Comments with respect to this document should be sent to:
Dated: January 28, 1998
On December 31, 1997, the Department of Transportation (the Department) issued a Show Cause Order, in which it tentatively approved the proposed American Airlines/TACA Group joint applications for statements of authorization to engage in reciprocal code-share services (subject to certain conditions), and invited interested persons to submit comments on that order. In reaching its conclusion, the Department found that approval of the agreement would "advance important public benefits," including operating efficiencies that would allow the parties to achieve "lower costs and enhanced service for U.S. and international consumers."(1) The Department also gave weight to the open skies agreements reached between the United States and various foreign applicants' homelands. The Department of Justice (DOJ) hereby submits its comments on that order.
The Department may grant the parties' proposed code-sharing applications only if it makes a positive finding that they are in the public interest. 49 U.S.C. § 40109. Effect on competition policy is a major factor in assessing whether a code-share agreement advances. Accordingly, these comments address whether the proposed code-sharing agreement promotes or enhances competition in the affected markets, rather than the narrower issue of whether the parties would violate the antitrust laws if they engaged in the joint activities contemplated by the proposed agreement.
The DOJ takes no position on the weight that the Department should give to open skies achieved as a precondition to its consideration of the current code-share application (or on whether open skies could be achieved without approval of this specific agreement). For the reasons noted below, however, the DOJ urges the Department to give little weight to the parties' proffered efficiencies and resulting claims of expanded networks and seamless service in the U.S. - Central American market. The claimed efficiency benefits that are unique to this transaction are very slight, yet the agreement presents some potential risks to competition that should be carefully weighed in the public interest analysis.
The DOJ concurs with the Department's views, as reflected in numerous decisions to grant airlines' requests for code-share authority, that these agreements have the potential to promote the public interest by creating consumer and pro-competitive benefits that airlines cannot provide on their own. Potential public interest benefits occur when an airline extends the reach of its route network by code-sharing on flights operated by an airline that operates a route network in another geographic region -- i.e, an end-to-end network combination.
Most domestic end-to-end network combinations involve code-share arrangements between major trunkline air carriers and commuter airlines that operate route networks linking the trunkline carriers' hubs with smaller cities in the region. Trunkline and commuter airlines' networks seldom overlap to any appreciable extent, because they operate different equipment and are more efficient at serving different markets and route structures.
International code-share agreements can also enable both the U.S. and foreign partner to extend significantly the reach of their individual networks. By linking largely end-to-end domestic and foreign networks, the code-share partners can offer "on-line" service to smaller cities located beyond its foreign partner's gateways, which it could not serve profitably with its own equipment, due to the small number of international passengers and cabotage prohibitions on foreign carriers serving local traffic. These network expanding agreements may generate two types of public interest benefits.
First, network expanding code-share agreements enable airlines to offer what many passengers consider to be the equivalent of single-line service to large numbers of passengers who otherwise would be limited to interline connections.(2) These passengers include "double connect" passengers whose routes include U.S. cities located beyond U.S. gateway airports, and foreign cities located beyond foreign gateway airports. Some "single connect" passengers may also be limited to interline connecting services if no single carrier operates between the interior city and a U.S. gateway, as well as between that U.S. gateway and the foreign gateway (or vice-versa for city-pair routes involving a foreign interior point). To the extent that passengers prefer code-share connecting services over traditional interline connecting services, the code-share agreements enable airlines to provide valuable services that otherwise would not be available in those markets.
In addition, by linking largely end-to-end route networks, international code-share agreements can inject additional competition into city-pair markets currently served by competitors' on-line connecting flights. If neither of the code-share partners currently serves that city-pair,(3) the new combined code-share service might be an effective competitive alternative to other carriers' existing on-line services, and constrain prices in that market.(4)
As in other network industries, competitive problems potentially occur in the overlap markets. In international code-sharing agreements linking largely end-to-end networks, those overlaps typically occur on U.S. gateway to foreign gateway routes currently served by both code-share partners. If numerous carriers operate international routes between the U.S. and the foreign region at issue, connecting at different U.S. and foreign gateways, the overlapping markets of concern may be limited to local passengers traveling between the U.S. and foreign gateway cities.
