COMMENTS OF THE DEPARTMENT OF JUSTICE
December 21, 2009
TABLE OF CONTENTS
Comments of the United States Department of Justice
The United States Department of Justice (DOJ) hereby submits these comments to the United States Department of Transportation (DOT) in the matter of the Joint Application of American Airlines, Inc. (American), British Airways PLC (British Airways), Iberia Lineas Aereas De Espana, S.A. (Iberia), Finnair OYJ (Finnair), and Royal Jordanian Airlines (collectively, Applicants) for Approval of and Antitrust Immunity for Alliance Agreements (the Joint Application or J.A.).
Applicants proposed agreements would result in competitive harm on certain transatlantic routes serving 2.5 million passengers annually. Fares between six pairs of cities (1) Boston and London, (2) Chicago and London, (3) Dallas and London, (4) Miami and London, (5) Miami and Madrid, and (6) New York and London could increase up to 15% under the proposed agreements. The Applicants claim substantial benefits will flow from an expanded alliance, but they have not shown that immunity is necessary to achieve these benefits. We therefore recommend that DOT impose conditions slot divestitures or carveouts, as appropriate on a grant of immunity to protect the public interest in competition.
American and British Airways founded the oneworld alliance in 1999. By 2008, oneworld had grown to include ten members operating flights to over 725 destinations worldwide and carrying nearly 330 million passengers.1 American, British Airways, and Iberia provide the vast majority of oneworld service between the U.S. and Europe.2
American which operates hubs in Dallas, Chicago, and Miami is the third largest carrier in the world with total revenues of about $23 billion in 2008. American serves 250 cities in 40 countries with more than 3,000 daily flights.3
British Airways is also among the worlds largest international airlines, serving 165 destinations in 77 countries. In 2008, British Airways carried over 33 million passengers and earned revenues of about $11 billion.4 British Airways operates a hub at Londons Heathrow Airport, the busiest of the five international airports in the London area.5 British Airways provides nonstop service to 17 U.S. cities from London, serving 15 of those from Heathrow.6
Iberia is the fourth largest European carrier, with 2008 revenues of about $4.5 billion.7 Iberia operates hubs in Madrid and Barcelona. The airline provides nonstop service from Madrid to Boston, Chicago, New York, Miami, Washington, D.C., and San Juan, Puerto Rico. On November 12, 2009, British Airways and Iberia reached an agreement to merge the two airlines.
Currently, without antitrust immunity, all oneworld members interact with one another with varying degrees of integration and across various markets. All oneworld members have agreed to provide alliance customers coordinated processes for reservations and baggage transfer, through-ticketing, frequent flyer reciprocity, and lounge sharing.8
In addition, American presently codeshares with British Airways and Iberia on numerous points behind and beyond their respective U.S., London, and Madrid gateways, with American placing its code on British Airways and Iberia flights to 64 destinations.9 With the exception of travel between Chicago and Manchester (a route operated by American), American and British Airways do not currently codeshare on their nonstop transatlantic routes. American codeshares on Iberia flights between the U.S. and Madrid, and on some Iberia flights beyond Madrid. Iberia places its code on American flights between the U.S. and Spain and some American domestic flights from Iberias U.S. gateways.10
American and British Airways sought antitrust immunity from DOT in 1997 and 2001. In 1997, American and British Airways were the two largest carriers providing service between the U.S. and Heathrow, and two of only four airlines allowed to serve those routes under the U.S.-U.K. Bermuda II Treaty. The 1997 immunity request anticipated new entry as a result of potential liberalization of the Bermuda II Treaty and a new open-skies treaty that would, among other things, remove restrictions on the number of airlines permitted to fly between the U.S. and the U.K. DOJ filed comments in that proceeding, concluding that the transaction raised significant competitive concerns. In particular, immunity would have eliminated nonstop competition between American and British Airways on six overlap routes between (1) Boston and Heathrow, (2) Chicago and Heathrow, (3) Dallas and Gatwick, (4) Los Angeles and Heathrow, (5) Miami and Heathrow, and (6) New York and Heathrow.11 DOT dismissed the application in July 1997 after it became clear that liberalization of the Bermuda II Treaty was unlikely.
In 2001, American and British Airways again requested antitrust immunity.12 Open skies still did not exist between the U.S. and the U.K. DOJ filed comments raising concern about five of the six markets at issue in the 1997 proceeding.13 DOJ recommended that DOT (1) require the airlines to divest slots to encourage a new entrant to offer service from Heathrow to Boston, Miami, and New York and (2) carve out from any grant of antitrust immunity coordination on routes between (a) Dallas and Gatwick and (b) Chicago and Heathrow.14 DOTs show cause order proposed to grant the application if (1) negotiation of an open-skies agreement succeeded and (2) American and British Airways agreed (a) to divest slots sufficient for 16 daily roundtrips from Heathrow and (b) to carve out from their agreements coordination on routes between (i) Chicago and London and (ii) Dallas and London.15 American and British Airways withdrew their application before DOT issued a final order.
