COMMENTS OF THE DEPARTMENT OF JUSTICE
December 17, 2001
TABLE OF CONTENTS
The United States Department of Justice ("DOJ") hereby submits comments on the applications of (1) American Airlines, Inc. ("American" or "AA") and British Airways Plc. ("British Airways" or "BA") under 49 U.S.C. §§41308-09 for approval and antitrust immunity for an alliance agreement between the two air carriers, (the "AA/BA transaction") and (2) United Airlines, Inc. ("United" or "UA"), British Midland Airways Limited ("bmi"), Austrian Airlines, Österreichische Luftverkehrs AG ("Austrian"), Lauda Air Luftfahrt AG ("Lauda"), Deutsche Lufthansa, A.G. ("Lufthansa" or "LH"), and Scandinavian Airlines System ("SAS") for approval and antitrust immunity for alliance agreements among those air carriers (the "UA/bmi transaction"). Under the applicable statute, the Department of Transportation must evaluate the competitive harm of a transaction, but may "trade" competitive harm in some markets for other public benefits, including improved competition in other markets, that are made possible by the transaction, and that cannot be achieved by less anticompetitive alternatives.
I. Introduction and Summary of Position
Capacity and pricing in U.S.-London markets have for decades been severely restricted and distorted by the United States' bilateral aviation treaties with the United Kingdom (Bermuda II and its predecessors), especially for service to and from London's Heathrow Airport ("LHR"). It has long been a goal of Department of Transportation ("DOT") to replace the Bermuda II bilateral with an Open Skies regime with no regulatory constraints on pricing and entry, and DOJ strongly supports that effort.
Replacing Bermuda II with Open Skies would eliminate a number of significant legal restrictions on free and open competition in the U.S.-U.K. aviation market. Bermuda II is among the most restrictive of the United States' bilateral aviation accords, severely limiting new entry between the U.S. and London's two major airports. It allows only four carriers (AA, BA, United and Virgin Atlantic Airways) to provide any service between the U.S. and LHR, and even those carriers may serve only ten U.S. cities. No LHR service to the U.S. may be provided by British Midland, Continental, Delta, Northwest, USAirways or any other North American or European airline. The bilateral further limits the total number of U.S. cities that may be served from London's Gatwick Airport ("LGW"), and Bermuda II also permits either government to place certain restrictions on capacity expansion in any U.S.-U.K. city pair, or to restrict carrier pricing in certain respects. The U.K. government has exercised these rights by imposing limits on connect carriers' freedom to undercut nonstop carrier fares from gateway points or to offer low fares to interior points in the U.S. or U.K. Because achieving Open Skies is related to approval of an immunized alliance for BA, these applications provide an opportunity to eliminate these legal restrictions and potentially enhance competition on many routes from the U.S. to London.
For consumers to realize the benefits of Open Skies and enhanced competition, however, DOT must be careful to pursue the goal it originally outlined during its last review of the AA/BA application. As DOT made clear then, its evaluation of the public interest required that it be satisfied that there was de facto, in addition to de jure, Open Skies with the United Kingdom before it would consider antitrust immunity. DOT defined "de facto Open Skies" to "include adequate provision for new and expanded U.S. carrier service through London airports, particularly Heathrow." DOT also stated in the context of the last application that "the ability of U.S. carriers to provide such service [at London] notwithstanding the constraints at Heathrow would be a critical consideration in our evaluation of the proposed Alliance." Order 99-7-22 at 2. The same reasoning should apply in this proceeding.
While consumers could clearly benefit from the removal of the Bermuda II restrictions, those benefits should not be purchased with a loss of competition where that loss can be avoided or mitigated by appropriate remedies. Approval of the AA/BA transaction threatens a substantial loss of competition and higher prices for a large number of consumers. Thus, without conditions to mitigate the harm, we would oppose the AA/BA transaction as we did three years ago.(1) We do not oppose including immunity for the UA/bmi transaction as part of an otherwise beneficial trade. Approval of the UA/bmi alliance presents no appreciable harm relative to the status quo because bmi is currently not an actual or potential competitor in U.S.-London markets -- it is prohibited by Bermuda II from operating to the U.S. from its LHR base.
Currently, AA and BA compete on a nonstop basis in six city pairs, and approval of the AA/BA alliance would significantly increase concentration in New York-LHR, Boston-LHR, Miami-LHR, Chicago-LHR, Los Angeles-LHR and DFW-London. In each of the markets except Los Angeles, the parties' combined shares of both frequency and time-sensitive business passengers exceed 50%, and they are the two largest carriers by these measures. Even if Bermuda II entry restrictions were removed, the scarcity of slots and related facilities at LHR, as well as network-related entry constraints, will make competitive entry unlikely into any of these markets except Miami-LHR.(2)
Divestiture of well-timed LHR slots and related facilities sufficient to permit nine new daily round trips by new entrants could substantially remedy the competitive harm in NYC-LHR and BOS-LHR. Those two markets have well-positioned potential entrants that would be likely to offer service if they had sufficient slots and facilities at LHR. But such slots and facilities are not likely to be available in a timely manner from any source other than divestiture from the parties. Moreover, unless divested slots are "earmarked" for particular markets (which we recommend against as inefficient), it is likely that slots for more than nine new dailies would have to be divested if DOT is to assure that existing competition in the New York and Boston markets will be preserved.
Even if LHR slots and facilities were available, there are no other new airlines well-positioned to enter and compete with the combined AA and BA in DFW-LON or ORD-LHR. In DFW, only AA and BA provide nonstop service today, and new entry on this hub-hub route is very unlikely. In ORD-LHR, only UA (and its new partner bmi) will offer competition. New entry is highly unlikely with approval of both alliances -- ORD-LHR would become a hub-hub route for both incumbents. The only condition that could mitigate the competitive harm in these markets is to withhold immunity from ("carve out") these routes. Even with carve outs, however, some residual harm will likely remain.
Entry is also unlikely in LAX-LHR, even with slot divestitures, although not as unlikely as for ORD and DFW. The competitive harm in LAX-LHR, however, is less certain than for the other overlaps, and it also may not be as large. Given this, DOT could reasonably decide to offset the potential for harm on the LAX-LHR route by providing one or more slots for more competition in other routes that might benefit from de facto Open Skies.
Consumer benefits specific to the transactions may be weighed against the unremedied competitive losses. Alliances potentially offer the benefit of lower prices and improved service in connecting markets that currently do not receive competitive on-line service. Although some consumer benefits in connecting markets can be expected here, the total level of those benefits is substantially smaller than portrayed by the parties.
The primary public interest benefit from the transactions is the fact that they may be the currency needed to purchase Open Skies. But Open Skies, by itself, offers limited benefits for U.S. consumers. Approval of both AA/BA and UA/bmi aligns the only two U.S. carriers currently permitted to operate at LHR (and that have the slots and facilities to do so) with the two largest slot-holding carriers at LHR. Because LHR is a severely slot constrained airport, entry by other carriers serving the U.S. in a post-Open Skies world would be minimal. For Open Skies to provide significant consumer benefits, removal of the legal prohibitions of Bermuda II must be accompanied by meaningful access to Heathrow for airlines serving the U.S. Such access requires additional slots and related facilities over and above the divestitures needed to cure the competitive harm created by the AA/BA transaction. Failure to provide additional access at LHR would tend to "lock in" a market structure that is largely a product of Bermuda II's restrictions on competitive service.
DOT should be mindful of the timing of access when balancing the harms and benefits of the alliances. Achieving de facto Open Skies now rather than at a future time is a cognizable benefit of the current proposal, but no proposal will enable entry to be instantaneous. The likely harm from immunizing AA and BA, however, will be much more immediate. Thus, DOT should seek a solution that will limit immunity (and the harm that might arise from it) until other carriers are able to actually begin operating at LHR to offset that harm. If Open Skies can be achieved with an appropriately conditioned AA/BA alliance, and if de facto Open Skies (as well as the conditions that address the competitive harms) will go into effect concurrently with the commencement of immunized operations, then U.S. consumers could benefit greatly from Open Skies in conjunction with approval and immunity for the AA/BA transaction.
II. Analytical Standards
Under the applicable statute, the Secretary must disapprove a proposed agreement if it "substantially reduces or eliminates competition," unless the Secretary finds that the agreement is necessary to meet a serious transportation need or to achieve important public benefits" and there is no less anticompetitive alternative. 49 U.S.C. §41309(b).
As explained more fully in our 1998 Comments (AA/BA I, DOJ Comments at 5-7), the appropriate framework for evaluating the competitive effects of an alliance is merger analysis. Our analytical framework as described in the Merger Guidelines involves identifying the relevant markets in which the firms compete, identifying the firms that compete in those markets, and measuring concentration. Where a transaction results in high concentration in a relevant market, DOJ undertakes a competitive effects analysis to determine whether the remaining competitors will find it profitable to increase prices or reduce the quality of their service after the merger. In examining likely competitive effects, DOJ considers factors such as entry barriers to evaluate the likelihood that high concentration will lead to increased market power in the particular circumstances of the markets at issue.
The Effect of the Events of September 11. For the most part, these Comments assume that the competitive situation will not be fundamentally altered by the September 11 attacks and their aftermath. AA and BA likewise have advised DOT that they believe this is the appropriate assumption for purposes of analyzing the proposals.(3) Although demand and carrier capacity have dropped significantly, our assumption is that, with a few exceptions discussed below, traffic and capacity will return to prior levels within the next year or two. However, predictions of the effects of the attacks on demand, costs and carrier viability are necessarily uncertain at this point.
