Resources, Community, and Economic Development Division B-238198 July 11,199O The Honorable John C. Danforth Ranking Minority Member Committee on Commerce, Science, and Transportation United States Senate The Honorable James L. Oberstar Chairman, Subcommittee on Aviation Committee on Public Works and Transportation House of Representatives This report responds to your request that we examine the effects of increased concentration on air fares and service at major airports around the country. We examined fares and service levels at 15 airports where one or two airlines handle most of the enplaning passengers. We compared fare levels at the concentrated airports with fares at 38 relatively unconcentrated airports. As arranged with your offices, unless you publicly announce its contents earlier, we plan no further distribution of this report until 30 days after the date of this letter. At that time, we will send copies to the Secretary of Transportation and to other interested parties. This work was performed under the general direction of Kenneth M. Mead, Director for Transportation Issues. Mr. Mead can be reached at (202) 275-1000. Major contributors are listed in appendix III. V J. Dexter Peach Assistant Comptroller General , libceeutiveSummary Airline deregulation was predicated on the belief that the industry was Purpose fundamentally competitive and that competition would ensure low prices and good service. However, allegations of high fares and service reductions, especially in cities where one or two airlines handle most of the traffic, have triggered congressional concern about the state of com- petition in many air travel markets. The Ranking Minority Member of the Senate Committee on Commerce, Science, and Transportation and the Chairman of the Subcommittee on Aviation, House Committee on Public Works and Transportation, asked GAOto examine trends in airline fares and service. GAOexamined airline yields-the fare per passenger mile-at 53 of the 75 busiest airports in the nation. Fifteen of the 53 airports have rela- tively high levels of market concentration in that one or two airlines handle most of the enplanements. Yields at these concentrated airports were compared with those at the 38 remaining airports, where there was more competition. GAOalso examined trends in departure frequen- cies and points served from the 15 concentrated airports1 GAOtestified on June 7,1989, on its preliminary findings on changes in fares and service at concentrated airports between 1985 and 1988. This report extends the analysis and now includes fares paid by almost 45 million travelers between 1985 and the second quarter 1989. This report also compares GAO'Sresults with those from studies by the Departments of Justice and Transportation and others. This report on major airports is one of a series of GAOreviews on compe- tition in the nation’s airline industry. Complementary work analyzes fares at airports serving small and medium-sized communities and examines how changes in the airline industry have affected the ability of new firms to enter the industry or of existing carriers to enter new markets. Over the past few years, a trend has developed toward the establish- Background ment of a dominant position by one or two airlines at a growing number of major airports. Airport dominance can result from an airline merger, or it might follow an airline’s decision to set up a hub at the airport. A hub airport is one where an airline consolidates and interchanges traffic from many other points in its system. A carrier that gains dominance lGA0 considered an airport market concentrated if one airline handled 60 percent of the enplane- ments or if two airlines handled 86 percent. Page 2 GAO/WED-99-102 Fares and Service at Major Airports Executive Summary over traffic at an airport can gain a significant advantage over its com- petitors on individual routes. At the same time that the major airlines were beginning to establish dominant positions at a number of airports around the nation, they also developed new operating and marketing strategies that can make it dif- ficult for new firms to enter the airline industry or for existing carriers to expand into markets controlled by other airlines. Increased market power can lead to higher prices or reduced service as airlines try to maximize their profits. In June 1989 testimony, GAO reported that fares, on the average, rose Results in Brief more at the concentrated airports than at the airports where there was more competition and were about 27 percent higher than at the airports with more competition. Extending the analysis to include the first two quarters of 1989 showed that fares remained close to 27 percent higher. Fares charged by the dominant carriers tended to rise as their airport market shares increased. Recent analyses of fares at concentrated air- ports by Justice Department staff and others have also found that fares are higher at concentrated airports. Service levels at the concentrated airports generally improved: there was more direct, or single plane, service to more places after the airport became concentrated. However, the number of routes where there was competition between carriers declined at most of the concentrated air- ports, and more routes were served by only one airline. Principal Findings Fares Have Risen In 1988, airline yields for all carriers at the concentrated airports aver- aged about 27 percent higher than yields at the relatively unconcen- trated airports. This difference persisted through the first two quarters of 1989. Yields were higher at the concentrated airports in 1985 as well, but the difference has widened as the dominant airlines have increased their share of the traffic at these airports. In 1988, the average yield earned by the dominant airlines at the concentrated airports was 20 cents per passenger mile, almost 38 percent higher than the average yield for all carriers at the unconcentrated airports. Page 3 GAO/RCED-90-102 Fares and Service at Major Airports Executive Summary The yields received by the dominant airlines are generally above the yields received by other carriers at the concentrated airports. GAOfound that yields rose following the establishment of the dominant positions. These dominant positions were established by mergers of competing car- riers, the establishment of hubs, and the extension of already dominant positions. The airlines have offered a number of explanations for the higher yields at concentrated airports, including shorter average trips out of hub air- ports? and better quality service (i.e., more service to more points that does not require a change of planes). GAOexamined these explanations and, after taking into account differences in average flight lengths, found that yields were still more than 20 percent higher at the concen- trated airports. Service Levels Have GAOexamined several measures of the level of service at the 15 concen- trated airports to see whether travelers at those cities received more or Improved but Competition less service as airport dominance grew. The number of destinations Has Declined served directly from concentrated airports, that is, with single plane ser- vice, increased 10 percent, and the number of daily departures increased 3 percent between May 1985 and May 1988. These increases are prima- rily comprised of large increases in service by the dominant carriers as they established or strengthened their hubs. At airports that were affected by mergers, the number of daily departures often declined. At the same time, competition has lessened at the concentrated airports. The number of destinations served directly by only one airline rose 25 percent, while the number of destinations served by four or more air- lines fell 52 percent. Other Analyses AlSo show A study comparing fares at hub and nonhub airports undertaken for the Higher Fares at Air Transport Association (ATA) included 14 of the 15 concentrated air- ports in this study. While the ATA study did not address the question of Concentrated A irports how market concentration affects air fares in the same way as GAO,ATA'S data showed that in 13 of the 14 concentrated hub airports GAO examined fares were above the industry average. The ATA study attrib- uted the higher fares to factors other than concentration, including the public’s willingness to pay for higher levels of service-i.e., more fre- quent, nonstop flights to more destinations. “Shorter trips spread fixed costs over fewer miles and so fares per mile are generally higher for shorter flights than longer ones. Page 4 GAO/RCED-90-102 Fares and Service at Major Airports ExecutiveSummary Analysts at the Justice Department recently examined airline fares at two airports that became concentrated following mergers of competing airlines and found that airline fares had risen significantly as a result of the mergers. Finally, the Department of Transportation recently pub- lished a multivolume study of competition in the airline industry. The study concluded that, while most air travelers had benefited under deregulation through more service and lower fares, there were “pockets of problems” including higher fares for passengers traveling to and from some highly concentrated airports. DOTconcluded that no action was warranted at this time. GAOis not making any recommendations in this report. However, in tes- Recommendations timony before the Congress, GAOoutlined the pros and cons of various options that might be taken to promote competition in the airline industry (see app. II). GAOalso is finalizing work on an econometric model that will help the Congress and the Department of Transportation decide what specific actions are most likely to promote competition and preserve the benefits of deregulation for the consumer. A report synthe- sizing all of GAO'Swork on competition in the airline industry, including appropriate recommendations and matters for congressional considera- tion, is planned for issuance early next year. As agreed with the requesters, GAOdid not obtain formal agency com- Agency Comments ments on this report. However, GAOhas met with DOTofficials and they concurred with the finding that fares are higher at concentrated airports. Page 5 GAO/RCED-90-102 Fares and Service at Major Airports , Contents Executive Summary 2 Chapter 1 12 Introduction The Airline Industry Under Deregulation Growing Concern About the Future of Airline 12 13 Competition Objectives, Scope, and Methodology 14 Chapter 2 21 The Airline Industry CAB Regulation Competition Under Deregulation: the Early Years 197% 21 22 Under Deregulation 1984 Competition Since 1984: Changes in Marketing and 24 Operating Practices Other Barriers to Competition 30 Industry Mergers and Consolidation 30 Chapter 3 32 Trends in Airline Yields Are Higher at Concentrated Airports Trends in Yields at the Concentrated Airports 32 33 Fares at 15 Trends at Different Concentrated Airports 35 Concentrated Airports several Other Factors Could Affect Yield Differences 46 Sensitivity Analysis 49 Chapter 4 52 Other Recent Analyses The DDJ’St. Louis Study 52 The Simat, Helliesen & Eichner Study for the Air 55 of Fares at Transport Association Concentrated Airports Study by Justice Department Analysts 60 DOI’Task Force on Competition Study 61 Chapter 5 64 Changesin Air Number of Destinations Served Directly Number of Daily Departures 65 66 PassengerService at Degree of Competition 67 15 Concentrated Other Analyses of Service Changes at Concentrated 70 Airports Airports Page 0 GAO/RCED90-102 Farea and Service at war Airporta Contents Chapter 6 Conclusions Appendixes Appendix I: Trends in Fares at Each of the 15 78 Concentrated Airports Appendix II: Policy Options Discussed in GAO Testimony 94 on Barriers to Competition in the Airline Industry Appendix III: Major Contributors to This Report 98 Tables Table 1.1: Comparison of GAO and DOT/CAB Origin- 18 Destination Data Fare Screens for 1988 Data Table 3.1: Airline Yields in 1988 at Concentrated Major 33 Airports Table 3.2: Annual Average Yields and Enplanement 36 Shares of Dominant Carriers at Minneapolis/St. Paul and St. Louis Compared to Yields at 38 Unconcentrated Airports Table 3.3: USAir-Piedmont Fare Changes at Syracuse 37 Compared to Changes in Fares at 38 Unconcentrated Airports Table 3.4: Northwest-Republic Yields at Detroit 38 Metropolitan Wayne County Airport and at 38 Unconcentrated Airports Table 3.5: Annual Average Yields and Enplanement 39 Shares of American Airlines at Nashville and Raleigh-Durham Compared to Yields at 38 Unconcentrated Airports Table 3.6: Delta Yields and Market Shares at Cincinnati 40 Compared to Yields at 38 Comparison Airports Table 3.7: Yields and Enplanement Shares at Charlotte for 41 USAir-Piedmont and Yields at 38 Unconcentrated Airports Table 3.8: Yields and Enplanement Shares at Dayton for 41 USAir-Piedmont and Yields at 38 Unconcentrated Airports Table 3.9: Yields and Enplanement Shares at Piedmont 42 Triad International Airport for U&%-Piedmont and Yields at 38 Unconcentrated Airports Table 3.10: Yields and Enplanement Shares at Memphis 43 for Northwest-Republic and Yields at 38 Unconcentrated Airports Page 7 GAO/RCED-!M-102 Fares and Service at MaJor Airports . Table 3.11: Yields and Enplanement Shares at Salt Lake 43 City for Delta-Western and Yields at 38 Unconcentrated Airports Table 3.12: Yields and Enplanement Shares at Pittsburgh 44 for USAir-Piedmont and Yields at 38 Unconcentrated Airports Table 3.13: Yields at Atlanta for Delta and Eastern 45 Airlines and Yields at 38 Unconcentrated Airports Table 3.14: Yields at Denver for United and Continental 46 and Yields at 38 Unconcentrated Airports Table 3.15: Differences in Yields at 15 Concentrated 48 Airports and 22 Unconcentrated Airports Table 3.16: Sensitivity Analysis of Assumptions 51 Employed to Calculate Yields at Concentrated and Unconcentrated Airports Table 4.1: Comparison of Round-Trip Fare Changes by 54 Carriers Dominating Hubs Table 4.2: Comparison of GAO and SH&E Origin- 59 Destination Data Fare Screens Table 4.3: DOT-Calculated Fare Premiums on Monopoly 62 Routes and at Concentrated Airports in 1988 Table 4.4: DOT-Calculated Fare Premiums at 62 Concentrated Hubs, 1984 and 1988 Table 5.1: Number of US. Destinations With Direct 65 Service From 15 Concentrated Airports During Month of May 1985-1988 Table 5.2: Number of Daily Flights to U.S. Destinations 67 From 15 Concentrated Airports During Month of May 1985-1988 Table 5.3: Number of U.S. Destinations to Which Only One 68 Airline Flew Directly From 15 Concentrated Airports in Month of May 1985-88 Table 5.4: Number of U.S. Destinations to Which Two or 69 Three Airlines Flew Directly From 15 Concentrated Airports in Month of May 1985-88 Table 5.5: Number of U.S. Destinations to Which Four or 70 More Airlines Flew Directly From 15 Concentrated Airports in Month of May 1985-88 Table 5.6: Competition Among Carriers Providing Direct 71 Service From St. Louis Table 5.7: Service Changes at St. Louis for Large Jet 71 Carriers (Domestic) June 1986 vs. June 1988 Page 8 GAO/WED-90-102 Fares and Service at Major Airports Contents Table 5.8: Number of Nonstop Destinations From 15 73 Concentrated Airports During Month of May 1985- 1988 Figures Figure 3.1: Average Yield for 12 Airports Where 35 Concentration Increased Figure 3.2: Average Yield for 22 and 38 Comparison 47 Airports Figure I. 1: Atlanta (Hartsfield Atlanta International 78 Airport) Figure 1.2: Charlotte (Charlotte/Douglas International 79 Airport) Figure 1.3: Cincinnati (Greater Cincinnati International 80 Airport) Figure 1.4: Dayton (Dayton International Airport) 81 Figure 1.5: Denver (Stapleton International Airport) 82 Figure 1.6: Detroit (Detroit Metropolitan Wayne County 83 Airport) Figure 1.7: Greensboro/High Point/Winston-Salem 84 (Piedmont Triad International Airport) Figure 1.8: Memphis (Memphis International Airport) 85 Figure 1.9: Minneapolis/St. Paul (Minneapolis/St. Paul 86 International Airport) Figure I. 10: Nashville (Nashville Metropolitan Airport) 87 Figure I. 11: Pittsburgh (Greater Pittsburgh International 88 Airport) Figure 1.12: Raleigh-Durham (Raleigh-Durham Airport) 89 Figure 1.13: St. Louis (Lambert-St. Louis International 90 Airport) Figure I. 14: Salt Lake City (Salt Lake City International 91 Airport) Figure 1.15: Syracuse (Hancock International Airport) 92 Figure 1.16: Average Yield for the 15 Concentrated 93 Airports Page 9 GAO/RCED-9@102 Fares and Service at Major Airports Contents Abbreviations ATA Air Transport Association CAB Civil Aeronautics Board CPI Consumer Price Index CRS Computerized Reservation System DOJ Department of Justice DOT Department of Transportation GAO General Accounting Office HHI Herfindahl-Hirschman Index O&D Origin and Destination SH&E Simat, Helliesen & Eichner, Inc. TACOS Travel agent commission overrides TWA Trans World Airlines Page 10 GAO/RCE!D90-102 Fares and Service at Major Airports Page 11 GAO/RCElMO-102 Fares and Service at Major Airports Chapter 1 , htroduetion For over 40 years the nation’s airline industry was subject to eco- nomic regulation by the Civil Aeronautics Board. Many economists and airline industry experts believed that government control over fares and service was inappropriate for this industry, and in 1978 the Congress passed the Airline Deregulation Act (P.L. No. 95-504). Proponents of deregulation believed that the free market, not gov- ernment regulation, should determine who should provide airline service and at what price. They believed that the airline industry was a naturally competitive one that should be treated like any other industry in which competition could be expected to flourish. In the early years following deregulation, many new firms entered the The Airline Industry industry and the existing carriers expanded their operations into new Under Deregulation markets. Between 1978 and 1984, as the proponents of airline deregula- tion had forecast, service offerings expanded and competition intensi- fied. While real (inflation-adjusted) fares rose due to sharp increases in fuel prices, they began to fall after 1981.’However, over time, many of the new entrants and some of the older carriers went out of business or merged with other airlines. After a series of bankruptcies and mergers between 1985 and 1988, the national air travel market became even more concentrated2 than when the industry was regulated by the federal government. For example, in 1978 the five largest airlines controlled 69 percent of the nation’s air travel market. Following the increase in com- petition the share of the five largest airlines fell to about 57 percent in 1985, but by late 1988 it had risen to 74 percent.3 Since deregulation, the airlines have reconfigured their operations into hub and spoke networks. Under a hub and spoke system, airlines bring many flights from “spoke” cities into a central “hub” airport, interchange the traffic, and send the flights back out to their final desti- nations. Airlines using hub and spoke operations maintain a large pres- ence at each hub airport and often dominate traffic at the hub. Thus, while competition has increased on many airline routes, certain airports have experienced increases in concentration. ‘Some proponents of deregulation held that becauseentry into airline markets was relatively easy, the threat of potential entry by would-be competitors would be sufficient to keep air fares low even in markets dominated by one airline. Airline markets were believed to be highly “contestable.” %oncentration is the degree to which sales in an industry or market are accounted for by a small number of firms. 3This is not necessarily the most important measure. Three airlines all competing over every route in the nation might well be better than five airlines each with a regional monopoly. Many routes remain highly competitive and the industry as a whole remains more competitive than before deregulation. Page 12 GAO/RCED-90-102 Fares and Service at Major Airports Chapter 1 Introduction Growing Concern operations, many major airports around the nation are now dominated About the Future of by one or two airlines. Once established, airline dominance might be rel- Airline Competition atively secure because it can be difficult for other airlines to establish or to significantly expand operations at airports where another carrier is already dominant. Marketing strategies, such as frequent flyer pro- grams, airline-owned computerized reservation systems, and travel agent commission overrides4 can be used to reinforce the dominance of the incumbent airline by keeping out potential competitors. In addition, physical constraints, such as inadequate gate space to accommodate new entrants and noise restrictions on the type of equipment that car- riers can use to serve some airports, can also limit entry by potential competitors. Concern is growing that the major airlines might be taking advantage of their positions at airports where they are dominant by charging higher fares for air travel out of those airports or by cutting back on service levels. Airport dominance, combined with other recent changes in the ways the major airlines provide and market their services, might have anti-competitive impacts and frustrate the goals of airline deregulation. Such concern caused the Ranking Minority Member of the Senate Com- mittee on Commerce, Science, and Transportation to ask us to examine fares and service at Lambert-St. Louis International Airport before and after the merger of Trans World Airlines (TWA) and Ozark Air Lines.” Both TWA and Ozark Air Lines used St. Louis as their primary hub air- port. Before the merger, TWA handled about 56 percent of the enplane- ments at St. Louis, but after the merger TWA handled 82 percent.6 We found that TWA’S fares for flights out of St. Louis rose substantially fol- lowing the merger in comparison with fare changes elsewhere. We also found that the number of carriers competing for traffic at St. Louis declined. More routes were served by only a single carrier, usually TWA, and far fewer routes were served by four or more carriers. Our fare 4Travel agent commission overrides are payments by airlines to agents above and beyond their normal commissions for increasing their bookings on a carrier’s flights. “Airline Competition, Fare and Service Changesat St. Louis Since the TWA-Ozark Merger (GAO/ RCED-8%217BR, Sept. 1988). “Enplanements are passengerboardings at the airport, and include both originating and connecting traffic. Page 13 GAO/RCFB-90-102 Fares and Service at Major Airports Chapter 1 Introduction findings for St. Louis were later confirmed in an analysis by the Depart- ment of Transportation (DCZ).7 At the request of the Ranking Minority Member of the Senate Committee Objectives,Scope,and on Commerce, Science, and Transportation and the current Chairman of Methodology the Subcommittee on Aviation, House Committee on Public Works and Transportation, we extended our St. Louis analysis to include fares and service at 14 other concentrated airports around the nation (see fol- lowing list). The 15 Concentrated Atlanta Charlotte Airports Cincinnati Dayton Denver Detroit Greensboro/High Point/Winston-Salem Memphis Minneapolis/St. Paul Nashville Pittsburgh Raleigh-Durham St. Louis Salt Lake City Syracuse Our objective was to determine whether . fares at major airports where one or two carriers handled most of the traffic were above fares for travel at other major airports where there was more competition; m dominant airlines charged higher fares than other carriers serving the airport; and l service levels had changed at concentrated airports. We