FTC Staff Comment Before the Federal Aviation Administration In the Matter of High Density Traffic Airports: Slot Allocation and Transfer Methods
us-ftc-advocacy-filings-a76d519548f9b529d6310ba18a102af4FTC Staff Comment Before the Federal Aviation Administration In the Matter of High Density Traffic Airports: Slot Allocation and Transfer MethodsAdvocacy Filing
Matter Number
V910024
- '''', COMMISSION AllTHORllED BEFORE THE FEDERAL AVIATION ADMINISTRATION DEPARTMENT OF TRANSPORTATION WASHINGTON, D.C. In The Matter of High Density Traffic Airports: Slot Allocation and Transfer Methods Docket No. 25758 Notice No. 88-18A 14 C.F.R. Part 93 Comment of the Staff of the Bureau of Economics * of the Federal Trade Commission November 15, 1991 * This comment represents the views of the staff of the Bureau of Economics of the Federal Trade Commission. They are not necessarily the views of the Commission or any individual Commissioner. Inquiries regarding this comment should be directed to Bruce H. Kobayashi (202-326-3363) of the FTC's Bureau of Economics.
I. Table of Contents Introduction and Summary 1 II. Expertise of the Staff of the Federal Trade Commission. . . . . . . . . . . . . . . 4 III. Background of and Issues contained in the SNPRM. . . .. 6 A. Background of the CUrrent Slot Rule . . . . 6 B. Background of the SNPRM and Proposed Changes to the FAA Slot Rule . . . . . . .. ..... 9 IV. An Analysis of the Proposed Changes in the Slot Rule 13 A. The Relationship Between Slot Share and Slot Use 14 (i) Theory and Hypotheses . . . . • . . . . . . . 14 (ii) Empirical Results . . . . . •. ... 26 (a) Average Slot Use . . . . . . . . . . . . 27 (b) Leased Versus Owned Slots . . . . . . . . 30 (c) Regression Analysis . . .. ... 39 B. Changes in the "Use or Lose" Rules . . . . 54 C. Restrictions on the Sale and Transfer of Lottery Slots . .. .... ... 59 V. Conclusion . ii 68
(., BEFORE THE FEDERAL AVIATION ADMINISTRATION DEPARTMENT OF TRANSPORTATION WASHINGTON, D.C. In The Matter of High Density Traffic Airports: Slot Allocation and Transfer Methods Docket No. 25758 Notice No. 88-18A 14 C.F.R. Part 93 Comment of the Staff of the Bureau of Economics 1 of the Federal Trade Commission I. Introduction and Summary The staff of the Bureau of Economics of the Federal Trade Commission (FTC) appreciates this opportunity to respond to the Federal Aviation Administration's (FAA) Supplemental Notice of Proposed Rulemaking (SNPRM).2 The FAA is seeking comments on several proposed amendments to the regulations governing allocation and transfer of landing slots at the four High Density Traffic Airports (HDTA).3 The general intent of the amendments now proposed is "to make slots available to new entrants through encouraging the sale 1 This comment represents the views of the staff of the Bureau of Economics of the Federal Trade Commission. They are not necessarily the views of the Commission or any individual Commissioner. Inquiries regarding this comment should be directed to Bruce H. Kobayashi (202-326-3363) of the FTC's Bureau of Economics. 2 See High Density Traffic Airports: Slot Allocation and Transfer Methods -- Supplemental Notice of Proposed RUlemaking, 56 Fed. Reg. 46674 (September 13, 1991). 3 14 C.F.R. Part 93. 1
,. {..
of marginally used slots by incumbent carriers, in order to
promote competition in the airline industry to the extent
possible without substantial disruption of existing air
service. ,,4 This comment discusses three major issues: (i) the
likely effect on slot usage of reallocating slots from large to
small carriers; (ii) the possible effects of proposed changes to
the FAA's "use or lose" rules; and (iii) the possible effects of
proposed additional restrictions on the sale and transfer of
lottery slots.
The proposed regulations and our comment focus on an
airline's slot usage rate at an RDTA. other potential
competition policy issues, such as the effect of the proposed
regulations on airline competition in individual city pair
markets, are not directly addressed by the proposed regulations,
and are outside the scope of this comment.
Our principal findings, based largely on empirical study of
historical slot use, are set out in detail in Part IV and are
briefly summarized here. First, two specific theories of
anticompetitive effect in the current regulatory environment are
tested by comparing carriers' shares of availaple slots to the
rate at which the carriers used them. The usage rate is measured
by the number of days the slot was used during the period studied
(May and June, 1990). On balance, the data do not suggest that
carriers with relatively large shares of the slots at a RDTA were
restricting service anticompetitively, by hoarding their slots or
4 See SNPRM, supra note 2, p. 46676.
2
t· using them at a low rate compared to carriers with smaller shares. At the most concentrated HDTA, Chicago's O'Hare Airport, statistical analysis shows that the carriers with larger shares of slots tended to use their slots more, not less, intensively. At New York's LaGuardia and at Washington's National airports, there was generally no statistically significant relationship between slot share and slot usage. At New York's Kennedy Airport, the relationship was negative and statistically significant, implying that high slot share was associated with lower usage; however, this finding is probably explained by the operating problems experienced by the largest carriers at that airport. In general, the data from the period studied do not support the anticompetitive theories about carriers' patterns of slot holding and usage. Next, possible effects of proposed changes to the "use or lose" rule are assessed. The data show that most carriers, on average, used their slots at a rate that was at or above the proposed minimum criterion of 90 percent usage on weekdays. Thus the change from the present criterion of 65 percent usage over all days may affect only a few slots. But because carriers with more slots and operations may enjoy greater flexibility in allocating flights among slots and rearranging schedules to ensure their slots meet the minimum "use or lose" standard, the 90 percent weekday rule could have a greater effect on the carriers with fewer slots. 3
Finally, we discuss the effect of proposed restrictions on the sale and transfer of slots obtained in a lottery. This issue is studied by examining t~e events that followed the 1986 slot lottery. The vast majority of the slots allocated then were traded or sold by the lottery recipient within two years, that is, within the minimum mandatory holding period proposed in the SNPRM. Almost all of the carriers that received slots in the 1986 lottery have since exited the industry, via merger or bankruptcy. If this experience is a guide to the future, many slots allocated in lotteries would be traded if there were no restrictions on transfer. In the absence of market power, slot transfers likely represent an efficient reallocation of resources from lower to higher valued uses, and imposing significant delays or other restrictions on transfers may reduce welfare. In addition, although the FAA's proposed restrictions on the sale and transfer of lottery slots may effectively deter a carrier from entering lotteries solely for the purpose of obtaining slots for resale, a more efficient way to achieve this goal may be to disqualify from future lotteries a carrier who has sold its lottery slots too quickly. II. Expertise of the Staff of the Federal Trade Commission. The FTC is an independent regulatory agency responsible for maintaining competition and safeguarding the interests of 5 consumers. In response to requests by federal, state, and 5 15 U.S.C. Parts 41 - 59. 4
local government bodies, the staff of the FTC often analyzes regulatory or legislative proposals that may affect competition or the efficiency of the economy. In the course of this work, as well as in antitrust and consumer protection research, nonpublic investigations, and litigation, the staff applies established principles and recent developments in economic theory to competition and consumer protection issues. The FTC staff has a longstanding interest in aviation issues, including the allocation and transfer of slots at HDTAs. The FTC staff filed comments in previous FAA administrative proceedings on slot allocation and transfer methods. 6 In addition, the staff has participated in various administrative proceedings involving airport access,7 and has issued staff reports on slot allocation and on airline deregulation. 8 6 See comments of the staff of the FTC, in Slot Transfer Methods, FAA Docket No. 24105, August 1984, and Slot Allocation Alternative Methods, FAA Docket No. 24110, 1984. 7 See comments of the staff of the FTC in Elimination of Airport Delays, FAA Docket 24206, Discussion Authority for Agreement to Shift Schedules, Department of Transportation, Docket No. 44634, February 23, 1987, and Charges for the Use of Metropolitan Washington Airports, FAA Docket No. 25204, April 13, 1987. See also Comments of the Bureaus of Economics, Competition and Consumer Protection of the Federal Trade Commission, Massport, Program for Airport Capacity Efficiency, February 29, 1988. 8 d . See D. Koran an J. Ogur, A~rport Access Problems: Lessons Learned from Slot Regulation by the FAA, Bureau of Economics Staff Report to the Federal Trade Commission, May 1983, and J. Ogur, M. vita and C. Wagner, The Deregulated Airline Industry: A Review of the Evidence, Bureau of Economics Staff Report to the Federal Trade Commission, January 1988. 5
Our present comment addresses certain issues relating to
economic efficiency and competition, based upon price theory and
welfare economics as understood in the interpretation and
enforcement of the antitrust laws. Except as noted, it does not
discuss other policy considerations, such as distributional
issues, that may be relevant to the FAA.
III. Background of and Issues contained in the SNPRM.
A. Background of the CUrrent Slot Rule
A "slot" is a right to take off or land during a specified
period of the day at one of four HDTAs (Chicago's O'Hare,
Washington, D. C. 's National, and New York's Kennedy and
LaGuardia airports).9 Slots were defined in 1968 to deal with
excess demand and noise problems at these four airports. Before
1985, the incumbent airlines at each HDTA decided how slots would
be allocated, with any reallocation requiring their unanimous
consent. The incumbent airlines could trade slots a~ong
10
themselves, one for one. Increasing competition in the post-
11
1977 deregulated era strained this method severely; in many
cases, the allocation system broke down completely, freezing the
9 See 14 C.F.R. Part 93.123 (1990).
10 See Department of Transportation, Federal Aviation
Administration, High Density Traffic Airports: Slot Allocation
and Transfer Methods, Final Rule; Request for Comment, ("FAA 1985
Final RUle") 14 C.F.R. Parts 11 and 93, 50 Fed. Reg. 52180, 52185
(1985) .
11 See D. M. Grether, R. M. Issac, and C. R. Plott, "The
Allocation of Landing Rights by Unanimity Among Competitors," 71
Am. Ecan. Rev. 166 (1981).
