Low-Cost Carriers and Low Fares: Competition and Concentration in the U.S. Airline Industry
Monday, May 12, 2003
Charles Najda Department of Economics
Stanford University Email: [emailprotected]
Abstract
This analysis will examine the current structure of the domestic airline market to determine the amount of market power incumbent carriers wield, the market dynamics that exist in short haul and long haul routes, and the impact of a new breed of air carrier, the low-cost carrier, on the distribution of airfares. The analysis seeks to determine if the pricing strategies of competitors differ depending on a low-cost carrier’s presence on a route. The presence of a low-cost carrier is a more important determinant of the competitiveness of a particular route than the extent of route and hub concentration on that route. Moreover, this paper argues that previous analyses overestimate the effect of route concentration, hubs, and other route specific characteristics on the distribution of market prices. The emerging significance of the low-cost carrier may indicate a shift in the structure of the airline market away from hub-and-spoke networks and towards point-to-point networks.
2
Acknowledgements
I would like to thank Professor Frank Wolak for his guidance and help in completing this thesis. He sparked my initial interest in this topic and for that I am very grateful. His thoughtful comments and his commitment to this project helped to make this honors thesis a success.
3
Table of Contents
1. INTRODUCTION..................................................................................................... 4
2. LOW-COST CARRIER: A DEFINITION ............................................................ 8
3. BACKGROUND: THE PRINCIPAL PLAYERS ............................................... 13
3.1. MAJOR CARRIER: AMERICAN AIRLINES ............................................... 13
3.2. MAJOR CARRIER: UNITED AIRLINES...................................................... 16
3.3. LOW-COST CARRIER: SOUTHWEST AIRLINES..................................... 18
3.4. LOW-COST CARRIER: JETBLUE AIRWAYS............................................ 20
4. IMPORTANCE OF CONCENTRATION & LOW-COST CARRIERS.......... 23
5. THE DATA.............................................................................................................. 30
6. METHODOLOGY ................................................................................................. 33
7. RESULTS ................................................................................................................ 54
8. CONCLUSION ....................................................................................................... 74
9. REGRESSION TABLES ....................................................................................... 78
10. INDEX OF TABLES .............................................................................................. 91
11. INDEX OF FIGURES ............................................................................................ 92
12. BIBLIOGRAPHY................................................................................................... 93
4
1. Introduction
This paper analyzes the current state of the airline market to determine the impact
of a new breed of air carrier, the low-cost carrier, on airfares, the market dynamics that
exist in short haul and long haul routes, and by what degree previous analyses overstate
the importance of hubs and concentration. Although, financial turmoil surrounds the
airline industry today, the evolution of the domestic airline market, in the United States,
continues. Since deregulation in 1978, the cost of air travel has fallen and the size of the
airline industry has grown rapidly. The early predictions of economists that airline
deregulation would improve consumer welfare have been proven to be correct
(Borenstein 1992). Over ten years after deregulation, Borenstein (1989) finds that high
levels of route and hub concentration are correlated with higher airfares. Hubs impact the
entire distribution of fares for all flights originating or terminating at those hub airports.
Moreover, potential competition has little if any disciplining affect on market prices.
The conclusions that Borenstein (1989) derives from an analysis of the airline
market are still valid today; moreover, without controlling for the presence of low-cost
carriers, hubs and high levels of concentration skew the price distribution and shift it
rightwards. Even after controlling for route specific characteristics such as distance, load
factor, and aircraft equipment route and airport concentration are estimated to raise prices
at each point of a route’s price distribution. Today’s airline market has evolved markedly
since the late 1980s. While low-cost carriers only carry 15% of the domestic traffic in the
United States, their impact on prices is significant and broadly distributed over the
marketplace (Parker Oct 17, 2002).This paper argues that most of the hub effect is due to
5
the lack of a low-cost carrier serving that hub and not associated with the hub itself. More
generally, the measurable change in price distributions associated with high levels of
route and airport concentration is overstated. Much of the concentration effect is directly
attributable to the presence of a low-cost airline or in this case the lack thereof.
The disciplining force of the low-cost carrier is apparent in the regional airline
market, where Richards (1996) finds that the potential presence of Southwest measurably
lowers airfares. Yet, Richards postulates that Southwest (a low-cost carrier) does not have
the same impact in the long haul market as it does in the regional market. This analysis
argues that low-cost carriers, such as Southwest and JetBlue, continue to discipline prices
in the regional market but also discipline prices in the long haul market. A significant
change in the distribution of prices is no longer a direct consequence of levels of route
and airport concentration when low-cost carriers are represented in the estimated
equations.
The impact of a low-cost carrier is independent of market distinctions and
subdivisions. On each route, regardless of its characteristics, a low-cost presence alters
the distribution of airfares to produce a distribution with a tighter range and lower median
fare. Borenstein shows that high levels of route and airport concentration, characteristic
of airline hubs, raise fares at the high-end of the market’s price distribution; however,
low-end fares are not untouched, these factors raise low-end fares as well. Thus measures
of route and airport concentration affect the range and height of each and every route
price distribution. Although, concentration levels may alter each fare in a distribution,
Borenstein argues that these measures impact the highest percentile fares to the greatest
extent. This paper determines that the largest shift in the expected price paid is
6
consistently at the upper tail of the price distribution. To this day, route and hub
concentration continue to primarily alter the shape of the composition of airfares to a
greater degree at higher price percentiles. This paper’s baseline estimates would imply
that the concentration effect is as much of a factor today as in the Borenstein (1989)
analysis; however, important interactions, not limited to concentration measures, call into
question the validity of this paper’s baseline equations. The level of concentration, along
with other often used route characteristics, no longer provides as accurate an account of
the airline market as possible.
A low-cost presence lowers prices at the low end, median, and high-end of ticket
prices in all market segments. In other words, this new breed of air carrier disciplines
market prices at any price point within a given market. The broadening impact of the
low-cost carrier across price points suggests that the airline market is evolving; moreover,
the level of many of the effects economists have attributed to concentration levels, hubs,
and slot constraints may in fact be overstated. Instead of measuring the relationship
between concentration levels, hubs, slot constraints and prices this paper argues that
previous econometric analyses may have actually been measuring the impact on a route
or airport of not having a low-cost carrier. An understanding of the significance of the
low-cost carrier requires insight into what constitutes a low-cost carrier. The ability of the
low-cost carrier to compete with established hub-and-spoke carriers on price represents
an evolution of the airline market in the United States.
This thesis argues that a change in the conventional method of analysis is
necessary to build an accurate dynamic model for estimating prices. That critical change
involves altering standard models of analysis in order to measure the impact of low-cost
7
carriers. Concentration levels are no longer the most important factor in determining the
level of market prices; moreover, measures of concentration may have never been as key
to understanding the dynamics of the airline market as previously suggested. Not
controlling for low-cost carriers, today’s airline market resembles the market of the late
1980s even though a variety of structural changes have taken place since then. The
measurable impact of the low-cost carriers has two implications. First, the presence of a
low-cost carrier causes the distribution of prices to flatten and shift left. The median fare
is lower and price volatility is lower. Second, the lack of a low-cost presence raises
prices. Borenstein is not only measuring the impact of concentration with controls for
route and airport dominance, but also is measuring the lack of a low-cost presence
through those variables.
Low-cost carriers cause price distributions to compact and shift lower across all
market segments. Hence, a large fraction of the concentration effect is directly due to the
lack of a low-cost presence. An examination of previous literature written on the airline
markets and the principle market participants provides a solid superstructure within
which a model can be built to determine the range of the low-cost carriers’ impact on the
marketplace. The low-cost carriers’ growing market share brings to light the consequence
of competitive entry in the airline industry. Moreover, a low-cost presence is correlated
with a flatter price distribution, a higher degree of competitiveness, and a decline in the
relevance of concentration, frequency, hubs, and capacity constraints in predicting market
prices. This result suggests any future analysis of the competitive structure of the airline
industry should take into account the low-cost effect.
8
2. Low-Cost Carrier: A Definition
This paper defines the low-cost carrier to be an airline that operates a point-to-
point network, pays employees below the industry average wage, and offers no frills
service. The two most prominent low-cost carriers, JetBlue and Southwest, both have
labor costs 30% to 40% lower than the mainline carriers. A traditional major carrier often
has a number of tools at its disposal, which it can use to deter entry or lessen the
competitiveness of recent entrants. These tools include predatory pricing, loyalty
programs, and congestion at the nation’s most popular airports. Yet, these tools are not
effective against low-cost carriers with point-to-point networks. A low-cost airline can
engage in Bertrand competition, with a high-cost competitor, without pricing at its own
marginal cost. The low-cost carrier can successfully neutralize the dominance of its
competitors, by competing on price.
The lower cost structure can be quantified by aggregating the cost savings of
point-to-point networks, wage savings, and savings from not providing numerous add-on
services. While labor costs are the largest single cost item for airlines, there are many
other costs. The pie chart in Figure 1 illustrates the composition of costs for the aggregate
airline industry in 2000. The cost differential between the low-cost and major carriers is
not only attributable to the wage differential. Although, the primary cost for any carrier is
labor related. Controlling labor costs can improve the bottom line. The operating cost
distribution below suggests that lowering labor costs by 10% can lower the average
airline’s total cost by 3.68%.
9
Figure 1 Airline Operating Cost Distribution
Operating Cost Distribution2000
Fuel, 12.80%Airports,
5.40%
Advertising, 1.20%
Labor, 36.80%
Commission, 11.30%
Food, 3.20%
Interest, 2.00% Other, 23.10%
Source: Air Transport Association, 2001.
The lower cost structure of a point-to-point network is a consequence of a number
of factors. They include: Airport congestion, which causes costly delays at hubs and is
not as prevalent at airports used by point-to-point carriers. For every major metropolitan
airport there are often two to three secondary airports. Low-cost carriers can achieve fast
turnarounds and pay less for leasing airport facilities at secondary airports like TF Green
airport outside Providence, Rhode Island. Low airport lease rates and gate costs also
contribute to the lower cost structure of low-cost carriers. Under utilized secondary
airports often levy lower charges for the use of their facilities.
In comparison, hubs require a large number of gates and personnel per flight, due
to the banks of flights that are used at hubs. The banks of flights result in the majority of
flights arriving and departing within 20-30 minutes of each other. These peak periods
10
result in a high demand for facilities and personnel for short periods of time. For
example, at its Dallas Fort Worth hub American operates banks of flights to make
connections convenient. While at neighboring Dallas Love Field Southwest spaces its
flights out due to the lower emphasis it places on connecting traffic. Like other hub-and-
spoke carriers, American Airlines has peak times when a considerable number of planes
land at its hubs and passengers rush off to get on their next flight. The system provides
customers a high level of convenience but creates operating inefficiencies. Employees
stand around between peaks. Planes sit on the ground longer and get caught in line
waiting to take off. The hub-and-spoke structure raises an airline’s costs at a hub
compared to operating that same hub with a de-peaked structure. In particular, the higher
number of personal required per flight to effectively operate a traditional hub may be an
important factor in the different cost structures of traditional and low-cost carriers.
“Spreading out peak travel times at hub airports by several more hours each day, a
concept known as rolling hubs, may be one way to make operations more efficient”
(Business Week Oct 23, 2002). Major carrier American Airlines is experimenting with a
rolling hub in Chicago in an attempt to achieve lower operating costs.
The two most prominent low-cost carriers, JetBlue and Southwest, both have
lower labor costs than the large incumbent carriers. Analysts estimate that Low-cost
carriers such as Southwest and JetBlue have labor costs 30% to 40% lower than the
mainline carriers. For example, United Airlines, American Airlines, Northwest Airlines,
and Continental Airlines all have costs at least 40% higher than Southwest. Although,
Delta Air Lines and Alaska Airlines have the lowest costs of the majors, each of them has
unit costs 30% higher than Southwest’s (Wall Street Journal Oct 9, 2002). Table 1
11
provides a break down of costs and revenue on an ASM (available seat mile) basis for
major and low-cost carriers. The comparison is the same for JetBlue, which has a cost
structure marginally lower than Southwest’s. Low-cost carriers have substantially lower
unit costs; however, they do not have substantially lower unit operating revenue.
Table 1 Operating Cost and Revenue for Carriers of Interest
Domestic Carrier Operating cost per available seat mile (cents) / 2001
Operating revenue per available seat mile (cents) / 2001
American Airlines 11.41 9.22 Continental Airlines 9.58 9.78 Delta Airlines 10.14 9.39 JetBlue Airways 6.81 8.26 Northwest Airlines 9.78 9.17 Southwest Airlines 7.54 8.51 United Airlines 12.00 9.80 Note: All figures are for the twelve month period ended December 31, 2001, except for the JetBlue figures which are for the first quarter of 2002. Source: 10-K, for fiscal year ended December, 31 2001 for each respective carrier, except for the JetBlue figures which are sourced from the firm’s 10-Q, for the quarterly period ended March, 31 2002.
The lack of unionization among low-cost airlines can be characterized as a myth.
Of the two largest low-cost airlines one is unionized and the other isn’t. Although, most
of Southwest’s workforce is unionized and all of JetBlue’s is not, the aggregate cost
structure of the two carriers is almost identical; unionization isn’t necessarily correlated
with high labor costs for low-cost carriers. These carriers also use fewer employees,
because they operate point-to-point networks.
The third and perhaps the most obvious attribute of the low-cost carrier is the no
frills service that these carriers provide passengers. Instead of providing passengers with
a menu of product choices priced within a range, the low-cost carriers offer a single type
of product, coach service. Low-cost carriers do not provide meals on flights, which
12
results in a savings of 5 to 10 dollars per coach passenger. No meals equates to a savings
of up to 3.2% from the average carrier’s operating cost (see Figure 1). These airlines lack
elaborate loyalty programs, which necessitate extra employees, to provide more
personalized service, and expensive facilities, like airport clubs. Low-cost airlines do not
provide costly services, which are only profit enhancing for a hub-and-spoke carrier able
to extract a high level of rents from customers with a high willingness to pay, business
travelers. The main advantage of the low-cost carrier is that it can compete on price, with
the high-cost traditional carriers. The functional structure of the low-cost carrier is
perceptible; moreover, this paper argues that the impact of this unique structure on the
airline market is just as evident.
13
3. Background: The Principal Players
To motivate the econometric model and analysis it is useful to survey briefly the
literature on the airline industry and the present condition of four important agents in the
airline market at present. The section on concentration and low-cost carriers examines the
analysis of past authors. This section provides the reader with a general understanding of
the two largest hub-and-spoke carriers, American Airlines and United Airlines, and the
two dominant low-cost carriers JetBlue Airways and Southwest Airlines. To explain how
two market forces, concentration and low-cost presence, evolved and currently impact the
market requires an understanding of these four representative airlines.
3.1. Major Carrier: American Airlines
American Airlines began as “Robertson Aircraft Corporation of Missouri, which
was the second aviation company to hold a U.S. airmail contract” (AMR Corp. 2002).
Charles Lindbergh flew the first mail freight flight for Robertson on the morning of April
15, 1926. The pace of consolidation increased rapidly as the nascent airline industry
grew. In 1930 Robertson Aircraft subsidiaries were incorporated into American Airways,
which was renamed American Airlines in 1934.
The introduction of the DC-7 on the New York to Los Angeles route in 1953 was
a harbinger of the importance of transcontinental routes to American’s profitability
throughout its history. In 1959 American became the first airline to offer nonstop coast to
coast jet service with the Boeing 707. This event validated the importance of
transcontinental routes to American. A move that would solidify the dominance of
14
American on many of its routes was the introduction of SABRE. “At the end of 1959 and
into the early 1960s, American, teaming up with IBM, introduced and implemented
SABRE (Semi-Automated Business Research Environment), the largest electronic data
processing system for business use” (AMR Corp. 2002). SABRE served as a powerful
device with which profitable price discrimination could be implemented through yield
management techniques.
Almost 15 years elapsed between the introduction of SABRE and the
implementation of two important complementary initiatives. First, American began
marketing SABRE to travel agents in 1975. SABRE’s presence at travel agencies
provided American with a direct conduit to market its travel to the consumer. American
could use the system to track the amount of business travel agents provided American.
Also, American implemented incentives tied to the amount of business an agent did with
American. The second initiative was the introduction of the Super Saver ticket. “On April
24, 1977, American introduced the most popular fare in its history, the Super Saver.
Initially offering discount fares from New York and California, Super Saver was
expanded to all of American's routes in March 1978 and later to Mexico and Canada”
(AMR Corp. 2002). The Super Saver allowed American to charge last minute travelers,
often business travelers, high fares while charging travelers, who plan ahead and as a
group have a more elastic demand curve, lower fares.
In a bid to build brand loyalty, “in 1981, American introduced the
AADVANTAGE travel awards program, a revolutionary marketing program to reward
frequent fliers” (AMR Corp. 2002). Borenstein (1992) argues that a locally dominant
airline can achieve market power through the use of frequent flyer programs. These
15
programs attract repeat business from lucrative business customers. The programs’
incentives lower the probability that a customer will choose a competing product.
However, every major carrier has a frequent flier program and all of those programs are
very similar. The similarity of these programs and the fact that carriers will provide
complimentary membership to encourage customers to switch (to their own program)
suggests that the market power derived from these programs is not material. Thus
SABRE and other innovations provided American with a platform from which it could
strengthen its market power. The success of these initiatives may have also spurred the
increasing importance of hubs.
Only 6 years after the introduction of SABRE American established its
Dallas/Fort Worth hub on June 11, 1981. The Dallas/Fort Worth hub increased
concentration levels and may have allowed American to raise prices. The success of its
first hub coupled with the use of SABRE and Super Saver tickets prompted American to
add “…new cities and routes to strengthen its hub-and-spoke networks” (AMR Corp.
