Low Cost Airline

Discussion Paper No.
11-052
The Construction of a
Low Cost Airline Network
Kai Hüschelrath, Kathrin Müller,
and Volodymyr Bilotkach
Discussion Paper No. 11-052
The Construction of a
Low Cost Airline Network
Kai Hüschelrath, Kathrin Müller,
and Volodymyr Bilotkach
Download this ZEW Discussion Paper from our ftp server:
http://ftp.zew.de/pub/zew-docs/dp/dp11052.pdf
Die Discussion Papers dienen einer möglichst schnellen Verbreitung von

neueren Forschungsarbeiten des ZEW. Die Beiträge liegen in alleiniger Verantwortung
der Autoren und stellen nicht notwendigerweise die Meinung des ZEW dar.
Discussion Papers are intended to make results of ZEW research promptly available to other
economists in order to encourage discussion and suggestions for revisions. The authors are solely
responsible for the contents which do not necessarily represent the opinion of the ZEW.
Non-technical Summary
In a number of industries, the success and sustainability of a new firm’s busi-

ness strategy is tied to its sequential decisions to enter multiple markets. In
all these industries, firms face the key challenge to optimize the sequence of
entries in a number of different markets - taking into account possible inter-
nal resource constraints and external barriers to entry - in order to operate
profitably and to build-up a sustainable market presence. In designing such
a sustainable entry pattern, a new entrant typically has to decide on the
optimal mixture of two distinct entry strategies: entering existing markets
and facing competition of incumbent firms and entering new markets which
can be expected to contribute to the overall profitability and success of the
company.
We focus on the domestic U.S. airline industry in order to empirically
estimate entry decisions and their timing. We distinguish between entry into
new markets, i.e. markets which have not been served directly by another
airline in the year prior to the entry, and entry into existing markets, i.e.
markets which were already served directly by another airline. In particular,
we investigate the construction of a low cost airline network. We choose
JetBlue as our unit of observation, first, because of its amazing success story.
JetBlue can be seen as the only significant and successful entrant in the
domestic U.S. airline industry in the last two decades. Within 10 years this
airline rapidly grew becoming the 9th largest U.S. airline in 2009. Second, we
chose JetBlue because it is the only entrant we can observe from its inception
and track its development over almost a decade.
Adopting duration models with time-varying covariates, we find that Jet-
Blue consistently avoided concentrated airports and targeted concentrated
routes; network economies also affected entry positively. For non-stop entry
into a route that has not been served on a non-stop basis before, our analy-
sis reveals that the carrier focused on thicker routes and secondary airports,
thereby avoiding direct confrontation with network carriers. Non-stop entry
into existing non-stop markets, however, shows that JetBlue concentrated on
longer-haul markets and avoided routes already operated by either other low
cost carriers or network carriers under bankruptcy protection.
Das Wichtigste in Kürze
In vielen Branchen hängt der Erfolg und die Nachhaltigkeit der Unterneh-
mensstrategie eines neuen Unternehmens auch von deren sequentiellen Ein-
trittsentscheidungen in unterschiedliche Märkte ab. Unter Berücksichtigung
der eigenen Ressourcen und möglicher Markteintrittsbarrieren spielt dabei
auch der Zeitpunkt des Markteintritts eine entscheidende Rolle. Beim De-
sign einer nachhaltigen sequentiellen Markteintrittsstrategie muss ein Un-
ternehmen typischer Weise auch über die für es optimale Mischung zweier
Möglichkeiten entscheiden: die Erschließung neuer Märkte, auf denen das
Unternehmen eine (vorläufige) Monopolstellung erhält und den Eintritt in
existierende Märkte, in denen es in Konkurrenz zu den etablierten Anbietern
tritt.
Am Beispiel der inneramerikanische Luftverkehrsbranche untersuchen wir
Markteintrittsstrategien und deren zeitliche Wahl. Wir unterscheiden zwi-
schen Markteintritten in neue Märkte, d.h. Strecken, die noch von keiner
anderen Fluggesellschaft direkt bedient werden, und Markteintritten in exis-
tierende Märkte, d.h. direkte Strecken, auf denen schon mindestens ein an-
derer Wettbewerber aktiv ist. Im Besonderen betrachten wir den Aufbau des
Netzwerkes der Billigfluggesellschaft JetBlue. JetBlue ist das bislang einzi-
ge wirklich relevante neu in den Markt eingetretene Unternehmen, das auch
profitabel wirtschaftet. Innerhalb von zehn Jahren wuchs es zur neuntgröß-
ten amerikanischen Fluglinie heran. JetBlue ist des Weiteren die einzige neu
eingetretene Fluggesellschaft, die wir von ihrer Gründung über fast ein Jahr-
zehnt verfolgen konnten.
Unsere Ergebnisse zeigen, dass JetBlue durchweg konzentrierte Flughä-
fen meidet aber insbesondere profitable Strecken mit geringer Wettbewerb-
sintensität als Geschäftsfeld auswählt. Des Weiteren beeinflussen Netzwerks-
gesichtspunkte die Aufnahme neuer Strecken wesentlich: Je mehr Umsteige-
verbindungen aufgrund einer neuen Direktstrecke angeboten werden können,
desto Wahrscheinlicher wird ein Markteintritt von JetBlue auf dieser Strecke.
Bezüglich der Erschließung neuer Märkte favorisiert JetBlue dichte Märkte
und Nebenflughäfen. Die Entscheidung auf existierenden Strecken in Kon-
kurrenz zu anderen Fluggesellschaften zu treten trifft JetBlue insbesondere
dann, wenn es sich um längere Strecken handelt und auf dieser Strecke noch
keine anderen Billigfluglinien ihre Dienste anbieten.
The Construction of a Low Cost
Airline Network
Facing competition and exploring new markets
Kathrin Müller a Kai Hüschelrath a,b
Volodymyr Bilotkach c
July 2011
The paper investigates the construction of a low cost airline

network by analyzing JetBlue Airways’ entry decisions into non-
stop domestic U.S. airport-pair markets between 2000 and 2009.
Adopting duration models with time-varying covariates, we find
that JetBlue consistently avoided concentrated airports and tar-
geted concentrated routes; network economies also affected entry
positively. For non-stop entry into a route that has not been
served on a non-stop basis before, our analysis reveals that the
carrier focused on thicker routes and secondary airports, thereby
avoiding direct confrontation with network carriers. Non-stop en-
try into existing non-stop markets, however, shows that JetBlue
concentrated on longer-haul markets and avoided routes already
operated by either other low cost carriers or network carriers un-
der bankruptcy protection.
Keywords Airline industry, network, entry, low cost carrier

JEL Classification: L11, L23, L93
a
ZEW Centre for European Economic Research, L7 1, 68161 Mannheim, Germany. We
would like to thank Jan Brueckner, Tim Hazledine, John Kwoka, Lars Persson, Nicolas
de Roos and Miguel Urdanoz for valuable comments on previous versions of the paper.
The usual disclaimer applies.
b
WHU Otto Beisheim School of Management, Burgplatz 2, 56179 Vallendar, Germany.
c
University of California, Irvine, 3151 Social Science Plaza, Irvine, CA 92697, USA.
1. Introduction
In a number of industries, the success and sustainability of a new firm’s busi-