To the extent that DOT finds that our assumptions about how quickly demand will return to "normal" are overly optimistic, DOT should be more cautious about approving this agreement. In those circumstances, competitive entry would be less likely, and the combinations of relatively strong, well-positioned competitors would be a greater threat to competition. Moreover, if a protracted drop-off in demand and an increase in cost leads to significant industry consolidation, the combination of AA and BA -- two of the largest and most financially stable airlines in the world -- becomes even more problematic, and the time horizon for replacement of the lost competition by new entry becomes even longer.
III. Competitive Analysis
United is one of the four largest airlines in the world and one of the four largest U.S.-U.K. competitors. As of the summer, 2001, UA provided nonstop service to London on at least a daily basis from seven U.S airports. All of its London service is at LHR, which it serves with a total of 15 daily frequencies, roughly the same frequency level as provided by American. Bmi is a regional airline based in the U.K. It does not currently provide any service between the U.S. and London,(4) but it is the second-largest carrier at LHR (in terms of slot holdings and daily operations) after British Airways. It serves roughly 20 points in the U.K. and Europe from LHR.(5) The other parties to the UA/bmi transaction (Lufthansa, Austrian, Lauda, and SAS) are European carriers that do not operate any service between the U.S. and London, although they do provide some service from LHR to cities in their home countries. Thus, there is no significant competitive overlap among the parties to the UA/bmi transaction in any U.S. markets.
American and British Airways are two of the four largest airlines in the world and two of the four largest competitors in markets connecting the U.S. and the U.K.(6) BA is the largest U.S.-London carrier by any measure. As of the summer, 2001, BA provided nonstop service to London on at least a daily basis from 21 U.S airports. It served LHR from 11 U.S. airports with a total of 26 daily frequencies. AA is the third-largest U.S.- London carrier in terms of cities served, and second-largest in terms of frequencies. As of the summer, 2001, American provided nonstop service to London on at least a daily basis from nine U.S airports, and it served LHR from six U.S. airports with a total of 15 daily frequencies.
Nonstop Overlaps. As was the case in 1998 when DOJ evaluated AA/BA I, American and British Airways compete on a nonstop basis in six city pairs: New York ("NYC"), Boston ("BOS"), Miami ("MIA"), Chicago ("ORD"), Los Angeles ("LAX") and Dallas ("DFW") to/from London. Exhibit DOJ-1 sets forth the competitive daily frequency levels for each nonstop overlap city pair in the year 2000, and with the post-September 11 frequency adjustments. The exhibit also notes entry and exit since 1998. Since 1998, the AA and BA frequency shares in the overlap routes have eroded somewhat, as other carriers have grown faster than the parties. Most notably, United entered with one frequency in BOS-LHR in 1999, and had (prior to September 11) expanded its service in ORD-LHR to three frequencies per day from two in 1998. In addition, Virgin Atlantic Airways ("Virgin" or "VS") has added one net U.S.-Heathrow frequency (increasing its New York service) since 1998.(7) Aside from the New York service, Virgin has merely moved one LHR frequency among several markets: from Miami (replaced by MIA-LGW) to Chicago, and then (after September 11) to BOS.(8)
To obtain passenger share information by carrier, DOJ has used bookings data provided by American from all the major computer reservations system vendors. The booking share data for the nonstop carriers(9) in each overlap route for the year 2000 are set forth in Exhibits DOJ-2A through DOJ-2F.(10) (11) The premium share data in these exhibits is for nonstop passengers traveling in first or business class (F, J, and C class tickets), which represent a reasonable proxy for the time-sensitive business traveler market.(12) Although these passengers account for only a small number of passengers on the plane, they account for a very large percentage of the total revenue from each flight. The exhibits also include each nonstop carrier's data for all passengers (regardless of booking class) traveling between the two cities, including those passengers that connect using one intermediate point. In general, AA and BA command a much larger share of the lucrative premium market than their share of all passengers would suggest. Moreover, AA and BA are the number one and two carriers of time-sensitive passengers (as measured by premium traffic) in five of the six overlap markets.
Competition for Corporate Contracts. Airlines are increasingly using corporate discount programs to compete for high yield business traffic, and a substantial amount of business travel is now at corporate fares. For example, American estimates that between one-half and two-thirds of its premium passengers in the overlap routes are traveling under corporate contracts.(13) AA and BA are significant competitors in the US-London markets and competition for corporate contract business is an important manifestation of that competition. [REDACTED TEXT]
To determine the competitive effects of the AA/BA transaction, it is first necessary to identify the relevant markets in which they compete. City pairs are relevant markets for the analysis of most airline mergers. Passenger demand is almost always specific to a given city pair; that is, passengers would not consider substituting travel in other city pairs if the price in their desired city pair increased by a small but significant amount. This conclusion is confirmed by the numerous examples of large price disparities between city pairs that are otherwise similar.(14) If groups of city pairs were part of larger markets, such price disparities likely would not persist.
For some passengers the market may be smaller than a city pair -- one-stop service may not be a reasonable substitute for nonstop service and service at one airport in a city may not be a reasonable substitute for service to another. If an airline can identify and price separately for those passengers, then service to them constitutes a separate market. Here, the evidence reveals that nonstop service in the relevant city pairs constitutes an antitrust market -- there is a large group of time-sensitive passengers who are willing to pay substantially more money for the convenience of nonstop service. The evidence also shows that service to LHR is a separate market from service to LGW for time-sensitive passengers.
In 1998, the evidence showed that passengers paying the highest fares overwhelmingly chose nonstop service, that corporate travel policies allowed employees to pay more for nonstop service and that businesses valued their employees' time and convenience. It also showed that airlines used fare restrictions and conditions to target these time-sensitive passengers. Based on that evidence, we concluded that nonstop service for such passengers was a separate market. The evidence today is much as it was three years ago and thus, our conclusion remains the same.(15)
Academic studies of fares in domestic markets(16) have shown that the number of nonstop competitors in a market has a significant effect on average fares in the market. Because reasonable connecting service is widely available domestically, if connecting service acted as an effective discipline on nonstop prices, then the entry or exit of a nonstop competitor should have little effect on fares. These results suggesting that connecting service does not discipline nonstop prices are consistent with DOJ's experience in investigating and analyzing recent airline transactions. Although these results were based on domestic markets, there is no reason to believe that international markets are systematically different in terms of the interaction between nonstop and connecting service. In fact, many of the parties' arguments, particularly those about changes in corporate travel policies which are discussed in detail below, should affect domestic as well as international travel, and thus, the effect of those policies on average fares are included in the prior studies. See, infra.
The available data suggests that very few nonstop passengers paying business fares appear willing to substitute to one-stop connections in most markets. Exhibit DOJ-3 summarizes the relative shares of nonstop and one-stop premium passengers on the six overlap routes. In all markets except Dallas (discussed below), the share of passengers in premium cabins choosing to fly nonstop exceeds 90%.
Exhibit DOJ-4 offers a more detailed breakdown of the data from Exhibit DOJ-3. The data show, contrary to AA/BA arguments,(17) that fewer than 1% of travelers on these routes choose to connect at any major European hub.(18) Connections at U.S. hubs also attract very few premium passengers in the city pairs other than DFW. Connect service to LHR attracted no more than 3% of the premium traffic in any of those city pairs. Connecting service to LGW garnered less than 1% of the traffic in each of those markets, as might be expected given the strong preference of time-sensitive passengers for service to LHR.
The Dallas-London market has a much higher share of connecting traffic than the other markets, but two-thirds of the connect traffic in this city pair is to LHR, which cannot be served on a nonstop basis from DFW. As shown in Exhibit DOJ-5, 16% of Dallas-London passengers traveled by connection, but 11% of the traffic used connecting services offered by the Heathrow carriers -- 10% on AA and 1% on UA.(19) This traffic pattern is thus far more indicative of passenger preference for LHR than of lack of preference for nonstop service.(20) It also demonstrates the strong pull the hub carrier has on premium passengers.(21)
AA and BA argue that corporate purchasers' increasing price sensitivity means that nonstop service is not a distinct market. In support, they cite a recent survey by American Express on corporate purchasing practices finding that (1) roughly two thirds of all corporations surveyed now have written corporate travel policies in place, (2) 76% of those have some type of policy in place requiring employees to use the lowest "logical" fare, and (3) 46% of that group include consideration of connecting options in determining lowest fare. They also argue that corporations are increasingly encouraging their employees to purchase discount fares with restrictions, and that they are increasingly negotiating discounts with airlines.(22)
But in fact, the evidence on corporate travel policies supports a conclusion that nonstop service is a separate market. The parties portray the American Express survey as showing that many corporations require use of connecting flights any time the connecting fare is cheaper. The survey, however, merely shows that some of the corporations surveyed include connecting fares in considering what is the lowest "logical" or convenient fare. Interviews with corporate travel managers disclose that corporate travel policies seldom require the use of connecting service over nonstop unless there is a large differential in price. For example, one corporation with more than $1 million in US-London travel annually from each of the parties explained to DOJ that, although its travel policy requires travelers to consider connecting alternatives, it does not require justification from a traveler who chooses more expensive nonstop service over connecting alternatives unless that price difference exceeds $500.
Moreover, contrary to suggestions made by both DOT (in prior cases) and the parties that the added circuity of connections becomes less significant with long-haul flights,(23) many corporations instead become more sensitive to their employees' convenience and are willing to pay large premiums to minimize that inconvenience. Thus, even among companies with strict travel policies, many permit their employees to incur the large added expense of business class where international flights, or flights in excess of 6-8 hours are involved.(24) The heavy usage of business class in U.S.-London markets, despite large premiums for such service, illustrates this willingness of large corporations to minimize traveler inconvenience.
