|This document is available in two formats: this web page (for browsing content) and PDF (comparable to original document formatting). To view the PDF you will need Acrobat Reader, which may be downloaded from the Adobe site. For an official signed copy, please contact the Antitrust Documents Group.|
COMMENTS OF THE UNITED STATES
TABLE OF CONTENTS
COMMENTS OF THE UNITED STATES
Delta Air Lines ("Delta") and US Airways (collectively, "the parties") propose a permanent exchange of more than 300 slots (representing the rights to more than 150 daily round trips) at LaGuardia Airport (LGA) and Ronald Reagan Washington National Airport (DCA) ("the transaction"). The parties seek a waiver from a Federal Aviation Administration ("FAA") order, which currently prohibit the permanent transfer of LGA slots. The FAA may grant the waiver if it is in the public interest. Its public interest inquiry is guided by several pro-competitive principles, including "avoiding unreasonable industry concentration, excessive market domination, monopoly powers, and other conditions" that would lead to an increase in fares;1 "encouraging ... an Air Transportation system relying on actual and potential competition" to provide low fares;2 and "encouraging entry into Air Transportation markets."3 The FAA seeks comments on its tentative decision to grant the requested waiver, which the FAA believes would be in the public interest if conditioned on the divestiture of limited DCA and LGA slots.4
The Department of Justice ("DOJ") offers these comments in support of the FAA's tentative decision.5 Our comments are based on an extensive investigation of the transaction during which we reviewed hundreds of thousands of documents from Delta, US Airways, [REDACTED]; analyzed data from the parties and public sources; [REDACTED]; and interviewed other industry participants and observers. Our comments are also based on our experience analyzing competition in the airline industry, including analyses and comments submitted in prior slots-related proceedings before the FAA.6 The DOJ has concluded that the FAA's proposed waiver with conditions will be in the public interest because it will free up slots for other carriers, facilitating entry at LGA and DCA, increasing competition and lowering fares for consumers, without interfering with the purported benefits of the transaction.
The Department of Transportation ("DOT")/FAA and Congress have long faced the challenge of managing congestion at the nation's busiest airports while ensuring efficient allocation of scarce airport capacity. LGA has been governed for over three years by a temporary order that prohibits the permanent transfer of slots. US Airways and Delta propose a transaction that is expressly prohibited by the LGA order and that would – contrary to the FAA's longstanding efforts to open LGA and other slot-controlled airports to more competition – reduce the likelihood of entry and diminish competition.
The availability of slots is a substantial barrier to entry at LGA and DCA, especially for low cost carriers ("LCCs"). Slot holdings at these airports are concentrated in the hands of large legacy carriers, primarily US Airways and Delta. Although LCC entry would undoubtedly benefit consumers, it could undermine the revenues and profits of the large carriers at LGA and DCA, giving them little incentive to sell or lease slots to LCCs.
The parties' transaction will make LCC entry at LGA and DCA less likely, depriving consumers of the lower fares and vigorous competition that LCCs bring to the marketplace. It will increase the share of slots held by Delta and US Airways, giving them more revenue and profits at risk due to entry, more markets for which it will be in their interest to forestall entry, and thus, less incentive to sell or lease slots to a potential entrant. Moreover, absent the transaction [REDACTED].
The transaction also will reduce competition between Delta and US Airways at DCA and LGA. US Airways and Delta are principal rivals at the airports. Post transaction, however, Delta will shrink substantially at DCA, reducing its ability to compete effectively with US Airways. Similarly, US Airways will shrink substantially at LGA, reducing its ability to compete effectively with Delta.
The FAA's proposed slot divestiture is necessary to protect consumers from competitive harm. It will open up DCA and LGA to entry by carriers that traditionally have found it extremely difficult to purchase slots. Such entry – particularly LCC entry – will benefit the public by increasing competition at LGA and DCA, bringing lower fares to consumers in New York and Washington.
The transaction will increase the slot holdings and shares of the dominant carriers at LGA and DCA. See Table 1 on p. 6 below. Pursuant to the transaction, US Airways will transfer 250 LGA slots and lease an additional 30 LGA slots to Delta; in turn, Delta will transfer 84 DCA slots to US Airways. This will increase the share of DCA slots held by US Airways – already the largest slot holder at DCA – from 44 percent to 54 percent.8 The transaction will increase the share of LGA slots held by Delta – currently the second largest LGA slot holder – from 24 percent to 49 percent, making it the largest slot holder at LGA. Delta's post transaction share of LGA slots will be substantially larger than US Airways' current share, meaning that the share of the largest slot holder at LGA will increase from 36 percent (US Airways' current share) to 49 percent (Delta's post transaction share). The HHI for slot holdings, a measure of slot concentration, will increase by 600 from 2394 to 2994 at LGA, and by 626 from 2756 to 3382 at DCA. Under the DOJ/FTC Horizontal Merger Guidelines, a merger producing such concentration increases in a highly concentrated market would be presumptively likely to create or enhance market power or facilitate its exercise.9
Source: DOJ calculations based on FAA data.