In short, the potential for consumer and pro-competitive benefits occurs in those markets where the code-share partners gain the ability to offer on-line services beyond their existing individual route systems. The competitive problems occur where their routes overlap with their code-share partners. With largely end-to-end combinations, the number of markets where pro-competitive benefits may be created is large, and thus the potential for promoting the public interest is relatively high. Similarly, the number of overlap markets is relatively low in end-to-end combinations, so the overall risk to competition is smaller.
The obverse is also true. With largely horizontal airline route combinations, the code-share partners' combined route network is not significantly larger than either of the existing networks. Consequently, they can jointly provide new on-line services to few city-pair markets currently served only by interline services, and they add significant competitive vigor to few city-pairs. The potential for code-share agreements between largely horizonal networks to create pro-competitive benefits and promote the public interest, therefore, is relatively low; and the risk to competition is relatively high.
When the Department considered largely end-to-end network combinations in previous international code-share agreements, it properly considered both pro-competitive benefits in beyond markets and risks to competition in overlap markets. If potential competitive harms in overlap markets could be minimized while preserving the pro-competitive gains in other markets, the public interest could be maximized by approving the applications with conditions. Thus, the Department properly incorporated certain "carve-outs" that were negotiated by the DOJ and the parties in the Delta/Swissair/Sabena/Austrian Airlines, United/Lufthansa, American/Canadian and United/Air Canada alliance agreements. With those carve-outs, the Department excluded certain overlap U.S. gateway to foreign gateway city pair routes from the scope of its grant of antitrust immunity to those parties, while preserving the potential for pro-competitive benefits for passengers traveling on all other city pair routes.
Carve-outs do not necessarily eliminate all risks to competition. For example, code-sharing firms may compete less aggressively in price or capacity on overlap routes if they fear that such competition might undermine the agreement to cooperate on beyond traffic. Also, if code-share partners obtain a dominant position in terms of daily frequencies (or on travel agents' CRS screens) in specific city pairs, or on routes between the U.S. and foreign countries or regions in general, they may be able to use their combined presence to deter the entry or expansion of other competitors in the market, or to drive smaller, unaligned carriers out of the market.
By any measure, the proposed American/TACA Code-Sharing Agreement is largely horizontal rather than end-to-end. As noted in the Department's Show Cause Order, American and a TACA carrier operate overlapping nonstop flights on virtually all routes between Miami, the principal Latin American hub in the United States, and Central American gateway cities -- including flights between Miami and Belize City (Belize), Guatemala City (Guatemala), Managua (Nicaragua), Panama City (Panama), San Jose (Costa Rica), San Salvador (El Salvador), Tegulcigapa and San Pedro (Honduras).(5) Last summer, these flights accounted for all 21 of American's daily nonstop U.S.- Central American flights, and seven of the TACA Group's eight daily nonstop U.S. - Central American flights.(6)
American can extend its existing network through code-sharing with TACA carriers by using TACA's regional network in Central America to extend its reach to passengers traveling between the United States and smaller Central American cities beyond the Central American gateways served by its nonstop flights. These cities, however, account for very few passengers traveling between the U.S. and Central America. Information provided by American to the DOJ indicates that in 1995 (the last full year for which we have data) an average of fewer than [redacted] American passengers per day, or less than [redacted] percent of its passengers on these routes, traveled beyond each of the Central American gateways that it served with nonstop flights.(7)
In both absolute and percentage terms, code-share service to the points would represent a very small expansion of American's existing network. And, only a fraction of those passengers would gain access to new on-line services in city-pair markets currently served only by interline services, or in city-pair markets currently served by a competitor's on-line flights but no American or a TACA on-line flights.
It appears that the code-share agreement cannot possibly extend the reach of American's existing network to new on-line services for single-connect city pairs in which one end point is a Central American point beyond American's Central American gateways. American would extend the scope of its network in Central America by gaining access to TACA's Central American network; but since TACA currently serves the only U.S. gateway that American uses for its Central American service (Miami), TACA can already provide on-line single-connect service for every city pair to which American would gain access by virtue of the proposed agreement. The only class of potential beneficiaries of the expansion of American's Central American route network, therefore, is that fraction of the [redacted] passengers per day whose double-connect itineraries included end points beyond American's existing Central American gateways and beyond its Miami gateway.
