994).
\57 United and American disagreed with our definition of the relevant
markets, arguing in favor of city pairs. We recognize that city
pairs can also be used to analyze various air markets. At the same
time, while some airports may serve as substitutes for other airports
in the same community, they are often not perfect substitutes. Only
an extensive analysis of the traffic of each airline at each of these
airports would reveal the extent to which the airports could serve as
substitutes.
POTENTIAL BENEFICIAL AND HARMFUL
EFFECTS ON CONSUMERS OF THE
ALLIANCE INITIALLY PROPOSED BY
UNITED AIRLINES AND DELTA AIR
LINES
========================================================= Appendix iii
The current alliance between United and Delta involves reciprocity
between their frequent flyer programs. We analyzed the potential
beneficial and harmful effects of such a structure in the body of
this report. However, the alliance between United and Delta could
incorporate code-sharing in the future. As a result, in this
appendix, we analyzed this alliance assuming a code-sharing
relationship. Our assumption is based on discussions with various
industry experts, who maintain that if the code-sharing alliance
between Northwest and Continental is implemented, the other major
airlines may move toward some sort of code-sharing alliance as a
competitive response. If the United-Delta alliance does proceed with
code-sharing, the potential for beneficial and harmful effects on
consumers could be significant.
UNITED-DELTA ALLIANCE WITH
CODE-SHARING WOULD HAVE NEW
DESTINATIONS AND FREQUENCIES
------------------------------------------------------- Appendix iii:1
When United and Delta originally proposed a code-sharing alliance,
they said that it responded to their customers' wishes and would
benefit consumers. The airlines explained that code-sharing would
create new on-line service, providing consumers with more flight
frequencies and improving connections. A comparison of the airlines'
estimates of a code-sharing alliance's benefits and our analysis of
these benefits follows.
ORIGINALLY PROPOSED
UNITED-DELTA ALLIANCE WOULD
HAVE PROVIDED MORE ON-LINE
DESTINATIONS FOR TRAVELERS
----------------------------------------------------- Appendix iii:1.1
According to United and Delta, a code-sharing alliance such as they
originally proposed would create new service in many U.S. cities.
The airlines reported that, under such an alliance, Delta would gain
access to 19 domestic points that it does not now serve and United
would gain access to 37 such points. In particular, 14 small and
medium-sized cities, such as Bangor, Maine, and Santa Barbara,
California, would gain new service. In total, the airlines said, if
a code-sharing alliance were extended as originally planned to the
airlines' commuter partners, Delta would extend its on-line network
to 108 new domestic points and United to 75 new domestic points.
Thus, multiplying the possible new destinations together, the
airlines said that 8,100 domestic city pairs could gain new on-line
service as a direct result of a code-sharing alliance. For example,
a passenger in Lincoln, Nebraska, could fly on-line to Sarasota,
Florida, for the first time on either United or Delta. The airlines
said they expected that on-line fares would be lower than interline
fares, which are generally the sum of the fares on each airline's
segments.
Our analysis showed that a code-sharing alliance between United and
Delta would serve fewer new airport pairs than the airlines
estimated.\58 Applying the same criteria that we used for our
analysis of the Northwest-Continental alliance, we first found that
such an alliance could create on-line single-connection service in
157 markets that served 2.6 million passengers in 1997. In 86 of
these markets, a United-Delta code-sharing alliance would produce
superior service to that already offered. (In 78 of these markets,
the United-Delta service would be improved on-line single-connection
service compared with existing double-connection service; competitors
provided existing double-connection service in the other 8 markets.)
However, in 71 markets, competing airlines' existing service would be
superior or at least equal to that which the alliance would provide.
We also found that a United-Delta code-sharing alliance could create
50 new double-connection markets. However, either one of the
alliance partners or a competitor already provided on-line service to
each of these markets. Although the United-Delta alliance would
provide superior service to that offered by competitors in 9 markets,
other competing airlines already provided superior or equal service
in 41 of these markets. Table III.1 provides our detailed analysis
of the new on-line benefits that consumers could expect from this
code-sharing alliance.
Table III.1
GAO's Analysis of the Benefits of a
United-Delta Code-Sharing Alliance:
Airport Pairs That Would Receive New On-
Line Service Through Single or Double
Connections
Total markets and average
number of passengers per
day who would receive new
United-Delta service
--------------------------
Number of Number of Number of
Average markets in markets in markets in
number which which which
of competing competing competing
passengers airlines airlines airlines
per day provide provide provide
Type of new on- Number of per market superior equal inferior
line service markets in 1997 service\a service\b service\c
------------------- ------------ ------------ ------------ ------------ ------------
Single-connection 157 46 20 51 86
Double-connection 50 6 33 8 9
-----------------------------------------------------------------------------------------
\a Superior service includes current direct and nonstop service by
competitors between airports that a United-Delta code-sharing
alliance could serve at best with single-connection service; and
direct, nonstop, or single-connection service by competitors between
airports that United-Delta could serve at best with double-connection
service. Direct service differs from nonstop service in that a
"direct" flight makes a scheduled stop between an origin and a
destination, but passengers flying between the origin and destination
are not required to change planes, while a nonstop flight between an
origin and a destination makes no scheduled stops en route.