In this proceeding, Applicants request immunity for a series of agreements. These include bilateral agreements between American and British Airways, Iberia, Finnair, and Royal Jordanian, and a multilateral coordination agreement among the five Applicants. Three Applicants American, British Airways, and Iberia also request approval of a Joint Business Agreement, which anticipates [REDACTED TEXT].16 Applicants contend that revenue sharing will promote metal neutrality and allow them to jointly tailor their service to serve customers better, rather than diverting passengers from one to another.17
Applicants claim that the various alliances included in the Joint Application will significantly improve customer convenience and choice, produce operating efficiencies and greater value for consumers.18 Specifically, they point to improvements in their ability to offer passengers connecting flights as the primary source of consumer benefits from antitrust immunity.19 Applicants maintain that the superior cooperation afforded by immunity will result in over $90 million in annual consumer benefits by way of lower prices.20
Applicants further state that the revenue sharing and closer integration called for by the alliance agreements will enable oneworld to compete more effectively against the immunized portions of the Star and SkyTeam alliances.21 Applicants claim that denying the Joint Application would doom oneworlds chances for long-term success in the alliance market place.22
Applicants also assert that the benefits of the proposed agreements justify unrestricted immunity,23 and that any carve-out would jeopardize the alliance and deprive consumers of substantial benefits.24 Finally, as in virtually every other application for antitrust immunity made to DOT, Applicants assert they will not proceed with the proposed alliance agreements without immunity, as the risk of litigation would be too great to bear.25
Congress has circumscribed the conditions for antitrust immunity for conduct like the proposed arrangement here that substantially harms competition. When a proposed arrangement substantially reduces or eliminates competition, DOT may approve and grant antitrust immunity when (1) the arrangement is necessary . . . to achieve important public benefits and (2) those benefits cannot be achieved by reasonably available alternatives that are materially less anticompetitive.26 The burden is on Applicants to justify their need for authority to engage in conduct that will likely restrict competition.
A grant of unrestricted immunity is likely to result in significant competitive harm in six transatlantic markets where American currently competes with British Airways and Iberia. Applicants documents and public filings [REDACTED TEXT],27 and American and Iberia are the only current nonstop competitors between Miami and Madrid. That competition would be lost if Applicants were to implement their agreements as proposed.
A joint venture is likely to harm competition if it would increase the participants ability or incentive to raise price or reduce output in any relevant market.28 The competitive effects of a joint venture in markets where the venture would eliminate competition (like the joint venture at issue in this proceeding) are similar to the competitive effects of a merger.29 DOJ, the Federal Trade Commission, and federal courts analyze the competitive effects of mergers using the principles contained in the Horizontal Merger Guidelines.30 DOT has previously cited this framework as well.31
Applicants proposed agreements will likely result in significant competitive harm in transatlantic markets where American currently competes on a nonstop basis with British Airways and Iberia.
In transatlantic routes covered by the Joint Application, nonstop service is a separate product from connect service. A variety of evidence supports this conclusion.
First, empirical analyses show that the number of nonstop carriers competing in a market has a significant impact on the average fares paid by customers.32
Second, nonstop fares on the six routes of concern are 28% higher on average than connecting fares.33
Third, in their internal models (referred to as quality of service index or QSI modeling34), airlines commonly assume that an airlines share of travel between cities will be higher if it offers nonstop service than if it offers connecting service.
Fourth, many businesses distinguish between nonstop and connecting travel. Numerous corporations have explicit guidelines governing when employees must consider onestop alternatives due to lower prices. Those guidelines often require a significant fare difference before the onestop option is mandated generally at least 10% and in some cases 25% or more. Some corporations actually require passengers to take nonstop service if available. This is not surprising given the value of employees time, especially for the types of employees likely to be dispatched on international travel.
As set forth in Table 1, Applicants compete on six nonstop transatlantic routes where they collectively possess about [REDACTED TEXT], even if connect service is included.
Source: Marketing Information Data Tapes35
As set forth in the Horizontal Merger Guidelines, the Herfindahl-Hirschman Index (HHI) is an aid to the interpretation of market data.36 As the HHI levels in Table 2 indicate, all these markets are highly concentrated now, and granting unconditional immunity will raise concentration to levels that the Horizontal Merger Guidelines presume would be likely to create or enhance market power:
Source: Marketing Information Data Tapes
In two markets Dallas to London and Miami to Madrid Applicants currently offer the only daily nonstop service. Applicants agreements would eliminate nonstop competition and result in very high levels of overall concentration in these two markets. In two other markets Boston to London and Miami to London Applicants are currently two of only three nonstop competitors and together would have over [REDACTED TEXT] shares of both markets.