The transaction will further a long-term trend of increased concentration at LGA and DCA that has been driven largely by industry consolidation rather than secondary slot market transactions. According to one study, the share of slots held by American Airlines, Delta, and US Airways at LGA increased from 27 percent in 1986 to 69 percent in 2006; at DCA, the share of slots held by the same three airlines increased from 25 percent in 1986 to 66 percent in 2006.10 Today, the slot shares of these three legacy carriers is more than 80 percent at both airports.11
The FAA has concluded that the increased concentration resulting from the transaction will lead Delta at LGA and US Airways at DCA to "rely on their increased dominance to maintain or enhance their premium fare structure in markets served at both airports."12 This is consistent with an extensive body of empirical work finding that airport concentration is associated with higher fares.13
In contrast to large incumbents' extensive slot holdings, LCCs currently hold only 6.8 percent of LGA slots and a mere 3.3 percent of DCA slots.14 Most LCC slot acquisitions at LGA and DCA have been the result of congressional or DOT/FAA action rather than secondary slot market transactions, and the few secondary slot market transactions involving LCCs have mostly been the result of sales due to extreme financial distress.15
The small presence of LCCs at LGA and DCA has deprived consumers of the vigorous competition and low fares that LCCs bring to the marketplace. Since airline deregulation began, it has become clear that open and fluid entry produces lower fares and better service. The effect of LCC entry on fares provides perhaps the most dramatic evidence. DOJ empirical work, discussed in Appendix A, suggests that the presence of an LCC on a nonstop route reduces fares by roughly 25 percent.16 This is consistent with an extensive economic literature showing the large effect of LCcs on fares relative to other classes of carriers.17 DOJ empirical work also suggests that airports with higher LCC penetration have much lower fares.18 The proposed transaction will make it less likely that consumers will realize the benefits of LCC competition at LGA and DCA.
In principle, limited or non-incumbent carriers could acquire slots in the secondary slot market or through FAA reallocation of underutilized slots. In practice, however, slots have not become available to these carriers through either method.
LGA and DCA slots are highly concentrated in the hands of Delta and US Airways, both of which have little incentive to sell or lease slots to other carriers that would compete with them. The disincentives to sell or lease to LCCs are particularly strong because their low fares – a substantial benefit to consumers – are a substantial threat to US Airways and Delta. Both carriers recognize that LCC entry at an airport erodes revenues and profits while decreasing fares for consumers. [REDACTED].
Concern about LCC entry is especially great at DCA and LGA, where limited LCC presence and slot controls protect high fares and profits for incumbent carriers. [REDACTED] US Airways and Delta understand that LCC entry would substantially lower their protected fares and profits at these airports. [REDACTED].
Despite FAA regulations designed to ensure that underutilized slots are reallocated to carriers that will use them efficiently, incumbent carriers continue to hoard slots, in part, to keep them out of the hands of new entrants. If US Airways or Delta could not meet the FAA's "use or lose" requirement for some of their slots (which mandates that slots be used 80 percent"of the time over a two-month period) they would be forced to return the slots to the FAA, which might reassign the slots to another carrier.24 Given the costs to them of LCC entry (i.e., reduced fares, revenues, and profits) it should not be surprising that US Airways, Delta, and other large incumbent carriers have adopted practices designed to meet the FAA's use or lose requirement at minimum cost, keeping slots from falling into the hands of other carriers.25
One way to minimize the cost of meeting the 80 percent use or lose requirement is to fly excessive frequencies (which increases slot utilization) or small planes (which reduces the cost of a flight). At LGA, US Airways has flown small planes [REDACTED].
Another way to hold onto slots without using them productively is referred to as "babysitting." Babysitting involves a carrier using flights on slots that exceeds the 80 percent threshold to cover the use or lose requirement for an unused slot in the same time period. For example, if a carrier has 10 flights over the threshold on slot A, and no flights on slot B, the carrier may "assign" those 10 flights to slot B; in other words, the carrier averages its flights over its slot holdings. [REDACTED] Carriers also "babysit" slots for each other. This involves the short term transfer of slots from a carrier incapable of meeting the use or lose requirement to another carrier capable of meeting the requirement, thereby protecting the slots from being returned to the FAA. [REDACTED].