\b Equal service means that competing airlines currently offer
service that matches a United-Delta code-sharing alliance's potential
single- or double-connection service.
\c Inferior service means that competing airlines currently offer (1)
no service or double-connection service between airports where a
United-Delta code-sharing alliance could offer single-connection
service or (2) no service between airports where a United-Delta
code-sharing alliance could provide double-connection service.
Source: GAO's analysis of information provided by BACK Associates
and Data Base Products, Inc.
--------------------
\58 See app. II for additional information on our methodology.
A UNITED-DELTA CODE-SHARING
ALLIANCE WOULD PROVIDE MORE
FREQUENCIES AND ROUTINGS
----------------------------------------------------- Appendix iii:1.2
According to airline officials, a code-sharing alliance would provide
more flight frequencies and better connections to 81 U.S. cities.
For example, United currently offers one daily nonstop round-trip
flight between Atlanta and San Francisco, and Delta offers six
flights in the same market. Under a code-sharing alliance, consumers
could choose among seven daily nonstop flights as if they were all
Delta or all United flights. United and Delta officials said that
combining each of their nonstop flights would result in a total of
4,600 new daily flight frequencies, affording over 2.3 million
passengers additional nonstop frequency options. Moreover, the
airlines projected that these additional nonstop frequency options
would attract new passengers to their alliance.\59 Our limited
analysis of flight frequencies suggests that the alliance could
result in new, possibly improved, route options.
--------------------
\59 The alliance projected 750,000 new passengers. Experts we
consulted, however, said that the model used was more appropriate for
determining the direction of a change rather than the absolute amount
of a change. Moreover, the model assumed that no competing airline
would make a competitive response, which might decrease the projected
number of passengers.
IF THE PARTNERS DID NOT
COMPETE, A UNITED-DELTA
CODE-SHARING ALLIANCE COULD
HARM CONSUMERS
------------------------------------------------------- Appendix iii:2
Because of the size of United's and Delta's respective route
networks, a code-sharing alliance would produce more overlapping
domestic routes than the other alliances and could harm many
travelers if the airlines did not compete. United and Delta
officials stated that their alliance would prove beneficial rather
than harmful because each airline would remain independent and would
continue to compete with the other.
According to United and Delta, neither airline would coordinate with
the other on operations, such as setting fares and schedules,
managing revenue, or acquiring aircraft. Without coordination in
these areas, they said, the potential for adversely affecting
competition is greatly diminished. According to both United and
Delta, if they had a code-sharing agreement, only the airline that
would actually provide transportation to a passenger on a given
flight segment would receive revenue from that passenger.\60 Each
airline would establish fares separately for the seats it sells. If
a passenger were to fly one segment on United and another on Delta,
the revenue would be prorated according to a standard agreement. The
airlines had also proposed that the "marketing carrier" (i.e., the
airline selling the ticket) would receive a cost-based distribution
fee, roughly equivalent to a travel agent's fee, for tickets sold
under its code. Thus, company officials said, the best way for each
airline to maximize its revenue and profits would be to sell as many
seats as possible on its own aircraft.
We interviewed industry experts when United and Delta were still
planning a code-sharing alliance. These experts maintained that such
an alliance would likely harm consumers by reducing competition
between the two airlines, eventually leading to higher fares. They
said that over time, airlines in this type of alliance would jointly
identify the markets where it would make financial sense for them to
reduce or eliminate capacity, especially those markets where the
partner airline had more flights. They said that because each
airline's management would retain a strong incentive to maximize
revenue and profit, each would benefit by reducing competition with
its alliance partner. However, they added that detecting the exact
point at which anticompetitive behavior is occurring, or could be
attributed to the existence of the alliance rather than to
independent business decisions, is extremely difficult. This same
reasoning--that alliance partners would be unlikely to compete in
markets that each served--was used by United when it filed a formal
statement opposing American's alliance with the TACA group.\61
--------------------
\60 Because of the airlines' concerns about confidentiality, we were
unable to conduct an independent review of the agreement.
\61 Comments of United Airlines, Inc., on American Airlines, Inc., et
al. and the TACA Group's Reciprocal Code-Sharing Services
Proceeding, Docket OST-96-1700-72 (Sept. 11, 1997). The TACA group
comprises the national airlines for Guatemala, Panama, Costa Rica,
Nicaragua, Honduras, and El Salvador.
A UNITED-DELTA CODE-SHARING
ALLIANCE WOULD REDUCE
COMPETITION IN SHARED
MARKETS
----------------------------------------------------- Appendix iii:2.1
If the airlines established a code-sharing alliance and competed less
with each other, approximately 27.8 million passengers in 550
distinct markets could be harmed through reductions in service and
increases in airfares. Figure III.1 shows the number of markets that
could be adversely affected by a code-sharing alliance if the two
airlines did not continue to compete. The figure shows, among other
things, that in 58 markets that served 4.3 million passengers in
1997, the alliance would become the only active competitor with a
market share of more than 10 percent.