6
previous allocation in place '2 and denying slots to new entrants. 13 starting in April 1986, revised FAA regulations have encouraged development of a market-based HDTA slot transfer system. 14 Initial allocation was determined by "grandfathering," giving the right to a slot to the carrier that had been operating it. '5 SUbject to exceptions and procedures described below, the revised regulations permit slot holders to sell, trade, or lease their slots, and permit slots to be held by any party. The FAA retains the right to repossess slots, which the regulations describe as operating privileges, not property . ht 16 rJ.g s. Slots may be withdrawn for such reasons as making necessary allocations for international flights and implementing 12 See FAA Final Rule, supra note 10, p. 52180. See also Comments of the Department of Justice, Slot Allocation: Alternative Methods and Slot Transfer Methods, Notice of Proposed Rulemaking, FAA Docket No. 24110 and 24105, August 6, 1984, pp. 11-12. 13 After the 1981 air controllers strike, slot restrictions were imposed for a limited time on an additional 18 airports. For a six-week period in 1982, the FAA allowed slots to be traded among airlines at these 22 airports. For a detailed discussion of the effects of this six-week market, see Koran and Ogur, supra note 8. Noting that over 190 slots were sold during this period despite the simultaneous existence of slot trading programs, they suggest that the large number of sales reflect the markets' ability to facilitate the movement of slots from low to high valued uses. 14 See FAA 1985 Final Rule, supra note 10, p. 52180, and 14 C.F.R. Parts 93.211--93.229. The rules were suspended for Newark, which is also an HDTA. 15 See 14 C.F.R. Part 93.215, and DOJ Comments, supra note 12, pp . 12 -14 . 16 See 14 C.F.R. Part 93.223(a). 7
17 competition goals. A priority system for these withdrawals is . hIt' 18 set out ~n t e regu a ~ons. Slots will be recalled for reallocation if used less than 65 percent of the time. 19 Lotteries distribute slots that are not currently allocated, such as those recalled for nonuse or voluntarily returned to the FAA. 20 Both incumbent and new entrant carriers may enter these lotteries, in which the order of choice is determined by random . t" d t 21 draw~ng, but new entran carr~ers are g~ven some a van age. Slots obtained in a lottery are sUbject to a sixty-day "must use" period before they can fully enter the slot market. 22 17 See 14 C.F.R. Part 93.223(a). For example, in March of 1986, the FAA withdrew 5 percent of the slots from incumbent carriers for distribution via lottery to new entrant carriers. This lottery is discussed in more detail in Section IV.C, infra. 18 This is done by assigning a withdrawal priority number to each slot. At the time of the initial allocation in 1985, each slot was randomly assigned a recall priority number, with the lowest numbered slots being the first to be withdrawn. Slots owned by carriers with eight or fewer slots are exempt from withdrawal, except those used less that 65 percent of the time over a two-month period. See 14 C.F.R. Part 93.223(f). 19 See 14 C. F. R. Part 93.227. ~ See 14 C.F.R. Part 93.225. 21 During the first lottery sequence, 25 percent of the slots (but no less than two) are set aside for new entrant carriers. See 14 C.F.R. Part 93.225(h). 22 A slot obtained by lottery may be traded one for one for any other slot at the same HDTA during the "must use" period; however, it cannot otherwise be sold or traded until it has been used 65 percent of the time during a sixty-day period. See 14 C.F.R. Part 93.221(a) (5). 8
B. Background of the SNPRM and Proposed Changes to the FAA Slot Rule This proceeding dates from 1988, when the FAA denied a petition to reallocate incumbent carriers' slots to new entrants and smaller carriers, but at the same time called for comment on the relationship between the slot allocation regulations and • • • 23 h several competltlon and market entry lssues. T e SNPRM . h d' 24 contlnues t e 1988 procee lng. The general intent of the amendments proposed in the SNPRM is "to make slots available to new entrants through encouraging the sale of marginally used slots by incumbent carriers, in order to promote competition in the airline industry to the extent possible without substantial disruption of existing air service. 11 25 The FAA believes that new entry and growth of smaller carriers can foster competition and that increased competition can benefit the public through reduced fares and greater services. But because competition among carriers at each HDTA is "intense", with none dominated by a single carrier, and each high density market area is also served by alternate 23 The call for public comment was required by P. L. 100- 457. See High Density Traffic Airports: Slot Allocation and Transfer Methods -- Notice of Proposed Rulemaking, 53 Fed. Reg. 51628 (December 22, 1988). 24 Like the 1988 proceeding, it also implements a statutory call for a rulemaking proceeding. Rather than specify particular issues, the law simply calls for consideration of "more efficient methods of allocating existing capacity at high density traffic airports in order to provide improved opportunities for operations by new entrant air carriers. 1I P. L. 101-508, Title IX, Part 9126. See SNPRM, supra note 2, p. 46675. "5 ~ See SNPRM, supra note 2, p. 46676. 9
airports without slot restrictions, the FAA believes it is uncertain whether new entry at the HDTAs would actually enhance competition in those markets. 26 The SNPRM's principal proposals would affect the "use or lose" threshold and the slot lotteries. The proposed new "use or lose" rule would require carriers to use their slots 90 percent of weekdays; slots used less frequently would be withdrawn. By contrast, the present rule requires carriers to use their slots 65 percent of all days, including both weekdays and weekends. In each case, the percentage usage is measured in two month reporting periods. The FAA suggests that this change will increase the number of slots available on the market and thus 27 enhance competitive opportunities for new entrants. If any 26 See SNPRM, supra note 2, p. 46677, 46679-80. Theoretical analysis suggests circumstances in which promoting entry artificially could reduce welfare. For example, welfare could fall if a slot reallocation resulted in an incumbent abandoning the only service to a destination and a new entrant adding marginal service to a different destination that was already served by many competitors. Borenstein suggests that such an outcome is likely, as the profits from "stealing business" from existing carriers on a high-traffic route likely outweigh the profits from being one of the few carriers serving a low-traffic route. See Borenstein, "On the Efficiency of Competitive Markets for Operating Licenses," 103 Q. J. Econ. 357 (1988). For a general discussion of the problems faced by regulators in making such decisions, see Koran and Ogur, supra note 8, Douglas and Miller, Economic Regulation of Domestic Air Transportation: Theory and Policy (1974) and Demsetz, "Information and Efficiency: Another Viewpoint," 12 J. L. & Econ. 1 (1969). 27 See SNPRM, supra note 2, at 46678. The current 65 percent "use or lose" threshold will continue to apply to slots held by carriers less than seven days per week (e.g., a slot held by one carrier on Tuesday and Thursday, and held by a second carrier the other five days of the week) . 10
slots are actually withdrawn for nonuse, they would be allocated to carriers qualifying as new entrants or limited incumbents. 28 The proposed new slot lottery rules would further restrict trading in slots obtained through lottery and would also further restrict lottery participation. The "must use" retention period would be extended, to require that a slot obtained in a lottery be used for two years before it could enter the market without restrictions. During this two year period, lottery slots could only be traded for other lottery slots at the same airport, and could be sold or leased only to a new entrant or limited incumbent. Now, a slot obtained in a lottery can be sold or traded without restrictions after a "must use" period of only sixty days, and during that period may be traded one-for-one for any other slot at the same airport. The FAA states that the proposed restrictions on transfer are designed to encourage continued use of slots by new entrants and limited incumbents. The FAA believes that result would foster competition, by discouraging lottery participation by parties interested more in liquidation than operation, and by reducing the smaller carriers' ability to choose between using their operating authority or selling it. 29 28 The FAA also proposes to require carriers to report slot usage on international flights, to reduce carriers' ability to count international flights toward the "use or lose" requirement, to relax the "use or lose" provisions for a period following bankruptcy or mergers, and to remove obsolete penalty provisions. 29 See SNPRM, supra note 2, p. 46676. 11
In addition, the proposed rules would change the lottery procedures to limit the set of carriers able to participate. First, the definition of "new entrant carriers" would be narrowed, by applying the term only to a carrier holding both appropriate Department of Transportation (DOT) economic authority . t' f' t 30 and an FAA part 121 or part 135 operatlng cer 1 lca e. Now, a carrier that has made substantial progress toward obtaining an FAA operating certificate is considered a "new entrant" and may t · . t 31 par lClpa e. Second, the role of large carriers in the lotteries would be reduced, by assigning primary participation to new entrants and a newly-defined class of "limited incumbents." A limited incumbent at an HDTA is defined as an air carrier or commuter operator that holds, or has held, fewer than twelve 32 slots. The basic lottery procedure, which would remain unchanged, calls for each participant to choose two slots in each round until the available slots are exhausted. New entrants are given an advantage by having 25 percent of the slots during the 30 The FAA states that experience with carriers not currently holding an FAA certificate "actually beginning and continuing operations after selecting slots has not been positive, and the Department believes that participation in current lotteries should be limited to those carriers most likely to be able to make efficient use of slots." See SNPRM, supra note 2, at 46676. 31 14 C.F.R. Part 93.225(g). 32 For example, a carrier that once held 20 slots but now holds four, after selling 16 to other carriers, would not qualify as a limited incumbent. The 12-slot threshold is consistent with the definition of "new entrant" in the statute that calls for the rulemaking proceeding. P. L. 101-506, Title IX, Part 9126; see SNPRM, supra note 2, pp. 46675, 46676. 12
first round reserved for them. The proposed rules would limit lottery participation to new entrants and limited incumbents (who could choose slots reserved for new entrants if they went unclaimed) until after they had finished their selections or had chosen enough slots to exceed the 12-slot "limited incumbent" definition. Only then could other incumbent carriers participate~ moreover, any slots they chose could only be held 33 until the next lottery. The FAA says this change is intended to "permit new entrant and limited incumbent carriers to acquire any available slots without competing with carriers already holding a substantial number of slots at the airport. ,,34 IV. An Analysis of the Proposed Changes in the Slot Rule This section assesses the possible consequences of the proposed changes by reviewing empirically the experience under the present regUlations. Part A, by examining the relationship between a carrier's share of the slots at an HDTA and its use of those slots, explores whether a reallocation of slots from carriers with a large number of slots at an HDTA to carriers with few slots appears likely to increase economic welfare. Data obtained from the FAA, detailing both how many slots carriers hold and the number of days they use them, does not suggest that 33 Current rules make no distinction between large and small incumbents during lotteries. Limited incumbents and certified new entrants would also be given preferential treatment under slot withdrawal rules. 34 See SNPRM, supra note 2, p. 46676. 13
carriers with relatively large shares of slots at an HDTA hoarded their slots or used them at comparatively low rates. This result is inconsistent with anticompetitive theories about carriers' patterns of slot holding and usage. Part B examines the effect of the proposed changes to the "use or lose" rules. Historical use data suggest that a 90 percent weekday rule may affect carriers with a small number of slots at an HDTA more than carriers with a large number of slots. Finally, Part C discusses the proposed additional limitations on trade in lottery slots. The vast majority of slots allocated to new entrants in the 1986 lotteries changed ownership within a short time period. If this experience is a guide to the future, many slots allocated in lotteries would be traded, if there were no restrictions on transfer. In the absence of market power, slot transfers likely represent an efficient reallocation of resources from lower to higher valued uses, and imposing significant delays or other restrictions on transfers may reduce welfare. A. The Relationship Between Slot Share and Slot Use (i) Theory and Hypotheses Commentators have pointed to higher concentration and prices for airline travel at HDTAs relative to non-HDTA airports as evidence of a lack of competition at HDTAs. For example, estimates of the effect of limited slot availability on airfares suggest that slot restrictions increase average or median fares eleven percent on short haul routes, and four percent for all 14
routes to and from the HDTAs. 35 There also has been a focus on
the levels of concentration at the HDTAs. 36 Based on slot
holdings, each of the HDTAs considered alone is either moderately
or highly concentrated, according to the standards of the
. . d l' 37 Department of Justlce Merger GUl e lnes. Table 1 lists the
unrestricted air carrier slot holdings and the Herfindahl-
Hirschmann Index (HHI) of concentration at the four HDTA's as of
~ See "Airline Competition: Effects of Airline Market
Concentration and Barriers to Entry on Airfares," GAO Report to
Congressional Requesters, RCED-91-101. Borenstein found a
similar 3-5 percent increase in yields at Chicago's O'Hare
airport. His study focused on hubs, and therefore did not
examine the other three HDTAs. See S. Borenstein, "Hubs and High
Fares: Dominance and Market Power in the u.s. Airline Industry,"
20 Rand J. Econ. 344 (1989).
36 See, e.g., U.S. General Accounting Office, "Airline
Competition: Industry operating and Marketing Practices Limit
Market Entry," U.S. General Accounting Office Report to
Congressional Requesters, RCED 90-147 (1990), pp. 21-30 ("GAO
Entry Report"). See also Testimony and Exhibits of America West
Airlines, Inc., DOT, USAir-Piedmont Acquisition case, Exhibit AWA
T-1,2, Docket 44719, September 21, 1987. Some critics have also
addressed distributional issues, observing that incumbent
carriers received a valuable (and salable) right for free. See,
e.g., FAA 1985 Final Rule, supra note 10, p. 52184. Such
distributional issues will not be addressed in this paper. While
it might be better for the Federal treasury that airlines pay for
slots, the Coase Theorem implies that in the absence of
transactions costs (which seem low), the initial allocation of
rights does not affect the allocative efficiency of the final
market allocation. See Ronald Coase, "The Problem of Social
Cost," 3 J. Law & Econ. 1 (1960).
37 See U.S. Department of Justice, "Merger Guidelines,"
Antitrust & Trade Regulation Report, Special Supplement (June
14, 1984). Under the DOJ Guidelines approach, markets are
"moderately concentrated" if their HHIs are between 1000 and 1800
and "highly concentrated" if their HHls are greater than 1800.
Our analysis focuses on overall slot concentration at an HDTA,
and not slot concentration during certain time periods. This
focus is appropriate as long as flights in different slot periods
compete significantly.