DOT has suggested elsewhere that the increasing use of corporate discount contracts and frequent flyer programs have reduced the importance of nonstop service to business travelers.(25) As discussed in our prior comments, these factors in fact reinforce the preference for nonstop service. AA/BA I, DOJ Comments at 10-12. Interviews with corporate travel managers reveal that a key consideration in choosing a preferred provider is whether the airline offers nonstop service on the routes where the corporation has significant levels of travel. If their primary selected carrier does not serve an important route nonstop, the corporation often signs contracts with other carriers who do. Likewise, business travelers tend to favor the frequent flyer program of the carrier that gives them the most opportunities to amass and redeem miles, i.e., the carrier with the most nonstop service from their home airport.
AA/BA argue that nonstop carriers have no ability to use fare restrictions to price discriminate against time-sensitive passengers who prefer nonstop service because enough of those passengers would switch to restricted fares after a price increase to make the increase unprofitable.(26) According to the parties, since many business passengers currently do not fly on full unrestricted fares (some travel on restricted fares, some travel on corporate contract fares) any attempt to increase unrestricted nonstop fares would be unprofitable because passengers would switch to connecting service or restricted nonstop fares. In essence, the parties would have DOT believe that the complicated pricing structures and sophisticated yield management systems that the carriers have constructed (at great cost) to allow them to segment demand and discriminate between business and leisure passengers are ineffective.
That is not the case: airlines have substantial ability to price discriminate among passengers. For example, knowing that many business passengers make reservations at the last minute, the carriers have placed significant advance purchase restrictions on low fares. It does not follow from the fact that some business passengers are at times able to satisfy fare restrictions that high yield, low elasticity passengers could or would switch to restricted fares in response to a price increase. Sophisticated yield management systems provide the airlines with additional techniques to differentiate among customers with differing levels of price sensitivity, for example by closing low fare buckets at peak business travel times. Moreover, contrary to the parties' assertion, the proliferation of unpublished fares(27) in fact increases the degree to which an airline can discriminate among customers by allowing the airlines to assess and respond on a case by case basis to a traveler's price sensitivity.(28)
The AA/BA regression analyses(29) that purport to show substantial switching among fare classes in response to changes in relative fares are seriously flawed and cannot be used to support any conclusions. The most significant of these flaws are: (1) although the regressions purport to measure the relative response of bookings in various classes to changes in relative prices, the parties do not use data on actual prices, but only the published fares;(30) (2) the analysis relies on estimates of elasticity for business class passengers that are based on coefficients in regressions that are not statistically significant;(31) and (3) the analysis attempts to use the data to estimate the diversion between nonstop and one-stop passengers in response to relative price changes, notwithstanding the parties' own observation that in their data, published fares for one-stop and nonstop service between the U.S. and London do not differ significantly from each other.(32)
The critical loss and critical elasticity analyses advanced by AA and BA(33) are also deeply flawed. Moreover, because they contemplate switching of passengers among nonstop service providers as well as from nonstop to connect service, the studies do not provide any useful evidence on the question of whether fares for connect service discipline nonstop fares.
In 1998, the evidence showed that the vast majority of business passengers preferred LHR to LGW and that LHR airline yields were strongly and consistently higher than LGW yields. As a result, we concluded that London Heathrow and London Gatwick were in separate markets for business passengers. Three years later, the evidence -- and thus our conclusion -- remains the same.
The evidence shows that premium passengers overwhelmingly go to LHR and that most attempts to compete with LHR service from LGW have been unsuccessful.(34) AA's experience in Boston is instructive. Between June 1998 and October 1999, AA served both Gatwick and Heathrow from Boston. In Exhibit DOJ-6 [Confidential], the number of AA high fare passengers reported in the DOT O&D survey is graphed for the two routes during the third quarter of 1999.(35) For passengers paying above $500 one-way, there is a very clear preference for Heathrow. For discounted coach traffic (generally $500-$600 one-way), as well as for business and first class seats (generally $1600 and up), basically all the passengers traveling on American are choosing to go to Heathrow, not Gatwick.(36) Not surprisingly, American exited the Boston-Gatwick route due to its poor profitability, but retained its Boston - Heathrow service.
American had a similar experience in Miami. Between May 1998 and April 1999, American operated flights from its Miami hub to both LHR and LGW. The Gatwick flight consistently lost large amounts of money each month [REDACTED TEXT]while the Heathrow flight was profitable. [REDACTED TEXT]. After a one year experiment, American exited the Miami-LGW market, while it remained in the Miami-LHR market.
Virgin learned the same lesson. In 2000 Virgin Atlantic operated flights to Newark from both LHR and LGW. The Heathrow flight obtained a yield that was 44% higher than the Gatwick flight, a result of lower demand by time-sensitive travelers for Gatwick services. Virgin notes that despite configuring its EWR-LGW planes with a reduced number of business seats (28 as opposed to 48 on EWR-LHR planes) the load factor on those seats was still only 60%.(37) In response to lower demand conditions after September 11, Virgin has withdrawn its EWR-LGW service, and replaced it with EWR-LHR service, using slots from what had been its third JFK-LHR service. In short, the EWR-LGW route was apparently not profitable enough to continue in a low demand environment. Virgin also made a similar move in Boston this fall. Instead of continuing its single BOS-LGW service, it has now shifted that service to LHR. Once again, LGW service to BOS has been shown to be a marginal proposition compared to service to LHR.(38)
Even BA has had difficulty realizing adequate revenues on Gatwick service in city pairs where it also offers Heathrow service.(39) Indeed, it recently canceled its JFK-LGW service. [REDACTED TEXT] BA's overall LGW strategy rests in substantial part on its recognition of business passengers' strong preference for LHR. [REDACTED TEXT]
In support of the argument that Heathrow and Gatwick are in the same market, AA and BA argue that (1) many passengers flying on unrestricted tickets use LGW; (41) (2) Continental provides profitable service to LGW from its EWR hub;(42) (3) LHR and LGW fares are usually identical;(43) and (4) EU and UK competition authorities have found that the two airports are in the same market.(44) They also argue that Stansted Airport is in the market. These arguments are not convincing.
First, the AA/BA claims concerning passengers flying on unrestricted LGW tickets rest on misleading comparisons. They argue that LHR and LGW are substitutes for many premium passengers because, in the six nonstop overlap routes, 18% of the passengers that flew to Gatwick in 2000 were unrestricted passengers, compared to 28% for LHR. This comparison does not address the relevant issue -- where they have a choice, what do premium passengers choose? In the three overlap markets where nonstop passengers have some choice of LGW or LHR (at Boston, New York, and Miami), passengers overwhelmingly (86%) choose LHR.(45) This demonstrates a strong preference among premium passengers for Heathrow.
Second, contrary to AA/BA arguments,(46) Continental's ability to sustain profitable service to LGW from its hub at EWR does not demonstrate that Gatwick is in the same market as Heathrow. Indeed, if anything, the premium passenger data for Newark further supports the conclusion that there is a significant passenger preference for Heathrow.(47) Exhibit DOJ-7 compares premium passenger share and total seat share for all Newark-London nonstop carriers. Continental, with its substantial hub advantages (strong frequent flyer base, attractive network for corporate discount bids, strong commission overrides) and double daily service comprising 22% of total Newark-London seats, can attract only 17% of the premium passengers. In contrast, AA and UA, which do not hub at EWR, and operate only one flight each, together carry 42% more premium passengers than Continental is able to attract to its LGW flights. The Exhibit also demonstrates the point, noted earlier, that Virgin's EWR-LGW service is unable to attract nearly as many premium customers as its LHR service. Once again, the evidence shows a strong preference for LHR, even among passengers using Continental's EWR hub, and suggests that insufficient numbers of passengers would switch from the LHR services in response to a small but significant increase in the LHR price.
Third, given the strong passenger preference for LHR, the similarity in prices among the two airports is uninformative, and not necessarily surprising. If the airlines serving LGW already know that there is no significant cross-elasticity between LHR and LGW, then they know that undercutting LHR fares would not lure many passengers from LHR, and cutting their price would therefore be unprofitable. They would lose revenue from the few high-fare passengers that would use their LGW services even without a price reduction, and would not gain an appreciable number of business travelers.
Fourth, a review of the EU and UK decisions cited by the parties reveals that in each case (1) the authorities did little or no empirical economic analysis to try to define markets, and (2) whether LHR/LGW were in the same market was largely (or entirely) irrelevant to the decision in the case. The decisions are summarized and distinguished from this case in Appendix A.
Finally, AA and BA argue that London's Stansted Airport should be included within the market,(48) but there has been only one daily round trip flown from that airport during a period when competition has been restricted from the other London airports (but not Stansted). This suggests strongly that service from Stansted cannot be relied on to discipline supra competitive pricing from the other London airports (either because it is not in the same relevant market or because service is not viable from there).(49)
As discussed more fully in §2 of the Merger Guidelines, there are several ways in which increases in concentration can lead to higher prices for consumers. First, mergers may enable remaining firms to raise price through "coordinated interaction." The fewer the firms in a market, the more likely it is that they will be able to reach and enforce an understanding to increase prices.(50) For example, with fewer competitors, there will be less likelihood of a fare sale, or of discount offers targeted at large corporate travelers in the city pair, and a greater likelihood that fare increases initiated by one competitor in the city pair will be matched by all the other significant competitors in the city pair. Coordinated effects are of relatively greater concern in markets such as these, where the available information allows close monitoring of prices and shares, and any "cheating" on an understanding can be detected and responded to quickly.