Figure III.1: Number of
Markets and Passengers Subject
to Potentially Decreased
Competition Under a
United-Delta Code-Sharing
Alliance
(See figure in printed
edition.)
Source: GAO's analysis of data provided by Data Base Products, Inc.,
on the top 5,000 origin and destination markets in 1997.
In many of the markets where the number of active competitors would
decrease if the two airlines did not continue to compete with each
other, the changes in concentration would be significant.\62 For
example, in the 179 markets where the creation of a code-sharing
alliance would decrease the number of active competitors from three
to two, the alliance would become the largest carrier in 151 and the
second largest carrier in the remaining 28. These 179 markets served
over 10.5 million passengers in 1997.
In some markets, the establishment of a code-sharing alliance between
United and Delta could have benefits that might, to some extent,
mitigate the harmful effects that the alliance might otherwise
create. Specifically, at airports where each airline alone has a
relatively insignificant competitive presence, a code-sharing
alliance would create a larger, potentially stronger entity, allowing
it to compete more effectively against other airlines. In 1997,
United and Delta each had a limited share (i.e., less than 10
percent) in 143 of the top 5,000 markets that served 11 million
passengers during that year. However, under a code-sharing alliance,
their combined market share would exceed 10 percent, and the
partnership would constitute a new active competitor. Of these 143
markets, 59 are currently dominated by one major airline, and a
United-Delta code-sharing alliance could increase competition for the
4.9 million passengers served in these markets. Nevertheless, our
analysis shows that the addition of the alliance as an active
competitor in these markets would make little overall difference
because the alliance would remain relatively small, especially
compared with other competitors.\63
--------------------
\62 The average change in the Hirschmann-Herfindahl Index (HHI) for
these 550 markets is 1,208, with a range of 235 to 4,470. This
suggests that a United-Delta code-sharing alliance would
substantially increase concentration in certain markets.
\63 The average change in the HHI for the 143 United-Delta markets
where the number of active competitors would increase because of the
alliance is 69, with a range of 1 to 193. This indicates that the
addition of the alliance as an active competitor would make
relatively little overall difference in these markets.
A UNITED-DELTA CODE-SHARING
ALLIANCE WOULD ESTABLISH
DOMINANT SHARES IN OVER 200
NEW MARKETS
----------------------------------------------------- Appendix iii:2.2
According to industry experts, if the partners of a code-sharing
alliance held a dominant share of the passenger traffic in shared
(i.e., overlapping) markets, they could more easily coordinate
capacity and maximize revenue, potentially harming passengers.
Specifically, the partners could choose to maximize revenue by
raising fares in selected markets where they hold a dominant
share--thereby harming consumers in these markets--even without
increasing their market share overall. Table III.2 shows that a
United-Delta code-sharing alliance would have more than a 50-percent
market share in 1,218 of the top 5,000 airport-pair markets. These
1,218 markets served more than 84 million passengers in 1997. The
alliance itself would add 213 markets (11.8 million passengers) to
the 1,005 markets already currently dominated by either United or
Delta.
Table III.2
Potential Effect of a United-Delta Code-
Sharing Alliance on Market Dominance
Number of markets
with more than 50-
percent market Number of
share passengers, 1997
------------------------------ ------------------ ------------------
United 273 30,227,283
Delta 732 42,681,590
Alliance (new) 213 11,754,050
Total 1,218 84,662,923
----------------------------------------------------------------------
Note: Market share represents the percentage of 1997 passenger
traffic carried by each airline.
Source: GAO's analysis of data provided by Data Base Products, Inc.,
on the top 5,000 origin and destination markets in 1997.
A UNITED-DELTA CODE-SHARING
ALLIANCE WOULD HAVE THE
POTENTIAL TO AFFECT
OPERATING AND MARKETING
BARRIERS TO ENTRY
----------------------------------------------------- Appendix iii:2.3
At the nation's slot-controlled and gate-constrained airports, a
United-Delta code-sharing alliance would not appreciably change the
level of concentration. Any potential harm to consumers would
probably be slight. For example, United's 48.3-percent share of
traffic at Chicago's slot-controlled O'Hare Airport and Delta's
76.8-percent share of traffic at Cincinnati's gate-constrained
airport would not be significantly affected by the combination, since
neither airline maintains a significant presence at the other's hub.
At O'Hare, a code-sharing alliance's share would increase to 51.7
percent; at Cincinnati, it would increase to 77.9 percent.
Nonaligned airlines, consumer groups and one industry expert stressed
the adverse effect on competition that a code-sharing alliance such
as that originally proposed between United and Delta could exercise
through computer reservation systems. The alliance could gain a
competitive advantage through multiple listings of the same
code-sharing flight on the reservation screen, increasing the
likelihood that the alliance's flights would be the first offered to
the consumer.
POTENTIAL BENEFICIAL AND HARMFUL
EFFECTS ON CONSUMERS OF A
CODE-SHARING ALLIANCE BETWEEN
AMERICAN AIRLINES AND US AIRWAYS
========================================================== Appendix IV
For American Airlines and US Airways, as for United and Delta, we
assumed that their alliance could proceed to a partnership involving
code-sharing. Because officials from both airlines said that they
would implement code-sharing if the other two alliances moved forward
with code-sharing arrangements, our analysis of the alliance's
potential beneficial and harmful effects assumes that they would do
the same. With code-sharing, the alliance could have significant
effects--both positive and negative--on consumers throughout the
United States.