focused on the period since concentration began to increase, roughly since 1985. We selected concentrated airports for analysis from among 7A Comparison of Air Fares and Services at St. Louis Before and After Trans World Airlines Acquired Ozark Airlines (sic), U.S. Department of Transportation, Office of Economics (DOT-P-37-89- 3, Jan. 1989). Page 14 GAO/RCFB-90-102 Fares and Service at Major Airports Chapter 1 Introduction the 75 busiest on the basis of enplanements. Our criteria for deciding that an airport was concentrated were that one airline handled at least 60 percent of the passengers enplaning at that airport or two airlines handled at least 85 percent of the enplaning passengers. A total of 22 airports met the concentration criteria. We chose enplanement share as the criterion, but others are possible, including the proportion of originating passengers handled by one carrier. Airlines have a smaller proportion of originating traffic than enplanements at their hubs because of the relatively large volume of non-originating, connecting passengers. When we calculated enplanement shares, we grouped together airlines under common ownership, such as Eastern and Conti- nental or Piedmont and USAir. From the total number of concentrated airports, we excluded six air- ports that met the concentration criteria but were in metropolitan areas served by more than one major commercial airport. Therefore, airports in the New York City, Los Angeles, Chicago, Houston, Baltimore/Wash- ington, San Francisco, and Dallas areas were not candidates even though airports in some of these cities met the concentration criteria. We elimi- nated airports in multi-airport cities because competition from carriers serving the other airport might offset, to some extent, the effects of con- centration. We also excluded one concentrated airport because it was outside the 48 contiguous states.8 All but one of the 15 airports we selected are hubs for one or more of the major airlines.g Some airports, such as Phoenix, are hubs, but are not concentrated by our criteria. We contrasted trends in yields (the fare per passenger mile) on routes from the 15 concentrated airports with yields on routes from a compar- ison group of 38 relatively unconcentrated airports listed below.LoThe airports used for comparison are those in the top 75 airports in the 48 contiguous states that did not meet our definition of concentration and were not in multi-airport cities. l1 We also compared the yields received by the dominant airline at each concentrated airport with the yields %ecause of their unusual geographic characteristics, we excluded airports outside the 48 states from both the concentrated and comparison airports in our study. ‘USAir, which acquired Piedmont Airlines, has assumedthe dominant position at Greensboro,N.C., but does not operate a hub there. “Passenger miles are the straight-line distances between the origin and destination, regardless of the route taken by the individual airlines. “Some of the airports in our control group are hubs (e.g., Phoenix), but they are not concentrated by our definition. Page 15 GAO/RCED-90-102 Fares and Service at Major Airports Chapter 1 Introduction earned by the other airlines serving the airport in order to further understand the effects of dominance. We examined trends in airline yields from the first quarter 1985 through the second quarter 1989, the most recent quarter for which fare data were available at the time this study was completed. Our analysis covers fares paid by almost 45 million travelers between 1985 and the second quarter 1989. The 38 &woncmt,r __-________ lated Albuquerque Austin Airports i n the Birmingham Comparison --- Group 1 Boston Buffalo Cleveland Columbus, OH El Paso Pt. Lauderdale Ft. Myers Hartford Indianapolis Jacksonville Kansas City Las Vegas Little Rock Louisville Miami Milwaukee New Orleans Norfolk/Virginia Beach Oklahoma City Omaha Orlando Philadelphia Phoenix Portland, OR Reno Richmond Rochester, NY Sacramento San Antonio San Diego Page 16 GAO/‘RCED-90-102Fares and Service at Major Airports Chapter 1 Introduction Seattle Tampa Tucson Tulsa West Palm Beach Because we are concerned with fares paid by travelers from the cities served by a concentrated airport, all of the yields calculated in this anal- ysis apply to traffic originating at such airports. To ensure that our analysis of trends in yields reflected changes in fares, as opposed to changes in the composition of the sample, we controlled for changes in the distribution of destinations and changes in the proportion of one- way and round-trip fares in the sample. For each combination of trip types (one-way or round-trip) and destination, we calculated the average yield for each quarter. We weighted the average yield for each combination according to the average amount of traffic for that combi- nation over the 18 quarters. In order to improve comparability between the concentrated airports and the comparison group, we examined a subset of 22 unconcentrated airports that excluded airports where average trip lengths were greater than 900 miles. Yields tend to be lower for longer flights because fares increase less than proportionately with mileage flown and, on average, the 38 airports in the comparison group had longer average trips than the 15 concentrated airports. In addition, we compared yields between the 15 concentrated and 38 unconcentrated airports for routes within each of several distance categories. We used the Origin and Destination (O&D) Survey data collected quar- terly by DOT in its 10 percent sample of airline tickets to make our yield comparisons. The airlines report detailed information on every tenth ticket to DOT and, after processing the data, DOT makes the data available for public use.‘” “All large, certificated route air carriers and their code sharing partners are required to submit O&D Survey data. Thus, only the smallest of domestic airlines offering scheduled service are exempt. Page 17 GAO/RCED-90-102 Fares aud Service at Major Airports . Chapter 1 Introduction Unfortunately, there are a variety of reporting errors in the O&D data. In particular, fares are occasionally misreported or miscoded.13As part of this study, we developed a new fare screen to eliminate fares that are too high or too low. I4Our fare screen was developed by examining avail- able fares in the Official Airline Guide and by discussing fares with industry experts. DOT is currently applying the high end of our fare screen to the latest submissions to identify any fares that are outside credible limits.15 Table 1.1 contrasts our new fare screen with D@S orig- inal screen. Based on an examination of listed fares, we developed a sep- arate fare screen for each year. Table 1 .I: Comparison of GAO and DOT/CAB Origin-Destination Data Fare Screens for 1988 Data DOT/CAB screen GAO screen Exclude if yield is Exclude if yield is less than greater than less than greater than Mileage category cents/mile cents/mile cents/mile cents/mile l-100 10.00 177.18 8 300 101-200 5.00 77.63 4 255 201-300 3.33 56.99 3 160 301-400 500 48.12 3 125 401-500 6.00 43.62 3 115 501-700 4.28 38.28 3 105 701-1000 5.00 32.93 3 80 1001-1300 4.61 29.67 3 65 1301-1600 500 27.75 3 55 1601-1900 430 26.40 3 50 1901-2200 4.54 25.34 3 40 2201-2500 4.40 24.63 3 40 above2500 4.28 2351 3 40 To determine whether the differences in yields were statistically signifi- cant, we contrasted the average yield at each concentrated airport with 13Forinternal uses, DOT had adopted a fare screen developed by the Civil Aeronautics Board to eliminate fares that were obviously too high or too low. However, the screen had not been adjusted for many years. As a result, over time many valid fares were being excluded. The Board's fare screen was used to develop SUMDOM, an internal data base. The data made available to the public and to data vendors are Data Bank lA, which does not screen out incorrect fares. Users can make their own adjustments. The criteria we developed more accurately screen the O&D Survey data. ‘“The unedited data sometimes have included recorded fares in the hundreds of thousands of dollars and as low as a few dollars when no such fares existed. ‘“DOT has elected to include very low reported fares, including the $0 fares paid by frequent flyers. Our focus is on fares actually paid by travelers, and so we exclude the $0 fares. Page 18 GAO/WED-90-102 Fares and Service at Major Airports Chapter 1 Introduction the average yield at the unconcentrated airports in 1988 and tested the difference between yields at the concentrated airports in 1985 and 1988 to see if they were significant. All the differences were statistically sig- nificant at the .OOl level except for the difference between the yield at Detroit and the unconcentrated airports in 1988; that is, there is only 1 chance in 1000 that there is no difference in yields. Other organizations, including D(JT, also have recently attempted to measure whether fares are higher at airports where one or two carriers dominate the traffic or where the airlines operate their hubs. Some of these studies have been undertaken in response to our analysis. We have reviewed the assumptions and methodologies underlying these alterna- tive approaches and, where possible, have attempted to reconcile these other findings with our own. Increased market power can lead to reduced levels of service as well as higher fares. Airlines can increase profits by cutting back on the number of flights or replacing direct service with connecting service. To test whether service levels had declined at the 15 concentrated airports, we examined service level data for the month of May of each year between 1985 and 1988. We compared the number of cities that could be reached by direct service, the total number of daily flights to all places, and the amount of competition as measured by the number of markets served by one carrier, by two or three carriers, or by four or more carriers. We did not find that service levels were reduced as concentration increased, but we did find that there were fewer routes with four or more carriers and more routes served by only one airline.‘” Finally, this report is part of a series of GAO reports on the state of com- petition in the airline industry. Prior studies examined how DOT fulfilled its role in overseeing airline mergers and what happened to airline fares and service at Lambert-St. Louis International Airport following the acquisition of Ozark Air Lines by TWA. I7The present report extends the analysis of St. Louis to all of the large concentrated airports in the 48 contiguous states that are not in multi-airport cities. ‘“The source of these air service data was the automated version of the Official AirIine Guide, which was purchased from an airline data vendor, I.P. Sharp, Inc. 17Airiine Competition: DOI% Implementation of Airline Regulatory Authority (GAO/RCED-89-93, June 1989); Airline Competition: Fare and Service Changesat St. Louis Since the TWA-Ozark Merger (GAO/RCED-%-217BR, Sept. 1988). Page 19 GAO/RCED-90-102 Fares and Service at Major Airports Chapter 1 introduction The present study does not attempt to measure precisely how market concentration and the various factors that protect dominant positions have affected airlines fares. We are currently developing an econometric model of the airline industry that will establish the linkage between fares and market conditions. In addition, the present study is concerned only with relatively large airports. We are also preparing two studies of fare trends at airports serving small and medium-sized cities following deregulation and since the industry has become more concentrated. Our review was conducted between September 1988 and November 1989. During that period we testified four times before the Congress on airline fares and service.18 IsFactors Affecting Concentration in the Airline Indust&GAO/T-RCED-88-65, Sept. 1988j; & Fares and Service at Concentrated Airports (GAO/T-RaD-89-37, June 1989); Barriers to Competi- tion in the Airline Industry (GAO/T-RCED-89-65, Sept. 20,1989); and Barriers to Competition in the Airline Industry (GAO/T-RCED-89-66, Sept. 21,1989). Page 20 GAOjRCED90402 Fares and Service at Major Airports Chapter 2 The Airline Industry Under Deregulation After 40 years of regulation by the Civil Aeronautics Board (CAB), in 1978 the Congress enacted the Airline Deregulation Act, which phased out economic regulation of the industry. In the early years of deregulation, many new firms entered the airline industry and, as expected by many airline industry analysts, service offerings expanded and fares fell. However, many other changes in the airline industry since 1978, unforeseen by the proponents of deregulation, have had a significant impact on the competitive environment. Most of the established carriers adapted to the new environment and developed operating and marketing strategies that made it difficult for the new airlines to successfully compete-and many failed. A wave of mergers and consolidations in the mid-1980s reinforced the trend toward a more concentrated airline industry. In 1938, the Congress chose to regulate the airline industry because of CAB Regulation concerns over safety, the airlines’ financial health, and perceived inequi- ties between airlines and other forms of transportation-since the other forms of transportation were regulated, while the airlines were not. Also, at the time economic regulation was imposed, many air carriers were near bankruptcy and service was unreliable. The Civil Aeronautics Act of 1938 (P.L. No. 75-706) created the Civil Aeronautics Authority (predecessor to CAB) and gave it authority over fares and market entry and exit similar in some ways to the Interstate Commerce Commission’s authority over railroads and motor carriers. CABcontrolled who could enter the industry and determined which air- lines could serve which individual city-pair markets. Airlines could neither add nor abandon routes without CABapproval. Furthermore, CAB refused to grant operating authority to any new trunk (major) airlines.’ Over time, the number of trunk carriers declined from 16 in 1938 to 10 by 1974. However, no trunk airline was allowed to go out of business entirely; instead, failing carriers were merged with healthier ones. CABalso tightly controlled fares. If a carrier wanted to raise or lower its fares, it had to file a tariff in advance with CAB.The Board could hold a hearing on the proposed fare change on its own initiative or upon the ‘CAB classified the airlines by the type of service provided. Tnmk airlines were those that had per- manent operating rights between major population areas. Local service airlines were created in the 1940s when CAB certified 19 carriers to provide service between smaller population centers and major airports. At the time the industry was deregulated, the number of local service airlines had shrunk to 8. Two of the former local service airlines, USAii and Piedmont, later became major carriers. Page 21 GAO/WED-90-102 Fares and Service at Major Airports Chapter 2 The Airline Industry Under Deregulation complaint by any person. Competitors could oppose the change or use the required waiting period to file a matching tariff. CABtried to set fares so that the airlines earned sufficient revenues to cover expenses and achieved a rate of return that CABbelieved necessary for financial viability. By the mid-1970s it was widely accepted that CABregulation had not been successful. The airlines rarely earned CAB’Starget rates of return. Moreover, because the airlines were precluded by CABfrom com- peting on the basis of fares, they often substituted service competition for price competition. This substitution increased airline operating costs. These higher costs could then be used to justify higher fares.2 Many economists argued that economic regulation of the airlines was inappropriate. The airline industry appeared to be inherently competi- tive and exhibited none of the characteristics of an industry that nor- mally required regulation.3 Evidence from unregulated intrastate air travel markets in California and Texas supported the argument that fares would fall, carriers could prosper, and safety could be maintained if the industry were deregulated. The Airline Deregulation Act of 1978 phased out federal control over Competition Under airline fares and routes. The Act allowed new airlines to form and made Deregulation: the it easier for existing carriers to expand operations into new markets and Early Years 1978-1984 deregulation aban don old ones. Economic theory suggested that, in the long run, would lead to increased competition. Greater competition would lead to lower fares, more service, and a wider variety of service offerings. More competition would also force airlines to become more efficient and reduce operating costs. Deregulation’s proponents believed that, in the absence of government regulation, airline competition would flourish. In those cases where only one airline served a particular market, the carrier would not be able to take advantage of its monopoly position because the principal form of capital in the airline industry (i.e., the airplanes) is highly mobile. Any 21nthe mid-1970s CAB adopted standards that made it more difficult for airlines to use such costs to justify fare increases. 31ndustriesin which competition is not expected to be feasible are sometimes called natural monopo- lies. An industry is a natural monopoly when the minimum average cost of production occurs at a rate of output generally sufficient to supply the entire market. If two firms split the market, each would be smaller than its optimally efficient size and each would have relatively high costs and an incentive to expand output. If both lower prices to sell more, price will generally fall faster than average cost becausea large portion of production costs in these industries is fixed, and competition becomesruinous. Ultimately, only one firm can survive in such a market. Virtually all public utilities are natural monopolies. Page 22 GAO/RCEDIO-102 Fares and Service at Major Airpor& Chapter 2 The Airline Industry Under Deregulation attempt by an airline with a monopoly in a particular market to raise fares and earn excessive profits would be short-lived because other car- riers would quickly enter the market and, by their competition, drive down the fares. Because every airline understood how its potential com- petitors would behave, the threat of entry by potential competitors was believed to be sufficient to discipline prices even in markets where only one airline offered service. The expectations of greater competition, more service, and lower fares were largely fulfilled during the first several years following airline deregulation. Between 1978 and 1984, the number of certificated air- lines almost tripled, from 44 to 114. Although the former trunk airlines still dominated the industry, their share of the traffic contracted while the share of the smaller airlines, including the new entrants, increased. Not only were there more carriers, but there were more carriers in more markets. Routes served by two or more airlines increased by 55 percent between 1978 and 1984, while those served by only one airline fell almost 10 percent. Fares also fell for most, although not for all, passengers. In constant dol- lars, the average fare fell 6 percent between 1978 and 1984 despite increases in airline operating costs that were higher than the general rate of inflation. By offering lower fares, new airlines forced the estab- lished carriers to offer substantial discounts. The proportion of travel made on discount fares increased from 39 percent in 1977 to 81 percent in 1984. Moreover, the discounts were deeper, increasing from 30 per- cent below full fare in 1977 to 51 percent below full fare in 1984. Fares, too, were more closely related to costs as carriers more fully incorpo- rated the distance taper into their fares4 and thereby reduced many of the cross-subsidies that had prevailed under CABregulation.” On the other hand, the airline industry on the whole did not perform very well financially in the early years of deregulation. From 1979 through 1984, the trunk and local service airlines lost $4 billion. These 4CAB largely set fares on the basis of distance. However, many airline operating costs do not vary directly with distance but with the number of takeoffs and landings. Other costs are periodic and do not vary at all with the distance flown. As a result, longer distance flights have lower costs per seat mile than shorter distance flights. The CAB fare formula did not adequately account for the distance taper. ‘The CAB fare formula subsidized travelers in short-distance, lightly traveled markets (who were charged a price that did not cover costs) at the expenseof travelers in long-distance, heavily traveled markets (who were charged a fare exceeding costs). Page 23 GAO/RCED-90-102 Fares and Service at Major Airports Chapter 2 The Airline Industry Under Deregulation losses can be traced to other factors as well as to deregulationP Eco- nomic recessions in 1980 and 1981-82 decreased the demand for air travel. The air traffic controllers’ strike and subsequent dismissal of much of the existing controller work force led to restrictions on the number of operations. Also, a 90 percent, constant-dollar increase in fuel costs between 1978 and 1981 led the airlines to raise fares. These fare increases, too, dampened traffic. Higher fares and negative economic growth combined to reduce traffic in 1980 and 1981. The industry had not experienced two successive years of negative traffic growth since World War II. Nevertheless, deregulation also played a role in the poor financial per- formance of the airline industry during the early years of deregulation, as intense fare competition reduced airline profits. The trunk airlines were most affected as their highly profitable routes attracted competi- tors and fare wars broke out. After the recession, lower fares stimulated traffic increases. Similarly, deregulation stimulated productivity improvements, but operating costs rose faster than productivity growth. Passenger miles per employee for scheduled airlines rose by more than 28 percent between 1978 and 1984 while seat miles per employee grew 33 percent. At the same time operating expenses increased nearly 94 percent. Airline deregulation, in combination with the dynamic environment in Competition Since which it occurred, led to an industry shake-out and to a very different 1984: Changesin airline industry from what had prevailed under CAB regulation. Three major developments emerged: Marketing and Operating Practices . the surviving major airlines reconfigured their route systems from linear systems to hub and spoke networks; the major carriers adopted a number of marketing practices that made it more difficult for potential competitors to challenge them in markets where they were dominant; and many new entrants and some of the original trunk carriers went out of business or merged with stronger airlines. “A number of studies have traced the early history of airline deregulation Among the notable are Policies for the Deregulated Airline Industry, CongressionalBudget Office (Washington, D.C.: July 1988); Deregulation: Increased Competition is Making Airlines More Efficient and Responsiveto Con- sumers (GAO/RCED-86-26, Nov. 1985); The Deregulated Airline Industry: A Review of the Evidence, Federal Trade Commission (Washington, D.C.: Jan. 1988); Steven Morrison and Clifford Winston, The Economic Effects of Airline