ness strategy is tied to its sequential decisions to enter multiple markets.
Prominent examples include fast food restaurants, supermarkets, banking
and transportation services. In all these industries, firms face the key chal-
lenge to optimize the sequence of entries in a number of different markets -
taking into account possible internal resource constraints and external bar-
riers to entry - in order to operate profitably and to build-up a sustainable
market presence. In designing such a sustainable entry pattern, a new en-
trant typically has to decide on the optimal mixture of two distinct entry
strategies: entering existing markets and facing competition of incumbent
firms and entering new markets which can be expected to contribute to the
overall profitability and success of the company.
The domestic U.S. airline industry provides a suitable environment for
an empirical assessment of the determinants of successful entry in general
and the importance of facing competition in existing markets and exploring
new markets in particular. On the one hand, this industry has experienced a
pronounced consolidation trend in the last decade, especially reflected in five
high-profile mergers1 and the demise of several smaller players such as ATA,
National and Aloha, along with a short-lived failed entry by Skybus. On the
other hand, despite this general consolidation trend, the domestic U.S. airline
industry witnessed one significant countervailing force: the market entry and
growth of JetBlue Airways.2 Since its first market appearance in February
2000 until the end of 2009, the low cost airline managed to build up a route
network with 60 destinations in 21 U.S. states and transported about 20
million domestic passengers (in 2009) making it the 9th largest airline in the
United States. Furthermore, despite its rapid growth, JetBlue Airways still
managed to realize an overall net income of USD 201 million, and therefore
belongs to the small group of profitable airlines.3
1
The mergers are American Airlines-Trans World Airlines (2001), US Airways-America
West Airlines (2005), Delta Air Lines-Northwest Airlines (2009), United Airlines-
Continental Airlines (2010), Southwest Airlines-AirTran Airways (2011).
2
We believe that it is too early at the time of this writing to consider Virgin America -
the youngest player in the domestic U.S. airline industry - as the second successful new
entrant. Since its market entry in 2008, the carrier only managed to turn profitable
for the first time in the third quarter of 2010.
3
Data source: Bureau of Transportation Statistics at
http://www.transtats.bts.gov/carriers.asp (accessed on 22 May 2011).
1
Although numerous examples of failed low cost carriers assert that a cost
advantage is not a sufficient condition for market success, JetBlue Airways is
different. One frequently cited distinctive characteristic is its innovative busi-
ness strategy that diverges from other low cost carriers in several important
dimensions. In addition to relying on secondary airports, JetBlue Airways
developed hub operations at New York’s largest airport (JFK). Furthermore,
the carrier offers high quality services including in-flight entertainment and
pre-assigned leather seats, some featuring more legroom than what is found
on traditional network carriers’ aircraft. It has also signed code-share agree-
ments with international carriers such as Lufthansa, Aer Lingus or Icelandair.
Last but not least, JetBlue Airways introduced long-haul services on a large
scale and therefore brought ‘low cost’ competition to a type of routes formerly
dominated by legacy network carriers.
Since the issue of sequential entry has been understudied in the empirical
literature, we analyze the factors that have driven JetBlue’s entry decisions,
from inception to the end of 2009. Our data analysis uses duration analysis
regression models, which have not previously been applied to airline entry
studies, but which are a popular tool in survival analysis literature. We find
that JetBlue consistently avoided concentrated airports and targeted concen-
trated routes; network economies also affected entry positively. For non-stop
entry into a route that has not been served on a non-stop basis before, our
analysis reveals that the carrier focused on thicker routes and secondary air-
ports, thereby avoiding direct confrontation with network carriers. Non-stop
entry into existing non-stop markets, however, shows that JetBlue concen-
trated on longer-haul markets, avoided slot-restricted airports, and routes
already operated by either other low cost carriers or network carriers under
bankruptcy protection.
In addition to developing an understanding of the entry strategy of a suc-
cessful low cost carrier, this study addresses the issue of the nature of entry
barriers in the U.S. airline industry. Our findings point to airport dominance
as a significant impediment to entry. We also find that the apparent entry
deterrence effect of airport dominance is not limited to hubs or large airports.
Furthermore, airport dominance deters entry into both markets where incum-
bents are present and on new non-stop routes. This suggests that network
carriers are able to use their airport dominance to prevent an entrant from
establishing a network with a hub at a different airport. At the same time,
JetBlue’s reliance on secondary airports, along with some evidence that the
2
airline avoided other low cost carriers, hints at the necessity for entrants into
the airline industry to differentiate their product as much as possible.
The remainder of the paper is structured as follows. The second section
provides a review of the literature on the determinants of entry into U.S.
airline markets. The third section gives some background on the entry and
growth of JetBlue Airways in the U.S. airline industry, followed by the pre-
sentation of our empirical analysis in Section 4. Section 5 concludes the paper
by summarizing the key results and deriving important conclusions for both
business strategy and public policy.
2. The determinants of entry into U.S. airline

markets
An airline network is constructed by multiple market entry decisions. In

determining these decisions, the airline management generally has to assess
both the external attractiveness of the candidate markets - determined by
potential customers, suppliers, competitors and partners - and the internal
capabilities and resources of the company that determine its ability to com-
pete in the respective candidate markets (see, e.g., Spulber, 2009, pp.433ff.).
Although admittedly a simplification, this section condenses down market en-
try decisions to answers of the following two questions: ’Is entry profitable?’
and ’Is entry possible?’
With respect to the profitability question, it can be expected that cur-
rent and expected profitability of a particular market typically is a key de-
terminant in the decision to enter the market. In general, it is reasonable
to assume that a profit-maximizing, risk-neutral firm will enter a market if
the net present value of expected post-entry profits is greater than the sunk
costs of entry. As post-entry profits depend on post-entry competition, the
entry decision therefore is connected to the entrant’s expectations about the
conduct and performance of the firms after entry. Furthermore, the level of
sunk costs incurred is a critical determinant of the entry decision (see e.g.
Besanko et al., 1996, 396 ff.). The higher the necessary sunk costs to enter
an industry; the higher is the risk of entry, and the lower the expected prof-
its. Additionally, the entry condition above clarifies that profits immediately
after entry are not necessary for a rational entry decision. It is sufficient
that, for instance, market growth expectations promise ample profits in the
3
future. With respect to airline markets, this condition means that an entry
decision must not be guided by the isolated profit expectations on the route
actually entered, but typically has to take account of the revenue and profit
contribution of the respective passengers over the entire network of the re-
spective airline. Concerning entry sequence, routes which are expected to be
most profitable should be entered first.
Although the expected profitability certainly is a key determinant of en-
try, empirical studies have regularly found evidence that abnormal profits are
not competed away by entry but remain persistent for longer time periods
(see Geroski (1995) for a general analysis; Joskow et al. (1994) for the U.S.
airline industry). This finding suggests that an entrant also has to address
the issue of the possible extent of entry into a particular market and implies
that a positive net present value (which at least outweighs sunk costs) is a
necessary but not sufficient condition for entry, as barriers to entry can reduce
or even eliminate entry incentives. For the U.S. airline industry, commen-
tators leave no doubt that several potentially significant (structural and/or
strategic) barriers to entry have developed after deregulation. For example,
in a report on ‘Aviation Competition - Challenges in Enhancing Competition
in Dominated Markets’, the US General Accounting Office (2001) identified
the following operating and marketing barriers, which might constrain new
entry into airline markets: access to airport facilities 4 , such as gates, ticket
counters, baggage handling and storage as well as take-off and landing slots;
frequent flyer programs; corporate incentive agreements; travel agent com-
mission overrides; flight frequency; and network size and breadth.
Given this general reasoning on the determinants of entry into airline
markets, the existing empirical research can broadly be separated into two
different strands. The first group of papers focuses on the estimation of struc-
tural models of entry decisions and consists of contributions by Bresnahan
and Reiss (1990, 1991), Reiss and Spiller (1989), Berry (1992), Dunn (2008),
and Ciliberto and Tamer (2009). Reiss and Spiller (1989) incorporate both
entry and price competition in a structural model, and investigate competi-
tion between differentiated direct and indirect services. They find that the
indirect services are significantly more competitive if a direct competitor is
also in the market. Dunn’s (2008) study investigates the decision of an airline
to offer high-quality non-stop service between cities, depending on whether
4
Scarcity of airport facilities (gates) in connection with control of gates have also been
identified as a crucial determinant of the hub premium (Ciliberto and Williams, 2010).
4
or not the carrier also offers a lower quality one-stop service. Dunn finds
that competition with rival one-stop service is an important determinant
of non-stop entry (complementary to direct competition between non-stop
entrants). The presence of a rival offering one-stop service in the market
reduces the probability of entry suggesting that there is competition between
one-stop and non-stop entry. Furthermore, Berry (1992) estimates a model
of airline entry with heterogeneous firms and finds that an airline’s market
share on routes departing from a particular airport is an important deter-
minant of entry into other routes from that airport. Ciliberto and Tamer
(2009) build on Berry’s contribution but relax the assumption that entry af-
fects the profitability of competing airlines symmetrically. They are able to
show significant heterogeneity in competition between airlines.
The second group of empirical papers follow a reduced form approach.
These studies estimate the likelihood of entry as a function of firm and mar-
ket characteristics. Starting with the contribution of Sinclair (1995) who
focuses on the importance of hub-and-spoke networks for route entry and
exit decisions, Boguslaski et al. (2004) estimate a model of city-pair entry for
Southwest Airlines using data from 1990 to 2000. In addition to a quantifi-
cation of the market characteristics which have influenced Southwest’s entry
decisions (such as especially high passenger density, short travel distances,
low income areas, prior airport presence and high route concentration), the
authors find evidence that Southwest’s entry strategies have changed signif-
icantly throughout the decade. Furthermore, Morrison and Winston (1990)
estimate probit entry models for several U.S. carriers before and after dereg-
ulation. They conclude that the airlines’ activity at origin and destination
airports is an important entry determinant. Finally, the study by Leder-
man and Januszewski (2003) estimates a reduced form model of entry into
airport-pair markets. The authors assume that an airline starts operating a
route as soon as the incremental profits - which depend on demand, cost and
expected competitive characteristics of the route - from serving that route
are positive. The model then explains entry as a function of the respective
airline’s own characteristics and the characteristics of all actual and potential
competitors on the route. The probit estimations with a dataset for the U.S.
domestic airline industry between 1996 and 2000 provide some evidence con-
sistent with the hypothesis that low cost carriers may aim at expanding the
variety of products in order to soften competition. Interestingly, Lederman
and Januszewski (2003) conclude that in order to be successful, low cost
5
carriers must either offer a differentiated product (i.e., enter new markets
in the terminology of this paper), or alternatively provide products similar
to existing ones but at lower prices (i.e., entry into existing markets in the
terminology of this paper).
In a nutshell, the review of the existing literature has shown that probit
models are often used to describe airline market entry. These models, how-
ever, do not take adequate account of the timing of entry decisions in general
and do not study these decisions from the inception of a new entrant in par-
ticular. Applying an econometric technique, which allows taking account of
the timing of entry (of a successful new entrant) is the distinguishing feature
of this study. Although duration analysis models have not been applied to
the airline industry, the model type has been used to study determinants
of both firm entry (e.g. Haveman and Nonnemaker, 2000; Fuentelsaz et al.,
2002; Fuentelsaz and Gómez, 2006) and exit (e.g. Audretsch and Mahmood,
1995; Disney et al., 2003; Mata and Portugal, 1994; Shane and Foo, 1999) in
a variety of industries.
3. Entry and growth of JetBlue Airways
Given the review of existing studies on the determinants of entry, this section
narrows the view down to the entry and growth of one particular airline: Jet-
Blue Airways. The unique position of JetBlue Airways as the only significant
and successful entrant in the domestic U.S. airline industry in the last two
decades justifies such a detailed investigation of its determinants. Before we
present our econometric approach in Section 4 - which concentrates solely on
the role of network construction in explaining the success of JetBlue Airways
- this section aims at providing some general background information on the
entry and growth of JetBlue Airways. In particular, after a brief general
characterization in Section 3.1, the subsequent Section 3.2 focuses on the
presentation and discussion of descriptive evidence on the entry pattern of
this carrier.
3.1. A brief characterization of JetBlue Airways
JetBlue Airways was founded by David Neeleman in February 1999. Neele-