A CODE-SHARING ALLIANCE BETWEEN
AMERICAN AND US AIRWAYS WOULD
INCLUDE NEW DESTINATIONS AND
FREQUENCIES
-------------------------------------------------------- Appendix IV:1
If American Airlines and US Airways ultimately decide to implement
code-sharing in a substantial number of markets, passengers could
benefit from more on-line destinations, better routes and
connections, and more flight frequencies. But because the alliance
does not currently include code-sharing between main American and
main US Airways, the two partners have not predicted any benefits.
AN AMERICAN-US AIRWAYS
CODE-SHARING ALLIANCE WOULD
OFFER CONSUMERS MORE ON-LINE
DESTINATIONS
------------------------------------------------------ Appendix IV:1.1
We found that a code-sharing alliance between American and US Airways
could create new single- and double-connection markets. Because
American currently serves 55 airports from its hubs that US Airways
does not serve and US Airways serves 76 airports from its hubs that
American does not serve, the alliance could, in theory, provide
on-line service in 4,180 new markets. However, many of these markets
would require more than two connections. When we eliminated these
impractical routes,\64 we found that an American-US Airways
code-sharing alliance could create 483 new single- or
double-connection markets.
We found that such an alliance could create on-line single-connection
service in 166 markets that served 2.0 million passengers in 1997.\65
In 47 of these markets, the alliance would produce superior service
to that already offered by a competitor. However, in 119 markets,
competing airlines' existing service would be superior or at least
equal to that created by the alliance. We also found that an
American-US Airways code-sharing alliance could create 317 new
double-connection markets. However, either one of the alliance
partners or a competitor already provided on-line service to each of
these markets. Although an American-US Airways code-sharing alliance
would provide superior service to that offered by competitors in 3
markets, other competing airlines already provided superior or equal
service to 314 markets. Table IV.1 provides our detailed analysis of
the new on-line benefits that consumers could expect from this
alliance.
Table IV.1
GAO's Analysis of Benefits Under an
American-US Airways Code-Sharing
Alliance: Airport Pairs That Would
Receive New On-Line Service Through
Single or Double Connections
Total markets and average
number of passengers per
day who would receive new
American-US Airways
service
--------------------------
Number of Number of Number of
markets in markets in markets in
Average which which which
number of competing competing competing
passengers airlines airlines airlines
per day per provide provide provide
Type of new on- Number of market in superior equal inferior
line service markets 1997 service\a service\b service\c
------------------- ------------ ------------ ------------ ------------ ------------
Single-connection 166 34 24 95 47
Double-connection 317 7 271 43 3
-----------------------------------------------------------------------------------------
\a Superior service includes current direct and nonstop service by
competitors between airports that an American-US Airways code-sharing
alliance could serve at best with single-connection service; and
direct, nonstop, or single-connection service by competitors between
airports that an American-US Airways code-sharing alliance could
serve at best with double-connection service. Direct service differs
from nonstop service in that a "direct" flight makes a scheduled stop
between an origin and a destination, but passengers flying between
the origin and destination are not required to change planes, while a
nonstop flight between an origin and a destination makes no scheduled
stops en route.
\b Equal service means that competing airlines currently offer
service that matches an American-US Airways code-sharing alliance's
potential single- or double-connection service.
\c Inferior service means that competing airlines currently offer (1)
no service or double-connection service between airports where an
American-US Airways code-sharing alliance could offer
single-connection service or (2) no service between airports where an
American-US Airways code-sharing alliance could provide
double-connection service.
Source: GAO's analysis of information provided by BACK Associates
and Data Base Products, Inc.
--------------------
\64 See app. II for additional information on our methodology.
\65 See app. II for additional information on our methodology.
IF THE PARTNERS DID NOT
COMPETE, AN AMERICAN-US AIRWAYS
CODE-SHARING ALLIANCE COULD
HARM CONSUMERS
-------------------------------------------------------- Appendix IV:2
Given the size of the partners' respective route networks, a
code-sharing alliance between American and US Airways would produce
much overlap and could harm many travelers if, over time, the
airlines did not continue to compete as independent companies.
AN AMERICAN-US AIRWAYS
CODE-SHARING ALLIANCE COULD
POTENTIALLY REDUCE
COMPETITION IN SHARED
MARKETS
------------------------------------------------------ Appendix IV:2.1
If the current American-US Airways alliance moved to a code-sharing
arrangement, it could reduce competition in 260 of the top 5,000
markets. These 260 markets served 13.3 million passengers in 1997.
Moreover, such an alliance could eliminate competition in 24 of the
markets, which served 2.0 million passengers in 1997. Such an
alliance could also decrease the number of active competitors from
three to two in 90 markets. Our analysis shows that in 80 of these
90 markets, the alliance would become the largest carrier, and in the
remaining 10 markets, it would become the second largest carrier (see
fig. IV.1).\66
Figure IV.1: Number of Markets
and Passengers Subject to
Potentially Decreased
Competition Under an
American-US Airways
Code-Sharing Alliance
(See figure in printed
edition.)