2002). In fact, “…competition has tended to decrease on direct routes to and from the
hub (airports)” (Borenstein 1992). Likewise it has been shown that decreased competition
often results in higher prices.
More recently, American earned record profits during the late 1990s by raising
business fares on a regular basis. American’s profitability seemed to validate the
dominance of the hub-and-spoke system. The collapse of the airline industry post
September 11th and the continued quagmire, in which the airlines find themselves,
suggests that American’s post deregulation strategy championed by Robert Crandall is no
longer effective. In fact, during the spring of 2003 American Airlines found itself
16
embroiled in a fight with its labor unions to lower its labor costs. The failure of the
unions to approve the required wage concessions would have resulted in American
declaring immediate bankruptcy. Throughout its storied history American has behaved
like a rational market participant, most recently building out a profit maximizing hub-
and-spoke network. However, American’s attempts to salvage its hub-and-spoke strategy
may prove to be a fatal mistake. The importance of the point-to-point carriers is rising
and shows no signs of abating.
3.2. Major Carrier: United Airlines
The recent turbulence United has experienced contrasts sharply with the calm that
enveloped the firm prior to deregulation. That calm reflects the conservative nature of
United’s management throughout its history. “When it comes to its business decisions,
United has been one of the more conservative airlines. For instance, before purchasing a
fleet of jets in the late 1950s, United conducted a year-long study to learn the economic
effects jet travel would have on its airline operations” (Corporation for Public
Broadcasting). While American was aggressively using new technologies and strategies,
United always took the more conservative route. United’s purchase of bankrupt PanAm’s
Asia-Pacific routes expanded the scope of its hub-and-spoke network. Ironically this
neoconservative purchase, to adopt and strengthen the hub system, may have ultimately
led to United filing for bankruptcy in November of 2002.
Although, United was often more conservative than its competitors when it came
to strategic business decisions the airline was nonetheless mired in financial managerial
trouble for much of its history. “The company ran through a string of six presidents
17
between 1970 and 1989, and changed its name twice as it continued its corporate
expansion until forced to divest and return to its core airline business in 1987. It was a
period of major hurdles for the airline's senior management team, which faced its biggest
challenge when the industry was deregulated in 1978” (UAL Corporation). Given that
United had difficulty maintaining profitability in a regulated environment, United’s
ability to adapt with the changing airline market post 1978 is remarkable.
United remained almost exclusively a domestic airline for several decades. But in
the mid 1980s and early 1990s, the airline broke into the international market. A
financially-troubled Pan Am sold its fabled Pacific Division to United in 1985. In 1990,
United added several European destinations, and a year later, Latin American routes
(Corporation for Public Broadcasting). At about the same time United began to construct
a set of fortress hubs and started to leverage its yield management technology to more
accurately match ticket prices to the individual demand curves of passengers.
The bankruptcy of United Airlines along with its peer US Airways begs the
question: Are the days of the hub-and-spoke carrier(s) numbered? United’s turbulent and
conservative history points to the success of the hub-and spoke system when coupled
with frequent flier programs and other anticompetitive practices. Even with all of its
troubles United managed to attain record profitability during the late 1990s with its hub-
and spoke network. The future for United, American, and the other major carriers is
unclear. The ability of these airlines to maintain high levels of concentration, let alone the
importance of high concentration levels, is diminishing as a new breed of carrier brings
the point-to-point network back into fashion.
18
American Airlines and United Airlines are typical hub-and-spoke carriers. Each
airline’s development showcases the major carrier’s market strategy since deregulation.
The goal of these strategies is to increase the level of profit each firm derives from its
own network. Borenstein (1989) argues that hub dominance results in higher prices. The
major carriers’ strategic activities involve implementing dominance enhancing programs,
like frequent flier programs, with the goal of raising profit levels. Thus the widespread
use of dominance enhancing programs suggests that concentration levels and hub airports
have an effect on price distributions and ultimately the fare each passenger pays.
3.3. Low-Cost Carrier: Southwest Airlines
Although, the history of the low-cost carrier is brief, this form of airline is now a
crucial factor in predicting market prices for domestic airline routes. In 1971 Southwest
Airlines inaugurated service between Dallas, Houston, and San Antonio Texas. Arguably
America's most unique airline Southwest pioneered the low-cost carrier and successful no
frills service at a low price. From its humble beginnings as an intra-Texas carrier to its
position today as the fourth largest carrier in the United States, Southwest has proven that
its business and network models work (Southwest Airlines). A pioneer of the single
aircraft type fleet, Southwest also initiated customer friendly initiatives such as electronic
ticketing while maintaining a competitive fare structure. The major carriers have
attempted to fend of Southwest through allegedly anticompetitive behavior and creating
low-cost subsidiaries, such as United Express, Continental Lite, and Delta Express, which
have all failed. The failure of competitors to dislodge Southwest from the markets it
19
serves reflects the strength the firm derives from its low-cost structure and extensive
route network.
An examination of the Southwest route map reveals that the airline serves
secondary airports such as Providence RI, instead of Boston Logan (Southwest.com).
Thus Southwest avoids competing directly with the major carriers by serving airports that
the majors have little if any service to (Reuters 03/12/02). Southwest only enters routes
that it can be profitable on, as opposed to routes where it would have to compete with
competitors, who have substantial market power. Southwest’s aggressive pricing strategy
encourages its competitors to price closer to marginal cost. A simple residual demand
example shows how a firm responds to the aggressive pricing of others by pricing more
aggressively itself. Case in point, Southwest directly competes with American on Los
Angeles area to San Jose flights and as a result American’s fares are half as expensive on
a price/cost ratio basis versus its own long haul flights (Southwest.com, AA.com).
Moreover, in late 2002 American responded to Southwest’s competitive pressure by
replacing mainline aircraft with cheaper to operate regional jets, from its American Eagle
subsidiary, in an attempt to lower its operating costs.
Southwest ends up carrying price conscious passengers, who are willing to forgo
certain services and fly from less desirable/convenient airports. Numerous travelers
prefer secondary airports that are often closer to suburban households and lack the
congestion of city centric airports. While Southwest certainly has brought competition to
the routes it has entered, Southwest has chosen to ignore high profit long haul routes
(Richards 1996, Pg. 33). Southwest’s strategic choice to minimize its competitive
interactions with American et al adds validity to the previous arguments by Borenstein
20
that American and United have considerable dominance over the market and have the
resources to preserve their dominance over lucrative routes.
Southwest’s success will highlight a couple important points, which will aid in
explaining JetBlue’s strategy. Southwest is a low-cost carrier that charges low prices for
its tickets, yet manages to maintain profitability year after year. An analysis of the
Southwest effect confirms that, “potential competition appears to be no substitute for
actual competition” (Borenstein 1992, Pg. 511). Richards (1996) finds that the pricing
strategies of incumbent carriers differ depending on Southwest’s presence. Southwest’s
presence can alter the competitiveness of a route and the composition of the distribution
of fares for a route.
3.4. Low-Cost Carrier: JetBlue Airways
Enter JetBlue, an upstart airline that is going head to head with the major carriers
on long haul transcontinental routes from John F. Kennedy (JFK) airport. David
Neeleman, veteran airline executive, started JetBlue 4 years ago. Mr. Neeleman is
familiar with the low-price airline business; he started Morris Air, which is a low fare
airline that he sold to Southwest in 1994 (New York Times 08/27/01). JetBlue started
with the largest initial capitalization of any airline, over 160 million dollars. Likewise,
powerful New York politicians, upset with high intra New York state fares, provided
JetBlue with a remarkable 75 slots at JFK, a slot restricted airport (New York Times
08/27/01). Since JetBlue obtained a large quantity of slots, capacity constraints do not
protect its competitors. JetBlue has the financial resources to be a formidable competitor
21
and has overcome slot restrictions at a premier airport through intense lobbying of the
Clinton Administration (New York Times 08/27/01).
JetBlue resembles Southwest in a number of ways. First, just like Southwest
JetBlue operates a fleet made up of a single type of aircraft, the Airbus A320
(JetBlue.com). While Southwest started with second hand aircraft, JetBlue’s A320s are
all straight from the factory. Operating a single type of airplane lowers JetBlue’s
operating costs. Moreover, newer aircraft are cheaper to operate than comparable older
aircraft. This may explain why JetBlue has a lower cost basis than Southwest. Second,
both Southwest and JetBlue have a single class of service, coach; however, JetBlue
outfitted its aircraft with “…live satellite TV at every leather seat” (New York Times
08/27/01). Third, at present both airlines have non-hostile relationships with their
respective employees (Time Magazine 07/10/01, Culture and Control 02/07/98). Most of
Southwest’s employees belong to unions, while JetBlue hopes to keep costs down by
keeping its workforce non-union (New York Times 08/27/01). From a distance JetBlue
appears to be a carbon copy of Southwest in every respect; however, the juxtaposition of
the two revels that JetBlue is modeling itself after image conscious Virgin Atlantic as
much as it is trying to emulate bare bones Southwest (New York Times 08/27/01).
Unlike other startups such as National Airlines, Pro Air, and Legend Airlines
JetBlue entered the airline industry with a multitude of advantages those failed airlines
lacked. As stated above, JetBlue started with a cash amount almost 5 times more than the
amount Legend Airlines, a failed startup partially funded by Las Vegas casinos, began
with. JetBlue also serves the primary airport for the New York City area, with millions of
potential customers, who lack a low-cost alternative within the New York city
22
metropolitan area (New York Times 08/27/01). These advantages are critical to JetBlue’s
future success.
Southwest Airlines and JetBlue Airways are typical low-cost point-to-point
carriers. Each airline’s expansion showcases the low-cost carrier’s market strategy in
today’s marketplace. The goal of these strategies, just as for the major carriers, is to
increase the level of profit each firm derives from its own network. Richards (1996)
argues that Southwest’s presence lowers airfares. The low-cost carriers’ strategic
activities involve implementing cost control programs, like high aircraft utilization rates,
with the goal of raising profit levels. These programs maintain a low-cost airline’s
competitive advantage, its cost structure. Thus the widespread use and success of cost
containment programs suggests that a low-cost presence on a route may have an effect on
the route’s fare distribution.
23
4. Importance of Concentration & Low-Cost Carriers
The academic work of economists was a significant force in the movement
towards the deregulation of the domestic airline industry in the early 1970s (Borenstein
1992). Post deregulation, most analysis of the airline markets focuses on the lack of
competition among the airlines; however, the early predictions of economists that airline
deregulation would improve consumer welfare have in fact been proven to be correct
(Borenstein 1992). Moreover, Borenstein (1992) argues that the most important lessons
that have been learned from the deregulation of the airline industry are not with regard
how to deregulate, but are a better understanding of the strategies and results of
competition in a complex and innovative service industry.
Before the deregulation of the airline industry many economists anticipated a
historic opportunity to study and examine a prominent industry transition from
government regulation to a competitive free market system. During the past twenty years
economists have produced numerous papers and analyses of the airline industry. The
papers written by authors such as Borenstein (1989, 1992) and others explain many
aspects of the airline industry. The importance of hub dominance in airline pricing
(Evans 1993) and the ability of low-cost carriers to have a significant affect on pricing
(Richards 1996) are critical pieces of evidence that lend support to this paper’s
hypothesis. Moreover, these articles serve as a foundation, which supports the conclusion
that JetBlue is a competitive force.
As the airline industry matures, in a parallel fashion the analysis and
understanding of the airline industry matures as well. During the 1980s, journal articles,
24
which argued that the airline industry would exhibit perfect contestability, accompanied
the recent deregulation of the airline industry (Beesley 1986). The analysis of Beesley
(1986) in the United States and Joy (1986) in Australia suggested that airlines could not
sustain ticket prices far in excess of marginal cost, because new entry could easily take
place due to the low barriers to entry of the airline industry.
However, “It has now been well established that airline pricing does not closely
reflect the perfect contestability ideal” (Borenstein 1989, P. 344). The airline market is
not perfectly contestable, because barriers to entry make it difficult for new entrants to
break into the airline market. Price discrimination is the norm not the exception, on routes
with high barriers to entry. As a result, much of the recent analysis of the airline industry
has focused on trying to explain what factors prevent entry, which would increase
competition and result in lower fares.
Borenstein (1989) chooses to focus on the importance of route and airport
dominance in determining the degree of market power exercised by an airline. The results
of the paper indicate that an airline’s share of passengers on a route and at the endpoint
airports significantly influences its ability to mark up price above cost. However, these
high markups tend to only benefit the dominant carriers, with smaller operators not
achieving any greater degree of pricing control. Borenstein echoes the argument of
Levine (1987) that airlines value airport dominance for the insulation such dominance
may give them from competition.
The two primary factors that determine the difficulty of entry into an airport are
slot and gate constraints (Borenstein 1989). Capacity restrictions, or more formally
control of a dominant share of slots or gates, “…can allow an airline to inhibit profitable
25
entry” (Borenstein 1989, P. 348). While no airline directly controls the capacity of any
airport, capacity restrictions can benefit incumbent carriers. “Slot restrictions have a
positive effect on yields” and along with airport dominance and computer reservation
systems, slot constraints provide carriers operating from slot-constricted airports with
greater market power (Richards 1996, P. 44). Not surprisingly, the originating and
terminating airports of long haul routes are either slot constricted or suffer severe
capacity restrictions due to a lack of infrastructure. Raising fares at an airport with no
capacity constraints attract entrants, which in turn force the fare back down. In contrast, if
an airport is slot restricted or capacity constrained, an incumbent can raise its fares and
competitors may be unable to enter in full force or even at all. Thus a sizable percentage
of the incumbents price increase will remain intact.
Route and hub concentration are critical aspects that must be considered when
examining the competitiveness of the airline markets. There are a number of programs
that airlines use to increase their ability to price discriminate. Marketing devices give an
advantage to a dominant carrier and amplify the market power of that incumbent airline
(Borenstein 1989). In particular, frequent flier programs encourage airline loyalty,
especially among customers who primarily fly from an airport that is a fortress hub of an
airline (Levine 1987). Fortress hubs are airports where the dominant carrier operates a
majority of the flights departing and arriving at the airport.
Both airline ownership of computer reservation systems (CRS) and scarce airport
facilities increase an airline’s localized market power (Evans 1993). American and
United were early entrants into the computer reservation service industry. Sabre
(American) and Apollo (United) signed up many travel agents early on and continue to be
26
two of the most dominant CRSs to this day (Borenstein 1992). However, Sabre is now an
independent company, although American is currently Sabre’s primary customer. The
effect of Sabre and Apollo is far reaching and creates bias in travel agent booking, called
the “halo effect” (Borenstein 1992). The “halo effect” is when a travel agent using an
Apollo system books on average a disproportionate share of his or her tickets on United
Airlines (Borenstein 1992). An airline’s CRS system enhances its market power;
however, the growth in internet bookings calls into whether these programs allow an
airline to more effectively defend its profits from entrants.
The results of the Borenstein (1989) analysis support the observation that an
airline charges higher prices when it has a dominant position at an airport. Moreover, at
hub airports the dominant carrier charges higher prices than it does in the remainder of its
system. Cost factors do not explain the discrepancy in prices (Borenstein 1989). The
primary conclusion is that an individual airline’s share of traffic on the route and at the
endpoint seems to be a principle determinant of a carrier’s ability to raise the price of its
product (Borenstein 1989). These results build upon and support the results of previous
authors, who wrote on the importance of route concentration; however, Borenstein (1989)
shows that analyses that focus only on route concentration invariably ignore one of the
most important factors in airline competition, airport concentration.
Evans (1993) reiterates a number of points of past authors concerning marketing
devices and slot restrictions; however, he disagrees with the conclusion of Borenstein
(1989) that route and airport concentration together are the primary factors that determine
the competitiveness of an airline route. Evans (1993) refines Borenstein’s (1989) earlier
article by distinguishing between the impact of route dominance versus airport
27
dominance on an airline’s level of market power. Evans tests whether the observed
dominance of most intercity-pair markets and airports in the United States airline industry
by single carriers confers pricing power on the dominant firm. Evans concludes that
airport dominance by a carrier does confer significant pricing power, whereas dominance
at the route level seems to confer no such pricing power. Thus, over time the identified
cause of the lack of competition transitioned from route concentration to route and airport
concentration to airport concentration. More recent articles shift the focus of the analysis
away from what causes a lack of competition to what forces bring about competition in
the airline markets.
Whinston and Collins (1992) conduct an event study analysis of People Express,
in order to measure the competitive structure of deregulated airline markets. Instead of
examining measures of concentration, Whinston and Collins (1992) consider an
alternative approach that uses the reactions of incumbent airlines’ stock prices to
announcements of entry by People Express to determine the competitive structure of the
airline markets. The stock reactions reveal significant route-specific profits. This
indicates that the contestable market model cannot be used to model airline routes. The
average incumbent on a route lost roughly $3 - $6 million in value when People Express
announced entry into that route (Whinston and Collins 1992). A decline in an incumbent
airline’s equity value suggests that equity market participants recognize that entry will
reduce an incumbent’s profits. While the analysis of Whinston and Collins is unique, the
use of an airline’s equity price as a proxy for competitiveness on a route may not be the
most accurate or direct way to measure competitiveness. Moreover, there are a number of
other factors that can influence the movement of an airline’s stock price. Richards
28
chooses to examine the airline markets in an innovative manner, but uses a more direct
and often used model to measure the competitiveness of airline routes.
Richards (1996) begins her paper by noting the work of Joesch and Zick (1994),
who along with others developed studies that indicate that airline markets are not
perfectly contestable. The overview of the industry is based upon the writings of other
authors including Borenstein (1992) and Whinston and Collins (1992). The relationship
between market concentration and airfares is statistically insignificant in the early 1980s
under regulation, but significant during the late 1980s and early 1990s (Joesch & Zick
1994). Airline markets are not perfectly contestable and concentration can explain why
airfares are high on particular routes. In particular there is a positive and statistically
significant relationship between airfares and hub concentration (Evans 1993). The market
conditions that cause high fares are now well explained; however, there has been little if
any discussion as to what type of market participant can cause significant and lasting
downward pressure on airfares.