AA/BA claim that, due to a number of factors that make perfect coordination in the airline industry difficult, the alliance cannot possibly facilitate coordinated interaction.(51) As the Merger Guidelines note, however:
Indeed, the fact that there is a history of collusive activity in the industry belies the claims of AA and BA that coordination is unlikely.(53)
Several characteristics of the airline industry increase the ability of carriers to engage in coordinated interaction. Most importantly, carriers have almost instantaneous knowledge of competitors' fare changes and the ability to quickly respond to any changes.(54) This makes it easier to detect cheating and rapidly punish a carrier deviating from the coordinated price.(55) The fact that many passengers travel at unpublished fares does not make successful coordination unlikely or impossible. A sufficiently large number of passengers still travel at the published fares to give airlines the incentive to coordinate, and any coordinated increase in published fares would result in enormous competitive harm. In addition, since many unpublished fares are a percentage discount off the published fares, coordination on published fares would also significantly impact passengers traveling on corporate discounts or other unpublished fares. Furthermore, although information on unpublished fare competition is certainly less perfect than for published fares, carriers are still able, from ARC and CRS data, to identify corporations and travel agencies where they are losing business and usually the competitor that is gaining business at their expense. Carriers thus have the ability to identify and retaliate against competitors reducing even off-tariff fares.(56) Finally, the capacity constraints at LHR further increase the danger of coordinated interaction in this case by making it less likely that any airline would respond to price increases by expanding its capacity.
Mergers may also enable firms to increase prices unilaterally.(57) In markets where products or services are differentiated, the prices of a firm are constrained by the prices of the next best substitute.(58) If two firms are frequently the first and second choices of customers or particular groups of customers, it may be profitable to raise post-merger prices because some of the lost sales will be captured by the partner. Unilateral effects are particularly likely where firms bid for customers' business. As AA/BA concede,(59) in the case of the airline industry, business passengers view airlines as differentiated products and have preferences for particular airlines based on such factors as frequency, service offerings, and marketing programs such as frequent flyer programs. Corporations purchasing travel through corporate discount programs differentiate airlines based, in large part, on how the service offered by a particular airline matches the corporation's travel needs. A carrier's market share likely reflects "not only its relative appeal as the first choice to consumers . . . but also its relative appeal as a second choice, and hence as a competitive constraint to the first choice." Merger Guidelines § 2.211. The high shares of premium traffic carried by AA and BA in the overlap markets reflect the extent to which they constrain each others fares for corporate travel.
AA and BA have submitted a "critical loss analysis" that purports to demonstrate that, when combined, the two carriers would not have the market power to raise fares unilaterally.(60) That analysis does not provide firm support for any conclusions about the ability of the immunized alliance to use market power to raise fares. The analysis employs overly simplistic and unrealistic assumptions about the strategies that a combined AA/BA might use to raise price, and about the yield management tools they can employ to retain price-sensitive passengers. When some of those assumptions are relaxed in plausible ways (for example by allowing the airline to replace lost local passengers with flow passengers), the critical diversion rates rise well above the benchmark examples cited by the parties. As those diversion rates rise, the ability of AA and BA to successfully raise price through a unilateral exercise in market power increases as well. Even taking the analysis as given, however, some of the parties' own "estimates" of elasticity (as well as those conducted by others) suggest that business travelers may be sufficiently inelastic with respect to price to make a price increase profitable.
In 1998, DOJ concluded that a combination of AA and BA would substantially reduce competition in each of the six overlap markets, eliminating all nonstop competition in DFW-LON and BOS-LHR, reducing the number of competitors from three to two in ORD-LHR and MIA-LHR, and from four to three in NYC-LHR and LAX-LHR. Moreover, AA and BA were each others' primary competitors for the premium time-sensitive business passengers, and the loss of that "inter-brand" competition would be significant, particularly in NYC-LHR. Finally, we expressed concern about competitive losses in markets between interior U.S. points and London where the two carriers competed on a connect, or connect versus nonstop, basis. AA/BA I, DOJ Comments at 16-17.
Exhibit 1 permits a comparison of current competitive service levels to those that existed in 1998. During that period, there has been some expansion by United and Virgin, as noted above.(61) The net effect of these competitive changes has been to erode somewhat the frequency shares held by American and British Airways, particularly in the Boston market.(62) Nonetheless, this transaction would produce substantial increases in concentration in the market for business passengers:
Unless there are realistic prospects for entry, significant price increases can be expected in each of the five overlap markets described above.
While the AA/BA alliance would eliminate some competition on this route, it is less clear than in other markets how significantly such a loss would be felt by consumers.
As noted previously, a growing number of large corporations are entering into corporate discount agreements. The corporate contracts typically provide for system-wide discounts off published fares, in return for a corporation's agreement to provide a minimum market share or revenue commitment on the contracting carrier's flights. In addition to system-wide discounts, the carriers sometimes offer steeper discounts in exchange for a higher market share commitment on specific city-pair routes -- generally those routes where more than one carrier provides service. Thus, the intensity of airline competition that exists for nonstop service on key city pairs determines the system-wide discounts a corporation is able to negotiate, and also the level of city pair-specific discounts as well.(69) Almost all corporations have multiple contracts, and the contracting process is an iterative one, involving simultaneous negotiations with multiple airlines. Within that process, airlines compete both before and after the contracts are signed for corporate passengers in key city pairs. Finally, because corporate fares are usually determined as a discount off the published fares, city pair competition continues to be an important determinant of actual prices paid under corporate contracts. Hence, while airlines compete for corporate customers on a network-to-network basis, and competition at that level is significant,(70) rivalry at the individual city-pair level is a vital competitive component.
The US-London markets are among the largest and most lucrative business travel markets in the world. [REDACTED TEXT] There are a large number of corporate purchasers representing a huge amount of commerce for whom American and British Airways are among a few well-positioned bidders for U.S.-London services. [REDACTED TEXT](71)
For many customers London service is a significant travel purchase, which means that these customers will be willing to incur significant transaction costs in making sure they get the best deal possible for their London travel. [REDACTED TEXT] The importance of London is reflected in the city-pair specific provisions of corporate contracts. [REDACTED TEXT] Large corporate customers play AA and BA off against each other to get better deals on London. One very large purchaser that has system-wide contracts with American, United and British Airways informed DOJ in an interview that the bidding among the three carriers has lowered its U.S.-London prices, and indicated that representatives of each of the three periodically offer additional inducements to shift shares of the customer's New York-London business to their carriers.
"A merger is not likely to create or enhance market power or to facilitate its exercise, if entry into the market is so easy that market participants, after the merger, ... could not profitably maintain a price increase above pre-merger levels." Merger Guidelines, §3.0. Such entry must be timely, likely and sufficient in magnitude, character and scope to counteract post-merger price increases. Even after the regulatory constraints flowing from Bermuda II are removed, entry into the relevant markets will be severely constrained.
DOJ's assessment in 1998 was that slots and facilities limitations at Heathrow made the prospect of competitive entry into the relevant markets in the event of post-transaction price increases highly unlikely. AA/BA I, DOJ Comments at 18-23. Entry conditions have not improved in the intervening three years.
The available evidence indicates that it will be very difficult for other carriers to obtain slots to begin or expand US-LHR service, especially in the short run. Not only is airside capacity scarce, it is also difficult to obtain aircraft parking spaces and terminal facilities necessary to operate new US-LHR service. Even if carriers can get arrival and departure runway space, airport authorities will refuse to allocate the slot if other facilities are unavailable.
Under EU slot allocation rules, new entrants (defined as carriers operating less than four slots on the day for which the slot is requested) get 50% of any slots available through capacity expansions or forfeited slots. The other 50% is divided among incumbent LHR carriers. Carriers that wish to receive new entrant slots apply to the slot coordinator and are placed in the queue for carriers awaiting slots. New slots are allocated according to position in the queue. Airport Coordination Ltd ("ACL"), the slot coordinator for LHR and LGW, has stated that "the opportunities to accommodate new entrant US carriers from the allocation of pool slots in the first two seasons are extremely limited."(72)
Some carriers have entered LHR with new entrant slots since the parties agreed to pursue their last transaction in 1996, but much of this entry occurred during a period when the rate of new slot creation (though still low) was higher than the current rate. In addition, most of the new entrants operate less than daily service. Of four new airlines beginning LHR operations in the summer 2001 season, three have only one or two weekly operations. [BA 0002165] Between the summer 1995 and summer 2000 seasons, 47 carriers were allocated new entrant slots at LHR, of which only nine have seven or more weekly operations. [BA 0002451-53] [REDACTED TEXT]
The parties' documents belie any claim that new entrant slots will enable other carriers to begin LHR service after Open Skies. [REDACTED TEXT]
Although straight slot sales are prohibited under European Union rules, slot trades are permitted and a secondary market in LHR does exist. Slot trades sometimes involve one carrier trading a low-value slot (which may have been obtained solely for the purpose of making a trade) plus cash for a more valuable slot. The frequency of slot trades at LHR is not known, but the parties' documents and other evidence demonstrate that it is difficult to purchase slots, particularly at peak times.
ACL likewise states that, while it is difficult to quantify the amount of slot trading at LHR, "it is ACL's observation that there is a general unwillingness on the part of incumbent Heathrow carriers to divest of slots and the market is illiquid."(73)
Virgin, which has been actively trying to obtain slots to increase its US-LHR service, has succeeded in getting slots in the secondary market for only three daily frequencies since 1996. In 1999 Virgin was forced to move its LHR-Miami service to Gatwick in order to get slots for LHR-Chicago and it was only able to switch its Boston service to LHR when it exited the Chicago market.
Some US carriers are members of alliances with other carriers holding LHR slots: United (British Midland/Lufthansa), Northwest (KLM), Delta (Air France/Alitalia). It is unlikely that alliance partners will be the source of sufficient slots to support entry to ameliorate the competitive losses in the overlap markets, or to support significant levels of new U.S.-LHR frequencies after Open Skies.