Source: GAO's analysis of data provided by Data Base Products, Inc.,
on the top 5,000 origin and destination markets in 1997.
The harmful effects that could result from reducing the number of
competitors in 260 markets could be mitigated by the benefits of
increasing competition in some individual markets. Each of the two
airlines has a limited share (i.e., less than 10 percent) in 60 of
the top 5,000 markets. These 60 markets served 5.2 million
passengers in 1997. Under a code-sharing alliance, the airlines'
share in these markets would exceed 10 percent, and the alliance
would represent a new active competitor. Of these 60 markets, 15 are
currently dominated by a single airline. In these 15 markets, the
alliance's creation could enhance competition, benefiting the 2.3
million passengers who were served in these markets during 1997.
However, on average, the addition of the alliance as an active
competitor would make relatively little difference in the level of
concentration in these markets.\67
--------------------
\66 The average change in the Hirschmann-Herfindahl Index (HHI) for
these 260 markets is 1,217, with a range from 228 to 4,477. This
suggests that an American-US Airways code-sharing alliance would
significantly increase concentration in a number of origin and
destination markets.
\67 Under the American-US Airways alliance, the average change in the
HHI for these 60 markets is 62, with a range of 5 to 156. This
indicates that the addition of the alliance as an active competitor
would make relatively little overall difference in these markets.
AN AMERICAN-US AIRWAYS
CODE-SHARING ALLIANCE WOULD
INCREASE DOMINANCE IN
CERTAIN MARKETS
------------------------------------------------------ Appendix IV:2.2
With code-sharing, an American-US Airways alliance would hold a
majority share in 977 of the top 5,000 origin and destination
markets. These 977 markets served 58.2 million passengers in 1997.
According to 1997 data, American had a dominant share in 278 markets
that served about 25.1 million passengers, and US Airways had a
majority share in 596 markets that served about 29 million
passengers. Thus, a code-sharing alliance would give the partners a
majority share in 103 additional markets. This increase in dominance
is significant because it would allow the partners to raise fares in
selected markets--thereby potentially harming consumers in these
markets--without increasing their market share overall. These 103
markets served nearly 4.2 million passengers in 1997 (see table
IV.2). These markets include routes between American's hubs in
Dallas or Miami and US Airways's hubs in Philadelphia and Pittsburgh
where the alliance would carry more than 80 percent of the
passengers.
Table IV.2
Potential Effect of an American-US
Airways Code-Sharing Alliance on Market
Dominance
Number of markets
with more than 50-
percent market Number of
share passengers, 1997
------------------------------ ------------------ ------------------
American 278 25,050,145
US Airways 596 28,995,941
Alliance (new) 103 4,171,404
Total 977 58,217,490
----------------------------------------------------------------------
Note:Market share represents the percentage of 1997 passenger traffic
carried by each airline.
Source: GAO's analysis of data provided by Data Base Products, Inc.,
on the top 5,000 origin and destination markets in 1997.
AN AMERICAN-US AIRWAYS
CODE-SHARING ALLIANCE WOULD
ALSO INCREASE BARRIERS TO
ENTRY
------------------------------------------------------ Appendix IV:2.3
Operating barriers could be an issue for this alliance if it proceeds
to code-sharing. Both airlines have significant presences at
slot-controlled Washington Reagan National and New York LaGuardia
airports. US Airways holds the largest percentage of slots at
Washington Reagan National (35.4 percent). Under an alliance, it
would control 49.0 percent of the slots there. US Airways also holds
the largest percentage of slots at New York LaGuardia (27.0 percent).
Under an alliance, it would control 44.5 percent of the slots there.
The change in concentration that would occur under an alliance at
gate-constrained airports would be less significant. US Airways
already has more than 80 percent of the market (as measured by 1997
enplanements) at the Pittsburgh airport. We have previously reported
fares more than 20 percent higher in constant dollars since
deregulation at this airport.\68
However, because American does not have a significant market share at
Pittsburgh (less than 2 percent), an alliance would not significantly
increase the partners' market share.
In addition, some nonaligned airlines, consumer groups, and an
industry expert stressed the adverse effect on competition that a
code-sharing alliance between American and US Airways could exercise
through computer reservation systems. The alliance could gain a
competitive advantage through multiple listings of the same
code-sharing flight on the reservation screen, increasing the
likelihood that the alliance's flights would be the first offered to
the consumer.
--------------------
\68 Airline Deregulation: Barriers to Entry Continue to Limit
Competition in Several Key Domestic Markets (GAO/RCED-97-4, Oct. 18,
1996).
IMPLEMENTATION OF ALL THREE
ALLIANCES AS CODE-SHARING
ARRANGEMENTS COULD SUBSTANTIALLY
AFFECT COMPETITION
=========================================================== Appendix V
If all three of the alliances were to move forward as code-sharing
arrangements, many questions would arise not only about the
beneficial and harmful effects that could be attributed directly to
the individual alliances but also about the cumulative effect of the
alliances on competition in the industry, particularly for
new-entrant and nonaligned airlines.