In a fifth market Chicago to London implementation of Applicants agreements would reduce the number of nonstop competitors from three to two for part of the year and from four to three for the other part. American and British Airways currently compete against only two nonstop competitors: (1) United, which operates it largest U.S. hub at OHare in Chicago, and (2) Virgin, which only offers seasonal service.37 [REDACTED TEXT], Virgins single daily frequency during the part of the year when it is operating puts it at a disadvantage in competing for corporate customers.38
In a sixth market New York to London American and British Airways are two of the five carriers currently offering nonstop service and together currently control about [REDACTED TEXT] of the market for all passengers. That [REDACTED TEXT] share understates Applicants competitive significance for business travelers who typically pay substantially higher fares than the fares paid by coach passengers. American and British Airways have about [REDACTED TEXT] of nonstop business travel between New York and London, compared to [REDACTED TEXT] of other travel.39 Delta and Continental lag far behind American, British Airways, and Virgin in the share of the nonstop business traffic they carry between New York and London [REDACTED TEXT] respectively). This deficit is likely due to the fact that neither Delta nor Continental offers as many flights as the others, particularly at times business passengers prefer. Frequency, as well as time of day, can be an important driver for corporate contract selections and for individual business passengers.40 British Airways provides nine flights per day on this route. American and Virgin each offer five daily flights. Continental has three flights per day, and Delta has only two daily flights.41
If Applicants were to implement their agreements, it is likely that competition in the six, heavily traveled routes discussed above would be diminished significantly. Numerous economic studies of the domestic U.S. airline industry have shown that reducing the number of nonstop carriers in a market directly affects fares.42 A cross-sectional analysis of third quarter 2008 fare data for U.S. carriers on transatlantic routes similarly shows that (1) fares paid by nonstop passengers in markets with only one nonstop competitor are 15% higher than fares paid by nonstop passengers in markets with two nonstop competitors and (2) fares paid by nonstop passengers in markets with two nonstop competitors are 6.6% higher than fares paid by nonstop passengers in markets with three nonstop competitors.43
The point of these analyses is that nonstop competition regardless of the presence of connecting competition has a direct effect on nonstop pricing. All the routes studied by DOJ had connecting competition, yet the number of nonstop competitors still had significant price effects. Put another way, for a substantial number of nonstop consumers, connecting competition does not discipline price in any meaningful sense. DOJs findings are consistent with the empirical research (including research in peer-reviewed journals) showing that the number of nonstop competitors between a pair of cities has a significant impact on the prices paid by passengers traveling between those cities, especially where the number of nonstop competitors is three or fewer.44
Were Applicants to coordinate fares as proposed, passengers traveling between the six pairs of cities identified above (1) Boston and London, (2) Chicago and London, (3) Dallas and London, (4) Miami and London, (5) Miami and Madrid, and (6) New York and London would be likely to pay fares significantly higher than the fares they would pay were Applicants to continue to compete. Those increased fares are directly attributable to the loss of nonstop competition.45
Applicants have submitted an economic report criticizing findings submitted by DOJ in the Star Alliance proceeding about the competitive effect of the number of nonstop competitors and purporting to show that immunizing their nonstop overlap routes would not result in higher prices.46 The report rests on faulty data and contains econometric errors. As explained in Appendix B, which responds to Applicants submission, once those problems are corrected, Applicants empirical analyses show that the number of independent nonstop competitors substantially affects pricing on routes.
Applicants claim that because they intend to increase overall capacity, they will have no ability or incentive to raise local fares.47 Even if Applicants were to increase total capacity, they would still have the ability and incentive to increase fares for nonstop service in markets where implementing their agreements would result in increased market power. Applicants use highly sophisticated computer systems that allocate seat availability among alternative routes. These systems seek to maximize yield on each flight, and one way to maximize yield is to restrict capacity (that is, to limit seat availability) and thereby impose higher fares in markets where airlines have the ability to do so like nonstop, hub-to-hub routes where they face little nonstop competition. Lacking nonstop competition on a route, Applicants would have a strong incentive to limit seat availability and raise fares for nonstop travel, even on routes where they may add capacity to serve connecting passengers traveling between different pairs of cities.
A transaction is unlikely to have anticompetitive effects in a particular market if entry into that market is so easy that merging parties would be unable to raise price after their merger.48 Entry may prevent adverse competitive effects when it would be timely, likely, and sufficient in its magnitude, character and scope.49 To be sufficient, entry must replace the competition lost due to the transaction. For transactions that eliminate competition between two competitors with large market shares, entry by a new competitor that will obtain a small market share is unlikely to be sufficient.
For the following reasons, entry is unlikely to prevent Applicants from raising price.
In four of the markets (1) Chicago and London, (2) Dallas and London, (3) Miami and London, and (4) Miami and Madrid Applicants would have hubs at both ends, a competitive advantage no new entrant can match. Entry is thus unlikely because the entrant would neither have access to feed traffic from connecting routes nor enjoy the significant marketing advantages Applicants possess in their hubs.50
Entry is unlikely to prevent adverse competitive effects in any of the five markets of concern involving London. Entry is unlikely because slots at Heathrow the most attractive London airport for a potential entrant are difficult to obtain.51
Applicants documents [REDACTED TEXT].52 Entrants also prefer Heathrow because it offers more connecting opportunities than Londons other international airports.53 Offering connecting opportunities also appears to be an advantage in securing highly profitable corporate business traffic. Heathrows superiority as a connect point is supported by the carriers internal documents as well as DOJs interviews with corporate travel managers. Many of those travel managers said they preferred to contract with a carrier that serves Heathrow even if the corporation was indifferent as to which London airport its local travelers arrived at or departed from because Heathrow offered better connecting opportunities for their travelers who were heading to points beyond London.