The residual demand model highlights the impact of an airline such as Southwest.
The model demonstrates that a firm responds to the aggressive pricing of competitors by
pricing more aggressively itself. This economic model represents the core of the
argument that the presence of a low-cost carrier causes airfares to fall (Richards 1996).
Richards estimates a series of fare equations to ascertain the effects of actual and
potential competition by Southwest on markets for air travel. The results of the analysis
suggest that pricing strategies of competitors differ depending on Southwest’s presence
on a route (Richards 1996). The particular or actual presence of a low-cost competitor
seems to be a more important determinant of fares than conventional measures of
29
concentration. Thus Richards (1996) offers the reader an alternative hypothesis as to what
factor influences competitiveness of the airline industry to the greatest extent. The
presence of a low fare carrier may be a more important determinant of the ticket pricing
on a particular route than the extent of route and hub concentration on that route. The
analysis and conclusion of Richards (1996) parallels the analysis and forthcoming
conclusion of this paper.
The major carriers successfully use dominance enhancing strategies to raise
concentration levels that in turn raise the market price for air travel. Moreover, low-cost
entry disciplines prices on short haul routes. The rapid expansion of the low-cost
presence in the airline markets over the last half decade and the continued dominance of
the major carriers together raise the following question. Are concentration levels or a
low-cost presence the more important predictor of market fares? The juxtaposition of
these opposing factors in this paper provides an answer to the question of which factor is
critical to understanding the dynamics of pricing in the airline markets.
30
5. The Data
In order to determine whether or not low-cost carriers significantly impact the
pricing of tickets in airline markets this paper will conduct an econometric analysis using
the Department of Transportation’s Databank 1B database. The Databank 1B database is
a domestic only subset of the Databank 1A database. The database is a 10% random
sample of all domestic tickets sold in the United States. This analysis employs the second
quarter 2002 dataset. This particular dataset includes valuable pricing information on a
per ticket basis. This analysis will focus on nonstop flights that do not include plane
changes or stops. The effect of plane changes and stops is well documented in papers
written by Borenstein (1989) and others. The inclusion of parameters to test for the effect
of on-plane stops and plane changes would not significantly alter the results of this
analysis, because these variables do not significantly alter Borenstein’s results. This
dataset includes over 1,650,000 individual ticket observations. The dataset includes 35
carriers, 415 airports, and 9150 different routes. Hence, the dataset for the regressions
contains a total of 21,573 route carrier pairs.
The Origin and Destination survey also includes data on length of the observed
route and number of passengers on the route. The Herfindahl-Hirschman Indices are
derived from this dataset. Low-cost market share measures are constructed from the
dataset by segregating the market share measures based on whether or not the carrier is
deemed to be a low-cost carrier. For the purposes of this analysis, there are 11 low-cost
carriers. A comprehensive list of all the represented airlines is included in Table 5.
Routes of 500 miles or less are included in the data, since Southwest primarily flies short
31
haul routes. Unlike Richards (1996), long haul routes of over 1000 miles are included in
the dataset to measure the impact of JetBlue and Southwest on the long haul market.
Furthermore, this dataset has served as the primary source of data for many of the articles
discussed in the section on concentration and low-cost carriers.
To compute cost per available seat mile (CASM) figures for all 35 airlines,
subsections of Department of Transportation’s Schedule T-1 and Schedule P-12 datasets
are utilized to construct the CASM figure of the dataset for this analysis. The schedule T-
1 dataset summarizes the T-100 by providing monthly compilations, by carrier and
service classification, of such items as available seat miles, available ton miles, revenue
passenger miles, revenue miles flown, and revenue departures performed. The Schedule
P-12 provides quarterly profit and loss statements for carriers with annual operating
revenues of $20 million or more. The data include operating revenues, operating
expenses, depreciation and amortization, operating profit, income tax, and net income.
The 2002 Schedule P-12 dataset does not include total operating expenses figures for
Airtran Airways, Northwest Airlines, Horizon Air, and TCI Skyking. As a result, the
average CASM (for the 31 other carriers) is substituted for the actual value. Rerunning
the set of regressions, without these carriers, does not significantly alter the estimated
coefficients for any of the equations. The cost per available seat mile (CASM) figures are
calculated for the year 2002.
Frequency, load factor, and equipment variables are constructed using data
aggregated from the Department of Transportation T-100 Domestic Segment dataset for
the months of April, May, and June 2002. Thus the union of the three datasets
corresponds to the second quarter of 2002. The T-100 Domestic Segment dataset contains
32
data reported by US air carriers operating non-stop between airports located within the
boundaries of the United States and its territories. The data fields contain information by
aircraft type and service class for departures performed, available capacity and seats,
passengers enplaned, freight and mail enplaned, scheduled departures, aircraft hours
ramp-to-ramp and airborne. Unlike the Databank 1B database, the T-100 Domestic
Segment database is not a 10% sample, but in fact is a comprehensive dataset. The T-100
dataset is similar in form to the DOT Service Segment Data, used in analysis of
Borenstein (1989) and colleagues. The Department of Transportation Databank 1B,
Schedule T-1, Schedule P-12, and T-100 Domestic Segment databases are all publicly
available through the recently constructed Bureau of Transportation Statistics (BTS)
TransStats website.1
1 The BTS TransStats website can be found at the following web address: http://www.transtats.bts.gov.
33
6. Methodology
Having outlined the progression of academic thought on the factors that influence
ticket prices and having argued that low-cost carriers have a fundamentally different cost
structure than their competitors, an analysis examining the impact of the low-cost carrier
is warranted. The 3 econometric models developed in this section of the thesis estimate
the marginal benefit of low-cost entry by quantifying the impact of several measures of
concentration on the market price for air travel. The three models include a baseline
model that closely resembles Borenstein’s (1989) equation, a 6-segment distance model
that includes the baseline variables plus dummy variables measuring the change in price
over discrete 500 mile segments, and a short/long haul segment model that includes the
baseline variables plus 2 indicators that classify a route as either belonging to a regional
market or a long haul market.
The price equation is estimated as a function of costs, service quality, market
demand characteristics, and other factors such as route concentration, route
characteristics, hub concentration, low-cost carrier route concentration, and low-cost
carrier hub concentration. Roundtrip tickets are treated as two one way tickets and
roundtrip fares are adjusted accordingly. The equations that follow will be used in
conjunction with the constructed dataset to show that low-cost carriers have a significant
impact on prices and may have a greater impact on prices than concentrations levels.
The baseline, 6-segment distance, and short/long haul segment models capture the
impact of mainline carriers’ market power on price. Following Borenstein (1989), each of
the three models is regressed on the sample 20th, 50th, and 80th percentile fare for each
34
route-carrier pair. As Borenstein notes, the segmentation of each equation into three
regressions provides insight into the impact of concentration levels and low-cost
participation on different strata of the ticket price spectrum. The paper includes three
basic models, regressed with and without a set of low-cost carriers and additionally
regressed on three distinct percentiles of price. As a result, 18 regressions are run in order
to provide a comprehensive analysis of the current state of the airline markets. These 18
equations converge to three hybrid models that seek to explain (determine the
significance of) two primary variables, which can affect the supply curve for air travel,
the level of competition, and concentration in the airline markets.
The baseline equation, based on a Borenstein (1989) pricing equation, provides a
logical starting point from which to begin further analysis exploring the impact of low-
cost carriers and the current state of the regional (short haul) and long haul airline
markets. Since all analysis in this paper focuses on nonstop tickets (roundtrip and one
way) the baseline equation excludes variables measuring the impact of stops and
directness on pricing. In fact, the results of the Borenstein (1989) regression clearly show
that stops do not have a significant impact on price, but that the circuitousness of a route
raises prices in a statistically significant manner. The baseline equation, regressed on
second quarter 2002 Department of Transportation data, brings Borenstein’s (1989)
analysis to the present day and brings to light any changes that have occurred in the
market structure over the last 15 years.
The baseline equation is also estimated with a set of variables that measure the
importance of the low-cost carriers. These variables include low-cost market share and a
dummy variable that indicates whether or not an endpoint airport is a hub of a low-cost
35
airline. The low-cost variables are analogous to the POTCOMP (potential competition)
variable that Borenstein (1989) includes in his second set of regressions. However, the
POTCOMP measure does not distinguished between mainline and low-cost carriers. The
measure only includes carriers that have a presence on the route of less than 1% of the
total traffic of the route in question. Thus the measure only quantifies the effect of
potential competition and not actual competition. It is important to note that, “(the)
inclusion of the potential competition variable does not change the results (of the
regression) substantially, though it may have a significant effect itself at least on lower
fares” (Borenstein 1989, P. 359). Borenstein’s results suggest low-cost carriers may only
be an important factor acting on prices at the 20th percentile. The inclusion of the low-
cost variables may indicate whether Borenstein’s (1989) initial conclusion still holds. The
juxtaposition of these two sets of regressions should reveal the impact of the low-cost
carrier on pricing in the airline market.
36
Baseline Equation
ln Pij = άi + β1 ln MKTDISTj + γ1 ln LOADFACij + γ2 ln EQUIPij + γ3 ln FREQij + δ1 ln
ORGSHAREij + δ2 ln ORGHERFj + δ3 RUTSHAREij + δ4 RUTHERFj + δ5 ln
COMPCOSTij + δ6 TOURISTj + δ7 HUBj + δ8 LWCSHUBj + δ9 SLOTj
Baseline Equation with Low-Cost Variables
ln Pij = άi + β1 ln MKTDISTj + γ1 ln LOADFACij + γ2 ln EQUIPij + γ3 ln FREQij + δ1 ln
ORGSHAREij + δ2 ln ORGHERFj + δ3 RUTSHAREij + δ4 RUTHERFj + δ5 ln
ORGLWCSHAREj + δ6 LWCSTSHAREj + δ7 LWCSTHERFj + δ8 ln
COMPCOSTij + δ9 TOURISTj + δ10 HUBj + δ11 LWCSHUBj + δ12 SLOTj
Market distance can impact the price of a ticket due to the change in cost structure
over different route lengths. Takeoffs and landings are highly inefficient and require up to
50% of a flight’s fuel. As a result, longer routes, on which the fixed cost of the takeoff
and landing is spread out, are less expensive to operate per mile. However, different
market distances are not just associated with direct costs to the airlines. Using a
continuous variable such as MKTDIS establishes a linear relationship between route
distance and price. This linear relationship may indicate a general trend as route length
increases; however, a significant coefficient on MKTDIS does not answer the question of
whether or not distance is linearly related to price. In order to determine if the
relationship between distance and price varies over length it is necessary to test discrete
segments of length.
37
This paper categorizes different route lengths as separate markets in order to test
the hypothesis that each of these market segments has a unique competitive structure.
High prices may categorize certain route lengths due to the variability of concentration
levels over several types of routes. To quantify the impact of the distance of a route on
ticket pricing this analysis employs a second set of equations that employ 6 dummy
variables representing 6 consecutive 500 mile increments in route length. The baseline
and excluded variable is MKTDIS_1, the 1-500 mile market segment. With these 6
dummy variables it is possible to construct a segment by segment relationship between
price and distance. As with the baseline equation, the six segment equation is constructed
in its base form and with 4 low-cost variables to measure the impact of low-cost carriers
on pricing over the 6 distance segments. In particular, these equations may identify the
relative importance of low-cost carriers over distance. Do low-cost carriers only impact
pricing on short haul routes as Richards (1996) postulates?
38
Six Segment Distance Equation
ln Pij = άi + β1 ln MKTDISTj + γ1 ln LOADFACij + γ2 ln EQUIPij + γ3 ln FREQij + δ1 ln
ORGSHAREij + δ2 ln ORGHERFj + δ3 RUTSHAREij + δ4 RUTHERFj + δ5 ln
COMPCOSTij + δ6 TOURISTj + δ7 HUBj + δ8 SLOTj + δ9 MKTDIS_1j + δ10
MKTDIS_2j + δ11 MKTDIS_3j + δ12 MKTDIS_4j + δ13 MKTDIS_5j + δ14
MKTDIS_6j
Six Segment Distance Equation with Low-Cost Variables
ln Pij = άi + β1 ln MKTDISTj + γ1 ln LOADFACij + γ2 ln EQUIPij + γ3 ln FREQij + δ1 ln
ORGSHAREij + δ2 ln ORGHERFj + δ3 RUTSHAREij + δ4 RUTHERFj + δ5 ln
ORGLWCSHAREj + δ6 LWCSTSHAREj + δ7 LWCSTHERFj + δ8 ln
COMPCOSTij + δ9 TOURISTj + δ10 HUBj + δ11 LWCSHUBj + δ12 SLOTj + δ13
MKTDIS_1j + δ14 MKTDIS_2j + δ15 MKTDIS_3j + δ16 MKTDIS_4j + δ17
MKTDIS_5j + δ18 MKTDIS_6j
The impact of low-cost carriers may vary over market distance; moreover, there
are two primary market segments in the United States, short haul (regional) routes and
long haul routes. This paper categorizes regional routes as being 0-1500 miles in length.
Long haul routes are defined to be 1501-3000 miles in length. It is not a coincidence that
this dichotomy loosely relates to the set of routes that serve meals and those that do not.
Traditionally low-cost carriers, such as Southwest, serve regional routes and avoid
competing on higher profile long haul routes. Recently, Southwest is adding longer routes
39
and maintaining the same level of in-flight service that it offers on other flights.
Moreover, JetBlue with hubs at John F. Kennedy Airport (JFK) and Long Beach Airport
(LGB) is directly targeting long haul transcontinental routes, prior to which these routes
lacked a low-cost presence. Although, JetBlue’s presence on long haul routes (1500 miles
in length or more) is still small this paper seeks to determine if that presence significantly
disciplines the pricing behavior of the incumbent nationwide carriers.
The two additional indicator variables (MKTDISREG and MKTDISLNG)
provide a direct means with which to measure the growing importance of the low-cost
carrier. Richards (1996) articulates the importance of Southwest in routes less than 1000
miles in length. In a similar vein, Borenstein (1989) suggests that the impact of potential
competition on prices is rather small. He observes that these potential competitors are not
materially important for the 50th and above percentile airfares. The airline market has
evolved since the mid 1990s and because low-cost carriers no longer only fly routes
under 1000 miles in length it is important to determine the extent of the importance of the
presence of low-cost carriers in the two primary market segments, regional and long haul.
40
Short/Long Haul Market Segment Equation
ln Pij = άi + β1 ln MKTDISTj + γ1 ln LOADFACij + γ2 ln EQUIPij + γ3 ln FREQij + δ1 ln
ORGSHAREij + δ2 ln ORGHERFj + δ3 RUTSHAREij + δ4 RUTHERFj + δ5 ln
COMPCOSTij + δ6 TOURISTj + δ7 HUBj + δ8 SLOTj + δ9 MKTDISREGj + δ10
MKTDISLNGj
Short/Long Haul Market Segment Equation with Low-Cost Variables
ln Pij = άi + β1 ln MKTDISTj + γ1 ln LOADFACij + γ2 ln EQUIPij + γ3 ln FREQij + δ1 ln
ORGSHAREij + δ2 ln ORGHERFj + δ3 RUTSHAREij + δ4 RUTHERFj + δ5 ln
ORGLWCSHAREj + δ6 LWCSTSHAREj + δ7 LWCSTHERFj + δ8 ln
COMPCOSTij + δ9 TOURISTj + δ10 HUBj + δ11 LWCSHUBj + δ12 SLOTj + δ13
MKTDISREGj + δ14 MKTDISLNGj
MKTDIST (market distance) is the nonstop distance measured in miles from one
endpoint of a route to another. Longer routes should have higher prices; however, time
spent cruising at altitude is proportionally higher on longer routes, which suggests that
longer routes have a lower per mile cost. The higher per mile cost of short routes is due to
the large amount of fuel that is burned during takeoff and landing. DISTANCE should be
positive, but costs increase proportionately more slowly as the route distance lengthens.
The average flight distance on a major carrier is less than a low-cost airline. The median
flight segment for the major carriers is 821.0 miles and is 866.0 miles for the low-cost
carriers. However, the difference is small suggesting that neither the major nor low-cost
41
carriers have a cost advantage from flying a much longer set of routes than other carrier
group. Low-cost carriers may lower prices in markets and fly a high proportion of short
haul flights. DISTANCE may not be significant when controlling for the presence of low-
cost carriers, depending on whether the low-cost carrier or operating cost factor has a
greater impact on price. At the time of the Borenstein (1989) article low-cost carriers had
a diminutive presence in the United States.
Ceteris paribus a longer route should have a higher market price than a shorter
route with the same characteristics. Longer routes require more fuel and are more costly
to operate (on an absolute basis) compared to shorter routes. However, most of the fuel
for a flight is burned during takeoff and landing. When the plane is at cruising altitude the
fuel efficiency of the plane is higher. Hence longer transcontinental flights have a lower
cost per mile than flights of 500 miles or less. As a result, an increased distance may not
result in as large of an increase in price as initially thought. It is important to note that the
dominant low-cost carrier, Southwest, primarily operates short haul routes, yet still
charges lower fares than its competitors. Distance should explain a large percentage of
the price level in a market, but in no way does distance explain the disparity in prices
between carriers.
The transcontinental market illustrates this disparity suitably. JetBlue provides its
product at a price that undercuts its competition by up to 75%. An unrestricted coach
ticket from JFK to LAX on American or United costs approximately $2500. In
comparison, a comparable ticket on JetBlue costs $598. This example also raises the
possibility that past analyses have overestimated the impact of distance, because low-cost
competition did not exist on long haul routes until recently. Prior to low-cost competition,
42
long haul prices may have had proportionally higher prices because airlines had more
market power on the longer routes. The possible discrepancy in past analyses bolsters the
importance of including variables in the price equation to control for the presence of low-
cost carriers.