If all three alliances were implemented as code-sharing agreements,
some consumers would benefit from extended route networks but other
consumers could be harmed if competition were to decline. On the one
hand, consumers would gain access to new airport pairs served by the
three alliances, as well as additional flight frequencies, new routes
with better connections, and expanded frequent flyer programs. On
the other hand, according to the industry experts we interviewed, the
alliances would stimulate little growth in passenger traffic and
would generally shift passengers either among themselves or away from
other nonaligned airlines in various markets. No airline partner
currently plans to add new flights or airplanes in any given market.
Moreover, if the formation of code-sharing alliances created an
environment in which the partners competed less vigorously, the
number of competitors could be reduced in hundreds of domestic
airport-pair markets that were among the top 5,000 in 1997,
potentially affecting tens of millions of passengers. In addition,
the number of markets dominated by the alliances would increase by
about 10 percent, causing over two-thirds of U.S. travelers to fly
in markets dominated by a single airline. Operating barriers could
increase at the slot-controlled airports in New York and Washington,
D.C., and if more markets were dominated by the alliances, marketing
barriers such as those represented by combined frequent flyer
programs could make entry by new airlines more difficult. Finally,
if all three alliances were to move forward as code-sharing
arrangements, the computer reservation systems that travel agents use
to book airline tickets could begin to display each alliance's
flights twice--once under each partner's code. Independent or
new-entrant airlines then might have more difficulty getting their
flights listed prominently in travel agents' displays.
CONSUMERS COULD REALIZE SOME
BENEFITS IF ALL THREE ALLIANCES
WERE TO PROCEED AS CODE-SHARING
AGREEMENTS
--------------------------------------------------------- Appendix V:1
If the three alliances were to proceed as code-sharing agreements,
they would be likely to create some new on-line destinations, allow
some new or improved routes and connections, and expand frequent
flyer and club benefits to members. However, because of some overlap
among the alliances, the total number of unique, new markets would be
smaller than the sum of such markets for each of the three alliances,
and fewer passengers would be likely to benefit from the alliances
than some of the airlines have predicted because their estimates
assume no competitive responses from other airlines. Overall, the
industry experts we interviewed indicated that the alliances would do
little to stimulate growth in passenger traffic because they would
mainly shift passengers among themselves, or from other airlines, in
various markets.
Two alliances would be likely to create new frequencies and better
connections (assuming no schedule changes by the airlines).
Northwest-Continental officials and United-Delta officials did not
count their possible new frequencies and routings in the same manner,
and because American and US Airways originally proposed a much more
limited alliance, they did not calculate how many new frequencies and
routes a code-sharing alliance would make possible.
Many consumers may also benefit from the expanded frequent flyer
options available under the current alliance agreements. However,
the particular frequent flyer benefits will vary by alliance for
consumers. In addition, the ability of consumers to obtain awards
may depend on the availability of frequent flyer seats, the number of
miles required to obtain awards, and the types of restrictions (e.g.,
blackout dates) that the airlines specify. Moreover, as noted
earlier, award requirements may change over time.
For methodological reasons, we were unable to quantify potential
cumulative benefits (e.g., new routes or flight frequencies) that
could be created by the alliances, but we believe that these benefits
could be substantial.
IF PARTNERS DID NOT COMPETE,
CODE-SHARING ALLIANCES COULD
HARM CONSUMERS
--------------------------------------------------------- Appendix V:2
It is difficult to determine whether three code-sharing alliances
would reduce competition, but industry experts' concerns and the
airlines' past records give cause for concern. There is agreement
among some industry experts that competition would be likely to
decline over time as the partners recognized their interdependence
and began to maintain fares above the competitive level. Such an
outcome is consistent with widely held economic principles that
associate less competition with fewer competitors. As the ties
between major airlines were strengthened, the opportunities would
increase for airlines to recognize their interdependence. If this
should occur, competition would suffer and fares would rise.
COMPETITION COULD DECLINE IN
SHARED MARKETS
------------------------------------------------------- Appendix V:2.1
If all three alliances were to proceed with code-sharing, then the
number of competitors could decline in some domestic airport-pair
markets. As figure V.1 shows, 78 of the top 5,000 markets would
become single-airline markets if the alliance partners did not
compete with each other. Overall, concentration would also increase.
Figure V.1: Number of Markets
and Passengers Subject to
Potentially Decreased
Competition Under Three
Code-Sharing Alliances
(See figure in printed
edition.)
\a Includes 17 markets where the three alliances would reduce the
number of active competitors from four to two.
\b Includes 23 markets where the three alliances would reduce the
number of active competitors from five to three.
\c Includes two markets where the three alliances would reduce the
number of active competitors from six to four.
Source: GAO's analysis of data provided by Data Base Products, Inc.,
on the top 5,000 origin and destination markets in 1997.