15
June 30, 1990. New York's LaGuardia, at 1241, and Washington's National, at 1282, were "moderately concentrated." New York's Kennedy, at 1816, and chicago's O'Hare, at 3126, were "highly concentrated." Concentrated markets for airport slots could create conditions for the exercise of market power or other t · t' t' t' 38 an lcompe 1 lve ac lon. 38 Concern about excess concentration in slots at Washington National and LaGuardia was an important part of the ruling of a Department of Transportation administrative law judge in 1987 that would have prevented USAir from merging with Piedmont. See Recommended Decision of Administrative Law Judge Ronnie A. Yoder, USAir-Piedmont Acquisition Case, DOT Docket 44719, September 21, 1987, at 90-102. This decision was later overturned by the Secretary of Transportation. 16
TABLE 1 DOMESTIC SLOT HOLDINGS AND CONCENTRATION SLOT HOLDER (6-30-90) LAGUARDIA NATIONAL UNRESTRICTED SLOTS (LGA) (DCA) NUMBER SHARE NUMBER SHARE OF SLOTS OF SLOTS UNITED (UAL) 45 7.28 28 4.87 AMERICAN (AAL) 57 9.22 35 6.09 USAIR (USA) 39 151 24.43 148 25.74 EASTERN (EAL) 72 11.65 82 14.26 NORTHWEST (NWA) 45 7.28 53 9.22 DELTA (DAL) (BNK) 40 56 9.06 47 8.17 SHAWMUT BANK 24 3.88 44 7.65 PAN AM (PAA) 64 10.36 38 6.61 CONTINENTAL (COA) 29 4.69 48 8.35 TRUMP (TPS) 57 9.22 28 4.87 AIR WISCONSIN (AWl) 0 0 0 0 MIDWAY (MID) 14 2.27 16 2.78 AMERICA WEST (AWE) 0 0 4 0.7 MIDWEST EXPR. (MEP) 4 0.65 4 0.7 AM. TRANS AIR (AMT) 0 0 0 0 FLYING TIGER (FTL) 0 0 0 0 COMMAND AIR (CMD) 0 0 O' 0 UNITED PARCEL (UPS) 0 0 0 0 MGM AIR (MGM) 0 0 0 0 FED. AV. ADM. (FAA) 49 0 N 667 575 HHI 1241 1282 39 Although Eastern and Continental were merged in 1986, control of Eastern was taken away from Texas Air (Continental's parent entity) by a Federal bankruptcy court on April 18, 1990. We therefore treat Eastern and Continental as separate carriers. 40 This bank now holds slots formerly held by TWA. See Table 4, showing TWA as the only significant slot user that did not hold any itself. 17
TABLE 1 (CONTINUED) KENNEDY (JFK) O'HARE (ORO) ALL HDTAs UAL AAL USA EAL NWA DAL BNK PAA COA TPS AWl MID AWE MEP AMT FTL CMD UPS MGM FAA TOTALS HHI #SLOTS 6 21 20 17 12 9 60 42 o o o o 1 o o 1 2 1 1 9 202 SHARE 3.11 10.88 10.36 8.81 6.22 4.66 31. 09 21. 76 o o o o 0.52 o o 0.52 1. 04 0.52 0.52 1816 #SLOTS 658 491 45 21 71 64 36 4 36 o 44 o 7 2 5 3 o o o 1 1488 SHARE 44.25 33.02 3.03 1. 41 4.77 4.37 2.42 0.27 2.42 o 2.96 o 0.47 0.13 0.34 0.2 o o o 3122 #SLOTS 737 604 364 192 181 167 164 148 113 85 44 30 12 10 5 4 2 1 1 59 2932 SHARE 25.14 20.60 12.41 6.55 6.17 6.00 5.59 5.05 3.85 2.90 1. 50 1. 02 0.41 0.34 0.17 0.14 0.07 0.03 0.03 2.01 Source: Tabulated from FAA Slot Data, May-June 1990. At the outset, it should be noted that neither higher concentration nor higher prices at HDTAs necessarily implies anticompetitive effects or causes. First, economic theory predicts that in competitive markets, fares at slot constrained airports will be higher, ceteris paribus, than in markets where there are no slot restrictions. Equilibrium competitive prices 18
· d 't t 41 for air travel must reflect slots' assoc1ate scarC1 y ren s. Indeed, if travel prices were not higher at slot constrained airports, this would imply that landing rights at HDTAs were not a valuable resource, so slots would not trade at positive prices. 42 Higher fares at HDTAs may merely reflect scarcity rents, rather than monopoly rents. Second, there may be a procompetitive reason for seemingly high concentration in airport slot markets: in competitive markets, more efficient firms, which offer their customers better combinations of price and service, generally will gain market share at the expense of the less efficient. 43 That is, an industry might be relatively concentrated because its larger firms are more efficient, not because they are less competitive. Thus, it is possible that only a few airlines hold a majority of the slots as a consequence of those carriers' greater efficiency. 41 Scarcity rents arise when competition for a resource in limited supply causes the price of that resource to rise above the marginal cost of providing the last unit. In the HDTA setting, scarcity rents would reflect the FAA's decision to limit the number of slots. In contrast, monopoly rents result from anticompetitive output restrictions. Here, monopoly rents would be additional rents due to slot holders' decisions to exercise market power by restricting slot use in a way that prevents existing slots from moving to their highest valued use. 42 In a competitive market, a slot's value would equal the net present value of its associated scarcity rents. 43 For example, the apparent high concentration at O'Hare may be the result of its use as a major hub by united and American. Use of these hub and spoke operations can increase efficiency. See, S. Morrison and C. Winston, The Economic Effects of Airline Deregulation, Washington, D. C.: Brookings (1986). For a general statement of this proposition, see e.g., H. Demsetz, The Market Concentration Doctrine (1973). 19
Finally, the concentration figures for HDTAs may overstate the potential for anticompetitive activity because they do not describe antitrust markets. In antitrust analysis, before concentration is measured markets must be defined. A concentration number for something that is not an antitrust market, that does not include relevant product and geographic competitive factors, will be misleading. Each of the HDTAs is near one or more regional airports without slot constraints (Midway at Chicago, Newark at New York, and Dulles and Baltimore- Washington International at Washington) .44 If flights from these airports are in the same antitrust market as flights from 44 See SNPRM, supra note 2, p. 46677 and FAA 1985 Final Rule, supra note 10, p. 52166. 20
the HDTAs,45 their competitive presence can constrain the prices
of flights from the HDTAs. 46
45 An illustration of how the HDTAs and their regional
competitors could be considered to be in the same antitrust
market is provided by the positions the Department of Justice
("DOJ") has taken in the Eastern Airlines bankruptcy proceedings.
The DOJ opposed United's proposed purchase of Eastern's National
Airport slots and facilities, on the grounds that it would
"lessen competition" between washington and other cities because
the airline is already the dominant carrier at Dulles
International Airport and the acquisition would have also given
it control of 20 percent of the slots at National. See "Justice
Dept. to Dispute Deal for Eastern Slots," Washington Post,
(February 15, 1991) p. B2i 60 Antitrust & Trade Regulation Report
152 (January 31, 1991), 186 (February 7, 1991), and 272 (February
21, 1991). But the DOJ did not object to Northwest Airlines'
purchase of those assets even though Northwest had more slots
than united at National prior to the bankruptcy sale. Similarly,
despite the fact that United was the largest carrier at O'Hare,
the DOJ did not object to United's purchase of Eastern's O'Hare
slots, a position consistent with considering Midway Airport
flights to be in the same antitrust market as O'Hare flights.
46 The fact that slots are scarce at one airport, and thus
sell for positive prices, while they are not scarce at another
airport in the same region is not alone an indication of whether
or not the two airports are in the same antitrust market. Two
regional airports can be in the same antitrust market even when
. prices for slots vary between the airports. However, an HDTA
alone could be a relevant antitrust market, and anticompetitive
price increases could occur, even when there is another non-slot
constrained airport in the region that is a substitute at current
prices. In antitrust law, the ambiguity of using price alone to
determine antitrust markets is sometimes called the "cellophane
trap" because of its association with the "cellophane" case, u.s.
v. E.I. du Pont de Nemours & Co., 351 U.S. 377, 404 (1956). The
term refers to a failure to ta}:e into account whether an observed
price is supracompetitive rather than competitive when
determining whether certain products or services are in the same
antitrust market. For a further discussion, see Richard A.
Posner, Antitrust Law: An Economic Perspective (1976), pp. 127-
129. In addition, Borenstein, supra note 35, p. 348, notes that
local airports may constitute separate antitrust markets if
airlines can price discriminate among passengers at different
airports. (See also DOJ Guidelines, paragraph 2.13, supra note
37.)