The transaction will exacerbate the disincentives of US Airways and Delta to sell or lease slots to other carriers. With increased slot shares at LGA and DCA, the parties will have more revenue and profit at risk due to entry, more markets for which it will be in their interest to forestall entry and thus, even less incentive than exists today to sell or lease slots to a potential entrant.32
At LGA, the transaction also will eliminate an important opportunity for LCCs to acquire slots. Apart from DOT/FAA or Congressional action, virtually the only way LCCs have been able to enter LGA is by acquiring slots sold by carriers in financial distress. [REDACTED].
The transaction also will reduce competition between US Airways and Delta at DCA and LGA. The parties currently compete on a number of DCA and LGA nonstop routes;34 in the past, they have competed on many others.35 As US Airways shrinks at LGA and Delta shrinks at DCA, each carrier is likely to withdraw from some routes on which they currently compete, resulting in an immediate loss of competition. [REDACTED].
In the longer run, competition between Delta and US Airways will be lost across a larger number of routes. Hub and spoke carriers, like US Airways and Delta, typically use slots to connect DCA and LGA to their "core" cities, such as hubs or focus cities where they have frequent service. Flights from DCA or LGA to core cities usually are highly profitable because large numbers of passengers fly them, often to connect to a final destination, which generates additional revenue for the airline. When deciding how to allocate its slots at a slot-controlled airport, a hub and spoke carrier will first allocate slots to its most profitable routes, typically core routes, and work its way down to other, less profitable routes. If a carrier has a small portfolio of slots, it is likely to allocate almost all of its slots to its core routes. If a carrier has slots in excess of those needed to serve core routes, it may allocate some of its slots to non-core routes.
US Airways and Delta currently have excess slots at LGA and DCA, respectively, allowing them to serve non-core routes. Their slot portfolios also are large enough that each may find it profitable to shift service from one route to another in response to profit opportunities. Indeed, Delta and US Airways have shifted service to enter LGA and DCA nonstop routes in direct competition with one another.38 Delta's recent actions with respect to its DCA slots provides a good example of the kind of longer-run competition that will to be lost with the transaction. [REDACTED] Post transaction, Delta at DCA and US Airways at LGA will hold a much smaller portfolio of slots. Each is likely to use its slots to serve primarily its core and shuttle routes; neither is likely to have excess slots to shift from those routes to other routes in response to profit opportunities.43
It is unlikely that other incumbent carriers will completely replace the lost competition between US Airways and Delta. Other carriers at LGA and DCA generally serve their hubs and major focus cities, rarely entering or exiting other routes.44 This suggests they generally are unlikely to shift service in response to a fare increase. The one exception is American Airlines, which has entered or exited several LGA routes.45 However, even accounting for American's ability to shift service at LGA, the transaction will reduce the number of carriers with "excess" slots to discipline a fare increase by the dominant carrier from two to one at LGA and from one to zero at DCA. As a result, the dominant carrier at each airport, Delta at LGA and US Airways at DCA, will be better able to raise fares and earn a (or increase its) premium after the transaction.
It also is unlikely that service from other New York or Washington area airports will completely offset lost competition between US Airways and Delta. Although other airports may be acceptable substitutes for some passengers (particularly price-sensitive passengers), they clearly are not close substitutes for other passengers, and competition among carriers at LGA and DCA matters. Indeed, data cited in the Notice show sometimes significant differences in average fares at the various airports,46 and the high values attached to slots and the carriers' efforts to protect these slots show there is differentiation between LGA and DCA and other area airports. While differences in average fares are not necessarily dispositive of market definition issues, the magnitude and persistence of the differences strongly suggest that the airports are not substitutes for some passengers.
Delta and US Airways claim that the transaction will yield significant benefits at LGA and DCA in the form of new service and increased connectivity.50 In particular, the parties argue that the transaction will permit Delta to create a hub, increasing annual capacity at LGA by more than two million seats. They also argue that the transaction will result in US Airways increasing annual capacity at DCA by one million seats. These benefits likely are overstated.
It is not surprising that the parties project increased capacity and traffic over current levels. At LGA, US Airways underutilizes its slotS.51 The appropriate baseline for measuring increased capacity and traffic is not US Airways' present inefficient use of LGA slots but how those slots would be used absent the transaction. In this case, [REDACTED].
At DCA, the appropriate baseline against which to measure any increase in capacity and traffic resulting from the transaction is not Delta's present operations, but its planned operations absent the transaction. As discussed above, [REDACTED].