man as well as several other JetBlue key executives were former Southwest
6
employees. In September 1999, JetBlue was awarded 75 take-off and landing
slots at New York’s JFK airport, followed by the granting of formal U.S. au-
thorisation in February 2000. JetBlue started operations on 11 February 2000
with services from New York JFK to Buffalo and Fort Lauderdale, rapidly
extending its route network in the following years. As of December 2009, the
carrier’s network included 60 destinations in 21 U.S. states, complemented
by destinations in eleven countries in the Caribbean and Latin America. Jet-
Blue operates a base at New York’s JFK airport, and has developed focus
city5 operations in Boston, Orlando, Fort Lauderdale, Long Beach, and San
Juan (Puerto Rico). In 2004, JetBlue transported about 11.6 million pas-
sengers on U.S. domestic flights. This number increased to about 20 million
passengers in 2009 - a share of about 3 percent of all domestic passengers
- making JetBlue the 9th largest airline in the United States.6 Despite the
rapid growth of JetBlue in partly difficult periods for the U.S. economy, the
airline realized an overall net income of $201 million from inception until the
end of 2009 (after subtracting the net losses experienced in four of the ten
business years7 ).
Although JetBlue is usually classified as low cost carrier, its business
strategy has several specific characteristics. First, the airline provides high
quality service in several important service dimensions, such as in-flight enter-
tainment and pre-assigned leather seats with more legroom. Second, JetBlue
does not only concentrate - like most other low cost carriers - on short- and
medium-haul markets but also entered long-haul markets typically only of-
fered by the network carriers. Third, JetBlue has recently started entering
into alliance agreements with foreign and domestic network carriers such as
Lufthansa8 , Aer Lingus and Icelandair (code-share agreements) or American
Airlines (interline agreement). JetBlue is considered a likely future mem-
ber of one of the three global airlines alliances: Star Alliance, SkyTeam and
oneworld.9
5
A focus city is typically defined as a location that is not a hub, but from which the
airline has non-stop flights to several destinations other than its hubs.
6
Data source: Bureau of Transportation Statistics at
http://www.transtats.bts.gov/carriers.asp (accessed on 22 May 2011).
7
The net losses were realised in 2000 (USD 21 million), 2005 (USD 20 million),
2006 (USD 1 million) and 2008 (USD 76 million). For the raw data, see
http://www.transtats.bts.gov/ (accessed on 22 May 2011).
8
See http://www.jetblue.com/about/ourcompany/lufthansa/ for a detailed characterisa-
tion of the agreement with Lufthansa (accessed on 22 May 2011).
9
Since Lufthansa acquired a 19 percent stake in JetBlue in December 2007, Star Alliance
is the most likely choice.
7
Despite its rather unconventional business strategy, a quick look at the
cost side of JetBlue reveals that it actually is a ‘low cost’ airline. While the
network carriers show average costs of 10.96 cents per available seat mile
(excluding fuel) in 2009, the average value for the low cost carrier group
drops to 7.06 cents. In 2009, JetBlue averaged 6.62 cents per available seat
mile, which is clearly below even the average cost level in the group of low
cost carriers.10 Complementary to the low cost-low fare approach, JetBlue
offers a high quality product as confirmed, e.g., by the Airline Quality Rating
(AQR) Scores11 , which always show a top rank for JetBlue Airways in both
the entire group of major airlines, and the sub-group of low cost airlines since
its first appearance in the rating in 2003.
3.2. Patterns of entry by JetBlue Airways
Given the brief general characterization of JetBlue Airways as a rather atyp-

ical low cost carrier, this section narrows the focus down to the patterns of
entry by JetBlue Airways. As a starting point for such a discussion, Figure 1
provides an overview of the entry activities of JetBlue Airways between 2000
and 2009.
35 35
30 30
25 25
Number of entries
Number of entries
20 20
15 15
10 10
5 5
0 0
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Entries into existing markets Entries into new markets ≤ 750 miles 751-1500 miles > 1500 miles
(a) Entries by type of market (b) Entries by haul
Figure 1: Entry activity of JetBlue Airways (2000-2009)

Source: U.S. DOT, T-100 Domestic Segment Data, authors’ calculations
Panel (a) of Figure 1 plots the number of non-stop market entries by Jet-
Blue Airways between 2000 and 2009. In addition to the respective overall
numbers, the panel also provides a differentiation between entries in existing
10
Data source: US DOT Form 41 via BTS, Schedule T2 & P6 & P52.
11
The AQR is a common method of comparing airline quality on combined multiple per-
formance criteria. AQR scores for the calendar year are based on 15 elements in four
major categories of airline performance: On-time performance (OT), denied boardings
(DB), mishandled baggage (MB) and customer complaints (CC). The AQR is derived
by Wichita State University (now in cooperation with Purdue University).
8
markets (i.e., markets which were already served directly by another airline
when JetBlue Airways entered) and entries in new markets (i.e., markets
which have not been served directly by another airline in the year prior to
the entry of JetBlue Airways). As shown in the panel, overall entry activ-
ity by JetBlue Airways has been substantial. Between 2000 and 2009, the
airline entered 131 domestic markets12 with a clear peak in 2006, where 31
new routes were introduced. As revealed further by the panel, entry into
new markets played a significant role in the business strategy of JetBlue Air-
ways. On average, about 40 percent of all entries created new routes, with
14 percent in 2005 and 83 percent in 2002 delineating the spectrum. Despite
the significance of entry into new markets, in 2009, only 30 percent of the
20 million JetBlue Airways passengers traveled in new markets while the re-
maining 70 percent flew in existing markets. Furthermore, JetBlue’s entry
decisions have apparently been successful in the sense that the airline only
left 23 routes out of the 131 routes it entered between 2000 and 2009. Of the
23 exits, 13 occurred in 2008, most probably in response to reduced demand
due to economic recession.
Panel (b) of Figure 1 breaks down entries by length of haul. As shown
in the panel, entry activity has been substantial in all three categories. In
sum, over the entire sample period, 42 entries (about 32 percent) took place
in long-haul markets above 1500 miles while 53 entries (about 40 percent)
were observed in medium-haul markets (751-1500 miles). The remaining 36
entries (about 28 percent) took place in short-haul markets up to 750 miles.
As further shown in panel (b), there is significant variation in the entry
activity of JetBlue Airways. While long-haul routes is the only category that
shows entry activity in every year since the birth of JetBlue Airways, short-
haul and medium-haul markets show higher absolute peaks (in 2006 and 2008
with 15 entries each).
Additionally, descriptive data analysis reveals that the share of long-haul
passengers for JetBlue Airways is significantly larger than for Southwest. Al-
though JetBlue’s entry waves in short-haul markets in 2006 and medium-haul
markets in 2008 inevitably led to a drop in the share of long-haul passengers
from its peak of 36 percent in 2005 to 23 percent in 2009, its share is still
substantially larger than in case of Southwest (8 percent). Although future
12
Non-stop services to unincorporated territories, such as Puerto Rico, are not counted
as domestic entries. Between 2000 and 2009, JetBlue started 11 non-stop services to
destinations in Puerto Rico.
9
growth of JetBlue Airways will likely lead to a further convergence - basi-
cally because the number of (potentially profitable) long-haul market entries
is limited - the focus of JetBlue Airways on long-haul routes in its first years
of existence is clearly reflected in the descriptive data analysis.13
4. Empirical Analysis
Given the review of existing evidence on the determinants of entry into air-
line markets and the identification of JetBlue Airways as a successful and
innovative new entrant, this section aims at investigating the determinants
of entry by JetBlue Airways in a rigorous econometric framework. Section
4.1 concentrates on hypothesis development, while Section 4.2 presents the
regression models and Section 4.3 describes the data set. Section 4.4 is de-
voted to the presentation of the empirical results separated between entries
into new markets and entries into existing markets.
4.1. Hypothesis development
We have identified three main categories of determinants of entry: route