Perhaps the most striking evidence of this preference is the fact that, after open skies, major carriers in large part shifted service between the U.S. and Gatwick to Heathrow. Few flights remain between the U.S. and Gatwick. Delta offers one flight to Gatwick and one to Heathrow from Atlanta, and offers one Gatwick frequency from Cincinnati (where it has no nonstop Heathrow competition). US Airways offers a daily flight to Gatwick from Charlotte (where it has no nonstop Heathrow competition). British Airways and Virgin operate a handful of daily flights to Gatwick, primarily from tourist destinations like Las Vegas, Orlando, and Tampa. Table 3 illustrates this shift:
Entry at Heathrow would be difficult.55 There are currently no pool slots available for new entrant transatlantic service at Heathrow, leaving entrants only the secondary market from which to secure slots.56 The slot coordinator for London airports has opined that Heathrow slots might be available in the secondary market if an entrant were willing to be flexible about timing of slots and accept commercially sub-optimal timings.57 Because time of day is important for arrivals from the United States,58 the likely availability of only sub-optimal slots at Heathrow is another factor discouraging entry.59
Although a number of carriers have gained access to Heathrow for service to U.S. cities since open skies took effect about a year ago, that does not show entry is easy.60 During the first open-skies season in the summer of 2008, there were 19 more daily flight frequencies between the U.S. and Heathrow than in the summer of 2007,61 with slots to accommodate these flights coming from a variety of sources, including outright purchases,62 slot leases,63 use by U.S. carriers of slots provided by a European alliance partner,64 and new frequencies self-funded by Heathrow incumbents shifting slots to U.S. routes from other markets.65
But, significantly, virtually all the new flights between Heathrow and the United States merely involve carriers that shifted service involving their hubs in the United States from Gatwick to Heathrow. There were two exceptions Northwest offered new service between Seattle and Heathrow, and Air France offered new service between Los Angeles and Heathrow but each airline abandoned its new route after less than a year. Thus, there are no examples of truly new entry into Heathrow from which to infer that entry is likely to prevent competitive harm were Applicants to implement their agreements.
Taken together, these factors suggest that entry is unlikely to prevent Applicants from raising fares in the five markets involving London.
Applicants must demonstrate that immunity is necessary to achieve the claimed public benefits of their agreements.66 Applicants, however, assign undue weight to these benefits67 and downplay their incentives to cooperate absent immunity.
Applicants make a variety of assertions about the passenger benefits that will flow from an immunity grant, including providing better integrated and more extensive networks through expanded codesharing,68 offering lower fares as a result of a reduction in double marginalization,69 and enhancing inter-alliance competition. As discussed below, the likely magnitude and value of any of these alleged benefits is overstated.
Americans modeling suggests that very few gains from new codesharing will result from passengers carried between small (non-gateway) cities in the U.S. and small (non-gateway) cities in Europe that is, the routes none of the alliance partners currently serve online. [REDACTED TEXT] .70 Because these are routes that by definition no alliance partner serves on-line today, we believe that any argument that antitrust immunity is necessary to allow Applicants to cooperate on these routes should be considered critically.
As the bulk of the routes alliances typically serve already have good service from the alliance partners, there is little incremental value from providing better connections to behind and beyond travelers using hubs on both sides of the Atlantic to reach their destinations. For example, Table 4 illustrates that the vast majority of passengers flying on a member of the immunized SkyTeam alliance travel either nonstop or on two segments. Specifically, only [REDACTED TEXT] of all SkyTeam marketed tickets involve three segments of travel. Among the two-segment tickets, [REDACTED TEXT] of passengers make their journey by traveling online and never connect to a flight operated by another Skyteam carrier.
Source: Marketing Information Data Tapes71
[REDACTED TEXT] .72 , Applicants were able to achieve much of what adds value from aligning without immunity. Notably, many of the routes upon which American and British Airways still do not permit the other to codeshare are routes .73
Applicants cite the elimination of double marginalization as a benefit of the proposed immunity grant.74 Although alliances can lead to lower fares by reducing incentives for each carrier to impose an additional markup on connecting traffic, immunity is not necessary to realize that result.
Previous economic studies of fares offered by international airline alliances in the 1990s did find that immunized alliance carriers charged interline fares that were lower than the interline fares charged by non-immunized alliance carriers.75 These studies, however, did not prove that airlines could eliminate double marginalization only by engaging in activity that raised antitrust concerns.