Even assuming the transaction will increase capacity and traffic, the FAA's proposed slot divestiture is not likely to interfere substantially with those benefits. After the divestiture, the parties still will be able to offer significant new service that could benefit consumers. At LGA, Delta will be able to operate 120 new round trips; at DCA, US Airways will be able to operate 28 new round trips. There is little evidence suggesting that a smaller transaction – as would result if the parties accepted the terms of the FAA's proposed waiver – would be unprofitable for the parties. [REDACTED].
The consumer benefits from new entry – particularly LCC entry – that likely will result from the FAA's proposed divestiture almost certainly will outweigh any loss from US Airways and Delta making minor modifications to their proposed schedules. LCCs are likely to use larger planes than US Airways and Delta, which would increase capacity at LGA and DCA,55 and they likely would charge lower fares on routes they entered in competition with other carriers. Thus, the marginal benefit to consumers of the divestiture is likely to be large. The modifications to Delta's and US Airways's proposed schedules are likely to have only a small effect on benefits to consumers. The parties already offer extensive service at the airports, which suggests that marginal benefit to consumers of their planned additional frequencies, [REDACTED], is likely to be relatively small.
The Notice requests comments on three options through which US Airways and Delta could divest slots: (1) private sales with bi-weekly reports to the FAA on the details of the sales efforts; (2) anonymous, cash-only sales in which the FAA forwards the highest bid to the seller for acceptance or rejection; and (3) a hybrid of the first and second options, in which FAA would manage the selling process, but the identity of the buyer would not be concealed, and the terms and consideration would be negotiated by the seller. Although a number of details remain to be determined, DOJ favors the second option if that method is implemented in a sound way.
The first and third mechanisms suffer from a significant shortcoming in that they allow the sellers to know and consider the identity of a potential buyer. Any mechanism that allows the seller to choose the buyer would permit discrimination against buyers inclined to use slots to compete against the parties. As defined in the Notice, the pool of eligible buyers includes a number of carriers that US Airways and Delta would know to be highly unlikely to compete aggressively with them, such as Canadian carriers that are not permitted to fly between two points within the United States. The first and third options may therefore result in sales to carriers least likely to compete with the parties rather than to those likely to generate the greatest consumer benefit from the use of the slots.57
The second option, which preserves the anonymity of potential buyers, is the preferable option. Under this option, the FAA would specify a bid closing date and time, slot purchasers' identities would not be revealed, and sales would be on a cash-only basis. The FAA would forward the highest bid to the selling carrier and the carrier would have three days to accept or reject the bid. The Notice, however, does not explain what happens if the selling carrier rejects the bid. We encourage the FAA either to strike the provision allowing the selling carrier to reject the highest bid or to specify a narrow set of reasons a bid may be rejected (e.g., if the bid does not meet a pre-determined "reserve" price).58
We also recommend expanding the restriction on re-sales and leases of slots purchased pursuant to whichever process is adopted. The Notice proposes "precluding the carriers purchasing the slot interests acquired pursuant to this proceeding from re-selling, or leasing, them to any carriers that are not eligible under the terms of the final action we take in this proceeding." The purpose of the provision is to ensure that the divested slots stay in the hands of new entrants or limited incumbents. We are concerned, however, that the FAA's proposal creates a loophole in which an eligible carrier could purchase slots through the divestiture process and effectively resell them to a non-qualifying carrier by substituting some other set of "old" slots. We therefore recommend precluding, for some reasonable period, purchasers from selling or leasing any slots to carriers not eligible under the terms of the final action taken in this proceeding.
We also note that any purchaser of divested slots will need access to sufficient ground facilities to use the slots effectively. This may be particularly problematic at LGA, where Delta will likely control the facilities the buyer would need. The FAA should consider ways to ensure that the successful purchaser will obtain access to these facilities.
The FAA's proposed slot divestiture serves the public interest by ensuring entry and protecting competition, without interfering with the purported benefits of the transaction. LGA and DCA have been largely closed to LCCs. Absent appropriate mitigation, the proposed slots transaction between US Airways and Delta will further reduce the likelihood that LCCs and other limited or non-incumbent carriers will be able to establish a significant presence at the airports. It also will reduce competition between US Airways and Delta at LGA and DCA. The FAA's proposed divestiture is necessary to protect consumers from this harm. The FAA has appropriately limited buyers of divested slots to carriers that historically have found it difficult to acquire slots to initiate or expand service at DCA and LGA. The entry facilitated by the divestitures – which is likely to include LCC entry – will result in greater competition at DCA and LGA, increasing service and substantially reducing fares for consumers flying to or from these airports.