LOADFAC (load factor) is the mean load factor of all flights of an airline that
serves the observed route. The per passenger cost of a flights decreases as the load factor
rises, which suggests that LOADFAC has a negative coefficient. Also low-cost carriers
tend to have higher load factors than their competitors. This may result in prices falling
on routes with a higher load factor. However, flights with high load factors tend to
operate during peak travel times in the morning and late afternoon. Congestion during
peak travel times raises the cost of a flight, both in terms of delays and increased personal
needed to handle the banks of flights that arrive and depart within a small window of
time. Also, the highest load factor flights tend to originate or terminate in hub cities,
which have high levels of congestion during peak periods. Congestion impacts low-cost
carriers to a lesser degree, since these firms operate out of less crowded airports. The
weakening correlation between congestion and load factor over the last decade would
suggest that a higher load factor would lower prices. Due to the growing presence of low-
cost carriers and the lower operating cost of full flights (on a per capita basis) the effect
of LOADFAC is uncertain.
EQUIP (equipment) is the mean size of the planes on flights that carry passengers
on the observed route and airline. Larger equipment tends to have lower operating costs
on per passenger basis. Also, larger planes operate more efficiently on longer routes,
which may indicate a degree of interaction between DISTANCE and EQUIP. On the
43
other hand, larger planes, usually with two aisles, are thought to be preferred by travelers.
The higher preference for higher capacity aircraft may increase demand on routes with
larger aircraft, which would in turn raise prices. Likewise, the high-cost major carriers
tend to operate a higher proportion of larger aircraft than low-cost rivals, such as
Southwest and JetBlue. For this reason, a larger mean aircraft size may indicate higher
prices.
FREQ (frequency) is the average frequency of flights available to passengers on
the observed airline and route. Controlling for other factors, a higher frequency of flights
raises the value of the product to the passenger. Passengers traveling on business have a
high opportunity cost for travel and value the convenience increased frequency provides
them. One would expect that increased value leads to higher demand and finally higher
prices. Higher frequency also allows airlines to operate aircraft more efficiently and for
longer utilization periods. Increased efficiency lowers an airline’s marginal cost and
ticket prices. Low-cost carriers tend to operate routes with a high frequency of service to
achieve cost reductions. With the widespread influence of low-cost carriers one would
expect that a high frequency may lower ticket prices. Thus the estimated effect of FREQ
on price is uncertain.
ORGSHARE (origin & dest. market share) is the weighted average of the
observed airline’s share of passenger originations at the two endpoints of a route.
ORGSHARE is weighted by the ratio of passengers on the route who start their trips at
the two endpoints divided by the total number of passengers, who originate travel at all
endpoints. An increase in ORGSHARE may allow airlines to exercise greater market
power. With greater market power an airline can raise prices to increase its profits.
44
Because the vast majority of flights still take place on hub-and-spoke carriers the
coefficient of ORGSHARE may diverge from ORGLWCSHARE.
ORGHERF (origin & dest. Herfindahl) is the weighted average of the Herfindahl
indexes of passengers’ originations at the two endpoints of the observed route (measured
from zero to one). The weighting of ORGHERF is the same as ORGSHARE. The
competitive factors that provide an airline with increased market power correspond to a
dominant carrier having many small competitors. However, an airline may prefer having
a single larger competitor in order to facilitate illegal, but profitable collusion. Case in
point, when American Airlines CEO Robert Crandall called Braniff International Airlines
CEO Howard Putnam to collude, he suggested doing so would be easy due to their large
shares of the market in question. This scenario is less likely to be viable now due to the
extensive presence of price aggressive competitors. In particular low-cost carriers are
unlikely to engage in collusion and prefer instead to compete on price with higher cost
incumbents. Both competitive models can result in higher prices, if the dominant carrier
exercises its market power or engages in collusion. An airline may prefer either
competitive landscape and as a result the effect of ORGHERF on fares is uncertain.
RUTSHARE (route market share) is the observed airline’s share of passengers on
the observed route (origin to destination traffic). Elementary economic theory would
suggest that increased route share corresponds to increased market power, which results
in higher prices for the consumer. The passenger will pay a higher price whether or not
the dominant firm’s market power has an umbrella effect. One can expect that increased
route share results in higher prices.
45
RUTHERF (route Herfindahl) is the Herfindahl index for passengers on the
observed route (measured from zero to one). If the observed route has a small number of
large participants, collusion is logistically easier. Through collusive practices the carriers
can raise the consumer’s cost of travel. However, if there is a dominant firm with a
competitive advantage and many smaller participants, the dominant firm’s pricing power
may not extend to its competitors. The change in the mean price on the observed route
depends not only on the value of the Herfindahl index, but also on the competitive state
of the dominant firm.
ORGLWCSHARE (low-cost origin & dest. market share) is a weighted average
of low-cost carriers’ share of the daily passengers at the endpoints of the route in
question. As in ORGSHARE the weighting is proportional to the number of total
passengers who originate travel at the two endpoints, not just passengers traveling on
low-cost carriers. Assuming increases in ORGSHARE will lead to increases in market
power; increases in ORGLWCSHARE indicate that low-cost carriers transport a larger
percentage of the passengers at the endpoint airports. An increase may result in lower
levels of market power and prices. Ceteris paribus including the type of the route, the
number of firms, and each company’s market share, ticket prices will be higher in a
market with no low-cost carriers compared to the same market with a single low-cost
carrier replacing one of the traditional airlines. Higher ORGLWCSHARE may lead to
reduced levels of market power and lower prices. However, low-cost carriers charge
lower fares than the major carriers. A larger market share for a low-cost carrier may
cause downward pressure on ticket prices.
46
LWCSTSHARE (low-cost route market share) is the share of all origin and
destination passengers on that route transported by low-cost carriers. The larger the
percentage of people transported by low-cost carriers the lower the airfare will likely be.
Since RUTSHARE, which measures the importance of concentration levels, is also
measured it may be the case that RUTSHARE is no longer significant when
LWCSTSHARE is included. This would indicate that low-cost route share, not general
levels of route concentration, explain the movements in ticket prices. If RUTSHARE has
a positive sign and LWCSTSHARE has a negative sign then each has an opposite impact
on pricing than the other variable. Such a result would be consistent with basic economic
theory, which suggests that concentration raises prices and entry lowers prices.
LWCSTHREF (low-cost route Herfindahl) is the Herfindahl index (measured
from zero to one) for the origin and destination passengers carried on low-cost carriers on
a particular route. The measure is included to show the likely discrepancy between the
normal Herfindahl index (RUTHREF) and this variable, which only measures the market
share of low-cost carriers.
COMPCOST (comparable cost) is the weighted average of the cost per seat mile
of every airline on the observed route other than the observed firm. Each carrier’s share
of passengers on the observed route determines its corresponding weighting. Clearly,
each airline’s system wide cost per seat mile only approximates the actual cost per seat
mile on the observed route. The system wide figure is an accurate, but not necessarily
precise proxy. If the marginal cost of its competitors rises, the observed carrier can raise
prices without altering its market share. Of course a firm may choose to not raise its
prices in order to raise its market share. Low-cost carriers have lower operating costs than
47
their competitors; however, low-cost carriers do not price tickets as if they had the same
cost basis. On a global basis it appears that a rise in COMPCOST may cause prices to
rise. However, the elasticity of COMPCOST (for a specific route) may be higher
dependent on whether or not the observed airline is a low-cost carrier. As the market
share of low-cost carriers has grown the elasticity of COMPCOST is likely to have fallen.
As a result, COMPCOST may play a less significant role in the pricing equation than in
the Borenstein (1989) analysis.
In addition, conventional wisdom suggests that there is a tight and positive
relationship between an airline’s cost per seat mile and the price it charges to the
consumer. Likewise, if the production costs of an airline’s competitors rises then the
observed airline would be expected to raise its price in response to capture a higher level
of profits. However, low-cost carriers often price tickets far below the major airlines’
level in order to gain market share, which can insulate a low-cost carrier from a
competitor’s predatory pricing scheme. An analysis of the models will explain whether
the low-cost carrier, with its unique cost structure, is a factor in market prices.
TOURIST (tourist route) measures the degree to which the observed route is a
tourist or leisure route. TOURIST is a dummy variable set to 1 if one of the endpoints of
the route is vacation oriented and 0 otherwise. This variable is critical, especially with
this paper’s focus on low-cost carriers. It may be the case that low fare carriers charge
less for tickets. However, the misconception that low-cost carriers only serve tourist
oriented airports, is just that, a misconception. The expansion of Southwest’s route
network and emergence of JetBlue is a direct result of low-cost carriers targeting non
tourist routes. Controlling for tourist oriented routes in the regression equation removes
48
an important factor that might otherwise obscure the significance of the variables testing
the significance of low-cost carrier presence. Tourists tend to have a more elastic demand
curve than other travelers. The highly elastic demand curve on tourist routes suggests that
a higher proportion of tourists on a route will lower prices. Thus, TOURIST should have
a negative coefficient.
Table 2 Tourist Oriented Airports
Airport Airport Code Aspen, Colorado ASE Fort Lauderdale International, Florida FLL Hilo International, Hawaii ITO Honolulu International, Hawaii HNL Palm Springs Metropolitan Area, California PSP Jacksonville International, Florida JAX Kahului Airport, Hawaii OGG Kona, Hawaii KOA Las Vegas Mccarran Intl, Nevada LAS Melbourne Intl, Florida MLB New Orleans International, Louisiana MSY Orlando International, Florida MCO Reno/Tahoe Int'l, Nevada RNO Tampa International, Florida TPA West Palm Beach International, Florida PBI Note: The compiled list of tourist airports is derived from Richards (1996).
HUB is a dummy variable equal to 1 if one of the endpoints of the observed route
is a hub of a major carrier, and 0 otherwise. Note that the major carriers are limited in this
analysis to American Airlines, United Airlines, Northwest Airlines, Delta Airlines, and
Continental Airlines. See below on page 49 for a list of hub cities, segregated by major
carrier. The analysis of Borenstein (1989) and others suggests that airlines wield higher
levels of market power at hub airports. As a result consumers face higher prices for
travel, if their route includes a hub airport at one of the endpoints. One expects that the
sign of HUB will be positive.
49
Table 3 Major Carrier Hubs
Major Carrier Hub Airports American Airlines Chicago O’Hare International – ORD
Dallas Fort Worth – DFW Lambert St. Louis – STL Miami International – MIA San Juan Luis Muñoz Marin – SJU
Continental Airlines Houston Intercontinental – IAH Newark International – EWR
Delta Airlines Atlanta Hartsfield International – ATL Cincinnati Northern Kentucky – CVG Dallas Fort Worth – DFW Salt Lake City – SLC
Northwest Airlines Detroit International – DTW Memphis International – MEM Minneapolis-St. Paul – MSP Tokyo Narita – NRT
United Airlines Chicago O’Hare International – ORD Denver International Airport – DEN San Francisco International – SFO Washington Dulles – IAD
Note: The hubs for the five largest major carriers in the United States are self-reported. This list is built from airline documents taken from the corporate websites of each respective carrier. Each of these documents and websites is listed in the bibliography.
LWCSHUB (low-cost hub at origin or dest.) is a dummy variable equal to 1 if one
of the endpoints of the observed route is the hub of a low-cost carrier, and 0 otherwise.
Table 4 on page 50 is a list of low-cost carrier hubs. Dominant airlines tend to have a
higher level of market power at a hub, where a high concentration of flights are operated
by the dominant firm. However, a low-cost hub on an observed route may result in lower
prices. A low-cost carrier’s dominant presence may indicate that most of the airfares on
50
the route are priced at a level set by the lowest cost producer, JetBlue or Southwest.
Instead of wielding traditional market power in order to raise profits, low-cost carriers
compete on price in order to win market share, which ultimately raises profits.
Table 4 Low-Cost Hubs
Low-Cost Carrier Hub Airports JetBlue Airways John F. Kennedy Airport – JFK
Long Beach Airport – LGB
Southwest Airlines Chicago Midway – MDW Dallas Love Field – DAL Phoenix Sky Harbor International – PHX
Note: Only Southwest and JetBlue have prominent hub airports. Hence, the primary airports for the other low-cost carriers are not included in this analysis. Also, the hubs for these low-cost carriers are self-reported. This list is built from airline documents taken from the corporate websites of each respective carrier. Each of these documents and websites is listed in the bibliography.
SLOT is a dummy variable equal to 1 if one of the endpoint airports of the
observed route is a slot restricted airport and 0 otherwise. It should be noted that slot
restrictions have been lifted at Chicago O’Hare, New York LaGuardia, and New York
JFK. Slot restrictions are currently only in place at Reagan National Airport, which
serves Washington D.C.
MKTDIS_1j, MKTDIS_2j, MKTDIS_3j, MKTDIS_4j, MKTDIS_5j, MKTDIS_6j.
(market distance) The MKTDIS variables control for the distance of a route with finer
granularity than MKTDIST. Each of the six variables is a dummy variable that takes on
the value 1 if the route length falls within the given MKTDIS variable’s range and 0
otherwise. These variables will help determine the extent of pricing differences, which
are tied to the length of a route. MKTDIS_1 refers to routes from 1 to 500 miles in
length, MKTDIS_2 refers to routes from 501 to 1000 miles in length, MKTDIS_3 refers
51
to routes from 1001 to 1500 miles in length, MKTDIS_4 refers to routes from 1501 to
2000 miles in length, MKTDIS_5 refers to routes from 2001 to 2500 miles in length,
MKTDIS_6 refers to routes from 2501 to 3000 miles in length.
MKTDISREG (regional market distance group) variable is an indicator variable
which takes on the value 1 when the observed route is between 1 and 1500 miles in
length and 0 when the observed route is 1501 to 3000 miles in length. This variable along
with MKTDISLNG may indicate the extent of low-cost carrier penetration. While low-
cost pioneer Southwest has avoided flying routes in excess of 1000 miles recently low-
cost carriers are expanding their long haul presence with JetBlue the most notable
example. Since MKTDISREG and MKTDISLNG are perfectly collinear, MKTDISREG
is excluded from the regression and as a result serves as a baseline measure.
MKTDISLNG (long haul market distance group) variable is an indicator variable
which takes on the value 1 when the observed route is between 1501 and 3000 miles in
length and 0 when the observed route is 1 to 1500 miles in length. This variable along
with MKTDISREG may point out the scope of low-cost carrier penetration. JetBlue is the
only low-cost carrier primarily targeting long haul routes. However, Southwest has a
small presence on long haul routes. In 2002 Southwest inaugurated its longest route from
Baltimore/Washington, D.C. (BWI) to Los Angeles. The new route has a distance of
2,329 miles (Silicon Valley / San Jose Business Journal 2002). Hence, the MKTDISLNG
variable may serve as a proxy for the impact of Southwest and JetBlue on long haul
transcontinental routes. Since MKTDISREG and MKTDISLNG are perfectly collinear,
MKTDISLNG is included in the regression and measures the difference in pricing from
the baseline regional market.
52
Table 5 includes a comprehensive list of the airlines included in the dataset
constructed for this analysis. Since the Databank 1B dataset is a 10% random sample of
all ticket travel, the airlines included in Table 5 are randomly selected, with a weighting
proportional to the market share of each carrier. The low-cost carrier designation is based
on the definition of a low-cost carrier described in section 2 of this paper. That definition
is not an industry standard, because there is no standard system for determining if an
airline is a low-cost carrier. The definition included in this paper selects the major low-
cost carriers, Airtran, Frontier, Southwest, and JetBlue. Moreover, the other low-cost
carriers, in Table 5, are small in comparison and so should not impact the estimated
results in a significant manner.
The equations and methodology described in this section are used to determine
whether low-cost carriers impact the distribution of prices in different segments of the
airline market. The equations include a comprehensive set of parameters to estimate
which factors influence the distribution of market prices. Controls for route specific
characteristics are included to maximize the accuracy of the estimated equations.
Moreover, these estimated equations will also establish whether including the low-cost
parameters was warranted. If the low-cost variables are not significant the low-cost
equations will resemble the baseline equations; otherwise, any differences between the
two sets of equations may characterize what the effect of a low-cost presence is on fare
distributions.
53
Table 5 Comprehensive Carrier List
Carrier Code Carrier Name Low-Cost Carrier 9N Trans States Airlines 0 AA American Airlines 0 AQ Aloha Airlines 0 AS Alaska Airlines 0 B6 JetBlue Airways 1 CO Continental Airlines 0 CS Continental Micro 0 DL Delta Air Lines 0 EV Atlantic Southeast Airlines 0 F9 Frontier Airlines 1 FL Airtran Airways 1 G4 Allegiant Air 0 HA Hawaiian Airlines 0 HP America West Airlines 0 JI Midway Airlines 0 MQ Simmons Airlines 0 N7 National Airlines 1 NJ Vanguard Airlines 1 NK Spirit Airlines 1 NW Northwest Airlines 0 OH Comair 0 OW OneWorld Alliance 0 PN Pan America Airways 0 QX Horizon Air 0 RU TCI SkyKing 0 SM Sunworld Airlines 1 SY Sun Country Airlines 1 TZ America Trans Air 1 UA United Airlines 0 US US Airways 0 WN Southwest Airlines 1 XJ Mesaba Airlines 0 XP Casino Express Airlines 1 YX Midwest Express Airlines 0 ZW Air Wisconsin 0 Note: 1 designates that an airline is a low-cost carrier and 0 designates that an airline is not a low-cost carrier. Also, although Sunworld Airlines, Sun Country Airlines, Casino Express Airlines, and America Trans Air are considered to be low-cost carriers in this analysis, each of these airlines primarily flies charter flights. Thus the impact of these carriers is expected not to be material to the results of this analysis.