characteristics, airport characteristics, and demographic characteristics. In
the following, we provide motivations for the respective variables in the three
categories.
Route characteristics
Six different route-specific variables enter our empirical analysis: Distance,

number of passengers, route HHI, LCC competition and Chapter 11. First,
with respect to the distance variable, it was suggested by the descriptive anal-
ysis above that although LCCs traditionally focused on short- and medium-
haul markets, JetBlue’s business strategy has a focus on long-haul routes. We
therefore expect that entry becomes more likely with route distance. Second,
route density was identified as a key determinant of LCC entry in previous
studies such as Ito and Lee (2003), basically because the revenue and profit
opportunities are expected to be larger. We therefore expect that the likeli-
hood of entry increases with the number of passengers traveling in a certain
13
These figures exclude JetBlue Airways’ (mostly long-haul) routes to Puerto Rico and
its international flights to the Caribbean and Latin America.
10
market. Third, a high route HHI suggests, other things equal, that competi-
tion on the respective route is less intensive and the respective market players
might enjoy market power leading to higher fares and supracompetitive prof-
its (see, e.g., Morrison and Winston, 1990). Since the most profitable routes
should be entered first, the higher a route’s HHI the earlier JetBlue’s entry
should occur. Fourth, the route presence of another low cost carrier can be
expected to reduce the profitability of entry as the respective carrier already
serves a substantial fraction of the entire low cost carrier demand potential.
We therefore expect that the existence of other low cost carriers on a route
hamper entry by JetBlue. Fifth, carriers flying under Chapter 11 bankruptcy
protection are often able to take advantage of this status to negotiate hard-
to-cut costs with employees, suppliers, and contractors and may therefore be
able to reduce fares. Furthermore, as shown by Busse (2002), airlines under
financial strain are more likely to initiate price wars. Ceteris paribus, we
therefore expect a reduced profitability of entry into routes which are also
served by airlines under Chapter 11 bankruptcy protection suggesting a neg-
ative relationship with entry activity on these routes. Sixth, although LCCs
typically focus on point-to-point traffic, it was identified above that JetBlue
focuses their operations on a hub and several focus cities. Exploiting network
economies might therefore be of major relevance for JetBlue Airways. Ceteris
paribus and given the results of previous research on airport dominance by
Borenstein (1989), we expect that network economies, as measured by the
number of potential new one-stop connections, make entry more attractive
in order to construct a sound network.
Airport characteristics
Four different airport-specific variables enter our empirical analysis: sec-

ondary airport, slot-restricted airport, airport HHI and passenger facility
charge (PFC). First, with respect to the secondary airport variable, it is a
well-documented observation that LCCs often use secondary airports as they
have lower charges, allow for shorter turnover times and increase the likeli-
hood of providing on-time flights (see, e.g., Ito and Lee, 2003). We expect
that JetBlue prefers entry on routes involving a secondary airport. Second,
LCCs tend to avoid slot-restricted airports as they are expensive to use, have
longer turnover times and increase the likelihood of delays. We expect entry
for routes that involve a slot-restricted airport is less likely. Third, LCCs tend
to avoid highly concentrated airports as the dominance of another carrier con-
11
strains growth potential, might increase airport-specific operating costs and
contains the risk of aggressive responses by the incumbent carrier (see, e.g.,
Hofer et al., 2008; Lin et al., 2002). We expect that the probability of entry
decreases for routes that involve a highly concentrated airport. Fourth, the
higher the PFC, the less lucrative is market entry other things equal. We
expect that entry occurs less often on routes that involve an airport with
high PFCs.
Demographic characteristics
Three different demographic variables enter our empirical analysis: popula-

tion, average weekly wage and unemployment rate. These variables are mea-
sured at the corresponding Metropolitan Statistical Area (MSA) level, and
obtained from the U.S. Census Bureau and the Bureau of Labour Statistics.
In regressions, we will use mean values for origin and destination airport’s
MSAs as independent variables. The expected relationships between these
variables and the hazard rate of entry are intuitive. Specifically, we expect
higher population and higher weekly wage to be associated with a higher
probability of entry, other things equal, as both these variables are measures
of market size. Higher unemployment rate will be associated with a lower
entry probability, due to lower expected demand.
4.2. The regression models
Analysis of network construction involves studying not only which routes

the airline decides to serve with non-stop flights, but also at what point in
time the entries take place. Investigating the timing of entry distinguishes
our approach from previous studies on the determinants of market entry by
low cost carriers (e.g. Boguslaski et al., 2004; Oliveira, 2008). A convenient
set of models which make it possible to account for the sequence of entry
are duration models commonly used in survival analysis, but also suitable
for entry analysis (see e.g. Haveman and Nonnemaker, 2000). These models
explain either the hazard rate λ(t) of entry (hazard rate metric) or the time
until entry takes place (accelerated failure-time metric). In the case of the
hazard rate metric, the hazard rate can be understood as the probability of
starting to serve a route directly within a short interval of time, conditional
on not having entered that route up to the starting time of the interval (see
12
Wooldridge, 2002). Formally, the hazard rate for market i at time t is given
by
P [t ≤ Ti < t + h∣t ≤ Ti ]
λi (t) = lim+ ,
h→0 h
where Ti denotes the duration between the foundation of JetBlue in the fourth
quarter of 1999, and the quarter when the carrier entered market i.
To estimate the effect of certain time-constant xi and time-varying covari-
ates zit on the hazard rate, we use a proportional hazard model, expressed
by
′ ′
λi (t∣xi , zit ) = λ0 (t) exp(xi βx + zit βz ).
As the baseline hazard λ0 (t)14 is time-dependent, but not influenced by

the covariates, each market has the same baseline hazard of entry. Therefore,
comparing market i to market m, we obtain
′ ′
λ(t∣xi , zit ) exp(xi βx + zit βz )
= ′ ,
λ(t∣xi , zmt ) exp(x′m βx + zmt βz )
which is called hazard ratio.