Deregulation, The Brookings Institution (Washington, DC.: 1986). - Page 24 GAO/RCED-90-102 Fares and Service at Major Airports . Chapter 2 The Airline Industry Under Deregulation Establ .ishment of Hub and Deregulation not only gave the airlines freedom over fares, it also allowed them to enter and exit routes without obtaining prior CAB Spoke Systems approval. This new flexibility allowed the airlines to reconfigure their route systems into hub and spoke networks. Hub and spoke operations concentrate most of an airline’s operations at one or a very few “hub” airports and connect virtually every other airport (spoke) in the car- rier’s system via nonstop service to the hub. The hubbing airline will schedule flights so as to bring in travelers from many spokes to the hub, transfer the passengers among planes, and send them off to their final destinations, all in a relatively short period of time. This whole process is repeated several times each day. This system has produced substantial benefits for both the airlines and the traveling public. While there are not sufficient passengers in most airport-pair markets to justify multiple daily nonstop flights, by com- bining passengers bound for many different places and flying them to a hub where they can transfer to flights to their desired destinations, the airline can effectively offer numerous daily one-stop flights in many air- port-pair markets. Hub operations make it easier for many travelers to secure flights departing and arriving at times that best match their pre- ferred departure and arrival times.7 Hub and spoke systems also allow the airlines to better use their airplanes. However, hub and spoke systems require the hubbing airline to handle many simultaneous departures and arrivals several times a day. Because of its numerous departures and arrivals, the hubbing carrier will control many of the gates or concourses. The hubbing carrier may even have exclusive-use rights to its terminal. Given the overall limits on an airport’s capacity to handle traffic and the size of the local originating air travel market, hubbing will often result in one carrier handling most of the enplanements at the airport where it has its hub- bing operations. Often an airport will expand its capacity in order to accommodate a car- rier that decides to set up a hub there. The carrier and the airport will typically enter into a long-term lease agreement for space at the facility. The revenues from the lease payments will be used to underwrite the airport bonds sold to pay for the capacity expansion and thereby lower the cost of borrowing. As a quid pro quo, the airline may require the ‘The difference between the traveler’s preferred time and the scheduled time of departure is called “frequency delay.” Travelers value minimkiig delays of any type, so increased frequencies have a positive impact on traveler welfare. Page 25 GAO/RCEDSO-102 Fares and Service at Major Airports Chapter 2 The Airline Industry Under Deregulation airport to include a majority-in-interest clause in the lease agreement, giving the airline a large say in any future airport construction activities that would affect its lease payments. By establishing a hub at a city, the airline can gain recognition as that city’s airline. Since the hubbing airline offers so much service out of the airport, travelers living in the city served by that airline might think of it first when planning to fly. The airlines, however, have gone beyond relying simply on market identification for securing the local traffic base at the hub airport. They have also adopted a number of sophisti- cated marketing techniques that can deter new entrants from chal- lenging the dominant carriers at their hubs. Frequent Flyer Programs Frequent flyer programs were designed to create brand loyalty, and it appears that they have been successful. Frequent flyer programs factor prominently in determining a traveler’s choice of airline, In our recent survey of 32 travel agents, three-fourths told us that their business travel customers choose their flights on the basis of their membership in frequent flyer programs more than half the time. Some frequent flyer programs are designed so that the awards increase in value as higher mileage thresholds are achieved. Because awards are paid only after thresholds are met, the traveler who has collected some, but not all, of the mileage needed to reach the desired award is unlikely to switch to another carrier. In addition, most of the major airlines’ plans set dead- lines for accumulating mileage to earn awards, so that a traveler can only reach the higher awards levels if mileage is earned quickly. Thus, members of frequent flyer programs will concentrate their air travel on a single carrier, and they will tend to prefer the carrier that flies to the most destinations and to the greatest variety of business and vacation destinations from the traveler’s city of origin. The dominant carrier at the airport, especially if it is the hub carrier, will likely be the one that offers the greatest number and variety of destinations and, therefore, offers the best opportunity to earn and to benefit from frequent flyer awards. Frequent flyer programs, therefore, can discourage potential competi- tors from challenging an incumbent airline at an airport where it is dom- inant. Airline passengers might not respond to the lower fares offered by a new entrant unless they are low enough to compensate for the loss of expected benefits from earning frequent flyer mileage. Indeed, because frequent flyers are often business travelers whose fares are paid by their employers, they lack incentive to switch to a new, low-fare Page 26 GAO/RCED-90-102 Fares and Service at Major Airports Chapter 2 The Airline Industry Under Deregulation carrier and have considerable incentive to stay with the incumbent. W ith respect to frequent flyer programs, a recent DOT report concluded that frequent flyer programs stabilize and protect the existing market shares of incumbent airlines and make it more difficult for smaller air carriers to compete successfully in some markets.8 Computerized Reservation More than 80 percent of all air travel today is booked through travel agents, and 95 percent of all travel agencies use at least one of the five Systems airline-owned computerized reservation systems (CFSS) to book flights? Initially, the airlines that marketed their systems to travel agents used the CRSS to gain an advantage over their competitors by biasing the com- puter screen display so that their flights would be listed first. CABfound this practice to be anti-competitive and prohibited it in 1984. Neverthe- less, there is strong evidence that anti-competitive impacts continue.*O CRSScan continue to have anti-competitive impacts in two ways. First, although the systems no longer bias the screen displays to favor the flights of the cRs-owning airline, the ens owners continue to get a dispro- portionate share of bookings from agents using their systems. Second, travel agents continue to favor the airline that owns the CRS the agent uses because the cas-owning airline maintains supportive business rela- tionships with its network of travel agent subscribers-the so-called “halo effect.” These bookings for cas owners are revenues lost to the airlines that do not own CR%. While available data do not allow us to identify the traffic and revenue impacts in different airport markets, the CRS-OWDiDg carriers attempt to sign up most of the agents in the area served by their hub airport. Thus, sAlrllne Marketing Practices: Travel Agencies, Frequent-Flyer Programs, and Computer Reservation Systems, U.S. Department of Transportation, Secretary’s Task Force on Competition in the U.S. Domestic Airline Industry (Washington, D.C.: Feb. 1990). “The five CRSsare SABRE, owned by American Airlines, with a 43 percent share of revenues; Apollo, owned by a consortium of airlines but principally by United Airlines and USAir, with a 32 percent market share; PARS, owned by TWA and Northwest Airlines, with a 10 percent share; System One, owned by Texas Air Corp., with a market share of 10 percent; and DATAS II, owned by Delta Airlines, with a market share of 5 percent. “‘In our 1986 report, Airline Competition: Impact of Computerized Reservation Systems (GAO/ RCED-86-74, May 1986), we found that it was likely that anti-competitive impacts continued to be a problem, and we recommendedthat DOT study the issue. In May 1988, DCJIissued its report, Study of Airline Computer Reservation Systems (DOT-P-37-88-2,May 1988), which, while presenting consider- able evidence that CRS owners were earning excessive profits, drew no conclusions. In September 1988, we testified before the Subcommittee on Aviation, HouseCommittee on Public Works and Transportation, on the DOT report and presented policy options for congressionalconsideration. Page 27 GAO/RCED90-102 Fares and Service at Major Airports Chapter 2 The Airline Industry Under Deregulation in St. Louis, where TWA has its hub, TWA’S PARS system controls 77 per- cent of the CRS market; in Dallas, where American Airlines has a hub, its SABRE system has a 91 percent share; and in Denver, where United Air- lines operates a hub, United’s Apollo system has a 76 percent market share. An airline that sought to introduce competing service from these points would find that the local travel agents usually subscribe to the dominant airline’s CRS and exhibit a preference for the flights of the CRS- owning dominant carrier. Not only might potential new entrants find that the available market is smaller than expected because of the “halo effect,” they might also find themselves at an added cost disadvantage because they must pay a booking fee to the CRS-Owning airline for each seat booked by a travel agent. This amounts to an added sales cost of about $2.00 per seat for the airline seeking to expand into a market where most of the local travel agents subscribe to the CRS of the dominant carrier. Part of the booking fee pays for the service provided, but most analyses of this issue conclude that booking fees are substantially greater than the cost of providing the service. Airlines have no choice but to pay these fees to all the cRs-owning airlines or else forego access to much of the available air travel market. In a recent analysis of booking fees, DOTfound that the two largest sys- tems, SABRE and Apollo, generate considerably more revenues from booking fees to their owners, American and United Airlines respec- tively, than those carriers paid out in booking fees to other CRS vendors. The other cas-owning airlines either paid out in booking fees as much as their systems earned or they were net payers of booking fees. DOT also reviewed recent CRS vendor estimates of incremental revenues and con- cluded that incremental revenues in 1988 might have been as high as $2 to $3 billion. This is much higher than DOT’S prior estimates.” Yield Management Yield management is an attempt by the airlines to optimize the pas- senger mix on each flight departure in terms of those paying full fares, those paying discount fares, and those paying deep discount fares. The development of CRSS and the evolution of sophisticated computer sys- tems allow the airlines to deal with large volumes of frequently changing data. The airlines can change their prices on a seat-by-seat basis as often as every 15 minutes. As a result, the airlines make ’‘Airline Marketing Practices: Travel Agencies, Frequent-Flyer Programs, and Computer Reservation Systems. Seealso Study of Airline Computer Reservation Systems. Page 28 GAO/WED-90-102 Fares and Service at Major Airports Chapter 2 The Airline Industry Under Deregulation thousands of fare changes each day. This flexibility also permits incum- bents to make rapid price adjustments in response to potential competi- tion from an entrant. Thus, an incumbent carrier, enjoying an established reputation and offering a wide variety of destinations, can lower prices quickly on routes where it is challenged, thereby frus- trating an entrant’s attempt to attract traffic by undercutting the incumbent’s higher fares. Travel Agent Commission The growing importance of travel agents has led airlines to develop Overrides incentive systems designed to increase their share of travel agent book- ings. One such system, the travel agent commission override, is designed to reward the travel agent for bookings on an airline above and beyond those the agent would have made otherwise. The commission override is often based on all the traffic that the agent books on a particular airline. For example, if the agent normally books $100,000 worth of business each month on a particular carrier and earns a 10 percent commission, the airline might agree to pay a 13 percent commission on all bookings if the agent books at least $120,000 per month. Commission overrides gen- erally apply to total agency sales, but they can be targeted at particular markets and particular flights. DGr concluded in its report on CRSS that commission overrides significantly increase the number of tickets an agent will book on a particular airline. A 1988 Louis Harris survey showed that 51 percent of agency locations reported that they “usually” (24 percent) or “sometimes” (27 percent) chose an air carrier in order to get override commissions.12 Overrides raise the marketing costs of all airlines that pay them, and all carriers have the option of paying override commissions. Nevertheless, for several reasons override commissions might be a more effective strategy for keeping out would-be competitors than for breaking into a market at an airport where another carrier is already dominant. The dominant carrier need only pay an override commission to the travel agents for increased bookings on flights that are threatened by a poten- tial competitor. This might be a relatively small share of the dominant carrier’s total traffic at the airport. The carrier attempting to establish competing service might pay the same absolute amount of override com- missions to induce travel agents to steer passengers to its flights, but these commissions would comprise a proportion of the entrant’s reve- nues from the service out of that airport much larger than that of the ’*“1988 Louis Harris Survey,” Travel Weekly, as cited in Airline Marketing Practices: Travel Agen- cies, kequent-Flyer Programs, and Computer Reservation Systems, pp. 26-27. Page 29 GAO/RCED-90-102 Fares and Service at Major Airports Chapter 2 The Airline Industry Under Deregulation dominant carrier. Potential entrants might ultimately conclude that it is not feasible to start serving this market. Resources in the airline industry are proving to be less mobile than Other Barriers to thought when the industry was first deregulated. In addition to the bar- Competition riers to entry created by airline marketing practices and the control over facilities at some hub airports, entry is also restricted at some airports because of noise and congestion problems. At four of the busiest airports in the nation (O’Hare in Chicago, LaGuardia and JFK in New York, and National in Washington) the number of takeoff and landing slots have been limited since 1969 and the slots are controlled by the airlines that have operated at those airports historically. Some studies have found that fares are higher at slot-controlled airports. Other airports restrict traffic and the type of equipment that can be flown in order to reduce the noise burden on the airport’s neighbors. While noise restrictions are necessary, they also discourage new firms from entering the market. For example, in the past new carriers began operations with used aircraft. However, noise restrictions at many air- ports limit the use of older aircraft.13 Thus, a new carrier may have to purchase relatively new airplanes if it wants to compete. The third major development that has affected competition in the airline Industry Mergers and industry in recent years has been the decline in the number of firms Consolidation providing most domestic passenger service. Shortly after deregulation, intrastate carriers, such as Southwest Airlines and PSA (Pacific South- west Airlines); charter carriers, like Capitol and World; and entirely new carriers, like People Express and America West, began interstate ser- vice. Because these carriers often had lower wage scales and offered fewer service amenities, they had significant cost advantages over the carriers formerly regulated by CAB, and were able to offer substantially lower fares. The market share of the new carriers increased while that of the trunks declined. The share of domestic traffic handled by the trunks fell from almost 90 percent prior to deregulation to 72 percent by 1985. By 1985, more than 20 new carriers had begun interstate service with jet airplanes. ‘“The Federal Aviation Administration designates aircraft as either Stage II or Stage III depending on how much noise they make. As of 1989, about 60 percent of the fleet was Stage II type planes. Even older, Stage I airplanes are no longer allowed to fly anywhere in the U.S. These stages are defined in Federal Aviation Regulation (FAR) Part 36, Sections 36.1 (f)(3) and (f)(5). Page 30 GAO/RCED-90-102 Fares and Service at Major Airports Chapter 2 The Airline Industry Under Deregulation But, over time, the major airlines responded to the new competition by adopting the marketing strategies discussed above and by learning to selectively match the fares of the new entrants. Many of the new entrants and some of the former trunks went bankrupt or were merged with the larger firms. D(Tr ultimately approved all of the mergers it reviewed, including some mergers between carriers that shared a common hub airport. In other cases, both merging carriers served the same region of the nation. This merger activity reached a peak in 1986, when DOT approved 14 applications for acquisition, consolidation, or merger.‘” The reduced competition and the loss of potential competition caused by the mergers and bankruptcies aroused concern even among those who ardently supported deregulation. Nevertheless, deregulation continues to provide significant benefits to the traveling public. While our work focuses on fare and service changes at airports where one or two airlines dominate the traffic, the vast majority of passengers now fly on routes served by at least two airlines; the majority of the nation’s largest 200 airports are not concentrated; and airline fares, adjusted for inflation, on most routes remain below pre-deregulation levels. Economists at the Brookings Institution estimate that airline deregulation continues to generate $10 billion annually in savings to the traveling public. The concern that has been raised centers on airports where one or two carriers dominate the traffic. Our analysis focuses on these so-called concentrated airports and compares fares and service levels at these air- ports with fares and service levels at airports that are not concentrated. 14Theapprovals led to seven mergers among airlines classified as “major” or “national” air carriers and six acquisitions of assetsor of smaller carriers. Page 31 GAO/‘RCED-90-102Fares and Service at Major Airports Trends in Airline Fares at 15 Concentrated AiIpOtiS Airports where one or two carriers handle most of the enplaning traffic have higher fares than airports where the traffic is less con- centrated. Moreover, the data show that fares tend to rise as concen- tration increases. While many factors can influence fare changes, the evidence that we have collected strongly suggests that fares and con- centration at an airport are related. Fares are higher at concentrated airports than at relatively less concentrated ones, and the evidence suggests the gap is increasing. The average yield, or fare per passenger mile, at the 15 airports where Yields Are Higher at one or two carriers handle most of the enplanements is higher than the Concentrated Airports average yield of the airlines at the 38 relatively unconcentrated airports that comprise our comparison group. Many of the 15 concentrated air- ports have been dominated by one or two carriers for some time, and the average yield at these concentrated airports exceeded the average yield at the unconcentrated airports throughout the 1985-1989 period. In 1988, the average yield at the concentrated airports was 27.2 percent higher than the average yield at the 38 comparison airports, but the entire 27.2 percent yield differential did not arise over the 1985-89 period. In 1985, the average yield at the concentrated airports was already 18.8 percent higher than the average at the 38 unconcentrated airports. Nevertheless, the gap has increased.’ In 1988, average yields for all carriers at 14 of the 15 concentrated air- ports were higher than the average yield earned by carriers serving the 38 unconcentrated airports. At 13 of the 15 concentrated airports, the yield of the dominant airline(s) was higher than the yield received by the other airlines serving these airports.* At three of the concentrated airports the yield received by the dominant airline was about 50 percent higher than the yield earned by the other carriers at the airport, and it was at least 15 percent higher at 10 of the airports. Moreover, the yields of the dominant airlines at the concentrated airports were consistently higher than those at the unconcentrated airports. Fare data for the first two quarters of 1989 indicate that these relationships are continuing. Table 3.1 shows yield data and market shares for the dominant carriers at each of the 15 concentrated airports in 1988. ‘These results were obtained when the traffic and sample distributions were held constant. The change is even greater when shifts in traffic and the sample distribution are not taken into account. In that case, yields at the concentrated airports grow from being 17.8 percent higher in 1985 to 29.2 percent higher in 1988. “Yields for American Airlines, the dominant carrier, at Raleigh-Durham are also higher if actual data, rather than weighted data, are used. Page 32 GAO/RCED-90-102 Fares and Service at Major Airports Chapter 3 Trends in Airllne Fares at 15 Concentrated Airports Table 3.1: Airline Yields in 1988 at Concentrated Major Airports Yield in cents, Market shares in percentages Share of dominant Yield of dominant Yield of nondominant Yield of all Airport carrier carrier carriers carriers Atlanta (Delta) 58 25.9 17.4 23.7 Atlanta flexas Air? 36 21.6 17.4 23.7 Nashville (American) 62 21.3 19.7 20.3 Charlotte (USAir)” 93 26.8 18.6 23.7 Cincinnati (Delta) 78 22.0 18.5 20.5 Dayton (USAir)” 79 20.7 20.1 20.4 Denver (Texas Air) 42 17.7 15.8 16.9 Denver (United) 45 16.6 15.8 16.9 Detroit (Northwestlb 59 16.4 12.9 14.3 Greensboro (USAir) 64 26.2 20.4 23.5 Memphis (Northwest) 83 23.3 23.5 23.4 Minneapolis (Northwest) 78 17.1 14.9 16.4 Pittsburah 0JSAirl” 87 19.8 14.3 17.6 Raleigh-Durham (American) 69 20.5 21.5 21.3 Salt Lake City (Delta) 80 21.6 14.5 18.3 St. Louis (TWA) 82 19.4 15.9 18.1 Syracuse (USAir)a 61 21.0 13.9 16.2 All 15 airportsC 20.0 16.0 18.5 =Because USAir’s takeover of Piedmont was approved In October 1987, yrelds and enplanement shares for USAir Include Piedmont data bNorthwest has had 60 percent or more of the enplanements at Detroit at other times between 1985 and 1989. ‘In contrast, yields at the 38 unconcentrated airports in 1988 averaged 14 5 cents per passenger mile. On average, yields at the unconcentrated airports declined from 14.7 Trends in Yields at the cents per passenger mile in the first quarter of 1985 to about 12.4 cents ConcentratedAirports in the second quarter of 1986 and remained at about that level through the second quarter of 1987. After that, the average yield at the uncon- centrated airports began to increase and reached 17.2 cents in the first quarter of 1989. The average yield at the unconcentrated airports declined to 16.4 cents in the second quarter 1989. Between 1985 and mid-1986 the average yield at the concentrated air- ports also fell and then recovered. As was true for the unconcentrated airports, the average yield at the concentrated airports began to increase in the third quarter of 1987 and rose to 21.7 cents in the first Page 33 GAO/RCED-90-102 Fares and Service at Major Airports Chapter 3 Trends in Airline Fares at 15 Concentrated AilpOrts quarter of 1989. The average yield at the concentrated airports also fell in the second quarter 1989 when it averaged 20.5 cents per passenger mile. Despite the similarity in the yield trends at the two groups of air- ports, the difference in the average yield at concentrated and unconcen- trated airports increased from 2.5 cents per passenger mile in the first quarter 1985 to 4.2 cents in the second quarter 1989. Yields have risen nationwide since 1985, but they have risen more at the concentrated airports. Yields at the concentrated airports were 27.8 per- cent higher in the second quarter 1989 than they were in the second quarter of 1985. Yield increases at these airports ranged from 10 per- cent to 68 percent. Over the same period, yields at the unconcentrated airports rose 20.4 percent.” Some of the concentrated airports have long been dominated by one or two airlines, and airlines serving those airports have achieved average yields above those earned by carriers at the unconcentrated airports throughout the period we examined. However, as concentration has increased the difference in yields has grown. Figure 3.1 contrasts yields at the 12 of the 15 airports where concentration has increased substan- tially since 1985 with yields at the 38 relatively unconcentrated airports. 