54
7. Results
The results of this analysis bring up to date Borenstein’s (1989) results and
interpretations while providing insight into the impact of low-cost carriers on airfares, the
market dynamics of short haul and long haul routes, and to what extent previous analyses
overstate the importance of hubs and concentration. Tables 9 through 26 include the
regression results for each of the 18 regressions. Tables 9, 11, and 13 are the baseline
regression results and are in large part consistent with Borenstein’s results. An individual
airline’s market share on a route and to a lesser extent at the endpoint airports still
determines a carrier’s ability to raise prices. At the 20th, 50th, and 80th percentile fares the
Route Market Share (RUTSHARE) coefficient is significant at the 1% level. The
variation of the coefficient value over price is low indicating that route concentration
provides pricing power to airlines at any fare level. Controlling for route specific
characteristics, the equation estimates that a 1% increase in a carrier’s share on a route
results in an increase in prices between 0.15% and 0.21%. Of note is the narrow range of
the price impact. Borenstein (1989) estimates an analogous range of 0.03% to 0.22%.
Thus the impact of route concentration is more prominent today, especially at the low-
end of fares.
The highest levels of route and endpoint airport concentration are often found on
routes that originate or terminate at a major nationwide carrier’s hub. Table 3 enumerates
the primary hubs for each major carrier. The baseline equation estimates that the presence
of a hub at either or both endpoints of a route raises prices between 6.8% and 29.2%.
These coefficients from the 20th, 50th, and 80th percentile equations are significant at the
55
1% level. The effect of a hub is significant, large, and proportional to the type of fare.
Ceteris paribus, this equation predicts that a hub raises the median fare by 17.5%. The
stronger effect of hubs on the higher end, 29.2% vs. 17.5% at the median, suggests that
loyalty programs targeted at business travelers provide a hub carrier with increased
market power. This result is consistent with Borenstein’s (1989) hypothesis that travelers
that purchase full fare tickets are more likely to participate in frequent flier programs.
While route share and hub effects are measurable and significant over a range of
prices, origin and destination market share (ORGSHARE) only impacts the highest
priced tickets. At the 80th percentile fare, the equation (represented in Table 13) estimates
that a 1% increase in a carrier’s share of a route’s endpoint airport traffic results in a
0.032% rise in prices. This result is significant at the 1% level. At the 20th percentile and
median price the effect of origin and destination market share is negligible and not
significant at the 10% level. The effect of origin and destination market share is not
consistent over across different types of airfares; moreover, this result diverges from the
equation Borenstein (1989) estimates. The growth in air traffic at secondary airports and
the stagnation of traffic growth at the largest most concentrated airports is one possible
explanation for the declining effect of origin and destination market share, especially at
the low-end of the market.
The effect of a slot restricted airport (SLOT) parallels the hub effect.2 The
baseline equation estimates that a slot restricted airport at the origin or destination causes
the 20th, 50th, and 80th percentile fares to rise 6.8%, 11.2%, and 22.3% respectively. The
80th percentile estimate is significant at the 1% level, while the median and 20th percentile
2 Borenstein (1989) does not include a parameter to measure whether airports are slot restricted. However, this parameter is included to highlight the effect of capacity constraints, in this case government enforced, on the marketplace.
56
estimates are statistically significant at the 2% level. The uniformly positive and
increasing effect of slot constraints across airfares suggests that capacity constraints,
whether imposed by the government or the market, reduce that ability of competitors to
enter a market and discipline prices. More generally, this analysis proposes that capacity
constraints hinder competitive entry and raise airfares.
The remaining set of estimated parameters are, with the exception of one
parameter, consistent with expectations when significant. Tourist oriented routes,
represented by the parameter TOURIST, are expected to have airfares 10.6% lower than
non-tourist routes at the median fare. While the signed relationship between TOURIST
and HUB is reverse, the increases in magnitudes are parallel across types of fares. Tourist
oriented routes differ from the baseline route the least at the lowest fare and the most at
the highest fare. This estimated outcome agrees with the premise that tourist oriented
routes carry a higher percentage of price sensitive tourists. These price sensitive
passengers are more likely to purchase the lowest priced fares; moreover, on these routes
frequent flier programs do not markedly enhance the market dominance of the observed
carrier. Price sensitive travelers are not very loyal to any single carrier.
The market distance parameter (MKTDIST) is positive and significant at the each
regressed fare percentile. This estimated result is as expected; although longer routes are
more efficient to operate on a per mile basis, a longer route is still more costly and as
such necessitates a higher airfare. The load factor variable (LOADFAC) is negative and
significant at the 1% level in the 80th percentile baseline equation (Table 13). For the
median fare LOADFAC is negative and significant at the 2% level and is not significant
for the lowest fares. This relationship is inverse to the one Borenstein predicts, where
57
LOADFAC is negative and only significant at the 20th percentile. However, the result of
this analysis is consistent with the argument that higher load factors indicate that a plane
is more efficiently utilized, which in turn lowers a carrier’s cost and the airfares that
carrier will charge on the observed route. Equipment (EQUIP) is estimated to effect each
price percentile in a negative and significant (at the 1% level) manner. Larger equipment
is more efficient to operate and allows an airline to lower fares.
The effect of the frequency of flights on a route (FREQ), served by a particular
carrier, is positive and significant at the 1% level over each fare percentile. A higher
frequency of flights is associated with more traveled business routes. This estimate
corroborates the argument that travelers with a high value of time will pay a premium for
expanded service. On the other hand, the effect of competitor’s cost (COMPCOST) is not
significant at any fare level. The impact of the weighted route Herfindahl (RUTHERF) is
only significant in the estimated equation regressed on the 80th percentile fare. The result
of the 3 baseline estimates upholds the conclusion that ceteris paribus, routes with a
single dominant carrier have higher prices. Moreover, market dominance of an airline is
enhanced when that observed carrier controls a large percentage of a route’s traffic and
has a number of small competitors.
While Borenstein (1989) finds that potential competition has little if any
disciplining affect on market prices, this analysis argues that low-cost carriers have a
significant and far-reaching impact on the airline market.3 The estimated baseline
equations with low-cost variables, presented in Tables 10, 12, and 14, reveal that the
3 It is important to note that Borenstein (1989) does not control for low-cost carriers. Instead, the paper controls for potential competition, which is not an accurate proxy for measuring the magnitude of low-cost presence. If the Borenstein had controlled for low-cost carriers the paper may have estimated lower magnitudes for the concentration and hub effects. Hence, the negligible effect of potential competition does not imply that low-cost carriers had little if any impact on the airline market in the late 1980s.
58
effect of a low-cost carrier is uniform across every type of fare. The low-cost route
market share (LWCSTSHARE) parameter is significant at the 1% level for all three
percentiles. At the 20th percentile, a 1% increase in the share of passengers carried by all
the low-cost carriers on the route results in a 0.14% drop in price. A 1% rise in low-cost
route market share lowers the median fare by 0.44%. The effect on the 80th percentile is
the highest; a 1% rise in the low-cost route market share lowers the estimated price by
0.98%. The chart below illustrates the dramatic rise in the magnitude of the low-cost
route share effect across fare types.
Figure 2 Low-Cost Effect on Routes
Low-Cost Route Market Share
-1.20
-1.00
-0.80
-0.60
-0.40
-0.20
0.0020th 50th 80th
Price Percentile
Coe
ffic
ient
Val
ue
Baseline Equation with Low-Cost Variables
Note: The low-cost route market share parameter is significant at the 1% level for all three percentiles.
59
The presence of a low-cost carrier is not only important on routes; the level of
low-cost endpoint airport market share (estimated with the parameter ORGLWCSHARE)
also impacts market fares. If low-cost carriers control 50% of the originating passengers
at the two endpoints of a route this analysis estimates that the observed carrier’s price
would fall by 0.005%, 0.026%, and 0.028% for the 20th, median, and 80th percentiles
respectively. Thus controlling for the observed carrier’s share on the route increases in
low-cost market share, either on the route or at the endpoint airports, lower the carrier’s
low-end prices and the median and high-end prices by 5 times the magnitude of the low-
end effect.
Controlling for low-cost route market share reveals a downward force on prices;
however, the effect of a low-cost hub at either or both of the endpoints is not significant
at any of the 3 fare levels. The lack of a negative and significant low-cost hub effect is
not consistent with the other low-cost parameters. However, low-cost hubs are not always
exclusively served by low-cost carriers. For example, JetBlue’s primary hub is at New
York John F. Kennedy (JFK) airport. Since this is a major metropolitan airport, with a
large base of business travelers, the low-cost hub does not significantly affect the fares of
flights terminating and originating the JFK.
The effect on prices of the presence of a low-cost carrier on a route, as explained
above, is negative and significant. In contrast, the effect of the level of low-cost
participation on a route, as measured by the low-cost route Herfindahl, is positive and
significant at the 50th and 80th percentile price. This result indicates that having one low-
cost carrier on a route does not provide the maximum low-cost effect and downward
movement in market prices. A lower Herfindahl index (HHI) reduces the magnitude of
60
the estimated effect of the low-cost route Herfindahl (LWCSTHERF) parameter. Thus
both the presence and level of the low-cost presence on a route is necessary to determine
the full low-cost route effect.
The estimated baseline equation with low-cost variables also reveals that the low-
cost effect is not limited to the 4 low cost variables. The low-cost effect extends to
concentration and route characteristic parameters as well. In fact, this analysis argues that
many of the concentration effects and the hub effect that Borenstein (1989) and others
suggest are critical to understanding airline pricing are significantly less important than
previously suggested. Case in point, the baseline equation estimates that the presence of a
hub at either or both endpoints of a route raises prices between 6.8% and 29.2%.
Controlling for the presence of low-cost carriers, with 4 low cost variables, the estimated
effect of a hub ranges from 1.4% to 10.5%. Thus at the low-end the hub effect fell by
80% and at the high-end the hub effect declined by 64%.
Without a low-cost presence a carrier at a hub can expect to add a 29.2% premium
to the highest fares; however, once a low-cost carrier enters the hub market the
incumbent carrier’s hub premium is only 10.5%. Much of the hub effect is actually due to
the lack of a low-cost presence. Low-cost carriers carry 12.1% of all domestic traffic
originating and terminating at a major carrier’s hub airport. The low-cost carriers’ small
hub market share results in a comparatively large decline in both the hub effect and the
level of localized market dominance. In the estimated baseline equations with low-cost
variables, the HUB variable is no longer significant in the 20th percentile equation; the
HUB variable is significant at the 1% level in the median and 80th percentile equations.
The low-cost impact on hubs is noteworthy, because 49.3% of all airline traffic originates
61
and or terminates at a major carrier’s hub. Thus understanding how low-cost carriers
affect the hub effect provides some intuition for their impact on the larger airline market.
Figure 3 diagrams that low-cost effect on the hub parameter across the three fare
percentiles.
Figure 3 Impact of a Low-Cost Presence on Hubs
Hub Airport at Origin or Destination
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
20th 50th 80th
Price Percentile
Coe
ffic
ient
Val
ue
Baseline Equation Baseline Equation with Low-Cost Variables
Note: HUB is significant at the 1% level in the baseline equation at all three percentiles. It is significant at the 1% level in the 50th and 80th percentile low-cost equations and is not significant in the 20th percentile equation.
The estimated change in price associated with hubs is overstated; moreover, the
effect of frequency and slot constraints on the market price is also over estimated. The
estimated effect of the frequency of flights on a route (FREQ), served by a particular
carrier, is positive and significant at the 1% level over each fare percentile. This is the
case for both the baseline and low-cost baseline equations. The coefficients of FREQ in
62
the low-cost baseline equation are 0.07, 0.05, and 0.14 for the 20th, 50th, and 80th
percentiles respectively. The addition on the 4 low-cost variables lowers the baseline
equation’s FREQ parameters by 25% at the 20th percentile, 55% at the median, and 42%
at the 80th percentile. Thus the additional market power that carriers may derive from
offering a high frequency of flights is significantly lower when a low-cost carrier is
present on a route. Moreover, a large percentage of the frequency effect is attributable to
lack of a low-cost presence on the route, not the actual frequency level. Figure 4 provides
a visual representation of the impact the low-cost parameters on the coefficient value of
FREQ across fares.
Figure 4 Impact of a Low-Cost Presence on Frequency
Frequency
0.00
0.05
0.10
0.15
0.20
0.25
0.30
20th 50th 80th
Price Percentile
Coe
ffic
ient
Val
ue
Baseline Equation Baseline Equation with Low-Cost Variables
Note: FREQ is significant at the 1% level in the baseline equation at the median and 80th percentile. FREQ is significant in the 20th percentile baseline equation at the 2% level. It is not significant in any of the low-cost equations.
63
The effect of slot constraints on prices is over estimated. The estimated effect of
the slot constraints on a route (SLOT) is not significant at the 10% level for any fare
percentile in low-cost baseline equations. The baseline equation estimates that a slot
restricted airport at the origin or destination causes the 20th, 50th, and 80th percentile fares
to rise 6.8%, 11.2%, and 22.3% respectively. However, none of the SLOT coefficients
are significant in low-cost baseline equations, thus the SLOT effect is nonexistent when
controlling for low-cost carriers. Thus the additional market power that carriers may
derive from barriers to entry, like slot constraints, are not material with the presence of a
low-cost carrier. Moreover, the entire slot effect can be ascribed to the lack of a low-cost
presence. Figure 5, below, provides exemplifies the impact the low-cost variables on the
coefficient value of SLOT across each price level.
64
Figure 5 Impact of a Low-Cost Presence on Slot Constraints
Slot Restricted Airport
0.00
0.05
0.10
0.15
0.20
0.25
20th 50th 80th
Price Percentile
Coe
ffic
ient
Val
ue
Baseline Equation Baseline Equation with Low-Cost Variables
Note: SLOT is significant at the 1% level in the baseline equation at the median and 80th percentile. SLOT is not significant in the 20th percentile baseline equation. It is also not significant in the 20th and 50th percentile low-cost equations. SLOT is significant at the 10% level in the low-cost equation at the 80th percentile.
Including low-cost parameters in the baseline equation has a similar de-
magnifying effect on EQUIP. At each fare percentile the coefficient value drops, but
remains significant. Without a low-cost presence the downward effect of larger aircraft is
overestimated. Thus the cost savings implied in EQUIP are lower than the baseline
equation suggests. The effect of TOURIST is also overestimated in the baseline equation.
The inclusion of low-cost controls lowers the absolute value of the TOURIST effect to
between 12.5% and 36.6%. Thus a large percentage of the TOURIST effect is dependent
65
upon a low-cost presence or lack thereof. This result is consistent with the fact that low-
cost carriers transport 40.8% of all tourist traffic. Furthermore, this low-cost presence
effect is of note because 28.2% of all traffic travels on tourist oriented routes. Even
though tourist oriented routes carry a higher percentage of price sensitive passengers
(Richards 1996), the low-cost effect is still significant. A comparatively higher low-cost
market share (in the range of 40% to 50%) is needed to impact (induce a low-cost effect
on) a price sensitive market segment, whereas on relatively price insensitive routes, such
as those originating and or terminating at a hub, a much lower market share of 10-15% is
necessary.
The low-cost parameters have no material impact on the market distance
(MKTDIS), route Herfindahl (RUTHERF), or the cost parameter (COMPCOST).
Likewise, there is no significant change in the route market share (RUTSHARE) variable.
Thus Borenstein (1989) and the baseline equation of this analysis do not overestimate the
impact of route concentration on prices. The route concentration effect continues to be
strong and an important factor in estimating market prices. The stability of the route
concentration effect contrasts sharply with the overestimated effects associated with
hubs, airport concentration, slot constraints, etc.
The three estimated six segment distance equations quantify the change in market
prices over 500 mile market segments. The inclusion of the 500 mile market distance
dummy variables does not change the estimated coefficients for the variables described
thus far. The baseline six segment distance equation, which does not contain low-cost
variables, brings to light the rising price premium linked to longer routes. A flight 501 to
1000 miles in length is estimated to have a price 0.5% higher than a flight of 1 to 500
66
miles in length, for the median fare. This analysis estimates effect of a 2501 to 3000 mile
route over the baseline distance is a 43.1% increase in price, for the median fare. This
positive relationship between price and the segmented distance effect also holds for the
20th and 80th percentile equations. Figure 6 diagrams the growth in the price premium
over the baseline route length as the route distance rises. Table 7 includes a list of
significance levels for the market distance coefficients included in Figure 6. It is
important to note that this graduated increase appears to take a log linear form, which is
at odds with the linear increase in route length that the indicator variables measure. The
price of tickets on longer routes rises more quickly than the route length. A route 1 to 500
miles longer than the baseline is estimated to have a 0.5% premium, while a route 1500 to
2000 miles longer has a 29.4% premium. The longest routes command a high premium;
however, a large percentage of the estimated distance effect is in point of fact due to the
lack of a low-cost presence.
67
Figure 6 Impact of a Low-Cost Presence on the Market Distance Effect
Market Distance Effect over Price Percentiles
-0.100
0.000
0.100
0.200
0.300
0.400
0.500
501-1000 1001-1500 1501-2000 2001-2500 2501-3000
Market Distance (miles)
Coef
ficie
nt V
alue
20th percentile 20th percentile with low-cost 50th percentile50th percentile with low-cost 80th percentile 80th percentile with low-cost
The six segment distance equations with low-cost parameters quantify the low-
cost effect within distinct 500 mile market segments. Across each fare type and each fare
percentile the effect of the market distance on price is lower when controlling for the
presence of low-cost carriers. The 501-1000 mile segment parameter is not significant at
any fare percentile in the six segment equation with low-cost variables. The 1501-2000
mile parameter’s level of significance changes with the introduction of the low-cost
parameters. At the 20th percentile the MKTDIS_4 (1501-2000 mile) variable is
significant at the 10% level and is no longer significant with the low-cost measures. The
same low-cost effect is apparent at the 80th percentile as well. MKTDIS_4 is significant
at the 20% level, but in the estimated equation with the low-cost variables MKTDIS_4 is
no longer significant.