United already has immunized alliance relationships, through the Star Alliance, with a number of carriers with LHR slots,(74) and is already authorized under Bermuda II to serve U.S.-LHR routes. If slots for US-LHR service were truly available from an alliance "pool," one would expect to see slot transfers from Star Alliance partners to United. We are aware of no such permanent one-way transfers. AA and BA point to a number of slot trades among Star Alliance partners to support their position, but these trades are mutually beneficial one-for-one swaps to solve timing problems rather than one carrier terminating service on a route and transferring the slots to a partner to begin new service.(75)
Other than the Star Alliance, the LHR slot holdings of alliance partners are very small, particularly during peak transatlantic arrival and departure times. About 63% of daily US-LHR arrivals are during the 600 to 1000 hours.(76) The oneworld (AA and BA) and Star alliances hold a combined average of 74.6% of total LHR arrival slots during those hours, compared to 4.2% for SkyTeam (Delta et al.) and 2.8% for Wings (Northwest et al.).(77) Similarly, during the 1000 to 1600 hours, when about 70% of the US-LHR departures occur, oneworld and Star hold an average of 76.7% of the hourly departure slots compared to about 4% for SkyTeam and 1.9% for Wings.(78)
Most alliances involve codesharing and joint frequent flyer programs but do not entail comprehensive (or any) revenue pooling. (An exception to this is the Northwest/KLM alliance, which does involve some revenue sharing, although not on KLM's European routes.)(79) The operating carrier typically gets almost all the revenue from a flight with the partner getting a commission for any tickets sold under their code. Under such an arrangement it would not likely be in the interest of the alliance partners to transfer valuable LHR slots to US carriers for transatlantic service.(80)
In most cases, the LHR service currently provided by the alliance partners of US carriers is extremely important to their networks. London is a top business destination from almost every city in the world and competitive service is critical to the partner airlines. Expectations that, for example, Air France would reduce LHR-Paris service to provide Delta with slots therefore seems unrealistic. AA/BA have identified what they call "non-strategic" operations of alliance partners (i.e., service to non-hub cities) that they claim could be used for new US-LHR service: six daily round trips by KLM and six by Air France.(81) Most of these slots, however, are not at times suitable for transatlantic service. During the peak US-LHR arrival and departure times discussed above, KLM has two "non-strategic" arrival slots and one departure slot and Air France has two "non-strategic" arrival slots and two departure slots. During hours when Detroit/Minneapolis-London flights currently arrive and depart (Northwest's most likely use for LHR slots), KLM has one "non-strategic" arrival slot and one departure slot.(82) Even assuming all slots from the "non-strategic" operations of alliance partners were made available for new US-LHR service, there are not enough such slots to even move existing LGW service to LHR, much less start new service to ameliorate competitive concerns. To the extent such slot sources are looked to generate Open Skies benefits, the amount of LHR service they will support is small.
Finally, most of the LHR service provided by the alliance partners is narrowbody service to Europe. To convert these slots to widebody service to the US, a carrier would need a place to park the larger aircraft and sufficient terminal capacity to handle the larger number of passengers. Bmi, the second largest slot-holder at LHR with about 13.5% of the slots, has stated that it is impossible for it to convert large numbers of slots to transatlantic service due to the unavailability of widebody stands and insufficient terminal capacity.(83)
Aircraft gates and stands are very scarce at LHR, particularly facilities for widebody aircraft during peak periods. Similarly, the terminals at LHR are at capacity, and it would be difficult to find space for additional widebody flights. There is little chance that additional capacity will become available in the short run. The UK has recently approved construction of a new terminal ("T5"), a project that has been under consideration for several years, but that has been slowed due to environmental opposition to LHR expansion. BA estimates that in a best case scenario if T5 is approved in 2001 and there is no judicial challenge (which seems unlikely), construction could begin in 2002 and be completed in 2007. [BA 0006501-31]
The difficulty of accommodating additional US service at LHR is illustrated by BA documents discussing their ability to move current US-Gatwick operations to LHR and co-locate AA and BA operations at LHR. BA, with large slot holdings at both LHR and LGW, has more flexibility than any other carrier to shift services, yet [REDACTED TEXT]
As we concluded in our last analysis, in addition to physical barriers to entry, lack of scale and scope may prevent entry on some routes, in particular those where competitors have hubs at one or both endpoints. AA/BA I, DOJ Comments at 28-29. Industry developments since 1998 have not diminished the entry-deterring effects of hubs. It is well-established that entry into hub markets is less likely to occur and less likely to be successful (unless the entrant has a hub at the other end point), and that hub carriers are able to charge a premium for nonstop service from the hub. Each of the six overlap routes has an AA or BA hub on one or both endpoints, and three will have alliance hubs at both endpoints.
Entry by nonhub carriers is difficult in part because such a carrier does not have the connecting feed of the hub carrier and is forced to rely solely on local traffic. Even where the local market is large enough to support service, the hub carrier has a significant advantage due to the "S-curve" effect that gives the largest carrier at an origin a disproportionate share of the traffic. This effect reflects both the increased quality of service offered by the largest carrier (e.g., more frequency to most destinations), as well as its ability to use corporate contracts and marketing devices such as frequent flyer programs and commission overrides to get a larger share of the traffic.
Due to these barriers related to airport dominance, we concluded in our last comments that no carrier was likely to enter two of the three hub-hub overlap routes (London-Dallas/Chicago).(84) Since that time Virgin entered (and exited after 9/11) ORD-LHR, and no carrier has entered the London-Dallas market. While our analysis would not have predicted Virgin's entry into the ORD-LHR market, its subsequent exit suggests that that route was the most marginal of its LHR routes and, as discussed below, it re-entry after approval of the pending applications would be more difficult, and unlikely.
To be considered a plausible entry candidate, an airline needs to have a route network that puts it in position to compete for key traffic streams, in addition to the necessary equipment and facilities. With the exception of bmi's potential entry into MIA-LHR, none of the potential entrants into any of the overlap markets has or is likely to acquire (outside of a divestiture) the LHR slots and facilities necessary for entry, for the reasons discussed above. The following discussion considers whether entry would be likely if slots were not a constraint.
BOS-LHR. Both Delta and USAirways have significant operations at Boston, and hence they have the type of "presence" that would give them the incentive and ability to attract corporate business to LHR services.
NYC-LHR. The New York-LHR market is one with a large base of frequent business passengers. To compete seriously for such customers in a market where several other carriers offer substantial frequency from multiple airports would require sufficient resources to provide at least three to four daily round trips, including at least one from Newark.(85) At least two airlines - Delta and Continental - have the presence at New York to be considered plausible entry candidates. Continental's hub at Newark would give it a strong set of "products" to bundle with its London service in bids to corporate customers. Delta's East Coast Shuttle and transatlantic hub at JFK give it a substantial presence at New York as well.
MIA-LHR. Bmi has announced it would likely enter this market. Bmi has both the endpoint presence and the LHR facilities needed to make it a plausible entry candidate into this market, which is not currently served by its alliance partner, United. Therefore, despite the fact that this market has an AA/BA alliance hub at both endpoints, there is at least one likely entrant.
LAX-LHR. Even if LHR slots and facilities were available to permit LAX-LHR entry, it is not clear that any new entrants have route networks that put them in a good position to enter. Other than United and AA, no U.S. carrier that currently operates international service has a significant share of enplanements at LAX. British midland's London operations might give it a traffic base for entry, but it will be allied with United if the pending applications are approved.
ORD-LHR. ORD is a major hub for both AA and United. LHR is a hub for BA and british midland. If the pending applications are approved, a would-be entrant would face two incumbents with hubs at both endpoints. These circumstances would likely make successful entry materially more difficult than the conditions Virgin faced when it decided to enter this market in 1999 (and exit in 2001).(86)
DFW-LON. Even if the alliance does not switch its DFW service to LHR in order to provide "beyond-beyond" connecting opportunities,(87) DFW-LON will connect a major alliance hub with a city with dominant alliance presence. Although Delta has a small domestic hub at DFW, it does not operate any transatlantic service from DFW, and would be highly unlikely to enter the route even if prices rose.
Like other international alliances, this alliance has the potential to benefit consumers by lowering fares and improving service.(88) In markets where the parties are primarily vertically related (that is, markets where they can provide interline service) cooperative pricing of interline service by an integrated alliance often results in lower prices for consumers. Alliance partners can also increase the quality of traditional interline service by coordinating their schedules to minimize layovers and offering the consumer more seamless "on-line" connections.
Experts hired by the parties, Professors Brueckner and Ordover, estimated these interline benefits at $55 to $69 million annually.(89) As explained below, these results substantially overstate the potential annual benefits. The AA/BA estimates are based on work by Brueckner showing that average fares for carriers operating with antitrust immunity in an Open Skies environment are 23% lower than traditional, non-alliance interline fares.(90) DOJ analysis replicates and extends these results and finds that prices for such alliances are statistically no different from those for single carrier, on-line service. However, DOJ also found that fares for AA/BA interline itineraries are already significantly below the fares for traditional interline service, although not as low as fares for prior immunized alliances. Assuming that the AA/BA alliance would be able to achieve connectivity benefits that were similar to those achieved by earlier immunized alliances that effect translates into fares roughly 7-10% lower than those currently offered, not the 23% used by Brueckner. Total annual savings to consumers, assuming fare decreases in the 7-10% range are realized, would be approximately $10-14 million. Appendix B describes in more detail the analysis generating these conclusions.
Despite the potential for benefits to interline passengers, this alliance differs from previous alliances in two important ways. First, unlike prior alliances, the competitive overlap markets in this case are significantly larger than the routes where connectivity benefits are likely. Analysis by the party's experts claims that there were 127,767 passengers who traveled on AA/BA interline routing in 2000 who paid $173 million in fares.(91) DOJ estimates that AA and BA carried over [REDACTED TEXT] paying over [REDACTED TEXT] in the overlap routes during the same period.(92) It would take a post-transaction price increase for those passengers of only 1% to negate the potential connectivity gains from the transaction.