3When comparing quarterly data, we use only the corresponding quarters of each year to avoid the problem of seasonal influences. Page 34 GAO/RCED-90-102 Fares and Service at Major Airports Chapter 3 Trends in Airline Fares at 15 Concentrated Ailp0l-t-S Figure 3.1: Average Yield for 12 Airports Where Concentration Increased 34 Average Yields (Cents per Mile) 32 30 28 26 24 22 18 16 14 12 lQ65 M85 3085 4Q65 lQ86 2Q86 3Q86 4086 1067 2087 3087 4Q87 1088 x288 3Q86 4Q68 lQ89 2Q99 First Quarter 1985 through Second Quarter 1969 - 12 Airports Where Conoenlraiion increased mmmm 38ComparisonAirports The 15 concentrated airports that we examined differed in how and Trends at Different when they became dominated, and the trends in fares reflect those dif- ConcentratedAirports ferences. The 15 airports include several that became concentrated as the result of mergers, others that became concentrated since 1985 after an airline set up or expanded a hub, and others that were concentrated even before 1985. Two of the 15 are dominated by two carriers.4 Yield trends at the 15 concentrated airports tend to track changes in concen- tration levels. Appendix I shows the trend in yields at each concentrated airport from the first quarter 1985 through the second quarter 1989. Airports Where Yields Five of the 15 airports we examined experienced large increases in con- centration as a result of mergers. At two of these airports, single airline Rose Following Mergers dominance was created by the mergers of two carriers hubbing at the 4Thesecategories overlap for some of the airports. For example, we have classified Denver as an airport characterized by two hubbing carriers, but it also was affected by mergers. Page 35 GAO/RCED-SO-102Fares and Service at Major Airports Chapter 3 Trends in Airline Fares at 15 Concentrated Ail-pOrts airport. TWA’S acquisition of Ozark Air Lines eliminated a hubbing com- petitor at St. Louis, and Northwest Airline’s takeover of Republic Air- lines eliminated a hubbing competitor at Minneapolis/St. Paul. Both these mergers were first proposed in early 1986 but were not wholly consummated until late 1986. It is not possible to set a precise date as to when the mergers were far enough along to affect competition and prices. Therefore, we contrast yields during 1985-the year preceding the merger-with yields in the years following the mergers. Table 3.2 shows market shares and yields for the dominant carriers at Minneap- olis/St. Paul and St. Louis and includes yields for the 38 unconcentrated airports. At Minneapolis/St. Paul, Northwest’s average yields were 5.7 percent higher in 1987 than in 1985, but they were 13.9 percent above 1985 levels in 1988. The situation in St. Louis was similar. In 1987 TWA’S yields were 5.1 percent higher than in 1985, and in 1988 they were about 11 percent higher. At the same time, yields at the 38 unconcen- trated airports were actually 5.8 percent lower in 1987 than in 1985, and by 1988 they were 5.5 percent over 1985 levels. According to a recent study of fares and service at these two airports by analysts at the Department of Justice, the mergers led to higher fares at these airports.” Table 3.2: Annual Average Yields and Enplanement Shares of Dominant Shares in percentages, Yields In cents Carriers at Minneapolis/St. Paul and St. Northwestb at Louis Compared to Yields at 38 Minneapolis/St. Paul TWAb at St. Louis Yield at the Unconcentrated Airports Year Share Yield Share Yield 38 airports 1985 42 15.0 59 17.5 . 13.8 1986 52 14.2 61 15.7 12.6 1987 79 15.9 82 18.4 13.0 1988 78 17.1 82 19.4 14.5 198ga 79 18.9 82 21.4 16.7 aThe 1989 data are for first two quarters only. bEnplanement shares are for Northwest and Republrc (Minneapoks/St. Paul) and for T W A and Ozark (St Louis) In 1986 Northwest and T W A enplanement shares were 43 percent and 57 percent, respectively, during the first three quarters of 1986 Yield data are for Northwest and Republic and for T W A and Ozark In 1985 and 1986 Syracuse’s Hancock International Airport also became concentrated as a result of mergers. Between 1977 and 1980, USAir had the largest share “Gregory J. Werden, Andrew S. Joskow, and Richard L. Johnson, The Effects of Mergers on Economic Performance: Two Case Studies from the Airline Industry, Antitrust Division, U.S. Department of Justice (Washington, DC.: n.d.). Page 36 GAO/RCRD-SO-102Fares and Service at Major Airports Chapter 3 Trends in Airline Fares at 15 Concentrated AirpOrts of enplanements with about 38 percent of the market. In 1979, Empire Airlines was formed and established Syracuse as its hub. Although USAir continued to have the largest share of enplanements, by 1985, Empire and USAir each handled about 28 percent of the enplaning pas- sengers at Syracuse. In 1986 Piedmont took over Empire, and in 1987 D(JTapproved the merger of USAir and Piedmont. In 1988, USAir had 61 percent of the enplanements at Syracuse. Quarterly data on fares at Syracuse show that fares have risen more rapidly since USAir has taken over Piedmont. Although the USAir-Pied- mont merger was first approved by DOT in late October 1987, the actual integration of the two airlines proceeded slowly. USAir’s fares rose less rapidly at Syracuse during the first two quarters of 1988 than at the 38 unconcentrated airports. However, by the third quarter 1988 USAir’s fares at Syracuse had begun to rise faster than fares at the 38 unconcen- trated airports. Table 3.3 shows changes in yields for USAir-Piedmont at Syracuse and for the 38 unconcentrated airports. Table 3.3: USAir-Piedmont Fare Changes at Syracuse Compared to Changes in Percentage change from same period in prior year Fares at 38 Unconcentrated Airports Year and quarter USAir at Syracusea 38 airports 1988 I - 1.5 12.9 198811 13.8 17.5 1988 III 11.7 9.3 19881V 12.2 7.6 1989 I 33.5 21.8 198911 14.7 12.4 aPercentage changes based on USAIr and Piedmont data The recent study by Department of Justice analysts compared the effect of a merger on Detroit to the merger impacts at St. Louis and Minneap- olis/St. Paul. The analysts expected little or no effect from the merger on Detroit fares because only one of the merging carriers operated a hub at Detroit Metropolitan Airport. They found that the merger caused almost no change in fares at Detroit. Our data show a similar result. We found that the fares of the dominant carrier at Detroit rose by about the same amount as fares at the unconcentrated airports during the 1985-89 period. Table 3.4 shows enplanement shares for Northwest-Republic and yield data for Northwest-Republic at Detroit and for the 38 unconcen- trated airports. Page 37 GAO@CED-90-102 Fares and Service at Major Airports Chapter 3 Trends in Airline Fares at 15 Concentrated AilpOrts Table 3.4: Northwest-Republic Yields at Detroit Metropolitan Wayne County Shares in percentages, Yields in cents, Yield differences in percentages Airport and at 38 Unconcentrated Northwest-Republicb Yield at the 38 Difference in Airports Year Enplanement Share Yield airports yieldsC 1985 44 15.9 13.8 15.9 1986 61 15.2 12.6 20.5 1987 60 15.6 130 20.1 1988 59 16.4 14.5 12.8 1 98ga 64 19.2 16.7 16.0 ‘The 1989 data for first two quarters only. bEnplanement share for Repubk in 1985, Republic and Northwest in 1986 Yield data for Northwest and Republic in 1985 and 1986. ‘Percentage difference between dominant airline’s yield and yield at the 38 unconcentrated airports was calculated pnor to roundmg of yield data Denver’s Stapleton International Airport became concentrated when People Express, which had recently purchased Frontier, was acquired by Continental in late 1986. Denver and Atlanta, the only airports among the 15 to be dominated by two carriers, are discussed later in this chapter. Yields Increased Following At airports where a carrier established a dominant position by setting up a hub during the period we examined, yields rose following the the Establishment of Hub increase in concentration. American Airlines established a hub at Nash- Operations ville in the first half of 1986 and at Raleigh-Durham in the middle of 1987. In both cases, American’s yields increased following the establish- ment of hub operations. American’s yields had been about 20 percent below those of other airlines serving Nashville and about 25 percent lower than others serving Raleigh-Durham in the year before the hubs were set up. After American became the dominant carrier, its yields rose much faster than those of the other carriers serving these airports. In the second quarter 1989 American’s yields at Nashville were about 9 percent higher than those of other carriers and were only 3 percent below those of other carriers at Raleigh-Durham. Table 3.5 shows enplanement shares and yields for American at Raleigh-Durham and at Nashville and includes yield data for the 38 unconcentrated airports. Page 38 GAO/RCED90-102 Fares and Service at Major Airports Chapter 3 Trends in Airline Fares at 15 Concentrated AirpOrts Table 3.5: Annual Average Yields and Enplanement Shares of American Shares in percentages, Yields in cents Airlines at Nashville and Raleigh-Durham Americ;;ra;aleigh- American at Compared to Yields at 38 Nashville Yield at the Unconcentrated Airports Year Share Yield Share Yield 38 airports 1985 3 11.6 19 17.3 13.8 1986 4 13.7 45 17.6 12.6 1987 41 16.8 59 19.5 13.0 1988 69 20.5 62 21.3 14.5 19a9a 78 246 72 24.0 16.7 aThe 1989 data are for first two quarters only In establishing its hubs at Nashville and Raleigh-Durham, American greatly increased its number of short distance flights. Between the fourth quarter 1985 and the fourth quarter 1988, the share of Amer- ican’s traffic in the O-500 mile category increased from less than 1 per- cent to 18 percent in Nashville. Between the fourth quarter 1986 and the fourth quarter 1988, the share of American’s traffic in the O-500 mile category grew from 9 percent to 20 percent in Raleigh-Durham. Since short distance flights have higher yields, the change in traffic mix most likely accounts for at least some of the observed increase in average yields.” At Cincinnati, Delta Airlines had been the largest carrier in terms of enplanements for more than a decade, but did not dominate Cincinnati air travel until it doubled the number of flight operations in late 1986 and early 1987. Yields at Cincinnati for Delta increased about 14 percent in 1987 while yields at the comparison airports increased about 3 per- cent. At Cincinnati the fares of the nondominant carriers, while still lower than Delta’s, have increased more than Delta’s since 1987. Table 3.6 shows yield data and enplanement shares for Delta at Cincinnati and yields at the 38 unconcentrated airports. tiYields are higher on short distance flights becausecost per passengermile are higher. Some airline costs do not vary with miles flown but with the number of takeoffs and landings or other factors than distance. Page 39 GAO/RCED-90-102 Fares and Service at Major Airports Chapter 3 Trends in Airline Fares at 16 Concentrated Ailp0l-t-S Table 3.6: Delta Yields and Market Shares at Cincinnati Compared to Yields Shares in percentages, Yields in cents at 38 Comparison Airports Delta Yield of other Cincinnati Yield at the 38 Year Share Yield carriers airports 1985 48 18.8 15.7 13.8 1986 47 17.7 15.1 12.6 1987 72 20.2 16.2 13.0 1988 78 22.0 18.5 14.5 198ga 83 24.9 21.7 16.7 aThe 1989 data are for first two quarters only Already Dominant Airlines In most situations where airports have always been dominated by one carrier, yields increased as concentration increased. Charlotte/Douglas Maintained Higher Yields International Airport has long been a concentrated airport. Domination by Eastern Airlines in the 1970s and early 1980s was replaced by domi- nation by Piedmont (now part of USAir) in 1982. Despite the fact that Charlotte has long been concentrated, USAir has extended its dominance so that Charlotte is the most concentrated airport of the 15 we examined. In 1985 Piedmont handled about three-fourths of the enplanements at Charlotte; by the first half of 1989, USAir-Piedmont handled more than 94 percent of the enplanements. Yields have long been relatively high at Charlotte, but as USAir-Pied- mont’s share of enplanements increased, yields also increased. USAir- Piedmont’s yields at Charlotte rose 32 percent as its share of enplane- ments increased 16 percentage points between 1985 and 1988. During the same period, the average yield at unconcentrated airports rose less than 6 percent. Table 3.7 shows yields and enplanement shares for USAir-Piedmont at Charlotte and yields at the 38 unconcentrated airports. Page 40 GAO/RCED90-102 Fares and Service at Major Airports Chapter 3 Trends in Airline Fares at 15 Concentrated AhpOrts Table 3.7: Yields and Enplanement Shares at Charlotte for USAir-Piedmont Shares In percentages, Yields in cents, Yield differences in percentages and Yields at 38 Unconcentrated Airports USAir-Piedmont Yield at the 38 Difference in Year Enplanement shareb Yield airports yieldsC 1985 77 20 3 138 47.5 1986 81 21.2 126 68.2 1987 89 22.7 130 75.4 1988 93 26 8 14.5 85.0 198ga 94 30.5 16.7 83.0 aThe 1989 data for first two quarters only bEnplanement share for Piedmont only In 198587 ‘Percentage drfference between dominant arrlrne’s yield and yreld at the 38 unconcentrated arrports was calculated prior to rounding of yield data Dayton, another Piedmont hub taken over by USAir, has been domi- nated by USAir-Piedmont throughout the 1985-89 period. The domi- nance has not been as dramatic as at Charlotte but, as at Charlotte, yields have risen and the gap between the dominant carrier’s yields and the yields at unconcentrated airports has widened, as USAir has increased its dominant position. Dayton was one of the airports whose dominant carrier’s yields were below those of the other carriers at the airport. Between 1985 and 1987 Piedmont’s yields were around 4 per- cent lower than those of the other carriers serving Dayton. Since 1988, however, USAir’s yields have been above those of the other carriers and in the first two quarters of 1989 have averaged more than 10 percent higher. Table 3.8 shows yields and enplanement shares for USAir-Pied- mont at Dayton and yields at the 38 unconcentrated airports. Table 3.8: Yields and Enplanement Shares at Dayton for USAir-Piedmont Shares in percentages, Yields In cents, Yield differences in percentages and Yields at 38 Unconcentrated Airports USAir-Piedmont Yield at the 38 Difference in Year Enplanement shareb Yield airports yieldsC 1985 66 17.2 13.8 25 0 1986 65 163 12.6 29.0 1987 71 17.6 13.0 35.8 1988 79 20.7 14.5 42.7 198ga 80 23.2 16.7 39.5 aThe 1989 data are for first two quarters only bEnplanement share for Predmont only In 198587. ‘Percentage difference between dominant arrlrne’s yreld and yreld at the 38 unconcentrated arrports was calculated prior to rounding of yield data. Page 41 GAO/RCED-90-102 Fares and Service at Major Airports Chapter 3 Trends in Airline Fares at 15 Concentrated AilpOrts Piedmont Triad International Airport, serving the Greensboro/High Point/Winston-Salem area, is unique among the concentrated airports we examined in that it is not a hub airport for the dominant carrier. Piedmont has had the largest presence at Greensboro over the entire period we reviewed. As was the case at Charlotte and Dayton, USAir- Piedmont’s market share has grown at Greensboro, and yields have fol- lowed suit. Piedmont’s market share was about 55 percent in 1985, and USAir has increased this to 64 percent by 1988. However, USAir-Pied- mont’s yields have risen much more rapidly. Table 3.9 shows yields and enplanement shares for USAir-Piedmont at Greensboro and yields at the 38 unconcentrated airports. Table 3.9: Yields and Enplanement Shares at Piedmont Triad International Shares in percentages, Yields in cents, Yield differences In percentages Airport for USAir-Piedmont and Yields at USAir-Piedmont Yield at the 38 Difference in 38 Unconcentrated Airports Year Enplanement shareb Yield airports yieldsc 1985 55 17.5 13.8 27.0 1986 57 18.9 12.6 49.9 1987 56 20.8 13.0 60.3 1988 64 26.2 14.5 80.4 1989" 66 31.2 16.7 87.5 aThe 1989 data are for first two quarters only bEnplanement share for Piedmont only in 1985-87 CPercentage difference between dommant airline’s yield and yield at the 38 unconcentrated airports was calculated prior to rounding of yield data Memphis International Airport was a hub for Republic Airlines and is now a hub for Northwest Airlines. Since Northwest had no presence in Memphis before it acquired Republic, the merger, per se, did not affect competition at the Memphis airport. Until 1982 Delta had the largest share of enplanements at Memphis. In 1982 Republic became the domi- nant carrier, and since 1984 Delta’s share of the traffic has plummeted. By 1988 Northwest had 83 percent of the market. As is the case at most concentrated major airports, yields are higher at Memphis than they are at the 38 unconcentrated airports. However, changes in yields have lagged behind changes in the market share of the dominant carrier. In 1986 and 1987 concentration increased, but yields did not rise appreciably until 1988, when the growth in Northwest’s enplanement share had stopped. In early 1989, yields rose only slightly. The gap between yields at Memphis and unconcentrated airports has narrowed, but carrier yields at Memphis are still almost 44 percent Page 42 GAO/RCED-90-102 Fares and Service at Major Airports Chapter 3 Trends in Airline Fares at 15 Concentrated AilpOrts above those at the 38 unconcentrated airports. Table 3.10 shows yields and enplanement shares for Northwest-Republic at Memphis and yields at the 38 unconcentrated airports. Table 3.10: Yields and Enplanement Shares at Memphis for Northwest- Shares In percentages, Yields In cents, Yield differences in percentages Republic and Yields at 38 Northwest-Republic Yield at the 38 Difference in Unconcentrated Airports Year Enplanement shareb Yield airoorts vieldsC 1985 64 21.2 13.8 54.3 1986 74 20.7 12.6 64.5 1987 85 21.7 13.0 67.8 1988 83 23.3 14.5 60.8 1989" 82 23.9 16.7 43.7 aThe 1989 data are for frrst two quarters only. bEnplanement share for Republrc only In 1985. =Percentage difference between dominant arrlme’s yreld and yield at the 38 unconcentrated airports was calculated prior to rounding of yield data Western Airlines had its hub at Salt Lake City before it was taken over by Delta in early 1987. Western’s yields had declined from the third quarter 1985 through the second quarter 1986. Delta proposed its purchase of Western in the third quarter 1986. Since Delta took over the hub at Salt Lake in early 1987, yields have risen as Delta has increased the share of enplanements held by the dominant carrier. Table 3.11 shows yields and enplanement shares for Delta-Western at Salt Lake City and yields at the 38 unconcentrated airports. Table 3.11: Yields and Enplanement Shares at Salt Lake City for Delta- Shares In percentages, Yields in cents, Yreld differences in percentages Western and Yields at 38 Delta-Western Yield at the 38 Difference in Unconcentrated Airports Year Enplanement shareb Yield airports vieldsC 1985 74 169 13.8 23.2 1986 74 16.1 12.6 27.4 1987 77 17.5 13.0 34.8 1988 80 21 6 14.5 49.2 1989" 82 23.5 16.7 40.9 aThe 1989 data are for first two quarters only. bEnplanement share for Western only In 1985 and 1986. CPercentage difference between dominant airline’s yield and yield at the 38 unconcentrated airports was calculated prior to rounding of yield data Page 43 GAO/RCED-90-102 Fares and Service at Major Airports Chapter 3 Trends in Airline Fares at 15 Concentrated AirpOrts At Pittsburgh, where concentration levels were high and the dominant carrier’s market share increased one to three percentage points each year, yields did not increase until the first half of 1989. Nevertheless, they remained substantially above yields at unconcentrated airports. At Pittsburgh, USAir has accounted for 80 percent or more of the enplane- ments during the entire period under review. Its yields declined some- what in 1986 and 1987, but rose again in 1988 and by the first half of 1989 were still more than 40 percent above the yields earned at the 38 relatively unconcentrated airports. The decline may have reflected changes in the distribution of USAir’s traffic. The proportion of pas- senger miles flown in the lowest distance/highest yield category (O-500 miles) fell from 43 percent to 31 percent between the fourth quarter 1985 and the fourth quarter 1988. Table 3.12 shows yields and enplane- ment shares for US&r-Piedmont at Pittsburgh and yields at the 38 unconcentrated airports. Table 3.12: Yields and Enplanement Shares at Pittsburgh for USAir-Piedmont Shares in percentages, Yields In cents, Yield differences in percentages and Yields at 38 Unconcentrated Airports USAir-Piedmont Yield at the 38 Difference in Year Enolanement shareb Yield airoorts vieldsC 1985 80 20.8 13.8 51.4 1986 83 19.3 12.6 53.4 1987 84 18.7 13.0 44.2 1988 87 19.8 14.5 36.4 198ga 89 23.4 16.7 40.4 aThe 1989 data are for first two quarters only. bEnplanement share for USAir only In 1985-87 ‘Percentage difference between dominant airline’s yield and yield at the 38 unconcentrated airports was calculated prior to rounding of yield data. Cities W ith Two Domin .ant At the Atlanta and Denver airports two airlines dominated the traffic. At Atlanta, before the Eastern Airlines strike, Delta Airlines handled Airlines Provide an almost 60 percent of the enplaning passengers while more than one- Additional Perspective third was handled by Eastern Airlines. Up until the time of the strike, the two-carrier concentration level was substantially