68
The largest flattening of the change in the price distribution caused by the market
segment effect is at the high end of the distribution. A larger percentage change in a
percentile airfare is not due to a higher level of low-cost market share in that market
segment. Table 6 provides a breakdown of low-cost market share by 500 mile segments.
Figure 6 and the summary statistics in the table below refute the assertion that a large
low-cost presence is required to significantly alter the distribution of airfares in a market
segment. Longer routes, with a lower percentage of price sensitive passengers, require a
lower low-cost presence to compact and shift the price distribution lower. This result has
two implications. First, the lower market presence of low cost carriers in longer routes
implies that higher barriers to entry exist on longer routes. However, the late arrival of
low-cost carriers to the long haul market may also explain the market share distribution in
Table 6. Second, shorter routes with already compact price distributions require a
comparatively large low-cost market share to produce a measurable low-cost effect on
these routes. If prices are already low a comparatively larger low-cost presence is needed
to discipline prices further.
Table 6 Low-Cost Market Share by Market Segment
Market Segment Total Passengers Low-Cost Passengers Low-Cost Market Share0-500 miles 2201362 987145 44.84%501-1000 miles 2116413 576565 27.24%1001-1500 miles 1024699 294356 28.73%1501-2000 miles 576565 123472 21.42%2001-2500 miles 375908 71013 18.89%2502-3000 miles 155277 18533 11.94%
In general, including the low-cost variables causes the estimated market distance
coefficients values to rise and level of significance to fall. This trend along with the fall
in the coefficient values supports the assertion that the effect of distance on prices is
69
overstated in the estimated baseline equations, which lack low-cost variables. Moreover,
since the baseline results are similar to Borenstein (1989) it is expected that the inclusion
of low-cost variables in the estimated equations would produce a similar result. Table 7
summarizes the significance levels of each of the market distance variables. The
summary establishes that the market distance is, on average, a less important predictor of
price than the regular six segment equation suggests.
Table 7 Summary of Significance Levels for Market Distance Variables4
Market Segment Price Percentile (miles) 20th 20th low-cost 50th 50th low-cost 80th 80th low-cost501-1000 0.265 0.548 0.821 0.714 0.808 0.1621001-1500 0.016 0.049 0.010 0.043 0.196 0.8681501-2000 0.058 0.175 0.000 0.004 0.182 0.9302001-2500 0.000 0.001 0.000 0.000 0.000 0.1542501-3000 0.032 0.390 0.000 0.000 0.000 0.035
An important hypothesis of this paper is that the low-cost effect is quantifiable in
the regional and the long haul market as well. This paper estimates that 83.2% of airline
traffic is regional and the remainder is long haul. In place of the 500 mile segmented
distance variables two dummy variables are used. A representative route for the long haul
market segment is New York (JFK) to Los Angeles (LAX). This route is the most
traveled long haul route in the United States; moreover, the low-cost market share on the
route is 0.06%. American Airlines is the dominant carrier on the route with a market
share of 46.3%. The most traveled regional route is Washington National (DCA) to New
York LaGuardia (LGA). Delta Airlines has the highest market share on the route at
53.5%. Also, there is no low-cost presence on the DCA to LGA route. Although, the most
4 The left column represents each 500 mile market segment that was included in the six segment market distance regressions. Each of the table’s interior cells correspond to an estimated significance level for a particular 500 mile market segment estimated in an equation regressed on a particular price percentile, noted in the column heading.
70
prominent regional and long haul markets lack a nontrivial low-cost presence, the low-
cost effect is significant in both market segments.
The long haul route, with the highest number of low-cost passengers, is Chicago
Midway (MDW) to Las Vegas (LAS). The dominant carriers on this route include
American Trans Air with 43.1%, National Airlines with 31.0%, and Southwest with a
25.8% market share. In fact, low-cost carriers transport 100% of the traffic on this tourist
oriented route. In the regional market, the most traveled route with a low-cost presence is
Los Angeles (LAX) to Oakland (OAK). Southwest dominates this regional route with a
market share of 92.6%; Southwest is the only low-cost carrier on the route. LAX to OAK
exemplifies the low-cost carrier’s preference for having either one or both of the endpoint
airports be a secondary airport. Even though low-cost carriers often do not compete with
the major airlines directly, the low cost presence is felt across market segments.
The short/long haul market segment equations indicate that the low-cost effect
extends across the entire airline market; moreover this effect is significant at each
measured fare percentile. Note that the effects of the other parameters discussed
previously remain the same unless otherwise noted. Tables 21-26 contain the summary
results of each of the short/long haul equations. The estimated results of three baseline
short/long haul equations demonstrate the high price premium associated with long haul
routes versus regional routes. Case in point, the effect of the long haul market distance
group variable on price is estimated to be 0.06, 0.17, and 0.14 at the 20th, 50th, and 80th
percentiles respectively. Thus, at the median airfare, a long haul route is estimated to
have a price 17% higher than a regional route controlling for route specific characteristics
besides length.
71
The result at the median mirrors the results at the 20th and 80th percentiles;
moreover, each of the long haul market distance group (MKTDISLNG) parameters is
significant at the 1% level. Thus there is a considerable price premium associated with
the long haul market that cannot be explained by the higher total (though lower per mile)
cost of long haul routes. A possible explanation for the price premium is that low-cost
carriers have a much smaller exposure to the long haul market than the regional market.
This lower market penetration rate triggers a lower level of competitive pressure on the
incumbent major carriers, which are not inclined to lower ticket prices. As a result, long
haul routes carry a price premium and make obvious the dominance of the major carriers.
Just as the price premium associated with the frequency of flights is overestimated
so is the price difference between regional and long haul routes at each of the three
measured fare percentiles. At the 20th, 50th, and 80th percentiles the long haul distance
effect (the price premium over a comparable regional route) is estimated to be 4.1%,
11.4%, and 5.4%. The 20th percentile coefficient is significant at the 10% level, the
median coefficient is significant at the 1% level, and the 80th percentile coefficient is
significant at the 5% level. Not only has the estimated magnitude of each MKTDISLNG
variable fallen, but the level of significance has fallen as well. The declining level of
significance indicates that regional and long haul price distinctions are less important
than the baseline short/long haul equation would imply. Moreover, between 33% and
62% of the long haul market distance effect is explained by the lack of a low-cost
presence. Figure 7 illustrates the significant impact of low-cost carriers on the long haul
market, across all types of fares.
72
Figure 7 Impact of a Low-Cost Presence on the Long Haul Effect
Long Haul Market Distance Group
0.000
0.020
0.040
0.060
0.080
0.100
0.120
0.140
0.160
0.180
20th 50th 80th
Price Percentile
Coe
ffic
ient
Val
ue
Short/Long Haul Equation Short/Long Haul Equation with Low-Cost Variables
Note: MKTDISLNG is significant at the 1% level in all three baseline short/long haul equations. It is also significant at the 10% level, 1% level, and 2% level in the 20th, 50th, and 80th percentile low-cost equations respectively.
An implication of the short/long haul equation results is that there is a difference
in pricing between regional and short haul routes regardless of a low-cost presence;
however, the more important observation is that low-cost carriers significantly lower the
difference in prices between regional and long haul routes. Thus the low-cost carrier is no
longer pigeon holed to the regional market; however, the regional market still accounts
for 83.2% of all airline traffic. The three largest long haul low-cost carriers Southwest,
America Trans Air, and JetBlue account for 75.4% of all long haul low-cost traffic. Table
8 summarizes the low-cost carriers that serve the long haul market (routes 1501-3000
miles in length) and provides each carrier’s market share. It is important to note that
73
Southwest is the largest low-cost carrier in the long haul market as well as the regional
market. JetBlue is the third largest low-cost carrier and only carries 3.12% of all long
haul traffic. A carrier’s share of the low-cost traffic may be proportional to that carrier’s
share of the low-cost effect. Thus much of the low-cost effect on MKTDISLNG and
more generally HUB, SLOT, FREQ, and ORGSHARE can be directly attributed to
Southwest and not JetBlue.
Table 8 Low-Cost Airline Market Share in the Long Haul Market
Airline Passengers Share of Low-Cost Traffic Share of all TrafficJetBlue Airways 33496 15.70% 3.12%Frontier Airlines 7585 3.56% 0.71%National Airlines 30244 14.18% 2.82%Vanguard Airlines 1482 0.69% 0.14%Spirit Airlines 11644 5.46% 1.08%Sun Country Airlines 1095 0.51% 0.10%America Trans Air 36405 17.07% 3.39%Southwest Airlines 90971 42.65% 8.47%Casino Express Airlines 298 0.14% 0.03%
Low-cost carriers carry only 19.8% of all long haul traffic; however, the impact of
these airlines on the pricing structure of the long haul market is large and significant.
Competitive entry does not have to be large, but has to be low-cost to affectively
discipline prices in the long haul market. Moreover, low-cost carriers only carry one fifth
of all long haul traffic, but their presence or lack thereof explains 33% to 62% of the
market distance effect measured by MKTDISLNG. This analysis argues that the
estimated effect of a low-cost presence must be taken into account in order to derive an
accurate understanding of the current state of the airline industry. Not taking into account
the low-cost presence in a marketplace results in an overestimation of the estimated effect
of critical market factors, such as airport concentration, hub presence, and flight
frequency on the distribution of prices.
74
8. Conclusion
In the current airline market, the impact of a new breed of air carrier, the low-cost
carrier, on airfares, is significant and broadly distributed. Previous industrial organization
literature argues that concentration measures are critical to understanding the dynamics of
pricing in the airline markets. The previous analyses of Borenstein (1989) and others
lacked a key insight: the low-cost carrier is an important market participant. However, the
Borenstein paper lays a solid groundwork that forms the base of this thesis. Without the
low-cost parameters in the 3 sets of equations, the estimated results of this paper closely
resemble Borenstein’s results. The inclusion of the low-cost parameters in the baseline
equation results in a set of estimated results that provide a more accurate accounting of
what factors really impact airline industry price distributions and more specifically the
prices that each consumer pays.
The low-cost effect results in the compaction of price distributions and the
lowering of the associated central tendency of those distributions across all market
segments. Moreover, a low-cost presence is correlated with a higher degree of
competitiveness and a decline in the relevance of concentration, frequency, hubs, and
capacity constraints in predicting market prices. The low-cost effect is statistically
significant and measurable across the entire distribution of airfares. The effect is also
significant and measurable for a route of variable length. A low-cost presence lowers
prices at the low end, median, and high-end of the price distribution on regional and long
haul routes. In fact, the largest degree of flattening occurs at the high end of the price
75
distribution. Thus routes with the highest levels of market dominance would experience
the largest change in their distribution of prices, with the entry of a low-cost carrier.
Moreover, a low-cost presence is estimated to cause a flatter price distribution, a
higher degree of competitiveness, and a decline in the relevance of concentration,
frequency, hubs, and capacity constraints in predicting market prices. This new breed of
air carrier disciplines the incumbent major carriers that rely on market dominance to
extract higher rents to support the higher cost basis of the hub-and-spoke networks these
airlines use. The disciplining effect of low-cost carriers on the premium major airlines
derive from hub airports may indicate that the higher cost basis of a traditional hub is no
longer justified. Moreover, the low-cost airline model, which is built around using a
point-to-point network, should continue to be a successful model for new entrants. While
it is unlikely that any major carriers will completely abandon their hub-and-spoke
networks, the partial dismantling of those hubs, in the form of de-peaking, will continue.
Market forces have and will continue to cause these structural changes. Furthermore,
there are policies can accelerate this shift and the magnitude of the low-cost presence in
the airline markets.
This paper argues that a shift in critical thought is necessary to analyze today’s
airline industry; moreover, policies to encourage competition and a flatter price
distribution will only bring about change if these policies are designed to encourage a
larger low-cost presence in the marketplace. Policies that try to reduce concentration
levels or dismantle hub airports will face stiff political opposition. The estimated baseline
results suggest that regulations targeting these factors would change the distribution of
airfares. In reality, the concentration and hub effects are significantly smaller than this
76
paper first proposes. Targeting these factors would not result in a dramatic change in the
competitiveness of airline routes as measured by the shape and location of the fare
distribution curve. Moreover, this kind of targeted policy would incite fierce lobbying by
the major nationwide carriers. The recent success of these carriers to obtain multi-billion
dollar subsidies from the federal government points to the strength of the corporate
airline lobby and airline union lobby.
Given that traditional policies, which encourage lower levels of concentration, are
relatively ineffectual and politically infeasible, is there a viable alternative? Instead of
attempting to reduce the impact of factors with a negative effect on the market, a
successful policy must try to raise the impact of factors with a positive effect on the
market. A viable policy must provide ample opportunity for low-cost entry in order to
compact the fare distribution and return rents to the American consumer. The primary
barrier to low-cost entry is the lack of airport infrastructure at many of America’s largest
metropolitan airports. In order to achieve the disciplining effect of a low-cost presence,
the government should implement policies that encourage capacity constrained airports to
build new infrastructure. Moreover, regulatory barriers to airport expansion should be
loosened. A lower level of airport expansion regulation would encourage airports to
undertake needed infrastructure improvements to increase capacity.
A build out of airport infrastructure, targeted at the most congested airports,
would encourage low-cost carriers to enter a greater number of non-secondary markets.
The newly constructed airport resources need to be allocated so that incumbent carriers
do not receive a preferential claim to those resources. New resources alone do not
discipline markets and compact the price distribution. Low-cost carriers must have access
77
to those resources; otherwise, the infrastructure projects are nothing more than a subsidy
to the major carriers. With greater access to once congested markets, the low-cost effect,
on the market price distribution in these congested markets, should continue to increase
in magnitude. Although, hubs still have higher fares than other airports, even when
controlling for low-cost carriers, the significant role of the low-cost carrier in today’s
airline market indicates that dominance and market power do not go unchecked in the
U.S. airline industry; moreover, this paper estimates that the entry of a low-cost carrier
lowers fares while at the same time it lessens the effect of concentration and increases
competition in the U.S. airline industry.
78
9. Regression Tables
Table 9 – Baseline Equation – 20th Percentile5
Dependent Variable: Price (ln) Beta Std. Error Std. Beta t Sig. (Constant) 2.433 0.128 19.067 0.000 Market Distance (ln) 0.337 0.009 0.530 35.922 0.000 Load Factor (ln) 0.008 0.018 0.005 0.430 0.667 Equipment (ln) -0.080 0.016 -0.273 -5.130 0.000 Frequency (ln) 0.097 0.016 0.322 6.187 0.000 Origin & Dest. Market Share (ln) -0.034 0.008 -0.073 -4.297 0.000 Origin & Dest. Herfindahl (ln) 0.044 0.013 0.051 3.316 0.001 Route Market Share 0.154 0.034 0.104 4.477 0.000 Route Herfindahl 0.033 0.042 0.014 0.792 0.428 Comparable Cost (ln) -0.010 0.007 -0.032 -1.417 0.157 Tourist Route -0.077 0.017 -0.058 -4.598 0.000 Hub Airport at Origin or Dest. 0.068 0.014 0.062 4.940 0.000 Slot Restricted Airport 0.073 0.048 0.018 1.530 0.126
Table 10 – Baseline Equation – 20th Percentile with Low-Cost Variables
Dependent Variable: Price (ln) Beta Std. Error Std. Beta t Sig. (Constant) 2.433 0.128 18.940 0.000 Market Distance (ln) 0.337 0.009 0.528 35.788 0.000 Load Factor (ln) 0.013 0.017 0.009 0.755 0.450 Equipment (ln) -0.055 0.016 -0.188 -3.429 0.001 Frequency (ln) 0.073 0.016 0.243 4.546 0.000 Origin & Dest. Market Share (ln) -0.037 0.008 -0.079 -4.648 0.000 Origin & Dest. Herfindahl (ln) 0.070 0.015 0.080 4.762 0.000 Route Market Share 0.147 0.034 0.098 4.262 0.000 Route Herfindahl 0.033 0.045 0.014 0.738 0.460 Low-Cost Origin & Dest. Market Share (ln) -0.011 0.005 -0.033 -2.200 0.028 Low-Cost Route Market Share -0.146 0.078 -0.084 -1.866 0.062 Low-Cost Route Herfindahl -0.003 0.091 -0.002 -0.035 0.972 Comparable Cost (ln) -0.010 0.007 -0.033 -1.478 0.139 Tourist Route -0.066 0.017 -0.050 -3.824 0.000 Hub Airport at Origin or Dest. 0.014 0.015 0.013 0.891 0.373 Low-Cost Hub at Origin or Dest. -0.034 0.023 -0.019 -1.507 0.132 Slot Restricted Airport 0.036 0.048 0.009 0.755 0.451 5 In each of the summary tables the estimated equations are presented in the following format. Beta represents the coefficient value of each regressed variable. Std. Error is the standard error associated with each variable. Likewise, Std. Beta is the standardized beta coefficient value. The t-statistic for each beta is listed under the heading t. Finally, Sig. stands for the significance level of each parameter. A table with the R2 figures and number of observations for each of the 18 equations is included in this paper as Table 27.