5. These comments are filed in response to the FAA's tentative decision. In a letter dated March 17, 2010, the parties requested the DOJ to defer filing any comments in this matter until it had an opportunity to review the parties' submissions, in particular their proposed divestitures. The details of the parties' proposed divestitures, however, have not been sufficiently developed to permit a full analysis by the DOJ at this time.
6. See Comments of the United States Department of Justice, Congestion and Delay Reduction at Chicago O'Hare International Airport, Docket No. FAA-2005-20704 (May 24, 2005); Comments of the United States Department of Justice, Notice of Alternative Policy Options for Managing Capacity a LaGuardia Airport and Proposed Extension of Lottery Allocation, Docket No. FAA-2001-9854 (June 20, 2002).
Delta has taken inconsistent positions on the competitive effects of slot shares and concentration at DCA and LGA. During Delta's bankruptcy three years ago, US Airways considered acquiring Delta. Delta resisted US Airways' overtures, arguing that the merger would cause competitive harm at DCA and LGA. See, e.g., [REDACTED] At that time, Delta argued that slot shares resulting from the merger – levels that are approximately the same as the shares that would result from the present proposed transaction – raised substantial competitive concerns. See Hearing on the State of the Airline Industry: The Potential Impact of Airline Mergers and Industry Consolidation Before the S. Comm. on Commerce, Science and Transportation, 110th Congo (Jan. 24, 2007) (testimony of Gerald Grinstein, CEO of Delta Air Lines) ("The combined carrier would overwhelming [sic] dominate at these unique airports with restricted entry due to slot controls ... At Washington National, a merged US Airways-Delta would operate nearly four times more slots as its next largest competitor ... At New York-LaGuardia, the combined carrier would operate almost twice as many slots as the next largest competitor .... "). Delta's current position is precisely the opposite.
l3. See Severin Borenstein & Nancy L. Rose, How Airline Markets Work ... Or Do They? Regulatory Reform in the Airline Industry 49 (Nat'l Bureau of Econ. Research, Working Paper No. 13452, 2007) (collecting cites).
16. See Appendix A, Section A.
18. For instance, an increase in LCC share at an airport from 0 to 20 percent corresponds with a decrease of 8 to 18 percentage points in the airport-wide average "fare premium." See Appendix A, Section B.
24. See Operating Limitations at New York LaGuardia Airport, 71 Fed. Reg. 77,854, 77,860 (2006) (minimum use requirement for LGA); 14 C.F.R. ï½§ 93.227 (West 2005) (minimum use requirement for DCA). When the minimum use rule was instituted in 1985, the FAA cited DOJ concerns that a use or lose provision was necessary to prevent large carriers from hoarding slots in an attempt to restrict service to drive up fares or foreclose entry by smaller competitors. See 50 Fed. Reg. 52,180, 52,188-89 (1985).
25. 0ther reasons why an incumbent might underutilize slots include a desire to preserve some flexibility for future operations. As explained in previous DOJ comments, transaction costs for airlines considering new service at the affected airports and uncertainty over the scope of the property right being sold may also be contributing factors. See Comments of the United States Department of Justice, Congestion and Delay Reduction at Chicago O'Hare International Airport, Docket No. F AA-2005-20704 (May 24, 2005).
26. US Airways has a much lower average seats flown per slot than other legacy carriers at each airport. See Appendix A, Section C, Table A-3a.
32. In the event LGA or DCA slots became available for lease or sale, a large incumbent would have a strong incentive to outbid other carriers for slots to prevent entry. Cf. Congestion and Delay Reduction at Chicago O'Hare International Airport, Docket No. FAA-2005-20704 (May 24, 2005) at 6 ("Strategically purchasing available slots can be an effective entry deterrent, especially since multiple slot holdings required for significant entry rarely come up for sale. "). The transaction will make that incentive even stronger for Delta at LGA and US Airways at DCA.
38. Appendix A, Table A-6a shows that over the last eight years, Delta has entered and exited several nonstop DCA routes on which it competed with US Airways. Table A-6b shows that US Airways and Delta have entered and exited several nonstop LGA routes on which they competed with one another.
45. American Airlines is excluded from Appendix A, Table A-7b because the list of routes it has entered or exited is fairly extensive.