unchanged, although Delta had increased its share somewhat relative to Eastern’s. Yields at Atlanta fell between 1985 and 1987. However, yields rose in 1988 and yields for both Eastern and Delta rose sharply in the first quarter of 1989. In the second quarter of 1989, Delta’s yield remained high, while Eastern’s plummeted as a result of the machinists’ strike. As at most of the other concentrated airports, yields at Atlanta for the Page 44 GAO/RCED-90-102 Fares and Service at Major Airports Chapter 3 Trends in Airline Fares at 16 Concentrated AirpOrts dominant carriers are substantially higher than yields at the unconcen- trated airports. In addition, the yields for Delta, the carrier with the larger share of the enplanements at Atlanta, have been consistently higher than Eastern’s yields. Table 3.13 shows yield data for Atlanta and for the 38 unconcentrated airports. Table 3.13: Yields at Atlanta for Delta and Eastern Airlines and Yields at 38 Yields In cents Unconcentrated Airports Atlanta Yields Yield at the 38 Year Delta Easternb All Carriers airports 1985 24.0 20.6 22.1 13.8 1986 22.7 18.4 20.6 12.6 1987 21.8 17.9 19.9 13.0 1988 25.9 21.6 23.7 14.5 198ga 29.3 22.4 26.4 16.7 aData for 1989 are for first two quarters only blncludes data for Continental and People Express Both United Airlines and Continental Airlines operate hubs at Denver. Before the third quarter 1986, Frontier Airlines also enjoyed a major presence at Denver, and average yields at Denver ranged from 9 to 26 percent below the average for the comparison group. After Continental took over People Express and Frontier, yields at Denver increased. Prior to the third quarter of 1986, Denver was not a concentrated airport, and the average yield at Denver was lower than that for the 38 unconcen- trated airports. This changed in 1987. Yields for both United and Conti- nental rose rapidly, and Denver experienced an exceptionally large increase in yields during the first two quarters of 1989. For the first two quarters of 1989, Continental’s yields were 33 percent higher than those at the unconcentrated airports while United’s were almost 20 percent higher. For Denver’s airport as a whole, yields were 23 percent higher than the yields at the unconcentrated airports during the first two quarters of 1989. Table 3.14 shows yield data for Denver and for the 38 unconcentrated airports. Page 46 GAO/RCED-SO-102Fares and Service at Major Airports Chapter 3 Trends in Airline Fares at 15 Concentrated Airports Table 3.14: Yields at Denver for United and Continental and Yields at 38 Ytelds in cents Unconcentrated Airports Denver Yields Yield at the 38 Year United Continentalb All carriers Airports 1985 12.0 11.9 12.2 13.8 1986 10.4 10.2 10.6 12.6 1987 14.2 15.8 14.8 13.0 1988 16.6 17.7 16.9 14.5 1989" 19.9 22.2 20.4 167 aData for 1989 are for first two quarters only blncludes data for Eastern, Frontier, and People Express In addition to concentration and market power, other factors could Several Other Factors account for the differences in yields at the 15 concentrated and 38 rela- Could Affect Yield tively unconcentrated airports and the differences between the domi- Differences nant carriers and the other airlines serving the concentrated airports. One factor that might account for the differences between the airports is length of haul. We compared yield changes at the 15 concentrated air- ports with yield changes at a subset of our comparison group of airports that excluded airports where average trip lengths were much longer than those of the concentrated airports. We excluded airports with longer average trip distances (greater than 900 miles) because yields are generally lower for longer trips. When we compared the 15 airports to this smaller comparison group of 22 airports, the difference in yields narrowed, although the trends remained the same (see fig. 3.2). Page 46 GAO/RCEIN6-102 Fares and Service at Major Airports Chapter 3 Trends in Airline Fares at 15 Concentrated AllpOrts Figure 3.2: Average Yield for 22 and 38 Comparison Airports 34 Average Yields (Cants per Mile) 32 30 28 26 24 22 20 18 16 14 12 lQ85 2Q65 3Q85 4Q85 la86 2086 3486 4Q56 1QW Ma7 3Q37 4087 lQ88 3Q63 3-8 4Q88 lQ89 2Q89 First Quarter 1985 through Second Quarter 1989 - 22 Comparable-Distance ComparisonAirports mH=M 38 Comparison Airports Contrasting yields at the 15 concentrated airports with yields at the subset of 22 unconcentrated airports having comparable average trip distances shows that yields at the concentrated airports were 21 percent higher in 1988. This difference is 6.2 percentage points smaller than the difference we observed between the 15 concentrated and the larger group of 38 unconcentrated airports. Table 3.15 shows that, as was the case with the 38 airports, the difference in yields between the 15 con- centrated and the 22 unconcentrated airports has widened over time. Page 47 GAO/RCED-90-102 Fares and Service at Major Airports Chapter 3 Trends in AirEne Fares at 15 Concentrated AilpOrts Table 3.15: Differences in Yields at 15 Concentrated Airports and 22 Yields in cents, Yield differences in percentages Unconcentrated Airports Yield at 15 Yield at 22 Difference in Year concentrated airports unconcentrated airports yieldsb 1985 16.3 14.6 11.7 1986 15.2 13.4 13.7 1987 16.4 13.7 19.8 1988 18.5 15.3 21.0 1989” 21.0 17.4 20.7 aThe 1989 data are for first two quarters only bPercentagedifference between yield at the 15 concentrated airports and yield at the 22 unconcen- trated airports was calculated phor to rounding of yield data In preparing the testimony on fares at concentrated airports that we presented before the Senate Committee on Commerce, Science, and Transportation in June 1989, we interviewed officials with all of the major airlines except United, whose officials declined to meet with us. According to some of these industry spokespersons, yields could be higher at the concentrated than at the unconcentrated airports because traffic out of the concentrated hub airports was more often nonstop or direct, while traffic out of the unconcentrated airports often had to be routed through hubs. The industry spokespersons claimed that nonstop or direct service is more desirable and could command higher fares than connecting service. While we did not control for this directly, we did compare the average number of coupons per traveler out of the concen- trated airports with the average number at the unconcentrated airports. For the fourth quarter 1988, the average number of coupons was 2.26 for the concentrated airports and 2.28 for the unconcentrated airports. This comparison suggests that the type of service was not materially different for the two groups of airports. Regarding the difference between the dominant airline and the other air- lines at the concentrated airports, the dominant carrier probably pro- vides nonstop or direct service more often than other airlines at the concentrated airport, which may be providing connecting service through other hubs. In addition, dominant airlines may command a higher proportion of higher yield, short haul traffic at the hubs. According to our data on direct service, the dominant carriers earn somewhat lower yields on average than the other carriers serving the concentrated airports. However, considerable variation among the 15 airports makes it difficult to draw any firm conclusions about yield dif- ferences between the dominant and nondominant airlines for different types of service. Page 48 GAO/RCED-SO-102Fares and Service at Major Airports , Chapter 3 Trends in Airline Fares at 15 Concentrated AilpOrts While we did not adjust our data for the proportion of traffic carried by the dominant and nondominant carriers when we calculated the average yield, we did break down the yield data into mileage blocks using 500 mile increments. We found that for the shortest distance category (O-500 miles) the dominant airlines’ yields at the concentrated airports were consistently higher than the yields for trips of a similar distance out of the comparison airports. In addition, at 11 of the 15 airports the yields on the short haul flights of the dominant airline were higher than those of the other airlines serving the concentrated airport. In the longest mileage category, these differences persisted, suggesting that even though the dominant airlines may have a higher proportion of the short haul traffic, their higher share does not account for all of the difference in average yield. In preparing this study, we had to choose which fares to include, which Sensitivity Analysis factors to control, and how to treat anomalies in data and reporting. We believe that our assumptions and adjustments give the most accurate picture possible of fare levels and trends, but it is important to know the effect of these assumptions on the results. Therefore, we attempted to determine how each assumption or adjustment affected the outcome by relaxing each assumption and recalculating the difference between fares at the concentrated and unconcentrated airports. We undertook the sen- sitivity analysis for 1988 annual data to test how assumptions affected our finding that fares were about 27 percent higher at concentrated airports. Weighted Data Because we were interested in trends in fares at concentrated airports over time as well as differences between concentrated and unconcen- trated airports, we weighted the fare data to take into account changes in the distribution of traveler destinations, changes in the proportions of one-way and round-trip tickets in the sample, and changes in the pro- portions of trips taken on the dominant and nondominant carriers. We wanted changes in yields to reflect fare changes and not changes in the trips taken. Therefore, for each combination of fare type (one-way or round-trip), type of carrier (dominant or nondominant) and destination, we calculated the average yield for each quarter. We weighted the average yield for each combination according to the average amount of traffic for that combination over the 18 quarters. For example, if one- way trips from Denver to Chicago on United Airlines averaged 0.1 per- cent of all trips on United over the 18 quarters, we weighted the results for each quarter so that the proportion was always the same. Page 49 GAO/RCED90-102 Fares and Service at Major Airports Chapter 3 Trends in Airline Fares at 16 Concentrated AirpOrEs While this weighting is appropriate for analyzing changes over time, DOT officials have criticized its use in comparing fare levels at different groups of airports at the same time. Unweighted, actual data might give different results. Therefore, we re-estimated 1988 fares at concentrated and unconcentrated airports without applying the weights. We found a slight increase in the difference. W ith unweighted data, fares were 29 percent higher at the concentrated airports. This result might be expected, since we had observed that the difference was growing over time. Adjusting for changes in traffic distribution had somewhat damp- ened the gap in the latter years, The following adjustments use the actual, unadjusted data for 1988. Alaska Airlines Because of reporting problems, Alaska Airlines was excluded from the database. Alaska had reported fares many times greater than those actually charged. To avoid biasing the results at those airports where Alaska offered a significant amount of service, we excluded the carrier from our analysis entirely. Alaska Airlines has since rectified the problem, and fare data for recent periods are more accurate. We re-esti- mated 1988 yields including Alaska Airlines, but the effect was small. The difference in yields, using unweighted data, between concentrated and unconcentrated airports including Alaska Airlines was 28.6 percent. Interline Fares Because we were interested not only in average fares at concentrated airports but also in the fares of the carriers that dominated those air- ports, we eliminated from our database trips that required the traveler to change airlines. By doing so we avoided the difficulty of trying to apportion the fare between the carriers. Interline trips are becoming less common but still comprise a measurable segment of total air travel, and including them gives a more complete picture. Since certain economies are associated with on-line connections, the a priori assumption would be that interline fares are higher. Recalculating yields including interline trips did produce slightly higher yields, but the yields were higher at both concentrated and unconcentrated airports. After including interline tickets, yields at the concentrated airports were 30 percent higher than at unconcentrated ones. Multiple Coupon Trips Our database was restricted to trips involving only two coupons in each direction. This restriction admits into the database only direct flights and those with one change of plane in each direction. W ith this restric- tion the database captured most of the traffic, but excluded some Page 60 GAO/RCED90-102 Fares and Service at Major Airports Chapter 3 Trends in Airline Fares at 15 Concentrated AilpOrts trips-especially those where the final destination was a small city. Fares might be expected to be lower for trips requiring multiple plane changes. We expanded our database to included three coupons in each direction, that is, flights requiring two plane changes in each direction, but again found little change in the size of the yields and the difference between yields at concentrated and unconcentrated airports. As expected, yields were slightly lower, but concentrated and unconcen- trated airports were equally affected. Low End Fare Screens We excluded from our data set fares that were either obviously too high or too low based on a review of listed fares, including $0 and nominal fares paid by frequent flyers. Our interest was in examining fares actu- ally paid for individual trips. Some analysts, however, may include $0 fares because they believe that free travel earned in frequent flyer pro- grams should be included in calculating the average fare for travel out of an airport. We re-estimated yields without screening out fares that were too low and, as might be expected, yields fell. Yields were about 6 percent lower at both the concentrated and the unconcentrated airports, and there was a small change in the difference. Yields at the concen- trated airports were still 27.6 percent above those at the unconcentrated airports. The results from relaxing each of the various assumptions discussed above appear in table 3.16. Table 3.16: Sensitivity Analysis of Assumptions Employed to Calculate Yields in cents, Yield differences In percentages Yields at Concentrated and Average yield Difference in Unconcentrated Airports Assumption Concentrated Unconcentrated yieldsa All assumDtlons in Dlace 185 145 27.2 Unweiclhted data 182 14 1 29.2 Alaska Airlines included 18.2 14 1 28.6 Interline fares included 19.0 14.6 30.0 Three couDon trbs included 181 14.0 29.4 No low end yield screen 170 13.3 27.6 ‘Differences In yields were calculated prior to roundmg of yield data Page 51 GAO/RCED90-102 Fares and Service at Major Airports Chapter 4 Other RecentAnalyses of Fares at ConcentratedAirports Over the past year several studies have examined fares at concentrated airports and at airports where the major carriers have set up their hubs.l Some of these analyses were undertaken in response to findings we reported in testimony before the Congress.2According to the results reported in Chapter 3, fares are significantly higher at major airports where one or two carriers dominate the traffic than at airports where enplaning passengers are distributed more widely among different carriers. The issues that we address in this report have also been examined by industry and other government agencies. The Air Transport Association recently commissioned a study of fares and service at hub airports. DOT has examined fare and service changes at St. Louis following the TWA- Ozark merger and has recently completed a study of fares at its own sample of concentrated airports. Finally, analysts at the Justice Depart- ment have recently assessedthe relationship between fares and concen- tration at three airports that became concentrated following the merger of hubbing carriers.3 We reported in September 1988 that average fares had risen for travel The DCYI’ St. Louis out of St. Louis following the merger of TWA and Ozark Air Lines.4 We Study also testified on our findings at hearings before the Senate Committee on Commerce, Science, and Transportation.5 DCIT also testified at those hear- ings, and the Department was asked to prepare a response to our finding that, following the merger of TWAand Ozark Airlines (a merger DOT approved over the objections of the Justice Department), fares for St. Louis travel had risen and competition at St. Louis had declined. DOT issued its report in January 1989, and the Department took issue with our findings on several grounds. First, DOT claimed our analysis I ISome of these studies try to explain why air fares might be higher at concentrated ah-ports by estimating an econometric model. We are also in the process of estimating a model of airline pricing behavior that will focus on how the various barriers to airline market entry affect airline fares. “DOT report on fare and service changes at St. Louis (DCJf-P-37-89-3);Hub Operations: An Analysis of Airline Hub and Spoke Systems Since Deregulation, prepared for the Air Transport Association by Siiat, Helliesen & Eichner, Inc. (May 1989). In addition, DOT recently issued a series of reports by the Secretary’s Task Force on Competition in the U.S. Domestic Airline Industry that includes a study of fares at most of the concentrated airports in this study. “Werden, Joskow, and Johnson. lGA0 report on fare and service changes at St. Louis (GAO/RCED-8%217BR). ‘Factors Affecting Concentration (GAO/T-RCED-88-65). Page 52 GAO/RCED-90-102 Fares and Service at Major Airports Chapter 4 Other Recent Analyses of Fares at Concentrated Airports was based on faulty data. The data that DW had provided us contained a fare filter that was out-of-date and led to the exclusion of many valid fares. In fact, we first identified the fare-filter problem and brought it to the attention of DOT.We created a new fare filter which DOTlater adopted. D m was also critical of the periods we chose to compare, the first three quarters of 1986 and 1987, and noted (as we had in our report) that fares were especially low in 1986, making the fare increases in 1987 seem larger than if some other base period were chosen. Finally, based on the expected effect of the merger on market power and, there- fore, on fare changes, we had separated the routes out of St. Louis into four categories: (1) markets where TWAand Ozark were the only carriers offering nonstop service prior to the merger, (2) markets in which TWA, Ozark, and at least one other carrier offered nonstop service, (3) mar- kets where only TWAor Ozark offered nonstop service, and (4) markets that received nonstop service only from TWAor Ozark and at least one other carrier. We expected the largest increase in fares to occur in mar- kets where TWAand Ozark had been the only carriers providing nonstop service prior to the merger, since the merger would lead to a monopoly on these routes. Our results were inconclusive. The fare increase we reported was relatively small in those markets where the merger pro- duced a monopoly and was largest in markets where TWAand Ozark competed along with other carriers. DOTcited this unexpected result as further evidence that our analysis was flawed. D m re-estimated the fare changes at St. Louis using a much less restric- tive constraint on allowable yields. DOTexamined fares for the same 67 routes that we analyzed and employed the same division based on com- petitive market categories. D m used the first half of 1985 as the base period and examined the trend in fares through the first half of 1988. DOTfound that TWA'Sfares in 1987 were 10 percent above those in 1985 and 20 percent over those in 1986. The increase reported by DOT,there- fore, is even larger than the 13 to 18 percent increase we reported. DOT also found that TWA'Sfare increases were greater than the rise in either the airline component of the Consumer Price Index (CPI) or the overall CPI. DOTreported that average fares rose an additional 11 percent in the first half of 1988, while, at the same time, the airline component of the CPI was unchanged and the overall CPI rose 4 percent. Between the first half of 1985 and the first half of 1988, DCTfound that TWA'Sfares for these 67 nonstop markets rose 22.7 percent, more than twice the increase in either measure of inflation. Page 53 GAO/RCED-90-102 Fares and Service at Major Airports Chapter 4 Other Recent Analyses of Fares at Concentrated Airports DW attributed the relatively large increase in TWA'Sfares out of St. Louis to demand factors. Between 1985 and 1988, St. Louis origin and destina- tion traffic increased by an average of 6.9 percent annually whereas, nationwide air travel grew 3.3 percent over the same period. In exam- ining why TWA'Sfares might have risen, DOTpresented data on TWA's system-wide operating costs. But, between March 1985 and June 1988, TWA'Soperating costs per available seat mile fell 16 percent from 8.82 cents to 7.41 cents. DOTdid not comment on why the fare increase out of St. Louis coincided with this sizable reduction in operating costs. DOT also presented evidence on fare changes at other concentrated hub airports to demonstrate that TWA'Sincreases were not atypical (see table 4.1). However, these data reinforce our point that fares have increased at airports where concentration has increased. Only Pittsburgh, which was concentrated throughout the 1985-88 period, and Atlanta, a two- carrier hub that was also concentrated throughout the period we examined, did not show a substantial increase. These results are fully consistent with the data we presented in Chapter 3. DOT found that fares were relatively unchanged in the 38 city pair mar- kets where TWAand Ozark did not compete before the merger, but like GAO,~o'r noted that fares rose most in those markets where TWAand Ozark offered nonstop service along with other airlines, not in those markets where TWAgained a monopoly after absorbing Ozark. Based on this circumstance, DCXfound no basis for concluding that the merger had significantly affected fares. Table 4.1: Comparison of Round-Trip Fare Changes by Carriers Dominating Percent fare change 1985- Hub@ Hub Carrier 88 Atlanta Delta +5 Charlotte Piedmont +34 Clnclnnati Delta +25 Detroit Northwest +27 Minneapolis Northwest +21 Plttsburah USAir -6 Raleigh American +35 St. Louis TWA $22 Salt Lake Cltv Delta $26 ?5ource: DOT report on fare and service changes at St LOUIS(DOT-P-37-89-3). However, our anomalous result occurred because we included the New York City-St. Louis route among those in which TWAand Ozark competed Page 54 GAO/RCED90-102 Fares and Service at Major Airports Chapter 4 Other Recent Analyses of Fares at Concentrated Airports with other carriers. However, the only other carrier in this case was People Express, an airline that went out of business during the period covered by our study. Thus, this route could have been classified with those where the merger resulted in a monopoly. Because fare increases on this route were larger than on any of the other 67 we examined (39 percent) and because it is such a heavily traveled route, transferring it to the post-merger monopoly category eliminates most of the anomaly in our results. W ith the New York City-St. Louis