21
, . In sum, the possibility of other reasons for both high prices and high concentration at HDTAs limits the usefulness of these variables alone to inform public policy decisions. To avoid the ambiguities of tests that emphasize price effects, our empirical work uses a measure of output, instead of price, in an attempt to differentiate between procompetitive and 47 anticompetitive uses of slots. Focus on output is also consistent with the current and proposed FAA "use or lose" rules and will allow the likely effect of the proposed rules to be examined directlY.~ 47 Leading antitrust scholars have suggested using changes in "physical" output to distinguish between procompetitive and anticompetitive practices. See, e.g., Richard A. Posner, "The Rule of Reason and the Economic Approach: Reflections on the Sylvania Decision," 45 Univ. Chicago L. Rev. 1 (1977), and Frank H. Easterbrook, "The Limits of Antitrust," 63 Texas L. Rev. 1 (1984) . 48 Because of data and time limitations, our empirical work does not take into account output choices other than the number of days a slot is used. A hypothetical monopolist also could reduce output by flying smaller planes. Since slot use and aircraft size are complements, a hypothetical monopolist would be expected to address both. Output tests that do not take into account all output choices by a hypothetical monopolist also can be ambiguous. When the unobserved and observed measures of quantity (e.g., days the slot was used and seats per plane) are highly complementary, a hypothetical monopolist may increase price, and both the observed and unobserved quantity above the efficient levels provided in a competitive market. See Keith B. Leffler, "Ambiguous Changes in Product Quality," 72 Am. Econ. Rev. 956 (1982). However, the use of smaller planes does not necessarily imply anticompetitive behavior, even if it means higher prices. Aircraft operating characteristics may make it more efficient to use smaller planes on shorter routes. In an efficient slot market, passengers on shorter routes with smaller planes will pay a premium if this represents the higher per- passenger scarcity rent associated with that slot. 22
The hypotheses to be tested in this comment all derive from theoretical predictions about the output-setting choices made by relatively large firms compared to those of smaller firms. The \ first hypothesis to be tested rests on the standard market power scenario, in which dominant firms have a larger incentive to reduce output than the fringe firms. standard theory predicts that the fringe will increase output in response to a contraction of output by the dominant firm. 49 It has been suggested that the carriers with a large number of slots at a HDTA will attempt to restrict output by reducing the number of days on which a slot is used. so In the context of an airport slot market, this implies that larger carriers will use their slots less intensively than the smaller "fringe" carriers. Conversely, if slots are being used efficiently, the larger carriers' slot usage would be as great or greater than slot usage by smaller carriers. The testable implication of such a procompetitive theory is that slot usage will be positively related to the market share of the carrier using the slot. In contrast, the anticompetitive theory implies the reverse, i.e., that the dominant or colluding 49 See A. A. Alchian and W. R. Allen, Exchange and Production: Compe~ition, Coordination and Control, 3rd. ed., (1983), pp. 266-268. 50 For example, airlines holding at least three slots in a period can distribute two flights over three slots, thus using three slots 66 percent of the time (above the minimum 65 percent required by the FAA), instead of using two slots 100 percent of the time. See SNPRM, supra note 2, p. 46678. 23
carriers will use their slots less than the competitive f . 51 rlnge. Another form of the anticompetitive hypothesis is that dominant carriers would be unwilling to sell slots to potential new entrants at competitive rates,52 but instead would lease slots to selected other carriers, their purpose being to deter entry into HDTAs by carriers likely to increase overall slot 53 use. The testable implications of this hypothesis are that 51 This analysis assumes the number of slots allocated per hour is set so that full utilization of all the slots in a period would not induce large congestion effects. If congestion effects are induced by full utilization, a reduction in slot use may be efficient. That is, a dominant carrier may wish to purchase and not use a slot in order to reduce the costs associated with excess congestion and delays imposed on the large number of their flights operated in that slot period. Carriers with a small number of slots in that slot period would be affected less and would have a smaller incentive to internalize these costs. Concerns about excess congestion have led the FAA to reject Cnlls to expand slot capacity at the HDTAs. See SNPRM, supra note 2, p. 46677. 52 See Recommended Decision of Administrative Law JUdge Ronnie A. Yoder, USAir-Piedmont Acquisition Case, supra note 38 at 93, and Testimony and Exhibits of America West, supra note 36, pp 8-11. However, one cannot distinguish this anticompetitive hypothesis from the situation where a potential entrant makes a false claim that it cannot obtain slots from private parties, pressed in order to gain them from the government for free. The current regulatory regime, which includes a slot market, cannot inhibit entry more than the previous regime of nontransferable historical allocations, in which entrants were unable to obtain slots under any circumstances. See Koran and Ogur, supra note 8. For a similar argument, see John R. Lott, Jr., "Licensing and Nontransferable Rents," 77 Am. Econ. Rev. 453 (1987). For a general discussion of entry, see Demsetz, "Barriers to Entry," 72 Am. Econ. Rev. 47 (1982). 53 See, e.g., GAO, supra note 36. In this scenario, airlines choose their competitors by leasing slots to them. In this way, the dominant firm(s) can ensure that less efficient firms will be competing with them for the same passengers, while (continued ... ) 24
the dominant air carriers will be net lessors of slots and that these leased slots will be used relatively less intensively than slots held and operated by the dominant carriers. 54 The two principal hypotheses tested in the comment deal solely with an airline's slot usage rate. Broader concerns, such as the likely effect of the proposed slot regulations on airline competition in individual city pair markets, are not addressed in this comment. Furthermore, given that the slot regulations do not directly address such concerns (e.g., through explicit route regulation), accurate prediction of the effect of the proposed regulations may be difficult. 55 This is a cross-sectional study, examining data from a two-month sample period. If the 53 ( ••• continued) simultaneously deterring the entry of more efficient firms. For a theoretical discussion of this effect, see, e.g., K. E. Rockett, "Choosing the Competition and Patent Licensing," 21 Rand J. of Econ. 161 (1990), which examines a single incumbent (with a patent) facing a limited number of potential entrants. Weakening either the single incumbent or limited entrant assumptions weakens this theoretical result. Thus, this theoretical result would be less likely to apply to airline markets where there is more than one incumbent or more than one equally efficient potential entrant. For a general discussion of the fragility of the entry deterrence models, see D. Malueg and M. Schwartz, "Preemptive Investment, Toehold Entry, and the MimiCking Principle," 22 Rand J. Econ. 1 (1991), and M. Waldman, "The Role of MUltiple Potential Entrants/Sequential Entry in Noncooperative Entry Deterrence," 22 Rand J. Econ. 446 (1991). 54 See, e.g., GAO Entry Report, supra note 36, Chapter 2. 55 The FAA notes that "there is no ex ante way to estimate the system impact or even whether it is positive or negative." See SNPRM, supra note 2, p. 46680. To the extent anticompetitive problems exist, such problems can be addressed through the enforcement of the antitrust laws. See, e.g., the discussion of the Eastern bankruptcy proceedings in note 45, supra. See, also, Borenstein, supra note 26. 25
sample is not representative, perhaps due to recent changes in the industry or at the specific airports considered, then the results ~ay not reflect current conditions in the market for slots. (ii) Empirical Results The data were obtained from the FAA's Office of Slot Administration. The FAA collects daily ownership and usage data for monitoring airline compliance with ownership and usage rules. The slot office keeps this data for the two previous years, thus limiting the time periods available for examination. Data from the period May-June 1990 were chosen because it lies between the extremes of airlines· seasonal demands. May and June generally have lower demand for air travel than July and August and higher demand than January and February. In addition, May-June 1990 was after the time period when Eastern Airlines was unable to operate a large proportion of its planes due to the strike by its employees in March 1989. May-June 1990 was also prior to the August 1990 Iraqi invasion of Kuwait and the resulting decline in demand for air services. This time period consists of 44 weekdays and 61 days in total. The dataset contains 2242 total air carrier slots at O'Hare, 1323 at LaGuardia, 956 at National, and 580 at Kennedy. statistical analysis is based on the set of slots meeting three criteria: (i) the slot was held and operated by a domestic airline on a domestic route; (ii) use of the slot was not restricted by the FAA; and (iii) the slot was not exempt from the 26
56 FAA's "use or lose" rules. This set contains 1488 slots at O'Hare, 667 at LaGuardia, 575 at National and 202 at Kennedy.57 (a) Average Slot Use Table 2 shows the average slot use by carrier at the four HDTAs measured by the proportion of total days used (TPCT) and weekdays used (WPCT). Examining the data, there is .no obvious pattern supporting the anticompetitive hypothesis. That is, the data do not suggest that the intensity of use of a slot was negatively related to the share of slots held by the carrier. At the most concentrated HDTA, O'Hare, almost all of the slot holders used their slots on average at least 90 percent of the time. The only exceptions were American Trans Air, based on all days, and Midwest Express, based on weekdays. Examining the numbers for the other HDTAs reveal similar high rates of use. At National, only Eastern, at 86 percent, was below a 90 percent average use rate, based on weekdays; at Laguardia, only Northwest, at 89 percent, was below 90 percent. Finally, at Kennedy the data show four carriers were below 90 percent, but 56 International routes, and thus foreign airlines, are subject to different slot allocation rules. 57 At O'Hare, 220 slots were not allocated or otherwise restricted by the FAA, 88 were international slots, 7 were domestic slots being used with international flights, and 4 were international slots operated domestically. At National, 147 slots were not allocated or otherwise restricted by the FAA. At LaGuardia, 298 slots were not allocated or restricted by the FAA, 36 were international slots, 23 were domestic slots operated internationally, and 15 were international slots operated domestically. Finally, at Kennedy, 219 slots were not allocated or restricted by the FAA, 74 were held and operated internationally, and 11 were domestic slots operated internationally. 27
for three of these the data are probably inaccurate. For Pan Am, the use rate was 88 percent for both all days and weekdays, but for Eastern, Flying Tiger, and MGM Grand, the rates were so extremely low (in two cases, zero) as to suggest that data are missing. 58 Thus, except for a few cases, all slot holders at the four HDTAs, on average, used their slots at a rate more than twenty-five percentage points above the current FAA "use or lose" threshold. 58 Of the slots held by Eastern, only those leased to other carriers were reported as being used during the period studied. But no slots were withdrawn from carriers for "use or lose" rule violations during May-June 1990. 28