79
Table 11 – Baseline Equation – 50th Percentile
Dependent Variable: Price (ln) Beta Std. Error Std. Beta t Sig. (Constant) 3.542 0.118 30.037 0.000 Market Distance (ln) 0.256 0.009 0.451 29.484 0.000 Load Factor (ln) -0.039 0.016 -0.030 -2.401 0.016 Equipment (ln) -0.120 0.014 -0.459 -8.331 0.000 Frequency (ln) 0.128 0.015 0.476 8.820 0.000 Origin & Dest. Market Share (ln) -0.009 0.007 -0.020 -1.155 0.248 Origin & Dest. Herfindahl (ln) 0.000 0.012 0.000 0.025 0.980 Route Market Share 0.207 0.032 0.156 6.497 0.000 Route Herfindahl 0.017 0.039 0.008 0.446 0.655 Comparable Cost (ln) 0.005 0.006 0.019 0.819 0.413 Tourist Route -0.106 0.016 -0.090 -6.858 0.000 Hub Airport at Origin or Dest. 0.175 0.013 0.180 13.761 0.000 Slot Restricted Airport 0.112 0.044 0.030 2.531 0.011 Table 12 – Baseline Equation – 50th Percentile with Low-Cost Variables
Dependent Variable: Price (ln) Beta Std. Error Std. Beta t Sig. (Constant) 3.374 0.115 29.330 0.000 Market Distance (ln) 0.253 0.008 0.446 30.070 0.000 Load Factor (ln) -0.021 0.016 -0.017 -1.369 0.171 Equipment (ln) -0.047 0.014 -0.180 -3.267 0.001 Frequency (ln) 0.057 0.014 0.213 3.976 0.000 Origin & Dest. Market Share (ln) -0.021 0.007 -0.050 -2.911 0.004 Origin & Dest. Herfindahl (ln) 0.079 0.013 0.102 6.055 0.000 Route Market Share 0.177 0.031 0.133 5.748 0.000 Route Herfindahl -0.010 0.040 -0.005 -0.260 0.795 Low-Cost Origin & Dest. Market Share (ln) -0.053 0.004 -0.178 -11.784 0.000 Low-Cost Route Market Share -0.446 0.070 -0.289 -6.376 0.000 Low-Cost Route Herfindahl 0.213 0.082 0.114 2.603 0.009 Comparable Cost (ln) 0.001 0.006 0.003 0.132 0.895 Tourist Route -0.071 0.015 -0.060 -4.583 0.000 Hub Airport at Origin or Dest. 0.056 0.014 0.058 4.047 0.000 Low-Cost Hub at Origin or Dest. 0.001 0.020 0.000 0.039 0.969 Slot Restricted Airport 0.019 0.043 0.005 0.456 0.648
80
Table 13 – Baseline Equation – 80th Percentile
Dependent Variable: Price (ln) Beta Std. Error Std. Beta t Sig. (Constant) 3.546 0.134 26.553 0.000 Market Distance (ln) 0.358 0.010 0.516 36.484 0.000 Load Factor (ln) -0.106 0.018 -0.068 -5.762 0.000 Equipment (ln) -0.214 0.016 -0.666 -13.069 0.000 Frequency (ln) 0.251 0.016 0.761 15.233 0.000 Origin & Dest. Market Share (ln) 0.032 0.008 0.062 3.811 0.000 Origin & Dest. Herfindahl (ln) -0.052 0.014 -0.055 -3.776 0.000 Route Market Share 0.189 0.036 0.116 5.237 0.000 Route Herfindahl 0.110 0.044 0.044 2.490 0.013 Comparable Cost (ln) 0.006 0.007 0.017 0.785 0.432 Tourist Route -0.237 0.018 -0.165 -13.501 0.000 Hub Airport at Origin or Dest. 0.292 0.014 0.245 20.256 0.000 Slot Restricted Airport 0.223 0.050 0.050 4.458 0.000 Table 14 – Baseline Equation – 80th Percentile with Low-Cost Variables
Dependent Variable: Price (ln) Beta Std. Error Std. Beta t Sig. (Constant) 3.355 0.126 26.713 0.000 Market Distance (ln) 0.354 0.009 0.510 38.498 0.000 Load Factor (ln) -0.080 0.017 -0.051 -4.684 0.000 Equipment (ln) -0.102 0.016 -0.318 -6.484 0.000 Frequency (ln) 0.145 0.016 0.441 9.198 0.000 Origin & Dest. Market Share (ln) 0.015 0.008 0.030 1.941 0.052 Origin & Dest. Herfindahl (ln) 0.038 0.014 0.040 2.686 0.007 Route Market Share 0.138 0.034 0.085 4.095 0.000 Route Herfindahl 0.031 0.044 0.012 0.711 0.477 Low-Cost Origin & Dest. Market Share (ln) -0.055 0.005 -0.152 -11.215 0.000 Low-Cost Route Market Share -0.984 0.076 -0.521 -12.879 0.000 Low-Cost Route Herfindahl 0.577 0.089 0.252 6.460 0.000 Comparable Cost (ln) 0.000 0.007 0.001 0.029 0.977 Tourist Route -0.175 0.017 -0.122 -10.397 0.000 Hub Airport at Origin or Dest. 0.105 0.015 0.088 6.907 0.000 Low-Cost Hub at Origin or Dest. -0.019 0.022 -0.010 -0.852 0.394 Slot Restricted Airport 0.075 0.047 0.017 1.605 0.109
81
Table 15 – Six Segment Distance Equation – 20th Percentile
Dependent Variable: Price (ln) Beta Std. Error Std. Beta t Sig. (Constant) 2.755 0.157 17.506 0.000 Market Distance (ln) 0.285 0.019 0.447 14.873 0.000 Load Factor (ln) 0.005 0.017 0.004 0.285 0.776 Equipment (ln) -0.090 0.016 -0.310 -5.752 0.000 Frequency (ln) 0.108 0.016 0.362 6.838 0.000 Origin & Dest. Market Share (ln) -0.030 0.008 -0.064 -3.729 0.000 Origin & Dest. Herfindahl (ln) 0.037 0.013 0.043 2.770 0.006 Route Market Share 0.134 0.035 0.091 3.876 0.000 Route Herfindahl 0.033 0.042 0.014 0.779 0.436 Comparable Cost (ln) -0.013 0.007 -0.040 -1.806 0.071 Tourist Route -0.087 0.017 -0.066 -5.169 0.000 Hub Airport at Origin or Dest. 0.071 0.014 0.066 5.138 0.000 Slot Restricted Airport 0.077 0.048 0.019 1.612 0.107 Market Distance of 501-1000 miles 0.026 0.023 0.024 1.116 0.265 Market Distance of 1001-1500 miles 0.080 0.033 0.055 2.415 0.016 Market Distance of 1501-2000 miles 0.079 0.042 0.041 1.897 0.058 Market Distance of 2001-2500 miles 0.217 0.049 0.091 4.411 0.000 Market Distance of 2501-3000 miles 0.134 0.063 0.035 2.147 0.032
82
Table 16 – Six Segment Distance Equation – 20th Percentile with Low-Cost Variables
Dependent Variable: Price (ln) Beta Std. Error Std. Beta t Sig. (Constant) 2.683 0.159 16.889 0.000 Market Distance (ln) 0.299 0.019 0.468 15.504 0.000 Load Factor (ln) 0.010 0.017 0.007 0.595 0.552 Equipment (ln) -0.067 0.016 -0.230 -4.091 0.000 Frequency (ln) 0.086 0.016 0.286 5.213 0.000 Origin & Dest. Market Share (ln) -0.033 0.008 -0.072 -4.154 0.000 Origin & Dest. Herfindahl (ln) 0.060 0.015 0.069 4.010 0.000 Route Market Share 0.133 0.035 0.090 3.862 0.000 Route Herfindahl 0.031 0.045 0.014 0.698 0.485 Low-Cost Origin & Dest. Market Share (ln) -0.007 0.005 -0.022 -1.408 0.159 Low-Cost Route Market Share -0.118 0.079 -0.069 -1.491 0.136 Low-Cost Route Herfindahl -0.020 0.092 -0.010 -0.222 0.824 Comparable Cost (ln) -0.012 0.007 -0.039 -1.719 0.086 Tourist Route -0.079 0.017 -0.060 -4.519 0.000 Hub Airport at Origin or Dest. 0.021 0.016 0.020 1.342 0.180 Low-Cost Hub at Origin or Dest. -0.043 0.023 -0.024 -1.857 0.063 Slot Restricted Airport 0.044 0.048 0.011 0.912 0.362 Market Distance of 501-1000 miles 0.014 0.023 0.013 0.600 0.548 Market Distance of 1001-1500 miles 0.066 0.033 0.045 1.970 0.049 Market Distance of 1501-2000 miles 0.057 0.042 0.030 1.355 0.175 Market Distance of 2001-2500 miles 0.165 0.050 0.070 3.325 0.001 Market Distance of 2501-3000 miles 0.055 0.064 0.014 0.860 0.390
83
Table 17 – Six Segment Distance Equation – 50th Percentile
Dependent Variable: Price (ln) Beta Std. Error Std. Beta t Sig. (Constant) 4.031 0.144 27.994 0.000 Market Distance (ln) 0.180 0.018 0.316 10.268 0.000 Load Factor (ln) -0.044 0.016 -0.035 -2.736 0.006 Equipment (ln) -0.142 0.014 -0.546 -9.887 0.000 Frequency (ln) 0.151 0.014 0.565 10.426 0.000 Origin & Dest. Market Share (ln) 0.001 0.007 0.002 0.129 0.898 Origin & Dest. Herfindahl (ln) -0.014 0.012 -0.018 -1.142 0.254 Route Market Share 0.168 0.032 0.127 5.307 0.000 Route Herfindahl 0.027 0.039 0.013 0.702 0.483 Comparable Cost (ln) 0.000 0.006 -0.002 -0.070 0.944 Tourist Route -0.113 0.015 -0.096 -7.306 0.000 Hub Airport at Origin or Dest. 0.188 0.013 0.195 14.854 0.000 Slot Restricted Airport 0.123 0.044 0.034 2.827 0.005 Market Distance of 501-1000 miles 0.005 0.021 0.005 0.226 0.821 Market Distance of 1001-1500 miles 0.079 0.030 0.060 2.579 0.010 Market Distance of 1501-2000 miles 0.143 0.038 0.083 3.730 0.000 Market Distance of 2001-2500 miles 0.294 0.045 0.139 6.548 0.000 Market Distance of 2501-3000 miles 0.431 0.057 0.125 7.533 0.000
84
Table 18 – Six Segment Distance Equation – 50th Percentile with Low-Cost Variables
Dependent Variable: Price (ln) Beta Std. Error Std. Beta t Sig. (Constant) 3.712 0.142 26.171 0.000 Market Distance (ln) 0.205 0.017 0.360 11.906 0.000 Load Factor (ln) -0.026 0.016 -0.021 -1.680 0.093 Equipment (ln) -0.069 0.015 -0.265 -4.723 0.000 Frequency (ln) 0.080 0.015 0.298 5.435 0.000 Origin & Dest. Market Share (ln) -0.014 0.007 -0.034 -1.974 0.048 Origin & Dest. Herfindahl (ln) 0.063 0.013 0.081 4.720 0.000 Route Market Share 0.156 0.031 0.118 5.039 0.000 Route Herfindahl -0.004 0.040 -0.002 -0.095 0.925 Low-Cost Origin & Dest. Market Share (ln) -0.047 0.005 -0.159 -10.311 0.000 Low-Cost Route Market Share -0.388 0.071 -0.253 -5.498 0.000 Low-Cost Route Herfindahl 0.174 0.082 0.094 2.121 0.034 Comparable Cost (ln) -0.003 0.006 -0.009 -0.421 0.674 Tourist Route -0.082 0.016 -0.070 -5.271 0.000 Hub Airport at Origin or Dest. 0.075 0.014 0.077 5.288 0.000 Low-Cost Hub at Origin or Dest. -0.013 0.020 -0.008 -0.641 0.522 Slot Restricted Airport 0.037 0.043 0.010 0.859 0.390 Market Distance of 501-1000 miles -0.008 0.021 -0.008 -0.366 0.714 Market Distance of 1001-1500 miles 0.060 0.030 0.046 2.021 0.043 Market Distance of 1501-2000 miles 0.108 0.038 0.062 2.858 0.004 Market Distance of 2001-2500 miles 0.185 0.044 0.087 4.166 0.000 Market Distance of 2501-3000 miles 0.241 0.057 0.070 4.235 0.000
85
Table 19 – Six Segment Distance Equation – 80th Percentile
Dependent Variable: Price (ln) Beta Std. Error Std. Beta t Sig. (Constant) 3.927 0.164 24.003 0.000 Market Distance (ln) 0.302 0.020 0.432 15.180 0.000 Load Factor (ln) -0.109 0.018 -0.070 -6.011 0.000 Equipment (ln) -0.232 0.016 -0.727 -14.222 0.000 Frequency (ln) 0.270 0.016 0.821 16.380 0.000 Origin & Dest. Market Share (ln) 0.042 0.008 0.081 4.994 0.000 Origin & Dest. Herfindahl (ln) -0.064 0.014 -0.067 -4.593 0.000 Route Market Share 0.159 0.036 0.098 4.421 0.000 Route Herfindahl 0.131 0.044 0.052 2.978 0.003 Comparable Cost (ln) 0.002 0.007 0.005 0.246 0.806 Tourist Route -0.240 0.018 -0.166 -13.677 0.000 Hub Airport at Origin or Dest. 0.309 0.014 0.260 21.412 0.000 Slot Restricted Airport 0.232 0.049 0.052 4.687 0.000 Market Distance of 501-1000 miles -0.006 0.024 -0.005 -0.243 0.808 Market Distance of 1001-1500 miles 0.045 0.035 0.028 1.294 0.196 Market Distance of 1501-2000 miles 0.058 0.044 0.027 1.334 0.182 Market Distance of 2001-2500 miles 0.260 0.051 0.100 5.091 0.000 Market Distance of 2501-3000 miles 0.460 0.065 0.108 7.071 0.000
86
Table 20 – Six Segment Distance Equation – 80th Percentile with Low-Cost Variables
Dependent Variable: Price (ln) Beta Std. Error Std. Beta t Sig. (Constant) 3.448 0.155 22.178 0.000 Market Distance (ln) 0.346 0.019 0.495 18.351 0.000 Load Factor (ln) -0.082 0.017 -0.053 -4.806 0.000 Equipment (ln) -0.113 0.016 -0.354 -7.070 0.000 Frequency (ln) 0.156 0.016 0.474 9.681 0.000 Origin & Dest. Market Share (ln) 0.019 0.008 0.037 2.402 0.016 Origin & Dest. Herfindahl (ln) 0.032 0.015 0.033 2.189 0.029 Route Market Share 0.133 0.034 0.082 3.940 0.000 Route Herfindahl 0.043 0.044 0.017 0.968 0.333 Low-Cost Origin & Dest. Market Share (ln) -0.052 0.005 -0.143 -10.390 0.000 Low-Cost Route Market Share -0.940 0.077 -0.500 -12.159 0.000 Low-Cost Route Herfindahl 0.542 0.090 0.237 6.027 0.000 Comparable Cost (ln) -0.001 0.007 -0.002 -0.086 0.931 Tourist Route -0.180 0.017 -0.125 -10.535 0.000 Hub Airport at Origin or Dest. 0.119 0.015 0.100 7.710 0.000 Low-Cost Hub at Origin or Dest. -0.025 0.022 -0.012 -1.102 0.270 Slot Restricted Airport 0.084 0.047 0.019 1.796 0.073 Market Distance of 501-1000 miles -0.031 0.023 -0.027 -1.400 0.162 Market Distance of 1001-1500 miles 0.005 0.033 0.003 0.166 0.868 Market Distance of 1501-2000 miles -0.004 0.041 -0.002 -0.088 0.930 Market Distance of 2001-2500 miles 0.069 0.049 0.027 1.425 0.154 Market Distance of 2501-3000 miles 0.132 0.062 0.031 2.110 0.035
87
Table 21 – Short/Long Haul Market Segment Equation – 20th Percentile
Dependent Variable: Price (ln) Beta Std. Error Std. Beta t Sig. (Constant) 2.558 0.131 19.475 0.000 Market Distance (ln) 0.318 0.011 0.500 29.465 0.000 Load Factor (ln) 0.006 0.017 0.004 0.323 0.746 Equipment (ln) -0.086 0.016 -0.295 -5.497 0.000 Frequency (ln) 0.104 0.016 0.347 6.584 0.000 Origin & Dest. Market Share (ln) -0.032 0.008 -0.068 -3.956 0.000 Origin & Dest. Herfindahl (ln) 0.039 0.013 0.045 2.922 0.003 Route Market Share 0.141 0.035 0.095 4.076 0.000 Route Herfindahl 0.028 0.042 0.012 0.666 0.505 Comparable Cost (ln) -0.011 0.007 -0.036 -1.609 0.108 Tourist Route -0.083 0.017 -0.063 -4.943 0.000 Hub Airport at Origin or Dest. 0.068 0.014 0.063 4.919 0.000 Slot Restricted Airport 0.077 0.048 0.019 1.610 0.108 Long Haul Market Distance Group 0.061 0.021 0.042 2.858 0.004 Table 22 – Short/Long Haul Market Segment Equation – 20th Percentile with Low-Cost Variables
Dependent Variable: Price (ln) Beta Std. Error Std. Beta t Sig. (Constant) 2.536 0.133 19.072 0.000 Market Distance (ln) 0.324 0.011 0.507 29.877 0.000 Load Factor (ln) 0.011 0.017 0.008 0.632 0.528 Equipment (ln) -0.063 0.016 -0.215 -3.870 0.000 Frequency (ln) 0.082 0.016 0.272 5.000 0.000 Origin & Dest. Market Share (ln) -0.034 0.008 -0.074 -4.298 0.000 Origin & Dest. Herfindahl (ln) 0.061 0.015 0.071 4.143 0.000 Route Market Share 0.138 0.035 0.093 4.002 0.000 Route Herfindahl 0.028 0.045 0.012 0.620 0.536 Low-Cost Origin & Dest. Market Share (ln) -0.007 0.005 -0.022 -1.459 0.145 Low-Cost Route Market Share -0.136 0.078 -0.079 -1.744 0.081 Low-Cost Route Herfindahl -0.005 0.091 -0.002 -0.055 0.956 Comparable Cost (ln) -0.011 0.007 -0.035 -1.555 0.120 Tourist Route -0.074 0.017 -0.056 -4.256 0.000 Hub Airport at Origin or Dest. 0.018 0.016 0.016 1.138 0.255 Low-Cost Hub at Origin or Dest. -0.039 0.023 -0.022 -1.702 0.089 Slot Restricted Airport 0.042 0.048 0.010 0.891 0.373 Long Haul Market Distance Group 0.041 0.022 0.028 1.870 0.061
88
Table 23 – Short/Long Haul Market Segment Equation – 50th Percentile
Dependent Variable: Price (ln) Beta Std. Error Std. Beta t Sig. (Constant) 3.829 0.121 31.748 0.000 Market Distance (ln) 0.209 0.010 0.368 21.110 0.000 Load Factor (ln) -0.043 0.016 -0.034 -2.655 0.008 Equipment (ln) -0.133 0.014 -0.512 -9.277 0.000 Frequency (ln) 0.143 0.014 0.533 9.837 0.000 Origin & Dest. Market Share (ln) -0.003 0.007 -0.008 -0.426 0.670 Origin & Dest. Herfindahl (ln) -0.011 0.012 -0.014 -0.912 0.362 Route Market Share 0.179 0.032 0.135 5.623 0.000 Route Herfindahl 0.015 0.039 0.007 0.386 0.699 Comparable Cost (ln) 0.002 0.006 0.006 0.244 0.807 Tourist Route -0.109 0.015 -0.093 -7.056 0.000 Hub Airport at Origin or Dest. 0.180 0.013 0.186 14.202 0.000 Slot Restricted Airport 0.121 0.044 0.033 2.757 0.006 Long Haul Market Distance Group 0.171 0.020 0.131 8.691 0.000 Table 24 – Short/Long Haul Market Segment Equation – 50th Percentile with Low-Cost Variables