route included with the post-merger monopoly group, fare increases were largest in those mar- kets where TWAand Ozark competed, and fares rose by roughly the same amount on those routes regardless of whether or not other carriers also served the route. DOT also included the New York City-St. Louis route among those in which TWA and Ozark competed with other carriers. Since DOT found the anomaly to be smaller than we estimated, including St. Louis-New York with the other monopoly routes would have likely yielded the expected result- fares rose most in markets where the merger created a monopoly. Anomalies notwithstanding, DOT’S results, as do ours, show unequivo- cally that fares rose most in those markets where TWA and Ozark com- peted before the merger. According to D&S data, comparing the first half of 1985 with the same period in 1988 reveals that fares rose 17.7 percent on routes where the merger created a monopoly (excluding New York) and 39 percent on routes where TWA and Ozark competed along with other airlines. On routes where they did not compete before the merger, fares rose by only 1.0 to 1.5 percent. Among the factors that might be mitigating fare increases out of St. Louis is competition from a low fare competitor, Southwest Airlines. DOT’S data showed that fares rose least on routes where Southwest competed. The results of the Simat, Helliesen & Eichner @H&E)study, undertaken The Simat, Helliesen & for the Air Transport Association, were presented to the Senate Com- Eichner Study for the mittee on Commerce, Science, and Transportation on the same day that Air Transport we testified before the Committee on our preliminary findings on air fares at concentrated airports. The SH&Estudy examines changes in air Association fares and service at 30 hub and 30 nonhub airports since 1980 and attempts to show that fares are not generally higher at hub airports and that where they are higher it is the result of better service and other factors. Following the testimony presented at the hearing, the Ranking Minority Member asked us to undertake an assessment of SH&E’S findings. We Page 55 GAO/RCED-90-102 Fares and Service at Major Airports Chapter 4 Other Recent Analyses of Fares at Concentrated Airports reviewed the SH&Ereport and submitted our findings for the hearing record. We found that SH&E'Sdata were consistent with our finding that fares are higher at airports where one or two carriers handle most of the enplanements. W ith respect to the causes of the fare differences, how- ever, we found that SH&E'Sanalysis contained some serious methodolog- ical problems and, therefore, cannot be used to disprove the hypothesis that fares and airport dominance are related. While we compared fares at concentrated airports with fares at uncon- centrated airports, SH&Econtrasted fares at 30 hub airports with fares at 30 nonhub airports. Although many hubs are dominated by one or two carriers, others are not. Some of the unconcentrated airports making up our comparison group are hub airports for some of the smaller airlines. We do not believe that hubs, per se, are the issue. Indeed, we recognize that hub and spoke networks have given many travelers greater choice in how to make their trips. More alternatives mean more competition and can lead to lower fares for those traveling through hubs. By choosing to examine whether fares are higher at hub airports, SH&E grouped airports where the hubbing airline has significant market power with airports where the hubbing carrier wields much less market power. Several of the hub airports that SH&Eincluded in its analysis are not ones where one would expect fares to be above average. Some of these hub airports are not highly concentrated-that is, they are not dominated by the hubbing airline. Other hubs included by SH&Eare located in cities with more than one airport. In multi-airport cities, com- petition from carriers operating out of other airport(s) could offset some of the advantages a dominant carrier might have at the concentrated airport. For example, SH&Eincluded Dallas Love Field as a hub airport. Love Field qualifies as concentrated since Southwest Airlines handles 100 percent of the enplaning passengers there. However, Southwest’s ability to exploit its monopoly position at Love Field is limited by com- petition from carriers serving Dallas/I% Worth International Airport. We excluded from our analysis airports in multi-airport cities because airport dominance will be less important if alternative air service is available from a nearby airport. SH&E'Sdata show that fares were above average at 21 hub airports and below average at 9 other hub airports. However, the hub airports where fares were lower are all either located in multi-airport cities or are not concentrated. When airports in multi-airport cities (Dallas Love Field, Page 56 GAO/RCFD90-102 Fares and Service at Major Airports . Chapter 4 Other Recent Analyses of Fares at Concentrated Airports Houston Hobby, Detroit Metropolitan Wayne County,” and Chicago Midway) and unconcentrated hub airports (Las Vegas, Kansas City, Mil- waukee, Orlando, and Phoenix)-are excluded, SH&E’S data show that average fares were higher than the industry average at all of the con- centrated hub airports. Six of the hubs that, according to SH&E, have fares below the national average were those of smaller carriers (Southwest at Dallas/Love Field and Houston/Hobby, America West at Las Vegas and Phoenix, Braniff at Kansas City,i and Midway at Chicago/Midway). These carriers either dominated traffic at the smaller airport in a multi-airport city (South- west and Midway) or did not dominate traffic at their hubs. Even if these carriers did dominate an airport, they might not be able to exercise the same kind of market power as the larger dominant airlines because they lack the ability to erect effectively the same entry barriers as the larger airlines. For example, the frequent flyer programs of the smaller carriers are not as attractive as those offered by the majors because the smaller carriers do not fly to as many places. The less exten- sive route systems of the smaller carriers limit the traveler’s ability to earn free travel and offer limited opportunities for spending the bonuses. Once the distinction between hubs and concentrated market power is made, we find that SH&E’S data are consistent with our finding that air fares are higher at concentrated airports. SH&E presented the results of an econometric model and claimed its results show that airport dominance is not a significant factor in explaining the variation in air fares. However, our review of SH&E’S model indicated that it contained serious methodological problems that invalidated its results. SH&E employed multiple regression analysis to estimate how air fares (as measured by a fare index) are affected by airport dominance (as mea- sured by the Herfindahl-Hirschman Index, or HHP ) and other factors. “Detroit was also included in our analysis, but it has become a multi-airport city since Southwest Airlines commencedoperations out of Detroit City Airport in July 1988. rBraniff has since declared bankruptcy. “The HHI is the sum of the squares of the market shares of the fii in the market. Thus, the highest possible HHI is 10,000 when one firm controls 100 percent of the market (100 X 100 = 10,000). This measure assigns a higher value to the situation where one fii dominates a market than to cases where the firms have relatively equal shares. For example, the HHI for a market where three firms held 80, 10, and 10 percent shares would be 6,600, while a 40,30, and 30 percent distribution would produce an HHI of 3,400. Page 57 GAO/RCED90-102 Fares and Service at Major Airports Chapter 4 Other Recent Analyses of Fares at Concentrated Airports Multiple regression analysis attempts to explain the variation in a dependent variable (in this case air fares) by correlating it with the vari- ation in the independent variables (in this case service levels, concentra- tion factors, load factor, et al.) that are thought to explain the behavior of the dependent variable. SH&Efound that the relationship between air fares and airport dominance (the HHI) was not statistically significant when the influence of other factors, especially service quality, was taken into account. However, the model developed by SH&Esuffered from a serious methodological problem: the independent, or explana- tory, variables were not independent of each other. If the independent variables in an econometric model are not truly inde- pendent but are instead highly correlated with each other, then the regression model is unable to separate the individual effects and, there- fore, to elicit much confidence in the results. In general, whenever two or more closely related explanatory factors are in the regression model, the results will not show clearly which of them has the most significant impact on the dependent variable. Statisticians call this condition mul- ticollinearity, and when it exists, the measures of statistical significance are biased toward concluding nonsignificance. SH&Ejustified excluding the variable measuring airport concentration on the finding that it was statistically insignificant. SH&Eargued that high fares are caused not by high airport concentration levels, but by other factors, such as high levels of service quality. High levels of service are costly to produce, and people are willing to pay more for this service. SH&Eincluded several variables that were sup- posed to be proxies for service quality, but these variables were all highly correlated with each other and with the variable representing concentration. In addition to multicollinearity, we identified several other problems with SH&E'Sanalysis. These included problems with variable measure- ment, database development, and the treatment of airlines that are owned in common. By not taking into account multi-airport cities in its analysis, SH&E'S variable representing concentration and market power was poorly mea- sured. For example, Dallas/Love Field has the highest HHI, a perfect 10,000, but Southwest Airlines’ market power is limited by competition from carriers serving the larger Dallas airport. By assigning a high value for the concentration variable to observations where concentration will Page 58 GAO/RCED-90-102 Fares and Service at Major Airports -_I Chapter 4 Other Recent Analyses of Fares at Concentrated Airports have little impact, the SH&E model undercut the variable’s explanatory power. As we did, SH&Eused DOT’SData Bank 1A in its analysis. When using Data Bank lA, it is necessary to edit the fare data to exclude errone- ously recorded fares that are obviously too high or too low. In preparing our analysis, we developed a new edit procedure, which DOThas endorsed and which at least one airline data vendor has adopted. SW&E, probably unaware that we had developed this new screen, applied instead a fare screen that excluded only fares over $2 per m ile and $0 fares. Our fare screen, based on published fare data, recognizes the dis- tance taper in airline fares. A comparison between our fare screen and SH&E’Ssuggests that SH&E’Sscreen allows many fares into the database that are too high, especially for longer distances, while it excludes some valid short distance fares (see table 4.2). Table 4.2: Comparison of GAO and SH&E Origin-Destination Data Fare Screens SH&E screen GAO screen Exclude if yield is Exclude if yield is equal to greater than less than greater than Mileage category cents/mile cents/mile cents/mile cents/mile l-100 0 200 8 300 lOl- 200 0 200 4 255 201-300 0 200 3 160 301-400 0 200 3 125 401-500 0 200 3 115 501-700 0 200 3 105 701-1,000 0 200 3 80 1,001-l ,300 0 200 3 65 1.301-1.600 0 200 3 55 1,601-l,900 0 200 3 50 1,901-2,200 0 200 3 40 2,201-2,500 0 200 3 40 above2.500 0 200 3 40 Finally, we treated as a single carrier airlines that were jointly owned. We reasoned that if one airline was owned by another, it would not be expected to compete with the parent. SH&Etreated jointly owned airlines as different carriers. This was a problem in cities such as Syracuse where USAir and Piedmont, which were jointly owned and in the pro- cess of being merged, each had a large market share. Classifying an air- port such as Syracuse as more competitive than it actually is tends to Page 59 GAO/RCED90-102 Fares and Service at Major Airports Chapter 4 Other Recent Analyses of Fares at Concentrated Airports bias the results. If concentration is a factor explaining airline fares, then misrepresenting competitive conditions will produce misleading results. Analysts at the Antitrust Division of the Justice Department (DOJ) have Study by Justice recently completed an examination of the effects on fares and service of Department Analysts two airline mergers (approved by DOT but opposed by Justice)-TWA’s acquisition of Ozark Air Lines and Northwest’s acquisition of Republic Airlines. Justice’s analysts attempted to measure the impact of the mergers on fares and service at St. Louis, Minneapolis/St. Paul, and Detroit. The Justice Department analysts attempted to isolate the effect of the merger, taking into account other factors such as costs, the presence of potential entrants, and other variables.” The DOJ analysts do not com- pare yields at the concentrated airports to yields at other airports, but rather attempt to predict how much higher (or lower) yields are due to the mergers. The DOJ analysts found that the Northwest-Republic merger affected fares at Minneapolis/St. Paul by roughly the amounts DOJ had predicted when it opposed the merger. According to the DOJ analysts, fares at Min- neapolis/St. Paul, where Northwest and Republic both operated hubs and had a number of overlapping routes, were 5.6 percent higher because of the merger. The DOJ analysts found this to be a significant increase. On long distance routes (more than 1000 miles) where North- west and Republic competed before the merger, yields were 7.5 percent higher. On long distance routes where they were potential competitors, yields were estimated to be 7.6 percent higher because of the merger. On shorter distance routes the impact of the merger on fares was smaller but still pronounced. Yields on routes out of Minneapolis/St. Paul where they did not compete directly rose only 2.9 percent. At Detroit, on the other hand, only Republic operated a hub before the merger, and fares did not rise following the merger but in fact fell slightly, by 0.8 percent. The TWA-Ozark merger also produced an effect at St. Louis similar to that predicted by DOJ when it opposed that merger. Although the overall “The Justice Department analysts employed regression analysis to estimate predictive equations that both forecast and backcast yields on routes out of the dominated airports The forecasting equation predicts what yields would have been in the period after the mergers had they not occurred. The backcasting equation predicts what yields would have been before the mergers had they already occurred. The predictions are combined to form a single estimate of the effect of the mergers on yields of a particular city-pair market. Page 60 GAO/RCED90-102 Fares and Service at Major Airports Chapter 4 Other Recent Analyses of Fares at Concentrated Airports estimated impact on airline yields was relatively small-an increase of 1.5 percent-yields did increase significantly-about 4.5 percent-on routes where TWA and Ozark competed directly or where they were at least potential competitors. Fares fell on other routes, causing the overall effect to be small. While, in many ways, DOJ’S approach differs from the approach we fol- lowed, its findings are consistent with, and supportive of, the ones presented here. Yields are higher at concentrated airports and increased concentration leads to higher fares. recently published the results of a g-month study of the state of DOT Task Force on DOT competition in the nation’s airline industry. DOT’S results are consistent Competition Study with those from most other analyses in that the Department concludes that, on balance, deregulation is working. Air fares are lower, service levels have increased, and greater numbers of people can afford to fly. However, DCK also found “pockets of problems,” including higher fares at concentrated air traffic hubs.l” DOT focused on air traffic hubs and defined a hub to be concentrated if one carrier had more than 75 percent of the enplanements. Under this criterion, eight hubs were judged to be concentrated-Charlotte, Cincin- nati, Dayton, Memphis, Minneapolis/St. Paul, Pittsburgh, St. Louis, and Salt Lake City. DOr also identified 8 two-carrier concentrated hubs- Atlanta, Chicago, Dallas, Denver, El Paso, Houston, Nashville, and Raleigh-Durham. Thus, DOT’S analysis includes 12 of the 15 concentrated airports we examined. *I DOT’S approach differed from ours in a number of respects. The task force compared yields at the concentrated hubs to the industry average, rather than a control group of less concentrated hubs. DOT calculated a fare premium, adjusted for distance and density factors, for concen- trated single-carrier hubs, concentrated two-carrier hubs, and monopoly routes. DOT also calculated the fare premium for the 15 concentrated air- ports used in our analysis (see table 4.3). “‘Air traffic hubs are communities, rather than airports, accounting for a certain percentageof the nation’s travel. For example, the Chicago hub is served by O’Hare and Midway airports. ’‘Because we excluded airports in multi-airport cities, each of these twelve airports accounted for all traffic at the hub it served. Page 61 GAO/RCEXb90-102 Fares and Service at Major Airports Chapter 4 Other Recent Analyses of Fares at Concentrated Airports Table 4.3: DOT-Calculated Fare Premiums on Monopoly Routes and at Market type Average fare premium Percent premium Concentrated Airports in 1988 Monopoly Routes $16.59 14.0% 8 Concentrated Hubs (Srnale carrier) $22.30 18.7% 8 Concentrated Hubs (Two earners) $10.42 8.9% 15 GAO Concentrated Hubs $21.44 18.4% DCTcalculated fare premiums for both single-carrier and two-carrier concentrated hubs and compared the 1988 premium with that of 1984. The single-carrier concentrated hubs show the largest premiums. On average, two-carrier concentrated hubs show premiums about half as great. However, Atlanta, which is a two-carrier hub, has the largest pre- mium of any concentrated hub (see table 4.4). Table 4.4: DOT-Calculated Fare Premiums at Concentrated Hubs, 1984 and 1988 Premiums in percentage, Shares In percentage 1988 1984 Dominant carrier Dominant carrier Hub Premium enolanement share Premium enolanement share Single Carrier. Charlotte 27.1 90 22.7 75 Cincinnati 34.1 78 29.5 56 Dayton 17.3 75 10.2 63 Memphis 28.8 86 28.1 47 Minneabolis/St.Paul I I 19.7 78 12.0 48 Pittsburah 10.4 86 16.3 77 St. Louis 17.8 82 16.4 58 Salt Lake City 16.7 80 9.9 71 Two Carrier: Atlanta 40 2 93 38.8 93 Chicaao -1.2 72 27.5 68 Dallas 18.5 79 9.8 68 Denver -5.4 85 -6.0 65 El Paso -18 0 73 -30.5 82 Houston 6.7 76 -9.0 51 Nashville 10.3 71 17.4 38 Raleiqh-Durham 9.6 80 11.9 52 Page 62 GAO/RCED-90-102 Fares and Service at Major Airports Chapter 4 Other Recent Analyses of Fares at Concentrated Airports DOT found that average fares were greater in most distance and density categories for single-carrier concentrated hubs, and that the most signif- icant premiums were in markets of more than 100 passengers per day and for distances ranging between 250 and 1000 miles. Passengerstrav- eling in these distance- and density-market categories paid 71 percent of the total 1988 premiums at the 8 single-carrier concentrated hubs. In fact, fares in dense markets at the concentrated hubs were frequently higher than fares in less dense markets even for the same distance cate- gories. This is the reverse of what normally happens in the airline industry. Scale economies and competition usually result in lower fares in densely traveled markets. The DOT task force report further buttresses the finding that fares are higher at concentrated airports. Moreover, the study contains consider- able information on airline operating and marketing practices that limit market entry and protect dominant, incumbent positions. Yet, the study does not explore any policy options that DOT or the Congress might con- sider to address the limits to competition.