From the hazard rate expression it is easy to see that for a binary covariate
xk shifting from zero to one, the hazard ratio is
′ ′
λ(t∣xi , zit , xk = 1) exp(xi βx + zit βz + 1 ⋅ βk )
= ′ ′ = exp(βk ),
λ(t∣xi , zit , xk = 0) exp(xi βx + zit βz )
which gives the coefficients an easy interpretation. For a binary covariate

xk shifting from zero to one, the hazard increases by exp(βk ) − 1. The same
interpretation applies for a discrete change of a covariate by one unit. As
a semi-parametric estimation method proposed by Cox (1972) imposes no
restrictions on the shape of the baseline hazard and therefore allows the
baseline hazard to be as flexible as possible, Cox regression is used for the
analysis.
Unfortunately, hazard rates are not easily conceivable and less concrete
than a corresponding time measure. With Cox regression, predicting actual
failure times is not possible. In order to be able to give an estimate on how
14
The baseline hazard is the hazard rate of observations with zero covariates. The covari-
ates shift the baseline hazard multiplicatively.
13
the covariates of interest influence the mean time to entry, we also apply a
parametric, and thus more rigid, type of duration models. As Cleves et al.
(2004) note, the use of the accelerated failure-time metric is justified if pre-
dicting the effects on failure time is desired. Duration models in accelerated
time metric are written as
ln(ti ) = xi βx + ln(τi ) τ ∼ e.g. log-logistic
Exponentiated coefficients in accelerated failure-time models can be in-

terpreted as time ratios for a change of a dummy variable from zero to one
or a one-unit change in the corresponding covariate. In other words, if the
covariate x1 increases by one unit, the predicted time of entry increases by
exp(β1 ) − 1. Dependent on the distribution of τ we can distinguish between
several parametric models: the exponential model, the Weibull regression, the
Gompertz regression, the log-normal regression, and the log-logistic regres-
sion. We choose among these five parametric models by comparing the values
of the Akaike information criterion (AIC). According to AIC, the model with
the lowest value of the AIC is preferred. In our case, it turns out that the
data is best described when using either the log-logistic or the log-normal
duration model. Results do not differ between these models and are pre-
sented for the log-logistic model. For comparison, results for the log-normal
duration model can be found in Table 5 in the Appendix.
4.3. Data and Descriptive Statistics
For our empirical analysis, we combined data from several sources. First,
we use the U.S. Department of Transportation (U.S. DOT) DB1B Market
Origin and Destination Survey to identify a sample of possible airport-pair
routes JetBlue Airways might enter. All routes which are served at least
via two stops qualify for potential non-stop entry. Second, we add infor-
mation on population, average income, and unemployment of the respective
Metropolitan Statistical Areas from the U.S. Census Bureau and the Bureau
of Labour Statistics. In all our estimations, we restrict the sample to routes
which connect the 200 largest Metropolitan Statistical Areas (MSA).
In our regressions, we aim to explain network development. In addition
to an analysis of the entire sample, we especially investigate a split of the
sample into non-stop entries which result in a new non-stop connection (entry
14
into new markets), and non-stop entries, which took place in markets which
have been served directly by another carrier (entry into existing markets).
For entry into new markets we identified all routes which are only served via
one- or two-stop connections.
We use traffic data from the U.S. DOT in order to identify non-stop
market entry of JetBlue Airways and to construct some of our explanatory
variables. More precisely, we use the T-100 Domestic Segment database for
the period from 2000 to 2009. This data set contains monthly domestic
non-stop segment data reported by U.S. airlines when both origin and des-
tination airports are located within the boundaries of the United States and
its territories. We used T-100’s information on origin, destination, available
capacity, number of departures, and number of passengers for each of the
major carriers15 to construct a quarterly panel data-set of non-directional
non-stop airport-pair markets. We dropped airline-route observations with
fewer than 12 quarterly departures and airline-route observations which were
only served for one quarter between 2000 and 2009. An entry event of Jet-
Blue Airways is determined by the quarter when we first observe the airline
providing non-stop scheduled services16 . Furthermore, we were able to con-
struct some of the airport characteristics such as airport concentration, the
number of routes JetBlue already serves from an airport, or the airport’s pas-
senger share from T-100 data. Data on passenger facility charges collected
by airports was retrieved from the Federal Aviation Administration (FAA).
Non-stop entry is explained by route, airport, and corresponding demo-
graphic characteristics. Distance is measured as the non-stop distance in 100
miles. Existing market demand is approximated by total number of passen-
gers. As described in Section 4.1, we also account for market concentration
(Route HHI ), low cost carrier competition (LCC comp.), possible advan-
tages of carriers operating under Chapter 11 protection (Chapter 11 route)
and possibilities of exploiting network economies (Netw. economies). With
regard to airport characteristics, we include a dummy variable to indicate
15
The T-100 data set also includes traffic data for regional carriers who support the major
airlines. Although most of these typically small carriers are legally independent, their
economic existence is often tied to a large network carrier. For example, in many
instances, regional carriers do not issue their own tickets but refer to the network
carrier for all flight bookings. For our analysis, regional carriers are merged to the
respective major carrier for which they operate on a specific route.
16
We cross-checked this methodology with information on route entries provided from
JetBlue at http://www.jetblue.com/about/ourcompany/history.asp and it turned out
that all entries were correctly identified.
15
that at least one end point is a secondary airport (Secondary airp.); another
variable indicates whether either of the endpoints is a slot-restricted airport
(Slot restr. airp.); and we include a variable that indicates whether either
of the endpoints is not a major hub (Non-HUB )17 . Furthermore, we include
the mean level of passenger facility charges (PFC ), and mean concentration
(Airp. HHI ) at the respective airports. To capture demographic character-
istics, the number of inhabitants (Population), average income (Income)18 ,
and the unemployment rate (Unempl.) are included.
Given this detailed overview of the data, Table 1 provides summary statis-
tics of the variables included, while Table 4 in the Appendix summarizes the
description of the variables, both for the entire sample, and for the two sub-
samples discussed above.
Table 1: Summary statistics

All markets One- or Existing
two-stop non-stop
markets markets
Variable Mean Std. D. Mean Std. D. Mean Std. D.
Distance 11.094 7.558 11.632 7.724 8.432 6.005
Passengers 8.763 28.352 2.143 3.962 41.573 58.450
Route HHI 55.581 24.712 54.691 25.116 59.990 22.084
Chapter 11 route 0.326 0.469 0.372 0.483 0.101 0.301
LCC comp. 0.310 0.614 0.222 0.513 0.746 0.845
Netw. economies 1.259 3.880 1.135 3.779 1.876 4.290
Existing market 0.167 0.373 0.000 0.000 1.000 0.000
Secondary airp. 0.201 0.401 0.200 0.400 0.205 0.404
Slot restr. airp. 0.069 0.254 0.049 0.216 0.169 0.375
Airp. HHI 37.258 14.730 37.334 14.979 36.878 13.425
Non-HUB 0.781 0.414 0.872 0.334 0.330 0.470
PFC 2.649 0.756 2.632 0.758 2.731 0.742
Population 2.382 2.621 2.089 2.478 3.832 2.820
Income 576.107 75.723 565.377 71.114 629.290 75.424
Unempl. 5.270 1.585 5.264 1.584 5.299 1.585
Observations 420,219 349,670 70,549

Source: U.S. DOT T100 Market Segment and DB1B Origin and Destination Survey,
U.S. Census Bureau, Bureau of Labor Statistics, authors’ calculations.
17
A route is classified to be between non-HUBs if one of the airport’s passenger share is
below 0.25 percent.
18
All price variables such as average weekly wage as well as PFC are inflation adjusted.
16
4.4. Empirical Results
Our estimation results are presented in Table 2 for the Cox proportional
hazard model and in Table 3 for the log-logistic duration model. We first
report results for the entire sample; then we break the sample into cases
of entry into new markets and entry into existing markets. The distinction
between the two types of markets is simple: new markets correspond to
airport pairs that had no non-stop service before entry of JetBlue (such as
New York JFK - Long Beach route); whereas existing routes are markets
where other airlines were present before JetBlue entered (such as New York
JFK-Los Angeles (LAX) route). Out of the 124 entries covered by our data
analysis, 45 were new entries, and 79 are classified as entries into existing
routes.
As reported in Table 3, the following four factors appear in all three
regressions as robust predictors of JetBlue’s entry decisions. First, JetBlue
was more likely to enter more concentrated airport-pair routes. Holding other
variables constant, the hazard rate of entry increases by about 1.7 percent if
the route’s HHI, as measured on a 0-100 scale, increases by 1 unit. As can be
retrieved from the log-logistic regression in Table 3, a one-unit increase in the
route’s HHI corresponds to an decrease in mean time to entry of 1.4 percent.
Thus, we find a large effect on the hazard of entry and its corresponding time
to entry for a relative small increase in market concentration. However, this
result is very intuitive - other things equal - as more concentrated markets are
typically associated with higher profits, which invite entrants. Second, the
carrier shied away from concentrated airports. Comparing the magnitudes
of the two coefficients, we see that airport concentration appears as a strong
entry deterrent. For instance, comparing the effect of a decrease in airport
concentration19 with that of an otherwise equal increase in route concentra-
tion20 , we see that the former is considerably more likely to attract entry by
JetBlue than the latter. The same result applies in the accelerated failure-
time metric. While an one point increase in route HHI decreases the mean
time to entry by 1.4 percent, a one point increase in each of the respective
airport’s HHIs increases the mean time to entry by 7.7 percent.
19
A one point decrease in the mean airport HHI, i.e. a one point increase at each of the
two airports, increases the hazard of entry by 9.2 percent.
20
A one point increase in the route HHI increases the hazard by 1.7 percent.
17
Table 2: Cox Regressions on non-stop entry
All non-stop Non-stop entry Non-stop entry
entries into connecting into existing
markets non-stop markets
Variables coeff. s.e. coeff. s.e. coeff. s.e.
Distance 0.276*** (0.064) 0.089 (0.088) 0.407*** (0.083)
2
Distance -0.008*** (0.002) -0.003 (0.003) -0.011*** (0.003)
Passengers -0.000 (0.001) 0.056*** (0.017) 0.000 (0.002)
Route HHI 0.017*** (0.005) 0.018** (0.009) 0.018*** (0.007)
LCC comp. -0.180 (0.136) -0.304 (0.277) -0.366** (0.172)
Chapter 11 route -0.621** (0.282) -0.426 (0.485) -0.874** (0.424)
Netw. economies 0.108*** (0.010) 0.114*** (0.018) 0.106*** (0.013)
Existing market 1.086*** (0.310)
Secondary airp. 0.559** (0.225) 1.406*** (0.421) 0.160 (0.294)
Slot restr. airp. -0.963*** (0.328) -0.743 (0.594) -0.916** (0.409)
Airp. HHI -0.097*** (0.013) -0.057*** (0.021) -0.118*** (0.016)
Non-HUB -1.318* (0.708) -1.414 (0.952) 1.435 (1.323)
Airp. HHI × 0.009 (0.023) 0.006 (0.027) -0.083 (0.059)
Non-HUB
PFC 0.148 (0.174) 0.030 (0.241) 0.348 (0.273)
Population 0.190*** (0.038) 0.167*** (0.061) 0.186*** (0.052)
Income 0.004*** (0.001) 0.004 (0.002) 0.004** (0.002)
Unempl. -0.358** (0.169) -0.238 (0.271) -0.513** (0.234)
Observations 420,219 349,670 70,549