A second significant difference between the AA/BA alliance and prior immunized alliances is that the parties' incentive to reduce capacity available for local traffic may reduce any benefits that might otherwise accrue to connecting passengers from joining two airline networks together. Such developments would contrast sharply with other alliances, which have increased alliance capacity dramatically in hub-hub routes, resulting in an explosive growth in the number of connecting passengers carried by the alliance.(93) [REDACTED TEXT]
IV. Possible Remedies and Public Interest Conditions
DOT may approve and immunize an anticompetitive transaction if it determines that the transaction is necessary to advance important public benefits that outweigh the anticompetitive effects, and that the public benefits cannot be achieved by reasonably available alternatives that are materially less anticompetitive. 49 U.S.C. §41309(b)(1). This standard would permit DOT to approve this transaction if it finds that the transaction is needed to achieve Open Skies with the United Kingdom, that it is the least anticompetitive way to achieve it, and that the benefits from Open Skies outweigh the competitive harm from this transaction. Whether there are reasonably available, less anticompetitive means to achieve Open Skies is essentially a political question that DOJ has not attempted to answer. What is apparent is that the current proposal could result in Open Skies in the near future. If an alliance between BA and some U.S. carrier is related to Open Skies, waiting for a less anticompetitive alliance than the AA/BA deal would postpone Open Skies until well into the future. If DOT decides it is in the public interest to act now, however, DOT can, and should, impose conditions designed to reduce or eliminate the anticompetitive effects of the AA/BA transaction where possible.
As to the third question -- whether the benefits of Open Skies outweigh the anticompetitive effects of the transaction (appropriately conditioned to reduce anticompetitive effects) -- a true Open Skies bilateral regime for this huge aviation market would provide enormous public benefits, if the U.S. can be assured that entry conditions are such that there will be de facto as well as de jure Open Skies. That will not occur unless additional slots are made available for service to the U.S., over and above those needed to remedy competitive harm.
As explained more fully below, the additional slots would simultaneously serve three important purposes: first, they would provide compensating benefits for the unremedied harm from the AA/BA alliance; second, they would permit DOT to allocate all slots through an efficient market mechanism while still seeking to preserve existing competition in NYC-LHR and BOS-LHR; and third, the additional slots could be sufficient to conclude that de facto Open Skies has been achieved.
LHR Slots and Facilities. The likely anticompetitive effects of this transaction in two of the affected markets could be substantially ameliorated if conditions were imposed to ensure that adequate LHR slots and facilities were available to entrants into those markets. As discussed above, NYC-LHR and BOS-LHR have one or more airlines that are in position to enter if they have the needed LHR resources. In total, these remedies would require the availability of sufficient slots and facilities to permit the operation of at least nine daily round trips -- two for BOS-LHR and seven for NYC-LHR.
In Boston, prior to September 11, BA flew three daily round trips in this market and AA operated two. In order for an entrant to replicate the rivalry of AA and BA in this market, slots and facilities sufficient to operate two daily round trips are required to replace the service currently supplied by the smaller of the two carriers.
In New York, prior to September 11, AA operated seven daily round trips in the NYC-LHR market and BA operated nine. United and Virgin operated four each. To replace the competition lost from AA, the smaller of the two merging competitors, DOT should make sure that well-timed slots and facilities sufficient to operate at least seven daily round trips are made available to entrants into this market.
Routes for Which Entry is Unlikely -- Carve-Outs. Even if LHR slots were available, entry is unlikely in ORD-LHR and DFW-LON. For those markets, DOT should try to minimize the competitive harm by imposing carve-out conditions.
In past cases, DOJ has advocated and DOT has adopted carve-out provisions. Under a carve-out, DOT limits the immunity so that it does not extend to "pricing, inventory or yield management coordination, or pooling of revenues, with respect to unrestricted coach-class fares or any business or first-class fares for local U.S.-point-of-sale passengers flying nonstop" in the specified city pair markets.(96) Even with the carve-out, the parties have immunity to fix prices and coordinate yield management for other fare products in the markets and may jointly offer corporate discounts that include unrestricted fares.(97) They also have immunity jointly to agree on capacity, frequency and aircraft configuration in the specified markets.
DOJ has some concern about whether carve-outs fully preserve pre-alliance competition in the affected local nonstop markets. We have reviewed available data in an attempt to determine whether the carve-outs that have been imposed to date have been effective in preserving competition, but that data is inconclusive. Despite the uncertainties, carve-outs still offer some promise of reducing the loss of the competition that determines the level of published unrestricted fares in the affected markets, and DOJ recommends that DOT impose them in this case with respect to AA/BA operations on DFW-LON and ORD-LHR.
LAX-LHR. New entry is unlikely in this market, even with slot divestitures, although not as unlikely as for ORD and DFW. We do not recommend a carve-out, however, because the competitive harm in LAX-LHR is less certain than for the other overlaps, and it also may not be as large. Given this, DOT could reasonably decide to offset the potential for harm on the LAX-LHR route by providing one or more slots for more competition in other routes that might benefit from de facto Open Skies.
In sum, the entry of new carriers with seven and two daily round trips in NYC and BOS together with the carve out of DFW and ORD, will substantially reduce (although not eliminate) the competitive harms that are likely to arise from combining the operations of AA and BA. Some residual competitive harm is likely to remain in the carve-out markets, as well as in LAX-LHR.
To meet DOT's public interest standard, DOT must go beyond maintaining the status quo. Because of physical constraints at LHR, even completely neutralizing the competitive harm from combining AA and BA would do little more than preserve and perhaps solidify the concentrated market structure that evolved under the Bermuda II agreement.(98) Therefore, DOJ endorses the goal DOT clearly stated in its last proceeding, where it noted that the public interest required de facto Open Skies with the United Kingdom before it would consider antitrust immunity. DOT defined "de facto Open Skies" to "include adequate provision for new and expanded U.S. carrier service through London airports, particularly Heathrow." DOT also stated in the context of the last application that "the ability of U.S. carriers to provide such service [at London] notwithstanding the constraints at Heathrow would be a critical consideration in our evaluation of the proposed Alliance." Order 99-7-22 at 2. DOT should apply the same reasoning in this proceeding. To achieve de facto Open Skies, DOT must provide for slots and related facilities in addition to those needed to remedy competitive harm in the NYC and BOS markets.
DOJ generally favors a market-based allocation system that would allow entrants to purchase divested slots with few strings attached. The marketplace generally does a better job than regulators of selecting the most efficient competitor. This approach does not, however, ensure that assets will be allocated to the markets with the greatest competitive harm. Under a market-based allocation system, it is likely that slots for more than nine daily frequencies would have to be made available in order to assure that NYC and BOS receive sufficient service to replace that provided by AA. For example, if a slot auction is held it seems highly likely that carriers would use the first purchased slots to serve important hubs such as Atlanta, Detroit, Houston or Philadelphia rather than Boston. While this result would clearly afford competitive benefits to passengers in cities getting new LHR service, it would leave the competitive harm in Boston unameliorated. Accordingly, a higher number of divested slots would make it more likely that slots would actually be used to provide service in the harmed markets.
If DOT chooses a non-market mechanism, like a carrier selection process, to award slots for specific markets, slots for as few as nine dailies would suffice to maintain the status quo in Boston and New York. Those slots would, however, have to be restricted in use to assure entry in the harmed markets. This approach is significantly less desireable than a market-based approach because the "earmarked" slots are unlikely to be going to their highest and best use.
Combining the US-London service of AA and BA through an immunized alliance will result in immediate competitive harm. DOT should therefore delay the effectiveness of any immunity order until carriers that receive the divested slots and facilities are in a position to begin new service to remedy the competitive harm. The delay could involve phase-ins of immunized operations by AA/BA in certain markets, and should specify dates certain rather than any performance measures by new entrants.(99) A date certain maximizes the incentive of new entrants to develop their competitive service as quickly as possible.
In sum, the achievement of Open Skies with the U.K. holds the potential for significant benefits for U.S. consumers. With appropriate conditions designed to ameliorate the competitve impact of the AA/BA transaction and to ensure de facto Open Skies, antitrust immunity for the proposed alliances in conjunction with Open Skies would advance the public interest.
Dated: December 17, 2001
1. Docket OST-97-2058, Joint Application of American Airlines, Inc. and British Airways Plc. For Approval of and Antitrust Immunity for Alliance Agreement ("AA/BA I"), Comments of the Department of Justice, May 21, 1998 ("DOJ Comments").
2. In MIA-LHR, the increased concentration is unlikely to result in a substantial lessening of competition because approval will be linked to Open Skies, and with Open Skies bmi, which has the necessary LHR slots, is well-positioned and likely to enter.
3. See, Joint Answer of AA and BA in Opposition to Petition of Continental Airlines, Inc. for Reconsideration, at 3.
4. However, it has initiated service in 2001 between Manchester and Chicago and Manchester and Washington, D.C.
5. Docket OST-2002-10575, Joint Application of United et al., Exhibit JA-12, p. 5.
6. The fourth significant U.S.-LHR competitor is Virgin, with a total of 9 daily LHR frequencies to 6 U.S. airports. Prior to September 11, Virgin also served 4 U.S. airports from LGW.
7. Subsequent to September 11, that service was moved from John F. Kennedy Airport (JFK) to Newark (EWR) to replace Virgin's discontinued EWR-LGW service.