‘” DOT’S recently released National Transportation Plan calls for increased spending to expand the capacity of the aviation system. I3Yet, the concentrated airports that are experiencing higher fares are not all capacity constrained, and the bar- riers to market entry identified in Chapter 2 will not be eliminated by building more runways and new terminals. “For a discussion of the pros and cons of various policy options, see app. II, which contains an excerpt from Barriers to Competition (GAO/T-RCED-89-65). See also Competition in the Airline Computerized Reservation System Industry (GAO/T-RCED-88-62, Sept. 1988) and Effects of Airline Entry Barriers on Fares (GAO/T-RCED-90-62, April 1990). Copies of these publications can be obtained by writing or calling GAO (see information on the inside of back cover of this report). ‘“Moving America: New Directions, New Opportunities, U.S. Department of Transportation (Wash- ington, DC.: Feb. 1990). Page 63 GAO/RCED-90-102 Fares and Service at Major Airports Chapter 5 Changesin Air PassengerService at 15 ConcentratedAirports In addition to raising concerns over higher fares, airport dominance by one or two airlines has prompted concern that service levels could decline at concentrated airports. Increased market power, combined with effective barriers to new entry, could cause reduced service levels at the concentrated airports as the dominant carriers discover they have less need to respond to competitive pressures by offering high levels of service. The term “service levels” can be used to mean either the quantity of service available or the quality of service delivered. The quality of ser- vice delivered includes such things as the quality of in-flight meals, the friendliness of ticketing and on-board personnel, the percentage of late flights, and the amount of lost baggage. We did not examine trends in these qualitative attributes of air travel. Some, such as the quality of the food served, are simply too subjective, while for others, such as on-time performance, the data were either unavailable for appropriate time periods or did not allow meaningful comparisons. Instead, we examined trends in the quantity of service available. Specif- ically, we examined trends in the number of routes served directly, the number of flights, and the number of airlines competing for traffic on routes out of the concentrated airports. Direct service includes both non- stop service to destinations and service with stops but not requiring the passenger to change planes. To assess changes in service levels at concentrated airports, we com- pared service level data for the month of May from 1985 through 1988. We compared the number of cities that could be reached by direct ser- vice, the total number of daily flights to all places, and the amount of competition as measured by the number of markets served by one car- rier, by two or three carriers, or by four or more carriers. The service data for carriers operating out of the 15 concentrated air- ports between May 1985 and May 1988 show an overall increase in the amount of service offered and in the number of places that can be reached by direct air service. However, at most of the airports affected by mergers, the number of daily flights decreased. In addition, the amount of competition declined on many routes out of the 15 concen- trated airports. More routes were served by only one carrier and fewer could be considered highly competitive. We considered routes highly competitive if four or more carriers provided direct service. Page 64 GAO/RCED-90-102 Fares and Service at Major Airporta . Chapter 5 Changes in Air Passenger Service at 15 Concentrated Airports The number of destinations that can be reached by direct air service Number of from the 15 concentrated airports increased at 10 of the airports, Destinations Served declined at 3 of the airports, and remained about the same at the other 2 Directly (changed less than 5 percent). Overall, the number of places served directly increased 10 percent, but there was considerable variation among the 15 airports. The improvement was most pronounced at air- ports where the airlines established or built up hubs during this period. At airports in Charlotte, Cincinnati, Detroit, Memphis, Nashville, Raleigh-Durham, and Syracuse hubs were established or built up between 1985 and 1988, and all except those in Charlotte and Detroit registered large increases in the number of destinations that could be reached with direct air service. Table 5.1 shows data on the number of routes with direct service at the 15 concentrated airports. Table 5.1: Number of U.S. Destinations With Direct Service From 15 Percentage Concentrated Airports During Month of change May 1985-88 Concentrated airport 1985 1986 1987 1988 1985-88 Atlanta 152 151 143 150 -1 Charlotte 85 88 94 91 7 Cincinnati 74 82 89 102 38 Dayton 50 69 63 60 20 Denver 158 158 151 147 -7 Detroit 121 117 121 125 3 Greensboro 53 46 45 48 -9 Memphis 92 97 101 106 15 MlnneapolQSt Paul 123 133 124 134 9 Nashville 62 81 85 80 29 Pittsburgh 114 111 116 128 12 Ralelah-Durham 55 54 59 85 55 Salt Lake City 86 84 91 102 19 St. Louis 136 132 131 126 -7 Syracuse 51 57 62 69 35 Total 1412 1460 1475 1553 10 aPercentage changes rounded to nearest whole number Source I P Sharp, Inc Raleigh-Durham experienced the largest increase. Because American Airlines established its hub there, the number of cities served directly from Raleigh-Durham increased from 55 in May 1985 to 85 in May 1988, a 55 percent increase. Of the airports where hubs were set up or expanded during this period, Detroit and Charlotte registered the Page 65 GAO/RCED90-102 Fares and Service at Major Airports Chapter 5 Changes in Air Passenger Service at 15 Concentrated Airports smallest improvements, with increases of just 3 percent and 7 percent, respectively, in the number of destinations served directly. Greensboro, the only concentrated airport that we examined which was not a hub airport, experienced the largest relative reduction in the number of places served directly-a 9 percent decline-while Denver, a two-carrier hub, experienced the largest absolute reduction in the number of places served directly-l 1 fewer destinations could be reached without changing planes in 1988 than in 1985. Also 10 fewer places were served directly from St. Louis in 1988 than in 1985. Number of Daily percent between May 1985 and May 1988, but 7 of the 15 concentrated Departures airports experienced a decline in the average number of daily flights. Departures out of Raleigh-Durham more than doubled while the number of flights from Nashville and Dayton increased more than 50 percent. On the other hand, the number of daily flights out of Denver fell 20 percent, while flights out of Minneapolis/St.Paul, Detroit, and Memphis-air- ports affected by the Northwest-Republic merger-fell between 10 and 16 percent. W ith respect to the number of flights offered, the patterns for the domi- nant and the nondominant firms differed. Daily departures offered by the dominant carriers grew 50 percent, and the proportion of total departures accounted for by the dominant firms grew by 46 percent. * Conversely, the number of daily departures out of these 15 airports by the nondominant carriers fell almost 50 percent. Table 5.2 provides flight frequency data for the 15 concentrated airports. Page 66 GAO/RCED-90-102 Fares and Service at Major Airports Chapter 5 Changes in Air Passenger Service at 15 Concentrated Airports Table 5.2: Number of Daily Flights to U.S. Destinations From 15 Concentrated Percentage Airports During the Month of May 1985- Number of daily flights change 88 Concentrated airport 1985 1986 1987 1988 1985-88b Atlanta 1162 1215 1193 1207 4 Nashwile 170 258 292 269 58 Charlotte 409 434 402 466 14 Cincinnati 284 264 340 405 43 Dayton 138 209 207 215 56 Denver 900 887 805 718 -20 Detroit 635 648 572 569 -10 Greensboro 128 110 111 108 -16 Memphis 413 404 375 346 -16 Minneapolis 563 615 507 506 -10 Pittsburgh 512 520 523 559 9 Raleigh-Durham 142 154 160 286 101 Salt Lake City 299 292 323 341 14 St. Louis 646 699 608 605 -6 Syracuse 175 162 177 169 -3 TotaP 6578 6872 6595 6769 3 Note, Dally flight data calculated from monthly data and rounded to nearest whole numbers Tolumns may not add to totals due to rounding bPercentage changes rounded to nearest whole number Source: I.P Sharp, Inc. As the data on service levels suggest, the dominant carriers at the con- Degreeof Competition centrated airports have increased their dominance over the past few years. As a result, on many routes out of the 15 concentrated airports the air traveler’s ability to choose among airlines has narrowed. Although there was an overall increase of 10 percent in the number of domestic destinations served directly from the 15 concentrated airports between 1985 and 1988, there was a 25 percent increase in the number of domestic destinations from the 15 concentrated airports that were served directly by only one carrier. Of the 15 concentrated airports we examined, 13 experienced an increase in the number of monopoly routes. One of the two exceptions was Atlanta, where concentration did not change greatly over the 1985-88 period. Atlanta is a hub for two carriers, Delta and Eastern Airlines. However, since the Eastern Airlines strike and subsequent bankruptcy, Eastern has substantially reduced its operations out of Atlanta. Therefore, the one airport that did not experi- ence an increase in one-carrier routes between 1985 and 1988 has seen a Page 67 GAO/RCED-90-102 Fares and Service at Major Airports Chapter 5 Changes in Air Passenger Service at 15 Concentrated Airports substantial decline in the activity of the principal competitor. Table 5.3 provides data on the number of routes where there has been a monopoly on direct service. Table 5.3: Number of U.S. Destinations to Which Only One Airline Flew Directly Percentage From 15 Concentrated Airports in Month change of May 1985-88 Concentrated airport 1985 1986 1987 1988 1 985-88a Atlanta 70 58 55 60 -14 Nashville 40 51 49 50 25 Charlotte 46 52 72 74 61 Cincinnati 43 47 53 66 53 Dayton 36 43 41 42 17 Denver 62 65 65 74 19 Detroit 59 51 67 65 10 Greensboro 39 33 33 38 -3 Memphis 51 61 75 86 69 Mrnneapolis 64 66 64 88 38 Pittsburah 81 78 89 90 11 Raleigh-Durham 39 33 39 50 28 Salt Lake City 52 56 64 72 38 St. Louis 80 69 87 84 5 Syracuse 2% 3% 39 46 64 Total 790 801 912 985 25 aPercentagechanges rounded to nearest whole numbers Source: I.P. Sharp, Inc. In some cases, the increase in the number of direct routes served by a single carrier may simply reflect added service by an airline at its hub and, as such, represents a net improvement in service offerings and traveler welfare. One carrier offering direct service is better than no direct service at all. However, at nine of the concentrated airports, the increase in the number of direct routes served by only one carrier is greater than the increase in the number of routes with direct service. If more routes are served by one carrier, and there is no corresponding increase in the total number of routes served directly, then it follows that fewer routes are served by two or more carriers. For example, at Charlotte there were 6 more routes with direct service in May 1988 than in May 1985. However, the number of direct routes served by only one airline grew by 28, from 46 to 74 routes. For the 15 concentrated airports as a group, no change occurred in the number of routes served by 2 or 3 airlines between 1985 and 1988, since Page 68 GAO/RCED90-102 Fares and Service at Major Airports Chapter 5 Changes in Air Passenger Service at 15 Concentrated Airports Table 5.5: Number of U.S. Destinations to Which Four or More Airlines Flew Percentage Directly From 15 Concentrated Airports change in Month of May 1985-88 Concentrated airport 1985 1988 1987 1988 1985-88= Atlanta 8 14 4 9 13 Nashville 1 7 4 2 100 Charlotte 5 4 0 1 -80 Cincinnati 4 1 1 1 -75 Dayton 1 3 3 0 -100 Denver 23 27 9 12 -48 Detroit 23 27 9 14 -39 Greensboro 3 2 2 0 -100 Memohis a 6 1 0 -100 Minneapolis 12 12 6 4 -67 Pittsburgh 1 3 3 3 200 Raleigh-Durham 2 1 1 1 -50 Salt Lake CitvI 3 3 2 2 -33 St. Louis 10 12 2 2 -80 Syracuse 2 0 0 0 -100 Total 106 122 47 51 -52 aPercentagechanges rounded to nearest whole numbers. Source: I.P. Sharp, Inc Other studies examining air fares also have examined trends in service Other Analyses of levels at concentrated airports. The approaches taken and the perform- Service Changesat ante measures examined are usually different from ours, and so the results of these other analyses are not strictly comparable. None of the ConcentratedAirports studies we reviewed contradicts our results. DOT St. Louis Study The Ranking Minority Member of the Senate Committee on Commerce, Science, and Transportation asked D(X to respond to our findings on fare and service changes reported in our September 1988 testimony before the Committee.1 Our analysis of service changes at Lambert-St. Louis International Airport following the TwA-Ozarkmerger compared service levels in June 1986 with those prevailing in June 1987. We found that TWA had increased the number of places served directly or with nonstop service from St. Louis. In 1987, TWA provided direct service to six more cities and offered nonstop service to four more cities than TWA and ‘Factors Affecting Concentration (GAO/T-RCED-88-65); GAO report on fare and service changes at St. Louis(GAO/RCED-88-217BR). Page 70 GAO/RCED-90-102 Fares and Service at Major Airports Chapter 5 Changes in Air Passenger Service at 15 Concentrated Airports increases at some were offset by reductions at others. However, no clear patterns emerged among gainers or losers. Table 5.4 provides data on the number of routes where two or three carriers provide direct service. Table 5.4: Number of U.S. Destinations to Which Two or Three Airlines Flew Percentage Directly From 15 Concentrated Airports change in Month of May 1985-88 Concentrated airport 1985 1988 1987 1988 1985-88 Atlanta 74 79 84 81 9 Nashville 21 23 32 28 33 Charlotte 34 32 22 16 -53 Clnclnnati 27 34 35 35 30 Davton 13 23 19 18 38 Denver 73 66 77 61 -16 Detroit 39 39 45 46 18 Greensboro 11 11 IO IO -9 Memphis 33 30 25 20 -39 Minneapolis 47 55 34 42 -11 Plttsburah ---.-- a 32 30 24 35 9 Raleigh-Durham 14 20 19 34 143 Salt Lake Cltv 31 25 25 28 -10 St. LOUIS 46 51 42 40 -13 Syracuse 21 19 23 23 10 Total 516 537 516 517 0 aPercentage changes rounded to nearest whole numbers Source: I.P. Sharp, Inc. On the other hand, the number of routes served by 4 or more carriers fell 52 percent. Although there are fewer routes in this category than in the single-carrier and two- or three-carrier categories, the results are the least ambiguous. In May 1988, four airports had no routes where 4 or more carriers competed, and most of the others registered dramatic reductions. The three airports where the number of routes served by 4 or more carriers increased gained only 4 such routes while the other 12 airports lost 59 routes served by 4 or more carriers. Table 5.5 provides data on the number of routes with direct service by at least 4 carriers. Page 69 GAO/RCED-90-102 Fares and Service at Major Airports Chapter 5 Changes in Air Passenger Service at 15 Concentrated Airports other carriers, 10 continued to receive competitive nonstop service in 1988. The exception was New York where People Express ceased pro- viding service after it was taken over by Texas Air. TWA reduced weekly flights in these markets by almost 15 percent while all carriers reduced flights about 11 percent. Some of TWA'S service reductions were offset by the expanded operations of other airlines. In 36 markets where either TWA or Ozark provided the only nonstop ser- vice before the merger, TWA continued to serve all but two following the merger. TWA provided a total of 850 nonstop flights each week in 1988, while TWA and Ozark together provided 760 flights per week in 1986. In addition, DOT found two markets that were no longer monopolized, Phoenix and Cincinnati. Finally, in the 5 markets where either TWA or Ozark competed with others before the merger, DOT reported that TWA had increased nonstop service by 29 percent while the other carriers serving these markets had reduced service by 19 percent. The primary difference between DOT'S analysis of the merger impacts at St. Louis and our study is that we did not report changes in the number of flights. DOT recorded a 7 percent decline in nonstop departures between 1986 and 1988. Although we did not present them in either our testimony or our subsequent report, we did collect data on the number of departures. Our data showed a 6 percent decline between 1985 and 1988. W ith respect to the merging carriers, Dm found that departures declined 8 percent between 1986 and 1988. Our data show a 6 percent decline over that period. However, TWA'S departures in 1988 were 19 percent higher than TWA and Ozark’s combined in 1985. Study by Justice The recent study by Justice Department staff also noted reductions in the volume of competitive service at some of the concentrated airports.2 Depar tment Staff The Justice study examined changes in the number of departures, avail- able seats, and the number of cities receiving nonstop service out of two airports where a merger eliminated a major competitor, St. Louis and Minneapolis/St.Paul. Justice had opposed the TWA-Ozarkand the North- west-Republic mergers, both of which eliminated a major competitor at these airports. This study also examined Detroit, a hub airport for only one of the merging partners, Republic. The approach in our analysis of the available service data differs in many ways from that of the Justice study. These differences largely 'Werden,Joskow,andJohmon. Page 72 GAO/RCEJI-90-102 Fares and Service at Major Airports Chapter 5 Changes in Air Passenger Service at 15 Concentrated Airports Ozark combined before the merger. On the other hand, other carriers serving St. Louis reduced the number of points served directly from 83 in 1986 to 66 in 1987. They also reduced by seven the number of places served nonstop. Our analysis also showed that there was a decline in competition on routes out of St. Louis. Table 5.6 provides DOT data on the competitive status of direct service routes out of St. Louis. Table 5.6: Competition Among Carriers Providing Direct Service From St.Louis Number of routes served June Percentage Number of carriers 1986 1987 change Four or more carriers 15 7 -53 Three or more carriers 30 16 -47 Two or more carriers 64 36 -44 One carrier 60 85 42 Total routes with direct service 124 121 -2 The report prepared by DOT, at the request of the Ranking Minority Member of the Senate Committee on Commerce, Science, and Transpor- tation, reexamined the question of service changes at St. Louis. Rather than focus on the periods immediately preceding and following the merger, Dm compared service levels in June 1986 with those prevailing in June 1988. DOT also examined nonstop rather than direct service. DOT reported almost no change in the number of cities receiving nonstop jet service from St. Louis. Some cities lost nonstop jet service from St. Louis, but others were added. Table 5.7 summarizes D&S findings on service changes. Table 5.7: Service Changes at St. Louis for Large Jet Carriers (Domestic) June Weekly departures Weekly seats 1986 vs. June 1988 June June Carrier 1986 1988 1986 1988 TWA/Ozark 2541 2331 347,160 316,886 Other Carriers 579 574 68,690 69,852 Total 3120 2905 415,850 386,738 According to DOT, some carriers, especially Southwest Airlines, increased service out of St. Louis since the merger. DOT also analyzed service changes using the same categories of market competition we devised to analyze fares. In the 18 markets where TWA and Ozark were the only carriers providing nonstop jet service before the merger, TWA reduced service in 16 markets and offered 26 percent fewer weekly flights in 1988. In the 11 markets where TWA and Ozark competed along with Page 71 GAO/RCED90-102 Fares and Service at Major Airports Conelusions There is general agreement that the increased competition spawned by deregulation has led to an overall reduction in real airline fares, that is, fares adjusted for inflation. However, it is also true that fares have fallen more in some markets than in others and, in some markets, fares probably are higher today than they would have been had the industry continued to be regulated. Under regulation, airline fares did not accurately reflect the differences in the costs of serving different markets. As a rule, passengers flying on heavily traveled, long-distance routes paid fares higher than those nec- essary to cover costs, and their fares subsidized passengers in lightly traveled, short-distance markets. Deregulation permitted the airlines to set fares without obtaining prior government approval and has allowed the airlines to rationalize the fare structure and make the pricing of air travel more economically efficient. Rationalizing air fares by eliminating or reducing the previous subsidies has meant that, in general, fares are now relatively higher in short distance markets than in long distance ones than they were under CABregulation. It is also true that airline fares are higher today in both real and nominal terms than they were in 1986, the peak of the recent wave of mergers involving some of the nation’s largest airlines. Many airline industry analysts believe that fares in 1986 were too low and were not consistent, in the long run, with a financially healthy industry. These analysts claim that fares had to rise if firms in the industry were to earn ade- quate rates of return. While air fares have risen since 1986, real fares are still well below 1979 levels. Thus, in some markets, higher fares might be consistent with improved economic efficiency, and higher fares industry-wide than those pre- vailing in 1986 may be necessary if carriers are going to earn sufficient revenues to buy new planes and provide investors an adequate return on their investments. Congressional concern over higher fares has cen- tered on fare increases that reflect growing market power, not with those that reflect cost differences. Our review focused on trends in fares and service at 15 airports around the nation dominated by one or two carriers. We found that the yields earned by the dominant airlines at these concentrated airports were con- sistently higher than yields at a comparison group of unconcentrated airports, and that at most of the concentrated airports the yields received by the dominant carriers were considerably higher than the yields earned by the other airlines serving those airports. According to Page 74 GAO/RCED-90-102 Fares and Service at Major Airports . Chapter 6 Changes in Air Passenger Service at 15 Concentrated Airports relate to the different purposes of the studies. Nevertheless, despite the dissimilarities of purpose and approach, common threads link the con- clusions. Justice found a large increase in the number of cities receiving nonstop service from Minneapolis and a somewhat smaller increase out of Detroit following the Northwest-Republic merger. Justice found no increase at St. Louis while we recorded an increase. Table 5.8 shows our service data using nonstop, instead of direct operations. Table 5.8: Number of Nonstop Destinations From 15 Concentrated Percentage Airports During Month of May 19851988 change Concentrated Airport 1985 1988 1987 1988 1985-l 988 Atlanta 118 115 117 114 -3 Nashville 28 48 52 49 75 Charlotte 61 64 69 76 25 Cincinnati 44 42 56 66 50 Dayton 37 42 40 40 8 Denver 102 105 108 108 6 Detroit 63 69 75 75 19 Greensboro 22 21 20 19 -14 Memphis 61 69 82 83 36 Minneapolis 71 79 89 90 27 Pittsburgh 88 91 92 96 9 Raleiah-Durham 30 28 26 52 73 Salt Lake City 49 55 63 64 31 St. Louis 87 91 99 100 15 Syracuse 25 24 28 29 16 Total 888 943 1016 1061 20 Percentage changes rounded to nearest whole numbers. Source: I.P. Sharp, Inc Page 73 GAO/RCED-!W102 Fares and Service at Major Airports Chapter 6 Conclusions Our analysis focuses on concentrated airports and how concentration in certain markets might lead to higher fares and to the erosion of the ben- efits of deregulation. In a related study, we are examining which changes in airline operating and marketing practices have resulted in barriers to entry and how such barriers might be reduced or eliminated. We are concerned that, if airline markets become highly concentrated, then the benefits of deregulation to the traveling public might be reduced. We do not believe that our results or the results of other studies show that airline deregulation has failed. Although the analyses we reviewed concur that fares are higher when there is less competition, the conclu- sion we draw from these analyses is that competition must be strength- ened and that barriers to successful competition be reduced. Thus, the issue before the Congress should not be whether the airline industry needs to be reregulated but rather what steps can be taken to revitalize competition in markets where competition has been reduced. Page 76 GAO/RCED96-102 Fares and Service at Major Airports Chapter 6 Conclusions the most recent period for which data were available, the gap between yields at concentrated and unconcentrated airports is widening. In addi- tion, yields increased as concentration increased even at airports that were already highly concentrated. Controlling for differences in average length of haul at unconcentrated airports did not appreciably alter our finding that yields were higher at concentrated airports. The issue of airline dominance and higher fares has been addressed in several other studies over the past 2 years. In some cases, these studies were undertaken in response to our analyses and were designed to counter our finding that fares are higher at concentrated airports. These alternative analyses generally do not contradict our finding that fares are higher at airports where one or two carriers handle most of the enplaning passengers, and whatever differences exist between the results reported by these other studies and our finding can usually be traced to differences in the evaluative methodologies. Some of these other studies attempt to attribute the fare difference to higher service levels, but the one econometric analysis attempting to make this case suffered from serious methodological problems and cannot be relied on to dispute the hypothesis that fares are higher at concentrated airports because of the market power of the dominant airlines. W ith respect to service offerings, we found some increases in the number of places served and in the number of daily flights, but in most cases the increases were on the part of the dominant carrier, offset by reduced offerings from nondominant carriers. In addition, many trav- elers have less choice among airlines as more routes out of the concen- trated airports are being served by only a single airline, usually the dominant carrier. Overall, deregulation has led to lower airline fares for most travelers, and the establishment of hubs has allowed the airlines to realize impor- tant operating efficiencies. While more passengers travel on competitive routes than was the case prior to deregulation, growing concentration, especially at hub airports, has led to fewer competitors on many routes. Over the past few years, numerous mergers and bankruptcies have reduced the number of airlines providing the vast majority of U.S. air passenger service. The mergers and bankruptcies that led to increased concentration cannot be easily undone. At the same time, changes in air- line marketing and operating practices make it more difficult for new airlines to enter the industry or for existing carriers to expand into mar- kets where another carrier already dominates the traffic. Page 75 GAO/RCED90-102 Fares and Service at Major Airports . Appendix I Trends in Fares at Each of the 15 ConcentratedAirports Figure 1.1: Atlanta (Hartsfield Atlanta InternatIonal Airport) 34 Average Yields (Cents per Mile) 32 30 ,,“‘-. 