The 1990s were a time of flux for airline alliances as airlines experimented with different partner alignments and degrees of coordination and integration. Indeed, most of the immunized alliance relationships included in these earlier studies featured only minimal levels of revenue sharing.76 Empirical work suggests that in more recent years (after the period studied in those earlier papers), airlines participating in alliances whether immunized or not have learned to manage their inventories and pricing activities to provide more competitive fares when forming a connection with another airline.77 Specifically, using 2005-2008 data, DOJ has found that connecting fares offered by non-immunized alliances for transatlantic routes are no more expensive than fares offered by immunized alliances.78
Applicants economic experts challenge DOJs study comparing immunized connecting fares to other types of connecting fares.79 As set forth in Appendix B, Applicants experts do not control for important factors that can change fares on individual routes over time, and they also make numerous data errors.80
Applicants maintain they must have immunity for oneworld to achieve parity with the SkyTeam and Star alliances, which received immunity. Applicants also suggest that consumers benefit from competition between alliances, particularly immunized alliances.
Immunity is not necessary for effective alliance competition. oneworld on its website proclaims that it is already the preferred alliance, citing to the advanced level of cooperation among its members and the travel industry awards captured by the alliance. Yet its members are parties to only two immunized relationships, one between American and Finnair, the other between American and LAN. Indeed, few, if any, corporate travel managers whom DOJ interviewed have stated a desire for increased inter-alliance competition.81 Even when a particular corporation has a contract with an alliance, that contract seldom encompasses all members of the alliance or results in lower fares than if the corporation had negotiated separate contracts with the carriers.82 Quite simply, corporate customers are looking for well-executed service to specific cities provided at convenient times and competitive prices.83 We do not believe Applicants have demonstrated that antitrust immunity is a necessary prerequisite for offering any of these components.84
Applicants insist that they will not implement the proposed agreements without total immunity. Even assuming Applicants might not enter into alliances structured exactly as those set forth in the Joint Application, and that oneworld must attain parity with the Star and SkyTeam alliances, Applicants likely would engage in some sort of cooperation that would provide nearly identical benefits to consumers as those likely to result from the Joint Application agreements. Carriers routinely enter into commercial relationships with each other and make significant long term investments in such relationships, without immunity from the antitrust laws. Today, oneworld members cooperate on premium customer benefits, frequent flyer redemption, customer handling (for example, interline electronic ticketing, through baggage check-in), and codesharing (where regulatory conditions permit).85
If past is prologue, Applicants are likely to move closer together, within the bounds of the antitrust laws, even without immunity.86 In their failed 2001 application for antitrust immunity, American and British Airways stated they would not proceed with the alliance expansion absent immunity.87 Yet American and British Airways relationship has evolved substantially since that time. [REDACTED TEXT] 88
Given the risk of significant competitive harm, any grant of immunity should be limited to protect competition. Three types of remedies could potentially provide that protection: (1) unencumbered slot divestitures, (2) earmarked slot divestitures, and (3) carve-outs from immunity. DOJ sets forth some considerations about each below.
DOT could require Applicants to divest slots to encourage new entry. DOJ recommended this approach in the 2001 proceeding involving American and British Airways.
The advantage of this approach as a potential remedy is that it requires little continuing oversight and enables market participants to decide how to use slots most efficiently after their divestiture. The limitation of this approach is that the purchaser of divested slots need not use the slots on the overlap routes to replace the loss of competition caused by implementing Applicants agreements. Recognizing this complication, DOJ recommended in the 2001 proceeding that DOT order the divestiture of more than the number of slots necessary to remedy the loss of competition on the specific overlap routes to protect against the possibility that some of the divested slots would be used to serve other routes.
Even then, complications remain. Ordering too many divestitures would unduly interfere with market forces; while ordering too few would not replace the lost competition and would fail to protect consumers. The efficacy of this approach also requires some evidence that there are potential entrants that would serve the overlap markets if they had the ability to acquire divested slots.
Another potential remedy is ordering slot divestitures that are earmarked for use in the markets that would be harmed by Applicants agreements. DOT had indicated that it would order earmarked divestitures in the 2001 proceeding before the application was withdrawn, and competition authorities in other jurisdictions have ordered similar relief.91 If a new entrant were found, this remedy could restore the competition that would be lost as a result of Applicants agreements.
There are also some disadvantages to this approach. As DOJ noted in its comments regarding the 2001 proceeding, earmarking slots risks inefficien[cy] if market conditions change and the divested slots could be used more productively elsewhere. Any remedy of this type thus would require continuing oversight to ensure that it does not unintentionally harm consumers.
Carve-outs, which have been used to remedy harm in several alliances, are relatively easy to administer and preserve the incentives of the current market participants to compete. Carve-outs may be the only effective remedy on routes where slot divestitures would not lead to entry. DOJ continues to believe that carve-outs can be an effective means of preventing competitive harm notwithstanding Applicants criticisms, which are addressed below.
Increasing Capacity: Applicants maintain that carve-outs will reduce their incentive to increase capacity because they will not share revenues from the carved-out passengers. As an initial matter, even if they were to implement their agreements, Applicants would not have identical interests because they remain separate entities that do not share revenues on domestic travel or international travel involving parts of the world not covered by the joint venture.92 Moreover, alliances have long been successful in creating new connections and expanding capacity. Carve-outs will not change that dynamic here. Connecting demand can be expected to increase when oneworld partners expand codesharing in markets beyond their hubs.93 Additional capacity will be needed to accommodate these additional connecting passengers, evewithout immunity. In short, nothing about carve-outs destroys Applicants incentives to incapacity.