51. See Appendix A, Section C, which describes the extent to which US Airways underutilizes its LGA slots.
55. See Appendix A, Section C, Table A-3a.
57. Under well-accepted economic theory, it could be argued that no matter which carrier purchases the slots, as long as transaction costs are low and secondary sales are permitted, slots should be bought and sold until they are put to their highest-valued use. If this theory is correct, the parties would be willing to sell slots to the highest bidder (as a sale to anyone other than the highest bidder would nonetheless result in the slots ultimately making their way into the hands of the highest bidder). So, if the theory holds, a rule requiring sale to the highest bidder should not harm the parties. In practice, however, as described above, a number of factors inhibit efficiency-enhancing transfers, and we are not confident that the secondary market is sufficiently liquid to achieve these results.
In order to help quantify the potential benefits of new entry at DCA and LGA, which under the DOT notice would likely come from "low-cost carriers" ("LCCs"), we empirically tested the extent to which entry or exit by LCCs on a route has historically changed fares for that route.1 As LCCs have little presence at DCA and LGA, we estimated the impact of entry or exit of a nonstop LCC competitor on routes with no existing nonstop LCC competitors.
The results below are derived from representative difference in means calculations of LCC impact on prices. Specifically, we calculate the percentage price change from LCC entry or exit on a route and determine the average change across all routes.2 Data for this exercise are from the DOT's DBIB database, which contains quarterly revenues and passenger counts based on a sample of tickets, and OAG Aviation, which contains flight frequencies by airport and airline. Our data cover the time period from third quarter 2002 through third quarter 2009. To estimate price, we calculated the average price per "coach class" passenger per quarter on nonstop routes.3 The average price for all quarters on a route with LCC presence was then compared to the average price for all quarters on the same route without LCC presence to derive the change in fares from LCC entry or exit on a route.4
A carrier was determined to be in the route if it flew at least 60 non-stop flights in each direction on a route in a given quarter. If the carrier flew fewer than 30 flights in each direction on the route, then the carrier was not considered in the route. When a carrier flew between 30 and 60 flights in a quarter, the route-quarter observation was not included in the analysis as it was unclear whether the carrier was a significant competitor in that quarter. The results below are robust to alternate thresholds.
Table A-1a identifies the average price impact of the presence sofa LCC competitor under the baseline scenario described above, as well as for particular subsets of the overall data, and Table A-1b identifies the median impact on the number of passengers under the same scenarios. It finds an economically significant impact from the presence of an LCC on nonstop route-level prices, ranging from a 21 % to 27% average price decrease and a 68% to 118% median increase in number of passengers depending on the data examined. Average price and passenger volume effects are statistically significant at a 99% confidence level. The size of the price effect that we find here is, if anything, on the low end of estimates of the price effect of LCC entry found in the economic literature.5
The routes in the analysis can be limited to those where at least one of the two airports is in the top 50 airports nationwide (based on annual domestic flights), and the length of the route is less than 1500 miles. The two restrictions combined limit the analysis to routes that are similar to those that can be found at DCA and LGA, which both have perimeter rules that restrict flight distance. The columns of the table place restrictions on the number of legacy incumbent carriers on a route. For routes similar to those found at DCA and LGA, and with only one or two legacy carriers, the Lee effect ranges from a 24% to 27% average decrease in fares and a 76% to 108% median increase in the number of passengers. Average fare and passenger volume effects are again statistically significant at a 99% confidence level. In conclusion, the effect of a Lee's presence on fares and the number of passengers on a route is significant.6
B. AIRPORT-WIDE FARES AS A FUNCTION OF LCC SHARE
The evidence in Section A demonstrates the significant impact of LCC presence on route level prices. In this section, we examine how average fares paid by domestic passengers who travel on nonstop flights to or from a given airport vary over time as the share of total nonstop flights that are flown by LCCs changes at that airport.7 We also look at how the total number of domestic nonstop passengers changes as a function of LCC presence. The large consumer benefits generated by LCC competition are easily discernable at the airport level as well.
LCCs fly many fewer flights at DCA and LGA compared to other comparably-sized airports. In the third quarter of 2009, LCCs flew 3.8% of all domestic nonstop flights at DCA and 9.5% at LGA, compared to an average of 31.9% across all other airports among the top 50, ranked by number of nonstop domestic flights. In addition, while airports have on average seen significant growth in LCC presence (relative to other carriers) in recent years, the LCC presence at DCA and LGA has been more static. From third quarter 2002 through third quarter 2009, the mean LCC share at top 50 airports excluding DCA and LGA increased by 7.9 percentage points, from 24% to 31.9%. At DCA, in contrast, LCC share increased by only 2.5 percentage points (1.3% to 3.8%), and at LGA by 3.6 percentage points (5.9% to 9.5%), over the same period. Both of these facts can be attributed in part to the existence of slot constraints that inhibit entry by LCCs.