TABLE 2 AVERAGE SLOT USE Average use rates below 90 percent in bold. SLOT NUMBER TPCT WPCT HOLDER OF SLOTS CHICAGO - O'HARE AIRPORT UNITED 658 0.94 0.97 AMERICAN 491 0.93 0.93 NORTHWEST 71 0.90 0.94 DELTA 64 0.94 0.96 USAIR 45 0.90 0.95 AIR WISCONSIN 44 0.92 0.97 CONTINENTAL 36 0.92 0.93 SHAWMUT BANK 36 0.91 0.93 EASTERN 21 0.97 0.97 AMERICA WEST 7 1. 00 1. 00 AMERICAN TRANS 5 0.81 0.90 PAN AM 4 0.94 0.95 FLYING TIGER 3 0.92 0.91 MIDWEST EXPRESS 2 0.91 0.88 FAA 1 0.89 0.89 WASHINGTON, D.C. - NATIONAL AIRPORT USAIR EASTERN NORTHWEST CONTINENTAL DELTA SHAWMUT BANK PAN AM AMERICAN UNITED TRUMP MIDWAY MIDWEST EXPRESS AMERICA WEST 148 82 53 48 47 44 38 35 28 28 16 4 4 0.93 0.86 0.89 0.92 0.94 0.89 0.93 0.95 0.90 0.94 0.91 0.88 1. 00 0.97 0.86 0.91 0.96 0.96 0.91 0.95 0.96 0.95 0.93 0.97 0.99 1. 00 TPCT = number of days slot was used + 61. WPCT = number of weekdays slot was used + 44. 29
TABLE 2 (CONTINUED) SLOT HOLDER NUMBER OF SLOTS TPCT WPCT NEW YORK- LAGUARDIA AIRPORT USAIR 151 0.91 0.95 EASTERN 72 0.93 0.94 PAN AM 64 0.88 0.91 TRUMP 57 0.93 0.93 AMERICAN 57 0.90 0.92 DELTA 56 0.96 0.97 UNITED 45 0.90 0.94 NORTHWEST 45 0.87 0.89 CONTINENTAL 29 0.90 0.94 SHAWMUT BANK 24 0.91 0.91 MIDWAY 14 0.89 0.94 MIDWEST EXPRESS 4 0.84 0.98 FAA 49 0.89 0.90 NEW YORK - KENNEDY AIRPORT SHAWMUT BANK 60 0.91 0.91 PAN AM 42 0.88 0.88 AMERICAN 21 0.90 0.90 USAIR 20 0.91 0.92 EASTERN 17 0.16 0.17 NORTHWEST 12 0.94 0.94 DELTA 9 0.95 0.95 UNITED 6 0.97 0.98 COMMAND 2 0.99 0.99 FLYING TIGER 1 0 0 MGM GRAND AIR 1 0 0 UPS 1 0.90 0.91 AMERICA WEST 1 1. 00 1. 00 FAA 9 0.90 0.89 Source: Tabulated from FAA Slot Data, May-June 1990. (b) Leased Versus Owned Slots Tables 3A and 3B compare leased slots with owned and operated slots. The left side of Table 3A shows each slot holder's rate of use of its owned and operated (0&0) slots, and the right side of Table 3A shows the rate of use of its slots 30
I , that were used by others. Table 3B lists the differences in the rate of use between each slot holder's 0&0 slots and its slots leased to others, along with two measures of the st~tistical significance of this difference. The anticompetitive leasing hypothesis implies that this difference should be positive, i.e., that the slot holder's 0&0 slots are used at a higher rate than the slots leased to others. 59 At O'Hare, the three largest airlines (in terms of slot market share) leased to carriers who used the slots more than did the lessors, based on weekday usage, with the difference being statistically significant for American. Based on all-days usage, the slots united leased to others were used slightly more than united's 0&0 slots and the slots American leased to others were used slightly less than its 0&0 slots; neither difference was significant, however. Northwest's all-days difference was negative and significant. Examining the other HDTAs in the Table, in only one case (USAir at National) was the difference between 0&0 and leased slot usage for one of the top three slot holders positive and significant. contrary to the prediction of the anticompetitive leasing theory, it was the carriers with smaller slot shares, rather than the larger ones, that were more likely to use their 0&0 slots more intensively. On balance, these results do not support the hypothesis that carriers with 59 No inference can be drawn from observing equal rates of use between 0&0 and leased slots because such an observation is consistent with both the competitive and anticompetitive hypotheses. 31
large slot shares cause anticompetitive harm through the way they lease slots to other carriers. 32
TABLE 3A RATES OF USE IN OWNED AND OPERATED VERSUS LEASED SLOTS positive and statistically significant differences in bold OWNED AND OPERATED LEASED N TPCT WPCT N TPCT WPCT CHICAGO - O'HARE UNITED 639 0.940 0.966 9 0.942 0.970 AMERICAN 411 0.928 0.931 44 0.928 0.960 NORTHWEST 33 0.858 0.929 33 0.942 0.946 DELTA 39 0.942 0.970 21 0.929 0.935 USAIR 36 0.902 0.954 6 0.902 0.909 CONTINENTAL 17 0.934 0.958 18 0.899 0.909 EASTERN 12 0.993 0.998 9 0.934 0.932 WASHINGTON, D.C. - NATIONAL USAIR 137 0.934 0.972 6 0.879 0.897 EASTERN 75 0.863 0.868 1 0.934 0.932 CONTINENTAL 39 0.923 0.960 9 0.903 0.960 NORTHWEST 29 0.882 0.902 19 0.923 0.943 DELTA 41 0.938 0.954 4 0.946 0.983 PAN AM 34 0.928 0.948 2 0.877 0.955 AMERICAN 21 0.970 0.970 9 0.949 0.965 MIDWAY 14 0.913 0.979 1 0.820 0.886 NEW YORK - LAGUARDIA USAIR 129 0.904 0.954 12 0.941 0.939 EASTERN 63 0.934 0.934 6 0.912 0.977 PAN AM 51 0.872 0.897 10 0.916 0.959 DELTA 44 0.980 0.984 10 0.888 0.925 AMERICAN 34 0.874 0.910 17 0.938 0.945 NORTHWEST 15 0.885 0.891 20 0.893 0.920 UNITED 31 0.919 0.957 2 0.918 0.943 CONTINENTAL 11 0.918 0.955 13 0.877 0.927 MIDWAY 13 0.905 0.965 1 0.672 0.681 NEW YORK - KENNEDY PAN AM 29 0.865 0.858 7 0.925 0.938 USAIR 14 0.901 0.899 4 0.943 0.955 EASTERN 14 0.000 0.000 3 0.918 0.955 AMERICAN 11 0.933 0.934 1 0.705 0.681 DELTA 4 1.000 1.000 5 0.918 0.918 UNITED 5 0.987 0.982 1 0.869 1.000 33
TABLE 3B DIFFERENCE AND TESTS OF STATISTICAL SIGNIFICANCE ALL DAYS DIFF = TPCTQ&o - TPCTLEASED DIFF WEEKDAYS DIFF = WPCTQ&o - WPCTLEASED DIFF CHICAGO - O'HARE UNITED AMERICAN NORTHWEST DELTA USAIR CONTINENTAL EASTERN -0.002 o -0.084** 0.013 0.000 0.035 0.059** -0.09 0.02 -3.88 0.67 0.00 1. 34 3.22 -0.09 0.02 -3.88 0.73 0.00 1. 34 2.78 -0.004 -0.029** -0.017 0.035** 0.045** 0.049** 0.066** -0.20 -2.03 -0.98 2.00 2.63 2.88 3.96 -0.42 -2.50 -0.97 2.09 2.52 2.89 3.41 WASHINGTON, D.C. - NATIONAL USAIR EASTERN CONTINENTAL NORTHWEST DELTA PAN AM AMERICAN MIDWAY 0.055** -0.071 0.020 -0.041 -0.008 0.051** 0.021 0.930 2.27 -0.58 0.62 -1. 26 -0.20 0.73 0.90 1. 21 1.24 0.68 -1. 40 -0.43 2.73 0.70 0.075** -0.064 0.000 -0.041 -0.029** -0.007 0.005 0.093** 3.50 -0.52 0.20 -1. 38 -0.73 -0.11 0.23 3.26 1.17 0.21 -1. 56 -2.15 -0.45 0.19 NEW YORK - LAGUARDIA USAIR EASTERN PAN AM DELTA AMERICAN NORTHWEST UNITED CONTINENTAL MIDWAY -0.037** 0.022 -0.044 0.092** -0.064** -0.008 0.001 0.041 0.233** -2.08 0.61 -1.14 7.05 -2.26 -0.17 0.01 1. 06 3.34 -3.09 0.83 -1. 39 3.64 -2.30 -0.17 0.01 1. 03 34 0.015 -0.043** -0.062** 0.059** -0.035 -0.029 0.014 0.028 0.284** 1. 04 -1.12 -1.79 4.13 -1.17 -0.76 0.58 0.74 11.43 1. 02 -2.76 -3.20 2.14 -1.19 -0.71 0.25 0.73
TABLE 3B ( CONTINUED) NEW YORK - KENNEDY PAN AM -0.060 -0.75 -1. 09 -0.08 -1. 00 -1. 61 USAIR -0.042 -0.31 -0.55 -0.056 -0.41 -0.76 EASTERN -0.918** -52.6 -21.17 -0.955** -52.19 -21.00 AMERICAN 0.228 2.63 0.253 2.89 DELTA 0.082 1. 31 1. 48 0.082 1. 32 1. 50 UNITED 0.118** 7.80 -0.018 -0.87 t e - t statistic, with variances of distribution of leased and 0&0 slots assumed equal. t U - t statistic, with variance of distribution of leased and 0&0 slots estimated separately. ** Significant at .05 level * Significant at .1 level Source: Tabulated from FAA slot data, May-June 1990. Table 4 lists the net leasing positions of specific carriers. The anticompetitive leasing hypothesis predicts that carriers with a large share of the slots at an HDTA will tend to be net lessors, so the hypothesis would be supported if the difference between the number of slots operated and the number of slots held is negative. But the data show otherwise. Instead, the carriers with relatively large slot shares were almost always net lessees. At O'Hare, the carrier with the largest number of slots, United, leased 20 more slots from others than others leased from united; for the second largest, American, the comparable figure is 34. At National and LaGuardia the top three slot holders were net lessees. The largest slot holder at National, USAir, was a 35
net lessee of slots vis a vis every slot holder it dealt with. 6o Finally, at Kennedy, the top two slot holders that are air carriers, Pan Am and American, were n~t lessees; moreover, the largest holder of slots at Kennedy, Shawmut Bank, could also be considered a net lessee for these purposes, by comparing the number of slots held by Shawmut Bank in trust for TWA (60) to the number of slots TWA operated (62). 60 It has even positions with Delta (one slot), is a net lessor to Northwest (4 slots leased from Northwest, 3 slots leased to Northwest), and leases one slot from continental. In addition, USAir leases two slots to TWA and leases 4 slots from the bank that holds TWA's slots. Thus, although USAir's relatively intense use of its 0&0 slots, compared to the usage rates of its lessees, would be consistent with the anticompetitive leasing hypothesis, the fact that it operates more slots than it owns is not. 36
TABLE 4 SLOTS OWNED, OPERATED, AND NET LEASING POSITION Slot holders with positions as net lessors in bold SLOT HOLDER SLOTS HELD SLOTS OPERATED SLOTS OPERATED - SLOTS HELD CHICAGO - O'HARE UNITED AMERICAN NORTHWEST DELTA USAIR AIR WISCONSIN CONTINENTAL SHAWMUT BANK TWA EASTERN AMERICA WEST AMERICAN TRANS AIR PAN AM FLYING TIGER MIDWEST EXPRESS FAA TOTAL 658 491 71 64 45 44 36 36 o 21 7 5 4 3 2 1 1488 678 525 55 52 46 45 38 o 27 13 7 o 2 o o o 1488 20 34 -16 -12 1 1 2 -36 27 -8 o -5 -2 -3 -2 -1 o WASHINGTON, D.C. - NATIONAL USAIR EASTERN NORTHWEST CONTINENTAL DELTA SHAWMUT BANK TWA PAN AM AMERICAN UNITED TRUMP MIDWAY MIDWEST EXPRESS AMERICA WEST TOTAL 148 82 53 43 47 44 o 38 35 28 28 16 4 4 575 37 152 82 57 47 46 o 33 39 39 28 28 16 4 4 575 4 o 4 -1 -1 -44 33 1 1 o o o o o o
TABLE 4 (CONTINUED) SLOT HOLDER SLOTS SLOTS SLOTS OPERATED - HELD OPERATED SLOTS HELD NEW YORK LAGUARDIA USAIR 151 163 12 EASTERN 72 78 6 PAN AM 64 70 6 TRUMP 57 57 0 DELTA 56 63 7 AMERICAN 57 64 7 UNITED 45 45 0 NORTHWEST 45 41 -4 CONTINENTAL 29 28 -1 SHAWMUT BANK 24 0 -24 TWA 0 32 32 MIDWAY 14 18 4 MIDWEST EXPRESS 4 6 2 AMERICA WEST 0 2 2 FAA 49 0 -49 TOTAL 667 667 0 NEW YORK - KENNEDY SHAWMUT BANK 60 0 -60 TWA 0 62 62 PAN AM 42 48 6 AMERICAN 21 24 3 USAIR 20 19 -1 EASTERN 17 14 -3 NORTHWEST 12 0 -12 DELTA 9 5 -4 FAA 9 0 -9 UNITED 6 6 0 COMMAND 2 2 0 MGM GRAND AIR 1 1 0 FLYING TIGER 1 1 0 UPS 1 1 0 AMERICA WEST 1 2 1 METRO AIR, N.E. 0 13 13 PAN AM EXPRESS 0 4 4 TOTAL 202 202 0 38
(c) Regression Analysis A multiple regression setting was also used to examine the anticompetitive and competitive hypotheses. In these regressions, the dependent variable is the number of days a slot was used. The dependent variable is measured by using either all 61 days in the period or only the 44 weekdays. To test the anticompetitive usage hypothesis -- that slots held by carriers with a large share of the slots at an HDTA will use their slots less intensively than carriers with a small slot share -- against the procompetitive hypothesis, which suggests the opposite, the slot holder's share of slots at an HDTA was included as an independent variable. To examine the anticompetitive leasing hypothesis -- that carriers with relatively more slots will lease their slots to carriers that use them at lower rates -- a dummy variable that denotes whether or not a slot was leased and a variable that interacts this dummy variable with the slot share variable were included as independent variables. To control for other factors that may affect slot use, dummy variables that denote whether a slot was traded or was operated by more than one carrier during the time period were included as independent variables. To control for differences in use in peak versus off- peak periods, dummy variables that denote the beginning of the time period defined by the slot were included as independent variables. To control for differences in slot use resulting from the fact that a carrier has an extensive route structure (as opposed to differences in slot use resulting from a carrier's 39