Markets 13,052 11,800 2,583
Entries 124 45 79
Log-likelihood -845.2 -302.8 -440.3
χ2 599.6 208.0 302.6
Notes: *** p<0.01, ** p<0.05, * p<0.1, standard errors in parentheses. Interval censoring is
accounted for using the Efron approximation.
Source: U.S. DOT T100 Market Segment Data, U.S. DOT DB1B Origin and Destination
Survey, U.S. Census Bureau, Bureau of Labor Statistics, authors’ calculations.
18
Table 3: Log-logistic Regressions on non-stop entry (accelerated failure-time
metric)
Distance -0.214*** (0.071) -0.078 (0.066) -0.460*** (0.139)
Distance2 0.005** (0.002) 0.002 (0.002) 0.014*** (0.005)
Passengers -0.001 (0.001) -0.046*** (0.018) -0.003* (0.002)
Route HHI -0.014** (0.006) -0.016* (0.008) -0.018** (0.009)
LCC comp. 0.266** (0.128) 0.420* (0.236) 0.466** (0.193)
Chapter 11 route 0.304 (0.242) -0.017 (0.290) 0.572 (0.490)
Netw. economies -0.140*** (0.038) -0.117*** (0.044) -0.243*** (0.072)
Existing market -1.062*** (0.334)
Secondary airp. -0.761*** (0.259) -1.604*** (0.583) -0.329 (0.343)
Slot restr. airp. 0.937*** (0.355) 0.815 (0.576) 0.800 (0.498)
Airp. HHI 0.074*** (0.016) 0.046** (0.021) 0.095*** (0.023)
Non-HUB 1.316* (0.675) 1.604* (0.910) -0.824 (1.299)
Airp. HHI × 0.001 (0.021) -0.007 (0.023) 0.060 (0.053)
Non-HUB
PFC -0.202 (0.190) -0.109 (0.218) -0.367 (0.358)
Population -0.233*** (0.054) -0.192*** (0.069) -0.257*** (0.080)
Income -0.001 (0.001) -0.002 (0.002) -0.001 (0.002)
Unempl. 0.588*** (0.135) 0.373*** (0.136) 0.873*** (0.215)
Constant 7.136*** (1.431) 8.628*** (2.254) 6.920*** (2.347)
Observations 420,219 349,670 70,549
Markets 13,052 11,800 2,583
Entries 124 45 79
Log-likelihood -404.0 -186.6 -189.4
χ2 660.1 231.6 352.0
Notes: *** p<0.01, ** p<0.05, * p<0.1, standard errors in parentheses.
19
This finding is in line with Hofer et al. (2008) who find that the largest
components of price premiums are those from airport concentration, rather
than route concentration. Third, JetBlue is apparently more likely to en-
ter a route, if the carrier can sufficiently exploit network economies. If the
number of new one-stop connections JetBlue can serve after a non-stop en-
try between two airports increases by one, the hazard of entry increases by
11 percent (= exp(0.108) − 1). In terms of time to entry, one new poten-
tial one-stop connection decreases the mean time to entry by 13 percent
(= exp(−0.140) − 1). This is also understandable as a significant sunk in-
vestment into airport facilities, marketing etc. is necessary to start off a
new airport presence. Furthermore, this result can be interpreted as a clear
indication that JetBlue rather wanted to construct a hub-and-spoke than
a point-to-point network. Hub and focus city development eases (aircraft)
scheduling and boosts the possibilities of JetBlue to offer connecting flights.
Fourth, the effect of population on the likelihood of entry is also robust and
significant in all specifications. An increase in one million inhabitants in
each of the endpoint cities increases the hazard of entry by 21 percent and
decreases time to entry by 21 percent.
With respect to the remaining variables, we observe the following effects.
Distance exhibits a significant effect in the entire sample, and for entries into
existing markets. Consistent with what is believed about JetBlue’s strategy,
the carrier is more likely to enter longer-haul routes already served by its
competitors. The effect of distance is decreasing, as evidenced by the negative
coefficient on distance squared. The effect is mainly driven by entry into
existing non-stop markets, since the effect is not statistically significant for
entry into connecting markets. Number of passengers served on the market
predicts entry into new routes, but not into existing markets. This result
simply implies that JetBlue did its homework to identify markets with many
connecting passengers (traveling with one or two stops en route), and without
non-stop services. Presence of other low cost carrier(s) serves as an important
deterrent for entry into existing markets. JetBlue also tried to avoid routes,
served by the airlines under Chapter 11 bankruptcy protection. Although this
result might appear counter-intuitive at first, our discussion above indicates
that both Chapter 11-specific cost advantages and an increased likelihood of
price wars can explain why JetBlue tended to avoid those markets. However,
entry deterrence of Chapter 11 bankruptcy protection cannot be confirmed
in the accelerated failure-time metric.
20
Interestingly, support for the commonly accepted wisdom that low cost
carriers tend to choose secondary airports appears mixed. It is true that
JetBlue is more likely to choose secondary gateways when entering new mar-
kets; however, the corresponding coefficient is not significant for regression
analyzing the carrier’s entry into existing routes. Part of this finding can be
explained by the observation that JetBlue operates many of its (long-haul)
flights from New York JFK and Boston Logan to mostly primary airports at
the West Coast.
Coming back to the issue of airport concentration, the negative relation-
ship between airport HHI and the likelihood of entry could have two expla-
nations. First, we could interpret this result as implying that JetBlue tries
avoiding hub airports, which tend to be concentrated. The second explana-
tion relates to economies of scale: airports located in smaller metropolitan
areas might only sustain limited services, and end up being monopolized by
a single airline, feeding traffic from the airport to its hub. The airline oper-
ating the nearest hub will then end up as a dominant carrier in such small
gateways. To account for this possibility, we have included a non-hub air-
port dummy, and interacted that dummy with the airport HHI. To identify
non-hub airports, we have used the Federal Aviation Administration’s clas-
sification: non-hub airports are defined as airports serving more than 10,000
but less than 0.25 percent of entire domestic passenger traffic annually. Our
results provide very little support to the contention that the effect of airport
concentration on entry is specific to airports of certain size.
Last but not least, our results show the expected effects of demographic
variables on entry, even though the corresponding coefficients are not always
significant. Interestingly, comparing the coefficients of our income variable
with the respective results reported in the study of Boguslaski et al. (2004)
reveals substantial differences. While Southwest is found to target on highly
price elastic passengers in low income cities with favorable operational condi-
tions, our analysis for JetBlue fails to find such an effect. This result therefore
supports out earlier observation that the business strategies (and customer
groups) of both carriers differ significantly.
With respect to the implications of our results for entry barriers in the
airline industry, we can say the following. First, the fact that JetBlue fo-
cused on secondary airports establishing new non-stop service indicates that
it avoided direct confrontation with the network carriers. The other impor-
tant finding is that JetBlue was more likely to enter routes if it was able to
21
exploit network economies by increasing the number of connection flights.
Collectively, the two results imply that airport presence itself may be an im-
portant entry barrier, consistent with Berry (1990) and Borenstein (1989).
Not only airport presence itself, but also the extent of this presence is a
defining factor, as clearly evidenced by the relationship between airport HHI
and likelihood of entry. As for the route-level entry barriers; the identity
of your competitors apparently matters a lot. We find some evidence that
JetBlue avoided routes, where it would have to compete with the low cost
and currently financially distressed carriers. This suggests that the carrier
evaluated the likely post-entry competition when making entry decisions.
5. Conclusion
In the last decade, the domestic U.S. airline industry has experienced a sub-
stantial consolidation trend. In addition to a number of high level mergers,
several smaller carriers had to leave the industry. The only significant coun-
tervailing force of this development has been the entry and growth of JetBlue
Airways. Since its first market appearance in February 2000 until the end
of 2009, the low cost airline managed to build up a route network with 60
destinations in 21 U.S. states and transported about 20 million passengers
(in 2009) making it the 9th largest airline in the United States. Furthermore,
despite its rapid growth, JetBlue Airways still managed to realize an overall
net income of USD 201 million, and therefore belongs to the small group of
profitable airlines.
Against this background, we analyze the factors that have driven Jet-
Blue’s entry decisions, from inception to the end of 2009. Our data analysis
uses Cox proportional regression models and, in order to provide a more intu-
itive picture, parametric duration models in accelerated failure-time metric.
These have not previously been applied to airline entry studies, but they
are a popular tool in the survival analysis literature. We find that JetBlue
consistently avoided concentrated airports and targeted concentrated routes;
network economies also affected entry positively. For non-stop entry into
a route that has not been served on a non-stop basis before, our analysis
reveals that the carrier focused on thicker routes and secondary airports,
thereby avoiding direct confrontation with network carriers. Non-stop entry
into existing non-stop markets, however, shows that JetBlue concentrated on
22
longer-haul markets, avoided slot-restricted airports, and routes already op-
erated by either other low cost carriers or network carriers under bankruptcy
protection.
In addition to developing an understanding of the entry strategy of a suc-
cessful low cost carrier, this study addresses the issue of the nature of entry
barriers in the U.S. airline industry. Our findings point to airport dominance
as a significant impediment to entry. We also find that the apparent entry
deterrence effect of airport dominance is not limited to hubs or large airports.
Furthermore, airport dominance deters entry into both markets where incum-
bents are present and on new non-stop routes. This suggests that network
carriers are able to use their airport dominance to prevent an entrant from
establishing a network with a hub at a different airport. At the same time,
JetBlue’s reliance on secondary airports, along with some evidence that the
airline avoided other low cost carriers, hints at the necessity for entrants into
the airline industry to differentiate their product as much as possible.
From a business strategy perspective, it can be concluded that successful
entry into the U.S. airline industry is still possible as long as the respective
entrant understands the key industry characteristics and growth needs and
is able to position itself taking into account its relative strengths and weak-
nesses. The case of JetBlue Airways has especially shown that significant
structural and strategic entry barriers can be overcome by a combination of
entry into existing and new markets driven by an innovative general business
strategy. Although entry into existing markets may yield higher revenues,
entry into new markets has the key advantage of avoiding the costs of com-
peting against incumbents and is therefore likely to contribute substantially
to the overall profitability and success of the company.
From a public policy perspective, it is very likely that consumers gain
substantially from the existence and growth of JetBlue Airways through sig-
nificant reductions in fares - first and foremost on the airport-pair actually
entered but also on adjacent city-pairs or on routes which face an elevated
probability of entry by JetBlue Airways. In order to keep and further extend
these benefits, antitrust authorities are not only well advised to monitor the
industry to identify potential forms of anticompetitive behavior by incumbent
firms but they should especially be skeptical with respect to any initiative of
network carriers to acquire or merge with JetBlue Airways. Given the effi-
ciency and significance of JetBlue, it is very likely that loosing this ’maverick
firm’ would cause substantial anticompetitive effects on many U.S. domestic
23
routes.
Although the paper investigates the entry strategy of one particular firm
in one particular industry, our results generally suggest that the art of suc-
cessful firm entry in industries with multiple markets includes both facing
competition by incumbents and exploring new markets. Although new firms
are well advised to differentiate their products in order to reduce competitive
pressure, successful and sustainable entry often cannot avoid overcoming en-
try barriers and competing with incumbents directly. Following the method-
ology of Paul Geroski (1991, 1995), industries which demand multiple entries
in different markets therefore are a nice example for the importance of both
imitative entry into existing markets and innovative entry into new markets.
Independent of an answer to the question which type of entry is more im-
portant for the company or social welfare, consumers will surely profit from
both increases in competitive pressure and spirit of innovation.
References
Audretsch, D. and T. Mahmood (1995): “Firm Survival: New Results
Using a Hazard Function,” Review of Economics and Statistics, 77, 97–103.
Berry, S. (1990): “Airport Presence as Product Differentiation,” American