We find that additional LCC presence at an airport over time is associated with large and statistically significant price decreases and volume increases at that airport. Our analysis shows, for example, that as LCC presence at an airport increases by 20 percentage points (from zero to 20%, say), the average airport-wide fare premium falls by an average of 8 to 18 percentage points, depending on the sample of airports we examine. Similarly, a 20 percentage point increase in LCC presence is associated with a 15 to 30% increase in the number of passengers at that airport.
Data are taken from the Department of Transportation DBIB database and OAG Aviation for third quarter 2002 through third quarter 2009.8 These data are used to compute the average airport-wide fare premium or discount relative to all airports using the method described in airport-wide fare premium or discount relative to all airports using the method described in Borenstein (2005).9 LCC presence is calculated as the percentage of total domestic nonstop flights at the airport operated by LCCs in a given quarter. We control for overall airport size effects with a variable representing the total number of domestic nonstop flights at that airport.
Table A-2 shows results from a series of fixed-effect regressions that separately estimate, within each airport, each of the change in airport fare premium/discount and the change in the total number of nonstop passengers as a function of LCC share at the airport, controlling for the total number of nonstop domestic flights at the airport and including quarterly dummy variables to measure time effects. These regressions are run using data from the top 50 and then the top 100 U.S. airports, measured by total nonstop domestic U.S. flights in third quarter of 2009.10 All of the estimated LCC share effects are statistically significant at the 99% level.11
These results show that, for instance, 10 extra percentage points of LCC share at an airport reduces on average the airport-wide price premium or discount by 4 to 9 percentage points, and increases the total number of passengers at the airport by 7.7 to 14.8%, depending on the sample used. In other words, additional LCC presence at an airport is associated with significantly lower average fares and higher passenger volumes at that airport.
[REDACTED] This section discusses the empirical evidence supporting the argument that the divestiture would increase the utilization of slots at each airport.
To accurately compare the size of aircraft used by each carrier, the analysis must account for the fact that commuter slots at DCA come with a restriction as to the size of the aircraft that can be flown.12 Roughly 20% of all slots at DCA are commuter slots, and as of August, 2009, over 27% of the slots for which US Airways was the published carrier were commuter slots. Using OAG schedule data for Tuesday, August 25, 2009, we identified the aircraft size of each flight flown.13 In order to set aside commuter slots, we dropped the proportion of flights with the smallest aircraft equal to the proportion of each carrier's slots that are commuter slots. Using the remaining data, we compared the average aircraft size of the different carriers at both DCA and LGA (where there are no commuter slot restrictions).
We find that US and Delta use smaller aircraft on average than do the carriers eligible to purchase the slots at both airports. Table A-3a provides a comparison of the average seats per "non-commuter" flight for US Airways, Delta, and the eligible carriers out of DCA and LGA on August 25, 2009. As can be seen from the table, US and Delta operate significantly smaller aircraft than do the carriers that are eligible to purchase the divested slots. Table A-3b reports the findings of a similar exercise applied to domestic flights at JFK, EWR, and ORD, three slot controlled or recently slot controlled airports. Table A-3b shows that the largest carrier at each airport uses significantly smaller aircraft than do the eligible buyers in a hypothetical divestiture of slots.14 Table A3-c presents the average load factors (passengers/seats) calculated from DOT's 2009 T-100 data for US, Delta, and eligible purchasers at DCA and LGA. The data show that US and Delta not only operate smaller aircraft, on average, they also carry fewer passengers per seat than the eligible purchasers.
While some of these differences in aircraft size and load factors at DCA and LGA may be explained by natural hubbing patterns, hubbing does not yield the same net benefits at slot-constrained airports as it does at unconstrained airports. This is because service to smaller markets potentially brought about by hubbing crowds out service and competition on larger routes. Considerably more customers are affected by the reduction in service on larger routes than benefit from potential additional service on smaller routes. Similarly, flights with relatively low load factors at slot-controlled airports crowd out flights that would have higher load factors. Neither crowding out effect is present at airports that are not slot-controlled, where the numbers of flights are not constrained by regulations.
D. EFFECTS OF PROPOSED DIVESTITURE ON PARTIES' ROUTE CHOICES
Table A-4. [REDACTED]
E. ADDITIONAL TABLES REFERENCED IN MAIN TEXT
Routes with one nonstop carrier
Note: Daily flights are calculated using an average of frequencies in each direction on a given route.
Source: DOT data
Routes with one nonstop carrier
Note: Daily flights are calculated using an average of frequencies in each direction on a given route.