Yield Management: Applicants also claim carve outs would lead each carriers yield management system to place a higher value on local traffic over connecting traffic. Specifically, they assert that nonstop passengers will inefficiently displace connecting passengers who value the transatlantic seat most.94
If alliance carriers allocated seats based solely their own revenue, as Applicants claim, no alliance could properly function as long as each carrier continued to operate parts of its own network outside the alliance agreement. Suppose, for example, that American is considering how to allocate a seat on an Atlanta-Miami flight after the alliance is approved. American could give the seat to a local passenger on Atlanta-Miami, to a connecting passenger traveling Atlanta-Miami-Buenos Aires, or to an Atlanta-Miami-London passenger (whose fare it would share with British Airways). In the first two cases, American keeps all the revenue because these American routes are outside the alliance. If American evaluated the seats based on what it would earn on that seat given its agreements, it could easily rank the alliance passenger last because most of the revenue from the transatlantic leg is shared, whereas American keeps 100% of the revenue otherwise.95
All alliances face this issue, and others have resolved it by a simple expedient: rather than trying to evaluate each ticket strictly based on revenue each carrier expects to earn, alliance partners agree on how to map fares into each others yield management systems.96 In essence, they pretend that an alliance partners fare is just like their own fare. Using these mappings to make sure that the fares being evaluated are comparable, each carrier treats fares partly booked on the others metal as its own. (In this way, partners without immunity also appear to have overcome the double marginalization problem.) As long as both partners contribute substantial networks to the alliance, each expects that at the end of the accounting period, the increased revenue from the greater number of connecting passengers carried will more than offset the loss that might be individually incurred on any one decision.
Scheduling Benefits: Applicants assert that carve-outs would reduce oneworlds ability to optimize schedules and create additional connections by adding new flights or spreading out existing flights (that is, eliminating wing tip flights).
With respect to adding new flights, the experience of other alliances operating with carve-outs belies the claim that carve-outs prevent the addition of new capacity. Since 1996, United and Lufthansa have operated in the immunized Star Alliance with two carved-out routes: (1) Chicago and Frankfurt and (2) Washington and Frankfurt. Capacity has risen 75% on the route between Chicago and Frankfurt and 105% on the route between Washington and Frankfurt, and the number of local passengers traveling on the routes has remained virtually unchanged.97 These alliance partners did not share revenues during this period, so the carrier operating a new flight received all the revenue from not only the local passengers (as would be the case with oneworld) but also the connecting passengers. Recognizing that it was in their mutual interest to increase capacity on the hub-hub routes to capture connecting passengers, they added flights.
With respect to spreading out existing flights, Applicants fail to acknowledge that competing carriers offer flights at about the same time because that is the time of peak demand, particularly from local passengers.98 Passengers benefit from having competing options at the time of day they prefer to fly.
Applicants point to the experience of American and Swiss in 2004. Both offered one nonstop flight per day between JFK and Zurich at almost exactly the same time in the early evening. After they entered a revenue-sharing agreement, they agreed that American would move its flight to 9 p.m., a time that was less desirable for local passengers but was more convenient for passengers seeking to connect through Zurich. Applicants claim that, although the profitability of the American flight fell, overall alliance profitability and traffic increased.99 They note that when the alliance began to fall apart, American moved its JFK-Zurich flight back to the peak time. To complete the story, when American moved its flight back to the time preferred by more passengers, Swiss independently added a 9 p.m. flight at JFK to accommodate late evening demand from passengers seeking to connect through Zurich,100 calling into question the claim that revenue sharing on the local traffic was necessary for the addition of the later JFK flight.
Applicants' proposed alliance agreements would significantly harm competition on the most significant transatlantic routes, resulting in fare increases of up to 15% for some passengers. Although the proposed alliance agreements may lead to public benifits, Applicants overstate their magnitude and value, particularly in relation to their current unimmunized alliance. We therefore recommend DOT impose conditions - slot divestitures or carveouts, as appropriate - to protect the public interest in competition.
Date: December 21, 2009
Certificate of Service
I hereby certify that a copy of the foregoing COMMENTS OF THE DEPARTMENT OF JUSTICE have been served this day by e-mail upon each of the following addresses:
2 Applicants Finnair and Royal Jordanian provide a very limited amount of transatlantic service. Finnair serves only the Helsinki-New York route, while Royal Jordanians transatlantic service to the U.S. is from Amman, Jordan, not Europe. The remaining members of oneworld Cathay Pacific, Japan Airlines, LAN, Malev, Qantas and its most recent addition, Mexicana do not offer service between the U.S. and Europe.
5 In addition to Heathrow, passengers traveling through London can also access international service at the Gatwick, Stansted, Luton, and London City airports (listed in descending order of passengers served).
10 Independent of its relationships with the other oneworld carriers, American has separate bilateral immunized relationships with Finnair and LAN. American also has non-immunized codeshare relationships with several non-oneworld carriers, including Alaska Airlines, Brussels Airlines and Gulf Air.