28 *** lQ65 2Q85 3Q05 4085 1086 26186 3086 4086 1087 2087 3Q87 4Q87 100% MB8 3088 4Q8a lQ89 2089 First Quarter 1985 through Second Quarter 1989 - Texas Air m--m Delta m Other Airlines Mama 38 Comparison Airports Note: Texas Air includes yield data for Eastern, Continental, and People Express. Page 78 GAO/RCED-90-102 Fares and Service at Major Airports . Page 77 GAO/RCEIMO-102 Fares and Service at Mejor Airports Appendix I Trends in Fares at Each of the 16 Concentrated Airports Figure 1.3: Cincinnati (Greater Cincinnati International Airport) 34 Average Yields (Cents per Mile) 32 39 28 26 24 22 lQ89 2085 3085 4085 lQ86 2486 3QS6 4QR lQ87 MB7 3Q87 4Q67 lQ98 M99 3Q88 4Qa8 lQ89 2Q99 First Quarter 1985 through Second Quarter 1989 - Delta - Other Airlines 00 8 m 38 ComparisonAh~~fts Page 80 GAO/RCED-90402 Fares and Service at Major Airports Appendix I Trends in Fares at Each of the 16 Concentrated Airports Figure 1.2: Charlotte (Charlotte/Douglas International Airport) 34 Average Yields (Cents per Mile) 32 30 28 26 9A IQ85 2Q85 3Q05 4085 lQ86 2Q86 3086 4488 1087 2067 3Q37 4087 lQ88 2Q88 3Q88 4Q88 lQ89 2Q89 First Quarter 1985 through Second Quarter 1999 - USAir Group m Other Airlines WHM 38ComparisonAirports Note: USAir Group indudes Piedmont and USAir yield data. Page 79 GAO/RCEIMO-102 Fares and Service at Major Airports Appendix I Trends in Fares at Each of the 15 Concentrated Airports Figure 1.5: Denver (Stapleton International Airport) 30 Average Yields (Cents per Mile) 28 25 24 22 20 18 16 12 10 lQ85 2465 3055 4085 lQ86 ZQ86 3086 4Q86 1087 MS7 3Q57 4487 lQ88 2Q88 3Q55 4QBB lass 2Q89 First Quarter 1985 through Second Quarter 1989 - Texas Air -1-1 United m Other Airlines q WHH 38ComparisonAirports Note: Texas Air includes yield data for Continental, Eastern, Frontier, and People Express. Page 82 GAO/RCED-90402 Fares and Service at Major Airporta Appendix I Trends in Fares at Each of the 16 Concentrated Airports Figure 1.4: Dayton (Dayton International Airport) 34 Average Yields (Cents per Mile) 32 30 28 26 24 22 20 18 16 y- - l a** ,~~amammmmmm8mmmmmmmmmmm0 IQ65 2Q85 3Q85 4Q85 lQ86 MB6 3Q86 4Q86 IQ07 2Q87 3Q87 4Q87 lQ88 x288 3Q88 4Q88 1089 2QB9 First Quarter 1985 through Second Quarter 1989 - USAir Group m Other Airlines nnnn 36ComparisonAirpwts Note: USAir Group indudes Piedmont and USAir yield data. Page 81 GAO/RCEKHO-102 Fares and Service at Mdor Airports Appendix I Trends in Fares at Each of the 15 Concentrated Airports Figure 1.7: Greensboro/High Point/Winston-Salem (Pfedmont Triad International Alrport) 34 Average Welds (Cants per Mile) 32 30 28 28 24 22 20 18 ++‘=.m. 16 #* - ~,,r+ I _laaaa*~““‘m' l mmmmmmmammL'~ 1085 2Q85 3Q85 4085 lQ86 2Q86 3086 4Q88 1 Q87 MB7 3087 4Q87 lQ88 2Q88 3088 4088 lQ89 2Q89 First Quarter 1985 through Second Quarter 1989 - USAir Group m Other Airlines n mw n 38 ComparisonAirports Note: USAir Group includes Piedmont and USAir yield data. Page 84 GAO/RCED-SO-102Fares and Service at Major Airports Appendix I Trends in Fares at Each of the 16 Concentrated Airports Figure 1.6: Detroit (Detroit Metropolitan Wayne County Airport) 32 Average Yields (Cents per Mile) 30 28 28 24 22 20 18 16 14 12 lQ85 2Q85 3Q85 4085 lQ88 2Q88 3086 4Q86 lQ87 2Q87 3Q87 4487 lQ88 2Q86 3Q88 4Q88 lQ69 2Q89 First Quarter 1985 through Second Quarter 1989 - Northwest (& Republic) I Other Airlines n m8 n 38 Comparison Airports Note: Northwest includes both Northwest and Republic yield data prior to their merger in late 1986. Page 83 GAO/RCED-90-102 Fares and Service at Major Airports Appendix I Trends in Fares at Each of the 16 Concentrated Airports Figure 1.9: Minneapolis/St. Paul (Minneapolis/St. Paul International Airport) 30 Average Yields (Centsper Mile) 28 26 24 22 20 18 16 14 12 10 lQ8!5 MB5 3485 4G85 1486 2Q89 3886 4486 lQ87 2Q87 3m7 4087 lQ88 2Qa8 3Q88 a68 lQ69 2Q89 First Quarter 1985 through Second Quarter 1989 - Northwest (& Republic) m Other Airlines n mam 38CompatisonAirporte Note: Northwest indudes both Northwest and Republic yield data prior to their merger in late 1996. Page 86 GAO/ICED-90-102 Fares and Service at Major Airporta . Appendix I Trends in Fares at Each of the 16 Concentrated Airports Figure 1.8: Memphis (Memphis International Airport) 34 Average Yields (Cfmts per Mile) 32 30 20 29 24 22 20 18 16 ma!5 2cw 3Q85 4485 lQ86 M86 3Q86 4Q89 lQ87 2487 3Q87 4Q87 lQ88 2488 3088 4Q88 lQ89 2Q89 First Quarter 1985 through Second Quarter 1969 - Northwest (a Republic) m Other Airlines n m W m 38CcmparisonAirports Note: Northwest includes both Northwest and Republic yield data prior to their merger in late 1986. Page 86 GAO/RCELNO-102 Fares and Service at Major Airports Appendix I Trends in Fares at Each of the 16 Concentrated Airports Figure 1.10: Nashville (Nashville Metropolitan Airport) 34 Avenge Yields (Cents per Mile) 32 30 28 25 24 22 20 18 16 4 +,@mmm&r 14 l -..... ,mmmmmmmmmmmmmmmmmmmmmm* •m--8~8mmmmm ,8a* 12 l mmmmmmmmmmmmmmmmmmmmmmmmmm8 1085 2085 3Q85 4Q8!5 IQ86 2Q85 3Q86 4486 IQ87 2Q87 3QB7 4Q87 lQ88 2488 3888 4Q58 IQ89 2Qs9 First Quarter 1985 through Second Quarter 1969 - American m Other Airlines ¤mm~ 38 ComparisonAirports Page 87 GA0/BcED-9o-102 Fares and Service at Major Airports Appendix I Trends in Fares at Each of the 15 Concentrated Airports Figure 1.11: Pittsburgh (Greater Pittsburgh International Airport) 32 Average Yields (Cants par Mile) 30 26 26 24 22 20 16 16 14 12 10 IQ65 2085 3Q55 4485 1Q95 M86 3086 4086 IQ57 2087 3Q57 4457 IQ55 2Q88 3Q55 4Q55 1089 2Q59 First Quarter 1965through Second Quarter 1989 - USAir Group - Other Airlines mn n n 38 ComparisonAirports Note: USAir Group indudes Piedmont and USAir yield data. Page 88 GAO/RCED-W-102 Fares and Service at Major Airports Appendix I Trends in Fares at Each of the 15 Concentrated Airports Figure 1.12: Raleigh-Durham (Raleigh-Durham Airport) 30 Average Yields (Cents per Mile) 25 25 24 22 20 18 5 1055 2Q55 3Q55 4Q55 IQ55 2Q55 3Q56 #a5 1087 Ma7 3Q57 4Q57 lQ55 2Q55 3Q55 4Q55 IQ59 2Q59 First Quarter 1985 through Second Quarter 1959 - American m Other Airlines n n n n 38ComparisonAirports Page 89 GAO/RCED-90-102 Fares and !&xv-ice at Major Airports Appendix I Trends in Fares at Each of the 16 Concentrated Airports Figure 1.13: St. Louis (Lambert-St. Louis International Airport) 32 Average Yields (Cants pftr Mile) 30 25 25 24 22 20 18 16 14 12 10 lQ85 2Q55 3Q55 4Q55 IQ55 M55 3Q55 4QS5 lQ57 2Q57 3Q57 4Q57 IQ55 2Q55 3088 4Q55 IQ59 2089 Fiat Quarter 1985 through Second Quarter 1959 - T W A (&Ozark) m Other Airlines n n nn 38 Comparison Airports Note: T W A includes both T W A and Ozark yield data prior to their merger in late 1986. Page 90 GAO/RCED-!W-102 Fares and Service at Major Airports Appendix I Trends in Fares at Each of the 16 Concentrated Airports Figure 1.14: Salt Lake City (Salt Lake City International Airport) 30 Average Yields (Cents per Mile) 25 25 24 22 20 18 16 14 12 10 5 lQB5 Ma5 2Q65 4Q85 IQ55 2Q55 3Q56 4Q55 IQ57 2Q57 3Q97 4Q57 1488 2Q55 3ce5 4Q58 IQ59 2Q59 First Quarter 1955 through Second Quarter 1959 - Delta (& Western) - Other Airlines mm88 38 Compsrison Airports Note: Delta includes both Delta and Western yield data prior to their merger in early 1987. Page 91 GAO/RCFD9@1O2 Fares and Service at Major Airports Appendix I Trends in Fares at Each of the 16 Concentrated Airports Figure 1.15: Syracuse (Hancock International Airport) b2 Average Yields (Cents per Mile) 30 28 26 24 22 20 18 16 14 12 10 lQ9!5 2Q85 3096 4(185 lQ99 M96 9Q96 4086 lQ97 zQ97 3Q87 4087 lQ99 2Q99 cm99 4088 1489 2Q99 First Quarter 1995 through Second Quarter 1999 - USAir Group m Other Airlines n w m m 38 Comparison Airports Note: USAir Group includes yield data for Piedmont, USAir, and Empire. Page 92 GAO/R(ZEIMO-102 Fares and Service at Major Airporta Appendix I Trends in Fares at Each of the 15 Concentrated Airports Figure 1.16: Average Yield for the 15 Concentrated Airports 34 Average Yields (Cents per Mile) 32 30 28 29 24 lQ85 2Q85 3Q85 4QB5 lQ86 2Q86 3QB6 4086 lQB7 2487 3087 4Q87 lQ89 2QM 3Q98 4Q88 1089 MB9 First Quarter 1985 through Second Quarter 1989 - 15 Concentrated Airports W m m m 38CompariwnAirprts Page 93 GAO/WED-~102 Fares and Service at Major Airports Appendix II Policy Options Discussedin GAO Testimony on Barriers to Competition in the Airline Industry We discussed policy options in our testimony before the Subcommittee on Aviation, Senate Committee on Commerce, Science, and Transporta- tion, on September 20, 1989. The discussion of policy options is reorinted below. The full written statement, Barriers to Comnetition in the Airline Industry (GAO/T-RCED-8%65), can be obtained at the address printed on the inside of the back cover of this report. The data we have gathered on potential barriers to entry in the airline industry indi- cate that some features of airline markets are likely to discourage entry. Slot con- trols, gate leases, and, at a few airports, noise restrictions are likely to restrict access to the essential facilities needed to establish competing service. While we do not have definite estimates yet from our econometric model of the impacts of these restrictions, we believe they are likely to restrict entry and inhibit competition. The effects of some of the airline marketing strategies are less clear. CRSs, as we indicated in our testimony last year, appear to have a clear anticompetitive effect, and we have urged DOT to consider possible remedies. Frequent flyer plans appear to present a clear potential for disadvantaging entrants. However, because of the lack of data on levels of use of these plans, it may not be possible even with the results of our econometric model to estimate these plans’ effects. TACOS appear to offer a less compelling basis for disadvantaging entrants. We do have some data on TACOS, however, that may be able to show their effect on fares. Code-sharing may have some anticompetitive effects, but also appears to offer some consumer advan- tages that may offset these effects. We recognize that the Committee is considering taking action to minimize the possible anticompetitive effects of the practices we have discussed. During the course of our work, we have identified various policy options. Though not an exhaustive list, our preliminary evaluation suggests that they can provide a framework for analysis and deliberation. All of these options involve important policy considerations and require a careful weighing of costs and benefits and an assessment of trade-offs. Airport facilities are essentially local responsibilities, yet most operate under federal Gate Access restrictions imposed by the Airport and Airway Improvement Act of 1982. This act requires that airports receiving federal grants be public use facilities, available for all to use on an equal basis. One policy option would be to extend additional federal restrictions on new leases so as to reduce the long-term control that leasing airlines acquire over the airport’s facilities. Airlines need some assurance of accessto an airport’s gates to justify their investment in providing service. However, it might be possible to provide this assur- ance without giving the airline the broad control over a gate that an exclusive-use lease provides. A preferential-use gate, for example, gives the leasing airline access to the gate whenever it needs it, while still making the gate available to others when it is unused. Several airports have acted to regain control over their facilities, either by requiring short-term or preferential leases or, as Omaha and Grand Rapids have done, by not renewing majority-in-interest clauses. Another policy option would be to reduce the federal restrictions that make the air- ports dependent on the airlines as a source of revenue. The Airport Development Page 94 GAO/RCED-90-102 Fares and Service at Major Airports Appendix II Policy Options Discussed in GAO Testimony on Barriers to Competition in the Airline Industry Acceleration Act of 1973, for example, prohibits the airports from imposing any direct passenger facility charges on the passengers using the airport. The airports argue that this act, by preventing the airports from charging the passengers directly, forces them to rely on the airlines as a source of revenue, thus giving the airlines more bargaining power in lease negotiations. Airlines believe that it is appropriate for them to control airport expansion, and also have been concerned that municipal authorities would use revenues from passenger facility charges for non-airline purposes. However, the 1982 Airport and Airway Improvement Act requires airport operators to provide the Secretary of Transportation with assur- ances that all local revenues will be expended for airport purposes as a precondition for obtaining federal airport grants. Passenger facility charges could help solve the funding problems that have prevented airport expansion and reduce the airports’ need to seek majority-in-interest clauses. A small number of airports have particularly stringent noise restrictions that, while not Noise Restrictions imposed by airlines, can be a substantial entry barrier. While all parties agree on the desirability of reducing airport noise, they disagree on the questions of the pace and strategy for doing so. These contentious issues have often set local and national interests at odds, and it is not clear how far federal efforts to impose national noise policies should go. Some airports (such as Boston and Denver) have adopted noise rules that have waivers to easeentry while still achieving the desired level of noise reduction. Further exploration of noise control strategies might identify other approaches that would allow airports to control noise while mlnimlzing adverse impacts on competition. In our view, the buy/sell rule for airport slots has been ineffective at encouraging entry Slot Restrictions into slot-controlled markets. Our analysis of FAA’s data indicates that no new entrants have been able to establish service by buying slots; that the number of slots sold has steadily declined; and that the slot market is increasingly becoming a short-term leasing market, ln which major carriers that have accumulated excess slots lease out rather than sell the ones they do not need. The leasing market, while permitted in FAA’s original formulation of the market, appears to have been considered the exception. It is now the exception that is becoming the rule. Several outside studies have found that the presence of slot controls increases airline fares significantly.’ By allowing a public right-the right to use the nation’s airspace-to be treated in some respects as a private asset that is not generally available on the open market, the present operation of the buy/sell rule not only restricts competition at the four slot-controlled airports, but can impede competition throughout the northeastern and midwestern United States. These airports are a critical part of any air traffic ‘See, for example, David R. Graham, Daniel P. Kaplan, and David S. Sibley, “Efficiency and Competi- tion in the Airline Industry,” Bell Journal of Economics,vol. 14, No. 1 (Spring 1983), pp. 135-136; Elizabeth E. Bailey, David R. Graham, and Camel P. Kaplan, Deregulating the Airlines (Cambridge: MIT Press, 1985); Gregory D. Call and Theodore E. Keeler, “Airline Heregulation, k’ares and Market Behavior: Some EmpiricalEvidence,”in Andrew F. Daughety(ed.),Analytical Studies in Transport Economics (Cambridge: Cambridge University Press, 1985), pp. 221-247; and Stephen A. Mornson YUZXfZd Winston, %npirical Implications and Tests of the Contestability Hypothesis,” Journal of zw and Economics,vol. 30 (April 1987), pp. 61-62. Page 95 GAO/‘RCED-90-102Fares and Service at Major Airports Appendix Jl Policy Options Discussed in GAO Testimony on Barriers to Competition in the Airline Industry network in the northeastern or midwestern parts of the United States. It is difficult for any carrier to become an effective competitor in these heavily populated parts of the country without access to these four airports. The short-run access to slots that leasing permits is a risky basis on which to invest in a long-term service com- mitment (e.g., by leasing gates and investing in advertising). We believe that something should be done to open up the slot market so that perma- nent entry becomes easier at slot-controlled airports. We are particularly concerned about proposals to extend slot restrictions as currently structured to other con- gested airports. One solution to this problem would be for the FAA to lease slots to the airlines rather than allow them to retain the control of slots that were given to them for nothing. Leasing would have the advantage both of generating revenue for the federal government and of opening up the slot market to new entrants. It would be essential, in establishing such a market, to recognize that airlines need to have assured access to slots for a long enough period to make reasonable investments in serving routes from that airport. It would be equally important, however, to ensure that the leases ran for a limited period of time so as to prevent the slots from becoming the de facto property of the leasing airlines (as gates have become at air- ports that have long-term gate leases). Lease terms could be staggered so that leases would be long enough to assure continuity of service while ensuring that some leases would come up for renewal each year, giving entrants an opportunity each year to bid on airport capacity. An alternative would be for DOT, under the provisions of the current buy/sell rule, periodically to withdraw a portion of the slots and reallocate them by lottery. Incumbent carriers would have the opportunity to buy the slots back from the win- ners of the lottery, but at least new entrants would have an opportunity to secure slots, either through the lottery itself, or by bidding on slots sold by lottery winners, In our testimony last year on CRSs,we discussed a number of policy options, ranging Computerized Reservation from divestiture of airline-owned CRSsto non-airline owners to modifications in vendor Systems contracts with travel agents. We continue to believe that further action is warranted to remedy the anticompetitive features of the CRSindustry. As we emphasized in our earlier testimony, action in one area, such as reducing or eliminating booking fees, could create problems in another area, such as increases in CRSsubscription fees to travel agents. Consequently, travel agents’bargaining power with CRSvendors would have to be increased by modifying restrictive contract provisions, e.g., length of contract terms and minimum use clauses. While DOTis making further investigations into the competitive impact of CRSs,it has not acted to open any regulatory proceedings, as we recommended it do last fall. It is especially important that DCJl’ begin to act since its CRSrules will sunset at the end of 1990. The three other airline marketing practices that we have discussed-frequent flyer plans, Other Airline Marketing TACOS,and code-sharing-have effects that are more difficult to measure. Frequent flyer Practices plans have proven to be extremely popular promotional tools, but they have the potential to reduce competition in markets where a single carrier has a dominant market share. Frequent flyer plans offer a literal free ride to their participants, but these free trips are Page 96 GAO/RCED-90-102 Fares and Service at Major Airports Appendix II Policy Options Discussed in GAO Testimony on Barriers to Competition in the Airline Industry paid for in the form of higher fares for the average traveler and possibly also in the form of excessive business travel. DOT,in its Information Directive of June 14, 1989, has requested information on frequent flyer plans which may help to resolve the question of their impact on competition. Travel agent commission overrides, overbooking privileges, and other volume incentives clearly have some effect on the pattern of airline bookings. They increase the cost of marketing tickets and thus may pose an entry barrier to entrants with less accessto capital than established airlines have. Code-sharing agree- ments offer some advantages to airline passengers,while also probably having some anticompetitive effects. All these practices are subject to regulation by DOT under its authority to regulate anticompetitive practices in the airline industry. Should anticompetitive effects of these practices be demonstrated, they could be either prohibited or modified in some way so as to reduce any anticompetitive impact. The popularity of frequent flyer plans may make action to reduce their anticompetitive effect unpalatable. For example, one modification short of outright prohibition would be to require that mileage be transferable from one plan to another or from one passenger to another. While this would reduce the potential anticompetitive effects because passengers could earn valuable miles on any airline, such a requirement could make the plans so unattractive to the airlines that they would withdraw them. If TACOSwere prohibited, airlines might well resort to other kinds of volume incen- tives. If code-sharing agreements were prohibited, airlines would probably just buy out their code-sharing partners or develop commuter subsidiaries internally, as sev- eral airlines have already done. An important part of the success of code-sharing has been the preference that code-shared flights are allowed in CRSs, where code- shared flights are generally listed ahead of interline flights. It would be possible to prohibit CRSs from listing code-shared and on-line connections ahead of interline connections, as the European CRS rules propose, but this would make it more diffi- cult for travel agents to find code-shared flights for passengers who prefer code- shared connections. Page 97 GAO/RCED-99-102 Fares and Service at Major Airports Appendix III Major Contributors to This Report James D. Noel, Assistant Director Resources, Francis P. Mulvey, Assistant Director Community, and Kim F. Coffman, Evaluator-in-Charge Thomas F. Noone, Senior Systems Analyst Economic Sandra J. Weiss, Senior Social Science Analyst Development Division, John C. Johnson, Staff Evaluator Washington, D.C. (340595) Page 98 GAO/RCEDBO-102 Fares and Service at Major Airporta
Airline Competition: Higher Fares and Reduced Competition at Concentrated Airports
Published by the Government Accountability Office on 1990-07-11.
Below is a raw (and likely hideous) rendition of the original report. (PDF)