Under the agreement, TACA could gain access to a larger number of passengers traveling beyond U.S. gateways to interior U.S. points served by American. But again, American already provides on-line service on virtually all of those single-connect routes; and potential new on-line services are also concentrated in double-connect markets.
The greatest potential for a substantial expansion of the applicants' networks is their plan to create a new gateway for Central American traffic at American's Dallas/Ft. Worth hub. For beyond passengers, this would create no new significant competitive options -- it would merely give American and TACA the opportunity to route beyond passengers over American's Dallas/Ft. Worth hub, in addition to its Miami hub. Moreover, it appears that every potential U.S. code-share partner for TACA would either create a new U.S. gateway for U.S.- Central American traffic at one of its hubs, or add U.S. - Central American flights at an existing Central American gateway.(8)
This almost exclusively horizontal American/TACA agreement stands in stark contrast to the largely end-to-end agreements that the Department has approved in the past. Most significantly, the Delta/Swissair/Sabena/Austrian Airlines, United/Lufthansa, American/Canadian and United/Air Canada alliances involved fewer problematic overlapping city pairs, and significantly greater opportunities for the code-share partners to extend the reach of their networks beyond foreign gateways.
As recognized by the Department, the risk of harm to overlapping city-pairs markets in this case is not trivial. In the overlapping nonstop Miami-Central American city pairs, American and TACA have combined market shares ranging from a low of 88% to a high of 100%.(9) Where competition exists Iberia is the most significant competitor.(10)
The Department has attempted to minimize the threat to competition on the overlapping routes by imposing several conditions to its approval of this agreement. If this agreement held out the potential for conferring pro-competitive benefits on large numbers of passengers, it might be appropriate to approve it subject to conditions crafted to minimize the accompanying competitive problems. But, the Department should recognize that it cannot eliminate the risks to competition with any conditions that it might impose, and this agreement does not offer significant pro-competitive efficiencies. Under these circumstances, the threats to competition that inevitably persist despite the best efforts to eliminate them through conditions should be a matter of concern.(11)
1. Order at 2.
2. When airline passengers have a choice between an on-line connecting service and an interline connecting service on the same route, experience indicates that most will choose the on-line service.
3. Typically, the code-share partners will be able to provide code-share service on single-connect routes that neither currently serves with on-line flights when the U.S. carrier provides service between a U.S. gateway city served by its partner, but not itself, and interior U.S. points (or vice-versa for single-connect routes involving foreign points beyond foreign gateway cities.
4. When both the U.S. and foreign code-share partner serve an overlapping foreign gateway to U.S. gateway route, the U.S. carrier will be able to offer code-share flights to foreign points that are linked to its foreign partner's gateway (and vice-versa for the foreign partner). Although one partner extends its route network in this situation, it does not offer an on-line single-connect service for any city-pairs that do not already have true on-line service by a single carrier. More importantly, it does not add any competitive discipline to those markets, because the carrier that controls the aircraft in the gateway to gateway markets maintains the ability to determine prices in the beyond markets served through its gateway. The competitive impact in those city-pair markets, therefore, is neutral.
5. Order at 7.
6. In the summer of 1997, a TACA airline also operated nonstop flights on one Miami-Central American city pair that was not also served by American (Miami-Roatan, Honduras). Official Airline Guide, June 9-15, 1997.
7. We do not have accurate data on foreign air carriers'
U.S. - Central American passengers who traveled beyond Central American gateways to smaller cities in the region. If their experience is similar to American's, however, the total average number of all such passengers would still be under [redacted] per day (American operates approximately 33% of all U.S. - Central American flights).
8. If the joint applicants had submitted an application to code-share only at Dallas/Ft. Worth, the balance between whatever pro-competitive benefits would arise from code-sharing on those routes and competitive harms, obviously, would have been quite different from the balance in this application.
9. Order at 26.
10. Iberia proposes to enter into its own code-share agreement with American on its Miami-Central American routes.
11. The risk to competition through coordination on overlap routes is particularly worrisome where, as here, both carriers have a history of illegal collusive behavior. See, United States v. American Airlines, Inc. 743 F.2d 1114 (5th Cir. 1984) (American Airlines invited Braniff to engage in price-fixing); U.S. v. Air Florida, et al. S.D. FL Cr. Dkt. 84-260 (1984) (TACA International nolo plea on price-fixing charge).