13 Comments of the Department of Justice, December 17, 2001 (U.S.-U.K. Alliance Case, Docket OST-2001-11029) at 31-32. DOJ did not raise concerns that passengers traveling between Los Angeles and Heathrow would be harmed by the proposed 2001 arrangements.
30 Fed. Trade Commn & Dept of Justice, Horizontal Merger Guidelines § 0.2 (1997), available at http://www.justice.gov/atr/public/guidelines/hmg.htm; see also, e.g., Chi. Bridge & Iron Co. v. F.T.C., 534 F.3d 410, 431 (5th Cir. 2008) (Merger Guidelines are often used as persuasive authority when deciding if a particular acquisition violates anti-trust laws.). For a more detailed description of DOJs approach to analyzing airline mergers, see the Statement of James J. OConnell before the Subcommittee of Aviation, Committee on Transportation and Infrastructure, U.S. House of Representatives (May 14, 2008) at 7-10, available at http://www.justice.gov/atr/public/testimony/233151.htm.
31 See Order 2009-4-5 at 7 (Docket OST-2008-0234) ([W]e primarily consider whether the alliance would significantly increase market concentration, whether the alliance raises concerns about potential anticompetitive effects in light of other factors, and whether new entry into the market would be timely, likely, and sufficient either to deter or counteract a proposed alliances potential for harm.).
34 To model the benefits of an alliance, airlines typically use QSI models to forecast traffic changes associated with better connectivity and codesharing arrangements. The basic QSI model is best described as a market-share allocation model that uses airline schedules as a basic input. Prices are assumed not to vary from historical levels. Given a fixed market size, passengers are assigned based on relative attractiveness of different airline offerings. For example, the QSI model assigns a higher value to nonstop flights than connecting alternatives.
35 Table 1 is based on data for 2008 Q2-Q4 (post-open skies), except all data for 2008 are used for Miami and Madrid. Air India has since exited both Chicago and New York. London includes Heathrow, Gatwick, and Stansted airports. New York includes JFK and Newark airports.
In 2001, DOJ alleged that nonstop flights between the United States and Heathrow were in different product markets than nonstop service between the United States and other London airports. See 2001 Comments, supra note 13, at 20-26. For purposes of analyzing Applicants current arrangement, it is not necessary to determine whether service to Heathrow is a separate market, although the evidence discussed in Section IV.B.4 below suggests that it is. Aggregating London airports neither changes carrier shares significantly nor alters the entry analysis.
37 Virgin began serving the route with one flight per day in April 2007. It reduced its service to five flights per week in the winter of 2008, and it cancelled all service on the route during the winter of 2009.
41 Although Deltas two daily frequencies depart from New York in time to arrive at Heathrow before 10 a.m., only two of Continentals three Newark-London flights arrive in London during the morning arrival period favored by business travelers. In contrast, three of Americans five daily flights reach London before 10 a.m., as do six of British Airwayss nine daily New York departures. [REDACTED TEXT] Id. at 1737.
42 See, e.g., Kamita, Analyzing the Effects of Temporary Antitrust Immunity: The Aloha-Hawaiian Immunity Agreement, Journal of Law and Economics (forthcoming 2009); Peters, Evaluating the Performance of Merger Simulation: Evidence from the U.S. Airline Industry, 49 Journal of Law and Economics 627 (2006); Joskow, Werden & Johnson, Entry, Exit and Performance in Airline Markets, 12 International Journal of Industrial Organization 457 (1994); Borenstein, The Evolution of U.S. Airline Competition, 6 Journal of Economic Perspectives 45 (1992); Borenstein, Hubs and High Fares: Airport Dominance and Market Power in the U.S. Airline Industry, 20 Rand Journal of Economics 344 (1989); Brueckner, Dyer & Spiller, Fare Determination in Hub and Spoke Networks, 23 Rand Journal of Economics 309 (1992); Morrison & Winston, Enhancing Performance in the Deregulated Air Transportation System, 1989 Brookings Papers: Microeconomics 61 (1989).
43 The fare-change findings for two-to-one routes are statistically significant. See Appendix A, Section I, for a description of the analysis performed. (The findings in Appendix A mirror those detailed in Appendix B to the Comments of the Department of Justice to the Show Cause Order, June 26, 2009 (Docket OST-2008-0234).) DOJs finding that the number of nonstop carriers on a route affects price is consistent with an affidavit from the Brattle Group submitted by Applicants. The Brattle Group estimates [REDACTED TEXT] J.A., Ex. 29.
45 See Appendix A at Section I (describing empirical evidence showing that a reduction in the number of competing airlines offering nonstop transatlantic flights may result in large, statistically significant price increases).
47 Supplemental Comments, supra note 24, at 4-6. Applicants offer examples of post-immunity capacity increases on hub-hub bridge routes in other alliances. Notably, however, they fail to address what happened to the fares paid by nonstop passengers traveling on such routes after those immunized carriers added capacity.