large presence at an HDTA) , the slot holder's national market share, based on revenue passenger miles, was included as an . d . bl 61 1ndepen ent var1a e. Finally, the slot withdrawal number was included as an independent variable to control for possible differences in the strength of the property right associated with 62 holding the slot. The following linear specification was estimated using ordinary least squares (OLS):63 61 These shares were obtained from Air Carrier Traffic statistics Monthly, u.s. DOT (May 1990) and (June 1990). 62 See the discussion in note 17, supra. The property right associated with a low numbered slot may be less clear than that associated with a high numbered slot because the low numbered slot is more likely to be withdrawn. If the absence of clearly defined property rights inhibits slot trading, low numbered slots would be less likely to move to their highest valued use. See A. Kleit, "competition without Apology: Market Power and Entry in the Deregulated Airline Industry," 14 Regulation, (forthcoming 1991) . 63 In addition to the OLS regressions reported in Table 5, several other sets of regressions were run. First, the relationship between Air Wisconsin and United was addressed. Although Air Wisconsin was independently owned and operated, it had a "code sharing" arrangement with United (i.e., it agreed to a large number of joint fares with united and was listed in the Official Airline Guide and on Computer Reservation Systems as UA*.) The regressions reported in Table 5.1 assume that United and Air Wisconsin are separate carriers. The regressions were also run for O'Hare under the assumption that united and Air Wisconsin were one carrier, but the results were not quantitatively or qualitatively different from those reported in Table 5.1. In addition, the regressions in Table 5.1 - 5.4 were estimated using a Tobit procedure with an upper limit at the maximum number of days: again there was very little difference in the results. Finally, potentially unused slots were included. The regressions reported in Table 5.1 - 5.4 do not include the few slots that, according to the data, were used less than 65 percent of the time (for example, at O'Hare those totaled four, held by American). Because no slots were withdrawn for failure to meet the FAA's "use or lose" standard during the period studied, these observations may represent missing data rather than nonuse. All of the regressions were also run with these (continued ... ) 40
NDOPER = ao + a,SLMS + a 2NAMS + a 3WITHD + a 4LEASE + a 5 LEASE*SLMS + a 6MULTO + a 7MULTO*SLMS + a 8 MULTH + a~LTH*SLMS + L atP t + e where NDOPER SLMS NAMS 64 WITHD LEASE 65 MULTH is the number of days (or weekdays) the slot was operated in the two month period, is the slot holder's share of the total number of nonrestricted and nonexempt slots at the end of the two month period, is the slot holder's national market share (based on Revenue Passenger Miles) for the two month period, is the FAA withdrawal priority number, is a dummy variable that equals one if the slot is operated by a single operator other than the slot holder, zero otherwise, is a dummy variable that equals one if the slot is operated by more than one operator, and was not traded in the two month period, zero otherwise, is a dummy variable that equals one if the slot holder changed during the two month period, is a dummy variable that equals one if the slct's hour or half-hour period starts at time t. The dummy variable for the latest slot period is left out. The anticompetitive usage hypothesis -- that carriers with a large share of the slots at an HDTA will use their slots less 63 ( ••• continued) observations (and their low usage rates) included; unless otherwise noted, no substantive changes arose. 64 Shawmut Bank took possession of the slots, as a trustee, from TWA in July 1989, and simultaneously agreed to long term leases of the slots back to TWA. Consequently, slots held by Shawmut Bank were assigned TWA's national market share. 65 Slots held by Shawmut Bank and operated by TWA were not counted as leased. ~ In cases where there was more than one slot holder, the slot was assigned, for purposes of this analysis, to the entity that held the slot the majority of the days during the two-month period. 41
intensively than carriers with a small slot share -- is Ho: a, < 0, and the procompetitive hypothesis is H,: a 1 ~ O. The anticompetitive leasing hypothesis -- that carriers with a large slot share will lease their slots to carriers who will use these slots at a lower rate -- is Ho: a 4 + asSLMS < 0 for SLMS large, and the procompetitive hypothesis again implies the opposite. The results from these regressions are reported in Table 5 for all four HDTAs. The results from each HDTA will be discussed separately, in turn. In general, the results do not support either anticompetitive hypothesis. 1. Chicago - O'Hare O'Hare is the most concentrated HDTA and is also the only HDTA that is primarily used as a hub (by American and united) . The high concentration might suggest that the potential for anticompetitive behavior would be greatest at O'Hare. 67 On the other hand, the fact that O'Hare is the only HDTA used primarily as a hub suggests that efficiency-based theories may better explain its concentration level. Table 5.1 reports the results for O'Hare. The anticompetitive usage hypothesis can be rejected for both the all-days and the weekday-only regressions. The coefficient on slot market share (SLMS) is positive and statistically significant in both cases.~ 67 See DOJ Guidelines, supra note 37. ~ In all cases, the coefficient is small in magnitude. For example, a coefficient of 2.62 implies that a 40 percent increase (continued ... ) 42
Examining the anticompetitive leasing hypothesis, the coefficients on the lease dummy variable (a 4 ) and on the lease dummy interacted with slot market share (as)' for the weekday regression, are both statistically significant. The regression implies that leased slots would be used more intensively than the lessor's owned and operated slots if the lessor's slot holding 69 share were above 20.5 percent. That is, according to the regression results, the two largest carriers at O'Hare, United and American, would lease slots to carriers that used them at rates greater than or equal to the rate at which united and American used their owned and operated slots. This finding is consistent with the findings in Table 3. In the all-days reg~ession, neither coefficient is significant. As for the other variables, the coefficients on the variables denoting whether a slot changed operators suggest that ~( ... continued) in slot share (e.g., roughly the difference between the leading carrier, united, and third largest carrier, Northwest) would result in the slot being used just over one additional day per two month period. Because all air carriers at O'Hare use their slots at high rates, this small difference may reflect differences in the abilities of large carriers with hub operations at O'Hare to cover mechanical or personnel problems, rather than planned differences in schedules. 69 The critical slot share is given by -a 4 /a S • That is, if a 4 is positive and as is negative, the estimated coefficients imply that a slot leased from a carrier with a slot share greater than this ratio is likely to be used more intensively than the lessor uses its 0&0 slots. On the other hand, if a 4 is negative and as is positive, the estimated coefficients imply that a slot leased from a carrier with a slot share lower than this ratio will be used more intensively than the lessor uses its 0&0 slots. 43
70 these slots were used less. Similarly, the coefficients on the variable that denotes whether a slot had m~re than one owner dnring the period studied suggest that these slots were used less. The coefficient on national market share is negative in both the all-days and weekday-only regressions and is statistically significant in the weekday-only regression. 71 The coefficient on the withdrawal number is positive and is statistically significant in the weekday regressions. 72 For expositional convenience, the coefficients on the time dummy nvariable are not reported. Overall, all regressions are statistically significant at the 1 percent level. The regression results for O'Hare support the procompetitive hypotheses. There was a positive and statistically significant relationship between slot share and rate of use. In addition, the regression based on weekdays suggests that slots leased by large carriers to others were used at a higher rate than their 70 The coefficient on the dummy variable MULTO is negative and significant in all the O'Hare regressions. The interaction term MULTO*SLMS is positive, but is not statistically significant in any regression. 71 At O'Hare, the two carriers with the largest slot shares are also the largest national carriers. The large negative coefficient on national market share likely reflects the relationship for other carriers at O'Hare, all who have small (less than 5 percent) slot shares there. 72 This possible relationship may reflect the assignment of highly valued or profitable flights to those slots least likely to be withdrawn (i.e., those with the highest withdrawal numbers). n Estimates of these coefficients are available upon request. 44
0&0 slots. Both results are inconsistent with the anticornpetitive hypotheses. TABLE 5.1 REGRESSION RESULTS CHICAGO - O'HARE AIRPORT DEPENDENT VARIABLE - NUMBER OF DAYS OPERATED, MAY-JUNE 1990 (t-statistics in parenthesis) VARIABLE CONST SLMS NAMS WITHD LEASE LEASE*SLMS MULTO MULTO*SLMS MULTH MULTH*SLMS N MEAN DEP. VAR. R-SQUARED ADJ R-SQUARED F LOG LIKELIHOOD ALL DAYS 52.30** 31.15 ) 2.15** ( 2.30) -0.29 ( -0.09) 0.0003 ( 1.46) 0.43 ( 0.92) -1. 57 ( -0.79) -1. 91** ( -2.71) 1. 08 0.40) 2.63 ( 1.39) -14.09** ( -2.47) 1480 56.90 0.29 0.27 15.59 -3923.29 COEFFICIENTS WEEKDAYS ONLY 41.79** ( 39.30) 2.62** ( 4.42) -11.81** ( -6.08) 0.0002* ( 1.94) -0.88** -3.00) 4.29** 3.39) -1.56** -3.50) 3.36** 1. 96) 1. 07 0.90) -4.41 ( -1.22) 1480 41. 96 0.25 0.23 12.99 -3247.49 * Significant at the .05 level (one-tailed test) ** Significant at the .025 level (one-tailed test) 45
2. Washington, D. C. - National Table 5.2 reports the results for National. The coefficients on slot market share, for both the all-days and weekday-only regressions, are negative, but small and not statistically significant. Thus, the data show no discernable relationship between slot market share and usage rate. The coefficients on the lease and lease interaction variables are also not statistically significant. The regression based on all days implies that carriers with a market share greater than 8.9 percent, which in the period studied would include uSAir, Northwest, and Eastern, leased slots to carriers that used them at a lower rate. Eastern has exited the airline business, and USAir and Northwest were both net lessees, positions inconsistent with the anticompetitive leasing hypothesis. Thus, the data do not suggest an anticompetitive problem with leased slots at National.~ The coefficient on national market share is positive and statistically significant in both regressions. The coefficients on the mUltiple operator and holder variables imply that slots operated by more than one carrier were used less intensively.75 ~ t d . I .. I . As no e prev10us y, USA1r 1S a net essee both 1n general and vis-a-vis each slot holder. See Table 4, supra, and the discussion in note 60. ~ Because the number of trades at Kennedy, LaGuardia and National was small, the multiple holder dummy variable and the interaction term containing this variable both could not be included in the regressions for these HDTAs. The coefficient on the interaction term, the variable included in the regressions, is not significant in any of the regressions. 46
The regression results for National present no discernable pattern relating slot share and the rate of slot use. Nor do the regression results suggest that slots leased by larger carriers to others were used at a significantly lower rate. 47
TABLE 5.2 REGRESSION RESULTS WASHINGTON, D.C. - NATIONAL AIRPORT DEPENDENT VARIABLE - NUMBER OF DAYS OPERATED, MAY-JUNE 1990 (t-statistics in parenthesis) VARIABLE CONST SLMS NAMS WITHD LEASE LEASE*SLMS MULTO MULTO*SLMS MULTH MULTH*SLMS N MEAN DEP. VAR. R-SQUARED ADJ R-SQUARED F LOG LIKELIHOOD ALL DAYS 56.44** 23.07) -1. 91 -0.71) 12.09** ( 2.81) -0.00005 ( -0.05) 1. 07 ( 0.93) -10.40 ( -1.12) -2.49 -1.45) 4.10 0.29) -1.06 -0.06) 572 55.92 0.19 0.15 5.09 -1711.92 COEFFICIENTS WEEKDAYS ONLY 40.62** 23.82) -0.37 ( -0.02) 11.13** 3.71) 0.0001 0.13) 0.70 0.88) -7.87 ( -1.22) -2.19* ( -1.83) 4.26 ( 0.43) -0.33 ( -0.03) 572 41.48 0.17 0.13 4.32 -1505.30 * Significant at the .05 level (one tailed test) ** Significant at the .025 level (one tailed test) 3. New York - LaGuardia Table 5.3 reports the results for Laguardia. The coefficient on the slot market share variable is negative in the all-days regression and positive in the weekday-only regression; 48
however, neither coefficient is statistically significant. For the weekday regressions, none of the coefficients on the lease and lease interaction variables is statistically significant. For the all-days regression, the coefficients are statistically significant and imply that carriers with slot shares greater than 18.9 percent leased slots to others that used them at a lower rate than their 0&0 slots. The only carrier with a share that high is USAir, which at LaGuardia was both a net lessee and leased to carriers that used these slots at a higher rate. 76 The coefficient on national market share is negative but not statistically significant, and again, slots operated or held by more than one carrier were used at a lower rate. The results for Laguardia are similar to those for National: the regression results present no discernable pattern relating slot share and the rate of slot use, and do not suggest that slots leased by large carriers to others were used at a significantly lower rate. ~ See Tables 3 and 4, supra. 49