Economic Review, Papers and Proceedings, 80, 394–399.
——— (1992): “Estimation of a Model of Entry in the Airline Industry,”

Econometrica, 60, 889–917.
Besanko, D., D. Dranove, and M. Shanley (1996): The Economics

of Strategy, New York: John Wiley & Sons.
Boguslaski, R., H. Ito, and D. Lee (2004): “Entry Patterns in the

Southwest Airlines Route System,” Review of Industrial Organization, 25,
317–350.
Borenstein, S. (1989): “Hubs and High Fares: Dominance and Market

Power in the US Airline Industry,” RAND Journal of Economics, 20, 344–
365.
Bresnahan, T. F. and P. C. Reiss (1990): “Entry in Monopoly Markets,”

The Review of Economic Studies, 57, 531 –553.
24
——— (1991): “Entry and Competition in Concentrated Markets,” The Jour-
nal of Political Economy, 99, 977–1009.
Busse, M. (2002): “Firm Financial Condition and Airline Price Wars,”

RAND Journal of Economics, 33, 298–318.
Ciliberto, F. and E. Tamer (2009): “Market Structure and Multiple

Equilibria in Airline Markets,” Econometrica, 77, 1791–1828.
Ciliberto, F. and J. Williams (2010): “Limited Access to Airport Fa-

cilities and Market Power in the Airline Industry,” Journal of Law and
Economics, 53, 467–495.
Cleves, M., W. Gould, and R. Gutierrez (2004): An Introduction to

Survival Analysis Using Stata, College Station, Texas: Stata Press.
Cox, D. (1972): “Regression Models and Life Tables,” Journal of the Royal
Statistical Society, 34, 187–220.
Disney, R., J. Haskel, and Y. Heden (2003): “Entry, Exit and Estab-
lishment Survival in UK Manufacturing,” Journal of Industrial Economics,
51, 91–112.
Dunn, A. (2008): “Do low-quality products affect high-quality entry? Multi-

product firms and nonstop entry in airline markets,” International Journal
of Industrial Organization, 26, 1074–1089.
Fuentelsaz, L. and J. Gómez (2006): “Multipoint Competition, Strate-

gic Similarity and Entry Into Geographic Markets,” Strategic Management
Journal, 27, 477–499.
Fuentelsaz, L., J. Gómez, and Y. Polo (2002): “Followers’ Entry

Timing: Evidence From the Spanish Banking Sector After Deregulation,”
Strategic Management Journal, 23, 245–264.
Geroski, P. (1995): “What Do We Know about Entry?” International

Journal of Industrial Organization, 13, 421–440.
Geroski, P. A. (1991): Market Dynamics and Entry, Oxford: Oxford Uni-

versity Press.
Haveman, H. and L. Nonnemaker (2000): “Competition in Multiple

Geographic Markets: The Impact on Growth and Market Entry,” Admin-
istrative Science Quarterly, 45, 232–267.
25
Hofer, C., R. Windle, and M. Dresner (2008): “Price Premiums and
Low Cost Carrier Competition,” Transportation Research Part E, 44, 864–
882.
Ito, H. and D. Lee (2003): “Low Cost Carrier Growth in the US Airline
Industry: Past, Present and Future,” Working paper, Cambridge.
Joskow, A., G. Werden, and R. Johnson (1994): “Entry, Exit, and

Performance in Airline Markets,” International Journal of Industrial Or-
ganization, 12, 457–471.
Lederman, M. and S. Januszewski (2003): “Entry Patterns of Low-Cost

Airlines,” Working paper, Cambridge.
Lin, J.-S., M. Dresner, and R. Windle (2002): “Determinants of Price