If the alliance partners did not compete, the harmful effects of
fewer competitors could, to a limited extent, be mitigated by an
increase in competition in some markets. In total, the three
alliances could add an active competitor to 328 of the top 5,000
markets. Of these 328 markets, which served almost 22 million
passengers in 1997, 97 are currently single-competitor markets. In
these markets, which served over 7.8 million passengers in 1997, the
alliances would add one or more active competitors.
MARKET DOMINANCE WOULD
INCREASE
------------------------------------------------------- Appendix V:2.2
If three code-sharing alliances were implemented, the number of
dominated markets would increase by 341 (about 10 percent), from
3,381 to 3,722 airport pairs--or about 75 percent of the top 5,000
markets in 1997.\69 Such an increase in market dominance is
significant, according to industry analysts, who predicted that
alliance partners might not be able to gain much market share overall
but might be able to increase revenues in individual markets where
they held a dominant position. In 1997, approximately 280 million
passengers, or over two-thirds of those who flew domestically, flew
in these markets.
--------------------
\69 Non-alliance airlines--such as TWA, AmericaWest, and
Southwest--dominated 1,035 of the top 5,000 routes in 1997. Over 96
million passengers flew on these routes. Thus, were the three
alliances to proceed to code-sharing and not act independently, they
would dominate 2,687 routes, on which over 183 million passengers
flew in 1997.
BARRIERS TO ENTRY COULD
INCREASE
------------------------------------------------------- Appendix V:2.3
Overall, airfares have decreased and service has improved since the
airline industry was deregulated in 1978. Nevertheless, operating
and marketing barriers have presented significant barriers to
competition. The existence of these barriers increases the
likelihood that additional concentration could harm consumers by
discouraging entry by other established or new entrant airlines, thus
allowing the alliance partners to raise their fares and/or reduce
their services. As we have previously pointed out, operating
restrictions such as slot controls and gate constraints can make it
more difficult for new carriers to enter a market. In all cases, the
alliances could add to the level of concentration at these airports,
as shown in table V.1.
Table V.1
Alliance Partners' Combined Market Share
at Slot-Controlled and Gate-Constrained
Airports
Postalliance market share\a
----------------------------------------
Prealliance
market share\a/ United- American-US Northwest-
Constraint Airport dominant airline Delta Airways Continental
------------ --------------- ---------------- ------------ ------------ ------------
Slot Chicago O'Hare 48.3/United 51.7 40.1 4.2
Washington 35.4/US Airways 24.0 49.0 14.4
Reagan National
New York 30.0/American 28.9 30.1 1.1\b
Kennedy
New York 27.0/US Airways 34.3 44.5 10.1
LaGuardia
Gate Charlotte 83.8/US Airways 3.3 85.3 1.4
Cincinnati 76.8/Delta 77.9 0.9 1.4
Detroit 77.8/Northwest 4.8 4.8 79.4
Minneapolis 80.5/Northwest 5.9 3.9 81.5
Newark 60.8/ 15.0 12.1 64.6
Continental
Pittsburgh 82.2/US Airways 3.6 83.1 2.5
-----------------------------------------------------------------------------------------
\a Market share is expressed as the percentage of total 1997
enplanements at each airport.
\b Continental did not serve New York's Kennedy Airport in 1997.
Source: GAO's analysis of DOT's data.
Although, in most cases, the percentage increase in market share
would be small, in every case, the alliance partner with the lesser
share might have an opportunity to improve its market position,
potentially increasing the difficulty for other airlines of gaining
access at these 10 important airports. In the complaint it filed
against Northwest and Continental, DOJ also noted that difficulty in
obtaining access to gate facilities impedes new entry. One
opportunity that the lesser partner might derive from its immediate
access to the dominant partner's strength at the airport is that its
flights might appear more attractive to consumers. For example,
under a code-sharing alliance, Texas consumers might find Continental
a more desirable airline to fly to Minneapolis, where Northwest
enjoys a dominant market share. This is because Continental's
presence in the alliance would allow it to market, and to offer, an
increased number of daily flights in these markets under its code on
Northwest's planes. Another opportunity for the lesser partner might
be to remove aircraft from airports where its operations are
unprofitable and to shift passengers to its partner's aircraft,
thereby strengthening its partner's position at that airport. For
example, if Northwest and Continental each operated flights from
Minneapolis to Cleveland, but Continental served that market only
with smaller commuter aircraft instead of larger jets, it might
choose to put its passengers on Northwest's jet flights.
Concentration at airports other than the 10 cited in table V.1 could
also increase, potentially preventing small and new-entrant carriers
from gaining market share at heavily concentrated airports.
We have also reported that airline sales and marketing practices may
make competitive entry more difficult for other airlines.\70 However,
we have not been able to quantify the effects of these barriers on
competition for any or all of the alliances. Nevertheless, marketing
practices such as frequent flyer programs and special bonuses to
travel agents for booking traffic on an incumbent airline may
encourage travelers to choose one airline over another on the basis
of factors other than the best fares. Such practices may be most
important if an airline is already dominant in a given market or
markets. Because the alliances could increase dominance by about 10
percent in the top 5,000 markets in 1997, marketing barriers in these
metropolitan areas would be likely to become more important. DOJ
also noted the effect that such practices had on impeding competition
from new entrants in the complaint it filed against Northwest and
Continental. Nonetheless, mitigating the effect of these practices
without banning them is difficult, and banning them involves a
trade-off between their potential anticompetitive effects and the
consumer benefits that some of them bring.