TABLE 5.3 REGRESSION RESULTS NEW YORK - LAGUARDIA AIRPORT DEPENDENT VARIABLE - NUMBER OF DAYS OPERATED, MAY-JUNE 1990 (t-statistics in parenthesis) VARIABLE ALL DAYS COEFFICIENTS WEEKDAYS ONLY CONST SLMS NAMS WITHD LEASE LEASE*SLMS MULTO MULTO*SLMS MULTH 50.11** 35.81** ( 14.29) ( 15.65) -5.40 2.03 ( -1. 62) ( 0.94) -1.86 -0.16 ( -0.54) ( -0.07) -0.0003 -0.0006 ( -0.35) ( -1. 06) -3.05** -0.75 ( -2.89) ( -1. 08) 24.75** 3.15 2.63) ( 0.51) -3.15** -1. 64** ( -2.66) ( -2.12) 12.28 5.43 ( 1.17 ) 0.80) MULTH*SLMS N MEAN DEP. VAR. R-SQUARED ADJ R-SQUARED F LOG LIKELIHOOD -65.07 ( -1.14) 614 55.76 0.19 0.13 3.12 -1797.38 -86.06** ( -2.30) 614 41. 40 0.18 0.12 2.95 -1535.23 * Significant at the .05 level (one tailed test) ** Significant at the .025 level (one tailed test) 50
4. New York - Kennedy Finally, Table 5.4 lists the regression results for Kennedy. The coefficient on slot market share for both the all-days and weekday-only regressions is negative and statistically significant. Although this result is consistent with the anticompetitive usage hypothesis, it may also be explained by special circumstances at Kennedy. The largest slot holder, with 31 percent of the slots, was a bank that entered a purchase-lease arrangement with TWA. The second largest slot holder was Pan Am. Both TWA and Pan Am had been experiencing financial difficulties on transatlantic flights. 77 Thus, the negative coefficient on market share may simply reflect these largest carriers' reduced output due to operating difficulties, rather than a monopolistic t t d t ' 78 ou pu re uc 10n. n Based on the set of unrestricted slots and international slots, international slots make up 23.5 percent of the slots at Kennedy, compared to 7.2 percent at LaGuardia, 4.9 percent at O'Hare, and 0 percent at National. Both Pan Am and TWA relied heavily on transatlantic flights for operating revenues, and many of these flights operated out of Kennedy. Pan Am received more than three-quarters of its operating revenues from international operations, over half of these revenues coming from transatlantic service. TWA received over a third of its operating revenues from its transatlantic service. See Secretary's Task Force on Competition in the u.S. Domestic Airline Industry: International Air Service," u.S. Department of Transportation, February 1990, pp. 14-15. Although these international flights used restricted international slots that are not in our dataset, passengers from these flights were an important source of traffic for these airlines' domestic flights, which used slots that are in our dataset. 78 Both carriers faced problems at various times due to fears of terrorism or increased competition on transatlantic routes. See Secretary's Task Force on Competition in the u.S. Domestic Airline Industry: International Air Service," supra note (continued ... ) 51
The coefficients on the lease and lease interaction variables imply that leased slots would be used more intensively than the lessor's owned and operated slots if the lessor's share were above 10 percent. This is not consistent with the anticompetitive leasing hypothesis. The coefficient on national market share is negative but not statistically significant in both regressions. 78 • d( ... contlnue) 77, p. 15 and pp. 20-1. These two carriers were the largest carriers serving the transatlantic market in 1989, but were facing stiff competition from lower cost entrants. Id., Table 111-39, pp. 93. 7hese difficulties SUbsequently resulted in both carriers selling, among other things, their transatlantic route authority from New York to other airlines. See, e.g., "Flight Plans: How Airlines stack Up," Wall Street Journal, Monday, June 17, 1991, p. B1, and "DOT Rejects Bids on 3 London Routes," Washington Post, Friday March 15, 1991, p. D1. Thus, market forces have already reduced the presence of these two carriers at Kennedy Airport. 52
TABLE 5.4 REGRESSION RESULTS NEW YORK - KENNEDY AIRPORT DEPENDENT VAhIABLE - NUMBER OF DAYS OPERATED, MAY-JUNE 1990 (t-statistics in parenthesis) VARIABLE CONST SLMS NAMS WITHD LEASE LEASE*SLMS MULTO MULTO*SLMS MULTH MULTH*SLMS N MEAN DEP. VAR. R-SQUARED ADJ R-SQUARED F LOG LIKELIHOOD ALL DAYS 47.62** ( 2.54) -26.84** ( -2.43) -8.41 ( -0.77) 0.0021 ( 0.61) -1. 87 ( -1.60) 13.59* 1. 95) -8.10** -2.59) 45.85** ( 2.86) 32.62 1. 27) 169 56.89 0.19 0.10 2.13 -491.44 COEFFICIENTS WEEKDAYS ONLY 40.45** ( 2.79) -22.89** ( -2.68) -7.60 ( -0.90) 0.0026 ( 0.18) -1.07 -1.17 ) 10.81** 2.01) -6.77** ( 2.80) 38.55** 3.11) 26.35 ( 1.33) 169 40.94 0.19 0.10 2.13 -448.15 * Significant at the .05 level (one tailed test) ** Significant at the .025 level (one tailed test) In sum, the data provide little support for the anticompetitive usage and anticornpetitive leasing hypotheses. A positive and significant relationship between slot share and slot use was found at O'Hare, and no discernable pattern relating slot 53
share and the rate of slot use was found at National and LaGuardia. The only result that might support the anticompetitive hypotheses, the negative and significant relationship between slot share and slot use found at Kennedy, is probably explained by the operating problems experienced by the largest carriers at that airport. In addition, the data do not suggest that slots leased by large carriers to others were used at a significantly lower rate, and suggest that the largest slot holders at HDTAs tended to be net lessees. B. Changes in the "Use or Lose" Rules The SNPRM contains a proposal to change the "use or lose" threshold from 65 percent based on all days to 90 percent based on weekday use only. To examine the potential impact of the proposed rule, Table 6 lists the number of slots, based on usage during the May-June 1990 reporting period, that would have complied under the proposed 90 percent "use or lose" threshold. According to the data, the rule would affect, based on current reporting practices, 15 to 20 percent of the slots at National, LaGuardia, and O'Hare, and over 30 percent of the slots at Kennedy. 54
TABLE 6 SLOT HOLDER COMPLIANCE WITH PROPOSED "USE OR LOSE" RULE NUMBER OF PERCENTAGE SLOTS USED OF SLOTS < 90% > 90% COMPLYING CHICAGO - O'HARE SHAWMUT BANK 8 28 77.8 AMERICAN 78 413 84.1 AMERICAN TRANS AIR 3 2 40.0 AMERICA WEST 0 7 100.0 AIR WISCONSIN 0 44 100.0 CONTINENTAL 8 28 77.8 DELTA 9 55 85.4 EASTERN 1 20 95.2 FLYING TIGER 1 2 66.7 MIDWEST EXPRESS 1 1 50.0 NORTHWEST 8 63 88.7 PAN AM 0 4 100.0 UNITED 94 564 85.7 USAIR 7 38 84.4 OVERALL 219 1269 85.3 VvASHINGTON, D. C. - NATIONAL SHAWMUT BANK 13 31 70.5 AMERICAN 6 29 82.9 AMERICA WEST 0 4 100 CONTINENTAL 3 45 93.8 DELTA 9 38 80.9 EASTERN 40 42 51.2 MIDWEST EXPRESS 0 4 100 MIDWAY 2 14 87.5 NORTHWEST 14 39 73.6 PAN AM 3 35 92.1 TRUMP 3 25 89.3 UNITED 4 24 85.7 USAIR 14 134 90.5 OVERALL 111 464 80.7 55
TABLE 6 (CONTINUED) NUMBER OF PERCENTAGE SLOTS USED OF SLOTS < 90% > 90% COMPLYING NEW YORK - LAGUARDIA SHAWMUT BANK 5 19 79.2 AMERICAN 12 45 79.0 CONTINENTAL 5 24 82.8 DELTA 4 52 92.9 EASTERN 9 63 87.5 MIDWEST EXPRESS 0 4 100 MIDWAY 1 13 92.9 NORTHWEST 12 33 73.3 PAN AM 19 45 70.3 TRUMP 7 50 87.7 UNITED 11 34 75.6 USAIR 17 134 88.7 OVERALL 118 549 82.2 NEW YORK - KENNEDY SHAWMUT BANK 19 41 68.3 AMERICAN 7 14 66.7 AMERICA WEST 0 2 100 COMMAND 0 2 100 DELTA 1 8 88.9 EASTERN 15 2 11. 8 FLYING TIGER 1 0 0 MGM 1 0 0 NORTHWEST 2 10 83.3 PAN AM 13 29 69 UNITED 0 6 100 UPS 0 1 100 USAIR 2 18 90 OVERALL 61 133 68.6 Source: Tabulated from FAA slot data. Table 6 may overstate the potential impacts of altering the "use or lose" rules. A carrier holding a large number of slots in a slot period might be able to comply with a new "use or lose" standard by moving flights among slots, taking flights assigned 56
to slots being used more than 90 percent of the time on weekdays and reassigning them to slots being used less than 90 percent of the time on weekdays. To illustrate, suppose an airline has five slots in a slot period, four of them listed as being used 39 weekdays in a reporting period (88 percent) and one listed as being used all 44 weekdays. At those rates, only one out of the five would comply with the proposed "use or lose" rule. However, by reassigning one flight from the 100 percent slot to each of the four slots being 88 percent slots, the holder could report all five slots being used 40 weekdays (90.9 percent), in 79 compliance with the proposed 90 percent "use or lose" rule. carriers with few slots per period would be less able to make such rearrangements. Thus, a higher "use or lose" threshold could affect these carriers more than larger carriers. If the five slots in the example above were not held by a single carrier, but instead each was held by a different one, then all four of the carriers with an 88 percent usage rate would have to 80 add flights to comply with the proposed rule. 79 A nonhypothetical case can be constructed from the data studied for this comment. During the 645-715am slot period at O'Hare, 17 out of the 19 American slots were not in compliance with the proposed "use or lose" rule; however, by reassignments among slots, American could have brought 18 of them into compliance. To keep the 19th slot, American would have had to add 28 weekday flights in the two month period. If instead it sold the slot, it would have had to cancel or shift to other slot periods 12 flights. 80 The ability of large carriers at HDTAs to reallocate flights among slots to ensure compliance led the FAA to propose previously in the NPRM that the "use or lose" thresholds be based upon the number of slots held in any time period. Specifically, (continued ... ) 57
The overall weekday use rates for May-June 1990 imply that almost all slot holders already could be close to complying with the proposed 90 percent rule. 81 Achieving complete compliance would no doubt call for schedule rearrangements and some added flights, but the high level of usage suggests that the rule change may not increase significantly either the rate at which slots are used or the numb0r of slots at risk of withdrawal for nonuse. However, where the proposed "use or lose" rule is binding, a too restrictive "use or lose" provision could impose unintended costs on air carriers and may reduce welfare. First, the proposed "use or lose" threshold may increase the probability that slots are withdrawn for reasons beyond the control of the h ld . . 82 slot 0 er or alr carrler. Although, as the FAA explains, 80 ( ••• continued) the use of lose threshold would have been 90 percent for carriers with ten or more slots in a hour, 80 percent for carriers with more than four but fewer than ten slots in an hour, and 65 percent for carriers with four or fewer slots in an hour. See SNPRM, supra note 2, p. 46674. The previous proposal to raise the "use or lose" percentage only for carriers with a larger number of slots per period has been dropped because carriers opposed it and because it would have made compliance and administration more difficult. The ability to re-allocate flights is not greater for larger carriers when cancellations in an hour are positively correlated. For example, if weather closes the airport, all slots in that period would be counted as not used. Thus, there would be no flights in that time period available to reallocate across slots. 81 See Table 2, supra. 82 Under the current "use or lose" reporting system, slots affected by flight cancellations (e.g., due to weather related problems or the recent phone outage in New York) are counted as not used, and theoretically, five weekday airport closings 5 PM (continued ... ) 58