8. Also, Delta initiated BOS-LGW service in May 2001 with one daily frequency to LGW.
9. CO operates its own aircraft only on routes between Newark and London Gatwick. Its share on other routes reflects ticket sales from its block seat arrangement with Virgin.
10. American uses such data internally, and refers to it as "CONCRS" data. The CRS data does not represent all tickets, however, since it represents only bookings made through travel agents. Moreover, unlike DOT's O&D data base, it does not contain information on prices paid, although it does provide the booking class used for each booking, allowing some inferences to be made about travel patterns of various categories of passengers. Despite these limitations, the CRS data is probably the best source of information available on traffic shares of all carriers by itinerary, as it contains much more complete information than the O&D data base on passengers booked on foreign carriers.
11. Shares have been computed for all origin-destination traffic in each direction and across both London airports for convenience of exposition and comparison. Similarly, New York shares aggregate JFK and EWR. Section III.B. below discusses specific differences between LGW and LHR in more detail.
12. This definition does not include time-sensitive passengers traveling on unrestricted coach tickets, and thus understates the number of passengers who would be harmed by a reduction in nonstop competition. Because some carriers use fare classes other than Y for selling unrestricted coach seats it is very difficult to make accurate comparisons between carriers, and we have therefore used F/J/C as a proxy for premium passengers.
13. Joint Reply of AA and BA, Appendix A, p. 23.
14. Examples in domestic markets include the disparities between markets that are served by low cost carriers and those that are not. Internationally, the fact that fares between New York and London are often much higher than fares between New York and points beyond London (e.g., Paris) also shows a lack of substitutability among city pairs.
15. Business passengers constitute a particularly high proportion of traffic to London. See, e.g., [REDACTED TEXT]
16. See, e.g., Andrew S. Joskow, et al., Entry, Exit and Performance in Airline Markets, 12 Int'l J. Indus. Org. 457 (1994); and E. Kim and V. Singal, Mergers and Market Power: Evidence from the Airline Industry, 83 American Economic Review 549 (1993).
17. E.g., Joint Application, Exhibit JA-8 at 13.
18. To compute the number of European connects in the data, all passengers connecting at five major European hubs (AMS, BRU, CDG, FAR, and ZRH) were summed.
19. A negligible amount of AA's traffic took a connection to LGW, but the rest connected to LHR.
20. It is also likely that the existing AA and BA nonstop service to Gatwick (with just 3 daily flights combined) may not offer enough capacity or time of day options. [REDACTED TEXT]
21. Thus, even if connect service to LHR were considered to be a substitute for nonstop service to LGW, AA's dominance of the connect traffic would suggest a competitive concern.
22. Joint Application at 54, citing the American Express Survey of Business Travel Management 2000-2001, B-18. See, also Joint Reply of AA and BA at 21-24 and Appendix A pp. 8-16.
23. E.g., DOT Order 96-3-33, Joint Application of Delta, Swissair, Sabena and Austrian, at 13.
24. American Express Survey at B26-B27. 39% of corporations use number of flight hours to determine whether employees may travel in business or first class.
25. Order 96-6-33, Joint Application of Delta, Swissair, Sabena and Austrian, at 13.
26. Joint Reply of AA and BA, Appendix A, pp. 1-2.
27. According to the parties more than half of BA's passengers fly on tickets with an with an unpublished fare. Joint Reply of AA and BA, Appendix B, p. 3.
28. In structuring its bid, the airline can adjust its discount offer depending on corporation's preference (or lack of preference) for connecting service. After the contracts are awarded, airlines can and do individualize their treatment of customers even further by waiving travel restrictions for particular corporations or travelers where there is a danger of losing them to competing airlines or fare products. These individualized responses allow the airline to keep fares high for less elastic customers while retaining the customers with a greater willingness and ability to switch.
29. Joint Reply of AA and BA, Appendix A.1.
30. See Appendix A.1, p. 2, footnote 4. Published prices can vary widely from the prices charged to many customers (for example, customers flying under corporate discount programs), which can have a significant and unpredictable effect on the analysis. Moreover, the AA/BA data attempts to control for seasonal effects by looking at relative booking shares, but many other potentially important factors in determining bookings (such as the larger business market in New York relative to other routes, for example) are ignored.
31. See Joint Reply of AA and BA, Appendix A.1, p. 21. The elasticity estimates also vary wildly: one is -0.3, the other -2.3.
32. See Joint Reply of AA and BA, Appendix A.1, p. 14. The lack of variation in the data means that no meaningful conclusion can be drawn from this analysis. The fact that published prices are the same should not be taken a evidence that nonstops and one-stops are in the same relevant market. Actual prices paid can vary significantly because of discount programs and the carriers' ability to affect seat availability through yield management.
33. Joint Reply of AA and BA, Appendices A.4 through A.6.
34. [REDACTED TEXT]
35. Data in this graph is taken from DOT's Origin and Destination (O&D) survey, which is a ten percent sample of all tickets. The data graphed herein is the actual sample data, so to obtain an approximate number of passengers for the quarter, the numbers shown in the graph must be multiplied by ten.
36. [REDACTED TEXT]
37. Answer of Virgin Atlantic Airways Limited, at ¶¶ 39-40.
38. Delta recently began daily BOS-LGW service. While it is too early to tell whether that service will be successful, there is no reason to believe that Delta will be more effective at attracting business passengers than American or Virgin.
39. [REDACTED TEXT]
40. BA 0006470.
41. Exhibit JA-8 at 20; Joint Reply of AA and BA, App. C, at 6-10.
42. See Joint Reply of AA and BA at 36.
43. Exhibit JA-8 at 25-26; Joint Reply of AA and BA, App. C, at 10-12.
44. Exhibit JA-8 at 23-25; Joint Reply of AA and BA, App. C, at 12-13.
45. Joint Reply of AA and BA, App. C, Table 3. Also, the AA/BA data premium passenger data, which use the F, C and Y fare classes, do not accurately reflect the flows of time-sensitive business traffic in these city pairs. Based on our review of the published fares mapped to the fare categories used by the parties, the "unrestricted coach" or "Y" fare category within the data source used by the parties may be both under-inclusive and over-inclusive, as some carriers use fare classes other than "Y" for unrestricted fares, and some restricted fares are sold within the "Y" fare class by some carriers. Using F/J/C class premium passenger measures instead of F/J/Y shows that the percentage of LGW premium passengers in the three overlap markets is just 9% of total premium traffic. Some of the difference in these estimates appears to be because F/J/C includes United's business class Heathrow customers, which are excluded when only F/J/Y are used. In addition, VS reports large numbers of Y class passengers, but sells many restricted tickets in that category, leading to an over expansive estimate of unrestricted passengers.
46. See Joint Reply of AA and BA at 36.
47. Out of all New York passengers to London, we focus on those travelers at Newark because they can most easily choose between the alternative London airports given existing service patterns. Regardless of whether JFK and EWR are in the same market, choices made by passengers at Newark present one test of whether Gatwick can offer significant competition to Heathrow.
48. Exhibit JA-8, p. 21.
49. [REDACTED TEXT]
50. The understanding could be tacit, and would not necessarily be illegal.
51. Joint Reply, Appendix B.
52. Merger Guidelines § 2.11.
53. E.g., United States v. Airline Tariff Publishing Co., et al. Civil Action No. 92-2854 (D.D.C., filed Dec. 21, 1992), 1994 W.L. 454730 (D.D.C.)("ATP"); United States v. American Airlines and Robert L. Crandall, 743 F.2d 1114 (5th Cir. 1984).
54. The suggestion that frequent fare changes make coordination difficult (Joint Reply, Appendix B at 6-7) ignores the sophisticated computer programs and teams of pricing analysts that carriers employ to analyze and quickly respond to fare changes.
55. AA/BA erroneously cite the Competitive Impact Statement filed by DOJ in ATP for the proposition that, absent facilitating mechanisms such as those prohibited by the Final Judgment is that case, coordination is impossible in the airline industry. In fact, the airlines have developed other means of coordinating fare increases since those restrictions became effective. For example, carriers frequently file increased fares on weekends (when relatively few tickets are purchased) and withdraw the increase if other carriers have not matched by Monday. In addition, the restrictions of the ATP decree do not apply to international fares. ATP, Final Judgment §II(I). See, 1994-2 Trade Cases ¶70,687.
56. Where significant volumes of business are subject to long term contracts there may be a strong incentive for firms to deviate from the coordinated price. Merger Guidelines § 2.12. While a large volume of high yield traffic is subject to corporate contracts, these contracts are not long term (generally a year), are not exclusive, and can be terminated on short notice by either party. Moreover, contracts come up on a rolling basis and no one contract represents a large percentage of the available business. Therefore, a carrier can quickly identify and retaliate against a competitor becoming "too aggressive" in going after contract business.
57. Merger Guidelines § 2.2.
58. Merger Guidelines § 2.21.
59. Joint Reply of AA and BA, Appendix B at 8-10.
60. Joint Reply of AA and BA, Appendices A.4 and A.6.
61. The parties also argue that Continental has "entered" all of the markets served by Virgin because it purchases seats from Virgin for resale under a "blocked space" arrangement. The overlap markets where Continental purchases seats from Virgin are NYC-LHR, BOS (or CHI)-LHR, and LAX-LHR. While such an arrangement probably does give Continental some level of competitive significance in these markets (particularly New York, where it has a hub), its significance is lessened by the fact that its costs are controlled by Virgin and Continental has no independent ability to expand. Moreover, Virgin's sale of seats to Continental reduces the capacity Virgin has available for sale.