Note: Daily frequency calculation exclude entry and exit quarters, except as noted below. If service dates from 2002 Q1, that quarter is only treated as an entry quarter if the average daily frequencies are less than half of the following quarter. Similarly, if service continues 2010 Q1, that quarter is only treated as an exit quarter if frequencies are less than half of the preceding quarter.
b. Since there are no quarters other than entry and exit quarters, average daily frequencies are taken from that quarter
c. Since entry and exit occurred in the same quarter, average daily frequencies are taken from that quarter
d. American flew this route beginning with 2nd quarter 2005.
e. AirTran flew this route from 2008Q3 to 2008Q4
f. Since the data begins in 2002Q1, the listed entry/exit carrier is the exit carrier only.
g. Both carriers operated only during the second quarter of each year. Subsequently, since 2010Q2 data is not available, it is unclear which, if either, carrier has exited.
h. American flew this route from 2002A1 through 2004Q4
i. CanJet flew this route from 2004Q2 through 2004Q3.
Source: DOT data
a. Since there are no quarters other than entry and exit quarters, average daily frequencies are taken from those quarters.
Source: DOT data
a. Since there are no quarters other than entry and exit quarters, average daily frequencies are taken from those quarters.
Source: DOT data
l. We have defined the LCCs to be B6, DH, F9, FL, NK, SY, TZ, VX, and WN for the purpose of this analysis. The "legacy" carriers have been defined as AA, AS, CO, DL, HP, NW, UA, US, YX, though HP and YX did not exist before deregulation.
2. An airline consumer price index is calculated in order to adjust prices for trends that are unrelated to the entry or exit of an LCC The index is based on the average price paid for nonstop travel within the continental U.S. (excluding Alaska) in each year-quarter using 2009 third quarter as the baseline.
4. As with many differences in means analyses that relate changes in the number of competitors to prices, these results could be biased if LCCs' selections of which route to enter are correlated with other variables that affect prices. However, this bias, if it exists, would lead to our estimates understating the true effect of LCC entry. The calculated entry coefficient will underestimate true effects if LCCs are more likely to enter routes for which they anticipate an upcoming upward shock to demand or pricing not caused by their entry. This is much more likely than an effect that runs in the opposite direction.
5. For example, the following papers find an LCC effect of roughly 50%: Transportation Research Board, National Research Council, Entry and Competition in the U.S. Airline Industry: Issues and Opportunities (1999); Martin Dresner, JiunSheng Chris Lin & Robert Windle, The Impact of Low-Cost Carriers on Airport and Route Competition, 30 J. Transp. Econ. & Pol'y 309 (1996).
6. Several different types of calculations were made to check for robustness. In fact, comparing the four quarters prior to entry or exit with the four quarters following entry or exit results in similar effects. Further, using regression analysis methods also results in similar effects.
8. We examine one-way fares, dividing roundtrip fares by two. We adjust fares over time based on CPI changes, as described in footnote 2. Legacy carrier fares that are classified as "first class" are discarded, though comparable fares for LCCs are retained due to inconsistencies in how LCCs categorize their fares in the data. If anything, this adjustment biases our estimates of the LCC impact downward.
9. As in Severin Borenstein, U.S. Domestic Airline Pricing, 1995-2004 (U.C. Berkeley Competition Pol'y Center, Working Paper No. CPC05-48, 2005), the measure of "airport-wide" fares here is calculated for domestic (excluding Alaska and Hawaii) flights to and from a given airport by comparing the sum of revenues within a 50 mile one-way distance band to the sum of their revenues if they were priced at the national average price of all flights within that 50 mile one-way distance band. The weighted average of these price differences over all flights at the airport represents the airport-wide premium or discount. So, for example, if an airport's weighted average fare were $150 and the nationwide weighted average for comparable distance routes were $115, its "fare premium" would be 31 %. In our regressions, we examine changes in this fare premium as a function of changes in LCC share and other factors.
10. LGA and DCA rank 12th and 18t\ respectively, on this metric. We exclude airports in Nantucket, Hyannis (Cape Cod), and Martha's Vineyard from the top 100 airports. These airports are primarily served by airlines that don't regularly report fare data to DOT, and thus they lack sufficient data to enable calculation of a fare premium in all quarters.
14. It should be noted that JetBlue is the second largest carrier at JFK and uses an average aircraft size of 135.6 seats. Jet Blue is eligible to purchase divested slots at both DCA and LGA. Eligibility of buyers at JFK, EWR, and ORD was based on the same criteria used at DCA and LGA.