50 The hub carriers frequent flyer base and relationships with travel agents make it difficult for an entrant to attract local passengers. See Gurrea, International Airline Code Sharing and Entry Deterrence, 1 Competition Policy and Antitrust 109 (2006); Lijesen, Nijkamp, Pels & Rietveld, The Home Carrier Advantage in Civil Aviation, 1 Competition Policy and Antitrust 215 (2006).
In DOJs empirical analysis of the 46 transatlantic hub routes that had nonstop service from one or two carriers, there was only one instance in the past three years of a non-hub carrier entering with regular service. See Appendix A, fn. 6.
51 Facility constraints pose less of a barrier to entry on the route between Miami and Madrid. While American and Iberia each have a hub at one end, neither of these airports has the access issues associated with Heathrow. Air Europa a non-immunized member of the Star Alliance with a small hub in Madrid recently began selling tickets for the five times per week flights its plans to start offering on the route in March 2010. Air Europa has little previous experience competing for U.S. business customers (the carrier only recently commenced service in JFK-Madrid).
53 See, e.g., AA-DOJ-CID#25295-AA-BA_0101592 [REDACTED TEXT]; Competition Commission, BAA Airports Market Investigation, Provisional findings report, 20 August 2008, Appendix 2.1 (finding that bulk of traffic at all London airports is on international service, but about a third of the passengers significantly above the level of the other London airports change aircraft at Heathrow).
54 Frequency counts are for U.S. originating flights to Gatwick or Heathrow. Frequency counts are by carrier and airport where a carrier has at least 60 flights in a quarter to count as having at least one frequency. New York includes carriers serving Newark and JFK airports. The list includes all routes with a net change of at least one service between the two London airports. Routes from U.S. cities with no change in service (Cincinnati, Charlotte, Las Vegas, Orlando, and Tampa) are not included.
55 In addition to open skies, another significant development since the 2001 proceeding is the opening at Heathrow of a new terminal, Terminal 5. The availability of that new facility has provided some added flexibility for accommodating the additional widebody service needed for long haul flights. However, the building of Terminal 5 was conditioned on a cap of 480,000 air transportation movements per year which ACL (the Heathrow slot coordinator) now views as the primary constraint on new service. This capacity constraint will remain until the 3rd runway opens in approximately 2018. ACL and BAA Response to Transportation Department Questions on Access to Slots and Facilities at London Airports, January 28, 2009 (Docket OST-2008-0252) (ACL Response).
59 The ACL and BAA stated that although a determined entrant might be able to obtain Gatwick slots on the secondary market, they did not see evidence of demand for new transatlantic services at Gatwick. ACL Response.
66 See, e.g., Competitor Collaboration Guidelines, supra note 33, § 3.36 (explaining that the proponents of a potentially anticompetitive collaborative agreement have the burden of showing the agreement is reasonably necessary to achieve cognizable benefits and there is no less restrictive means of achieving those benefits).
71 Data: Transatlantic passengers in MIDT, 2008 Q3. SkyTeam passengers are those with any combination of Delta, Northwest, Air France, Alitalia, and KLM as marketing carriers. On-line passengers traveled the entire itinerary on the same operating carrier.
73 Applicants also cite potential new nonstop service as a public benefit, specifically referencing the possibility of adding nonstop service [REDACTED TEXT] . See J.A., Ex. 13. While new service would likely benefit consumers, the promise of new service does not depend on the immunized status of the alliance. For example, [REDACTED TEXT]. See AA00965-996 at 980 [REDACTED TEXT]; AA-BA_0100743-789 at 754 [REDACTED TEXT] flyopenskies.com/docs/press/openskiesusec-golive.pdf. [REDACTED TEXT]. Moreover, American itself has argued in the past that simple codesharing and non-immunized joint marketing will support new nonstop service . . . just as effectively as eliminating all competition between [alliance partners]. Answer of American Airlines, Inc., August 9, 2006, Docket OST-2005-22922 at 4.
75 See Brueckner, J., and Whalen, T., 2000, The Price Effects of International Airline Alliances. Journal of Law and Economics, Vol. 43, pp.503-545. Brueckner, J, 2003, International Airfares in the Age of Alliances, Review of Economics and Statistics, Vol. 85, pp.105-118. Whalen, T., 2007, A Panel Data Analysis of Code-Sharing, Antitrust Immunity, and Open Skies Treaties in International Aviation Markets. Review of Industrial Organization, Vol. 30, pp.39-61.
77 [REDACTED TEXT] . AA-DOT-0004842-52. Furthermore, oneworlds website states that interline revenues have grown by a greater percentage than online. Those results seem unlikely if these fares still suffered from double marginalization.
95 Similarly, British Airways yield management system likely would favor a Glasgow-London-Tel Aviv passenger (for which British Airways would keep all the revenue) over a Glasgow-London-Dallas passenger (for which it would share revenue with American).
96 Applicants use such fare mapping methods today. See, e.g., AA 02544-46 [REDACTED TEXT] ; Joint Applicants Consolidated Response to Order 2008-12-11 at 12 (As members of oneworld, the JBA participants already have full fare class mapping. Thus, fare combinability will require only the addition of a rule in each carriers tariffs. . . .).