Some nonaligned airlines, consumer groups, and an industry expert
stressed the adverse effects on competition that code-sharing
alliances could exercise through computer reservation systems.
Through code-sharing, flights that previously appeared in these
systems under one airline's code could now appear twice--once under
the operating airline's code and once under the code-sharing
partner's. Connecting flights between the partners could appear
three times, once under each partner's code and once as a connecting
flight. Thus, it is likely that the creation of a code-sharing
alliance would increase the number of flights listed for the partners
on the first reservation screen, from which travel agents often book
flights. Where three alliances provided code-sharing flights in the
same markets, six or more code-sharing flights might appear on the
first screen, crowding out opportunities for other airlines.
--------------------
\70 See, for example, Aviation Competition: International Aviation
Alliances and the Influence of Airline Marketing Practices
(GAO/T-RCED-98-131, Mar. 19, 1998).
MAJOR CONTRIBUTORS TO THIS REPORT
=========================================================== Appendix 0
RESOURCES, COMMUNITY, AND ECONOMIC
DEVELOPMENT DIVISION, WASHINGTON,
D.C.
Aaron Casey
Elizabeth Eisenstadt
Tina Kinney
Steven Martin
Sara Ann Moessbauer
Marnie Shaul
SEATTLE REGIONAL OFFICE
Paul Aussendorf
Stan Stenersen
OFFICE OF GENERAL COUNSEL
David Hooper
OFFICE OF THE CHIEF ECONOMIST
Joseph Kile
RELATED GAO PRODUCTS
============================================================ Chapter 1
Aviation Competition: Proposed Domestic Airline Alliances Raise
Serious Issues (GAO/T-RCED-98-215, June 4, 1998).
Domestic Aviation: Service Problems and Limited Competition Continue
in Some Markets (GAO/T-RCED-98-176, Apr. 23, 1998).
Aviation Competition: International Aviation Alliances and the
Influence of Airline Marketing Practices (GAO/T-RCED-98-131, Mar.
19. 1998).
Airline Competition: Barriers to Entry Continue in Some Domestic
Markets (GAO/T-RCED-98-112, Mar. 5, 1998).
Domestic Aviation: Barriers Continue to Limit Competition
(GAO/T-RCED-98-32, Oct. 28, 1997).
Airline Deregulation: Addressing the Air Service Problems of Some
Communities (GAO/T-RCED-97-187, June 25, 1997).
International Aviation: Competition Issues in the U.S.-U.K. Market
(GAO/T-RCED-97-103, June 4, 1997).
Domestic Aviation: Barriers to Entry Continue to Limit Benefits of
Airline Deregulation (GAO/T-RCED-97-120, May 13, 1997).
Airline Deregulation: Barriers to Entry Continue to Limit
Competition in Several Key Domestic Markets (GAO/RCED-97-4, Oct. 18,
1996).
Domestic Aviation: Changes in Airfares, Service, and Safety Since
Airline Deregulation (GAO/T-RCED-96-126, Apr. 25, 1996).
Airline Deregulation: Changes in Airfares, Service, and Safety at
Small, Medium-Sized, and Large Communities (GAO/RCED-96-79, Apr. 19,
1996).
International Aviation: Airline Alliances Produce Benefits, but
Effect on Competition Is Uncertain (GAO/RCED-95-99, Apr. 6, 1995).
Airline Competition: Higher Fares and Less Competition Continue at
Concentrated Airports (GAO/RCED-93-171, July 15, 1993).
Computer Reservation Systems: Action Needed to Better Monitor the
CRS Industry and Eliminate CRS Biases (GAO/RCED-92-130, Mar. 20,
1992).
Airline Competition: Effects of Airline Market Concentration and
Barriers to Entry on Airfares (GAO/RCED-91-101, Apr. 26, 1991).
Airline Competition: Industry Operating and Marketing Practices
Limit Market Entry (GAO/RCED-90-147, Aug. 29, 1990).
Airline Competition: Higher Fares and Reduced Competition at
Concentrated Airports (GAO/RCED-90-102, July 11, 1990).
Airline Deregulation: Barriers to Competition in the Airline
Industry (GAO/T-RCED-89-65, Sept. 20, 1989).
Airline Competition: Fare and Service Changes at St. Louis Since
the TWA-Ozark Merger (GAO/RCED-88-217BR, Sept. 21, 1988).
Competition in the Airline Computerized Reservation Systems
(GAO/T-RCED-88-62, Sept. 14, 1988).
Airline Competition: Impact of Computerized Reservation Systems
(GAO/RCED-86-74, May 9, 1986).
Airline Takeoff and Landing Slots: Department of Transportation's
Slot Allocation Rule (GAO/RCED-86-92, Jan. 31, 1986).
Deregulation: Increased Competition Is Making Airlines More
Efficient and Responsive to Consumers (GAO/RCED-86-26, Nov. 6,
1985).
*** End of document. ***