the change to a 90 percent weekday rule will not reduce the allowable number of cancellations per period for a carrier operating a "normal" Monday through Friday schedule,83 under the current rules a carrier can greatly increase the number of allowable cancellations by using or leasing the slot on weekends. And finally, to protect a slot against withdrawal, a carrier may add flights just to insure compliance with the proposed usage constraint. 84 This behavior may be wasteful. C. Restrictions on the Sale and Transfer of Lottery Slots Finally, greater restrictions are proposed on the sale and transfer of lottery slots to deter those carriers interested only in obtaining slots for resale from entering the lottery.8S 82 ( ... continued) (e.g., due to bad weather) in a two-month period would put all carriers at an HDTA in violation of the proposed "UGe or lose" rules. A carrier can request a waiver from the "use or lose" rules only if it experienced problems on nine or more consecutive days in a period. See 14 C.F.R. Part 93.227. The SNPRM contains no proposal to amend the waiver procedure. 83 See SNPRM, supra note 2. Under the proposed 90 percent weekday rules, a carrier with one slot in a given hour at a given HDTA would have to schedule a flight on every weekday and wouid have a cushion of .5 flights per week (or equivalently, one weekday cancellation every two weeks). Now, the same airline scheduling weekday-only flights has a cushion of .45 flights a week. 84 See Comment of the DOJ, supra note 12. Flight operations in addition to those that would have been conducted with a less binding constraint are likely driven by compliance with a rule and not the marginal benefits and costs of conducting the flight. Thus, in the absence of a monopolistic output reduction, it is possible that these additional flights will not increase and may even decrease total welfare. 8S For a general discussion, see Richard A. Posner, "The Social Costs of Monopoly and Regulation," 83 J. Pol. Econ. 807 (1975) . 59
These proposed rules are apparently in response to the observation that slots allocated to new entrant carriers were often quickly sold to large carriers. The FAA notes that, at times, such sales occurred on or soon after the expiration of the 60 day minimum holding requirement.& Examination of the ownership history of slots obtained in the 1986 slot lottery confirms this and suggests that the vast majority of these slots were traded or otherwise changed ownership within two years. Table 7 lists this set of slots allocated in this 1986 FAA lottery by RDTA and by lottery recipient at three of the four HDTAs. 87 Slots allocated in this lottery were obtained by withdrawing approximately 5 percent of the slots at RDTAs from incumbent carriers. Table 7 also lists the donating carriers and the number of days the original lottery slot was held by the lottery recipient. & See SNPRM, supra note 2, p. 46676. 87 Data for the fourth HDTA, Kennedy Airport, were not available. 60
TABLE 7 1986 FAA LOTTERY SLOTS Slots held more than two years are listed ir, bold. LOTTERY DONOR SLOT WITHDRAWAL DAYS RECIPIENT PERIOD NUMBER ORIGINAL SLOT HELD BY LOTTERY RECIPIENT WASHINGTON, D.C. - NATIONAL AIR WISCONSIN NORTHWEST 1600 1121 91 MIDWAY 1600 1117 0 DELTA 1100 1223 0 NEW YORK AIR 1700 1188 185 FAA 2100 1588 0 NEW YORK AIR 1800 1303 185 BRANIFF NEW YORK AIR 1900 1281 358 PAN AM 1400 1666 217 AIR ATLANTIC PIEDMONT 1300 1127 0 EASTERN 1400 1076 241 FAA 800 1274 90 AMERICAN 900 1044 241 SKYBUS EASTERN 1500 1009 112 DELTA 1000 1230 112 NORTHWEST 1500 1058 112 PAN AM 900 1335 112 JET AMERICA FAA 2000 1475 0 FAA 2000 1097 0 EASTERN 700 1140 441 NEW YORK AIR 1900 1310 0 MIDWEST EXPRESS EASTERN 1000 1010 1894 FAA 1100 1138 0 PRESIDENTIAL USAIR 1300 1175 23 PIEDMONT 1200 1033 41 AMERICAN 700 1321 23 NEW YORK AIR 800 1150 23 TRANS TAR USAIR 1300 1175 39 PIEDMONT 1200 1033 17 61
TABLE 7 (CONTINUED) WESTERN USAIR 1200 1298 210 FAA 2100 1548 370 NEW YORK - LAGUARDIA AIR WISCONSIN EASTERN 1530 3186 92 EASTERN 1800 3436 92 EASTERN 1830 3507 92 NEW YORK AIR 1000 3709 0 UNITED 1500 3819 186 SKYBUS EASTERN 1430 3489 0 USAIR 1630 1630 38 AMERICAN 930 3201 0 EASTERN 1000 3198 0 MIDWEST EXPRESS EASTERN 1730 3064 501 EASTERN 1900 3591 1834 EASTERN 1030 3234 1834 AMERICAN 1130 3260 175 OZARK EASTERN 1400 3278 53 TWA 1400 3055 0 PRESIDENTIAL AMERICAN 800 3049 98 EASTERN 1100 3253 214 MIDWAY 2130 3136 222 EASTERN 2030 3124 214 NEW YORK AIR 2130 3029 214 AMERICAN 1200 3328 214 NEW YORK AIR 830 3347 145 WESTERN UNITED 1700 3264 183 NEW YORK AIR 1930 3848 210 FAA 1300 3107 0 USAIR 1600 1600 210 NEW YORK AIR 2130 3063 210 EASTERN 1200 3244 210 PIEDMONT 700 3309 103 62
TABLE 7 - (CONTINUED) WORLD EASTERN 700 3530 0 TWA 1000 3156 0 EASTERN 900 3230 90 FAA 1300 3195 0 PIEDMONT 2030 3750 88 PAN AM 830 3336 90 NEW YORK AIR 930 3283 91 CHICAGO - O'HARE AIR CAL USAIR 1415 7646 243 UNITED 1645 7021 549 UNITED 1645 7041 549 UNITED 745 7106 365 AMERICAN 1745 8265 3 AMERICAN 1915 7028 3 AMERICAN 1815 8281 3 AMERICAN 1515 7022 365 AMERICA WEST USAIR 1915 8087 0 NORTHWEST 1415 7412 1582 FAA 1215 7400 0 UNITED 1545 7809 0 FAA 1215 7264 0 BRANIFF OZARK 815 7816 661 EVERGREEN FAA 1045 8142 0 FAA 715 8453 0 SKYBUS BRITISH AIRWAYS 1015 8035 0 AMERICAN 945 7026 0 FAA 915 7272 0 NORTHWEST 1415 7630 0 REPUBLIC 1515 7794 0 FAA 1045 7400 0 AMERICAN 1615 8303 0 AMERICAN 915 7015 0 MCCLAIN BRITISH AIRWAYS 1415 7559 0 USAIR 1415 7107 0 FAA 1315 8391 96 AMERICAN 1245 7030 119 DELTA 1145 7054 105 PIEDMONT 2015 8565 344 FAA 1245 7580 96 FAA 1815 7112 0 63
TABLE 7 (CONTINUED) PRESIDENTIAL UNITED 1815 8048 0 UNITED 1615 8226 0 PAN AM AMERICAN 1915 7043 172 WESTERN FAA 1315 8308 0 FAA 1215 7135 113 UNITLD 2045 7006 274 FAA 1145 8195 0 WORLD REPUBLIC 1515 8410 0 UNITED 2015 7086 0 FAA 715 8453 0 UNITED 2015 7085 0 UNITED 2045 7080 0 USAIR 1915 8087 0 AMERICAN 745 7100 0 NOT PICKED AMERICAN 2045 7011 AMERICAN 1915 7018 UNITED 1615 7505 AMERICAN 645 7536 UNITED 1645 7138 AMERICAN 645 7307 PEOPLE'S EXPRESS 1015 8659 UNITED 2045 7410 UNITED 645 7653 UNITED 915 7093 DELTA 1045 7050 UNITED 645 7038 AMERICAN 645 7613 FRONTIER 2045 7204 SIMMONS AVIATION 715 8270 UNITED 945 7154 AMERICAN 2045 7024 AMERICAN 1045 7034 PIEDMONT 2045 7534 AMERICAN 1715 8537 AMERICAN 645 7116 UNITED 2015 7148 AMERICAN 645 7116 AMERICAN 645 7335 OZARK 745 8172 AMERICAN 645 7297 UNITED 645 7674 UNITED 645 7627 64
TABLE 7 (CONTINUED) UNITED UNITED SIMMONS AVIATION UNITED UNITED AMERICAN UNITED AMERICAN UNITED AMERICAN AMERICAN UNITED UNITED 1515 2045 715 915 645 915 645 645 2045 2045 2045 645 2045 7517 7340 8486 8209 7071 7020 7384 7245 7438 7060 7225 7673 7103 Only one of the lottery recipients at National and LaGuardia Midwest Express -- owned slots or operated flights at National as of June 30, 1990. At O'Hare, only America West and Air 88 Wisconsin remained as of June 30, 1990. All but four of the lottery slots changed hands within what would have been the two year "must use" periOd proposed in the SNPRM. Excluding these four slots, the average number of days a slot was held by the lottery recipient was 112 days at National, 89 118 days at LaGuardia, and 90 days at O'Hare. To the extent that small and new entrant carriers in the future will face the 88 Air Wisconsin is being purchased by United. Western Airlines was purchased by Delta and Ozark was purchased by TWA in 1986, and Air Cal was purchased by American in 1987. See Ogur, Wagner and Vita, supra note 8, Table 11-3, pp. 28-29. The remainder exited as scheduled air carriers. 89 This number does not include 41 lottery slots that were not picked at O'Hare. In addition, 19 of the slots at O'Hare were determined to have been returned or withdrawn by the FAA before the lottery recipient operated the slot. Not counting these 19 slots yield an average holding period of a little over five months (156 days). 65
same competitive pressures and economic incentives as they did following the 1986 lotteries, the data suggest that lottery slots allocated in future lotteries likely would be quickly traded, in the absence of rules prohibiting their transfer. In the absence of market power, the strength of a market system is its ability to direct resources to their highest valued uses. 90 Relying on nonmarket methods such as grandfathering and lotteries initially to allocate slots and prohibiting or delaying their subsequent transfer would sacrifice the benefits of an unrestricted slot market. 91 Because nonmarket mechanisms are unlikely to result in an efficient initial allocation of 92 slots, the presence of an unrestricted slot market with a short or nonexistent minimum holding period would permit initially misallocated slots be transferred to more highly valued uses. The FAA proposal would severely limit market reallocations by requiring the lottery participants to use lottery slots for two years before they could be transferred or sold freely. This prohibition on transfer, along with the stricter limits on primary participation in slot lotteries, will create, at least over a two year period, a user classification for new entrant and 90 See DOJ comments, supra note 12, p. 7. 91 Auctions would be a market based method of initial allocation. For a discussion of the legal issues that may preclude or severely restrict their use, see the DOJ comments, supra note 12, pp. 13-15. 92 See Comments of the DOJ, supra note 12, pp. 12-15. See also Koran and Ogur, supra note 8. 66
limited incumbent carriers (or small) airlines. If the lottery allocation has not resulted in the most efficient set of carriers receiving the lottery slots, the result could be to perpetuate an 93 initial misallocation of resources for up to two years. And given that it may be the larger, not the smaller, carriers who use slots most intensively, reallocation of slots to smaller carriers may induce reductions in slot utilization and consumer welfare. Whether the misallocation is significant would depend upon how many slots are affected. seeking slots without intending to use them might be deterred without the proposed two year "must use" period by barring from future slot lotteries a carrier who has sold its lottery slots too quickly. The proposed definition of a "limited incumbent" based on both current and past holdings is consistent with this approach. 94 If the costs of being excluded from future slot lotteries are large enough to cancel the gains from the slots obtained in the current lottery, the carrier will be 93 As in the case of the 90 percent "use or lose" rule, a carrier may choose to operate a flight for two years in order to obtain the valuable slot. This could result in much of the value of the slot being dissipated by the operation of the inefficient flight. See Posner, supra note 85. 94 See the discussion in note 32, supra. A smaller carrier's sale of lottery slots soon after the minimum holding period expired may simply mean its profits were lower than it expected; if so, it would be wrong to infer from the quick sale that the carrier wanted the slot solely in order to sell it for a profit. still, the costs imposed by that erroneous inference may not be large as long as slots remain transferable. If the smaller carrier would experience higher costs and lower profitability permanently, its future entry is unlikely anyway. If the carrier corrected its cost disadvantage, it could purchase a slot from lottery winner or an existing slot carrier. 67
,r ;. deterred from entering the current lottery for what the FAA finds to be undesirable purposes. And if the implicit penalties from lottery "debarment" at a single HDTA are not large enough, the carrier-specific prohibition could be extended, for example, to include other HDTAs, or slot lotteries at future HDTAs. such a system, coupled with the proposed selective screening of "new entrants," could reduce unwanted behavior and preserve favorable treatment for qualified small carriers with less potential for perpetuating an initial misallocation of resources. v. Conclusion Historical slot use at the four HDTAs is not consistent with the two anticompetitive theories studied. The data do not suggest that slots were being used at low rates, nor do the data suggest that carriers with larger shares of available slots used their slots at lower rates than carriers with smaller shares. Thus, the data do not present compelling reasons to believe that changing the existing slot allocation and transfer rules in the manner proposed would be necessary to increase slot usage rates to high levels. The proposed regulations and our comment address only slot usage, and not other aspects of airline competition. The proposed changes to the FAA's "use or lose" rules would require a rate of use approximately equal to the average rate of use already voluntarily chosen by most carriers. The rule is likely to affect a carrier with fewer slots more than a larger carrier with many slots. The proposed restrictions on lottery 68
participation and lottery slot sales may prevent slots from moving to their highest valued uses, and thus could reduce welfare. 69
Continue your research in ChatGPT or Claude
Connect Omnilex to search the legal corpus from your AI assistant.