Dependent Variable: Price (ln) Beta Std. Error Std. Beta t Sig. (Constant) 3.579 0.119 30.120 0.000 Market Distance (ln) 0.223 0.010 0.392 23.039 0.000 Load Factor (ln) -0.025 0.016 -0.020 -1.583 0.113 Equipment (ln) -0.060 0.014 -0.232 -4.156 0.000 Frequency (ln) 0.071 0.015 0.267 4.900 0.000 Origin & Dest. Market Share (ln) -0.017 0.007 -0.040 -2.305 0.021 Origin & Dest. Herfindahl (ln) 0.066 0.013 0.085 4.989 0.000 Route Market Share 0.161 0.031 0.122 5.220 0.000 Route Herfindahl -0.013 0.040 -0.006 -0.314 0.754 Low-Cost Origin & Dest. Market Share (ln) -0.048 0.005 -0.164 -10.650 0.000 Low-Cost Route Market Share -0.428 0.070 -0.279 -6.132 0.000 Low-Cost Route Herfindahl 0.211 0.082 0.114 2.590 0.010 Comparable Cost (ln) -0.001 0.006 -0.004 -0.176 0.860 Tourist Route -0.076 0.016 -0.065 -4.933 0.000 Hub Airport at Origin or Dest. 0.066 0.014 0.069 4.762 0.000 Low-Cost Hub at Origin or Dest. -0.009 0.020 -0.006 -0.458 0.647 Slot Restricted Airport 0.031 0.043 0.009 0.737 0.461 Long Haul Market Distance Group 0.114 0.019 0.088 5.901 0.000
89
Table 25 – Short/Long Haul Market Segment Equation – 80th Percentile
Dependent Variable: Price (ln) Beta Std. Error Std. Beta t Sig. (Constant) 3.779 0.137 27.533 0.000 Market Distance (ln) 0.321 0.011 0.459 28.435 0.000 Load Factor (ln) -0.108 0.018 -0.070 -5.935 0.000 Equipment (ln) -0.223 0.016 -0.699 -13.659 0.000 Frequency (ln) 0.261 0.016 0.795 15.834 0.000 Origin & Dest. Market Share (ln) 0.036 0.008 0.070 4.300 0.000 Origin & Dest. Herfindahl (ln) -0.060 0.014 -0.064 -4.341 0.000 Route Market Share 0.170 0.036 0.105 4.715 0.000 Route Herfindahl 0.110 0.044 0.043 2.490 0.013 Comparable Cost (ln) 0.003 0.007 0.010 0.449 0.654 Tourist Route -0.238 0.018 -0.165 -13.524 0.000 Hub Airport at Origin or Dest. 0.297 0.014 0.250 20.608 0.000 Slot Restricted Airport 0.230 0.050 0.051 4.626 0.000 Long Haul Market Distance Group 0.142 0.022 0.089 6.350 0.000 Table 26 – Short/Long Haul Market Segment Equation – 80th Percentile with Low-Cost Variables
Dependent Variable: Price (ln) Beta Std. Error Std. Beta t Sig. (Constant) 3.444 0.130 26.457 0.000 Market Distance (ln) 0.341 0.011 0.488 32.176 0.000 Load Factor (ln) -0.081 0.017 -0.052 -4.740 0.000 Equipment (ln) -0.106 0.016 -0.333 -6.705 0.000 Frequency (ln) 0.149 0.016 0.455 9.363 0.000 Origin & Dest. Market Share (ln) 0.017 0.008 0.033 2.117 0.034 Origin & Dest. Herfindahl (ln) 0.035 0.015 0.037 2.405 0.016 Route Market Share 0.136 0.034 0.084 4.020 0.000 Route Herfindahl 0.031 0.044 0.012 0.718 0.473 Low-Cost Origin & Dest. Market Share (ln) -0.053 0.005 -0.148 -10.769 0.000 Low-Cost Route Market Share -0.977 0.076 -0.519 -12.773 0.000 Low-Cost Route Herfindahl 0.577 0.089 0.253 6.463 0.000 Comparable Cost (ln) 0.000 0.007 0.000 0.000 1.000 Tourist Route -0.177 0.017 -0.123 -10.437 0.000 Hub Airport at Origin or Dest. 0.110 0.015 0.093 7.212 0.000 Low-Cost Hub at Origin or Dest. -0.024 0.022 -0.012 -1.073 0.284 Slot Restricted Airport 0.080 0.047 0.018 1.717 0.086 Long Haul Market Distance Group 0.053 0.021 0.033 2.519 0.012
90
Table 27 – Summary Statistics for the Regression Equations
Dependent Variable: Price 20th 50th 80th R2 N R2 N R2 N Baseline Equation 0.271 21,573 0.217 21,573 0.329 21,573 Baseline Equation (Low-Cost) 0.280 21,573 0.274 21,573 0.421 21,573 Six Segment Equation 0.267 21,573 0.230 21,573 0.341 21,573 Six Segment Equation (Low-Cost) 0.274 21,573 0.274 21,573 0.421 21,573 Short/Long Haul Equation 0.264 21,573 0.222 21,573 0.330 21,573 Short/Long Haul Equation (Low-Cost) 0.272 21,573 0.271 21,573 0.418 21,573
91
10. Index of Tables
Table 1 Operating Cost and Revenue for Carriers of Interest ..................................... 11 Table 2 Tourist Oriented Airports................................................................................ 48 Table 3 Major Carrier Hubs......................................................................................... 49 Table 4 Low-Cost Hubs ............................................................................................... 50 Table 5 Comprehensive Carrier List............................................................................ 53 Table 6 Low-Cost Market Share by Market Segment ................................................. 68 Table 7 Summary of Significance Levels for Market Distance Variables .................. 69 Table 8 Low-Cost Airline Market Share in the Long Haul Market............................. 73 Table 9 Baseline Equation – 20th Percentile ............................................................... 78 Table 10 Baseline Equation – 20th Percentile with Low-Cost Variables...................... 78 Table 11 Baseline Equation – 50th Percentile ............................................................... 79 Table 12 Baseline Equation – 50th Percentile with Low-Cost Variables...................... 79 Table 13 Baseline Equation – 80th Percentile ............................................................... 80 Table 14 Baseline Equation – 80th Percentile with Low-Cost Variables...................... 80 Table 15 Six Segment Distance Equation – 20th Percentile.......................................... 81 Table 16 Six Segment Distance Equation – 20th Percentile with Low-Cost Variables 82 Table 17 Six Segment Distance Equation – 50th Percentile.......................................... 83 Table 18 Six Segment Distance Equation – 50th Percentile with Low-Cost Variables 84 Table 19 Six Segment Distance Equation – 80th Percentile.......................................... 85 Table 20 Six Segment Distance Equation – 80th Percentile with Low-Cost Variables 86 Table 21 Short/Long Haul Market Segment Equation – 20th Percentile ...................... 87 Table 22 Short/Long Haul Market Segment Equation – 20th Percentile with Low-Cost
Variables ................................................................................................................... 87 Table 23 Short/Long Haul Market Segment Equation – 50th Percentile ...................... 88 Table 24 Short/Long Haul Market Segment Equation – 50th Percentile with Low-Cost
Variables ................................................................................................................... 88 Table 25 Short/Long Haul Market Segment Equation – 80th Percentile ...................... 89 Table 26 Short/Long Haul Market Segment Equation – 80th Percentile with Low-Cost
Variables ................................................................................................................... 89 Table 27 Summary Statistics for the Regression Equations .......................................... 90
92
11. Index of Figures
Figure 1 Airline Operating Cost Distribution............................................................... 9 Figure 2 Low-Cost Effect on Routes.......................................................................... 58 Figure 3 Impact of a Low-Cost Presence on Hubs..................................................... 61 Figure 4 Impact of a Low-Cost Presence on Frequency ............................................ 62 Figure 5 Impact of a Low-Cost Presence on Slot Constraints ................................... 64 Figure 6 Impact of a Low-Cost Presence on the Market Distance Effect .................. 67 Figure 7 Impact of a Low-Cost Presence on the Long Haul Effect ........................... 72
93
12. Bibliography
Air Transport Association. 2001. “Statement of Carol B. Hallett, President & Chief Executive Officer Air Transport Association of America Before the United States Senate Science, Transportation and Commerce Committee Hearing on Airline Labor Relations.” Online: January 25, 2003. http://www.airtransport.org/public/testimony/display2.asp?nid=889
AMR Corporation. 2002 “American Airlines History.” Online: January 9, 2003.
http://www.amrcorp.com AMR Corporation. 2002 “American Hub Profiles.” Online: January 25 2003.
http://www.amrcorp.com AMR Corporation. 2002 “Form 10-K, For Fiscal Year Ended December, 31 2001.”
Online: January 26 2003. http://www.edgar-online.com/bin/edgardoc/finSys_main.asp?dcn=0000950134-02-001661&nad=
Beesley, Michael E. 1986. “Commitment, Sunk Costs, and Entry to the Airline Industry.”
Journal of Transport Economics and Policy, v20, n2 (September): 173-190. Berry, Steven T. 1992. “Estimation of a Model of Entry in the Airline Industry.”
Econometrica, v60, n4 (July): 889-917. Bittlemayer, George. 1990. “Efficiency and Entry in a Simple Airline Network.”
International Journal of Industrial Organization, v8, n2: 245-257. Borenstein, Severin. 1989. “Hubs and High Fares: Dominance and Market Power in the
U.S. Airline Industry.” The RAND Journal of Economics, v20, Issue 3: 415-436. Borenstein, Severin. 1990. “Airline Mergers, Airport Dominance, and Market Power.”
American Economic Review, v80, n2 (May): 400-404. Borenstein, Severin. 1992. “The Evolution of U.S. Airline Competition.” Journal of
Economic Perspectives, v6, n2: 45-73. Borenstein, Severin and Rose, Nancy L. 1994. “Competition and Price Dispersion in the
U.S. Airline Industry.” Journal of Political Economy, v102, n4 (August): 653-683.
Brander, James A. and Zhang, Anming. 1993. “Dynamic Oligopoly Behavior in the
Airline Industry.” International Journal of Industrial Organization, v11, n3 (September): 407-435.
94
Brock, James W. 2000. “Industry Update: Airlines.” Review of Industrial Organization, v16, n1 (February): 41-51.
Brueckner, Jan K., Dyer, Nichola J., and Spiller, Pablo T. 1992. “Fare Determination in
Airline Hub-and-Spoke Networks.” The RAND Journal of Economics, v23, n3 (Autumn): 309-333.
“Chasing the Sun.” Corporation for Public Broadcasting, PBS. Online: April 27, 2003.
http://www.pbs.org/kcet/chasingthesun/companies/united.html Continental Airlines. 2002. “Form 10-K, For Fiscal Year Ended December, 31 2001.”
Online: January 26, 2003. http://www.edgar-online.com/bin/edgardoc/finSys_main.asp?dcn=0000319687-02-000006&nad=
Delta Airlines. 2002. “Form 10-K, For Fiscal Year Ended December, 31 2001.” Online:
January 26, 2003. http://investor.delta.com/edgar.cfm?Page=2 DiCarlo, Lisa. 2002. “JetBlue IPO Will Fly Right For Investors.” Forbes.com. February
12th. Donnelly, Sally. 2001. “Blue Skies for JetBlue.” Time Magazine. July 13th. Donnelly, Sally. 2001. “Blue Skies.” Time Magazine. August 27th. Reprinted version
taken from JetBlue.com. Dresner, Martin E., Lin, Jiun-Sheng Chris, and Windle, Robert. 1996. “The Impact of
Low-Cost Carriers on Airport and Route Competition.” Journal of Transport Economics and Policy, v30, n3 (September): 309-328.
Evans, William. 1993. “Localized Market Power in the U.S. Airline Industry.” The
Review of Economics and Statistics, v75, Issue 1: 66-75. Joesch, Jutta and Zick, Cathleen. 1994. “Evidence of Changing Contestability in
Commercial Airline Markets during the 1980s.” Journal of Consumer Affairs, v28, n1: 1-24.
JetBlue Airways. 2002 “Form 10-Q, For the Quarterly Period Ended March, 31 2002.”
Online: January 26 2003. http://investor.jetblue.com/ireye/ir_site.zhtml?ticker=jblu&script=1901
Joy, Stewart. 1986. “Contestable Market Analysis in the Australian Domestic Airline
Industry.” Journal of Transport Economics and Policy, v20, n2: 245-54. Leahy, A. S. 1994. “Concentration in the U.S. Airline Industry.” International Journal of
Transport Economics v21, n2 (June): 209-215.
95
Levine, M.E. 1987. “Airline Competition in Deregulated Markets: Theory, Firm Strategy, and Public Policy.” Yale Journal on Regulation, v4: 393-494.
Meyer, John R. and Menzies, Thomas R. 2000. “The Continuing Vigil: Maintaining
Competition in Deregulated Airline Markets.” Journal of Transport Economics and Policy, v34, n1 (January): 1-20.
Morrison, Steven A. and Winston, Clifford. 1990. “The Dynamics of Airline Pricing and
Competition.” American Economic Review, v80, n2 (May): 389-393. Northwest Airlines. 2002. “Form 10-K, For Fiscal Year Ended December, 31 2001.”
Online: January 26, 2003. http://ir.nwa.com/ireye/ir_site.zhtml?ticker=NWAC&script=800&layout=8
Peteraf, Margaret A.1995. “Sunk Costs, Contestability and Airline Monopoly Power.”
Review of Industrial Organization, v10, n3 (June): 289-306. Reynolds-Feighan, A. 2001. “Traffic Distribution in Low-Cost and Full-Service Carrier
Networks in the US Air Transportation Market.” Journal of Air Transport Management, v7, n5 (September): 265-275.
Richards, Krista. 1996. “The Effect of Southwest Airlines on U.S. Airline Markets.”
Research in Transportation Economics, v4: 33-47. Sanchez, Felix. 2002. “Airline Wants JetBlue Slots.” Reuters. March 12th. Southwest Airlines. 2002 “Form 10-K, For Fiscal Year Ended December, 31 2001.”
Online: January 26 2003. http://www.southwest.com/investor_relations/fs_sec_filings.html
Southwest Airlines. “We Weren’t Just Airborne Yesterday.” Online: April 27, 2003.
http://www.iflyswa.com/about_swa/airborne.html “Southwest to offer nonstop, cross-country flights.” Silicon Valley / San Jose Business
Journal. May 7, 2002. Online: April 27, 2003. http://sanjose.bizjournals.com/sanjose/stories/2002/05/06/daily21.html
UAL Corporation. 2002 “Form 10-K, For Fiscal Year Ended December, 31 2001.”
Online: January 26 2003. http://www.ual.com/page/framedpage/0,1449,1381,00.html
UAL Corporation. “United Airlines History.” Online: April 27, 2003.
http://www.ual.com/page/middlepage/0,1454,2286,00.html Vowles, Timothy M. 2001. “The Southwest Effect in Multi-Airport Regions.” Journal of
Air Transport Management, v7, n4 (July): 251-258.
96
Whinston, Michael D. and Collins, Scott C. 1992. “Entry and Competitive Structure in
Deregulated Airline Markets: An Event Study Analysis of People Express.” The RAND Journal of Economics, v23, n4: 445-462.
Winston, Clifford. 1993. “Economic Deregulation: Days of Reckoning for
Microeconomists.” Journal of Economic Literature, v31, n3 (September): 1263-1289.
Windle, Robert and Dresner, Martin. 1999. “Competitive Responses to Low-cost Carrier
Entry.” Transportation Research: Part E: Logistics and Transportation Review, v35, n1 (March): 59-75.