62. Overall, Virgin has only increased its U.S.-LHR frequencies by one daily flight. Since September 11, Virgin has dropped ORD-LHR service and shifted that LHR frequency to BOS, to replace its pre-existing BOS-LGW flight. This presents an issue as to how to evaluate the competitive effects of the transaction in ORD-LHR and BOS-LHR. We anticipate that, as traffic rebounds, Virgin may not shift the BOS-LHR frequency back to CHI if AA/BA and the pending alliance between United and british midland are approved, since it would in that case be facing competition from two alliances with hubs at both endpoints of the route. However, that also means that the parties' shares of BOS-LHR premium traffic set forth in Exhibit DOJ-2B overstate current concentration levels.
63. [REDACTED TEXT]
64. [REDACTED TEXT]
65. [REDACTED TEXT]
66. [REDACTED TEXT]
67. [REDACTED TEXT]
68. [REDACTED TEXT]
69. Indeed, system-wide contracts frequently have city-pair specific discounts and city-pair specific performance goals for the corporate customer.
70. See, Answer of Michael E. Levine to AA/BA Application, filed November 14, 2001.
71. Customer overlap at any point in time grossly understates AA/BA competition since many customers choose one or the other through the bidding process.
72. Response of Airport Coordination Ltd to Questions on Access to Slots and Facilities at London's Airports at 5. ACL stated that it might be possible to accommodate one daily frequency with a late evening arrival and an afternoon departure, as well as some less than daily (primarily weekend) service. Id.
73. Response of Airport Coordination Ltd to Questions on Access to Slots and Facilities at London's Airports at 5.
74. Star members with LHR slots include Lufthansa, SAS, Air New Zealand, Austrian, and Air Canada. In addition, British Midland, the second largest LHR slot holder has recently joined Star.
75. Most of UA's increase in US-LHR service has been funded by slots purchased from Pan Am that were historically used for UK-Germany fifth freedom operations. UA turned operation of these slots over to Lufthansa during a transition period, after which they were returned to UA for increased US service. We are aware of no net transfer of slots from LH to UA.
76. Exhibit JA-R-3.
77. Exhibit JA-R-4. This understates the percentage of peak slots effectively controlled by oneworld and Star, as several of the "non-aligned" carriers, while not alliance members have codesharing relationships with oneworld or Star carriers (e.g., JAL, Swissair).
78. Exhibits JA-R-3 and JA-R-4.
79. Even if carriers share revenues on transatlantic routes it might not be in the interest of the partner to transfer slots from European routes on which there is no revenue sharing.
80. Short-term leases among alliance partners are not a likely to develop, either. Under the applicable slot transfer rules, leases of slots are not allowed unless the designator code of the carrier that owns the slot is used on the flights that use the slot. Because, for example, Air France would not have traffic rights for US-UK service, it would have to transfer full ownership of a slot to Delta in order for its slots to be used for US-UK transatlantic service. [BA 0003954-61]
82. JA-R-8; OAG.
83. UA/bmi application at 50-51.
84. We predicted in 1998 that a carrier such as bmi would likely enter London-Miami, which is largely a leisure route (AA/BA I, DOJ Comments at 28-29), and bmi has said they would begin service post-Open Skies if their alliance with UA is immunized.
85. [REDACTED TEXT]
86. Virgin was not very successful in attracting U.S-originating premium passengers to its ORD service even when its competitors did not have linked hubs. It's share of such traffic was only 2.25% in 2000. Exhibit DOJ-8. Its prospects for attracting the high-yield passengers needed to support profitable service could only be worse if it faced two incumbents with hubs at both endpoints.
87. [REDACTED TEXT]
88. If DOJ were reviewing this transaction under the antitrust laws, we would examine whether any benefits were transaction-specific. Merger Guidelines § 4. If an alternative, less anticompetitive way to achieve the benefits exists, such benefits may not be cognizable under the Merger Guidelines. Given the timing and circumstances of this proceeding, we have not attempted to determine the extent to which benefits from the AA/BA alliance, including Open Skies, could reasonably be acheived through less anticompetitive transactions or methods. The foregoing discussion, therefore, assumes that all benefits are transaction specific.
89. Statement of Jan K. Brueckner, November 2, 2001; Statement of Janusz A. Ordover and Milena Novy-Marx, November 2, 2001.
90. Antitrust immunity for international alliances has been granted by DOT as a quid pro quo for Open Skies in the foreign carrier's home country. In all such cases to date, the Open Skies that resulted has been both de jure and de facto Open Skies.
91. It is likely that a significant portion of these passengers traveled in markets that already have on-line service from one or both of the parties.
92. Estimate based on confidential data supplied by the applicants to DOT in response to information requirement 14 (See Joint Application at 81-83). In this context, business passengers are defined as those who paid one-way fares in excess of $500. This estimate does not include passengers carried by other airlines.
93. See Department of Transportation, International Aviation Developments: Global Deregulation Takes Off (First Report), 1999 at 7-8.
94. [REDACTED TEXT]
95. [REDACTED TEXT]
96. E.g., Order 2000-5-13, Appendix 1. Such activities are not prohibited; they are simply not immunized from the antitrust laws.
97. "For immunity to apply, however: (1) in the case of corporate fare products and group fare products, local U.S. point-of-sale non-stop traffic [in the carve-out markets] shall constitute no more than 25% of a corporation's or group's anticipated travel (measured in flight segments) under its contract..." Order 2000-5-13, Appendix A, p. 1.
98. Some new service from Heathrow could occur even without the imposition of slot and facility divestiture conditions by DOT. For example, bmi has indicated some interest in serving Seattle-LHR and Denver-LHR after Open Skies goes into effect. Such new services, to the extent they are considered likely, should be considered as Open Skies benefits.
99. Based on statements by BAA, accommodation of significant new US-LHR service will take at least two years even if carriers could immediately obtain slots through AA/BA divestitures.
To measure the prices effects on connecting routes associated with the formation of immunized alliances, we assembled a data set of U.S.-Europe routes using the third quarter of each year from 1990 to 2000. The data were restricted to routes where the U.S. endpoint was a non-gateway airport, a restriction which eliminates all markets with nonstop service (i.e. the gateway-to-gateway markets). It also restricts the data to markets where every itinerary must include a US carrier segment. Because foreign carriers do not report to the DOT Origin and Destination Survey, if the market could be served solely by a foreign carrier, some competition would go unobserved in the data. This restriction is made to guarantee that all the competition in a particular market is observable in the data. We also eliminated open jaws, itineraries with more than 3 coupon segments in one direction, unreasonable fares and those with more than two different carriers in the itinerary.
The regression results described below use data aggregated up to the level of a route. This aggregation reduces the possibility that the results are driven by changes in the business/leisure passenger mix within a market between different carriers. Because the data are aggregated to the route level, many variables used in the regression are transformed into the percentage of passengers fitting that particular criteria. For example, rather than having a dummy variable to indicate that an alliance offers service in a particular market, we use the percentage of traffic in the market on carriers who have a code share alliance.(7) The variables used in these regressions are listed in the table below.
The regression results are presented in the table below. The regression is estimated using fixed effects for routes to capture all effects that are constant over time (for example, the distance between the city-pairs). The regression also includes dummy variables for each time period and carrier-specific effect variables, but those coefficients are omitted from the table for simplicity. The dependent variable in the regression is the natural log of average fare.
The results show that the fare charged for online (single carrier) service is 19.9% below the fare for non-alliance interline flights. The regression also predicts that the fare for unimmunized alliance flights is 9.6% lower than non-alliance interline service, roughly half way to the online fare, and immunized alliance fares are an additional 12.6% below that for a total of 22% lower fares. Brueckner argues that the immunity grant is solving a coordination problem and in his simulation of benefits, assumes AA/BA interline fares fall by his estimated effect of immunity, which is 23%. According to his theory, though, because the coordination problem is "fixed" when a single carrier prices the service, the benefit to interline passengers cannot exceed the pricing benefit associated with online fares. In fact, we find in our estimates that the 22% fare decrease associated with immunized alliances is statistically equivalent to the coefficient for online service.
Because AA/BA have an extensive marketing alliance, there is no reason to assume their current fares are identical to non-alliance interline fares. The regression indicates that current AA/BA interline fares are, in fact, 12.6% below nonalliance interline fares. AA/BA are still jointly providing lower fares than otherwise expected. Conversely, after immunization and Open Skies, there is no reason to expect according to Brueckner's coordination theory or the data that AA/BA interline fares should fall any lower than the levels charged by online carriers. If AA/BA interline fares fell to the level of online service, fares would decrease 7-10%, depending on whether one uses the coefficient on online service or the sum of the coefficients on the alliance and immunity variables (which are statistically identical). Taking all passengers in the 3rd quarter of 2000 who traveled on AA/BA interline observations, we calculated how much those passengers would have saved if fares were 7% lower and 10% lower and find consumers would have saved $2.4 million and $3.4 million, respectively. Roughly extrapolating that to an annual number by multiplying by four suggests consumers would save only $9.6 million to $13.6 million annually.
1. Case No. COMP/JV.19 (Aug. 11, 1999).
2. Case No. IV/M.1305 (Dec. 19, 1998).
3. The EU determined that entry would be difficult because "[s]lots at Heathrow, in particular, will be hard to come by, especially in the quantities needed to offer sufficiently frequent service to attract business passengers." Id. at ¶ 38.
4. Case No. IV/M.967 (Sept. 22, 1997).
5. UK Competition Commission, British Airways Plc and CityFlyer Express Ltd. (1999).
6. The Commission stated that "the development of new, effective, competition is likely to be a gradual process, particularly if BA and its subsidiaries are competing for any slots that become available. We believe that access to slots at Gatwick for BA's competitors is such that the threat of rapid new entry is, at best, limited; and we do not believe that the situation at Heathrow is any better." Id. at ¶ 2.172.
7. Regressions on unaggregated data produced very similar results.