Reactions to Entry in the U.S. Airline Industry,” Transportation Journal,
41, 5–22.
Mata, J. and P. Portugal (1994): “Life Duration of New Firms,” Journal

of Industrial Economics, 42, 227–245.
Morrison, S. and C. Winston (1990): “The Dynamics of Airline Pricing

and Competition,” American Economic Review, Papers and Proceedings,
80, 389–393.
Oliveira, A. V. (2008): “An empirical model of low-cost carrier entry,”

Transportation Research Part A: Policy & Practice, 42, 673–695.
Reiss, P. C. and P. T. Spiller (1989): “Competition and Entry in Small

Airline Markets,” Journal of Law and Economics, 32, S179–S202.
Shane, S. and M.-D. Foo (1999): “New Firm Survival: Institutional Ex-
planations for New Franchisor Mortality,” Management Science, 45, 142–
159.
Sinclair, R. A. (1995): “An empirical model of entry and exit in airline

markets,” Review of Industrial Organization, 10, 541–557.
Spulber, D. (2009): Economics and Management of Competitive Strategy,

New Jersey: World Scientific.
US General Accounting Office (2001): “Aviation Competition - Chal-

lenges in Enhancing Competition in Dominated Markets,” Gao/01-518t,
Washington D.C.
26
Wooldridge, J. (2002): Econometric Analysis of Cross Section and Panel
Data, Cambridge, MA: MIT Press.
27
A. Appendix
Table 4: Description of variables

Variable Description
Distance Non-stop distance in 100 miles
Passengers Sum of passengers (in thousands) in the previous period
(DB1B)
Route HHI Route HHI (0-100) in the previous period, passenger-
based (DB1B)
Chapter 11 route At least one Chapter11-airline serves the route
LCC comp. Number of other low-cost carrier serving the market
Netw. economies Number of potential new one-stop connection routes
(= # B6 routes Airport A + # B6 routes Airport B)
Existing market Route is already served directly by some other airline
Secondary airp. At least one airport is a secondary airport
Slot restr. airp. At least one airport is slot-restricted
Airp. HHI HHI (0-100) at the airports in the previous period
(Mean), passenger-based
Non-HUB Airport’s passenger share is below 0.25 percent (at least
one of the airp.)
PFC Passenger Facility Charge in 1995 USD (Mean)
Population Population estimate in Mio inhabitants (Mean)
Income Average weekly wage in 1995 USD (Mean)
Unempl. Unemployment rate (Mean)
28
Table 5: Log-normal Regressions on non-stop entry (accelerated failure-time
metric)
Distance -0.174*** (0.058) -0.085 (0.066) -0.368*** (0.141)
Distance2 0.004** (0.002) 0.002 (0.002) 0.011** (0.005)
Passengers -0.001 (0.002) -0.045** (0.019) -0.002 (0.002)
Route HHI -0.013** (0.005) -0.014* (0.007) -0.016 (0.010)
LCC comp. 0.240* (0.127) 0.425* (0.221) 0.393* (0.208)
Chapter 11 route 0.357 (0.229) 0.168 (0.271) 0.516 (0.532)
Netw. economies -0.144*** (0.031) -0.118*** (0.041) -0.259*** (0.097)
Existing market -0.955*** (0.303)
Secondary airp. -0.661*** (0.235) -1.179*** (0.442) -0.397 (0.367)
Slot restr. airp. 0.856** (0.335) 0.709 (0.513) 0.908 (0.563)
Airp. HHI 0.069*** (0.014) 0.043** (0.019) 0.093*** (0.026)
Non-HUB 1.529** (0.598) 1.648** (0.808) -1.260 (1.555)
Airp. HHI × -0.015 (0.018) -0.012 (0.021) 0.077 (0.061)
Non-HUB
PFC -0.127 (0.164) -0.012 (0.188) -0.618 (0.465)
Population -0.215*** (0.046) -0.178*** (0.060) -0.286*** (0.096)
Income -0.001 (0.001) -0.002 (0.002) -0.001 (0.002)
Unempl. 0.541*** (0.106) 0.370*** (0.127) 0.892*** (0.259)
Constant 7.298*** (1.299) 8.160*** (1.908) 7.444*** (2.723)
Observations 420,219 349,670 70,549
Markets 13,052 11,800 2,583
Entries 124 45 79
Log-likelihood -398.5 -180.8 -192.3
χ2 672.6 242.8 348.2
Notes: *** p<0.01, ** p<0.05, * p<0.1, standard errors in parentheses.
29

Low Cost Airline

Uploaded by

Copyright:

Available Formats

Low Cost Airline

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Low Cost Airline

Uploaded by

Copyright:

Available Formats

Discussion Paper No.

Download this ZEW Discussion Paper from our ftp server:

Die Dis­­cus­­si­­on Pape­rs die­­nen einer mög­­lichst schnel­­len Ver­­brei­­tung von

In a number of industries, the success and sustainability of a new firm’s busi-

Kathrin Müller a Kai Hüschelrath a,b

The paper investigates the construction of a low cost airline

Keywords Airline industry, network, entry, low cost carrier

In a number of industries, the success and sustainability of a new firm’s busi-

2. The determinants of entry into U.S. airline

An airline network is constructed by multiple market entry decisions. In

3. Entry and growth of JetBlue Airways

3.1. A brief characterization of JetBlue Airways

JetBlue Airways was founded by David Neeleman in February 1999. Neele-

3.2. Patterns of entry by JetBlue Airways

Given the brief general characterization of JetBlue Airways as a rather atyp-

(a) Entries by type of market (b) Entries by haul

Figure 1: Entry activity of JetBlue Airways (2000-2009)

4.1. Hypothesis development

We have identified three main categories of determinants of entry: route

Six different route-specific variables enter our empirical analysis: Distance,

Four different airport-specific variables enter our empirical analysis: sec-

Three different demographic variables enter our empirical analysis: popula-

4.2. The regression models

Analysis of network construction involves studying not only which routes

As the baseline hazard λ0 (t)14 is time-dependent, but not influenced by

which is called hazard ratio.

which gives the coefficients an easy interpretation. For a binary covariate

ln(ti ) = xi βx + ln(τi ) τ ∼ e.g. log-logistic

Exponentiated coefficients in accelerated failure-time models can be in-

4.3. Data and Descriptive Statistics

Table 1: Summary statistics

Observations 420,219 349,670 70,549

Observations 420,219 349,670 70,549

Berry, S. (1990): “Airport Presence as Product Differentiation,” American

——— (1992): “Estimation of a Model of Entry in the Airline Industry,”

Besanko, D., D. Dranove, and M. Shanley (1996): The Economics

Boguslaski, R., H. Ito, and D. Lee (2004): “Entry Patterns in the

Borenstein, S. (1989): “Hubs and High Fares: Dominance and Market

Bresnahan, T. F. and P. C. Reiss (1990): “Entry in Monopoly Markets,”

Busse, M. (2002): “Firm Financial Condition and Airline Price Wars,”

Ciliberto, F. and E. Tamer (2009): “Market Structure and Multiple

Ciliberto, F. and J. Williams (2010): “Limited Access to Airport Fa-

Cleves, M., W. Gould, and R. Gutierrez (2004): An Introduction to

Dunn, A. (2008): “Do low-quality products affect high-quality entry? Multi-

Fuentelsaz, L. and J. Gómez (2006): “Multipoint Competition, Strate-

Fuentelsaz, L., J. Gómez, and Y. Polo (2002): “Followers’ Entry

Geroski, P. (1995): “What Do We Know about Entry?” International

Geroski, P. A. (1991): Market Dynamics and Entry, Oxford: Oxford Uni-

Haveman, H. and L. Nonnemaker (2000): “Competition in Multiple

Joskow, A., G. Werden, and R. Johnson (1994): “Entry, Exit, and

Lederman, M. and S. Januszewski (2003): “Entry Patterns of Low-Cost

Lin, J.-S., M. Dresner, and R. Windle (2002): “Determinants of Price

Mata, J. and P. Portugal (1994): “Life Duration of New Firms,” Journal

Morrison, S. and C. Winston (1990): “The Dynamics of Airline Pricing

Oliveira, A. V. (2008): “An empirical model of low-cost carrier entry,”

Reiss, P. C. and P. T. Spiller (1989): “Competition and Entry in Small

Sinclair, R. A. (1995): “An empirical model of entry and exit in airline

Spulber, D. (2009): Economics and Management of Competitive Strategy,

US General Accounting Office (2001): “Aviation Competition - Chal-

Table 4: Description of variables

Die Discussion Papers dienen einer möglichst schnellen Verbreitung von